"
Geophysical Monograph 186
Amazonia and Global Change Michael Keller Mercedes Bustamante John Gash Pedro Silva Dias
Editors
~ American Geophysical Union Washington, DC
• • <\1
,I,·,
Published under the aegis of the AGU Books Board
CONTEN1S'
Kenneth R. ~inschwaner, Chair; Gray E. Bebout, Joseph E. Borovsky, Kenneth H. Brink, RalfR. Haese, Robert B. Jackson, W. Berry Lyons, Thomas Nicholson, Andrew Nyblade, Nancy N. Rabalais, A. SUIjalal Sharma, Darrell Strobel, Chunzai Wang, and Paul David Williams, members.
J Preface John Gash, Michael Keller, Mercedes Bustamante, and Pedro Silva Dias
Library of Congress Cataloging-in-Publication Data Amazonia and global change / Michael Keller ... [et al.]. p. em. (Geophysical monograph; 186) Includes bibliographical references and index. ISBN 978-0-87590-476-4 (alk. paper) 1. Rain forest ecology-Amazon River Region. 2. Biosphere-Research-Amazon River Region. 3. Climatic changes-Amazon River Region. 4. Amazon River Region-Climate. I. Keller, Michael, 1960-
ix
Section I: People and Land Change People and Environment in Amazonia: The LBA Experience and Other Perspectives M. Batistella, D. S. Alves, E. F. Moran, C. Souza Jr., R. Walker, and S. Walsh
QH112.A433 2009 577.34/1409811-dc22 2009040686 ISBN: 978-0-87590-476-4 ISSN: 0065-8448
Cover Photo: The Igarape Asu in the Instituto Nacional de Pesquisas da Amazonia (INPA) research catchment north of Manaus. Photo courtesy of John Gash.
Copyright 2009 by the American Geophysical Union 2000 Florida Avenue, N.W. Washington, DC 20009
1
The Changing Rates and Patterns of Deforestation and Land Use in Brazilian Amazonia Diogenes S. Alves, Douglas C. Morton, Mateus Batistella, Oar A. Roberts, and Carlos Souza Jr
11
Selective Logging and Its Relation to Deforestation Gregory P. Asner, Michael Keller, Marco Lentini, Frank Merry, and Carlos Souza Jr
25
The Spatial Distribution and Interannual Variability of Fire in Amazonia Wilfrid Schroeder, Ane Alencar, Eugenio Arima, and Alberto Setzer
43
The Expansion of Intensive Agriculture and Ranching in Brazilian Amazonia Robert Walker, Ruth DeFries, Maria del Carmen Vera-Diaz, Yosio Shimab,ukuro, and Adriano Venturieri
61
Scenarios of Future Amazonian Landscapes: Econometric and Dynamic ~imulation Models Stephen Perz, Joseph P. Messina, Eustaquio Reis, Robert Walker, and Stephen}. Walsh
83
Road Impacts in Brazilian Amazonia Alexander Pfaff, Alisson Barbieri, Thomas Ludewigs, Frank Merry, Stephen Perz, and Eustaquio Reis
101
Small Farmers and Deforestation in Amazonia Eduardo S. Brond/zio, Anthony Cak, Marcellus M. Caldas, Carlos Mena, Richard Bilsborrow, Celia T. Futemma, Thomas Ludewigs, Emilio F. Moran, and Mateus Batistella
11 7
Figures, tables and short excerpts may be reprinted in scientific books and journals ifthe source is properly cited.
Section II: Atmosphere and Climate Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, is granted by the American Geophysical Union for libraries and othet; users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee of $1.50 per copy plus $0.35 per page is paid directly to CCC, 222 Rosewood Dr., Danvers, MA 01923.0065-8448/09/$01.50+0.35. This consent does not extend to other kinds of copying, such as copying for creating new collective works or for resale. The reproduction of multiple copies and the use of full articles or the use of extracts, including figures and tables, for commercial purposes requires permission from the American Geophysical Union.
Understanding the Climate of Amazonia: Progress From LBA Carlos A. Nobre, Jose A. Marengo, and Paulo Artaxo
145
Characteristics of Amazonian Climate: Main Features Carlos A. Nobre, Guillermo O. Obregon, Jose A. Marengo, Rong Fu, and German Po veda
149
The Amazonian Boundary Layer and Mesoscale Circulations A. k. Betts, G. Fisch, C. von Randow, M. A. F. Silva Dias, J. c. P. Cohen, R. da Silva, and D. R. Fitzjarrald........ 163
Printed in the United States ofAmerica.
Natural Volatile Organic Compound Emissions From'Plants and Their Roles in Oxidant Balance and Particle Formation JOrgen Kesselmeier, Alex Guenther, Thorsten Hoffmann, Maria TereSa Piedade, and Jorg Warnke
i
183
Ecophysiology of'Forest and Savanna Vegetation . j. Lloyd, M. tjoulden, J. P. Ometto, S. Patino, N. M. Fyllas, and C. A. Quesada
Biomass Burning in Amazonia: Emissions, Long-Range Transport of Smoke and Its Regional and Remote Impacts
K. M. Longo, S. R. Freitas, M. O. Andreae, R. Yokelson, and P. Artaxo
207
Section IV: Surface Water
Aerosol Particles in Amazonia: Their Composition, Role in the Radiation Balance, Cloud Formation, and Nutrient Cycles
Paulo Artaxo, Luciana V. Rizzo, Melina Paixao, Silvia de Lucca, Paulo H. Oliveira, Luciene L. Lara, Kenia T. Wiedemann, Meintat O. Andreae, Brent Holben, Joel Schafer, Alexandre L. Correia, and Theotonio M. Pauliquevis
Surface Waters in Amazonia: Key Findings and Perspectives
John M. Melack, Reynaldo L. Victoria, and Javier Tomasella 233
485
The Role of Rivers in the Regional Carbon Balance
Jeffrey E. Richey, Alex V. Krusche, Mark S. Johnson, Hillandia B. da Cunha, and Maria V. Ballester
Modeling the Regional and Remote Climatic Impact of Deforestation
M. A. Silva Dias, R. Avissar, and P. Silva Dias
251
Evapotranspiration Humberto R. da Rocha, Antonio O. Manzi, and Jim Shuttleworth
261
.489
Water and Chemical Budgets at the Catchment Scale Including Nutrient Exports From Intact Forests and Disturbed Landscapes
Javier Tomasella, Christopher Neill, Ricardo Figueiredo, and Antonio D. Nobre
505
Floodplain Ecosystem Processes
Global Warming and Climate Change in Amazonia: Climate-Vegetation Feedback and Impacts on Water Resources
Jose Marengo, Carlos A. Nobre, Richard A. Betts, Peter M. Cox, Gilvan Sampaio, and Luis Salazar
463
John M. Melack, Evlyn M. L. M. Novo, Bruce R. Forsberg, Maria T. F. Piedade, and Laurence Maurice 273
525
Effects of Climatic Variability and Deforestation on Surface Water Regimes
Marcos Heil Costa, Michael T. Coe, and Jean Loup Guyot
543
Section III: Terrestrial Ecosystems Section V: Conclusions and Vision for the Future
Biogeochemistry and Ecology of Terrestrial Ecosystems of Amazonia
Yadvinder Malhi and Eric A. Davidson
293
Nutrient Limitations to Secondary Forest Regrowth
Eric A. Davidson and Luiz A. Martinelli
Results From LBA and a Vision for Future Amazonian Research
M. Batistella, P. Artaxo, C. Nobre, M. Bustamante, and F. Luizao
299
Index
,~
i I
The Maintenance of Soil Fertility in Amazonian Managed Systems
Flavio}. Luizao, Philip M. Fearnside, Carlos E. P. Cerri, and Johannes Lehmann
311
Sources and Sinks of Trace Gases in Amazonia and the Cerrado
M. M. C. Bustamante, M. Keller, and D. A. Silva
.337
The Production, Storage, and Flow of Carbon in Amazonian Forests
Yadvinder Malhi, Sassan Saatchi, Cecile Girardin, and Luiz E. O. C. Aragao
.355
Changes in Amazonian Forest Biomass, Dynamics, and Composition, 1980-2002
Oliver L. Phillips, Niro Higuchi, Simone Vieira, Timothy R. Baker, J:(uo-Jung Chao, and Simon L. Lewis
.373
Ecosystem Carbon Fluxes and Amazonian Forest Metabolism
Scott Saleska, Humberto da Rocha, Bart Kruijt, and Antonio Nobre
389
The Regional Carbon Budget
R. A. Houghton, Manuel Gloor, Jon Lloyd, and Christopher Potter
409
The Effects of Drought on Amazonian Rain Forests
P. Meir, P. M. Brando, D. Nepstad, S. Vasconcelos, A. C. L. Costa, E. Davidson, S. Almeida, R. A. Fisher, E. D. Sotta, D. Zarin, and G. Cardinot
429
Soil Carbon Dynamics
Susan Trumbore and Plfnio Barbosa de Camargo
451
555 565
y
PREFACE
J
/
Writing about Amazonia demands superlatives: the world's most extensive area of tropical forest, the world's greatest river, the world's most species-diverse ecosystem, the world's largest store of aboveground carbon; the list goes on. We add one more: Amazonia, subject of the largest, coordinated, scientific study into a specific region of the world's land surface. That study, an international experiment led by Brazil, is the Large-Scale Biosphere-Atmosphere Experiment in Amazonia, also known as LBA. The ambitious objective ofLBA was to understand how Amazonia functions as an entity, as a whole ecosystem. This task was made all the more urgent, yet equally all the more difficult, by the fact that Amazonia is in a state offlux. Climate change, combinedwith land cover change in the form of deforestation, has created a three-dimensional moving picture of interacting causes and effects: To capture this dynamic situation, LBA adopted the design philosophy that the big picture would only emerge from an understanding of the component pieces and the interactions between them, building up regional-scale understanding from local measurements. This book synthesizes the results of that LBA research, bringing together the most important new scientific results and the new understanding that has resulted. The statistics on LBA are impressive: nearly 2000 scientists (including over 500 Ph.D. and masters students) producing at least 1300 scientific papers. Reviewing all of this research in a single book is a daunting task and a process that inevitably reflects the personal perspectives of the editors and authors. Nevertheless, we hope to have covered the whole spectrum of research: the human dimensions, the meteorology and atmospheric chemistry, the ecology and biogeochemistry ofthe land surface, and the hydrology. Despite the integration of research within LBA, there is a continuing need to improve communications between disciplines and for individual scientists to see their own research in the context of the overall
effort to understand the Amazonian ecosystem. Recognizing this need, the Scientific Steering Committee of LBA asked us to edit this book, to bring all this research together within one cover. An important legacy ofLBA has been the training of a new generation of young environmental scientists who are now responsible for continuing the next phase of LBA. We envision that this book will be a resource to underpin that future research. LBA is a special program of the Brazilian Ministry of Science and Technology (MCT), and we acknowledge their continuing support in the planning and implementation of this research. Funds have been provided by a number of Brazilian national and state agencies and funding agencies in the United States, the European Union, and elsewhere. In particular, we acknowledge major funding from MCT, NASA, and the European Commission. We acknowledge a debt to Carlos Nobre and Diane Wickland for their vision and leadership, p~rticularly in the early stages of the design of LBA. Our personal thanks also go to Lorena Brewster for coordinating the Ibditing of this volume. John Gash Centre for Ecology and Hydrology Michael Keller International Institute ofTropical Forestry, USDA Forest Service NEON, Inc. Mercedes Bustamante University ofBrasilia Pedro Silva Dias University ofSao Paulo
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10. 1029/2009GM000883
ix
,"
People and Environment in Amazonia: The LBA Experience and Other Perspectives M. Batistella, l D. S. Alves,2 E. F. Moran,3 C. Souza Jr.,4 R. Walker,s and S. J. Walsh6 Amazonia is the arena for an ongoing extraordinary transfonnation of nature and society. This process of change can be depicted in many ways and by various disciplines, with emphasis on the biosphere or the atmosphere, as demonstrated by the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA). However, the human factors behind environmental change should not be neglected. This chapter introduces the section on people and environment in the region proposing an examination of the human dimensions of land use and land cover change from the LBA experience and other perspectives. As a basis for this approach, we provide a brief review on related topics and insights about opportunities for integrative research. Selected findings produced by LBA projects are highlighted and a synthetic view on research gaps, analytical gaps, data gaps, and policy implications of human dimension research in Amazonia is [presented. 1. MOTIVATIONS
The chapters in this section of the book examine a variety of human impacts on ecosystems, landscapes, and regions as a consequence of different processes occurring in Amazonia, for example, deforestation and land use change [Alves et al., this volume; Brondizio et al., this volume], selective log-
lEmbrapa Satellite Monitoring, Campinas, Brazil. 2Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos, Brazil. 3Department of Anthropology and Anthropological Center for Training and Research on Global Environmental Change, Indiana University, Bloomington, Indiana, USA. 4Instituto do Homem e Meio Ambiente da Amazonia, Belem, Brazil. 5Department of Geography, Michigan State University, East Lansing, Michigan, USA. 6Department of Geography, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA. Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2009GM000902
ging [Asner et al.~ this volume], fire occurrences [Schroeder et al., this volume], the expansion of intensive agriculture and ranching [Walker et al., this volume], road building, and development [Pfaff et al., this volume]. Scenarios of future Amazonian landscapes built with simulation models are also discussed [Perz et al., this volume]. Assuming the current transformation ofnature and society in Amazonia as an interactive process, this chapter proposes a broad examination of the human dimensions ofland use and land cover change from the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) and other perspectives. One way to approach this theme is to take the major scientific questions posed by LBA and identify situations where the human perspective is implicated, for example, when human agents are responsible, directly or indirectly, for changes in land use and land cover. Another way might be to examine the epistemology behind LBA to understand the role of science, and scientists, in formulating the mix of articulated disciplines and questions. A third way could be to examine biophysical dimensions, the climate change drivers, and their impacts on human society. Our goals are not to enumerate all these options but to present how a human-centered perspective came to be part of LBA and what insights we have gained to date from that perspective. The next section will focus on the major
2
BATISTELLA ET AL.
PEOPLE AND ENVIRONMENT IN AMAZONIA
questions proposed for LBA and why a human dimension was inherent, and indeed necessary, given how these questions were formulated. Our focus is on the human dimensions of land use and land cover change because it was under this topic that the substantive work on human dimensions was originally developed within LBA science. 2. AN OVERVIEW ON THE HUMAN DIMENSION RESEARCH IN LBA Taking the two fundamental questions of LBA, to understand how Amazonia functions as a regional system and how changes in land use, land cover, and climate might affect its functioning, it is evident that one could not answer these two questions by examining only the biophysical systems. At the center of change in Amazonia is the fact that the region is experiencing migration and settlement, high rates of deforestation and logging, the development of roads and urban centers, and land use intensification. These changes in land use and land cover will have an impact on Amazonia's longterm functioning. People are the agents of change, the agents that disturb the system, and also the ones that will suffer the consequences of such disturbance. LBA has learned to address how human-driven land use change affected land cover and how climate variability has been influenced by human changes in land cover and ecosystem functioning. People are a major force in the biosphere (and the atmosphere), but human actions are mediated by human institutions at a variety ofscales, from local, to regional, to national, and international. It is not simply individuals or households, which act upon the environment of Amazonia; it is also government, nongovernmental organizations, and other forms of organized groups. These groups do not always work in concert. They differ in their goals, and how these conflicting goals play out is part of human dimension research. Soon after its inception, LBA leading investigators recognized the need to develop a human dimensiOli research agenda within the program. As with the International GeosphereBiosphere Programme (IGBP), which realized in 1988 that it needed to invite social scientists to help understand the human drivers of global change, LBA became concerned early with the impact of environmental changes on people and of people on the environment. To explore this required social science, the challenge was how to include this human dimension expertise in an effective way, given that the original LBA questions were not conceived with this perspective in mind and were drawn up without social scientists' participation. The original formulation involved climatologists, biogeochemists, hydrologists, ecologists, and other biophysical scientists. As the LBA research progressed, the need for understanding the human dimensions of change in Amazo-
nia became so evident that even sensitivities to the theme by the Brazilian government, related to policy issues, were overcome, leading to recognized progress in articulating the natural and the social sciences [Batistella et al., 2008]. Four strategies were adopted to develop the· human dimension component of LBA. First, efforts were made to find ad hoc partnerships to jointly formulate scientific questions reflecting the social processes behind land use and land cover change and the impacts of environmental change upon human health. Second, efforts were made to examine the policy implications ofLBA for human resources, education, and training, particularly with regard to how it might affect the sustainable development of the region and the role of the state in such development. Third, I:BA sought to strengthen the bridges with the social sciences, by inviting social scientists to join its Scientific Steering Committee and later developing programmatic initiatives that were inclusive of social science questions [see Lahsen, 2002; Alves et al., 2004; Schor, 2005; Alves, 2007a, 2007b, 2008; Egler and Ibanez, 2007]. Finally, in the context of the transition to the second phase of LBA, a number of initiatives were taken to bring decision makers and stakeholders to present and discuss scientific findings and new integrative questions. Among the relevant results was a bibliographic study to review the scientific production during the 1990s in the region, with a focus on four major themes: populations, ethnic and cultural representations; agropastoral and extractive activities; industrial activities; and urban networks [Becker, 2007a]. The search found a reasonable amount of work in the social sciences, both in regional institutions, and outside the region. It revealed considerable depth in research on sociopolitical issues, particularly on modernization and social change, the expansion of the agropastoral frontier in the period 1960-1985, the impact of the current frontier expansion, and of new dynamics of regional change. The greatest amount of debate in the literature focused on land use, particularly the use of forest resources, the impacts of cattle ranching, extractivism, and the different ways, some of them predatory, ofland appropriation dominating the region. Other areas of considerable debate in the literature deal with deforestation, logging, and forest management and, more recently, urban development in the region as an alternative to rural living. The study identified some major topics for future research: the need for more attention to intraregional migration, the potential and limitations of various forms of production, and the role of cities and networks promoting regional development and change. Another key initiative to foster the integration of social sciences in LBA was the workshop "The Human Dimensions of Environmental Change and LBA," held in May 2004. It was a singular opportunity to address three main
concerns about the· importance of social sciences to LBA: (l) human dimensions,of environmental change in Amazonia, including the id,9rltification of research gaps and analytical tools to condu<j suc~ research; (2) data availability and quality; and (3)J1Olicy making. The workshop represented a rare chance for articulation among LBA and social scientists, allowing the formulation of research questions, not addressed in the original LBA scientific plan. A collection of selected papers produced by this workshop can be found in the work of Costa et al. [2007]. The workshop emphasized knowledge gaps and the need for new analytical data and tools of interest to LBA, particularly related to issues
3
of sustainability. Table 1 synthesizes the outcomes of the workshop. Far from being a complete picture, it offers a possible path for future initiatives integrating social and natural sciences in Amazonia. 3. HUMAN AND BIOPHYSICAL DIMENSIONS OF LAND USE AND LAND COVER CHANGE IN AMAZONIA: SELECTED FINDINGS AND OTHER PERSPECTIVES A variety of relevant topics were posed by LBA scientists and contributors to the discussion of human dimensions
Table 1. Research Gaps, Analytical Gaps, Data Gaps, and Policy Implications of Human Dimension Research in Amazonia
Topics
Institutions and Governance
Logistics and Regional Development
Production Systems
Populational Mobility
Research gaps
Role and weight of institutions; Public/private relations; Links between markets and the State
Differentiation of territorial units; Links with institutions
Urban and rural linkages
Inter-regional! intraregional patterns; Linkages with all other topics (e.g., LUCC)
Analytical gaps
Sociology of action
Network analysis
Environmental valuation; Regional accounting; Production functions; Data integration
Spatially explicit models
Data gaps
Case studies
Land zoning data; Census data (including census-tract level)
Environmental, social, and economic variables; Census data (including
Intramunicipal data (mostly from field work)
I
census~tract
Implications for policy making
Assessment of institutional lack and efficiency
Infrastructure and urban planning; Land zoning
level) Land zoning; Institutional building
Urbanization
Land Use/Cover Changes (LUCC)
Regional patterns for deforestation and abandonment dynamics; Linkages between agricultural production and LUCC (e.g., intensification, degradation); New occupation fronts Multiscale analysis Tools for characterization assessments ofland of urban areas use expansion and concentration; Intra-annual classification of crops and pastures; Analysis based on land parcels and agrarian structure Deforestation data Data for periods from the 1970s; between Intra-annual remote censuses sensing data; Agricultural production data; Land tenure data
Urban typology; Linkages between urban infrastructure and all other topics (e.g., LUCC)
Job creation; Infrastructure and Infrastructure and urban planning urban planning
Land zoning and planning; Deforestation control
4
PEOPLE AND ENVIRONMENT IN AMAZONIA
of Amazonian environmental changes. The economic questions, for example, their relation with deforestation and other social and environmental dimensions, have revealed multitiereci processes and highlighted the limitations of data gaps, analytical gaps, and incomplete knowledge [Perz et al., this volume]. The main challenges still reflect the need to balance economic development and nature conservation, including the difficulties of ensuring sustainability through market integration. In addition to these difficulties, the agrarian situation represents -distinct opportunities and limitations for actors [Costa, 2007a; Walker et al., this volume; Brondizio et al., this volume]. As a consequence, some areas show emerging production systems, while others maintain their business as usual [Costa, 2007b]. The contrast between macroeconomic approaches and case studies at local scales remains an important source of discussion. This emerges from the research about land use patterns and processes at various scales and from the challenge ofmultiscale integration, as revealed by the chapters of this section ofthe book. Two different problems can be recognized: (l) understanding the system functioning as a regional entity, a wellknown challenge of scale integration within LBA [LBA, 1996; Nobre et al., 2001] and (2) identifying and comparing different locations, looking for understanding of social processes with higher or lower chances of environmental and economic sustainability [Batistella and Moran, 2005]. Logistics and regional development are closely related with the economic system, but one can also find connections with other dimensions, particularly with geopolitics and policy making. Becker [2007b] emphasizes the singular dynamics referring to the soybean phenomenon in Amazonian frontiers, its production chains, its impact on the organization of actors, mainly due to the different roles of smallholders and largeholders, and policies related to infrastructure development. Walker et al. [this volume] discuss policies that created the preconditions for modem Amazonian agriculture as well as describing cattle ranching and soybean market set. tings and trajectories of expansion. The role of roads and networks to provide access to resources and markets has also been highlighted [Pfaff et al., this volume]. These artificial landscape corridors should be considered in territorial planning, land zoning, and geopolitics. However, the differentiation ofland units and institutional arrangements is far from being achieved. Moreover, understanding population mobility and strategies of occupation through the region will expose complex trajectories [Hogan et al., 2008]. In recent years, the population in Amazonia has grown increasingly urbanized, but the consequences of patterns and processes of urbanization to land use and land cover change remain a research topic to be developed.
BATISTELLAET AL.
The questions addressing land use and land cover change in Amazonia are central to LBA as they articulate with most of the research components of the experiment (Figure 1). However, some scientific gaps remain. For example, knowledge is still incomplete about the regional patterns of deforestation and land abandonment, the identification and quantification of land use intensification and land degradation processes, and the immediate tracking of new fronts of occupation [Alves, 2001]. Data gaps include deforestation assessments for dates before the 1970s, remote sensing seasonal data, regional and local data about the agrarian structure and agropastoral production. Without these inputs~ it is virtually impossible to carry out comprehensive multiscale analyses, intra-annual classifications of agricultural and pasture lands, as well as studies based on land parcels. These issues have clear policy implications, particularly for land zoning and deforestation monitoring. Several initiatives within LBA addressed changes in land use and land cover. Considering only LBA-Ecology, a NASA-funded program on terrestrial ecology, there were 38 projects under this research component (Figure 1). These projects used a variety of perspectives, as illustrated by Figure 2. On the other hand, only five projects addressed human dimensions of Amazonian change (Figure 3). In general, progress was made in understanding the relationship between certain types ofland use and forest conversion, in particular, with regard to logging, cattle ranching, and small agriculture. Also of late, there has been progress in understanding the dynamics of conversion to mechanized soybean production, the transformation of land from forest matrices to agropastoral production centered landscapes, and the dynamics of fire in their relation to altered regions through selective logging and other land uses. The expansion of intensive agriculture, the impact of roads and networks on deforestation, and scenario developments also produced relevant results through LBA. Chapters in this section present findings and considerations on these matters. Some aspects deserve special attention when examining the biophysical and human dimensions of land use and land cover change in Amazonia. Differences in soil quality, for example, explain much of the variance in the rates of secondary succession, in crop choices, and persistence of farmers on the rural properties [Moran et al., 2000]. This finding highlights that we cannot overlook soil quality assessment as a key variable that makes a real difference in social and environmental outcomes. Farmers with high quality soils were able to persist on their properties despite ups and downs of the economy over a 30-year period. When these soils were located favorably to markets, this advantage further multiplies. These differences in soil quality are particularly no-
5
Physical Climate System Water and Energy
..' ...--4----+
I...-_A~tm_o_sP_h_e_ric_tC~he_tru_'s_try~~_K0 t
t
Carbon Storage and Exchange
Biogeochemistry: Trace Gases and Nutrients
Land Surface Hydrology and Water Chemistry
t Land Use / Land Cover
E2H1...-~~H_uma~n_D_I_·m_e_ns_io_n_s~~----J Figure 1. Number of projects by research component of LBA-ECO (the projects funded during the phase of synthesis and integration are in parentheses).
table when we compare results across regions [Tucker et al., 1998]. In interregional comparisons, biophysical factors are often more explanatory than human factors, and broad differences in agricultural potential are more likely to be considered by policy makers, thereby further enhancing the natural effect of biophysical differences.
However, the use and management of land better explains the differences in ~he rate of secondary succession and agriculture intensification when we compare locations within a region [Moran arid Brondizio, 1998]. This is not surprising, as the detailed Idiowledge of a given property allows differences in management to be used in explanation. These
Figure 2. Research themes for projects under the component "Land Use and Land Cover" ofLBA-ECO.
6
PEOPLE AND ENVIRONMENT IN AMAZONIA
BATISTELLA ETAL.
of forest'reserves also play an important role in maintaining lower levels of fragmentation [Batistella et al., 2003; Batistella and Bront!'izio, 2004]. The policy imPlications of such considerations are crucial for further init~ives regarding settlement implementation. Development and conservation strategies can be informed by the results achieved, but regional dynamics and local context should be taken into account to avoid political failures. The path for a reasonable conceptual approach explaining the heterogeneous processes of colonization in Amazonia is far from achieved, but analyses of landscape structure and function can provide a unique way to understand the spatial characteristics of Amazonian land change. Land tenure, type of settlement, the developmental cycles, period and cohort effects also affect patterns of land use and land cover [McCracken et al., 1999]. The cohort effect, for example, persists despite period effects, i.e., events such as tight credit, hyperinflation, and other macroeconomic forces affect the magnitude but not the overall trajectory of deforestation [Evans et al., 2001]. On the other hand, the conservation of relatively large areas of forest within human settlements is more effective if dependent on institutions' self-organization relative to the needs of the population and to demarcating areas of reserve with rights given to local people vested in its protection [Batistella, 2001]. The understanding of human and environment interactions, in particular, the role of cohort, age, period effects, external capital, and household processes upon land cover trajectories in colonization areas, has progressed [Walker et al., 2000; Brondizio et al., 2002]. Among other initiatives, agent-based models incorporate household demographics and labor arrangements, land cover and allocation, soil quality and crop productivity, and spatial features of the farm lot [Deadman et al., 2004]. However, the role of social variables to understand the dynamics ofland use and land cover in Amazonia is still poorly explored. Social studies rarely investigate the outcomes in terms of land change, and land use/land cover assessments rarely include the social dimensions of land change [Turner et aI., 2004]. This research gap reveals an opportunity for future studies about the human-environment interactions in the region.
Figure 3. Research themes for projects under the component "Human Dimensions" ofLBA-ECO.
differences are commonly left out in aggregate analyses, the potential land degradation risks caused by deforestation where farmers may at best be treated as small or large, rather and associated soil erosion in Brazilian Amazonia [Lu et al., than having inherent different qualities in their stewardship 2007]. of the land. Vegetation change detection has long been regarded as a Monitoring land cover change in Amazonia has evolved challenge, especially in the moist tropical regions. Hybrid significantly during the last decade. Stages of secondary approaches combining image differencing and postclassisuccession can now be associated with spatial and spectral fication comparison are promising in detecting vegetation patterns that can be captured via analysis of remote sensed change trajectories, especially for vegetation gain and loss images, and further, can be used to estimate biomass and [Lu et al., 2008]. carbon, as well as to infer cycles ofproduction [Moran et al., The search for quantitative methods to analyze and de1994; Lu et al., 2005]. Spectral mixture analysis and classifi- scribe the structure of landscapes has also become a high ers using spatial, spectral, and textural information are better priority. Land use and land cover issues are at the core of able to capture the heterogeneity of landscapes with greater this perspective, due to their intricate dynamics and conaccuracy [Lu et al., 2004]. sequences in landscape structure and function. Landscape Classification of successional forest stages remains a fragmentation is the process whereby a landscape matrix challenging task because of the lack of sharp distinctions , is progressively subdivided into smaller and more isolated between adjacent stages and confusion between early suc- patches, mainly as a result of human land use activities. The cessional forest stages with degraded pasture and advanced design of Amazonian settlements affects the structure of successional forest stages with perennial plantations and landscapes and the processes of fragmentation. Orthogonal agroforestry. Accurate classification of these land covers settlement structure (such as the classic fishbone patterns and associated biomass estimation has become a significant found in Rondonia) produce greater forest fragmentation, factor in reducing the uncertainty of carbon emission and have lesser spatial complexity, and less interspersion between sequestration [Zarin et al., 2005; NeefJet al., 2006]. landscape classes than settlements designed as a function of Integration of remote sensing and geographic informa- topographic variability. The desi~ based on topography tion systems allows the evaluation and mapping of soil ero- plays an important role in maintaining or increasing forest sion risk in a large area. When high-quality topographic and interior habitat relative to the entire landscape area, lowerclimate data are not available, a surface cover index based ing the impact of forest fragmentation onthe occurrence and purely on remote-sensing data is useful to evaluate and map distribution of organisms. The maintenance of large patches
4. CONCLUDING REMARKS
I
j
1
The following seven chapters discuss human and biophysical dimensions of land use and land cover change in Amazonia assuming that the understanding of changes in Amazonian landscapes and regions depend on documentation of alterations in land cover and trajectories of land use. These land changes are intriguing processes to investigate,
7
as they produce relevant environmental outcomes and social feedbacks, such as land appropriation and conflicts, agricultural production systems, and human-dominated landscapes. The spatial nature of human-environment interactions in Amazonia brings up issues of scale and levels of analyses, providing opportunities for the study of spatially explicit processes, such as tropical deforestation and its impacts from region to household. The chapters in this section of the book provide a general picture about the accomplishments and limitations of this integrative research and also indicate promising grounds for future studies.
REFERENCES Alves, D. S. (2001), 0 processo de desmatamento na Amazonia, Parcerias Estrategicas, 12,259-275. Alves, D. S. (2007a), Science and technology and sustainable development in Brazilian Amazon, in Stability ofTropical Rainforest Margins, Linking Ecological, Economic and Social Constraints ofLand Use and Conservation, edited by T. Tscharntke et al., pp. 1-20, Springer, Germany. Alves, D. S. (2007b), Cenarios de Cobertura e Uso da Terra e Dimens6es Humanas no LBA, in Dimensoes Humanas da Biosfera-Atmosfera da Amazonia, edited by W. M. da Costa, B. K Becker, an? D. S. Alves, pp. 39-64, EDUSP, Sao Paulo. Alves, D. S. (2008),'Taking things public: A contribution to address human dimensiohs of environmental change, Philos. Trans. R. Soc. Ser. B, 363,11903-1909, doi:10. 1098/rstb.2007.0020. Alves, D. S., B. K, Becker, and M. Batistella (2004), Land cover/ land use change and human dimensions in the Large Scale Biosphere-Atmosphere Experiment in Amazonia (LBA), LUCC Newsletter, 10, 4-5. Alves, D. S., D. C. Morton, M. Batistella, D. A. Roberts, and C. Souza Jr. (2009), The changing rates and patterns of deforestation and land use in Brazilian Amazonia, Geophys. Monogr. Ser., doi:l0.1029/2008GM000722, this volume. Asner, G. P., M. Keller, M. Lentini, F. Merry, and C. Souza Jr. (2009), Selective logging and its relation to deforestation, Geophys. Monogr. Ser., doi:1O.1029/2008GM000723, this volume. Batistella, M. (2001), Landscape change and land uselland cover dynamics in Rondonia, Brazilian Amazon, Ph.D. dissertation, 399 pp., Indiana University, Bloomington, Ind. Batistella, M., and E. S. Brondizio (2004), Uma estrah:gia integrada de analise e monitoramento do impacto ambiental de assentamentos rurais na Amazonia, in Avaliat;:iio e Contabilizat;:iio de Impactos Ambientais, edited by A. R. Romeiro, pp. 74-86, Editora Unicamp, Campinas. Batistella, M., and E. F. Moran (2005), Dimens6es humanas do uso e cobertura das terras na Amazonia: uma contribui!;ao do LBA, Acta Amazonica, 35(2), 239-247. Batistella, M., S. Robeson, and E. F. Moran (2003), Settlement design, forest fragmentation, and landscape change in Rondonia, Amazonia, Photogramm. Eng. Remote Sens., 69(7), 805-812.
8
PEOPLE AND ENVIRONMENT IN AMAZONIA
Batistella, M., E. F. Moran, and D. S. Alves (Eds.) (2008), Amazonia: Natureza e Sociedade em Transformar;iio, 304 pp., EDUSP, Sao Paulo. (2007a), Sintese da prodm;ao cientifica em ciencias Becker, B. humanas na Amazonia: 1990-2002, in Dimensiies Humanas da Biosfera-Atmosfera da Amazonia, edited by W. M. da Costa, B. K. Becker, and D. S. Alves, pp. 13-38, EDUSP, Sao Paulo. Becker, B. K. (2007b), Reflex5es obre a geopolitica e a logistica da soja na Amazonia, in Dimensiies Humanas da BiosferaAtmosfera da Amazonia, edited by W. M. da Costa, B. K. Becker, and D. S. Alves, pp. 113-128; EDUSP, Sao Paulo. Brondizio, E. S., S. D. McCracken, E. F. Moran, AD. Siqueira, D. R. Nelson, and C. Rodriguez-Pedraza (2002), The colonist footprint: Towards a conceptual framework of deforestation trajectories among small farmers in Frontier Amazonia, in Land Use and Deforestation in the Amazon, edited by C. Wood and R. Porro, pp. 133-161, Univ. Press of Florida, Gainesville, Fla. Brondizio, E. S., A Cak, M. M. Caldas, C. Mena, R. Bilsborrow, C. T. Futemma, T. Ludewigs, E. F. Moran, and M. Batistella (2009), Small farmers and deforestation in Amazonia, Geophys. Monogr. Ser., doi:l0.l029/2008GM000716, this volume. Costa, F. A (2007a), A questao agniria na Amazonia e os desafios de urn novo desenvolvimento, in Dimensiies Humanas da Biosfera-Atmosfera da Amazonia, edited by W. M. da Costa, B. K. Becker, and D. S. Alves, pp. 129-166, EDUSP, Sao Paulo. Costa, W. M. (2007b), Tendencias recentes na Amazonia, in Dimensiies Humanas da Biosfera-Atmosfera da Amazonia, edited by W. M. da Costa, B. K. Becker, and D. S. Alves, pp. 81-1 I 1, EDUSP, Sao Paulo. Costa, W. M. da, B. K. Becker, and D. S. Alves (Eds.) (2007), Dimensiies Humanas da Biosfera-Atmosfera da Amazonia, 176 pp., EDUSP, Sao Paulo. Deadman, P., D. Robinson, E. Moran, and E. S. Brondizio (2004), Colonists household decision making and land use change in the Amazon rainforest: An agent-based simulation, Environ. Plann. B Plann. Des., 31, 693-709. Egler, P. C. G., and M. G. V. Ibafiez (2007), Construindo pontes entre gerac;;ao de conhecimentos e a formulac;;ao de politicas publicas, in Dimensiies Humanas da Biosfera-Atmosfera da Amazonia, edited by W. M. da Costa, B. K. Becker, and D. S. Alves, pp. 167-174, EDUSP, Sao Paulo. Evans, T. P., A Manire, F. de Castro, E. S. Brondizio, and S. D. McCracken (2001), A dynamic model of household decision making and parcel-levelland cover change in the eastern Arnazon, Ecol. Modell., 143(1-2), 95-113. Hogan, D. J., A de O. D'Antona, andR. L. Carmo (2008), Dinamica demognifica recente na Amazonia, in Amazonia: Natureza e Sociedade em Transformar;iio, edited by M. Batistella" E. F. Moran, and D. S. Alves, pp. 71-116, EDUSP, Sao Paulo. Lahsen, M. (2002), Brazilian climate epistemers' multiple epistemes: An exploration of shared meaning, diverse identities and geopolitics in Global Change Science, Discussion Paper 2002-01, Belfer Center for Science and International Affairs (BCSIA), Environment and Natural Resources Program, Kennedy School of Government, Harvard University, Cambridge, Mass.
f.
BATISTELLA ET AL. LBA (1996), Concise Science Plan. (Available at http://lba.cptec. inpe.br/lba/?p=3&lg=eng., accessed 19 April 2006). Lu, D., P. Mausel, M. Batistella, and E. F. Moran (2004), Comparison of land-cover classification methods in the Brazilian Amazon Basin, Photogramm. Eng. Remote Sens., 70(6),723-·,-731. Lu, D., M. Batistella, and E. F. Moran (2005), Satellite estimation of aboveground biomass and impacts of forest stand structure, Photogramm. Eng. Remote Sens., 71(8),967-974. Lu, D., M. Batistella, P. Mausel, and E. F. Moran (2007), Mapping and monitoring land degradation risks in the Western Brazilian Amazon using Multitemporal Landsat TMlETM+ images, Land Degrad. Dev., 18,41-54. Lu, D., M. Batistella, and E. F. Moran (2008), Integration of Landsat TM and SPOT HRG images for vegetation change detection in the Brazilian Amazon, Photogramm. Eng. Remote Sens., 73(4),421-430. McCracken, S., E. S. Brondizio, D. Nelson, E. F. Moran, AD. Siqueira, and C. Rodriguez-Pedraza (1999), Remote sensing and GIS at farm property level: Demography and deforestation in the Brazilian Amazon, Photogramm. Eng. Remote Sens., 65(11), 1311-1320. Moran, E. F., and E. S. Brondizio (1998), Land-use change after deforestation in Amazonia, in People and Pixels: Linking Remote Sensing and Social Science, edited by D. Liverman et aI., pp. 94-120, National Academy Press, Washington, D.C. Moran, E. F., E. S. Brondizio, P. Mausel, and Y. Wu (1994), Integrating Amazonian yegetation, land-use, and satellite data, BioScience, 44(5), 329-339. Moran, E. F., E. S. Brondizio, 1. M. Tucker, M. C. Silva-Forsberg, S. D. McCracken, and 1. Falesi (2000), Effects of soil fertility and land-use on forest succession in Amazonia, For. Ecol. Manage., 139, 93-108. Neeff, T., R. Lucas, J. R. dos Santos, E. S. Brondizio, and C. Freitas (2006), Area and age of secondary forests in Brazilian Amazonia, Ecosystem, 9, 609-623. Nobre, C. A, D. Wickland, and P. 1. Kabat (2001), The Large Scale Biosphere-Atmosphere Experiment in Amazonia (LBA), IGBP Newsletter, 45, 2-4. Perz, S., J. P. Messina, E. Reis, R. Walker, and S. 1. Walsh (2009), Scenarios of future Amazonian landscapes: Econometric and dynamic simulation models, Geophys. Monogr. Ser., doi: 10.10291 2008GM000736, this volume. Pfaff, A, A Barbieri, T. Ludewigs, F. Merry, S. Perz, and E. Reis (2009), Road impacts in Brazilian Amazonia, Geophys. Monogr. Ser., doi:IO.1029/2008GM000737, this volume. Schor, T. (2005), Ciencia e tecnologia: Uma interpretac;;ao da pesquisa na Amazonia-O caso do Experimento de Grande Escala da Biosfera-Atmosfera na Amazonia (LBA), Tese de Doutorado, Universidade de Sao Paulo, Sao Paulo. Schroeder, W., A Alencar, E. Arima, and A Setzer (2009), The spatial distribution and interannual variability of fire in Amazonia, Geophys. Monogr. Ser., doi:l0.1029/2008GM000724, this volume. Tucker, 1. M., E. S. Brondizio, and E. F. Moran (1998), Rates of forest regrowth in eastern Amazonia: A comparison of Aitamira
and Bragantina region,S; Para State, Brazil, Interciencia, 23(2), 64-73. Turner, B. L., E. F. Moran, and R. Rindfuss (2004), Integrated land-change sC2eyce and its relevance to the human sciences, in Land Change ~ence: Observing Monitoring, and Understanding Trajectories of Change on the Earth's Surface, edited by G. Gutman et aI., pp. 431-448, Springer, New York. Walker, R., E. F. Moran, and L. Anselin (2000), Deforestation and cattle ranching in the Brazilian Amazon: Extemal capital and household processes, World Dev., 28(4), 683-699. Walker, R., R. DeFries, M. del C. Vera-Diaz, Y. Shimabukuro, and A Venturieri (2009), The expansion of intensive agriculture and ranching in Brazilian Amazonia, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000735, this volume. Zarin, D., et al. (2005), Legacy of fire slows carbon accumulation in Amazonian forest regrowth, Front. Ecol. Environ., 3(7), 365-369.
9
D. S. Alves, Instituto Nacional de Pesquisas Espaciais (INPE), DPI (SRE 2), Avenida dos Astronautas 1758, CEP 12227-010, Sao Jose dos Campos, SP, Brazil. M. BatistelIa, Embrapa Satellite Monitoring, Avenida Soldado Passarinho 303, Fazenda Chapadiio, CEP 13070-115, Campinas, SP, Brazil. (
[email protected]) E. F. Moran, Department of Anthropology and Anthropological Center for Training and Research on Global Environmental Change, Indiana University, Bloomington, IN 47405, USA C. Souza Jr., Instituto do Homem e Meio Ambiente da Amazonia (Imazon), Rua Domingos Marreiros 2020, CEP 66060-160, Belem, PA, Brazil. R. Walker, Department of Geography, Michigan State University, 234 Geography Building, East Lansing, MI 48823, USA S. 1. Walsh, Department of Geography, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA
The Changing Rates and Patterns of Deforestation and Land Use in Brazilian Amazonia Diogenes S. Alves, l Douglas C. Morton,2Mateus Batistella,3 Dar A. Roberts,4 and Carlos Souza Jr. 5 Investigating the rates and patterns of land cover and land use change (LCLUC) in Amazonia is a central issue for Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) research. LCLUC, along with climatic changes, affects the biological, chemical, and physical functions of Amazonia, thereby linking environmental change at the local, regional, and global scales. Considerable research has focused on estimating rates of forest conversion in Amazonia, mainly through the use of satellite remote sensing, and evaluating factors that influence these rates. Beyond the rates of forest loss, LCLUC research in Amazonia has also considered the variety of agricultural uses that replace forest cover, forest degradation from logging and fire, and secondary vegetation on previously cleared lands.
I 1. INTRODUCTION Investigating the rates and patterns of land cover and land use change (LCLUC) in Amazonia is a central issue for Large-Scale Biosphere-Atmosphere (LBA) Experiment in Amazonia research [Keller et al., 2004] (see the LBA Extended Science Plan at http://lba.cptec.inpe.br/lba/site/ ?p=plano_cientifico_estendido&t=1). LCLUC, along with climatic changes, affects the biological, chemical, and physical functions of Amazonia, thereby linking environmental change at the local, regional, and global scales [Keller et al.,
lrnstituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos, Brazil. 2Goddard Space Flight Center, Greenbelt, Maryland, USA. 3Embrapa Satellite Monitoring, Campinas, Brazil. 4Department of Geography, University of California, Santa Barbara,California, USA. 5rnstituto do Homem e Meio Ambiente da Amazonia, Belem, Brazil. Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 1O.1029/2008GM000722
2004] (LBA ExteMed Science Plan at http://lba.cptec.inpe. br/lba/sitel?p=plaho_cientifico_estendido&t=1). Considerable research has1focused on estimating rates of forest conversion in Amazbnia, mainly through the use of satellite remote sensing and evaluating factors that influence these rates [e.g., Tardin etal., 1980; Fearnside et al., 1990; Fearnside, 1990; Skole and Tucker, 1993; Alves, 2002; Margulis, 2004; Chambers et aI., 2007]. Beyond the rates of forest loss, LCLUC research in Amazonia has also considered the variety of agricultural uses that replace forest cover, forest degradation from logging and fire, and secondary vegetation on previously cleared lands. LCLUC in Brazilian Amazonia is highly heterogeneous, both spatially and temporally, as are the varieties of agricultural uses that replace forest cover [e.g., Becker, 1997; Machado, 1998; Faminow, 1998; Alves, 2002, 2007a; Morton et al., 2006]. To capture this heterogeneity, we develop a framework in which deforestation is one transition stage in a continuum of land use and land cover changes and their associated impacts on ecosystems and landscapes of Amazonia. We refer to the sequence of land cover changes, from mature forests to agricultural uses and abandonment, as a land use trajectory. Individual events within a trajectory are described as transitions. .The current Amazonian landscape is an integrated measure ofthe disturbance history from different development phases over the past 50 years. Numerous studies have provided 11
12
ALVES ET AL.
DEFORESTATION AND LAND USE IN BRAZILIAN AMAZONIA
a multifaceted history of the driving forces behind frontier expansion and deforestation in the region [e.g., Vetho, 1976; Sawyer, 1984; Schmink and Wood, 1992; Machado, 1998; Maitgulis, 2004], and we will not duplicate those efforts here. Historically, variations in forest clearing rates were generally linked with changes in access to the region; thus, road building and migration were critical precursors to forest losses. During the 1980s and 1990s, large-scale colonization projects, credit incentives, and steady investment in the region led to annlJal forest losses of 1-3 million ha [INPE, 2001, 2007] (Figure 1). More recently, economic forces within and beyond Amazonia have exerted stronger controls over deforestation rates and postclearing land uses, including domestic and global demand for beef, soybeans, and wood products [Faminow, 1998; Andersen et at., 2002; Margulis, 2004; Veiga et at., 2004; Morton et at., 2006]. LBA contributed to the development of remote sensing methods to map land cover and land use in Amazonia [Roberts et at., 2002; Hess et at., 2003; Lu et at., 2004; Anderson et at., 2005; Morton et at., 2006] and to greater understanding of the patterns and processes of deforestation and the overall dynamics of LCLUC through field, remote sensing, modeling, and related studies [e.g., Atves, 2002; Asner et at., 2005; Soares Filho et at., 2004; Atves, 2007a]. LCLUC plays a central role in many elements of LBA research, since the
sum of recent LCLUC defines the spatial patterns of land cover and the relative proportion ofmature forest, secondary growth, or degraded forest of varying structural characteristics, wetlands, natural and planted pastures, and croplands in any Amazonian landscape. In this chapter, we summarize recent LBA research focused on regional-scale patterns and processes of forest modification, conversion, postclearing land use, and the fate of deforested land over time. Rates and patterns of deforestation are influenced by a range of economic, social, and political factors, and where possible, we describe linkages between these controls and deforestation activity. We con~ centrate on the large-scale dynamics that link a variety of LCLUC processes; specific transitions such as logging, fire, and individual agricultural uses will also be addressed in more detail in the following chapters. We begin with a summary of deforestation mapping and monitoring approaches of both forest loss and postclearing land use. Basin-wide dynamics offorest loss, agricultural land uses, and rates ofland abandonment are subsequently described to discuss recent spatial and temporal trends in LCLUC. Finally, we examine the deforestation and postclearing land use of Mato Grosso state in greater detail to highlight rapid changes over the past decade in this region and the development of grain production capacity.
i
2. CHARACTERIZING SPATIAL AND TEMPORAL VARIATION~,OFLANDCOVER AND USE
,/ Mapping or mjlhitoring land cover changes in Amazonia is challenging.~he region is very large, rapidly changing, and often covered by clouds. The study of deforestation and subsequent LCLUC has therefore relied on satellite remote sensing and periodic agricultural censuses to construct the spatial and temporal variations in deforestation and postclearing land uses. Satellite remote sensing has also been an integral paIt of investigations of the forest modification by logging and fire that often precedes deforestation and to the characterization of the spatial and temporal distribution of anthropogenic fires in Amazonia. Given the vast geographic extent of inundated forests and other wetlands, tropical forest, and savannas (Plate 1), methodological approaches to map or monitor LCLUC often require trade-offs in spatial or temporal resolution [Chambers et at., 2007]. Thus, deforestation mapping based on highresolution satellite data can only be completed once per year, since cloud-free satellite coverage is most reliable during the dry season months [Fearnside, 1990; Asner, 2001]. Monitoring changes in land management across Amazonia may only be possible at 5- or lO-year intervals from agricultural census data, given the amount of effort required to survey
6O'W
70'W
farmers across the basin. The required spatial and temporal resolution of any application will therefore influence the choice of a specific satellite sensor or data product. Satellite remote sensing analyses have mapped the spatial extent of deforestation [e.g., Fearnside et at., 1990; Fearnside, 1990; INPE, 2001; Atves, 2007b], selective logging [e.g., Asner et at., 2005; Souza et at., 2005], secondary forest [e.g., Roberts et at., 2002; Atves et at., 2003], and land use following clearing [e.g., Moran and Brondizio, 1998; Morton et at., 2006; Lu et at., 2008]. Region-wide deforestation in Brazilian Amazonia has been mapped with satellite data since the 1970s [e.g., Tardin et at., 1980; Fearnside et at., 1990; Fearnside, 1990; Skote and Tucker, 1993; Shimabukuro et at., 2007; Atves, 2007b]. Annual Landsat-based surveys estimate total deforested area in Brazilian Amazonia to have reached nearly 70 million ha by 2005 [INPE,200l, 2007] (Plate 1). Information regarding the timing offorest clearing activities has emerged recently with the launch of new moderate resolution sensors SPOT-VEGETATION (1998,1.1 km) and Moderate Resolution Imaging Spectroradiometer (MODIS; 2000, 2003, 250 m to 1 km). Near-daily coverage from these instruments can be combined to provide cloud-free data at weekly to monthly intervals to map land cover change [e.g., Carreiras et at., t002; Souza et at., 2003; Anderson et at.,
lso•w
3.5
3
"
C to to
.c
2.5 O'
c
~
!
2
! c
1.5
0
i
~
J:!II
C
0.5
H11--_---t
H._-JlI--_-----.
l~lml~l~
BB (1)
I -__--III-_--I
H
I---.J
~~~~~~lmlm~~~~~oo~~~~ro~~2~
94 (2)
(3)
10'
.l11)l'k,lolIE911l!1 ...
.oeror.tlauon 10 200II .WolJandIOJldWalOt_
Figure I, Interannual deforestation rates in Amazonia from Instituto Nacional de Pesquisas Espaciais (INPE) deforestation surveys [INPE, 2001, 2007; Alves, 2007b]. (1) Average annual rates for the 1978-1988 period; estimates for 1978 were produced after partial reanalysis ofblack-and-white Landsat MSS 1:500,000-scale images and maps by Tardin et al. [1980] to address inconsistencies between this study and later INPE surveys [Fearnside et al., 1990]; (2) Average annual rates for the 1992-1994 period. Statistics for the 1987-2000 period are based on visual interpretation of color composites of Landsat TM red, near-infrared and mid-infrared channels at the 1:250,000-scale. Later statistics are derived from digital processing of Landsat TM images mapping forest clearings of6.25 and larger.
.cernldO UnobOllrvad Aroa
13
300
c::=::::J kin
Plate 1. Brazilian Amazonia showing the areas of closed forest (58%), Cerrado woodland-savannas (14%), wetlands and water bodies (9%), and deforestation to 2005 (13%). Unobserved areas correspond to 2005 cloud cover (6%) or areas outside the limits of Brazilian Amazonia. Data sources: closed forest, woodland-savannas, and deforestation [INPE, 2007]; wetlands and water bodies [Hess et al., 2003].
14 DEFORESTATION AND LAND USE IN BRAZILIAN AMAZONIA
2005; Morton et al., 2005; Lu et al., 2008]. Moderate resolution data are not ideal for quantifying fine-scale land cover changes; deforestation monitoring algorithms only consider forest losles larger than several moderate resolution pixels, or approximately 25 ha [Morton et al., 2005; Shimabukuro et al., 2007]. MODIS-based deforestation monitoring provided the first regional understanding of the timing of forest cl~aring activities [Anderson et al., 2005; Kay, 2005; Shimabukuro et al., 2007]. For recent deforestation in Mato Grosso state, clearing activity began in 93% of deforested areas prior to the onset of dry season conditions [Kay, 2005]. Clearings initiated in the wet season averaged three to five times the size of those areas cleared during the dry season, indicating that mechanized clearing may be less dependent on climate conditions than previously thought [Kay, 2005]. Data from DETER, an operational deforestation monitoring system developed by Instituto Nacional de Pesquisas Espaciais, Sao Paulo, Brazil [Shimabukuro et al., 2007], show
ALVES ET AL.
NW (10"IO'58.9"S, 62"38'12.6"W)
Jam (l0·25'58.4"S. 62"26'13.2"W)
1986
1986
2003
2003 .Prlmary Forest Secondary Forest • .Pasture
• •
G,een Pasture
•
Waler
Bum Sca, Rock/Savanna
•
ConstmeUonlurban
Plate 3. Contrasting patterns offorest fragmentation and land cover in two regions of Rondonia. Old secondmy forest appears to be abundant in an area of the northwestern part of the state. Within a region of old settlement near the town of Jam, deforestation eliminates most forest cover between 1986 and 2003. Each scene is 21 Ian across.
Plate 2. (a) Regions of predominance offorest clearings of various sizes during 1991-1997 [after Alves, 2002]. Categories represent the clearing size that accounted for 50% or more ofthe total cleared area in the period; area depicted comprises the y.,o cells that accumulated the first 95% in the Lorenz curve shown in Figure 3 [after Alves, 2002]. (b) Same as Plate 2a but for 2000-2005.
that deforestation timing differed markedly between 2005 and 2006. In 2005, forest clearing in Mato Grosso was almost equally split between September and April (47%) and May and August (53%), suggesting a strong wet season clearing component (November and April, 31%). In 2006, less than 20% of all clearing between was identified before May. Less wet season clearing in 2006 is consistent with reductions in large, mechanized clearing activities compared to othe:r; recent years. Agricultural census data are a rich archive of regional information on agricultural production, land management decisions, and related ecosystem and economic impacts. For 1970-1985, censuses were carried out every 5 years; after this period, a single survey was conducted (1995/1996), and a new census was underway in 2007. Census data are generally available at the municipal scale; h9wever, due to frequent subdivisions of large municipalities, establishing a consistent unit of analysis to track changes over the entire 1970-1996 period would require very large, heterogeneous units, which, in some cases, would include entire states (see, for example, http://www.ipeadata.gov.br). Also, method-
ological issues from changes in the categories of data or the period of data cql1ection, and difficulties related to the logistics of data correction in Amazonia further complicate comparisons betfeen censuses. Despite these drawbacks, agricultural ce~ses constitute the most complete survey of agriculturarproduction, including the area under different land use categories, crop production, and agricultural inputs, allowing for detailed analyses of social, economic, and environmental aspects of agriculture in Amazonia and comparison of the region with other parts ofBrazil. During LBA, research based on data from the agricultural censuses evaluated the suite of positive and negative effects of deforestation in Amazonia [Andersen et al., 2002] and the dynamics of land abandonment and land use intensification during this period [Alves, 2007a]. Methodological advances in fusing satellite and census data captured the corresponding spatial detail and management information in both data types for studies of land cover change and future landscape scenarios [Cardille and Foley, 2003; Morton et al., 2009] . 3. LAND COVER AND LAND USE CHANGE: PATTERNS AND TRAJECTORIES IN AMAZONIA The same LCLUC trajectory can result from different suites of transitions, depending on the type of initial forest disturbance and the number ofpreceding land uses (Figure 2). For example, the forest to pasture trajectory can occur directly, if mature forest is clear-cut to plant grasses, or in-
15
directly if pastures are established following logging or crop cultivation. The likely transition pathway from forest to other land uses depends on the stage of frontier occupation and the site conditions, such as distance to existing settlements and roads [Alves, 2002], soils and topography [e.g., Machado, 1998], land tenure, household assets, and market conditions for specific forest or agricultural products [Batistella et al., 2003; Batistella and Moran, 2005]. These factors influence the probabilities for individual transitions within this diagram at a variety of spatial scales; the spatial extent of logging and deforestation were nearly equivalent during 2000-2002, yet logged forest within 25 km of major roads had a higher probability of being deforested than unlogged forest [Asner et al., 2006]. Census data from 1970 to 1995 show several key trends in LCLUC trajectories during the expansion of the agricultural frontier. During this period, the majority of deforested land was converted to pasture for cattle ranching. The relative importance of temporary crops was relatively stable across much of Amazonia except Mato Grosso (Table 1). The contribution of Amazonian cattle to the total Brazilian herd increased from 8% to 23%, driven by both an increase in pasture area and a doubling of the average stocking rates per hectare. The long-term lrnd use trajectories in a given region may be linked to different land use processes and socioeconomic factors. For example, cycles of land abandonment can be linked to shiftingicultivation or land rotation in established
8
63
15
Pasture
la)
Figure 2. (a) Diagram showing the most common transitions among land cover/use classes studied during LBA. Initial forest disturbances occur through clear-cutting (solid), fire (dashed), and logging (dotted). Subsequent transitions among pasture, cropland, secondary forest, and degraded forest cover types show the diversity of pathways that are possible for any land use trajectory. (b) Transition probabilities for postclearing land uses in Mato Grosso state during 2000-2005 for forest, secondary forest, and cerradlio woodland clearings >25 ha. Small deforestation events «25 ha, not shown) in Mato Grosso account for 15% of all deforestation. Percentages refer to the fraction of cleared area converted to specific land uses [Morton et al., 2006, 2007a].
ALVES ET AL.
16 DEFORESTAnON AND LAND USE IN BRAZILIAN AMAZONIA
17
i
Table 1. Evolution ofAggregate Land Use Statistics According to Brazilian Agricultural Census· 1970
1975
1980
1985
1995
Land Category, % Total Farm Area
Pasture Crops, temporary Crops, permanent Forest Abandoned landb All otherC
37.9 2.6 0.3 37.3 15.5 6.4
35.5 3.3 0.4 43.0 13.1 4.7
35.2 4.1 0.7 42.4 9.8 7.8
36.8 4.4 0.8 40.4 8.9 8.7
42.4 3.9 0.8 41.3 5.7 5.9
Temporary Crops in MT and in All Other States, % Total Farm Area
Mato Grosso state All other states
1.4 3.0
2.1 3.8
4.1 4.0
5.3 4.0
5.6 2.8
Cattle Head in Amazonia as a Fraction a/National Herd, %
8.2
9.0
12.7
14.7
23.3
Average Stocking Rate, headlha
0.30
0.30
0.40
0.40
0.70
·Aggregate data for the nine states belonging to Legal Amazonia: Acre, Amapa, Amazonas, Maranhao, Mato Grosso, Para, Rondonia, Roraima, and Tocantins. Source: http:// www.ipeadata.gov.br. bAbandoned land is defined as land unused for more than 4 years. cAll other includes land in rotation, planted forest, and other categories like swamps.
farms; cropland can rotate with pasture when grain prices are low. Although not shown in Figure 2, agroforestry, reforestation, urban expansion, and other types ofland use can also replace pastures or croplands. In addition, some land use trajectories can be influenced by a combination of factors, such as forest degradation from selective logging and fire [Nepstad et al., 1999], which fundamentally alter forest structure and land value. Particular LCLUC transitions generate unique patterns of forest loss. Different agrarian regimes, inc~uding farm size, the architecture of settlement projects, and different production and land management strategies can lead to diverse expressions of the same trajectory in landscape patterns. The composition and configuration of the landscapes produced have important consequences for the functioning of the biophysical systems in Amazonia and may help inform discussions of plausible development scenarios for the region. Within many older agricultural frontiers, concentrated deforestation activity in the vicinity of major roads and colonization projects [Machado, 1998; Alves, 2002] has led to landscapes dominated by pastures and cropland. The magnitude of forest clearing for agriculture in these areas often exceeds the limits prescribed by the Brazilian Forest Code [Alves et al., 2003; Alvez, 2007b].
The following sections review advances in understanding the evolution oflandscape patterns and the dominant longterm LCLUC trajectories in Amazonia.
eastern Para states and in smaller clearings in regions with higher densities of settlement projects in Para and Rondonia. Overall, large e-(earings on larger farms contributed the greatest fraction)bf total deforestation (Figure 3) [Alves, 2002]. During ~00-2005, the patterns in deforestation size show a bimodal distribution, with regions either dominated by very large (> 1000 ha, 25% of cells) or very small «50 ha, 51 % of cells) clearing sizes. Very large clearings in central Para, southern Amazonas, and central Roraima states suggest that these regions were recently exposed to the same degree of capital and technology that was previously found only in older frontier areas. Landscape patterns of forest conversion at the local scale reflect additional heterogeneity beyond clearing size (Plate 3). In 1986, central Rondonia near Jaru was already highly disturbed, consisting of nearly equal proportions of primary forest and pasture, with pasture concentrated along planned roads at 4-km intervals. By 2003, linear strips of forest from 1986 had been reduced to small forest fragments, many of which were less than 1 km across. Small patches of secondary forest mapped in 2003 occur exclusively along the margins offorest fragments that have never b~en cleared, suggesting that forest edges are taking on the spectral signature of secondary forest in the absence of any clearing. Patterns in a region of the nearby municipio of Ariquemes differ markedly with extensive tracts ofmature forest in both 1986 and 2003, no "fishbone" pattern from evenly spaced roads, and some large patches of secondary forest as much as 18 years old. Differences in fragmentation patterns reflect differences in the architecture of settlement and colonization
projects, whereas the persistence of secondary forest in the northwest is likely due to higher rainfall and poorer soils in this region. 3.2. Forest Degradation From Logging Selective logging is one of the most important drivers of forest degradation and land cover change in Amazonia. Logging is rarely practiced in a sustainable fashion. In fact, only 1248 ha of mature forests were harvested following the Forest Stewardship Council (FSC) standards in Amazonia in 2003 [Lentini et al., 2005]. The extensive network of secondary roads built by loggers and capital obtained by land owners selling timber help to accelerate the deforestation process near sawmill centers [Uhl et al., 1991; Verissimo et al., 1992]. Unmanaged logging practices lead to forest degradation through damage to forest structure and altered species composition [see Asner et al., this volume]. Using remote sensing techniques, Asner et al. [2005] estimated that the annual area affected by logging was 12,000-19,000 km2 between 2000 and 2002, equivalent to the average annual deforestation rate during this period of 18,000 ± 2900 km [INPE, 2007]. Logging and deforestation are not mutually exclusive; an average of 16% oflogged forests ¥e clear-cut in the first year following logging operations, with 33% deforested within 4 years of logging [Asner et al;, 2006]. Canopy damage and slash from logging operatiods increase the likelihood of fire damage in logged forests [Nepstad et al., 1999], although the extent of logged and burned forest has not been estimated for the entire Amazon region.
3.1. Landscape Patterns ofForest Conversion Deforestation in Amazonia has replaced the forest with a fragmented landscape ofpasture and agricultural areas, leaving few forest remnants where deforestation has been most concentrated. The total extent of deforestation in Amazonia until 2005, depicted in Plate 1, provides a first approximation of important regional patterns in forest loss. Major road and river networks are buffered by the outlines of historic deforestation and older frontier areas of eastern Para, Mato Grosso, and Rondonia states have greater forest loss than newer frontiers in central Para, Acre, or Amazonas states. Specific site conditions, including soil quality or topography, further influence both the location of forest clearing and the postclearing land uses, such that patterns of deforestation and land use may be locally consistent. The spatial patterns resulting from forest conversion may differ substantially across the basin as a function of clearing size (Plate 2). Deforestation between 1991 and 1997 occurred in very large clearings in central Mato Grosso and
,-.. ~
3.3. Forest Conversion to Pasture
100
= '-" ~
e
~
75
~
e
I-
~
.... .........
50
•••••••••••••••
~ ~
Q
l:l
~OJ
e
""
0 0
25 50 75 Number of cells (%)
100
!-1991-97mte • area
200ha!
Figure 3. Lorenz curve of the 1991-1997 deforestation rate calculated for \1..0 cells and accompanying cumulative curves of the area of forest clearings of two different sizes in the same period [after Alves, 2002].
According to census data, pasture has been the most common land use in Amazonia (Table 1). Typical processes of pasture establishment in Amazonia include the direct conversion of forest to pasture or a longer conversion trajectory beginning with an initial phase of annual crops before pasture establishment after a number ofyears [e.g., Millikan, 1992]. After establishment, pasture productivity typically remains high for 5 to 7 years before declining due to changes in soil fertility and pH, resulting in a progressive decrease in forage quality and increased weed invasion [Buschbacher, 1986; Serrao and Toledo, 1990]. As pasture quality degrades, pastures can either be reinvigorated through repeated cycles ofburning, reformed via fertilizer application and reseeding, or abandoned to secondary succession. The length oftime a pasture remains productive is highly dependent on pasture management practices, local climate, and soil quality [Serrao and Toledo, 1990;
18
DEFORESTATION AND LAND USE IN BRAZILIAN AMAZONIA
Moran, 1993; Dias Filho et al., 2000; Numata et al., 2007]. For example, pastures established on Alfisols or Ultiso1s in Rondonia that receive moderate levels of precipitation can remain prdtluctive for well over 20 years, whereas pastures established on Oxisols or in more humid or arid conditions show earlier evidence of degradation and higher rates of abandonment [Numata et ai., 2007]. Cattle ranching remains the dominant land use in Amazonia (Table 1), following important changes during the last decades. Faminow [1998] argues that a fundamental cause for the growth of the cattle herd was the considerable expansion of regional demand associated with urban growth. Andersen et ai. [2002] and Margulis [2004] reviewed the many motivations for cattle ranching and intensification of pasture use, concluding that ranching became profitable independent of subsidies due to the growth of urban demand and increased productivity. Veiga et al. [2004] observed a variety of market chains stimulated by local demands and markets outside Amazonia. Higher stocking rates are more commonly found in the most deforested areas, suggesting a transition to pasture use intensification [Alves, 2007a]. Taken together, these factors help to explain the continued predominance of pastures in land use trajectories in Amazonian landscapes. 3.4. Forest Conversion to Cropland In the context of LCLUC, we classify forest conversion to cropland according to the most common land use trajectories in recent decades. Cropland may directly follow deforestation or arise as part of a rotation cycle with secondary forest or pasture.' Direct conversion of forest to cropland occurs for both small-scale [e.g., Moran and Brondizio, 1998] and large-scale crop production [Morton et al., 2006]. In addition to subsistence crops, small farmers may also invest in other crops for local or national markets [Moran and Brondizio, 1998; Costa, 2007]. Forest conversion for soybean, maize, or other grain production follows the recent development of crop varieties specifically adapted to the soils and climate of some Amazon regions [Warnken, 1999; Jepson, 2006]. The dynamics of forest conversion for mechanized crop production in Mato Grosso is discussed in more detail in section 4.1. The nature ofrotation systems between cropland and forest or pasture depend on both farm size and market conditions. For small farms, crop areas may be used until soil nutrients are depleted and then abandoned for several years to allow forest vegetation to accumulate nutrients. The length of fallow rotations in a "slash-and-burn" or "chop-and-mulch" system depends on the rate of forest recovery and farm size [Denich et al., 2004]. On larger farms, market conditions
for beef or grains may determine the interannual patterns of pasture and cropland use or the frequency of fallow cycles. Cropland can be both a precursor to land consolidation for cattle ranching or an endpoint itself, independent offarm size. Census data suggest that in Amazonia, croplands established in the original phases of colonization were largely replaced by cattle ranching as more forest was converted [Alves, 2007a]. However, recent expansion of mechanized crop production was generated through new deforestation, savanna clearing, and transitions from pasture to cropland [Morton et al., 2006]. The diversity of transition pathways, crop types, and farm sizes in Amazonia highlights the spatial and temporal variability of cropland on the landscape. Deforestation dynamics in Mato Grosso state represents one case ofparticular interest because of specific sociodemographic, economic, and bioclimatic conditions, suggesting the establishment of a new land use system differing from those dominating in other parts of Amazonia. Mato Grosso had the highest deforestation rate during 1995-2005, accounting for 33-43% of the annual deforestation increment in the Brazilian Amazon [lNPE, 2007]. High rates of forest loss were driven by large clearing sizes [Alves, 2002; Morton et al., 2006; Ferreira et ai., 2007] (Plate 2 and Figure 3); large landholders (2:1000 ha) owned an estimated 84% and 82% of all land in private property statewide according to the 1985 and 1996 agricultural censuses, respectively [lBGE, 1996]. Although deforestation is associated with a variety of influences, economic factors have been largely linked with credit and economic opportunities for extensive cattle ranching operations and crop production such as soybeans, and with inter-regional differences in land prices [Fearnside, 2001; Andersen et al., 2002; Margulis, 2004; Morton et ai., 2006]. Deforestation in Mato Grosso is highly mechanized in comparison with other states. Two tractors, linked by a strong chain, are used to pull down trees in the transitional forests. Even in taller-stature forests, heavy machinery is used to manage manually felled trees. Piling and re-burning fore~t vegetation can reduce standing forest to bare soil in a matter of months. Unlike previous estimates of carbon losses from deforestation, where 20% of biomass is combusted, and the remainder decomposes over 10-30 years [Fearnside et al., 1993; Houghton et ai., 2000], mechanized forest clearing practices may result in near-complete combustion of aboveground woody biomass and woody roots [Morton et al., 2008]. Mechanization has the~eby increased the potential size of forest clearings and decreased the duration of the deforestation process. Combined advances in deforestation mapping and tracking the fate of cleared land provide spatial and temporal details regarding land cover transitions statewide. Vegetation
ALVES ET AL.
phenology, derived froriI time series of MODIS data, has proven useful for separating land cover types and following changes in land mauagement over time [Ratana et al., 2005; Morton et ai., 20.d6; Brown et al., 2007]. Figure 2b highlights the dyna~cs of 2000-2005 transitions among major cover types in'Mato Grosso state, showing the proportion of new deforestation, woodland savanna, and secondary forest conversions >25"ha as a function of postclearing land use. The main driver of forest loss in Mato Grosso is large-scale cattle production, yet direct conversion of forest to cropland contributed substantially to the number of large deforestation events and to woodland and secondary forest losses during this period [Morton et al., 2006, 2007a, 2007b]. Secondary forest is not a large component of the landscape in Mato Grosso compared with estimates for other regions, comprising only 11-14% of historic deforestation [Carreiras et ai., 2006; Morton et al., 2007a]. Detailed analysis of the source of secondary and degraded forests in Mato Grosso from abandonment, logging, and burning remains a research challenge. Expansion of soybeans and other mechanized crop varieties in Amazonia has renewed the debate over extensive versus intensive land uses, and about the social and environmental outcomes of agricultural expansion. Climate, soils, and topography are suitable for soybean cultivation in forested regions of northern Mato Grosso and surrounding areas [Jasinski et ai., 2005], and some authors have argued that soybean cultivation can be a competitive, intensive agriculture alternative over extensive and low-productive cattle ranching [e.g., Andersen et al., 2002; Margulis, 2004]. However, soybean production can contribute to pushing cattle ranching into new deforestation frontiers, as seen following its introduction in southern and west-central Brazil [Andersen et al., 2002], even if a detailed assessment of the role of soybeans in concentration of land tenure and income, rural outmigration, and loss of biodiversity has not yet been completed [Fearnside, 2001]. 3.5. Land Abandonment and Secondary Vegetation Growth
Considerable research has focused on mapping secondary forest at local and regional scales [Lucas et al., 1993; Moran et ai., 1994; Roberts et ai., 2002; Alves et al., 2003; Carreiras et ai., 2006]. Secondary forests are a potential carbon sink and can help recover hydrological and biogeochemical functioning after forest clearing [e.g., Brown and Lugo, 1990; Moran et al., 1994]. Secondary succession can develop following different pathways, including land rota~ tion during shifting cultivation and land abandonment after pasture degradation or immediately following forest clearing; species composition, vegetation structure, and rates of carbon uptake in secondary forests are highly dependent
19
upon soil type and prior land use [Alves et al., 1997; Moran et al., 2000; Lucas et ai., 2002; Zarin et al., 2005]. Census data and remote sensing analyses raise important questions about the long-term dynamics of secondary vegetation in Amazonia. The proportion of cleared land that was unused for more than 4 years as a percentage of farm area declined steadily, from 15.5% to 5.7%, during 1970-1995 (Table 1). This evidence is consistent with findings that rates of land abandonment were higher in newly established frontiers, while secondary vegetation tended to be re-cleared concurrently with the elimination of mature forest remnants in older settlement areas [Alves et al., 2003; Alves, 2007a]. Time series of satellite data show that secondary forest is a dynamic component of the landscape in the Ariquemes and Ii-Parana regions of Rondonia (Figure 4). In both regions, steady increases in pasture area resulted from more rapid re-clearing of secondary forest than pasture abandonment. Overall, the contribution of secondary forest remained stable or declined during 1986-2003, never exceeding 10% of the landscape. Declining rates of land abandonment in more intensively deforested areas indicate that over the long term, secondary forests may offset only a small fraction of the initial carbon emissions from deforestation [Alves et ai., 1997; Alves,2007a]. 4. cmJcLusIONS AND OUTLOOK ,
Brazilian Amaionia is one of most active regions of agricultural expansipn in the world. Clearing tropical forest is the primary means to increase the area of cattle pasture and crops, while related processes such as logging, fire for land clearing and management, land abandonment, and land use intensification are also key elements of the LCLUC dynamics. The conceptual model of transitions between multiple land cover and use states' illustrates the heterogeneity of LCLUC trajectories and their expression in landscape patterns across Amazonia. Characterizing the spatial patterns created by such processes represents an important methodological success in Amazonia, based on multiple data sources and a variety of analysis techniques, from which to investigate the role of LCLUC on the biophysical system. Agriculture in the region is becoming increasingly intensive, conducted by large-scale operators with sufficient access to capital. These shifts in the spatial and temporal dynamics of LCLUC are present in both census data and satellite remote sensing as a decrease in secondary forests, increase in pasture stocking rates, and rapid expansion of the area under mechanized agriculture. The rise of intensive production and the influence of both national and international market forces on land use have led to the development of new ecologically oriented certification schemes for beef
20
DEFORESTATION AND LAND USE IN BRAZILIAN AMAZONIA
A) Ji-Parana
Pasture
ALVES ET AL.
B) Ariquemes
311
40 315
30
30
211
25
20
20
111
115 10
10
1:
II
'~ofl1t+~HHft+t1~~Hilff "
6.
5
II
10
20
.\
101.---------------1
.
C} JI-Parana
D Ari uemes
~...;;.;;..;;;..;;;;;.=~------.....,
10
2Or-------.::;;..L.;=-=;;.;;L;=~
..
1:
l:
'"
..'"
'"~
in Brazilian Amazonia provides detailed estimates of forest loss on an annual basis. Advancement in near real-time monitoring of defor~tationin cerrado and closed forest and mapping selective,t10gging has generated essential data for environmental ~nitoring. Successes in remote sensing of deforestation in Amazonia serve as an important example of technical progress for other nations considering programs to reduce deforestation. Future research will continue to focus on the economic, social, and environmental elements of each forest loss trajectory, highlighting spatial and temporal heterogeneity in the causes and consequences of Amazon deforestation. Recent advances in remote sensing pave the way for additional efforts to quantify the basin-wide impacts offorest degradation from fire, forest fragmentation, and land abandonment to secondary forest. Findings from LBA also lay the groundwork for related research on the influence of specific land use transitions and spatial patterns of land cover for climate, biogeochemistry, and long-term agricultural productivity, as reported in the following chapters of this book.
10
REFERENCES
~
6.
0
r.
0
10
10
Secondary Forest Figure 4. Transitions among pasture (gray), secondary forest (dashed), and primary forest (black) for the Ii-Parana and Ariquemes regions of Rondonia state during 1986-2003. Dynamics for pasture in (a) Ji-Parana and (b) Ariquemes. Values above the x axis represent a gain of a specific class as a percentage of the landscape, and values below the axis represent a loss. Changes in secondary forest 'over time in a similar manner for (c) Ii-Parana and (d) Ariquemes. In Ji-Parana, pasture shows a general increase over time, with most pasture originating from areas that were previously pasture. Pasture loss is primarily to secondary forest. Secondary forest shows no significant increase over time in Ji-Parana, leading to a declining ratio of secondary forest to cleared lands. Large fluctuations between pasture and secondary forest in Ii-Parana during 1997-1999 are most likely due to early dry season imagery in these years leading to overestimating secondary forest. Rates of pasture abandonment to secondary forest were more stable in Ariquemes than in Ji-Parana. Both pasture and secondary forest show a general increase over time, resulting in a ratio of secondary forest to pasture of over 30% in Ariquemes.
and grain production in Amazonia. At the same time, high deforestation rates in older settlement areas, expansion of agricultural frontiers into new areas, and prevailingly low productivity ofland show the recurrence of historical trends. Thus, a diversity of actors remain influential in both "old"
and "new" frontiers presenting a challenge for delineating plausible future scenarios ofLCLUC in Amazonia. Advances in satellite remote sensing of deforestation and postclearing land use have led to high-quality data for both science and policy applications. Deforestation mapping
Alves, D. S. (2002), Space-time dynamics of deforestation in Brazilian Amazonia, Int. J. Remote Sens., 23, 2903-2908. Alves, D. S. (2007a), Cenarios de cobertura e uso da terra e dimensoes humanas no LBA, in Dimensoes Humanas da BiosferaAtmosfera daAmazonia, edited by W. M. da Costa, B. K. Becker, and D. S. Alves, pp. 39-63, EDUSP, Sao Paulo, Brazil. Alves, D. S. (2007b), Science and technology and sustainable development in Brazilian Amazon, in The Stability of Tropical Rainforest Margins, edited by T. Tscharntke et aI., pp. 493-512, Springer, Berlin, Germany. Alves, D. S., J. V. Soares, S. Amaral, E. M. K. Mello, S. A. S. Almeida, O. F. da Silva, and A. M. Silveira (1997), Biomass of primary and secondary vegetation in Rondonia, Western Brazilian Amazon, Global Change BioI., 3, 451--461. Alves, D. S., M. 1. S. Escada, J. L. G. Pereira, and C. A. Linhares (2003), Land use intensification and abandonment in Rondonia, Brazilian Amazonia, Int. J. Remote Sens., 24,899-903. Andersen, L. E., C. W. J. Granger, E. J. Reis, D. Weinhold, and S. Wunder (2002), The Economics of Deforestation: Dynamic Modeling ofAmazonia, Cambridge Univ. Press, Cambridge. Anderson, L. 0., Y. E. Shimabukuro, R. S. DeFries, and D. C. Morton (2005), Assessment of deforestation in near real time over the Brazilian Amazon using multitemporal fraction images derived from Terra MODIS, IEEE Geosci. Remote Sens. Lett., 2, 315-318. Asner, G.P. (2001), Cloud cover in Landsat observations of the Brazilian Amazon, Int. J. Remote Sens., 22, 3855-3862. Asner, G. P., D. E. Knapp, E. N. Broadbent, P. J. C. Oliveira, M. Keller, and J. N. Silva (2005), Selective logging in the Brazilian Amazon,Science,310,480--482.
21
Asner, G. P., E. N. Broadbent, P. J. C. Oliveira, M. Keller, D. E. Knapp, and J. N. M. Silva (2006), Condition and fate of logged forests in the Brazilian Amazon, Proc. Natl. Acad. Sci. U. S. A., 103,12,947-12,950, doi:1O.1073/pnas.0604093103. Asner, G. P., M. Keller, M. Lentini, F. Merry, and C. Souza Jr. (2009), Selective logging and its relation to deforestation, Geophys. Monogr. Ser., doi:lO.l029/2008GM000723, this volume. Batistella, M., and E. F. Moran (2005), Dimensoes humanas do uso e cobertura das terras na Amazonia: Uma contribuifYao do LBA, Acta Amazonica, 35,239-247. Batistella, M., S. Robeson, and E. F. Moran (2003), Settlement design, forest fragmentation, and landscape change in Rondonia, Amazonia, Photogramm. Eng. Remote Sens., 69, 805-812. Becker, B. K. (1997), Amazonia, 5th ed., ATICA, Sao Paulo. Brown, J. C., W. E. Jepson, J. H. Kastens, B. D. Wardlow, J. M. Lomas, and K. P. Price (2007), Multitemporal, moderate-spatial resolution remote sensing of modem agricultural production and land modification in the Brazilian Amazon, GIScience Remote Sens., 44,117-148. Brown, S., and A. Lugo (1990), Tropical secondary forests, J. Tropical Ecol., 6, 1-32. Buschbacher, R. (1986), Tropical deforestation and pasture development, Bioscience, 36,22-28. Cardille, J. A., and J. A. Foley (2003), Agricultural land-use change in Brazilian Amazonia between 1980 and 1995: Evidence from integrated satellite and census data, Remote Sens. Environ., 87, 551-562. Carreiras, J. M. B., y(. E. Shimabukuro, and J. M. C. Pereira (2002), Fraction images derived from SPOT-4 VEGETATION data to assess land-cover Ichange over the State ofMato Grosso, Brazil, Int. J. Remote Sen's., 23, 4979--4983. Carreiras, J. M. B., J. M. C. Pereira, M. L. Campagnolo, and Y. E. Shimabukuro (2006), Assessing the extent of agriculture/pasture and secondary succession forest in the Brazilian Legal Amazon using SPOT VEGETATION data, Remote Sens. Environ., 101, 283-298. Chambers, J. Q., G. P. Asner, D. C. Morton, L. O. Anderson, S. S. Saatchi, F. d. B. Espirito-'Santo, M. Palace, and C. Souza Jr. (2007), Regional ecosystem structure and function: Ecological insights from remote sensing of tropical forests, Trends Ecol. Evoi,22,414--423. Costa, W. M. (2007), Tendencias recentes na Amazonia: Os sistemas produtivos emergentes, in Dimensoes Humanas da Biosfera-Atmosfera da Amazonia, edited by W. M. da Costa, B. K. Becker, and D. S. Alves, pp. 81-11, EDUSP, Sao Paulo, Brazil. Denich, M., K. Vielhauer, M. S. de A. Kato, A. Block, O. R. Kato, T. D. de Abreu Sa, W. Lucke, and P. L. G. Vlek (2004), Mechanized land preparation in forest-based fallow systems: The experience from eastern Amazonia, Agroforestry Syst., 61, 91-106. Dias Filho, M. B., E. A. Davidson, and C. J. R. de Carvalho (2000), Linking biogeochemical cycles to cattle pasture management and sustainability in the Amazon Basin, in The Biogeochemistry ?fthe Amazon Basin, edited by M. McClain, R. L. Victoria, and J. E. Ritchey, pp. 85-105, Oxford Univ. Press, New York. Faminow, M. D. (1998), Cattle Deforestation and Development in the Amazon, CAB International, New York.
ALYES ET AL. 22
23
DEFORESTATION AND LAND USE IN BRAZILIAN AMAZONIA
Honzak G. M. Foody, P. 1. Curran, and C. Cor) , . Lucas, R . M 0" M , (1990) , The rate and extent of deforestation in F earnsl'd e, P .M ves (1993), Characterising tropical secondary forests usmg . Brazilian Amazonia, Environ. Conserv., 17, 213-226. multi-temporal Landsat sensor imagery, Int. J. Remote Sens., 14, Feaffii~de, P. M. (2001), Soybean cultivation as a threat to the envI3061-3067. ronment in Brazil, Environ. Conserv., 28, 23-38. Lucas, R. M., M. Honzak, I. do Amaral, P. J. Curran, and G. Jy,1. 'd P MAT Tardin and L. G. Meira Filho .(1990), F earnsl e, . ., . . , Foody (2002), Forest regeneration on abandoned clearances m Deforestation rate in Brazilian Amazonia, Reprint, InstItuto de Central Amazonia, Int. 1. Remote Sens., 23, 965-9~8 .. Pesquisas Espaciais, Sao Jose dos Campos, Brazil. . Machado, L. (1998), A fronteira agricola na ~mazoma, m G~ogra 'd P M N Leal Jr ., and F. M. Fernandes (1993), Ramfor' . F emusl e, . . , . fia e Meio Ambiente no Brasil, 2nd ed., edited b~ A. Chnstofoest burning and the global carbon budget: Biomass, combustIOn letti et aI., pp 181-217, Hucitec, Sao Paulo, BraZIl. efficiency, and charcoal' formation in the Brazilian Amazon, J. Deforestation in Brazilian Ama· S (2004) , Causes or 'J. d M arguIIS,. Geophys. Res., 98(D9), 16,733-16,743... zon, World Bank, Washington. (Available at http://www-w s. . N C L G Ferreira and F. MlzIara (2007), Deforesta, F errelra, . ., . , worldbank.org). . tion hotspots in the Brazilian Amazon: Evidence and causes as Millikan, B. H. (1992), Tropical deforestation, land degradatIOn, assessed from remote sensing and census data, Earth Interact., and society, lessons from Rondonia, Brazil, Lat. Am. Perspect., 11(1),1-16, doi:l0.1l75/EI201.1. 19(1),45-72. . '1' Hess, L. L., 1. M. Melack, E. M. L. M. N~vo, C. C. F. B.arbosa, oran, E .F, ( 1993) , Deforestation and land use m the Brazllan M and M. Gastil (2003), Dual-season mappmg of wetland mundaAmazon, Human Ecol., 21,1-21. tion and vegetation for the central Amazon basin, Remote Sens. and E. S. Brondizio (1998), Land-use change after M oran, E . F ., . . ki R Environ., 87, 404-428. deforestation in Amazonia, in People and PIxels: Lm ng eSkole C. A. Nobre, J. L. Hackler, K. T. D L H oughton,R"A' " mote Sensing and Social Science, edited by D. Liverman et aI., , Lawrence, and W. H. Chomentowski (2000), Annua~ ~luxes of 94-120 National Academy Press, Washington, D. C. carbon from deforestation and regrowth in the BraZIlian Ama- M~~~n, E. F.: E. Brondiz~o, P. Mausel, and Y. Wu ~1994), In~e . zon, Nature, 403, 301-304. grating Amazonian vegetation, land-use and satellite data, BwINPE (2007), Projeto PRODES. (Available at http://www.obt.mpe. Science, 44, 329-338. . br/prodes). Moran, E. F., E. S. Brondizio, J. M. Tucker, M. C. da S~lva-F.o:sberg, Instituto Brasileiro de Geografia e Estatistica (mGE) ~1996), Dados S. McCracken, and I. Falesi (2000), Effects of SOli fertility and do Censo Agropecuario. (Available at http://www.lbge.gov.~r). land-use on forest succession in Amazonia, For. Ecol. Manage., Instituto Nacional de Pesquisas Espaciais (INPE) (2001), MOllltor139,93-108. ing of the Brazilian Amazonian For~s~ by ~atelli~e, Reprint, In- Morton, D. C., R. S. DeFries, Y. E. Shimabukuro, L. O. Anderstituto Nacional de Pesquisas EspacIals, Sao Jose dos Campos, son, F. d. B. Espirito-Santo, M. C. Hansen,. an~ M. Carr~ll Brazil. . (2005) Rapid assessment of annual deforestatIOn m the Brazil. ki E W D C Morton , R. S. DeFries, Y. E. Shimabukuro, ian ~azon using MODIS data, Earth Interact., 9(8), E1139, J asms , . ., . , L. O. Anderson, and M. C. Hansen (2005), Physicalla~dscape doi: 1O.l175/E1139.1. correlates of the expansion of mechanized agricultur~ III Mato Morton, D. C., R. S. DeFries, Y. E. Shimab~kuro, L. O. A~der Grosso, Brazil, Earth Interact., 9(16), EI143 , dOl: 10.1175/ son, E. Arai, F. d. B. Espirito-Santo, R. Freitas, ~nd J. Mor~se~e
(2006), Cropland expansion changes deforestatIOn d~namlcs m the southern Brazilian Amazon, Proc. Nat!. Acad. SCI. U. S. A., 103, 14,637-14,641, doi:lO.1073/pnas.0~06377103.. 289-316. '1 Morton, D. C., Y. E. Shimabukuro, R. Freitas, E. Aral, an~ R. S. Kay, K. (2005), Estimating wet-season 'deforestation in the BrazlDeFries (2007a), Secondary forest dynamics and Cerradao ~oss ian Amazon using MOD13 250m data, Geography, M. S., 25, in Mato Grosso during 2001-2005 from MODIS phenology tIm.e Univ. of Maryland, College Park. series, paper presented at XIII Simp. Bras. Sens. ~emoto, Flo~l Keller, M., M. A. Silva-Dias, D. C. Nepstad, and M. O..Andre~e anopolis, Sta Catar, Brazil, 21-26 Apr 2007. (Available at http.!/ (2004), The large-scale biosphere-atmosphere expen~ent III www.dsr.inpe.br/sbsr2007Ibibliotecal). . Amazonia: Analyzing regional land use change e~fects, III Eco- Morton, D. c., Y. E. Shimabukuro, B. F. T. Rudorff, A. Lm~a, R. systems and Land Use Change, edited by R. deFnes, G. Asner, Freitas, and R. S. DeFries (2007b), Challenge for conservatIOn at and R. Houghton, pp. 321-334, AGU, Boston, Mass. . the agricultural frontier: Deforestation, fire, and land use dynam· . M A Verissimo , and L. Sobral (2005), Forest Facts m L ent 1m, ., . . . ics in Mato Grosso, Agua Ambiente, 2,5-20. the Brazilian Amazon 2003, Imazon, Belem. . Morton, D. C., R. S. DeFries, J. T. R!\nderson~ L. Glg~lO, Lu, D., P. Mausel, M. Batistella, and E. F. M~ran (2004): ~ompanSchroeder, and G. R. van der Werf (2008), Agrlcultur~l mtenslson of land-cover classification methods III the BraZIlian Amafication increases deforestation fire activity in Amazoma, Global zon Basin, Photogramm. Eng. Remote Sens., 70, 72~-731. . Change Bioi., 14, 2262-2275. Lu, D., M. Batistella, and E. F. Moran (2008), .IntegratlOn ofLa~d Morton, D. C., R. S. DeFries, and Y. E. Shimabukuro (2009), Cropsat TM and SPOT HRG images for vegetation change detectIOn land expansion in cerrado and transition forest ~cosystems: in the Brazilian Amazon, Photogramm. Eng. Remote Sens., 73, Quantifying habitat loss from satellite-based vegetation phenol-
EI143.1. . . W (2 006) , Producing a modern agricultural frontier: FIrmS 2 J epson,. and cooperatives in Eastern Mato Grosso, Econ. Geogr., 8 ,
w,.
421-430.
ogy, in Cerrado Land-Use and Conservation: Assessing Tradeoffs Between Human and Ecological Needs, edited by C. Klink, R. S. DeFries, and ~;(;avalcanti, Conservation Int., Washington, D. C., in press. F Nepstad, D. C., et <Jj. (1999), Large-scale impoverishment of Amazonian forestsi{y logging and fire, Nature, 398, 505-508. Numata, I., O. A. Chadwick, D. A. Roberts., 1. P. Schimel, F. F. Sampaio, F. C. ,Leonidas, and 1. V. Soares (2007), Temporal function of soil order, pastIlre age, and management, Rondonia, Brazil,Agric. Ecosyst. Environ., 118, 159-172. Ratana, P., A. R. Huete, and L. G. Ferreira (2005), Analysis of CelTado physiognomies and conversion in the MODIS seasonaltemporal domain, Earth Interact., 9, E1119, doi:l0.1175110873562(2005)009<0001 :AOCPAC>2.0.CO;2. Roberts, D. A., I. Numata, K. Holmes, G. Batista, T. Krug, A. Monteiro, B. Powell, and O. A. Chadwick (2002), Large area mapping ofland-cover change in Rondonia using multitemporal spectral mixture analysis and decision tree classifiers, 1. Geophys. Res., 107(020), 8073, doi:l0.1029/2001JD000374. Sawyer, D. (1984), Frontier expansion and retraction in Brazil, in Frontier Expansion in Amazonia, edited by M. Schmink, and C. H. Wood, pp. 180-203, Univ. of Florida Press, Gainesville. Schmink, M., and C. H. Wood (1992), Contested Frontiers in Amazonia, Columbia Univ. Press, New York. Serrao, E. A. S., and 1. M. Toledo (1990), The search "for sustainability in Amazonian pastures, in Alternatives to Deforestation: Steps Toward Sustainable Use ofthe Amazon Rain Forest, edited by A. B. Anderson, pp. 195-214, Columbia Univ. Press, New York. Shimabukuro, Y. E., V. Duarte, M. A. Moreira, E. Arai, D. M. Valeriano, L. O. Anderson, and F. d. B. Espirito-Santo (2007), Desflorestamento na Amazonia-Sistema DETER, in Sensor MODIS e Suas Aplicar;i5es Ambientais no Brasil, edited by B. F. T. Rudorff, Y. E. Shimabukuro, and J. C. Ceballos, pp. 389-401, Editora Parentese, Sao Jose dos Campos, Brazil. Skole, D., and C. Tucker (1993), Tropical deforestation and habitat fragmentation in the Amazon - Satellite data from 1978 to 1988, Science, 260, 1905-1910. Soares-Filho, B., A. Alencar, D. Nesptad, M. Cerqueira, M. C. V. Diaz, S. Rivero, L. Solorzano, and E. Voll (2004), Simulating the response of land-cover changes to road paving and governance along a major Amazon highway: The Santarem-Cuiaba cOlTidor, Global Change Bioi., 10, 745-764. Souza, C., Jr., L. Firestone, L. M. Silva, and D. Roberts (2003), Mapping forest degradation in the Eastern Amazon from SPOT 4 through spectral mixture models, Remote Sens. Environ., 87, 494-506.
Souza, C., Jr., D. A. Roberts, and M. A. Cochrane (2005), Combining spectral and spatial information to map canopy damages from selective logging and forest fires, Remote Sens. Environ., 98, 329-343. Tardin, A. T., D. C. L. Lee, R. 1. R. Santos, O. R. Assis, M. P. S. Barbosa, M. L. Moreira, M. T. Pereira, D. Silva, and C. P. Santos Filho (1980), Subprojeto Desmatamento: Convenio IBDF/ CNPq-INPE, Technical Report INPE-I649-RPE/I03, Instituto de Pesquisas Espaciais, Sao Jose dos Campos, Brazil. Uhl, C., A. Verissimo, M. M. Mattos, Z. Brandino, and I. C. G. Vieira (1991), Social, economic, and ecological consequences of selective logging in an Amazon frontier-The case of Tailandia, For. Ecol. Manage., 46, 243-273. Veiga, J. B., 1. F. Tourrand, M. G. Piketty, R. Poccard-Chapuis, A. M. Alves, and M. C. Thales (2004), Expansiio e Trajetorias da Pecuaria na Amazonia: Para, Brasil, Editora Universidade de Brasilia, Brasilia, Brazil. Velho, O. G. (1976), Capitalismo Autoritario e Campesinato, DIFEL, Sao Paulo, Brazil. Verissimo, A., P. BalTeto, M. Mattos, R. Tarifa, and C. Uhl (1992), Logging impacts and prospects for sustainable forest management in an old Amazonian frontier-The case of Paragominas, For. Ecol. Manage., 55, 169-199. Warnken, P. F. (1999), The Development and Growth of the Soybean Indusfly in Brazil, Iowa State Univ., Ames. Zarin, D. J., et al. (2005), Legacy of fire slows carbon accumulation in Amazonian forest regrowth, Front. Ecol. Environ., 3, 365-369. I
D. S. Alves, InStltuto Nacional de Pesquisas Espaciais (INPE), DPI (SRE 2), Avenida dos Astronautas 1758, CEP 12227-010, Sao Jose dos Campos SP, Brazil. ([email protected]) M. Batistella, Embrapa Satellite Monitoring, Avenida Soldado Passarinho, 303 CEP 13070-15, Campinas SP, Brazil. (mb@cnpm. embrapa.br) D. C. Morton, Goddard Space Flight Center, 8800 Greenbelt Road, Code 614.4, Greenbelt, MD 20771, USA. (douglas.morton@ nasa.gov, [email protected]) D. A. Roberts, Department of Geography, EH 1832, University of California Santa Barbara, Santa Barbara, CA 93117, USA. ([email protected]) C. Souza Jr., Instituto do Homem e Meio Ambiente da Amazonia (Imazon), Rua Domingos Marreiros 2020, CEP 66060-160, Belem PA, Brazil. ([email protected])
Selective Logging and Its Relation to Deforestation Gregory P. Asner,' Michael Keller,z,3 Marco Lentini,4 Frank Merry,5 and Carlos Souza Jr. 6 Selective logging is a major contributor to the social, economic, and ecological dynamics of Brazilian Amazonia. Logging activities have expanded from lowvolume floodplain harvests in past centuries to high-volume operations today that take about 25 million m 3ofwood from the forest each year. The most common highimpact conventional and often illegal logging practices result in major collateral forest damage, with cascading effects on ecosystem processes. Initial carbon losses and forest recovery rates following timber harvest are tightly linked to initial logging intensity, which drives changes in forest gap fraction, fragmentation, and the light environment. Other ecological processes affected by selective logging include nutrient cycling, hydrological function, and postharvest disturbance such as fire. This chapter synthesizes the ecological impacts of selective logging, in the context of the recent socioeconomic conditions throughout Brazilian Amazonia, as determined from field-based and remote sensing studies' carried out during the Large-Scale Biosphere-Atmosphere Experiment in Amazonia program.
I 1. INTRODUCTION
newab1e resource for the region. There is general consensus that selective logging is widespread and important to the economy; however, the industry has suffered from generally weak and inconsistent government oversight, low capital investment, and a lack of understanding of both forest ecology and management. This combination of conditions has prevented the development of a sustainable logging industry and has led to considerable ecological damage. In the past decade, the ecological, social, and geographic sciences have made important but disparate strides to understand the dynamics of selective logging in Amazonia, with a focus on Brazil where most studies have been conducted. Our goal here is to synthesize the work from these studies and to clarify our understanding of the ecological role of timber production. We focus on the contributions from the Large-scale Biosphere-Atmosphere Experiment in Amazonia (LBA) program. We start with a brief history of the logging industry in Brazil, including the pertinent aspects of the social, economic, and policy drivers oflogging practices. We then link this knowledge of the historical and contemporary conditions for the Amazon forest industry to recent scientific
Selective logging is an important land use in Amazonia. The logging industry is an economic engine that generates revenue, provides jobs, and has the potential to be a re-
'Department of Global Ecology, Carnegie Institution, Stanford, California, USA. 2International Institute of Tropical Forestry, USDA Forest Service, Rio Piedras, Puerto Rico. 3NEON, Inc., Boulder, Colorado, USA. 4Instituto Floresta Tropical, Belem, Brazil. sWoods Hole Research Center, Falmouth, Massachusetts, USA. 6Instituto do Homem e Meio Ambiente da Amazonia, Belem, Brazil.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000723
25
26
SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
findings demonstrating the effects of logging on the ecology of the region. Throughout the chapter, we also highlight the contributions of remote sensing as a tool to understand and monitor the course and consequences of selective logging in Amazonia. 2. SOCIOECONOMIC CHARACTERIZATION OF SELECTIVE LOGGING
2.1. Development ofLogging Frontiers in Amazonia European settlers had begun logging the Amazon forest by the seventeenth century [Rankin, 1985]. For the first three centuries of settlement, logging was restricted to low-volume harvest of floodplain forests along the main Amazonian rivers and was of secondary importance to other extractive industries such as Brazil nuts and rubber. It was not until the 1950s that industrial mills, mainly subsidiaries of large international companies such as Georgia Pacific, sprang up in the Amazon estuary to produce high-value sawn wood and veneer for export. Among the earliest examples of selective logging were two floodplain tree species known as virola (Virola surinamensis) and andiroba (Carapa guianensis) [Barros and Uhl, 1995; Pinedo-Vasquez et al., 2001; Zarin et al., 2001]. In the 1960s and 1970s, government policies and investments in infrastructure throughout Amazonia opened access to extensive portions of upland forests [Binswanger, 1991; Browder, 1988; Scholz, 2000]. An extensive and migratory logging industry emerged based on a low-cost raw material in newly forming economic frontiers with minimal governance [Uhl et al., 1997; Verissimo et ai., 1998,2002; Stone, 1998a]. The industry blossomed into a diverse sector with new products and extensive national markets, changing the nature of selective logging along the way. Instead of one or two key species destined for export, a domestic market based on rough-sawn wood and eventually pl~ood absorbed a greater variety of species. Notwithstanding the penetration of mahogany logging deep into the forest, the majority of selective logging operations followed the new roads to harvest high volumes. The new logging strategy created boom and bust economies, severe ecological damage, and a legacy of wasteful and marginally legal practices that still pervade the industry. After 3 decades of deforestation and unplanned selective logging, timber stocks in the old frontiers became largely depleted. Old logging frontiers (Plate 1), which closely follow the arc of deforestation in the states of Para, Mato Grosso, and Rondonia, still encompass 45% of the Amazonian logging centers, but they now generate only about 50% of the revenues and jobs of the timber industry [Lentini et al.,
2005]. The increasing scarcity of raw material stimulated the migration of firms to newer frontiers (intermediate and new frontiers in Plate 1). Roads that strike deep into the interior of Amazonia, mainly the BR-163 Cuiaba-Santarem Highway and, to a lesser extent, the BR-230 Transamazon Highway, have seen a dramatic surge in sawmills and logging. Nepstad et al. [1999] used sawmill surveys conducted by the Brazilian nongovernmental organization Instituto do Homem e Meio Ambiente da Amazonia (IMAZON) and showed that, for the period 1995-1996, logging centers were busy in nearly all states of Brazilian Amazonia (Plate 2a). The pattern of logging centers is similar to the detailed geographic distribution oflogged forest revealed in a remote sensing analysis for the years 1999-2002 (Plate 2b) [Asner et al., 2005]. In 2004, IMAZON catalogued 82 Amazonian logging centers encompassing 3132 timber mills, which consumed 24.5 million m 3 oflogs that produced lOA million m 3 of processed timber including sawn wood, veneers, plywood, and finished wood products (Table 1). This implies an average production yield of only 42% [Lentini et al., 2005]. More than 90% of the production from Brazilian Amazonia is currently concentrated in the states of Para, Mato Grosso and Rondonia. The total gross revenue of the timber industry in Brazilian Amazonia in 2004 was about $2.3 billion U.S. dollars generating approximately 380,000 jobs, including 124,000 direct jobs (processing and logging) and 255,000 indirect jobs. Although there are cost and market share differences between the new, intermediate and old frontiers, this frontier migration has not been accompanied by notable improvements in forest management and timber processing [Merry et al., 2006], as discussed in section 2.2.
2.2. Economics ofSelective Logging In their quest for high-quality raw materials, loggers seek new forest frontiers. The economics of the logging industry in Brazil directly influence the management approaches and therefore strongly affect ecological impacts and the longterm su~tainability of forest timber production. As we mentioned in section 2.1, selective logging has evolved from a single- or few-species model, typical of floodplain logging and mahogany harvest, to a model that can remove up to 40 m 3 ha- 1 and can comprise any variety of 50 or more species. This approach, called conventional logging, is widely used and is profitable. Spatial-economic models estimate the feasible extent of selective logging in Brazilian Amazonia, based on the expected costs for harvest timbe~ and log transport and the prices for logs in the logging centers [Stone, 1998b; Verissimo et al., 1998,2000]. Basically, these models (e.g., Plate 1) identify forests in which selective logging is economically viable and show widespread potential for
ASNER ET AL.
70·VV ~,
65·VV
55· VV
50· VV
45' VV
~
)~7
COLOMBIA
60· VV
27
I
VENEZUELA
~
5· N
"
o
5· S
LoggIng centers o
100,000 to 200,000 m3
•
200,000 to 600,000 m3
•
> 600,000 m3
Logging frontiers
PERU
BOLIVIA
o
Old (> 30 years)
_ _
Intermediary (10-30 years) New « 10 years)
_
Estuarine zone
NMain roads
o
Brazilian Amazon States
Plate 1. Geographic distribution oflogging centers in 2004 and harvestable regi~ns of the Brazilian Amazon [Lentini I et al., 2005].
the forest sector to play an important role in the development of emerging frontiers. Plates 1 and 2 demonstrate that logging centers have expanded sufficiently so that harvestable forests cover most of Brazilian Amazonia. The growth of the logging industry in Brazil has not been determined by new harvesting and processing technologies that are available to logging companies. Instead, the highdamage, high-waste approaches involved in many conventionallogging methods have persisted [Pereira et al., 2002; Holmes et ai., 2002; Asner et al., 2006]. Moreover, despite technological advances in forest management, whereby reduced-impact logging (RIL) methods can be employed to harvest wood while minimizing damage to the forest and maintaining economic returns [Sist, 2000], there has been poor adoption of good management practices. A brief list of reasons why RIL has not been widely adopted includes poorly defined property rights, high transaction costs associated with government bureaucracy, poor distribution of information on good forest management, and an entrenched rentseeking bureaucracy [Putz et al., 2000; Boltz et al., 2001]. It is these influences on firm decision-making that continue to
encourage the use of poor-quality selective logging practices over RIL. As discussed in section 3, this timber-harvesting environment results in ecological responses that have just recently been quantified during the LBA program. While RIL logging has many ecological benefits, the economic benefits of RIL are less certain [Putz et al., 2000]. Among the problems ofRIL logging is, ironically, the preservation of a nearly intact canopy. For all of its biodiversity, microclimatic, and fire protection benefits, an intact canopy keeps potential regenerating trees in the dark and thereby limits postlogging growth. A potential solution to this problem is the selective elimination of competitors around future harvest trees, known as liberation [Wadsworth and Zweede, 2006]. Dauber et al. [2005] modeled tree growth based on field data from an extensive network of plots with growth rates measured on more than 10,000 trees in Bolivian Amazonia. Tree growth was modeled for no treatment and for a silvicultural treatment where surrounding competitive trees and vines are killed for a 25-year cutting cycle in four regions. While modeled first-harvest volumes were considerably larger than the second-harvest volumes for all four
26
SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
findings demonstrating the effects oflogging on the ecology of the region. Throughout the chapter, we also highlight the contributions of remote sensing as a tool to understand and monitor tQe course and consequences of selective logging in Amazonia. 2. SOCIOECONOMIC CHARACTERIZATION OF SELECTIVE LOGGING 2.1. Development ofLogging Frontiers in Amazonia
2005]. The increasing scarcity ofraw material stimulated the migration of firms to newer frontiers (intermediate and new frontiers in Plate 1). Roads that strike deep into the interior of Amazonia, mainly the BR-163 Cuiaba-Santarem Highway and, to a lesser extent, the BR-230 Transamazon Highway, have seen a dramatic surge in sawmills and logging. Nepstad et al. [1999] used sawmill surveys conducted by the Brazilian nongovernmental organization Instituto do Homem e Meio Ambiente da Amazonia (IMAZON) and showed that, for the period 1995-1996, logging centers were busy in nearly all states of Brazilian Amazonia (Plate 2a). The pattern of logging centers is similar to the detailed geographic distribution oflogged forest revealed in a remote sensing analysis for the years 1999-2002 (Plate 2b) [Asner et aI., 2005]. In 2004, IMAZON catalogued 82 Amazonian logging centers encompassing 3132 timber mills, which consumed 24.5 million m3 of logs that produced 10.4 million m3 of processed timber including sawn wood, veneers, plywood, and finished wood products (Table 1). This implies an average production yield of only 42% [Lentini et al., 2005]. More than 90% of the production from Brazilian Amazonia is currently concentrated in the states of Para, Mato Grosso and Rondonia. The total gross revenue ofthe timber industry in Brazilian Amazonia in 2004 was about $2.3 billion U.S. dollars generating approximately 380,000 jobs, including 124,000 direct jobs (processing and logging) and 255,000 indirect jobs. Although there are cost and market share differences between the new, intermediate and old frontiers, this frontier migration has not been accompanied by notable improvements in forest management and timber processing [Merry et al., 2006], as discussed in section 2.2.
European settlers had begun logging the Amazon forest by the seventeenth century [Rankin, 1985]. For the first three centuries ofsettlement, logging was restricted to low-volume harvest of floodplain forests along the main Amazonian rivers and was of secondary importance to other extractive industries such as Brazil nuts and rubber. It was not until the 1950s that industrial mills, mainly subsidiaries of large international companies such as Georgia Pacific, sprang up in the Amazon estuary to produce high-value sawn wood and veneer for export. Among the earliest examples of selective logging were two floodplain tree species known as virola (Virola surinamensis) and andiroba (Carapa guianensis) [Barros and Uhl, 1995; Pinedo-Vasquez et al., 2001; Zarin et al., 2001]. In the 1960s and 1970s, government policies and investments in infrastructure throughout Amazonia opened access to extensive portions of upland forests [Binswanger, 1991; Browder, 1988; Scholz, 2000]. An extensive and migratory logging industry emerged based on a low-cost raw material in newly forming economic frontiers with minimal governance [Uhl et al., 1997; Verissimo et al., 1998,2002; Stone, 1998a]. The industry blossomed into a diverse sector with 2.2. Economics ofSelective Logging new products and extensive national markets, changing the nature of selective logging along the way. Instead of one or In their quest for high-quality raw materials, loggers seek two key species destined for export, a domestic market based new forest frontiers. The economics of the logging industry on rough-sawn wood and eventually pl)"Yood absorbed a in Brazil directly influence the management approaches and greater variety of species. Notwithstanding the penetration therefore strongly affect ecological impacts and the longof mahogany logging deep into the forest, the majority of term sw;tainability of forest timber production. As we menselective logging operations followed the new roads to har- tioned in section 2.1, selective logging has evolved from a vest high volumes. The new logging strategy created boom single- or few-species model, typical of floodplain logging and bust economies, severe ecological damage, and a legacy and mahogany harvest, to a model that can remove up to of wasteful and marginally legal practices that still pervade 40 m3 ha- I and can comprise any variety of 50 or more spethe industry. cies. This approach, called conventional logging, is widely After 3. decades of deforestation and unplanned selective used and is profitable. Spatial-economic models estimate the logging, timber stocks in the old frontiers became largely de- feasible extent of selective logging in Brazili\ln Amazonia, pleted. Old logging frontiers (Plate 1), which closely follow based on the expected costs for harvest timber and log transthe arc of deforestation in the states of Para, Mato Grosso, port and the prices for logs in the logging centers [Stone, and Rondonia, still encompass 45% of the Amazonian log- 1998b; Verissimo et al., 1998,2000]. Basically, these modging centers, but they now generate only about 50% of the els (e.g., Plate 1) identify forests in which selective logging revenues and jobs of the timber industry [Lentini et aI., is economically viable and show widespread potential for
ASNERET AL.
70·W
6S·W
60·W
~,
" ..
OOLOM."~
VENEZUELA
!
SS·W
G-"'·
SO·W
27
4S·W
S· N
N
A o
5· S
LoggIng centers o
100,000 to 200,000 m3
•
200,000 to 600,000 m3
•
> 600,000 m3
Logging frontiers
o
Old (> 30 years)
_
Intermediary (10-30 years)
_
New « 10 years)
_
Estuarine zone
NMainroads
o
Brazilian Amazon States
I
Plate 1. Geographic distribution of logging centers in 2004 and harvestable regipns of the Brazilian Amazon [Lentini et ai., 2005]. I
the forest sector to play an important role in the development of emerging frontiers. Plates 1 and 2 demonstrate that logging centers have expanded sufficiently so that harvestable forests cover most of Brazilian Amazonia. The growth of the logging industry in Brazil has not been determined by new harvesting and processing technologies that are available to logging companies. Instead, the highdamage, high-waste approaches involved in many conventionallogging methods have persisted [Pereira et al., 2002; Holmes et al., 2002; Asner et al., 2006]. Moreover, despite technological advances in forest management, whereby reduced-impact logging (RIL) methods can be employed to harvest wood while minimizing damage to the forest and maintaining economic returns [Sist, 2000], there has been poor adoption of good management practices. A brief list of reasons why RIL has not been widely adopted includes poorl)' defined property rights, high transaction costs associated with government bureaucracy, poor distribution of information on good forest management, and an entrenched rentseeking bureaucracy [Putz et al., 2000; Boltz et al., 2001]. It is these influences on firm decision-making that continue to
encourage the use ofpoor-quality selective logging practices over RIL. As discussed in section 3, this timber-harvesting environment results in ecological responses that have just recently been quantified during the LBA program. While RIL logging has many ecological benefits, the economic benefits of RIL are less certain [Putz et al., 2000]. Among the problems ofRIL logging is, ironically, the preservation of a nearly intact canopy. For all of its biodiversity, microclimatic, and fire protection benefits, an intact canopy keeps potential regenerating trees in the dark and thereby limits postlogging growth. A potential solution to this problem is the selective elimination of competitors around future harvest trees, known as liberation [Wadsworth and Zweede, 2006]. Dauber et al. [2005] modeled tree growth based on field data from an extensive network of plots with growth rates measured on more than 10,000 trees in Bolivian Amazonia. Tree growth was modeled for no treatment and for a silvicultural treatment where surrounding competitive trees and vines are killed for a 25-year cutting cycle in four regions. While modeled first-harvest volumes were considerably larger than the second-harvest volumes for all four
28
SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
ASNER ET AL.
29
Table 1. Socioeconomic Data on Selective Logging in 2004 for Brazilian Amazonia"
,
State Acre Amapa Amazonas Maranhao Mato Grosso Para Rondonia Roraima Combined
SON
O·
S'S
Lo ~ing Centers ( Imber Firms) 1(52) 1 (73) 3 (48) 1 (45) 26 (872) 33 (1,592) 16 (422) 1 (28) 82 (3,132)
Market
Logwood Consumption (x 1000 m 3 a-I)
Gross Income (million U.S. dollars)
Jobs Generated: Direct and Indirect
Exportation (%)
Regional (%)
420 130 490 430 8,010 11,150 3,700 130 24,460
41.6 9.3 55.9 31.7 673.9 1,113.6 368.9 15.9 2,310.7
5,729 2,228 11,344 6,817 108,569 183,741 58,818 2,375 379,621
83 34 64 9 19 50 27 79 36
12 67 18 35 9 11 11 21 11
"Data are retabulated from Lentini et at. [2005].
regions, in the best-case transitional (dry-to-moist forest) ecoregion, the second cut reached 64% of the volume of the first cut under silvicultural treatment compared to only 28% for untreated forest. Silvicultural treatments are costly and currently rarely implemented in Amazonia.
0·8
2.3. Role a/Illegal Logging
N
A b
_
1999-2000 Logging
_
200().2001 logging
_
2001·2002 Logging
_
Federal Conlervatlon Units
_
Indlgenoua ReSeNtS ~
o
__
-~===::11~
500
1.000
~Km
2,000
Plate 2. (a) Regional distribution oflogging centers in the Brazilian Amazon, 1995-1996, derived from sawmill surveys [Nepstad et at., 1999]. Reprinted by permission from Macmillan Publishers Ltd: Nature, copyright 1999. (b) Regional distribution oflogging damage to forests from 1999 to 2002 in the states ofPani (PA), Roraima (RR), Rondonia (RO), Acre (Ae), and northern Mato Grosso (MT) derived from satellite analysis [Asner et at., 2005].
Because it has been widespread, the practice of illegal logging requires some extra attention here. There are two legal mechanisms to gain permission to log forests in Brazilian Amazonia: forest management plans, regulated by specific policy instruments, and deforestation. Current Brazilian law allows the deforestation of 20% of the total area in rural Amazonian properties. In the past, both mechanisms were controlled by the federal environmental agency Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renovaveis (mAMA). Currently, the enactment of a new forestry law (Lei 11284/2006) created the first directives to decentralize the control over forest management plans and to delegate authorizations for deforestation to state agencies in an attempt to increase the transparency of these processes [Zarin et al., 2007]. The few available figures for the log wood volume generated through these mechanisms show that their sum was less than 15 million m 3 between 2000 and 2004. In contrast, surveys of the timber industry carried out in 1998 and 2004 [Lentini et al., 2005] show that the total Amazonian production during this period was above 24 million m 3, a figure directly supported by satellite estimates of Asner et al. [2005]. Combining these estimates, it becomes clear that during these years, at least 40% of the log produc~ tion in Amazonia was illegally generated. In the last several years, government and civil society have taken measures against illegal logging. Hundreds of forest management plans were canceled by mAMA in 2003-2004 in an attempt to halt illegal logging and decrease the defor-
estation rates, estimated by Instituto Nacional de Pesquisas Espaciais (INPE) (the National Space Agency) at 1.7 million ha in 2004 (PRODES: Assessment of deforestation in Brazilian Amazonia, 2005, see http://www.obt.inpe.br/prodes/ index.htrnl). In 2005, the Ministry of Environment, Federal Police, mAMA, and several governmental organizations conducted three large-scale operations targeting illegality in the forest sector and corruption. Despite these recent efforts, the scale of illegal operations remains difficult to precisely I estimate. Illegal logging, conducted without government approved management plan1 and without permits, has especially deleterious economic, social, and ecological effects. Economic impacts can be me.asured by losses in governmental taxes and poor development of economic sectors related to logging. From a societal perspective, formally regulated employment is replaced by informality leading to the low quality ofthese jobs, higher risks for forest workers, lower benefits, and generation of conflicts between loggers and traditional communities over land use. Ecologically, while legal management plans limit logging entries for extended periods (often 30 years), illegal logging promotes multiple logging entries into an area as market conditions change. Multiple entries result in forest impoverishment, a dramatic loss of biodiversity, and increased susceptibility to fire [Nepstad et al., 1999]. 3. ECOLOGICAL IMPACTS OF CURRENT LOGGING PRACTICES The ecological impacts of selective logging are directly related to harvest intensity, in terms of volume of wood removed per hectare and harvest method, which largely determines the level of collateral damage incurred during and after timber harvest. Harvest methods, ranging from largescale conventional logging using crawler tractors and/or wheeled skidders to carefully planned and managed RIL, are
30
SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
tightly linked to both the initial forest damage and to the longer-term ecological responseSTanging from canopy light environment to carbon cycling to fire regimes. The eff&ts of selective logging start with forest biophysical properties, where the forest canopy cover (measured by gap fraction or light availability) is immediately changed during harvest operations. Changes in the spatial pattern and extent of forest gaps then have cascading effects on rates of forest growth, hydrological processes, and the entire food web of a forest. These gaps 'can be arranged by forest-use stratum, including access roads, tree falls where logs are initially cut, skid trails used to remove logs from the forest, and log decks (commonly known as patios) used to load logs onto trucks (Plate 3). These four strata help to organize the ecological impacts as discussed below, and their pattern across the landscape, both in terms of number and density, exerts significant control over the rate of forest recovery for all organisms. Although roads, log decks, skids, and tree fall locations may be visible to satellites for a few years following timber harvest, the canopy rapidly closes (Plate 4), making the geography of logging difficult to track and misguiding the casual observer into thinking that forest recovery is rapid (Appendix A). In fact, while canopy closure is rapid, forest recovery, in terms of both carbon stocks and ecological processes, is far from rapid in upland rain forest areas of Amazonia. Selective logging alters forest ecological processes extending from changes in phenology to the way that carbon and nutrients are cycled and even to the emissions of trace gases such as nitrogen oxides and methane. Although LBA did not focus on tropical forest wildlife, it is important to acknowledge the impacts of logging here because they can be highly variable and often substantial. Several studies have documented how selective logging can cause biotic impoverishment of species and communities or, alternatively, can stimulate population growth of some species [Johns, 1992; Thiollay, 1992; Hill et al., 1995]. Both the magnitude and direction of ecological challge following harvest depend heavily upon the initial logging intensity and the subsequent spatial and temporal dynamics offorest gap fraction (see Fimbel et al. [2001] for an extensive synthesis). 3.1. Light Environment
A major ecological constraint over plant establishment and regeneration in tropical forests is the low light intensity in the understory [Johns et al., 1996]. In humid tropical forests, roughly 2-3% of photosynthetically active radiation (PAR) (400-700 nm) reaches the forest floor [Lee, 1987], and canopy gap fraction typically ranges from just 2 to 4% [e.g., Chazdon and Fetcher, 1984]. Plant productivity or carbon uptake by vegetation is tightly linked to PAR availabil-
ity [Monteith, 1972; Field et al., 1995]. Canopy gaps created by selective timber harvests have immediate effects on PAR interception, latent and sensible heat fluxes, water stress, and plant productivity in tropical forests [Healey et al., 2000; Pinard and Cropper, 2000]. Rates of forest regeneration can thus be linked to the size, number, and spatial arrangement of canopy gaps following harvest [Pereira et al., 2002]. The light environment following logging can be assessed in terms of ground damage resulting from timber harvest (e.g., skids, roads, and decks) and in terms of canopy gap fraction among these types of damage classes. Across a wide range of conventional and RIL harvest intensities (2.6-6.4 felled trees ha- 1) in the eastern Amazon, Asner et al. [2004b] found that the majority of ground damage occurred as skid trails (4-12%), whereas log decks and roads were only a small contributor to the total ground damage «2%). Feldpausch et al. [2005] identified a similar pattern among RIL plots in Mato Grosso, Brazil. However, despite similar harvest intensities, conventional logging causes more canopy damage than does RIL, either from the initial harvest or from tree falls that occur in the years following the logging event [Pereira et al., 2002; Schulze and Zweede, 2006]. Critically, neither the number of log decks nor their individual or total area is well correlated with the number of trees removed or intensity of tree harvesting (trees per hectare) [Asner et al., 2004a]. In contrast, the area of skids is often well correlated with the ground area damaged (square meters) per tree felled, but these features are among the most difficult to map in the field or from satellite sensors (Appendix A). In terms of light interception by the canopy following logging, field surveys across the damage classes show that gap fractions are highest in log decks and lowest in tree fall areas immediately following timber harvest [Feldpausch et al., 2005]. However, the small surface area of log decks reduces their contribution to a very small fraction of the areaintegrated effects of logging throughout a forest. In contrast, lower gap fractions from tree falls are spread throughout a harvested forest, resulting in a large contribution of these areas to the total stand-level canopy gap fraction [Asner et al., 2006]. Canopy openings at tree fall locations are highest at the point where the crown is removed and then decrease with distance from each felled crown (Figure 1). Following harvests in the eastern Amazon, the area affected by the felling of each tree was approximately 100 m in radius for conventional logging but only 50 m for RIL [Asner et al., 2004b]. The size and duration of the tree f~ll gaps have a significant impact on PAR interception and the resulting primary production following timber harvests [Huang et al., 2008]. Pereira et al. [2002] demonstrated the advantages of RIL methods in the maintenance of canopy cover following timber harvest. Feldpausch et al. [2005] observed that RIL
ASNER ET AL.
I[~J
Skids
LZ:J Roads
_
Decf<s • Harvesled Trees
31
I
..'
Plate 3. Plan view of an actual 100 ha logged area in the eastern Brazilian Amazon harvested using conventional methods [Pereira et al., 2002]. r !
reduces canopy damage but only when harvest intensities, calculated in terms of wood volume, are relatively high. In sum, the positive effects of RIL practices on canopy cover are mostly realized at greater harvest volumes; otherwise, they converge on conventional logging damage levels when extraction rates are very low.
the overall carbon laccounting is small. Most of the carbon contained in logs ttansported from forest to mills is rapidly cycled to the atmosphere because sawmill waste is generally burned. Log harvesting operations in the Amazon are inefficient and create a great deal of collateral damage [Verissimo et ~l., 1992]. Table 2 shows that 6 times as much carbon is com3.2. Carbon Cycling mitted to waste in the forest (including fallen and standing coarse woody debris and belowground debris) as is exported The total carbon budget from logging depends upon a bal- as roundwood. In well-managed, low-impact operations, the ance of the long-term storage of carbon in wood products, ratio of debris creation to exported wood can be as low as 2.4 the losses resulting from inefficient milling and processing, [Feldpausch et al., 2005]. The waste includes the portions the losses related to decomposition of wood resulting from of harvested trees that are not trucked out of the forest, trees logging damage to the forest, and the carbon gains over time killed by felling operations, and especially trees killed by from forest regrowth. Carbon lost as a result of the logging the operation of heavy equipment used to open forest roads in the Amazon expressed on an areal basis is summarized in and to skid logs out of the forest. Research indicates that Table 2. Industrial logging in the Amazon region of Brazil skidding operations are responsible for a majority of the colremoves about 19 to 40 m) ha- I of roundwood from the for- lateral damage and that improvements in skidding through est. Compared to other losses, the carbon lost as a result of . the use of proper equipment and planning can cut the damlog processing is small (~12% of the total). Finished prod- age and carbon loss by half [Pereira et al., 2002; Keller et ucts account for only 42% ofthe roundwood removed from al., 2004b]. Most of the waste is generated immediately or the forest [Lentini et al., 2005]. While these products even- within a year of logging. The loss of carbon from the ecotually decay, the lifetime of wood products from Amazon system to the atmosphere is not instantaneous. However, logging is unknown. The relevance of finished products to under the hot and humid conditions of the Amazon, the av-
32
SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
ASNER ET AL.
(b)
(a) OJ
..-
0.5 years postharvest
jO'4 8
b
.
33
I • .RIL 0-
CL
I
~Q. 0.3 &
0.5
r-...--~--'--~----r--======o
1.5 years postharvest 0.4
I•
RIL
--0-
I
CL
0.3
.~0,2
0.2
g ~ 0.1
0.1 0,0
0,0
0
20
40
80
100
80
(c) t:
ts
2.5 years postharvest 0.4
I• --0-
I CL
d
20
40
60
3.5 years postharvest 0.4
~ 0.3
0.3
Cl ~
0.2
80
100
----,,...--====:-1
0,6 ,-.,.....--.--.......
RIL
f!
c
o
(d) 0.6
.2
'---'-------'--~~--'------'--~'----'
I•
RIL
--0-
I
CL
eu
0.2
0
lij
U 0.1 0.0 20
0
40
80
100
80
Distance from felled tree crown (m)
o
20
40
60
80
100
Distance ffm felled tree crown (m)
Figure 1. Mean (plus/minus standard deviation) forest canopy gap fraction at incr~asing distance from felled tree crowns in conventionally logged (CL) and reduced-impact logging (RIL) sites in the eastern Amazon at (a) 0.5, (b) 1.5, (c) 2.5, and (d) 3.5 years following timber harvest [Asner et al., 2004b].
e
f
erage lifetime of coarse woody debris is about 5 to 7 years [Chambers et al., 2000; Palace et al., 2007]. Despite this basic understanding of carbon losses following timber harvest, the primary sources of data remain limited, and additional measurements are needed.
The belowground contribution to the carbon balance is especially uncertain because of the precarious knowledge of belowground carbon stocks in the Amazon [Keller et ai., 2001]. Root stocks are poorly quantified, and it is difficult to quantify small changes in soil carbon pools. In one study,
Table 2. Estimates of Carbon Loss From Logginga Mean (Mg C ha- 1)
Plate 4. Sequence ofLandsat images showing roads and log decks generated by selective logging in central Mato Grosso and the rapid apparent recovery of the canopy, even following fire [Souza et al., 2005]. Colors indicate different normalized difference fraction index from high canopy cover in greens to low canopy cover in pinks and whites. ,
II
... !
Roundwood Aboveground woody debris Belowground woody debris . Standing dead Total
5
26 6 5 42
Low (Mg C ha- ' )
High (Mg C ha- ' )
4 21 5 4 33
8 32 8 6 55
aRoundwood estimates for mean, high, and low harvest volumes are from Asner et al. [2005] and Nepstad et al. [1999]. Roundwood density is assumed to be 0.7 Mg m3, and the conversion factor for wood mass to C is 0.5 [Schlesinger, 1997]. The roundwood loss is adjusted to account for 42% of the wood t4at becomes durable products [Lentini et al., 2005]. Coarse woody debris estimates are based on the work of Keller et al. [2004b]. Belowground loss is calculated as 20% of aboveground coarse woody debris plus roundwood [Keller et al., 2001]. Standing dead is 20% ofaboveground coarse woody debris [Palace et al., 2007].
ASNER ET AL.
34 SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
.:. .l.
researchers quantified soil C02 flux following reducedimpact logging in the Tapaj6s National Forest of central Para State [Keller et al., 2005]. The authors found a slight increase in ~02 emissions from tree fall gaps and skid trails and a decrease in emissions from log decks, compared to background forest emissions. When the C02 emissions were aggregated spatially, the emissions from logged forest were statistically indistinguishable from background emissions. It is likely that increased C02 production related to the decomposition of dead roots is offset 'by the loss of root respiration resulting in no net change in C02 flux [Varner et ai., 2003; Silver et al., 2005]. The rate of regrowth in logged forests in the Amazon has been measured in only a few very limited cases in experimental plots, and those regrowth estimates have focused on estimation of future timber production [e.g., Silva et al., 1995]. We are not aware of any published studies of regrowth rates and carbon budgets following conventional industrial logging in the Amazon. Regrowth rates depend on a variety offactors dominated by light availability (gap fraction) and the efficiency oflight use by the canopies [Huang et ai., 2008]. In general, carbon uptake rates will be highest where a greater portion of the canopy has been opened. Therefore, gross oarbon uptake will be highest where the largest amount of slash has been generated, and this is where the largest gross carbon losses will be incurred from decomposition. Optimistic modeling scenarios suggest that timber
production may be maintained for up to 200 years under 30year rotations, provided that the market accepts new potentially commercial species with time [Alder and Silva, 2000; Keller et ai., 2004a]. However, without substantial silvicultural intervention [Dauber et al., 2005], fast growing species with low wood density will tend to replace slow growing species with high wood density, leading to forest stands with lower total carbon stocks [Keller et al., 2004a; Bunker et al., 2005]. In their simulations, Keller et al. [2004a] and Huang et al. [2008] predicted that Amazon forests would lose an average of 12-19 Mg C ha- I over the first 30 years of rotation (a new area is cut each year) and between 16 and 30 Mg C ha- I with reentry logging cycles over 200 years (Figure 2). It is important to point out that these models are based on average growth rates. Brienen and Zuidema [2007] have shown that skewed distributions ofthe tree growth should be considered for more realistic simulations of tropical forest production. 3.3. Other Biogeochemical Cycles
Beyond its direct effects on forest structure and the carbon cycle, selective logging can alter nutrient cycles and other key biogeochemical processes regulating forest productivity and neighboring aquatic systems. For example, nitrate losses from logged areas in Guyana vary in proportion to the area of soil disturbance surrounding harvested trees [Brou-
200 180 160 10
..c 140
U
0)
e.
120
10
E 100
"
.Q
lD
RIL Slow Growth CL Slow Growth RIL Fast Growth CL Fast Growth
80 60
•• 0
.'
; .. "
:
"
"
: 0'
:'
40 0
50
100
150
200
Year
I
J
L
!
Figure 2, Simulated recovery of aboveground biomass in eastern Amazon forest sites under 3D-year revisit scenarios using conventional logging (CL) and reduced-impact logging (RIL) extraction techniques [Keller et at., 2D04a].
wer, 1996]. Increases in~itrate availability in tropical forest soils stimulate denitrification and nitrogen oxide emissions [Keller and Reiners;'1994]. Logging often has its mostobvious effect on s9ns via increased erosion and runoff, with concomitant lo~s ofrtutrients and organic matter [McNabb et al., 1997; Brouwer, 1996; Johns et al., 1996]. Both the magnitude and direction of ecological change following harvest depend heavily upon the initial logging intensity and the subsequent spatial and temporal dynamics of forest gap fraction. Timber harvesting alters stocks and fluxes ofplant nutrients in the landscape, with potential implications for short-term recovery and long-term sustainability of logging practices in tropical forests [Brouwer, 1996; McNabb et al., 1997]. In particular, rock-derived nutrients such as phosphorus (P), calcium (Ca), and magnesium (Mg) are scarce in the most common soil types found in the eastern Amazon [Sanchez, 1976; Silver et ai., 2000]. Ultisols and oxisols contain little mobile inorganic P. Most P is found in organic pools or unavailable forms in the mineral soil, and both P and base cations (Ca, Mg, etc.) are recycled very efficiently in humid tropical forests [Vitousek and Sanford, 1986; S~ewart and Tiessen, 1987]. Organic pools such as wood, foliage, fine roots, and soil organic matter are critical sources of these scarce nutrients. Certain landscape features in selectively logged forests are more overtly impacted by harvest operations than others. Log decks and roads often undergo a complete removal of the surface root mat, litter layer, and soil organic matter, which are important nutrient storage pools for both rockderived nutrients and nitrogen. Selective logging operations often result in soil compaction, which can enhance reducing conditions in the log deck soils, resulting in higher pH and Ca and Mg concentrations [Olander et al., 2005]. These zones of changing redox can lead to changes in nitrogen oxide and methane production [Keller et al., 2005; see also Bustamante et al., this volume]. Under moderate logging intensities, no increases in soil nutrients are observed in the tree fall gaps despite the flush of nutrient-rich fresh foliage in the tree crown that is left behind after the bole wood is removed [Olander et al., 2005]. Calculations suggest that nutrient inputs from crown foliage in tree fall gaps are probably too small to detect against the background level of nutrients in the topsoils [Keller et al., 2004a]. The logging disturbances with the greatest spatial extent, skids and gaps, have the smallest immediate effect on soil nutrients, while those with the smallest spatial extent, roads and decks, have the largest impact. In the central Amazon, changes observed in soils 3-6 months after harvest were similar to those measured 16 years after logging, suggesting some interesting linkages between the mechanisms causing the immediate change and
35
those maintaining these changes over time [Olander et al., 2005]. The direct impacts on soil properties appear less important than the loss ofnutrients in bole wood in determining the sustainability of selective logging. 4. LOGGING AND DEFORESTATION 4.1. Converting Logged Forest to Clearings
Selective logging is one of many land uses on the actively developing frontiers in Brazilian Amazonia. Capital formed through selective logging may be invested in other economic activities such as cattle ranching or intensive agriculture [Schneider et al., 2000; Merry et ai., 2006]. Even for smallholders, revenues from logging may facilitate complete deforestation. This view of logging is juxtaposed with the view of selective logging as a potentially environmentally friendly land use alternative to deforestation. Along with capital formation, selective logging rapidly increases human access to the forests via logging roads [Souza et al., 2005]. While the causal relations between logging and clearing are still uncertain, the spatial association has been clearly demonstrated by Asner et al. [2006] through the analysis of high-resolutionr satellite images covering five states of Brazilian Amazonia from 1999 to 2004. They combined selective logging maps derived from the Carnegie Landsat Analysis System (CLAS) with deforestation maps from the INPE PRODES program, as well as geographic information systems layers showing road networks. Over more than 2,000,000 km2 of Amazonia, logging was most likely to occur within 10 km of paved roads (Figure 3). However, up to 30% of all mapped logging areas were 25-50 km from these arteries. Moreover, Asner et al. [2006] found that 16 ± 1% of selectively logged areas were deforested within 1 year of logging, with a subsequent annual deforestation rate of 5.4% for 4 years following timber harvests. The probability of deforestation for a logged forest was up to 4 times greater than for unlogged forests. During the early 2000s, selective logging was clearly a step toward full clear-cutting of the forest. 4.2. Forest Degradation and Fire Following Logging
Clear-cutting is not the only fate of selectively logged forests in Amazonia. As discussed, timber harvest operations result in a matrix of new roads throughout a forest, always leading back to more substantial paved road networks. These road networks increase general access to hunters, who may decrease faunal population densities, with cascading impacts on the forest food web [Peres, 2001; Peres et al.,
36 SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
ASNER ET AL.
37
0.90
.,'
0.80
"1 _2_8 "3"9 "6 Gap Class
7
_ _
0.70
2'!
or:(
'iii 0.60
'5
I-
....0 c
0.15
Logging Intact forest
I
;;;; 4 _10
5"0
0
:e0
CI.
e
II..
0.10
0.05
0.00 0-5
5-10
10-15 15·20 20-25 25-30 30-35 35-40 40-45 45·50
2000·2001
Distance Class from Major Roads (5 km Increments)
Figure 3. Proportion of total area oflogging or intact forest with distance from majorroads throughout the states ofPani, Mato Grosso, Rondonia, Acre, and Roraima in Brazilian Amazonia [Asner et al., 2006]. '
I
I
2006]. Canopy openings can also result in forest desiccation [Ray et aI., 2005], leading to increased fire susceptibility, as documented at local to regional levels [Cochrane et al., 1999; Nepstad et al., 1999; GelWing, 2002; Souza et al., 2003,2005]. The impacts of logging on forest fragmentation have recently been quantified over a four-state region of Brazilian Amazonia at a spatial resolution of 30 m [Broadbent et aI., 2008]. Annually from 1999 to 2002, deforestation and selective logging generated ~32,000 and 38,000 kIn of new forest edges, which equates to a growth rate in fragment edges of 0.8% and 3.1 %, respectively, for clear-cuttiJ.lg and selective logging. On the basis of the average published distance in which forest edge effects persist (100 m), Broadbent et al. [2008] estimated that deforestation resulted in an additional ~3000 kIn2 of "edge-affected" forest over the study period, whereas selective logging generated nearly 20,000 kIn2 of edge-affected forest, often deep into intact forest areas. From a fragmentation perspective, the high-impact logging practices of the recent past in Brazil have been a major force of forest fragmentation and change. In sum, a combination of logging, forest degradation, and deforestation is rendering Brazilian Amazonia as a new ecological landscape, with mosaics of cleared and forested land, the latter of which resides as a complex mosaic of secondary and logged primary forest [Fearnside and Guimariies, 1996;
Steininger, 1996; Cardille and Foley, 2003]. Just a fourstate area of forest affected by clearing and logging alone was enormous in 2000--2001 (Plate 5). Adding geographic detail on fires, degradation and hunting will be a challenging task, but advances in remote sensing are making it possible to map and monitor a range of changes in tropical forest environments (reviewed by Chambers et al. [2007]). Remote sensing research during LBA fueled many of the technique developments that make the mapping of selective logging more tractable today (Appendix A). 5. CONCLUDING REMARKS Selecti:ve logging is clearly an important form of land use in Amazonia, with the potential for sustainable forest use if these more careful methods of extraction and planning can be more widely adopted and enforced. From 2000 to 2005, about half ofthe selective logging in Brazilian Amazonia was illegal and hence unregulated [Barreto et al., 2006]. Moreover, at least through 2002, selective logging was regionally dominated by high-damage extraction operations that left the forest susceptible to drought and fire [Nepstad et al., 1999; GelWing, 2002]. Logging was often a precursor to deforestation [Asner et al., 2006] via the increased access to the forest that unofficial logging roads provide to inhabitants [Souza et al., 2005]. Furthermore, as of 2007, the total land under For-
Plate 5. Mosaic of selective logging (gap classes 1-10) and deforestation (D, blue), compiled by Asner et al. [2006]. Both the logging and deforestation cover the period 2000-2001. Gap classes 1-10 for logging indicate the area-integrated loss of canopy cover at l-km resolution. Class 1 is 0-10% canopy loss, class 2 is 11-20% canopy loss, etc. Gray polygons indicate federal conservation lands and indigenous reserves. This compilation was limited to the states of Para (PA), northern Mato Grosso (MT), Rondonia (RO), and Acre (AC) as indicated.
i est Stewardship Council green certification in Amazonia was merely 3.2 million ha (Forest Stewardship Council, Florestas Certificadas, 2005, available at http://www.fsc.org.br). which represents only 3% of total Amazonia timber production [Barreto et al., 2006]. Thus, in recent years, selective logging was not dominated by the kind of well-managed, lowimpact, green certification operations intended to preserve forest cover, structure, diversity, and function and to be an alternative to deforestation. Instead, it has been dominated by high-damage operations that leave the canopy in disarray, with cascading effects on the ecology ofthe region. This chapter has highlighted and synthesized much of what was learned about the ecological impacts of logging during the LBA program. Future studies will build upon these findings and hopefully will keep pace with the rapidly evolving timber industry policies in Brazil and other regions of the Amazon. APPENDIX A: REMOTE SENSING OF SELECTIVE LOGGING Detection of selective logging with moderate-resolution satellite imagery, such as Landsat thematic mapper (30-m pixel size), is challenged by the complex mixture of dead
and live vegetatio~, shadowing, and soils found throughout forest environments. From the satellite vantage point, forest damage caused by logging seems to disappear within 3 years or less, making detection of previously logged forest (> 1 year) very challenging [Stone and Lefebvre, 1998]. During LBA, remote sensing studies on logging in the Brazilian Amazon found that Landsat reflectance data have limited spectral resolution to detect logged forest from intact forest [Asner et al., 2002a; Souza et al., 2005]. Vegetation indices [Stone and Lefebvre, 1998; Souza et al., 2005; Broadbent et al., 2006] and texture filters [Asner et al., 2002a] also showed a limited capability for detection of logging. Improving the spatial resolution of reflectance data can help; Ito 4-m resolution IKONOS satellite data can readily detect forest canopy structure and canopy damage caused by selective logging [Asner et al., 2002b; Read et al., 2003; Souza et al., 2003]. However, the high cost of these images and additional computational challenges of extracting information severely limit the operational use of IKONOS and similar imagery. LBA research showed that the detection of logging at moderate spatial resolution is best accomplished at the subpixel scale using a technique called spectral mixture analysis
Table.At. Remote Sensing Techniques Applied to Selective Logging in the Br '1' A " Mappmg aZI Ian mazon
I:
Approach Studies yo I Isua Watrin and Rocha [1992J interpretation
I,
J:
.
~.
1 :1l J
Betection of logging landings and buffer
Decision tree
Change detection
CLASlite
NDFIand CCA
n
I L.
...I
r
orma Ize
Sensor Landsat TM
Stone and Lefebvre [1998J
LandsatTM
Matricardi et al. [2001J
LandsatTM
Santos et al. [2001J
LandsatTM
I ocal
Souza et al. [2003J
Souza et al. [2003J
SPOT-4
LandsatTM andETM+
LandsatTM
local
local
local
LandsatTM andETM+
http://claslite.ciw.edu
LandsatTM anYWhere andETM+, in the SPOT-415, world ASTER, ALI, and MODIS LandsatTM andETM+
map total logging area
Brazilian Amazon Brazilian Amazon local map total logging area (canopy dam age, clearings, and undamaged forest)
Asner et al. [2005, 2006J
Souza et al. [2005J
Objective
local
Souza and Barreto [2000J, LandsatTM Matricardi et al. [2001 J, andETM+ and Moilteiro et al. [2003J
Grar;a et al. [2005J Image segmentation
CLAS
Spatial Extent
five states of the Brazilian Amazon
local
map forest canopy damage associated with logging and burning map forest canopy damage associated with logging and burning
ASNER ET AL. i
Advantages
Disadvantages
It does not require sophisticated image processing techniques.
It is relatively simple to implement and satisfactory to estimate the total potential logging area. It has simple and intuitive classification rules.
It enhances forest canopy damaged areas.
It is laborintensive for large areas and may be user-biased to define the boundaries of logged forest.
Logging buffers vary across the landscape and do not reproduce the actual shape of the logged area. It has not been tested in very large areas, and classification rules may vary across the landscape. It requires two pairs of images and does not separate natural and anthropogenic forest changes. It has not been tested in very large areas, and segmentation rules may vary across the landscape.
map total It is relatively logging area simple to (canopy damage, implement and clearings, and satisfactorily undamaged estimate the total forest) logging area. Free software is available. It is highly It requires high automated and computation standardized to power and pairs very large areas. of images to detect forest change. It is highly It is limited to automated, run tropical forests. on a standard desktop computer, requires minimal training. It enhances forest canopy damaged areas.
39
It has not been tested in very large areas and does not separate
d d'ffi ' ,ance ematlc Ma PI C I erence fraction index; and CCA Contextual Clas'fi t' APpe~ us; LAS, Carnegie Landsat Analysis System' NDFI , Sl ca IOn Igonthm. ' ,
(SMA). Images obtained with SMA that show detailed fractional cover of soils, nbnphotosynthetic vegetation (NPV), and green vegetatiorr'tmhance our ability to detect logging infrastructure andjanopy damage. For example, soil fractional cover map~erived from SMA can enhance the detection oflog decks and roads [Souza and Barreto, 2000]; maps of NPV fraction enhance the expression of damaged and dead vegetation [Souza et al., 2003]; and the green vegetation fractional cover is sensitive to canopy openings [Asner et al., 2004a]. Several mapping techniques were tested and applied in local to large regional-scale studies of selective logging in Brazil (Table AI). These techniques vary in terms of the mapping goals, the approach and geographic extent, and reported accuracies. In terms of mapping goals, some techniques were developed to map the total potentially logged area, which includes forest canopy damaged and forest clearings and undamaged forest islands, while others were focused only on the mapping of forest canopy damage. The former group of techniques included visual interpretation of Landsat images [e.g., Stone and Lefebvre, 1998; Matricardi et at., 2001], manual and automated detections of)og decks with an estimated timber-harvesting buffer around the decks [Souza and Barreto, 2000], and highly automated SMA approaches combined with pattern recognition methods [Souza et at., 2005; Asner et at., 2005, 2006]. Future studies will likely focus on techniques that balance issues of satellite image quality, availability and cost, processing time, and the level of expertise required to produce verifiable maps of selective logging. LBA research paved the way for these current and future developments. AclO1owledgments. We thank the many individuals and agencies from Brazil, the United States, and elsewhere for years of financial, logistical, and scientific support required to develop an understanding of land use change and logging practices in the Amazon region. This work was supported by the NASA LBA-ECO program and the Gordon and Betty Moore Foundation.
REFERENCES Alder, D., and 1. N. M. Silva (2000), An empirical cohort model for management of terra fume forests in the Brazilian Amazon, For. Ecol. Manage., 130,141-157. Asner; G. P., M. Keller, R. Pereira, and 1. Zweede (2002a), Remote sensing of selective logging in Amazonia: Assessing limitations based on detailed field measurements, Landsat ETM+ and textural analysis, Remote Sens. Environ., 80, 483--496. Asner, G. P., M. Palace, M. Keller, R. Pereira Jr., J. N. M. Silva, and J. C. Zweede (2002b), Estimating canopy structure in an Amazon forest from laser rangefinder and IKONOS satellite observations, Biotropica, 34, 483--492.
Asner, G. P., M. Keller, R. Pereira, 1. C. Zweede, and 1. N. M. Silva (2004a), Canopy damage and recovery following selective logging in an Amazon forest: Integrating field and satellite studies, Ecol. Appl., 14,280-298. Asner, G. P., M. Keller, andJ. N. M. Silva (2004b), Spatial and temporal dynamics of forest canopy gaps following selective logging in the eastern Amazon, Global Change Bioi., 10, 765-783. Asner, G. P., D. E. Knapp, E. N. Broadbent, P. J. C. Oliveira, M. Keller, and 1. N. M. Silva (2005), Selective logging in the Brazilian Amazon, Science, 310,480--482. Asner, G. P., E. N. Broadbent, P. J. C. Oliveira, D. E. Knapp, M. Keller, and J. N. Silva (2006), Condition and fate oflogged forests in the Brazilian Amazon, Proc. Natl. Acad. Sci. U S. A., 103, 12,947-12,950. Barreto, P., C. Souza Jr., R. Noguer6n, A. Anderson, and R. Salomiio (2006), Human Pressure on the Brazilian Amazon Forests, 84 pp., World Resour. Inst., Washington, D. C. Barros, A. C., and C. Uhl (1995), Logging along the Amazon River and estuary: Patterns, problems and potential, For. Ecol. Manage., 77,87-105. Binswanger, H. P. (1991), Brazilian policies that encourage deforestation in the Amazon, World Dev., 19, 821-829. Boltz, F., D. R. Carter, T. P. Holmes, and R. Pereira Jr. (2001), Financial returns under uncertainty for conventional and reducedimpact logging in permanent production forests of the Brazilian Amazon, Ecol. Econ., 39, 387-398. Brienen, R. J. W., and P. A. Zuidema (2007), Incorporating persistent tree growth differences increases estimates of tropical timber yield, Front. Ecol. iEnviron., 5, 302-306. Broadbent, E. N., D:. 1. Zarin, G. P. Asner, M. Pena-Claros, A. Cooper, and R. Littell (2006), Forest structure and spectral properties after selectiVe logging in Bolivia, Ecol. Appl., 16, 11481163. Broadbent, E. N., G. P. Asner, M. Keller, D. E. Knapp, P. J. C. Oliveira, and J. N. Silva (2008), Forest fragmentation and edge effects from deforestation and selective logging in the Brazilian Amazon, Bioi. Conserv., 141, 1745-1757, doi:l0.10161 j .biocon.2008 .04.024. . Brouwer, L. C. (1996), Nutrient Cycling in Pristine and Logged Tropical Rain Forest, Tropenbos-Guyana Ser., vol. 1, TropenbosGuyana Programme, Georgetown, Guyana. Browder, J. O. (1988), Public policy and deforestation in the Brazilian Amazon, in Public Policies and the Misuse of the Forest Resource, edited by R. Repetto and M. Gillis, pp. 247-298, Cambridge Univ. Press, Cambridge, U. K. Bunker, D. E., F. DeClerck, 1. C. Bradford, R. K. Colwell, 1. Perfecto, O. L. Phillips, M. Sankaran, and S. Naeem (2005), Species loss and aboveground carbon storage in a tropical forest, Science, 310,1029-1031. Bustamante, M. M. C., M. Keller, andD. A. da Silva (2009), Sources and sinks of trace gases in Amazonia and the cerrado, Geophys. Monogr. Ser., doi: 1O.l029/2008GM000733, this volume. Cardille, J. A., and J. A. Foley (2003), Agricultural land-use change in Brazilian Amazonia between 1980 and 1995: Evidence from integrated satellite and census data, Remote Sens. Environ., 87, 551-562.
40
SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
ASNERET AL.
41
I
... I
..1 .~
,
"
J. '
Chambers, J. Q., N. Higuchi, J. P. Schimel, L. V. Ferreira, and J. M. Melack (2000), Decomposition and carbon cycling of dead trees in tropical forests of the central Amazon, Oecologia, 122, 380-388..,. Chambers, J. Q., G. P. Asner, D. C. Morton, L. O. Anderson, S. S. Saatchi, F. D. B. Espirito-Santo, M. Palace, and C. Souza Jr. (2007), Regional ecosystem structure and function: Ecological insights from remote sensing of tropical forests, Trends Ecol. Evol.:22,414-423. Chazdon, R. L., and N. Fetcher ~1984), Photosynthetic light environments in a lowland tropical rain forest in Costa Rica, J Ecol., 72,553-564. Cochrane, M. A, A Alencar, M. D. Schulze, C. M. Souza, D. C. Nepstad, P. Lefebvre, and E. A Davidson (1999), Positive feedbacks in the fire dynamic of closed canopy tropical forests, Science,284,1832-1835. Dauber, E., T. S. Fredericksen, and M. Pena-Claros (2005), Sustainability of timber harvesting in Bolivian tropical forests, For. Ecol. Manage., 214, 294-304. Fearnside, P. M., and W. M. Guimaraes (1996), Carbon uptake by secondary forests in Brazilian Amazonia, For. Ecol. Manage., 80,35-46. Feldpausch, T. R, S. Jirka, C. A M. Passos, F. Jasper, and S. Riha (2005), When big trees fall: Damage and carbon export by reduced impact logging in southern Amazonia, For. Ecol. Manage., 219,199-215. Field, C. B., J. T. Randerson, and C. M. Malmstrom (1995), Global net primary production: Combining ecology and remote sensing, Remote Sens. Environ., 51, 74-88. Fimbel, R A., A Grajal, and J. G. Robinson (Eds.) (2001), The Cutting Edge: Conserving Wildlife in Logged Tropical Forests, 700 pp., Columbia Univ. Press, New York. Gerwing, J. (2002), Degradation offorests through logging and fire in the eastern Brazilian Amazon, For. Ecol. Manage., 157, 131-141. Gra9a, P. M., L. A Santos, J. R. Soares, and J. V. Souza (2005), Desenvolvimento metodol6gico para detec91io e mapeamento de areas florestais sob explora9ao madeireira: Estudo de caso, regiao norte do Mato Grosso, in XII Simposio Brasileiro de Sensoriamento Remoto, pp. 1555-1562, Inst. Nac. Pesqui. de Espaciais, Sao Jose dos Campos, Brazil. Healey, J. R, C. Price, and J. Tay (2000), The 'cost of carbon retention by reduced impact logging, For. Ecol. Manage., 139, 237-255. Hill, J. K, K C. Hamer, L. A. Lace, and W. M. T. Banham (1995), Effects of selective logging on tropical forest butterflies on Buru, Indonesia, J App!. Eeal., 32, 754-760. Holmes, T. P., G. M. Blate, J. C. Zweede, R Pereira Jr., P. B. Barreto, F. Boltz, and R. Bauch (2002), Financial and ecological indicators of reduced impact logging performance in the eastern Amazon, For. Ecol. Manage., 163, 93-110. Huang, M., G. P. Asner, M. Keller, and J. A Berry (2008), An ecosystem modelfor tropical forest disturbance and selective logging, J Geophys. Res., 113, G01002, doi:l0.l029/2007JG000438. Johns, AD. (1992), Vertebrate responses to selective logging: Implications for the design of logging systems, Philos. Trans. R. Soc. London, Ser. B, 335,437-442.
Johns, J. S., P. Barreto, and C. Ubi (1996), Logging damage during planned and unplanned logging operations in the eastern Amazon, For. Eco!. Manage., 89, 59-77. Keller, M., and W. A Reiners (1994), Soil-atmosphere exchange of nitrous oxide, nitric oxide, and methane under secondary succession of pasture to forest in the Atlantic lowlands of Costa Rica, Global Biogeochem. Cycles, 8, 399-409. Keller, M., M. Palace, and G. Hurtt (2001), Biomass estimation in the Tapajos National Forest, Brazil-Examination of sampling and allometric uncertainties, For. Ecol. Manage., 154, 371-382. Keller, M., G. P. Asner, J. N. M. Silva, and M. Palace (2004a), Sustainability of selective logging of upland forests in the Brazilian Amazon: Carbon budgets and remote sensing as tools for evaluation oflogging effects, in Working Forests in the Tropics: Conservation through Sustainable Management?, edited by D. Zarin et aI., pp. 41-63, Columbia Univ. Press, New York. Keller, M., M. Palace, G. P. Asner, R. Pereira, and J. N. M. Silva (2004b), Coarse woody debris in undisturbed and logged forests in the eastern Brazilian Amazon, Global Change BioI., 10, 784-795. Keller, M., R K Varner, J. Dias, H. Silva, P. Crill, R de Oliveira Jr., and G. P. Asner (2005), Soil-atmosphere exchange of nitrous oxide, nitric oxide, methane, and carbon dioxide in logged and undisturbed forest in the Tapajos National Forest, Brazil, Earth Interact., 9(23), E1I25, doi: 10. 1175/E1I25. 1. Lee, D. W. (1987), The spectral distribution of radiation in two neotropical rainforests, Biotropica, 19, 161-166. Lentini, M., D. Pereira, D. Celentano, and R. Pereira (2005), Fatos Florestais da Amazonia, Inst. do Homem e Meio Ambiente da Amazonia, Belem, Brazil. Matricardi, E. AT., D. L. Skole, M. A Chomentowski, and M. A Cochrane (2001), Multi-temporal detection of selective logging in the Amazon using remote sensing, Spec. Rep. BSRSI Res. Adv. RA03-01\w, 27 pp., Trop. For. Inf. Cent., Mich. State Univ., East Lansing. McNabb, K L., M. S. Miller, B. G. Lockaby, B. J. Stokes, R G. Clawson, J. A Stanturf, and J. N. M. Silva (1997), Selection harvests in Amazonian rainforests: Long-term impacts on soil properties, For. Ecol. Manage., 93,153-160. Merry, F., G. Amacher, D. Nepstad, P. Lefebvre, E. Lima, and S. Bauch (2006),Industrial development on logging frontiers in the Brazilian Amazon, Int. J Sustain. Dev., 9, 277-296. Monteiro; A L., C. M. Souza Jr., and P. Barreto (2003), Detection of logging in Amazonian transition forests using spectral mixture models, Int. J Remote Sens., 24,151-159. Monteith, J. L. (1972), Solar radiation and productivity in tropical ecosystems, J Appl. Ecol., 9, 747~766. Nepstad, D. C., et al. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505-508. Olander, L. 0., M. M. Bustamante, G. P. Asner, E. Telles, Z. Prado, and B. P. Camargo (2005), Surface soil changes following selective logging in an eastern Amazon forest, Earth Interact., 9(4), E1I35, doi:l0.l175/E1I35.1. Palace, M., M. Keller, G. P. Asner, J. N. M. Silva, and C. Passos (2007), Necromass in undisturbed and logged forests in the Brazilian Amazon, For. Ecol. Manage., 238, 309-318.
Pereira, R., Jr., J. Zweede, G. P. Asner, and M. Keller (2002), Forest canopy damage and recovery in reduced-impact and conventional selective loggiflg in eastern Para, Brazil, For. Eco!. Manage., 168, 77-89.Y Peres, C. A. (2001)/ Synergistic effects of subsistence hunting and habitat fragmed'~tion oil Amazonian forest vertebrates, Conserv. BioI., 15,1490-1505. Peres, C. A, J. Barlow, and W. F. Laurance (2006), Detecting anthropogenic disturbance in tropical forest, Trends Ecol. Evo!., 21,227-229. Pinard, M. A, and W. P. Cropper (2000), Simulated effects oflogging on carbon storage in dipterocarp forest, J App!. Ecol., 37, 267-283. Pinedo-Vasquez, M., D. J. Zarin, K Coffey, C. Padoch, and F. Rabelo (2001), Post-boom logging in Amazonia, Hum. BioI., 29, 219-239. Putz, F. E., D. P. Dykstra, and R. Heinrich (2000), Why poor logging practices persist in the tropics, Conserv. BioI., 14, 951956. Rankin, J. M. (1985), Forestry in the Brazilian Amazon, in Amazonia, edited by G. Prance and T. Lovejoy, pp. 369-392, Pergamon, Oxford, U. K Ray, D., D. Nepstad, and P. Moutinho (2005), Micrometeorological and canopy controls of fire susceptibility in a fOl:ested Amazon landscape, Ecol. Appl., 15,1664-1678. Read, J. M., D. B. Clark, E. M. Venticinque, and M. P. Moreira (2003), Application of merged I-m and 4-m resolution satellite data to research and management in tropical forests, J Appl. Ecol., 40, 592-600. Sanchez, P. A (1976), Properties and Management ofSoils in the Tropics, 235 pp., John Wiley, New York. Santos, J. R., T. Krug, L. S. Araujo, L. G. Meira Filho, and C. A Almeida (2001), Dados multitemporais TMlLandsat aplicados ao estudo da dinamica de explora91io madeireira na Amazonia, inX Simposio Brasileiro de Sensoriamento Remoto, pp. 17511755, Inst. Nac. Pesqui. de Espaciais, Sao Jose dos Campos, Brazil. Schlesinger, W. H. (1997), Biogeochemistry: An Analysis of Global Change, 2nd ed., 588 pp., Academic, San Diego, Calif. Schneider, R, E. Arima, A Verissimo, P. Barreto, and C. Souza Jr. (2000), Sustainable Amazon: Limitations and Opportunities for Rural Development, Inst. do Homem e Meio Ambiente da Amazonia, Brasilia. Scholz, I. (2000), Overexploitation or Sustainable Management: Action Patterns of the Tropical Timber Industly: The Case of Para, Brazil, 1960-1997,441 pp., Frank Cass, London. Schulze, M., and J. Zweede (2006), Canopy dynamics in unlogged and logged forest stands in the eastern Amazon, For. Ecol. Manage., 236, 56-64. Silva, J. N. M., J. O. P. de Carvalho, J. Lopes, B. F. de Almeida, D. H. M. Costa, L. C. de Oliveira, J. K Vanclay, and J. P. Skovs~ gaard (1995), Growth and yield of a tropical rain forest in the Brazilian Amazon 13 years after logging, For. Ecol. Manage., 71,267-274. Silver, W. L., J. Neff, M. McGroddy, E. Veldkamp, M. Keller, and R Cosme (2000), Effects of soil texture on belowground carbon
and nutrient storage in a lowland Amazonian forest ecosystem, Ecosystems, 3,193-209. Silver, W. L., A W. Thompson, M. E. McGroddy, R. K. Varner, J. D. Dias, H. Silva, P. M. Crill, and M. Keller (2005), Fine root dynamics and trace gas fluxes in two lowland tropical forest soils, Global Change BioI., 11, 290-306. Sist, P. (2000), Reduced-impact logging in the tropics: Objectives, principles and impacts, Int. For. Rev., 2, 255-263 . Souza, C., and P. Barreto (2000), An alternative approach for detecting and monitoring selectively logged forests in the Amazon, Int. J Remote Sens., 21,173-179. Souza, C., L. A Firestone, L. Moreira, and D. A Roberts (2003), Mapping forest degradation in the eastern Amazon from SPOT 4 through spectral mixture models, Remote Sens. Environ., 87, 494-506. Souza, C. M., D. A Roberts, and M. A. Cochrane (2005), Combining spectral and spatial information to map canopy damage from selective logging and forest fires, Remote Sens. Environ., 98, 329-343. Steininger, M. K (1996), Tropical secondary forest regrowth in the Amazon: Age, area, and change estimation with thematic mapper data, Int. J Remote Sens., 17, 9-27. Stewart, J. W. B., and H. Tiessen (1987), Dynamics of soil organic phosphorus, Biogeochemistry, 4, 41-60. Stone, S. W. (l998a), Evolution of the timber industry along an aging frontier: The case of Paragominas (1990-95), World Dev., 26,433-448. Stone, S. W. (1998b), Using a geographic information system for applied policy analysis: The case of logging in the eastern Amazon, Ecol. Econ., p7, 43-61. Stone, T. A, and P. Lefebvre (1998), Using multi-temporal satellite data to evaluate selective logging in Para, Brazil, Int. J Remote Sens., 19,2517-2526. Thiollay, J. M. (1992), Influence of selective logging on bird species diversity in a Guiana rain forest, Conserv. Bioi., 6, 47-63. Ubi, C., P. Barreto, A Verissimo, E. Vidal, P. Amaral, A C. Barros, C. Souza, J. Johns, and J. Geiwing (1997), Natural resource management in the Brazilian Amazon, BioScience, 47,160-168. Varner, R K, M. Keller, J. R Robertson, J. D. Dias, H. Silva, P. M. Crill, M. McGroddy, and W. L. Silver (2003), Experimentally induced root mortality increased nitrous oxide emission from tropical forest soils, Geophys. Res. Lett., 30(3), 1144, doi: 10.1 029/2002GLO16164. Verissirno, A., P. Barreto, M. Mattos, R. Tarifa, and C. Ubi (1992), Logging impacts and prospects for sustainable forest management in an old Amazonian frontier: The case of Paragominas, For. Ecol. Manage., 55,169-199. Verissimo, A, C. S. Junior, S. Stone, and C. Ubi (1998), Zoning of timber extraction in the Brazilian Amazon, Conserv. Bioi., 12,128-136. Verissimo, A., C. Souza, and P. Amaral (2000), Identifica9ao de .Areas com Potencial para a Cria9ao de Florestas Nacionais na Amazonia Legal, report, 36 pp., Braz. Minist. of Environ., Brasilia. (Available at http://www.imazon.org.br/upload/im_ livros_OIO.zip)
42
I,
.. .Ii
SELECTIVE LOGGING AND ITS RELATION TO DEFORESTATION
Verissimo, A., E. Lima, and M. Lentini (2002), Palos Madeireiros do Estado do Pani, report, 72 pp., Inst. do Homem e Meio Ambiente da Amazonia, Belem, Brazil. (Available at http://www. imazon.oJg.br/publicacoes/publicacao.asp?id=Ill) Vitousek, P. M., and R. L. Sanford Jr. (1986), Nutrient cycling in moist tropical forest, Annu. Rev. Eco!. Syst., 17, 137-167. Wadsworth, F. H., and J. C. Zweede (2006), Liberation: Acceptable production of tropical forest timber, For. Ecol. Manage., 233,45-51. Watrin, O. S., and A. M. A. Rocha (1992), Levantamento da vegeta<;ao natural e do uso da terra no Municipio de Paragominas (PA) utilizando imagens TMlLandsat, Belem, Bo!. Pesqui. EMBRAPA, Cent. TecnolAgri. Aliment., 124, 40-80. Zarin, D., V. F. G. Pereira, H. Raffles, F. G. Rabelo, M. PinedoVasquez, and R. G. CongaIton (2001), Landscape change in tidal floodplain near the mouth of the Amazon River, For. Eco!. Manage., 154, 383-393.
Zarin, D., M. D. Schulze, E. Vidal, and M. Lentini (2007), Beyond reaping the first harvest: Management objectives for timber production in the Brazilian Amazon, Conserv. BioI., 21, 916-925.
J G. P. Asner, Department of Global Ecology, Carnegie Institution, 260 Panama Street, Stanford, CA 94305, USA. (gpa@ stanford.edu) M. Keller, NEON, Inc., 3223 Arapahoe Avenue, Boulder, CO 80303, USA. ([email protected]) M. Lentini, Instituto Floresta Tropical, Belem 66025-660, Brazil. ([email protected]) F. Merry, Woods Hole Research Center, Falmouth, MA 025401644, USA. ([email protected]) C. Souza Jr., Instituto do Homem e Meio Ambiente da Amazonia, Belem 66060-160, Brazil. ([email protected])
The Spatial Distribution and Interannual Variability of Fire in Amazonia Wilfrid Schroeder, I Ane Alencar,2 Eugenio Arima,3 and Alberto Setzer4 Charcoal evidence suggests that fires in Amazonian forests were an infrequent agent of forest disturbance prior to the twentieth century. However, the spatial and temporal distribution of fires changed dramatically during the past few decades. Fire has become one of the driving forces of land use and land cover change in Amazonia. Increasing human intervention in the region, in conjunction with climate anomalies, has exposed tropical forests to an unprecedented amount of vegetation fires with important consequences to the functioning of the complex Amazonian system and atmospheric concentrations of greenhouse gases. In this chapter, the main fire types in Amazonia are discussed: deforestation, maintenance, accidental, and natural fires. The major causes and consequences of vegetation fires are analyzed in light of their social, economic, and biophysical drivers. Satellite data are used to derive current maps describing the spatial and temporal distribution of fires in the region, highlighting some of the important linkages between human activities and climate conditions that combine to create unique anthropogenic fire regimes across Amazonia.
1. INTRODUCTION
1990; Ray et al., 2005; Uhl and Kaufmann, 1990]. However, since prehistorical times, humans have learned to manipulate fire and use it as a major hunting weapon and agricultural tool [Goudsblom, 1992]. The more recent history offire occurrence in Amazonia is marked by a contrast between low frequency natural fires and the growing dominance of anthropogenic fires as human occupation in the region has increased. In the past few decades, droughts related to El Nino-Southern Oscillation (ENSO) episodes, combined with the encroachment ofhuman settlements in the region and the development of transportation infrastructure have transformed fire into a major environmental threat to the Amazonian ecosystem and regional climate [Cochrane et al., 1999; Nepstad et al., 1999a; Alencar et al., 2004]. In the past, intensity and frequency offires were not severe enough to change the ecosystem, but nowadays, humans have transformed fire into a chronic, persistent element ofthe local landscape. In the Brazilian part of Amazonia alone, fire is currently the primary land clearing and management approach for an estimated four million farmers [Nepstad et al., 1999b]. The occurrence of major destructive fire seasons is no longer constrained to
Amazonian forests have long been disturbed by fires [Meggers, 1994]. Geological data provide evidence of charcoal deposits in soils of mature forests in the Amazon basin indicating historical, however infrequent, fire activity in the region [Sanford et al., 1985; Meggers, 1994]. The low historical fire frequency is largely explained by the high humidity and rainfall levels that characterize the region and which often prevent natural fires from developing [Goldammer, lEarth System Science Interdisciplinary Center, University of Maryland, College Park, Maryland, USA. 2Instituto de Pesquisa Ambiental da Amazonia, Brasilia, Brazil. 3Department of Environmental Studies, Hobart and William Smith Colleges, Geneva, New York, USA. 4Centro de Previsao do Tempo e Estudos ClimMicos, Instituto. Nacional de Pesquisas Espaciais, Sao Jose dos Campos, Brazil.
,
I
j
.L.. i
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1 029/2008GM000724 43
44
SCHROEDER ET AL.
SPATIAL DISTRIBUTION AND INTERANNUAL VARlABILITY OF FIRE IN AMAZONIA
45
I
Jl 1.
II
~
1
ENSO years and, combined with climate change, may accelerate the conversion of the rainforest into savanna-like vegetation [Nepstad et al., 2004). Vegetati~n fires in Amazonia have been monitored routinely since the mid-1980s with the use of satellites [Setzer and Pereira, 1991]. Satellite data are particularly useful for monitoring fires in Amazonia as they provide systematic information on fire activity for the entire region, including the most remote areas where ground-based observations are difficult. The number of operahng satellite systems with fire monitoring capabilities has significantly increased in the past decade, thereby augmenting our capacity to delineate the spatial and temporal patterns of fire distribution in the region. In this chapter, we describe the spatial and temporal distribution of vegetation fires with a focus on Brazilian Amazonia. We start with a discussion of the fundamental causes (section 1) and consequences (section 2) associated with the use of fire to promote land use and land cover change. In section 3, we present a brief overview of the main satellite systems and products which are available to monitor and evaluate fire activity in the region in order to (in sections 4 and 5) explore the spatial and temporal distributions of fire detections which can be derived using those products. 2. VEGETATION FIRES IN AMAZONIA: MAJOR CAUSES Vegetation fires vary according to cause, intensity, duration, and spatial pattern. In order to properly address the subject, it becomes necessary to draw a distinction between the different types of fire based on their physical properties, while incorporating the aspects and the implications of the policies designed to control them [Alencar et al., 1997; Nepstad et ai., 1999b]. Fires in Amazonia can be classified into four major groups.
2.1. Deforestation Fires These are intentional fires used to facilitate land clearing for forest conversion into crop production or pasture in the initial stages of frontier occupation and deforestation [Nepstad et ai., 1999b]. Typically, forests are cut down in the first months of the dry season, and the slashed biomass is left to dry under the Sun for 2 to 3 months, depending on the biomass volume, initial moisture content, and weather conditions [Sorrensen, 2004]. Fire is used as a cost effective technology to provide rapid transformation of the dried organic matter into short-lived fertilizing ash. This method is utilized in the deforestation process by small subsistence farmers, as well as by large-scale mechanized agriculture and cattle ranchers I-
...!
alike. The same process is used to convert young and advanced succession forests, locally known as "Capoeinis," in the shifting cultivation process used by an estimated half million small agricultural households in Brazilian Amazonia [Sorrensen, 2004). A survey conducted in five regions along the Arc of Deforestation between 1995 and 1996 suggested that approximately 13% ofthe area burned was due to deforestation fires [Alencar et al., 1997; Nepstad et al., 1999b]. The total annual deforested area estimated for Brazilian Legal Amazonia by the PRODES project in the same period peaked at 29,000 km2 • Deforestation rates averaged approximately 22,000 km2 a-I from 2000 to 2005 [INPE, 2005].
2.2. Maintenance Fires These are also intentional fires used in the management of pastures as well as for clearing crop residue. Pastures in Amazonia, particularly those recently established, are frequently invaded by pests, weeds, and other competing natural vegetation. Moreover, within a few years of planting, pastures lose vigor as soil fertility declines [Uhl and Buschbacher, 1985]. Declines in pasture productivity vary tremendously from a few years to more than a decade depending on the cattle stocking rate, grass species, and management practices such as rotation and control of erosion and leaching, etc. Cattle also prefer newer, tender sprouts as opposed to old unpalatable grasses. Hence, fire is used as an inexpensive means to control weeds, to restore part of the soil fertility, and to rejuvenate grasses. Unlike deforestation fires, maintenance fires are rapid and low in intensity due to the reduced amount of biomass fuel. It is suggested that this type of fire affected an area of 20% of the total area burned along the Arc of Deforestation during 1995-1996 [Alencar et al., 1997; Nepstad et al., 1999b].
2.3. Accidental Fires These are fires that escape control from intentional burning associated with nearby maintenance or conversion fires. Accidental fires are known to affect forest areas as well as rural properties in already deforested zones. In the case when forests are accidentally burned, the problem will normally be concentrated along forest edges in areas of active deforestation and land use [Nepstad et al., 1999b; Gascon et al., 2000; Cochrane and Laurance, 2002). Nevertheless, this type of fire may impact large regions during exceptionally dry years when fire lines can penetrate the forests and affect areas far away from their ignition sources [Elvidge et ai., 2001; Nelson, 2001]. Charcoal pits are another important ignition source of accidental fires, especially in Eastern Amazonia [Alencar et al., 2004]. Fires which accidentally
burn forests will possess different characteristics depending on the degree of alter$on of the affected areas. In relatively intact forests, fires ale low in intensity, move very slowly, and tend not to spJjad to large areas [Cochrane and Schulze, 1998; Cochrane.fi003). On the other hand, forests disturbed by logging or previous fires are much more prone to subsequent, long-lasting, intense fires that can bum extensive areas [Nepstad et' al., 1999a]. Fragmented forests are more susceptible to fires because of the larger amount of available dry matter and canopy openness to air Currents and winds which help feed the fire lines [Cochrane and Schulze, 1999; Cochrane, 2003; Alencar et ai., 2004]. The expansion ofeconomic activities and the increasing intensity and frequency of ENSO events may promote a future of more frequent and larger forest fires in the region [Nepstad et ai., 1999a). Alencar et al. [2006] suggest that forest fires during ENSO years can bum an area two times larger than that resulting from deforestation. Accidental fires affecting rural properties in already deforested areas can also cause significant damage to crops, plantations, pastures, and infrastructure, resulting in great economic losses [Alencar et al., 1997; Mendonr;a et al., 2004]. According to a survey performed over,five study sites along the Arc of Deforestation in 1996 [Nepstad et al., 1999b], those types of escaped fires were responsible for 47% of the area burned in that period, which represented an average rainfall year.
2.4. Natural Fires Natural fires are those caused by lightning strikes. Other natural causes include friction fires sparked by falling rocks, and landslides, volcanic fires, and prism fires caused by the Sun's light beams deflected by crystal rocks [Stott, 2000]. Those types of fires are much rarer than lightning fires and, to our knowledge, no case has ever been documented in Amazonia. Although lightning strikes along the intertropical convergence zone (ITCZ) are very common [Stott, 2000], quantification of natural fire events in Amazonia is difficult due to limited data. Anecdotal reports suggest, however, that natural fires are rather infrequent in the region as lightning is often accompanied by rain, which extinguishes the initial ignition and prevents flame propagation [Ramos-Neto and Pivel!o, 2000; Stott, 2000]. The litter material must also be dry and arranged properly to bum. Moreover, not all striking episodes have high amperage and low voltage necessary to convert the electrical charge into fire. In some cases, the lightning strike will be "cold" and blast without producing fire [Pyne, 2001]. In fact, the vast majority of fire events in Amazonia are caused by intentional or unintentional human action and very few can be attributed to natural causes [Goldammer, 1990).
Fire events detected by satellites are spatially concentrated on or near deforested fields [Cochrane, 2001; Cochrane and Laurance, 2002; Alencar et al., 2004, 2006]. The different types of fires described above may be influenced by social, economic, and political factors, as well as by biophysical conditions, resulting in distinct spatial and temporal patterns of fire activity across the region (see maps and description in sections 3-5 below). For instance, Alencar et al. [2004] found that accidental forest understory fires during ENSO and non-ENSO years are strongly correlated with distance to main roads, charcoal pits, and settlements. Arima et ai. [2007] showed that the probability of fire is positively correlated with the farmgate price of beef and soybean, even when controlled for the amount of rainfall and different soil types. Higher farmgate prices provide an economic incentive for the conversion of forests into agricultural land and consequently to the use of fire as a management tool. Cultural factors also help explain why certain areas are more fire prone than others. Simmons et al. [2004] suggested that the cohesiveness and identity of communities can influence the likelihood of accidental fires. Moran et al. [2006] suggest that communities that practice slash and bum agriculture have their own ways to cope with fire, particularly during ENSO years. Thus, more traditional communities tend to use their empirical knowledge to prevent escaped fires, in contrast to ~ewly formed ones. However, the uncontrolled fires ignite4 by humans during 2005 in Acre, which affected an area of approximately 300,000 ha of forests, indicate that an intense drought can foster fire tragedies anywhere in Amazonia. In terms of institutional factors, Arima et al. [2007] estimated that areas protected by the federal government, such as indigenous lands and conservation areas, reduced the probability of fire by 33% on average, keeping rainfall, and distance to deforestation and infrastructure constant. Nepstad et al. [2006a] showed that even inhabited reserves such as indigenous lands and extractive reserves successfully prevent fire. On average, fire occurrence outside those areas was four times higher (see also Bruner et al. [2001] for a discussion of the effectiveness of protected areas in tropical regions). Biophysical factors, particularly rainfall levels and water holding capacity of soils, also affect the likelihood of fires [Nepstad et al., 2004]. Arima et al. [2007] showed that the probability offire in Brazilian Amazonia decreased on average from 10% to virtually zero when rainfall increased from 1400 to 3000 mm a-I even controlling for distances to deforested areas and to infrastructure. Soil water holding capacity is also critical to fires particularly during severe droughts in El·Nmo years. For instance, Nepstad et al. [2001] estimated that nearly 1 million km2 of forests had become vulnerable to fire during the 1997-1998 El Nino because the available soil
46
SPATIAL DISTRIBUTION AND INTERANNUAL VARIABILITY OF FIRE IN AMAZONIA
SCHROEDER ET AL.
47
\
-
water to plants was depleted up to 10 m in depth. The deep roots ofAmazonian forests are giant pumps that extract water from the soil up to 18 m deep maintaining a humid forest understory tiuring the 3- to 4-month dry season [Nepstad et al., 1994], thereby reducing the probability offire spread.
~,
3. CONSEQUENCES OF VEGETATION FIRES IN AMAZONIA Fire has been one of the'most important agents of landscape transformation in Amazonia. In rural Amazonia, fire assures initial productivity ofthe recently deforested areas in the absence of technological alternatives and soil correction [Sorrensen, 2004]. In addition, fire is also the most affordable and reliable management tool used to control weeds, favoring grass species used for cattle grazing. Despite the important role of anthropogenic fire in the Amazonian production system, its intensive and uncontrolled use can have major consequences to the region's ecosystem as well as to its people. These consequences include impacts on ecological and biophysical processes, regional and local economies, and impacts on health and societal behavior. 3.1. Ecological and Biophysical Consequences Fires can affect ecological and biophysical processes at different scales. At a local or stand scale, the ecological consequences ofuncontrolled and more frequent fires in tropical forests include, but are not limited to (1) increased vulnerability of forests to recurrent fires [Nepstad et al., 1995; Cochrane and Schulze, 1999]; (2) changes in biodiversity including large-scale tree mortality [Barlow et al., 2003; Holdsworth and Uhl, 1997; Barbosa and Fearnside, 1999; Cochrane and Schulze, 1999; Gerwing, 2002; Haugaasen et al., 2003], changes in forest composition and fruit availability, and impacts on faunal populations [Barlow et al., 2002; Barlow and Peres, 2004a, 2006; Peres et al., 2003]; (3) changes to soil nutrient availability influencing vegetation recovery in areas of secondary forest regrowth [Bushbacher et al., 1988; Hughes et al., 2000; Moran et al., 2000; Zarin et al., 2005]. Although fires occur at the landscape scale, the increase offire activity in Amazonia can have major consequences to the regional and global climate as well [Nobre et al., 1991; Rosenfeld, 1999; Andreae et al., 2004; Artaxo et al., 2005]. In terms of local impacts, forest fires promote significant changes in forest structure. Several studies have reported considerable reduction in aboveground biomass of forests disturbed by logging and fire. In these forests, a single fire can kill from 15% to 50% of the standing trees [Holdsworth and Uhl, 1997; Barbosa and Fearnside, 1999; Cochrane
J"
r
r 1
and Schulze, 1999; Gerwing, 2002; Haugaasen et al., 2003], thereby reducing the canopy closure through leaf shedding. The decrease of the leaf area index favors the increase of incoming solar radiation, drying the forest interior and increasing the amount of dead material and the forest floor fuel layer [Uhl and Buschbacher, 1985; Uhl and Kauffman, 1990; Nepstad et al., 2001; Ray et al., 2005]. In general terms, when an understory fire kills trees, it perpetuates the formation of gaps and fuel material on the forest floor in subsequent years [Nepstad et al., 1995,2001; Cochrane and Schulze, 1999]. These effects on forest structure are also reproduced by logging operations, which are recognized as one of the main anthropogenic disturbances contributing to forest flammability [Holdsworth and Uhl, 1997; Cochrane et al., 1999]. This interaction between logging and fire creates a positive feedback, which enhances the forest flammability following the initial disturbance [Nepstad et al., 2001; Cochrane, 2003]. Fire is an important disturbance influencing the rate and pattern of ecological succession of tropical forests [Goldammer, 1990; Chazdon, 2003]. Forest regeneration in Amazonia is a slow process' which can span several decades [Steininger, 1996; Tucker et al., 1998; Zarin et al., 2005]. Exposure to subsequent fires can compromise the reestablishment of less resistant plant species and lead to changes in forest composition [Uhl et al., 1988; Uhl and Kauffman, 1990]. Recurrent fires can retard succession to a point where it becomes difficult to reestablish the natural recovery process [Goldammer, 1990; Chazdon, 2003]. In addition, consecutive fires affecting areas of forest regrowth will negatively impact the nutrient elemental pools [Hughes et al., 2000; Zarin et al., 2005], limiting the forest capacity to recover. Changes in forest structure and composition due to fire will also affect biodiversity [Laurance, 2003]. Barlow et al. [2002] found that several types of avifauna were negatively impacted by the large scale 1997-1998 ENSO forest fire that occurred in Tapaj6s/Arapiuns Extractive Reserve, in the low Amazon region. These fire disturbances reduced the abundance ,of invertebrate communities, decreasing the availability of prey density to some bird species [Haugaasen et al., 2003]. Moreover, the heat produced by surface fires stresses trees, reducing the food supply to vertebrate frugivores and causing a decrease of these populations in recently burned areas [Barlow and Peres, 2006; Peres et al., 2003]. However, few studies have addressed the long-term response of biodiversity to fire in the region. A study cqnducted in eastern Amazonia indicated that game vertebrates (e.g., tapir, deer, agouti) tend to return to previously burned areas, since they now have new source of foraging substrate (regrowing vegetation), and they are protected by the dense understory vegetation established years after the fire, making it diffi-
cult for hunters to access these areas [Carvalho and Nepstad, 2000]. Barlow mid Peres [2004b, 2006] also reported continued tree mortlitlity and changing population dynamics among vertebratelwhen. analyzing areas affected by fires 3 years after the i¢ial burning event. Large-scale forest fires and the increase offire activity,as a whole, in Amazonia also affect the regional and global climate systems. Climatic consequences offires are mainly observed through an increase in both direct and committed CO 2 emissions [Barlow and Peres, 2004a; Barbosa and Fearnside, 1999; Alencar et al., 2006], as well as the emissions of methane and other gases and substances [Fearnside, 1997; Potter et al., 2002; Davidson and Artaxo, 2004] and through changes in surface properties (e.g., albedo, evapotranspiration, sensible and latent heat fluxes). While biomass burning emissions have an important role associated with the processes that control radiation balance and cloud formation [Penner et al., 1992; Andreae et al., 2004; Koren et al., 2004], surface cover change through forest fragmentation is recognized by major climate models as a key element which could lead to the savannization of large areas and to an increase in the risk of wildfires [Hoffmann et al., 2Q03a; Betts et al., 2004; Cox et al., 2004; Cowling and Shin, 2006]. The increase of biomass burning emissions affects the incoming solar radiation in Amazonia (e.g., increase in diffuse radiation) [Nemani et al., 2003], and this tendency may alter forest structure by favoring particular species of the plant community (e.g., increasing liana density). In addition, physiological and biogeochemical processes in old-growth tropical forests can be influenced by changes in atmospheric composition and land surface dynamics which include (1) rising atmospheric CO 2 concentration, (2) an increase in land surface temperature, (3) changes in precipitation and ecosystem water availability, and (4) changes in disturbance frequency [Chambers and Silver, 2004]. Process-based ecosystem models used to simulate the impact of fire in promoting future changes in climatic patterns showed large declines in net primary productivity and release of carbon as a result of Amazonian forest dieback [Friend et al., 1997]. The negative impacts offires and biomass burning emissions can be exacerbated by ENSO events, which promote severe droughts in the region [Van der Well et al., 2004]. These effects of climate change constitute a positive feedback in which the degraded forests become less effective at sequestering carbon and regulating regional climate, while becoming more susceptible to fire [Nepstad et al., 2001]. In sum, fire impacts climate which is a major determinant of the biological activities of plants, including phenology, physiology, distribution, and plant-animal interactions [Wright, 2005]. Ifthe trend ofmore extreme droughts and increased fire activity in tropical moist forest continues, it may
result in replacement of tropical moist forest species with more drought-tolerant and fire-resistant forms of scrubby, open vegetation resembling the cerrado (scrub savanna) of central Brazil [Shukla et al., 1990]. 3.2. Economic Consequences There are several economic losses associated with vegetation fires. The most common results from the direct impacts associated with fires that escape control are, namely, the loss of cattle and crops, and damages to infrastructure. However, other economic consequences include basin-wide effects associated with airport closures due to smoke and power outages due to fires along power lines. In addition, forest fires contribute to reducing forest value to society while influencing investments in rural areas. Forest fires decrease the production and cause mortality of important nontimber forest product species such as fruit and medicine trees and vines [Peres et al. 2003; Shanley and Medina, 2005]. Peres et al. [2003] reported losses of fruit trees due to forest fires along the Rio Tapaj6s with implications to game frugivore vertebrates. Shanley and Medina [2005] reported a decrease of about 80% in the family consumption of economically important fruit trees after a forest fire. The use of fire ~s the main characteristic of the Amazonian agriculture frontier, where land use investments are low, and the risk 9f accidental fire is high [Nepstad et al., 1999b, 2001; Sortensen, 2004]. Every year, accidental or escaped fires fro~ agriculture and pasture fields cause major economic losses in the region. Fire affects small- and large-scale farmers and ranchers by burning infi'astructure such as fences, buildings, and equipment, leading to reduced production capacity [Alencar et al., 1997; Nepstad et al., 1999b; Mendom;a et al., 2004]. Other consequences include losses of crop fields, pasture, and cattle. The risk of such losses end up influencing land use type and management decisions, perpetuating land use practices that use fire and discouraging investment in more sustainable methods [Nepstad et al., 2001]. In this scenario of high fire risk, extensive cattle ranching and annual crops are preferable if compared to more vulnerable and intensive land uses such as perennial crops [Walker et al., 2000]. In summary, investment in more sustainable land management is likely to decline as fire risk increases [Nepstad et al., 2001]. The consequences of escaped fires affect more than nonforest land uses. Forest fires were estimated to represent a loss of approximately US$5 per hectare in terms of marketable adult trees in the Paragominas region [Mendom;a et al., 2004]. In that region alone, one of the most important logging centers in Amazonia, this monetary loss was estimated at more than US$13 million during the ENSO 1997-1998
SCHROEDER ET AL.
48 SPATIAL DISTRIBUTION AND INTERANNUAL VARIABILITY OF FIRE IN AMAZONIA
49
\
....... HI
J
.l-
...I
period [Mendonr;a et ai., 2004]. In other areas where the eco- Research (INPE) in Sao Jose dos Campos, Brazil. In July nomic loss per hectare of forest can be higher, forest man- 1985, l-km resolution images from the advanced very high agement becomes even more risky. The high rates of tree resolution radiometer (AVHRR) aboard the NOAA-9 satmortality pr~moted by forest fires discourage more sustaina- ellite were acquired and processed to provide weather and ble forest management practices such as reduced impact log- cloud information for the NASA-INPE Amazon Boundary ging [Nepstad et ai., 2001]. The chance oflosing the second Layer Experiment (ABLE 2A). Unexpectedly, the images and third harvest cycles due to uncontrolled fires is one of showed dozens oflarge burnings with smoke plumes spreadthe several factors that contribute to more intensive harvest ing for hundreds of kilometers over supposedly pristine foroperations. This logging pattern also increases the likelihood ested areas. The ABLE 2A experiment provided the basis of fire spread creating a positi~e feedback between logging for the interpretation of the chemical species measured [Andreae et ai., 1988] and also the sample cases to develop a practices and forest fire risk [Nepstad et ai., 2001]. The smoke from fires also can reduce the visibility leading detection technique for identifying active fires in the 4-llm to airport closures and cause shortcuts in power lines inter- spectral channel. INPE then processed the AVHRR images rupting energy transmission [Mendonr;a et ai., 2004]. Despite for 1987, which showed hundreds offire events and massive the apparent importance of this type of economic loss, these emissions of gases and particulates to the atmosphere [Setzer consequences of fire to the regional economy are still to be and Pereira, 1991]. In 1989, the National System for Forest Fire Prevenestimated. However, it is possible that a future of more intion and Combat (PREVFOGO) was established under the tensive fire activity in Amazonia can increase the awareness auspices of the Brazilian Institute for the Environment and of this type of economic loss and push for quantification and Natural Renewable Resources (IBAMA). The AVHRR investment on more effective public policies to control fire. instrument aboard the National Oceanic and Atmospheric Administration (NOAA) polar orbiting satellite series re3.3. Sociai Consequences mained the primary data provider for the INPE fire monIncreased incidence of smoke-induced respiratory ill- itoring system for nearly 10 years. During that time, fire nesses is normally observed at the peak offire seasons [Men- detection data processed by INPE were routinely dissemidonr;a et ai., 2004]. School closures are also reported during nated to PREVFOGO via facsimile. However, it was only days of extreme smoke concentration, for example, in Acre after the great 1997-1998 El Nino episode that the active fire in 2005. Along with the direct impacts to human health, for- processing by INPE gained national interest. Widespread est impoverishment due to fires can lead to reduced produc- forest fires were reported for various areas across Brazilian tivity, thereby influencing patterns of land abandonment and Amazonia during the peak El Nino months with major forest loss affecting Roraima state [Eividge et ai., 2001]. As poverty. Smoke-induced respiratory illness is responsible for an a result, a national program was established immediately average of 9000 in-patients every year during the burning after the 1997-1998 El Nino episode as a response from season in Amazonia [Mendonr;a et ai., 2004]. In fact, fire can the Brazilian federal government to environmental concerns affect the health of more than 13,000 people during ENSO raised by the national and international communities in face years. Despite the relatively low number of people affected of the damages caused by the fires. In May 1998, the Fire by respiratory illnesses, if compared to Amazonian popula- Prevention and Control Program for the Arc of Deforestation, the government costs to treat such illnesses were esti- tion (PROARCO) was established under the auspices of mated to reach US$10 million during the 1997-1998 ENSO. IBAMA,with financial support from the World Bank. The It is important to realize that these numbers are only based PROARCO program was designed to make intensive use on the cases that required hospitalization. Anecdotal evi- of satellite remote sensing products and geographic infordence suggests that the impact offire and smoke to rural and mation systems technology to provide near real-time active urban population health is underestimated, since most of the fire information and fire monitoring statistics for Brazilian respiratory problems, mainly in rural Amazonia,. tend to be Amazonia. The fire information was intended to support the regional strategic plans and help guide the field activities of treated at home. PREVFOGO. Following the establishment of PROARCO, the remote 4. MAPPING AND MONITORING FIRE EVENTS: sensing active fire database for Brazilian Amazonia was PAST, PRESENT, AND TRENDS gradually improved by incorporating additional satellite Routine active fire monitoring over Amazonia was initi- systems into routine fire monitoring operations undertaken ated during the mid-1980s at the National Institute for Space by IBAMA (http://www.dpi.inpe.br/proarco) and INPE
(http://www.cptec.inpe.brlqueimadas). The original pre-El Niiio fire monitorinK·capacity based on a single AVHRR sensor was rapidly.:enhanced by incorporating data from (l) the Geostatiofary Operational Environmental Satellite 4-km resolutiol).Anager positioned at 75°W longitude along the equator (GbES East), (2) the l-km resolution Moderate Resolution Imaging Spectroradiometer (MODIS) aboard the EOS-AM (Terra) and EOS-PM (Aqua) polar orbiting satellites, (3) the 0.55-km resolution Defense Meteorological Satellite Program (DMSP) nighttime-based detection using the Operational Linescan System (OLS) visible channel data, (4) the 2A-km resolution visible and infrared scanner (VIRS) aboard the Tropical Rainfall Measuring Mission (TRMM) low inclination orbit satellite, (5) the European geostationary Meteosat Second Generation (Meteosat-9) 3-km resolution instrument, and finally, (6) by incorporating data from additional AVHRR instruments flying aboard other NOAA polar orbiting satellites. The combined use of multiple polar orbiting and geostationary satellites was an important step toward reducing the response time during fire emergencies, while permitting improved delineation of fire spatial and temporal dynamics through the integration and c9mparison of multiple observations and the production of basin-wide fire statistics [Schroeder et ai., 2005]. Most remote sensing imaging systems with fire monitoring capability rely on the 4-llm spectral region to detect active fires. Active fire detection performance varies primarily as a function of a sensor's spatial resolution and algorithm used. Satellite sensors measure the total instantaneous fire-emitted radiation, which depends on the size and intensity of the flaming front. In fact, the actual flaming area of a vegetation fire will normally be constrained to a few tenths of a hectare, which, compared to the spatial resolution of the instruments described above, will represent only a small fraction of the projected pixel area. Consequently, coarser resolution satellite data will typically be associated with reduced detection capacity as larger flaming areas, or alternatively more intense fires, will be required to create a distinguishable signature from the area-averaged pixel data [Schroeder et ai., 2008b]. In this respect, the improved sensitivity to smaller fires provided by single daytime and nighttime observations from moderate resolution polar orbiting instruments serve to complement the high observation frequency data provided by coarse spatial resolution geostationary satellites. Application of different fire detection algorithms also provides the user community with additional fire product versions to choose from [Morisette et ai., 2005]. Despite having the longer satellite time series currently available, AVHRR data may be impacted by the systematic orbital drift of NOAA 7, NOAA 9, NOAA 11, and NOAA 14 satellites, which can have important consequences for interannual fire analyses
[Csiszar et ai., 2003]. Currently, the AVHRRINOAA 12 and the TRMM data provide two of the longest continuous fire data records available for Amazonia (approximately 10 years of data acquisition) systematically produced from a single satellite instrument and algorithm architecture. Despite its shorter time series of active fire data records for Amazonia, the MODIS instrument routinely provides very accurate image navigation information, which can be useful when finer spatial analyses are desired [Wolfe et ai., 2002]. Despite the sizeable number of products that are available for Brazilian Amazonia, very little information is at hand to characterize the individual burning events that are described by the active fire detection products. Sensor limitations commonly associated with low saturation levels prevent estimation of important parameters such as fire temperature and size for a significant fraction ofthe events mapped. However, altemate products can be used to help characterize the extent of burning at a particular location beyond the inherent limitations associated with the active fire data. These complementary products include burnt area mapping [e.g., Barbosa et ai., 1999; Gregoire et ai., 2003; Roy et ai., 2002; Simon et ai., 2004] and fire radiative power estimates [Kaufman et ai., 1996; Wooster et ai., 2005]. The bumt area mapping derived from MODIS is the first peer-reviewed global-scale product to be incorporated into the routine land surface products processin~ stream of a major satellite mission. Although preliminary assessment of its performance included part of Amazonia tRoy et ai., 2005], further research is still required to fully ~haracterize the potential for fire monitoring applications dver the region. Fire radiative power estimates are another relatively recent application in the field of remote sensing. Previous studies have demonstrated its effectiveness in quantifying the rates of biomass combustion for vegetation fires, which, in tum, can be used to derive estimates of gaseous emissions from burning [Kaufman et ai., 1996; Wooster, 2002; Wooster et ai., 2003]. Application of fire radiative power estimates to derive total fire emitted radiant energy [or fire radiative energy (FRE)] is dependent on the frequency of observations (for integration purposes) for which geostationary instruments are well suited. However, the coarse resolution and low saturation level of most imaging instruments aboard geostationary platforms, along with the problem of cloud coverage, may still prevent full derivation ofFRE estimates in many cases [Roberts et ai., 2005]. Future remote sensing systems with active fire monitoring capacity include two new series of polar orbiting and geostationary satellites which should enable routine imaging of Amazonia for the next two decades. The National Polar-orbiting Operational Environmental Satellite System (NPOESS) will replace the existing AVHRR sensor series with improved spatial, spectral, and radiometric resolutions.
50
.....
.. ii
.ll 1!
SPATIAL DISTRIBUTION AND INTERANNUAL VARIABILITY OF FIRE IN AMAZONIA
The proposed NPOESS orbital configuration is expected to produce three daytime image acquisitions with Equator crossing times of 0930, 1330, and 1730. The first NPOESS instrument fs expected to be operational by early 201Os. The Geostationary Environmental Operational Satellite Series-R, planned for launching by early 201Os, will replace the existing GOES imager series. GOES-R should provide improved spectral; spatial, and temporal resolution data, becoming an important tool for monitoring fire activity over Amazonia at very high observation frequencies (typically :S15 min for full hemisphere coverage). Another new trend in remote sensing deals with the concept of small satellite missions, which are targeted at instruments dedicated to map a fewer number of parameters (see Briess et ai. [2003] for an example of a small satellite mission dedicated to active fire detection). Their major advantage is reduced costs relative to other multimission platforms while satisfying the specific needs of a particular measurement (e.g., spectral and spatial sampling issues). 5. SPATIAL DISTRIBUTION, INTERANNUAL VARIABILITY, AND CHANGE RATES As described in the previous sections, the spatial distribution of vegetation fires in Brazilian Amazonia is strongly associated with human presence as well as with the regional biophysical conditions. Satellite maps of vegetation fire activity show that approximately 40% of the region is under the influence offires (Plate 1). Vegetation fires are primarily concentrated in the southern and eastern parts of the basin including the states of Rondonia and Mato Grosso (to the south) and Tocantins, Maranhao, and eastern Para (to the east) (Table 1). In those areas, improved infrastructure, in particular, the road network, serves to promote an accelerated process of land conversion and the more intensive use
SCHROEDER ET AL.
of land where fires playa significant role [Laurance et ai., 2001; Nepstad et al., 2001; Aiencar et ai., 2004; Arima et al., 2007; SO/TeIlSen, 2004] (see also Figure 3). While fires appear to be widely distributed in space especially in areas such as the states of Tocantins and Maranhao, analysis of their return frequency shows the existence of more complex spatial patterns (Plate 1). These patterns are a function of the type of application involved with the use of fire and therefore will reflect the characteristics associated with the two main categories described above, namely, conversion and maintenance fires. Conversion fires will usually be related to high intensity, long-lasting burning episodes as a result of larger fuel loads, and their distribution in space will tend to form a continuum of gradually expanding areas following the deforestation patterns (see section 5). Maintenance fires, however, will tend to be associated with lower intensity and shorter burning episodes which are typically scattered in space following the rural landscape configuration. These considerations make the detection of conversion fires from remote sensing imaging systems more likely, therefore creating clusters of high fire frequency over areas where forest conversion continues for two or more consecutive years. Under cloud-free conditions, open sky fires (e.g., conversion or maintenance) as small as 0.1 ha may be detected even by coarse resolution satellite imaging systems [Prins et ai., 1998]. However, for low-intensity understory burning, most of the radiant energy emitted by the fire will be intercepted by the canopy and, thereby, prevent detection from infrared imaging systems. An alternative approach for mapping understory fires, which relies on the application of nighttime imaging of visible light emitters, was used over part of Roraima state to map the extent of understory burning during the 1997-1998 El Nino episode [Eividge et ai., 2001]. However, basin-wide annual quantification of understory active fires, which can be compared to the open sky
0,'.-
.','
a)
\
,;
10'S
o 125250
500
Region
! I
j
"
~,
rt,I
Brazilian Amazonia Acre Amapa Amazonas Maranhao Mato Grosso Para Rondonia Roraima Tocantins
Number of Cells
LowConf (%)
MedConf (%)
HighConf (%)
168,726 5,306 4,899 52,082 11,283 30,705 41,421 8,201 7,660 9,577
44.3 46.1 61.9 77.3 3.1 13.4 38.9 21.3 54.6 2.2
11.6 16.2 12.3 14.6 3.7 11.0 10.8 11.0 13.6 6.3
11.4 16.6 11.5 5.0 12.5 19.5 10.3 11.9 12.6 19.3
32.7 21.1 14.3 3.1 80.7 56.1 40.0 55.8 19.2 72.2
750
~i~_iiiiiiiil~~i Kilometers 70'W
eo'w
50'W
eo'w
50'W
b)
O·
10'S
500
i
No Detection (%)
,
I· "'~ ',. I
t.'-
O·
o 125250
Table 1. Fractional Distribution of 0.05° x 0.05° Grid Cells With Associated Hot Spot Detections (From Plate 1) for Brazilian Amazonia and Its Nine States
»,
70'W
750 i Kilometers
Plate 1. (a) Areas of potential fire activity as mapped by daytime and nighttime AVHRRINOAA 12, GOES 08 and 12, and MODIS/Terra and Aqua data during 2003-2006 (4.15 million hot spots processed). Hot spot observations from all three fire products were aggregated using a 0.05° x 0.05° grid. High confidence areas are defined as those locations for which all three sens~rs (i.e., AVHRR, GOES imager, and MODIS) produced hot spot detections during the time period above, whereas medlUm and 10": confidenc~ areas had two and one sensor producing hot spot detections, respectively. , Are~s of 10': confidence are paltlCularly noticeable across Amazonas state and are usually associated with low rank: detections depicted by the GOES fire product. (b) Map of active fire detection frequency derived from MODISITerra and Aqua daytime and nighttime overpasses during 2003-2006. Data was aggregated using a 0.025° x 0.025° grid and further resampled to 0.05° x ?05° .gr~d using maximum value criterion. Color scheme represents the number of years with observed hot spot detectIOn wlthm each 0.05° x 0.05° grid cell. The distribution of indigenous reserves and federal and state (displayed as a single layer) conservation units is also shown in Plate lb. '
51
52
SPATIAL DISTRIBUTION AND INTERANNUAL VARIABILITY OF FIRE IN AMAZONIA
SCHROEDER ET AL. I
'II
.U
detections, remains to be determined. Other complementary studies have used change detection techniques and spectral indices applied to high and moderate resolution satellite data for monitoAng areas affected by understory fires [Alencar et aI., 2006; Shimabukuro et al., 2006; Souza et al., 2005]. These techniques focus on the effects of understory fires (e.g., tree mortality) as opposed to the instantaneous fireemitted radiation to derive estimates ofburning activity over closed-canopy forest and selective logging areas. Consequently, only areas where canopy damage is noticeable may be represented by those methods. The regional climate ofAmazonia is another important factor which can influence the spatial and temporal distribution offires in the region. The average biophysical conditions of Amazonia often pose a natural barrier for vegetation fires to develop [Uhl and Kauffman, 1990]. Temperatures remain stable throughout the year, annual average rainfall is in excess of 2000 mm for most of the region, and the average number of consecutive rainless days during the dry season is relatively small [AlvaM et al., 2002; da Rocha et aI., 2004]. These factors have a direct impact on the human activities in the region (e.g., limiting road traffic flow during the wet season) and consequently on the use offires. Land management through fire then becomes temporally constricted and tends to follow the onset of the dry season across the basin. Along the transition zone that separates the evergreen tropi-
cal forests from the cerrado type of vegetation, the contrast between dry and wet seasons becomes more pronounced, and the rainless periods can be more than 4 months. Under such conditions, the temporal distribution of fire use may also be influenced by social, economic, and political factors which will help determine at the local or regional scale the particular timing of fire use during the dry season period. In this kind of environment, a more stratified regional pattern may result (e.g., central Mato Grosso state; Plate 2). Land cover change and fire activity may vary as a function of economic incentives promoted by national and international market connections [Fearnside, 2001; Brown et al., 2005; Nepstad et al., 2006b]. For instance, the steady increase in soybean market price observed during the 20012004 period was followed by an equivalent increase in the total area planted in Amazonia [Morton et al., 2006], thereby pushing the annual deforestation rates and the number offire detections alike. The increase in the use of fires for land clearing during the 2001-2004 period was reflected in the growing number of detections mapped over densely forested areas for that same period (Figure 1). Major interannual variabilitY of fire activity in Amazonia can also be associated with extreme climatic events. Events such as El Nino and the recent warming of the tropical North Atlantic in 2005 [Marengo et al., 2008] are prone to increase forest flammability as a result of severe drought conditions,
e)1
.' ,
80
, - - - - - - - - - - - - - - - - - - - - - - - - r 180000
70
160000 140000
60 51
120000 ~
49
~ 50
ti
Q)
100000
0>
.lll 40 c Q)
e
~
80000 60000
2002
2003
2004
_VCFO-20
_VCF20-40
c::=:::::J VCF 60-80
-
2005 _VCF40-60
-HotSpots
lK
Annual Cloud Fraction
2000
45000 ,1;;1
1800
40000
,E!'
1600
35000
~ ~ 1400
C 0
19
~
-5
1::
E
'iii
,
30000 ~
'"
.!!!- 1200
~
.l!l
.c 0
1000
z
25000
tI
'0
O'
'0 o
C
'§
~
600
8.
(f)
20000 ~
0>
800
15000
Qi
E ::J
::J
Z
~
o 2001
'0
a.
20000
0
::c '0 Qi ..c
-oC/)
40000
10
tli
~ (f)
30 20
~
53
400
10000
200
5000
o
Z
o 1998 _
1999 2000
2001
2002
2003
2004
2005
2006
Roraima _Acre· . 0 .. Brazilian Amazonia
10'8
Figu~e 1. (a) MOD~S0erra daytime and nighttime fire product summary statistics during 2001-2005. (dashed line) Hot spot mter~nnual v~natlOn pres~nted as totals and (vertical bars) as annual percentage distribution using unifonn Vegeta-
o 125250
500
tIOn Contmuous FIelds (VCF) mtervals. (asterisks) Basin-wide average cloud-cover fraction derived from MODISlTerra shows little interannual variation «4%). (b) (dashed line) Annual number of hot spots detected by the TRMMNIRS instmment over the Brazilian Amazonia. Vertical bars describe hot spots detected over Roraima and Acre states, which were severely affected by the 1997-1998 EI Nino and the 2005 wanning ofthe tropical North Atlantic waters, respectively.
750
IKilome(ers
i 70'W
60'W
50'W
Plate 2. Map of peak month of hot spot detection as observed by MODISlTerra and Aqua, AVHRR-NOAA 12 and GOES 08, and 12 daytime and nighttime images during 2003-2006 (4.15 million hot spots processed). Data was aggregated using a 0.05° x 0.05° grid. Only cells having hot spot detections produced from two or more sensors (medium and high confidence cells from Plate la) are represented on the map.
which may develop in parts of the region. Increased forest flammability associated with higher risk of fires escaping control often lead to widespread forest fires affecting significantly large areas [Van der Werfet al., 2004; Alencar et aI.,
2006; Brown et al., 2006; Nepstad et al., 1999a]. Satellite active fire products will normally show strong peaks departing from the annual average in fire activity associated with such large-scale climate anomalies (Figure 1).
-----1'
1
II
j
54
...
:ri... .Ii
.Ii !!
...
SPATIAL DISTRIBUTION AND INTERANNUAL VARIABILITY OF FIRE IN AMAZONIA
Some of the most recent alternatives designed to cope with the increase in fire activity in Brazilian Amazonia were associated with the adoption of specific public policies including tempm-ary fire prohibition, increase in law enforcement, and the creation of new conservation areas. Fire prohibition was first proposed for Mato Grosso state in 2001 and was followed by other regions in subsequent years. It is used as a preventive mechanism limiting burning at the peak of the dry season as well as an emergency response when rapid reduction in fire activity is desired [Brown et al., 2006]. Successful application of a fire moratorium depends on the effectiveness oflaw enforcement and on community engagement. The creation of new conservation areas also depends on the effectiveness of law enforcement and park administration, and in some cases, the established areas may not withstand the threats of logging and fire [Ferreira et al., 1999; Laurance and Williamson, 2001; Pedlowsld et al., 2005]. Pressure is building along conservation units where the surrounding forests are being depleted (see examples in Plate 1). 6. SPATIAL AND NUMERICAL RELATIONSHIPS WITH DEFORESTATION RATES As described above, vegetation fires and deforestation activities in Amazonia are closely related. However, the numerical relationship between satellite derived hot spot counts and the spatially coincident deforestation estimates at any spatial scale remains mostly unresolved. Among the major factors limiting our capacity to establish a more accurate relationship between hot spot counts and deforestation area are the following: 1. Vegetation fires have a highly dynamic nature. Constant changes in fire size and temperature limit our ability to derive a mean fire property. 2. The mode of image acquisition is noncontinuous. Satellite images are usually acquired at intervals ranging from 15 min to 12 h for geostationary and polar orbiting satellites, respectively. 3. The imaging process limitations involved. Optically thick clouds can obscure fires and prevent their detection [Schroeder et al., 2008a]. 4. The forms of burning and fire type variations. The wide range of vegetation structures and fuel loads which characterize Amazonia will influence the detection offires accordingly [Schroeder et al., 2005]. Due to the limiting factors above, it is very likely that a significant fraction of the actual fires will have only a few observations made by most remote sensing products during the entire life cycle of the burning event. Consequently, the relationship between the total deforested area and the number of hot spots detected for a particular location is usu-
" 1.:1.
r... !
.
1
ally difficult to derive. Nevertheless, the spatial distribution and concentration of hot spots tend to follow the trend described by the deforestation rates (Figure 2). Figure 3 shows the percentage distribution of hot spots detected alongside a 260-km segment of highway BR-163 near Novo Progresso in Para state. Fire detection statistics were derived for two distinct periods (1999-2000 and 200~ 2005) using seven lO-km buffers across the road's main axis. The 70-km-across subregion described by the buffers represents one ofthe most intact tracts of forests found in the immediate vicinity of highway BR-163 during this period. Factors such as reduced road trafficability, especially in the wet season, and the increased distance to ports and markets have limited the expansion of human activities in this region relative to other areas. For comparison purposes, the corresponding percentage dish'ibution of the annual deforestation increment derived from higher-resolution Landsat Enhanced Thematic Mapper Plus (ETM+) imagery is also plotted in Figure 3. Fire and deforestation show very similar patterns for the two periods analyzed with equivalent changes in the spatial distribution over time. Most important in Figure 3 is the progress in deforestation and fire use away from the highway and deeper into the forested areas, which suggests the intrusion of human activities in previously undisturbed areas. Absolute deforestation rates increased by a factor of three within the 5-year period analyzed, while hot spot counts went up by as much as five times. It is important to note that active fire detection products based on contextual methods, such as the one used with Figure 3, can be affected by commission errors which might reinforce the relationship with deforestation (see Giglio et al. [1999] for a discussion of different types of fire detection methods). These errors may be observed over deforested sites surrounded by relatively homogeneous forests as a result of the high thermal contrast between the two areas which cause a false detection to be produced [Schroeder et al., 2008b]. Despite the good overall agreement between the two different data sets in Figure 3, the measure of correlation describing individual episodes (i.e., the relationship between the number of hot spots detected and the area in hectares of the overlapping deforestation polygon) remains low (,2 = 0.54 using 2004 data). There has been a long debate over the relationship between roads, deforestation, and fire use in Amazonia [see, for example, Nepstad et al., 2001; Laurance et al., 2001; Silveira, 2001; Camara et al., 2005]. Roads facilitate access to otherwise remote areas and therefore serve to promote land use expansion where deforestation and fires play a significant role. However, their importance in relation to other forces such as regional and global economic markets is still subject to major controversy. Nevertheless, as shown in Figure 3,
~
SCHROEDER ET AL.
J .J: ~
~
lioooo
,
)1
14000
70000
"
12000
.."
, 60000
10000
50000
1
~
g-
8000
~
~ <::
o
'"
40000
~
(5 J:
6000 ~
o
30000 4000 20000
--HotSpots . • . • . •Deforestation
10000
o 1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2000
0 2005
Figure 2. Deforestation estimates (PRODES) and AVHRRlNOAA 12 hot spot detections for the state of Mato Grosso during 1994-2005.
50
i
• Oaf 1999·2000 • Hot Spots 1999-2000 <>I Oaf 2004·2005 o Hot Spots 2004-2005
45 40 35 ~ 30 C1>
OJ
~ 25 C1>
~
If
20 15 10 5 0 10
20
30
40
50
60
70
Distance to Road (km)
Figure 3. Hot spots detected by the TRMMNIRS instrument and deforestation increments mapped with the Landsat enhanced thematic mapper plus (ETM+) instrument during 1999-2000 and 2004-2005 for a 260-km segment ofhighway BR-163 near Novo Progresso, Pani state. Values represent the percentage distribution using lO-km-wide buffer zones from the highway.
1
55
56
•...
-.... -"
....
:'". 1!.I
..11
,I
SPATIAL DISTRIBUTION AND INTERANNUAL VARIABILITY OF FIRE IN AMAZONIA
deforestation and fires are particularly concentrated along the major road corridors in the region. Future projections of forest transformation by deforestation and fires will be influenced b~ the assumptions incorporated into the models, which often lead to important differences in the predictions produced. The establishment of positive feedbacks between deforestation and fires is perhaps the most important aspect which helps explain the frequently observed pattern of increased forest degradation along major road corridors. Provided that human occupation in Amazonia will continue in the decades to come, it is very likely that deforestation and fires will also become even more frequent phenomena in the region, as forest fragmentation increases. Many modeling studies have suggested important regional biophysical changes induced by logging and fires which could ultimately lead to a gradual replacement of forests by savanna-like vegetation as drier and warmer conditions prevail [Nobre et al., 1991; Henderson-Sellers et al., 1993; HojJmann et al., 2003b]. 7. CONCLUSION As presented throughout this chapter, vegetation fires in Amazonia are very diverse in nature with highly heterogeneous spatial and temporal distributions. Fire is a fast and inexpensive tool currently used by millions of landholders in the region for converting forests into crop production and pasture as well as for managing their lands afterwards. Consequently, fires are strongly influenced by the human presence in the region. The widespread use offires in Amazonia has important impacts on various biophysical and climate dynamics processes, which extend from landscape to global scales. Fire intrusion over previously undisturbed areas may also trigger important feedback mechanisms which can disrupt the fragile environment of evergreen tropical forests. Major awareness of the extent of biomass burning and its associated effects over Amazonia was fostered in 1985 when the first evidences of the widespread use of fires in the region were noticed. Fire mapping and monitoring capabilities have been significantly enhanced since then, evolving from single A VHRR image monitoring in the 1980s to more than a dozen orbital platforms collecting over a hundred images every day. However, the spatial scales at which active fires occur (normally in the order of a few tenths of a hectare) and the degree of variability of fire properties as a function of time still pose significant challenges for the full characterization of the different fire regimes in Amazonia. Nevertheless, despite their limitations in terms of spatial resolution, and spectral and radiometric characteristics, which often preclude estimates of fire-related parameters such as temperature and area affected, the current suite of instruments
available for active fire monitoring have been providing important information to delineate the regional aspects of fire occurrence in Amazonia. Social scientists have gained a good understanding of the major causes and economic and social consequences offire in Amazonia. Biologists and ecologists now understand the effects of fire in the Amazonian forest ecosystem and are able to delineate several feedback processes that emerge from the complex associations between fire and vegetation dynamics. Likewise, climatologists have successfully described the impacts of fires on regional and global climate dynamics providing groundbreaking results that magnify the importance of widespread use offires in the tropics. Future research methods should integrate multiple data sets to provide new avenues for improved understanding of the dynamics of fire use in the tropics. Coupled human-climate models are then required to assess the implications ofvegetation fires for regional societies and global climate and to help delineate the multiple feedbacks and processes, which drive the interactions between humans and the environment. Acknowledgments. We would like to acknowledge the highly skilled and dedicated scientists studying vegetation fires in Amazonia, along with several local groups, who have either directly or indirectly contributed to the material discussed in this chapter. The AVHRR, GOES, and MODIS satellite active fire detection product data sets used in this chapter were kindly provided by INPE, the University of Wisconsin at Madison, and NASA, respectively. Funding for this research was available from NASA's LBA-Eco investigations: CD-II, LC-14, LC-02, LC-23, LC-24, LC-36, LC35, LC-37, and TG-03, from NASA project NNG06GD96A, from project Milenio-LBA2 (CNPq 420199/2005-5), and from NASA's Earth and Space Science Fellowship Program.
REFERENCES Alencar, A., D. Nepstad, E. Silva, F. Brown, P. Lefebvre, E. Mendosa, D. Almeida, and O. Carvalho Jr. (1997), 0 Uso do Fogo na Amazonia: Estudos de Caso ao Longo do Arco de Desmatamento, World Bank, Brasilia, Brazil. Alencar, A, L. Solorzano, and D. Nepstad (2004), Modeling forest understory fires in an eastern Amazonian landscape, Ecol. Appl., 14(4), S139-S149. Alencar, A, D. Nepstad, andM. C. VeraDiaz (2006), Forest understory fire in the Brazilian Amazon in Enso and non-Enso years: Area burned and committed carbon emissions, Earth Interact., 10(6), EI150, doi:IO.1175/EI150.1. Alvahl, R. C. S., R Gielow, H. R da Rocha, H. C: Freitas, J. M. Lopes, A. O. Manzi, C. Von Randow, M. A F. Silva Dias, O. M. R. Cabral, and M. J. Waterloo (2002), Intradiurnal and seasonal variability of soil temperature, heat flux, soil moisture content, and thermal properties under forest and pasture in Rondonia, J. Geophys. Res., 107(D20), 8043, doi:IO.1029/2001JD000599.
SCHROEDER ET AL.
57
Andreae, M. 0., E. V. Browkll, G. L. Gregory, R. C. Harris, G. F. Bruner, A G., R E. Gullison, R. E. Rice, and G. A B. da Fonseca Hill, G. W. Sachse, R W. Talbot, M. Garstang, D. J. Jacob, and (2001), Effectiveness ofparks in protecting tropical biodiversity, Science, 291,125-128. A L. Torres (1988), lliomass-burning emissions and associated Buschbacher, R, C. Ubi, and E. A Serrao (1988), Abandoned pashaze layers over Awt~zonia, J. Geophys. Res., 93,1509-1527. Andreae, M. 0., D. B"osenf\lld, P. Artaxo, A A Costa, G. P. Frank, tures in Eastern Amazonia. II. Nutrient stockes in the soil and K M. Longo,)ttd M. A F. Silva Dias (2004), Smoking rain vegetation, J. Eco!., 76,682-699. clouds over the Amazon, Science, 303, 1337-1342. Camara, G., A P. Dutra-Aguiar, M. I. Escada, S. Amaral, T. Carneiro, Arima, E., C. S. Simmons, R Walker, and M. A Cochrane (2007), A M. V. Monteiro, R. Araujo, I. Vieira, and B. Becker (2005), Fire in the Brazilian Amazon: A spatially explicit model for polAmazonian deforestation models, Science, 307, 1043-1044. Carvalho Jr., 0., and D. Nepstad (2000), Forest fire: Impacts on icy impact analysis, J. Reg. Sci., 43(3), 541-567. plant biomass and mammals populations in eastern Amazon, Artaxo, P., L. V. Gatti, A M. C. Leal, K M. Longo, S. R. Freitas, L. L. Lara, T. M. Pauliquevis, A. S. Procopio, and L. V. Rizzo First LBA Scientific Conference, Belem, Brazil, 26-30 June. (2005), Atmospheric chemistry in Amazonia: The forest and the Chambers, J. Q., and W. L. Silver (2004), Some aspects of ecophysiological and biogeochemical responses of tropical forests biomass burning emissions controlling the composition of the Amazonian atmosphere, Acta Amazonica, 35(2), 185-196. to atmospheric change, Phi/os. Trans. R. Soc. London, Ser. B, 359,463-476. Barbosa, R. I., and P. M. Fearnside (1999), Incendios na Amazonia Brasileira: Estimativa da emissao de gases do efeito estufa pela Chazdon, R. L. (2003), Tropical forest recovery: Legacies of human impact and natural disturbances, Perspect. Plant Eco!. Evol. queima de diferentes ecossistemas de Roraima na passagem do evento "E! Nino" (1997/1998), Acta Amazonica, 29(4), Syst., 6(1), 51-71. 513-534. Cochrane, M. A. (2001), Synergistic interactions between habitat Barbosa, P. M., J.-M. Gregoire, and J. M. C. Pereira (1999), An alfragmentation and fire in evergreen tropical forests, Conserv. gorithm for extracting burned areas from time series of AVHRR Bioi., 15(6),1515-1521. GAC data applied at a continental scale, Remote Sens. Environ., Cochrane, M. A (2003), Fire science for rainforests, Nature, 421, 913-919. 69,253-263. Barlow, J., and C. A. Peres (2004a), Ecological responses to EI Cochrane, M. A, and W. F. Laurance (2002), Fire as a large-scale Nino-induced fires in central Amazonia: Management implicaedge effect in Amazonia forests, J. Trap. Ecol., 18, 311-325. tions for flammable tropical forests, Phi/os. Trans. R. Soc. Lon- Cochrane, M. A, and M. D. Schulze (1998), Forest fires in the Brazilian Amazon, IConserv. Bioi., 12(5), 948-950. don, Ser. B, 359, 367-380. Barlow, J., and C. A Peres (2004b), Avifaunal responses to single Cochrane, M. A, and M. D. Schulze (1999), Fire as a recurrent event and recurrent wildfires in Amazonian forests, Ecol. Appl., 14(5), in tropical forests of the Eastern Amazon: Effects on forest struc1358-1373. ture, biomass, and ~pecies composition, Biotropica, 31(1), 2-16. Barlow, J., and C. A Peres (2006), Effects of single and recurrent Cochrane, M. A, A Alencar, M. Schulze, C. Souza Jr, D. C. Nepwildfires on fruit production and large vertebrate abundance in a stad, P. Lefebvre, and E. Davidson (1999), Positive feedbacks in central Amazonian forest, Biodivers. Conserv., 15(3),985-1012. the fire dynamic of closed canopy tropical forest, Science, 284, 1837-1841. Barlow, J., T. Haugaasen, and C. A Peres (2002), Effects ofground fires on understory bird assemblages in Amazonian forests, Bioi. Cowling, S. A, and Y. Shin (2006), Simulated ecosystem threshold responses to covarying temperature, precipitation and atmoConserv., 105(1), 157-169. Barlow, J., C. A. Peres, B. O. Langan, and T. Haugaasen (2003), spheric CO2 within a region of Amazonia, Global Ecol. BiogeLarge tree mortality and the decline of forest biomass following ogr., 15, 553-566. Amazonian wildfires, Ecol. Lett., 6(1), 6-8. Cox, P. M., R A Betts, M. Collins, P. P. Harris, C. Huntingford, Betts, R. A, P. M. Cox, M. Collins, P. P. Harris, C. Huntingford, and C. D. Jones (2004), Amazonian forest dieback under cliand C. D. Jones (2004), The role of ecosystem-atmosphere inmate-carbon cycle projections for the 21st century, Theor. Appl. teractions in simulated Amazonian precipitation decrease and Climatol., 78, 137-156. forest dieback under global climate warming, Theor. App!. Cli- Csiszar, I., A Abuelgasim, Z. Li, J. Jin, R. Fraser, and W.-M. Hao matol., 78(1-3), 157-175. (2003), Interannual changes of active fire detectability in North Briess, K, H. Jahn, E. Lorenz, D. Oertel, W. Skrbek, and B. ZhuAmerica from long-term records of the advanced very high resokov (2003), Fire recognition potential ofthe Bi-spectral InfraRed lutionradiometer,J. Geophys. Res., lO8(D2), 4075, doi:~0.10291 2001JD001373. Detection (BIRD) satellite, Int. J. Remote Sens., 24, 865-872. Brown; I. F., W. Schroeder, A Setzer, M. L. R. Maldonado, N. da Rocha, H. R, M. L. Goulden, S. D. Miller, M. C. Menton, L. D. Pantoja, A Duarte, and J. Marengo (2006), Monitoring fires in V. O. Pinto, H. C. de Freitas, and A M. Silva Figueira (2004), Southwestern Amazonia rain forests, Eos Trans. AGU, 87(26), . Seasonality ofwater and heat fluxes over a tropical forest in east253, doi: 10.102912006E0260001. ern Amazonia, Ecol. Appl., 14(4), S22-S32. Brown, J. C., M. Koeppe, B. Coles, and K P. Price (2005), Soy- Davidson, E., and P. Artaxo (2004), Globally significant changes in bean production and conversion of tropical forest in the Brabiological processes of the Amazon basin: Results ofthe Largezilian Amazon: The case of Vilhena, Rondonia, Ambio, 34(6), Scale Biosphere-Atmosphere Experiment, Global Change Bioi., 462-469. lO,519-529.
58
SPATIAL DISTRIBUTION AND INTERANNUAL VARIABILITY OF FIRE IN AMAZONIA
SCHROEDER ET AL.
~I
Elvidge, C. D., V. R. Robson, K. E. Baugh, J. B. Dietz, Y. E. Shimabukuro, T. Krug, E. M. Novo, and F. R. Echavarria (2001), DMSP-OLS estimation of tropical forest area impacted by surface fires in Roraima, Brazil: 1995 versus 1998, Int. J. Remote
(1996), Relationship between remotely sensed fire intensity and rate of emission of smoke: SCAR-C experiment, in Global Biomass Burning, edited by J. Levine, pp. 685-696, MIT Press, Cambridge, Mass. Koren, I., Y. J. Kaufman, L. A. Remer, and J. V. Martins (2004), Measurement of the effect of Amazon smoke on inhibition of cloud formation, Science, 303,1342-1345. Laurance, W. F. (2003), Slow bum: The insidious effects of surface fires on tropical forests, Trends Ecol. Evo!., 18(5),209-212. Laurance, W. F., and G. B. Williamson (2001), Positive feedbacks among forest fragmentation, drought and climate change in the Amazon, Conserv. Bioi., 15(6),1529-1535. Laurance, W. F., M. A. Cochrane, S. Bergen, P. M. Feamside, P. Delamonica, C. Barber, S. D'Angelo, and T. Fernandes (2001), The future of the Brazilian Amazon, Science, 291, 438-439. Marengo, J. A., C. A. Nobre, J. Tomasella, M. D. Oyama, G. S. de Oliveira, R. de Oliveira, H. Camargo, L. M. Alves, and I. F. Brown (2008), The drought of Amazonia in 2005, J. Clim., 21, 495-516. Meggers, B. J. (1994), Archeological evidence for the impact of mega-Nino events of Amazonia during the past two millennia, Clim. Change, 28, 321-338. Mendon'Ya, M. J. C., M. C. Vera-Diaz, D. Nepstad, R. Seroa da Motta, A. Alencar, J. C. Gomes, and R. A. Ortiz (2004), The economic cost of the use of fire in the Amazon, Ecol. Econ., 49, 89-105. Moran, E., E. S. Brondizio, J. M. Tucker, M. C. Silva-Forsberg, S. McCracken, and I. Falesi (2000), Effects of soil fertility and land-use on forest succession in Amazonia, For. Ecol. Manage., 139,93-108. Moran, E., R. Adams, B. Bakoyema, S. Fiorini, and B. Boucek (2006), Human strategies for coping with El Nino related drought in Amazonia, Clim. Change, 77,343-361. Morisette, J. T., L. Giglio, I. Csiszar, A. Setzer, W. Schroeder, D. Morton, and C. Justice (2005), Validation of MODIS active fire detection products derived from two algorithms, Earth Interact., 9(9), E1141, doi:l0.1175/E1141.1. Morton, D. C., R. S. DeFries, Y. E. Shimabukuro, L. O. Anderson, E. Arai, F. del bon Espirito-Santo, R. Freitas, R. Freitas, and J. Morisette (2006), Cropland expansion changes deforestation dynamics in the southern Brazilian Amazon, Proc. Natl. Acad. Sci. U. S. A., 103(39),14,637-14,641. Nelson, B. W. (2001), Fogo em florestas da Amazonia central em 1997, Proceedings of the X Brazilian Remote Sensing Symposium, Foz do Igua'Yu, Brazil, 21-26 April. Nemani, R. R., C. D. Keeling, H. Hashimoto, W. M. Jolly, C. J. Tucker, R. B. Myneni, and S. W. Running (2003), Climate driven increases in global terrestrial net primary production from 1982 to 1999, Science, 300, 1560-1563. Nepstad, D., G. Carvalho, A. C. Barros, A. Alencar, J. P. Capobianco, J. Bishop, P. Moutinho, P. Lefebvre, U. L. 'Silva Jr., and E. Prins (2001), Road paving, fire regime feedbacks, and the future of Amazon forests, For. Ecol. Manage., 154, 395-407. Nepstad, D., P. Lefebvre, U. Lopes da Silva, J. Tomasella, P. Schlesinger, L. Solorzano, P. Moutinho, D. Ray, and J. Guerreira Benito (2004), Amazon drought and its implications for forest
flammability and tree growth: A basin-wide analysis, Global Change Bioi., 10(5), 704--717. Nepstad, D., et al. (2006a), Inhibition of Amazon deforestation and fire by parksrfl'hd indigenous lands, Conserv. Bioi., 20(1), 65-73. Nepstad, D., C. ¥!Stickler, and O. T. Almeida (2006b), Globalization of the Amazon soy and beef industries: OppOitunities for conservation, Conserv. Bioi., 20(6), 1595-1603. Nepstad, D. C., C.R. Carvalho, E. A. Davidson, P. Jipp, P. Lefebvre, G. H. Negreiros, E. D. d. Silva, T. Stone, S. Trumbore, and S. Vieira (1994), The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures, Nature, 372, 666-669. Nepstad, D. C., P. Jipp, P. R. d. S. Moutinho, G. H. D. Negreiros, and S. Vieira (1995), Forest recovery following pasture abandonment in Amazonia: Canopy seasonality, fire resistance and ants, Evaluating and Monitoring the Health ofLarge-Scale Ecosystems, edited by D. Rapport et aI., pp. 333-349, Springer, New York. Nepstad, D. C., et al. (1999a), Large scale impoverishment ofAmazonian forests by logging and fire, Nature, 398, 505-508. Nepstad, D. C., A. G. Moreira, and A. Alencar (1999b), Flames in the rainforest: Origins, impacts and alternatives to Amazonian fires, The Pilot Program to Conserve of the Brazilian Rainforest, World Bank, Brasilia, Brazil. Nobre, C. A., P. J. Sellers, and J. Shukla (1991), Amazonian deforestation and regional climate change, J. Clim., 4, 957-988. Pedlowski, M. A., E. A. T. Matricardi, D. Skole, S. R. Cameron, W. Chomentowski, C. Fernandes, and A. Lisboa (2005), Conservation units: A new deforestation frontier in the Amazonian state of Rondonia, Brazil, Environ. Conserv., 32(2),149-155. Penner, J. E., R. E. Dickinson, and C. A. O'Neill (1992), Effects of aerosol from biomass burning on the global radiation budget, Science, 256(5062),1432-1434. Peres, C. A., J. Barlow, and T. Haugaasen (2003), Vertebrate responses to surface fires in a Central Amazonian forest, Oryx, 37, 97-109. Potter, C., V. Brooks-Genovese, S. Klooster, and A. Torregrosa (2002), Biomass burning emissions of reactive gases estimated from satellite data analysis and ecosystem modeling for the Brazilian Amazon region, J. Geophys. Res., 107(020), 8056, doi: 10.1029/2000JD000250. Prins, E. M., J. M. Feltz, W. P. Menzel, and D. E. Ward (1998), An overview of GOES-8 diurnal fire and smoke results for SCARBand 1995 fire season in South America, J. Geophys. Res., 103(024),31,821-31,835. Pyne, S. J. (2001), Fire: A BriefHistory, 224 pp. Univ. of Wash. Press, Seattle. . Ramos-Neto, M. B., and V. R. Pivello (2000), Lightning fires in a Brazilian savanna national park: Rethinking management strategies, Environ. Manage., 26(6), 675-684. Ray, D., D. Nepstad, and P. Moutinho (2005), Micrometeorological and canopy controls offire susceptibility in forested Amazon landscape, Ecol. Appl., 15(5),1664--1678. Roberts, G., M. J. Wooster, G. L.W. Perry, N. Drake, L.-M. Rebelo, F. Dipotso (2005), Retrieval of biomass combustion nites
I
59
and totals from fire radiative power observations: Application to southern Afric~ using geostationary SEVIRI imagery, J. Geophys. Res., 110, 021111, doi: 10. 1029/2005JD006018. Rosenfeld, D. (1999), TRMM observed first direct evidence of smoke from forest fires inhibiting rainfall, Geophys. Res. Lett., 26,3105-3108. Roy, D., P. E. Lewis, and C. O. Justice (2002), Burned area mapping using multi-temporal moderate spatial resolution data-a bi-directional model-based expectation approach, Remote Sens. Environ., 83, 263-286. Roy, D., Y. Jin, P. E. Lewis, and C. Justice (2005), Prototyping a global algorithm for systematic fire-affected area mapping using MODIS time series data, Remote Sens. Environ., 97, 137-162. Sanford, R. L., J. Saldarriaga, K. Clark, C. UbI, and R. Herrera (1985), Amazon rain-forest fires, Science, 227,53-55. Schroeder, W., J. T. Morisette, I. Csiszar, L. Giglio, D. Morton, and C. Justice (2005), Characterizing vegetation fire dynamics in Brazil through multisatellite data: Common trends and practical issues, Earth Interact., 9(13), E1120, doi:IO.11751 EI120.1. Schroeder, W., I. Csiszar, and J. Morisette (2008a), Quantifying the impact of cloud obscuration on remote sensing of active fires in the Brazilian Amazon, Remote Sens. Environ., 112(2), 456-470, doi: 10.1 016lj.rse.2007.05.004. Schroeder, W., E. Prins, L. Giglio, I. Csiszar, C. Schmidt, J. Morisette, and D. Morton (2008b), Validation of GOES and MODIS active fire detection products using ASTER and ETM+ data, Remote Sens. Environ., 112, 2711-2726. Setzer, A. W., and M. C. Pereira (1991), Amazonia biomass burnings in 1987 and l an estimate of their tropospheric emissions, Ambio, 20(1), 19~22. Shanley, P., and G. Medina (2005), Frutiferas e Plantas Uteis na Vida Amazonica, 300 pp., CIFOR, Imazon, Belem. Shimabukuro, Y. E., V. Duarte, E. Arai, R. M. de Freitas, D. M. Valeriano, I. F. Brown, and M. Maldonado (2006), Fraction images derived from Terra MODIS data for mapping burned area in Acre state, Brazilian Amazonia, paper presented at the IGARSS Annual Meeting, Denver, Colorado, July 31-August 04. Shukla, J., C. Nobre, and P. Sellers (1990), Amazon deforestation and climate change, Science, 247, 1322-1325. Silveira, J. P. (2001), Development of the Brazilian Amazon, Science,292,1651-1654. Simmons, C. S., R. T. Walker, C. H. Wood, E. Arima, and M. Cochrane (2004), Wildfires in Amazonia: A pilot study examining the role of farming systems, social capital, and fire contagion, J. Lat. Am. Geogr., 3(1), 81-95. Simon, M., S. Plummer, F. Fierens, J. J. Hoelzemann, and O. Arino (2004), Burnt area detection at global scale using ATSR-2: The GLOBSCAR products and their qualification, J. Geophys. Res., 109, D14S02, doi:l0.1029/2003JD003622. Sorrensen, C. (2004), Contributions of fire use study to land usel cover change framework: Understanding landscape change in agricultural frontiers, Hum. Ecol., 32(4), 395-420. Souza, C. M., Jr., D. A. Roberts, and M. A. Cochrane (2005), Combining spectral and spatial information to map canopy
62
:II.
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
Brazil's perennial dream to "develop" Amazonia has, in many respects, been met by agriculture, at the dawn of the twenty-first century. The region now boasts, in addition to a large anq,.growing population, a farm and ranch economy that accounts for 14% of all value added in agricultural activities nationwide. Amazonian beef supplies consumers in Sao Paulo, and even China, while soybeans produced in the region are shipped from Porto Velho and Santarem, on their way to European markets. As Brazil has emerged as one the world's agricultural powerhouses, it has stood, in part, on the firm foundation of a robust Amazonian sector. Agricultural development in Amazonia has often provoked controversy, given the tremendous ecological value of the region's environment. First with ranching, and now with the soybean boom, tractors and cattle have steadily marched across lands that for millennia supported only closed moist forest, resident ecosystems, and dispersed indigenous peoples. With ongoing trends in demand for Brazil's agricultural commodities, the advance will likely continue. The present chapter considers this expansion of intensive agriculture and ranching into Amazonia, in the interest of promoting sustainability science within the broader community of Large-Scale Biosphere-Atmosphere (LBA) Experiment in Amazonia scholarship. Effective development and environmental policy for the Amazon basin must be thoroughly grounded on an understanding of the region's agriculture. The chapter is organized as follows. We start by addressing the policies that created the preconditions for modern Amazonian agriculture, the outcome of a development process that spanned 40 years and continues to this day. The region is as remote as it is large, and has therefore required sustained government intervention to release its potential. The policy discussion is followed by sectoral descriptions, first, of cattle ranching and, second, of soy farming; for each, we consider market settings, as well trajectories of expansion using data from Brazilian censuses. Note that our focus is on soy, among all crops grown in the region. Although others are economically important, soy remains the premier case of "mechanized" agriculture in Amazonia, given its ascendancy in terms of income generated, landscape impacts, and likely continued expansion. That said, the emerging biofuel economy places sugarcane on the horizon as a new crop of potential importance to the region [Sawyer, 2008]. The sectoral descriptions are data rich but do not provide a conceptual framework capable of linking market conditions to impacts on the region's environment. To accomplish this, we frame ranching and soy farming in the classical land use model of von Thiinen, which we then deploy to explain Amazonian land cover dynamics. In turn, we address these dynamics with remote sensing data from deforestation studies in Mato Grosso, Para, and Rondonia. We conclude the
WALKER ET AL.
chapter by discussing possible scenarios of agricultural advance, and possible policy responses. Before beginning, we note that ambiguity surrounds the term, Amazonia, a biophysical region that crosses a number of national borders. It is also a political entity within the Brazilian state, referred to as "Amazonia Legal," a planning region that includes both moist tropical forests and the relatively dry cerrados ofTocantins, Mato Grosso, and Rondonia. The chapter focuses on Brazil, the country that has experienced the most significant ecological, economic, and social changes within the basin. However, we recognize the importance of Amazonia to all who share in its natural wealth, and we understand that forces of development affecting the region are by no means restricted to Brazil. Throughout the chapter, we will refer to Amazonia, by which we mean the Brazilian portion. Nevertheless, much of the exposition is of relevance to Brazil's geographic partners to the basin. 2. THE POLICY ENVIRONMENT Amazonia was relatively "empty" until quite recently, and in order to understand the dramatic arrival of agriculture and ranching, it is necessary to consider the Brazilian government's strenuous efforts in this regard. These efforts unfolded through the latter half of the twentieth century and involved a variety ofpolicies, some of which targeted the region specifically, and others meant for the national economy that nevertheless have had significant impact. We now consider each ofthese in turn. As mentioned, Amazonia enjoyed considerable prosperity during the rubber boom, with substantial economic interaction both domestically and across the Atlantic. That said, our focus is on the current expansion initiated in the 1960s, a period that has been referred to as the supply-side boom [Walker et al., 2008]. 3. POLICIES TARGETING AMAZONIA
3.1. National Integration Despite long-standing calls to "develop" Amazonia, serious efforts only begin with the military regime and Operation Amazonia, a series of government actions undertaken and laws passed during 1966 and 1967 meant to accommodate both economic and geopolitical concerns [Goulding et al., 1995]. The prime objective was infrastructure investment designed to link Amazonia with the south and northeastern parts of Brazil, an objective concretely symbolized by the completion of the Belem-Brasilia Highway in 1960 [Valverde and Dias, 1967]. At first favoring the lower basin, Operation Amazonia also sought to establish a western development pole in Manaus, with the creation of the free
\
trade zone in 1967. The thrust of Operation Amazonia continued under a more f~al planning process, with the First National Development Plan (1 PND 1972-1974), which continued efforts to Ii¢<: the. region with the rest of Brazil, and to stimulate both e.¢nomic growth and colonization [Browder, 1988; Hall, 1989; Mahar, 1979]. The prime fiscal instrument, promulgated through Law 5174, provided 100% tax exemptions for enterprises investing in Amazonia, especially in agriculture and livestock. Investors also received import and export tax exemptions, subsidized credit, and access to special funds from both domestic and international lenders [Browder, 1988; Hall,1987, 1989; Mahar, 1989; Santana, 1997]. The principal beneficiaries of such early programs were cattle ranching activities, consistent with investment patterns of the World and the Inter-American Development Banks [Hall, 1989]. In 1970, Amazonian development initiatives responded to specifically humanitarian concerns provoked by the 1970 drought in the Northeast with the National Integration Plan (PIN). PIN elaborated an extensive colonization program involving a three-tiered urban hierarchy and the creation of settlements, or Projetos de Integrayao e Colonizayao, designed to accommodate large populations ofpoor farmers and landless individuals. PIN also provided new investment funds for agriculture via the Fundo para Investimento Privado no Desenvolvimento (FIDAM) and accelerated infrastructure investments. These ultimately proved of key importance to agricultural expansion into Amazonia [Browder, 1988; Hall, 1989; Santana et al., 1997]. The Belem-Brasilia Highway (BR-OIO), important in establishing an early north-south link, passed entirely through the lower basin, but its route traversed, primarily, areas of cerrado. Significant penetration of the closed moist forest did not occur until PIN in the 1970s, with the oonstruction of two highways bisecting the region, the east-west TransAmazon Highway (BR-230) and the north-south Cuiaba-Santarem Highway (BR-163), one thousand kilometers to the west ofBelem-Brasilia. Although these highways remain mostly unpaved to this very day (despite recent progress on BR-163), they succeeded in opening the inner part of the basin and its tropical moist forests to substantial agricultural development.
3.2. Agro-Industrial Expansion In 1975, the PIN program was formally abandoned, and development policy focused exclusively on agro-industrial ventures. The Second National Development Plan (II PND 1975-1979) stressed the importance ofthe Amazon basin for generating foreign exchange and called for continuing infrastructure investments, as well as the promotion of exportoriented activities such as ranching, timber, and mineral
63
extraction [Browder, 1988; Hall, 1987, 1989; Mahar, 1979; Santana et al., 1997]. This was accomplished via the Programa de Polos Agropecuarios e Agrominerais da Amazonia, designed to attract investment with subsidies and tax relief. PIN's fiscal instrument, FIDAM, gave way to a mutual fund providing a wide range of incentives, the Fundo do Investimento da Amazonia (FINAM). As with earlier credit programs, FINAM paid special attention to cattle ranching, and by 1980, the region's federal planning agency at the time, Superintendencia de Desenvolvimento da Amazonia, had approved 469 cattle-related projects involving US$565 million [Hecht et al., 1988]. In 1980, the government initiated the Programa Grande Carajas (PGC) to tap the tremendous mineral wealth of the Carajas iron ore reserves, expected to satisfy around 7.5% of world iron demand [Hall, 1987, 1989]. The program also included an agricultural component and set aside tracts of land for this purpose. It is within the context of PGC Agricola that significant mention is first made of Amazonian soybeans and the issue of foot-and-mouth disease (FMD), which affected the market potential ofAmazonian beef at the time. Perhaps as important to the region's agricultural development was PGC's continuing emphasis on infrastructure. Under its auspices, two ports were constructed, a 900-km railway and the famed Tucurui hydroelectric dam on the Tocantins River, the largest ever built in a tropical rainforest region [Hall, 1987; 1989].
! 3.3. The Democratic Abertura In 1979, at the end of the II PND period, Brazil experienced the second oil crisis, which severely impacted proactive development initiatives. Thus, although a third national plan was conceived and implemented (III PND), it remained largely a paperbound effort. The World Bank did intervene with funding for the investments in Polo de Desenvolvimento Noroeste (POLONOROESTE) in 1982, aimed in large part at paving BR-364 through Rondonia [Woodward, 1988]. This significantly opened the western basin to development in a much more dramatic manner than the establishment of the free trade zone in Manaus in 1967. In fact, rapid deforestation associated with POLONOROESTE helped shift Amazonian development policy, in the early years of democracy, away from agricultural developmerit, toward long simmering concerns about the environment and indigenous populations. Thus, the First Amazon Development Plan of 1986 cut subsidies and tax incentives to agricultural interests, at the same time as it promoted extractive reserves and addressed Brazil's perennial issue ofland distribution and rural poverty. In the wake of reform following restoration of democracy in 1985, an extensive system of protected areas emerged,
64
lIf'
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
the National System of Nature Conservation Units (SNUC), meant to conserve both cultural and natural resources [Simmons, 2002). This legacy continues to evolve with recent additions s'bch as lands in the Terra do Meio in central Para state. Here, between 2004 and 2006, the federal government created the Reserva Extrativista do Riozinho do Anfrisio (7360 km2), the Estayao Ecol6gica da Terra do Meio (33,000 km2), the Parque Nacional Serra do Pardo (4,450 km2), and the Reserva Extrativista do Iriri (3990 km2). The govemment also grew more aggressive in defending indigenous land rights that had been neglected in the heady days of regional occupation spearheaded by the military regime. At the present time, substantial tracts of Amazonian forest remain intact in indigenous reserves, the conservation units of the SNUC program, and efforts by individual states to set lands aside from the path of development. 4. AMAZONIAN SPILLOVERS FROM BRAZILIAN POLICY AND ECONOMIC REFORM 4.1. The Plano Plurianual Process
The restoration of democracy ushered in a new age of national planning, designed to boost Brazilian economic performance overall, but with significant spillovers for Amazonia. The constitution of 1988 implemented the plano plurianual (pPA), a planning cycle to be undertaken by successive federal administrations. The first PPA with substantial impact on Amazonia was II PPA (1996-1999) or Brasil em Ayao. This was followed by III PPA (2000-2003), or Avanya Brasil, and IV PPA (2004-2007), or Brasil de Todos. Presently, the Lula da Silva administration is implementing a successor PPA referred to as 0 Programa de Acelerayao do Crescimento (PAC; 2007-2010). The PPAs are largely directed at infrastructure projects for roads, ports, waterways, airports, and electricity. In general, programs across administrations show continuity, at least with respect to Amazonia, with Brasil de Todos and Lula's PAC initiatives completing and extending earlier initiatives under Brasil em Ayao and Avanya Brasil. Although web resources are available to monitor progress (e.g., www.brasil.gov.br/pac/). assessment of implementation is difficult, given the complexity and range of tasks involved. Walker and Reis [2007] have released maps to the LBA data archives that enable a digital evaluation of the extension of the highway system pursuant to these programs, which grew from 400 km in 1968, the extent of Belem-Brasilia in the Amazonia Legal, to 56,663 km in 2001. In 1968, the system was entirely federal, whereas by 2001, state contributions to road building, enabled by the initial federal infrastructure, accounted for 37,410 km. That said, road-paving has comprised an important component
ti
E., ~.'
F
of the PPAs, and most all-season thoroughfares are found in the federal component of the system. Of 20,168 km of paved roads, 12,962 are federal. Transportation infrastructure investment has long been regarded as a key to economic development [Vance, 1986; Owen, 1987). 4.2. Macroeconomic Reform
In addition to the PPAs, administrations starting with Fernando Collor, and up to the present day with Luiz Inacio Lula, have pursued macroeconomic reforms aimed at boosting Brazil's competitiveness. These reforms, involving privatization of state enterprise, reduced public expenditures, and monetary reform, are consistent with the economic and political transformations that have been referred to as neoliberalism [Liverman and Vila, 2006). Although neoliberalism is not a monolithic matrix of institutional changes, macroeconomic reforms in the Brazilian case present a classic case of state retraction from economic enterprise, the cornerstone of neoliberalism [Mollo and Saad-Filho, 2004). Another important component of Brazil's neoliberal transformation has been monetary and fiscal policy, especially the Plano Real implemented by Fernando Henrique Cardoso. The Plano Real and its successor programs played and continue to playa critical role in the region's agriculture through currency stabilization and the control of inflation, which reached dizzying heights in the early 1990s. Although Brazilian currency initially appreciated under the plan, devaluation in 1999 with free-floating exchange rates opened the floodgates to international demand for Brazil's agricultural commodities, accelerating exports from Amazonia and the country as a whole [Brandao et al., 2005). 4.3. Innovation and Sanitary Controls
Technological innovation has played a crucial role in facilitating the penetration of Amazonia by mechanized agriculture and ranching. In the case of soy, early Brazilian farming exploited higher latitudes due to phenologicallimitations of the crop and its harvest cycle. The first cultivars developed specifically for low-latitude tropical areas were Tropical, BR-1O (Teresina), and BR-ll (Carajas) [Kiihl et al., 1985). More recently, Empresa Brasiloirade Pesquisa Agropecuario (EMBRAPA) has produced soy plants oflongjuvenile period such as BRS Milena, BRS Celeste, BR-27 (Serid6), BR-28 (Cariri), Embrapa 9 (Bays)1 and Embrapa 30 (Vale do Rio Doce). Since 2003, use of genetically modified seeds has become widespread, and from 2004 to 2007, genetically modified soy increased its domain from 34% to 65.9% of the total area planted (Soystats at www.soystats. com). Seed innovations, combined with fertilizer and lime-
WALKER ET AL.
stone applications and the adoption of no-till planting techniques, have raised Amazonian productivity to about 3 t ha-I, nearly 30% more th.ilti Brazil's average in2005 [IBGE-PAM, 2005; Almeida et!Ll., 1999; Souza et al., 2000a, 2000b). Innovations in ra~ng have been equally important, particularly with respect to pasture productivity. When development first began in Amazonia, pasture degradation proved a severe challenge, with weed infestations and declines in soil productivity leading to early abandonment [Goulding et al., 1995]. This has been largely resolved by the use of new forages. The expansion of Brazilian ranching into the cerrados ofthe center-west from Sao Paulo State and points south was enabled by the adoption and modification of African grasses. Amazonian ranchers soon followed suit. The improvement of forages has significantly improved the economic prospects of ranching in the region, thereby facilitating its expansion. The shift to planted pastures has been remarkable, throughout Brazil and especially Amazonia, where they increased by nearly 62% between 1985 and 1995 [Cattaneo, 2005]. In addition to pasture grass improvement, cattle ranching in Amazonia, and Brazil more generally, has benefited from concerted action by federal and state govemments to eliminate sanitary barriers to Brazilian beefin domestic and foreign markets [Walker et al., 2008]. This has been accomplished by controlling FMD. To keep their own stocks healthy, importers from many countries require certification by the International Organization ofAnimal Health (the OlE) that beef come from areas free of FMD. Brazilian efforts to eradicate FMD have followed a multipronged approach designed to create diseasefree zones, using vaccines that, if repeated every six months, effectively control it. Brazil also defines areas, or circuitos, based on likelihood of disease presence, ranging from zones of unknown risk to those completely disease free [Arima et al., 2005]. Since the late 1990s, this program has dramatically altered the prospects for Brazilian, and Amazonian, export. As recently as 1998, only two southern states (Rio Grande do SuI and Santa Catarina) were certified for international export by OIE. By 2005, certification had spread to almost all nonAmazonian states, as well as to Acre, Rondonia, Mato Grosso, and Tocantins, and considerable success had been met in controlling the disease in the State of Para. It is important to point out, however, that key consumers such as the United States and Japan require exporters to be entirely free ofFMD within their borders and, therefore, do not import beef in natura from Brazil. That said, many international markets have opened to Brazilian products, especially in Europe and the Middle East. 4.4. Policy Impacts Overall
A variety of policy initiatives, targeting both Amazonia and the Brazilian economy at large, have enabled and pro-
65
moted the expansion of intensive agriculture and ranching into the far reaches of Amazonia. Initiatives include, but are not limited to, infrastructure development, macroeconomic policy (both monetary and fiscal), crop and pasture grass improvements, and sanitary regulations. Given the broad spectrum of efforts involved, their application over a period of decades, and inherent difficulties in assigning cause and effect in the growth of an economy, it is problematic to determine which has proved decisive. Prima facie, it seems reasonable to attribute the agricultural expansion of the past four decades to policy synergism. Nevertheless, we make several pointed remarks in the interest of identifying specific effects of certain importance. First and foremost, of course, has been the extension of federal highway infrastructure, which lowered transportation costs to destinations outside the region [Pfaff, 1999; Simmons et al., 2007; Walker et al., 2008). This has been a long-term process spanning four decades. Part and parcel of this has been investment in infrastructure more generally, including air and water transport, as well as electricity. Although analysis of the impact of utilities on agricultural expansion into Amazonia is problematic, efforts have been made to identify transportation cost reductions, and they are substantial [Simmons et al., 2007; Walker et al., 2008). The travel time from ~elem to Sao Paulo, for example, dropped nearly 50% between 1968 and 1995, from about 100 h to just over 50 h by U;Uck. In that these calculations postdate the completion of Belem-Brasilia, and do not account for road investment since ~995, the total effect for the development period is probably much more significant. As for macroeconomic policy, much has been written about the early role of fiscal incentives in attracting capital from the south [e.g., Hecht, 1985]. This impact was probably overstated, given early agricultural investment in the absence of incentives [Walker et al., 2008] and continuing development under a regime of reduced subsidies [e.g., Helfand and Castro de Rezende, 2001a, 2001b; Cattaneo, 2001, Margulis, 2004]. A greater stimulus likely derives from monetary reform, and the Plano Real in particular, with the control of inflation and a free-floating exchange rate starting in 1999 that devalued the real against the dollar. That said, analyses have suggested that lowered exchange rates would reduce Amazonian agricultural expansion [Cattaneo, 2001]. Such analyses appear to be based on an assumption that Amazonian agriculture produces only for domestic markets, which were shocked by the income reduction of devaluation in 1999, leading to lower internal demand [Cattaneo, 2001; Wiebelt, 1995). In fact, the market situation has changed dramatically in only a few years, with Amazonia providing a large share of all Brazilian beef exports, in natura, not to mention the growth of export-oriented, soy farming in Mato
66
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
Grosso. Explosive growth in the production of soy and cattle is traceable precisely to the 1999 devaluation, with a tremendous buildup of herds in the north and soy planting in the center-~est, presumably Mato Grosso [Brandiio et al., 2005]. The dramatic arrival of Amazonian beef products on world markets was no doubt enabled by the fortuitous successes with FMD control, pursuant to policies enacted in the 1990s and discussed above. Walker et al. [2008] show that FMD infections have fallen by orders of magnitude since the Military era, with tens of' thousands of annual cases in the 1970s, to fewer than a hundred per year in the early part of the twenty-first century. We now tum to the implications of FMD control, and its companion policies, for the Amazonian herd. 5. CATTLE RANCHING Ranching has a very long history in the Amazon basin, starting hundreds ofyears ago on the extensive natural grasslands of Maraj6 Island at the mouth of the Amazon River [Arima and Uhl, 1997]. Nevertheless, when the region was opened to significant investment and migration in the early 1970s, its livestock sector represented a little more than 8% of the national herd. Several decades of expansion have changed this dramatically, and presently, more than 70 million animals populate the Amazonian landscape, accounting for over one third of Brazil's entire stock of 185 million head as of 2005. This growth has been widely dispersed across the region, although much of it is concentrated in Pani, Mato Grosso, and Rondonia, three states with a combined herd of 56 million animals (Table 1). Initially, Amazonian ranchers supplied the frontier's growing urban population, and very few meat packers sold to national markets [Faminow, 1997]. Now, nearly 90% of regional output leaves for other parts of Brazil, and Mato Grosso, Pani, Tocantins, Rondonia, and
Table 1. Growth in Cattle Herd Amazonian States Number of Cattle"
Acre Amapa Amazonas Mato Grosso Para Rondonia Roraima Tocantins
1990
2005
Average Annual Growth (% a-I)
400,085 69,619 637,299 9,041,258 6,182,090 1,718,697
2,313,185 96,599 1,197,171 26,651,500 18,063,669 11,349,452 507,000 7,961,926
30 2 5 12 12 35 NAb 5
o 4,309,160
"Date source: IBGE, Pesquisa Pecuaria Municipal. ~A, not available.
Acre all possess slaughterhouses with international trade permits [Arima et al., 2005]. Growth in Amazonia's share of Brazilian export has been explosive, as just discussed, nearly doubling in 5 years from a little over 5% in 2000 to almost 10% by 2005 (Table 2). These beef products satisfy demands around the world, with markets in Latin America, the European Union, the Middle East, and Asia, including China [Arima et al., 2005].
WALKER ET AL.
67
Table 2. Export of Brazilian and Amazonian Beef, 2000-2005" Brazil,'~ ! Am on A zon (%)
2000
2001
2002
2003
2004
2005
188,245 10,000 5.31
367,765 25,427 6.91
429,811 29,977 6.97
663,165 44,676 6.74
924,764 53,139 5.75
1,078,355 106,185 9.85
"Values given in metric tons. Boneless meat, refrigerated and frozen. Data source: ABIEC (www.abiec. com.br/abiec/estatisticas/voll_export.htrn (accessed 2 August 2004).
5.1. Ranching "Economics"
The expanding market situation, together with lowered production costs stemming from the synergies of the policy process elaborated above, has made Amazonian ranching profitable [Walker et al., 2000; Zen, 2002; Margulis, 2004; Arima et al., 2005; Piketty et al., 2005; Walker et al., 2008]. This might seem surprising, given earlier arguments that cattle ventures in the region were little more than corporate adventures, focused on subsidy capture and land speculation, not on productive land use [e.g., Hecht, 1985; Hecht and Cockburn, 1990]. To the contrary, Amazonian producers make good money by providing for domestic consumption, which enables southern producers to engage in a growing international trade, and by directly participating in the global markets themselves [Kaimowitz et al., 2004; Pacheco, 2005]. Given that FMD still limits the involvement of many Amazonian states in international trade, producers here have, to this point, played mainly a support role in the globalization of production in the south [Nehmi Filho and Pusch, 2003]. This circumstance is changing rapidly, with continuing sanitary improvements in the northern herds. Indeed, the natural incidence of diseases (FMD, Brucelosis) and ectoparasites is less in key Amazonian cattle regions than in southern and central parts of Brazil [Arima and Uhl, 1997]. Amazonian producers are poised to profit from the 8% to 10% price premium available through international trade. Early Amazonian ranchers exploited the natural pastures on the eastern part of Maraj6 Island and the extensive floodplains (varzea) of the Amazon river course. More recently, ranching has shifted to the uplands, or terra firme, both in areas of cerrado and closed forest. Although ranching in this environment originally mimicked the low input, low technology systems found in the natural grasslands, it has grown significantly more productive, and profitable, over time. Now, terra firme ranchers till their pastures of improved forage grasses, monitor their herds on paddocks, and rely on veterinary and technical expertise for vaccines and genetic materials. Further, they have rationalized their production systems by integrating specialized operations across flexible and efficient production chains [Poccard-Chapuis et al., 2005]. Smallholders and seasonally inundated lowland sites pro-
duce calves (cria) to be fattened (recria and engorda) on the improved pastures of largeholders on terra firme [PoccardChapuis et al., 2005; Arima and Uhl, 1997]. Profitability of terra firme ranching has recently been analyzed by various researchers [e.g., Margulis, 2004; Arima et al., 2005]. For example, Arima et al. [2005] compared economic performance of large operations (>5,000 head) in key Amazonian locations to traditional southern ranching districts. Their findings establish that Amazonian producers enjoy a higher rate of return on investments than their southern counterparts, due to low land prices and the resource base, with its abundant rainfall and solar energy. This advantage holds even though product prices in the south ate 10% to 20% higher given locations close to the prime markets. Although Amazonian soils are mainly Latosols with low fertility and high acidity, the fact remains that they are not much worse than those found in other cattle-producing parts of the country [Falesi, 1976; Fearnside, 1980; Adamoli et al., 1985]. Good moisture conditions, high insolation, and lack of frost compensate for soil limitations and allow for growth rates higher than elsewhere in the country [Margulis, 2004; Arima et al., 2005; Anualpec, 2003; Arima and Uhl, 1997]. Depending on type of operation (calving, fattening, etc.), large ranches in Amazonia enjoy 10-16% higher rates of animal growth, which translates into 5 to 10 additional kilograms of meat produced per hectare per year than on southern ranches (58-82 kg live growth ha- I versus 53-74 kg live growth per ha- I ). Such productivity advantages translate into higher profits, given low land prices [Arima et al., 2005; Anualpec, 2003; Barros, 2002]. The Arima et al. [2005] study documents land price differentials between north and south [see also Sawyer, 2008]. In Tupa, Sao Paulo, one of the traditional centers of southern production, they are nearly three times higher than in many important Amazonian production sites; here, a hectare of land goes for R$3300 compared to about $R1250 in Amazonia [Arima et al., 2005; Barros, 2002].Given ranches range into the tens of thousands of hectares, such a price differential represents a very large difference in capital costs reaching into the millions ofD.S. dollars for individual operations [A rima and Uhl, 1997].
With productivity advantages and lower costs for land, ranchers in Amazonia enjoy a higher profit rate than anywhere else in Brazil, despite the remoteness oftheir locations from major southern markets, and the impact of transportation costs on the prices they receive. The Tupa site yields a 4% internal rate of return (IRR); this does not compare well to Amazonian production, which yields an IRR nearly three times higher, at 12%. Evidently, Tupa is impacted by land costs that reflect the significantly higher rents obtainable from intensive agriculture [Arima et al., 2005], but is not an outlier among other southern locations in terms of low profit potential. Aggregate municipal calculations show an Amazonian return on investment at 5%, exceeding the average over all other states at 3.37% [Arima et al., 2005; Anualpec, 2003]. This reflects the fact that ranching is profitable across the board, for larg~ and small operations alike [Arima et al., 2005; Topall, 199~].
i 5.2. The Market Situation I
The profitability of Amazonian ranching is likely to translate into sectoral growth as markets continue expanding. Market expansion, in tum, will be driven by changes in supply and demand. As for the supply side, the dramatic insertion of Brazilian cattle 'products into the global market place is, in large part, a result of focused efforts to eradicate FMD. Recent outbreaks in Mato Grosso do SuI, Para, and Amazonas have dampened the initial upward spiral of Brazilian export, and 49 importing countries now impose certain restrictions on Brazilian products. Nevertheless, over the midrun exports are likely to surge again given that the infections appear to be the result of poor application of sanitary procedures rather than a new strain of the disease. Trade restrictions have generally not applied to all Brazilian states, and those unaffected by outbreak are continuing to export [Arima et al., 2005]. Since room for herd expansion is limited in cattle countries, such as Australia and Argentina, and since fear of bovine spongiform encephalopathy (BSE), or "I11ad cow" disease, restricts U.S. trade, Brazil, with its "undeveloped" lands, is the only country poised to supply any significant growth in world demand [Arima et al., 2005].
68
WALKER ET AL.
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
As it turns out, these markets can be expected to expand such protection, so once the subsidies are removed, they will significantly over the next few decades. As is well-known, find their goods quite favorably priced [Arima et al., 2005]. meat is a "superior" good in economic terms, which means The fact that Brazilian cattle are mainly range-fed will make that as incljIDes rise, consumers tend to eat more of it. Thus, their meats even more desirable, since concerns continue to extensive scope exists for growth of demand given increas- linger about BSE. Given the role that Amazonian production ing incomes in China, and in Brazil itself, countries that will plays in both national and international markets, and given add to the population of consumers searching for meat in su- the profitability of Amazonian ranching, this will no doubt permarkets around the world. In China, yearly consumption translate into considerable expansion of the region's herd. of beef is 4 kg per person on average, which compares to 44 kg per person in the United States [U.s. Department of 5.3. The Dynamics ofthe Amazonian Herd Agriculture, 1997]. Clearly, the Chinese will ultimately exThe evolving economic environment faced by ranchers in perience some of their economic boom at the dinner table. the Amazon basin has enabled their steady march across the Further intensifYing the potential demand for Brazilian region's wild areas. Table 1 shows this by state, for the years beef is the weakening of political support for agricultural 1990 and 2005. In 1990, the Amazonian herd of 18 milsubsidies in developed countries, which keeps their prices arlion animals was widely dispersed, and four states already tificially low and competitive. Brazilian producers enjoy no
possessed counts exceeding a million head (Mato Grosso, Para, Rondonia, and Tocantins). By 2005, the pattern had dispersed further, a~ ranching occupied the far-flung corners of the basin. ,~oreover, the regional herd had reached 74,000,000, a sto~ larger than found in most cattle-producing countries. By M05, Acre joined the four other states supporting over a million animals, and Roraima, the only state without cattle in 1990, had a herd exceeding 500,000. Figures 1 and 2 graphically illustrate herd sizes throughout Amazonia Legal for the years in question (1990 and 2005). As can be seen, the early distribution suggests an advance starting in cerrado areas, with sizeable herds found in the southern and the eastern parts of the basin. A nearly continuous arc ofmunicipalities supported cattle herds, ranging from Paragominas in the northeast (para State), down through the south of Para, Tocantins and Mato Grosso, then west into Rondonia. The
Legend (In Heads)
Legend (In Heads) c:=J No data or No Production
D
No data or No Production
. . 1-100,000 _ 100,001 - 200,000
_
1 - 100,000 100,001 - 200,000
_
300,001 - 400,000
_
400,001 - 500,000 > 500,000
_
_
200,001 - 300,000
Source: IBGE, Pesquisa Pecuaria Municipal
_
300,001 - 400,000
o
_
400,001 - 500,000 > 500,000
200,001 - 300,000
_
legacy ranching on the natural grasslands of Maraj6 Island is observable, as are several anomalous locations beyond the initial cattle frontier, such as Itaituba in central Para, Jurua in northern Mato Grosso, and Rio Branco in Acre. The current distribution of the Amazonian herd shows a consolidation and advance of the frontier. As can be seen in Figure 2, the cattle arc is now completely continuous,joining up municipalities from Para all the way to Acre, thousands of kilometers to the southwest. In addition, the cattle-producing area now contains practically the entire state ofMato Grosso, as well as a sizeable portion of Para, such that nearly half the basin shows appreciable production. The only large remaining area with a few animals is in Amazonas, although Roraima and Amapa have yet to develop significant herds. Nevertheless, ranching has jumped the Amazon River course in western Para state, and cattle now forage in Monte Alegre,
Cattle by Municipality The Legal Amazon - 2005
Cattle by municipality The Legal Amazon -1990
250
I
Figure 1. Cattle herd by municipio, 1990.
500 I
1,000 Kilometers I
_
69
Source: IBGE, Pesquisa Pecuaria Municipal
o
250
I
Figure 2. Cattle herd by municipio, 2005.
500 I
1,000 Kilometers
I
70
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
Alenquer, 6bidos, and Oriximina, to the north of Santarem. The dynamism of the Amazonian cattle economy is met by mechanized agriculture, our next topic of discussion. ~
6. THE EXPANSION OF SOY FARMING Like ranching, agriculture also has a long history in Amazonia: and its first riches were created, in part, by farming in the lower basin, where Portuguese colonists used slaves to cultivate or extract cocoa, coffee, cotton, and sugar cane [Santos, 1980]. That said, the current shape of farm production is a far cry from its early antecedents, and soy cultivation, in particular, represents a dramatic addition to Amazonian agriculture. Although the region produces many different crops, soy dominates by far in terms of revenues, yielding a 2005 crop valued at 7.8 billion reais, or about 61 % of the gross value of the harvest of annual crops in the region. Thus, it is a little wonder that soy, produced by highly capitalized production systems modernized far beyond what has traditionally passed as Amazonian farming, has come to symbolize the advancing edge of mechanized agriculture. The Amazonian soy boom of the past decade comprises an important part of national growth overall, which has expanded from an annual production of 20 million t in 1990, to 50 million t in 2004. This has led to an increase in soybean area planted from 115,847 to 215,972 km 2 in all Brazil. Such increases have made Brazil, the harvester of 28% of the global soybean crop, the world's second largest producer and exporter, supplying 27% of the world total. In comparison, the United States, the world's largest producer, supplies about 35% [USDA-FAS, 2004; IBGE, 2005]. Although traditional soy-producing states, such as Rio Grande do SuI, Mato Grosso do SuI, Sao Paulo, Minas Gerais, and Santa Catarina, supplied 54% of Brazil's total production in 1990, they had lowered their share to 37% by 1999. Currently, 33% of Brazilian soybeans are harvested in Amazonia Legal [CONAB, 2003; IBGE, 2005], a number that has grown dramatically from 24% in 1998. Evidently, this redistribution of production indicates a migration of soy production to the north, and the incorporation of the cheap, vast lands of the cerrado found there [Castro et al., 2001]. 6.1. The Supply Side: Infrastructure and New Cultivars
As with ranching, the boom of industrial agriculture in northern Brazil has been stimulated by auspicious changes in supply and demand, the latter ofwhich is linked to domestic and global markets. On the supply side, infrastructure investments have promoted soy production by bringing about a precipitous decline in transportation costs. As has already been discussed, the development of road networks has cut
days off travel times between northern and southern destinations in Brazil. The ports, waterways, and rails developed under various federal initiatives have also been especially important for Amazonian soy production. The establishment of major transshipment points on the Rio Madeira in Porto Velho, and a deep water port on the Amazon River itself, in Santarem, has decreased transportation costs for soybean produced in the upper basin. In eastern Amazonia, the Ferrovia Norte-SuI (north-south railway) now provides a rail link to connect soy producers in Maranhao and Tocantins to the Port of Itaqui on the Atlantic coast. Besides infrastructure expansion, the recent development of soybean cultivars suitable for hot humid conditions has proven decisive. Soybeans are naturally short-day plants adapted for growth in subtropical and temperate areas. Thus, as with ranching, early thinking about soy farming in Amazonia emphasized environmental limitations, and a popular and scientific consensus emerged that climatic conditions would ultimately inhibit the development of a robust soy economy above latitude 25° [McGrath and Vera-Diaz, 2006; Jordan, 1982; Sioli, 1973]. Adding further support to this general view was the fact that Brazilian soybeans were originally cultivated with great success between 20 0 S and 300 S, where U.S. cultivars were well adapted to the local climate and soil [EMBRAPA-SOJA, 2002]. Nevertheless, genetic modifications have opened the door to Amazonian production, and the current expansion into low latitudes is possible with new cultivars possessing long-juvenile genes, which delay flowering and maturity. Without long-juvenile genes, soybean plants grown at low-latitudes flower too soon; this makes them short and difficult to harvest mechanically [Hartwig and Kiihl, 1979; Sinclair et al., 2005; VeraDiaz et al., 2008]. 6.2. Growing Demand
As for growth in world demand, this has, in large part, been sparked by robust economic expansion in China, and the ri.sing global consumption of vegetable oils and soyconsuming poultry, swine, and livestock. Today an estimated 30% of the world's vegetable oil consumption and 70% of protein meal consumption are derived from soy (Soystats at www.soystats.com). which provides an effective and sanitary substitute for animal parts, the primary vector of disease transmission in animal-fattening operations [Rohter, 2003; Vera-Diaz et al., 2008]. Between 1990 and 2000, soybean global demand grew 68% (from 104.2 to 175.2 million t), and global consumption of soybean grain increased 41 % (from 104 to 146.7 million t) [AGRIANUAL, 2000; RCW, 2004]. During this same period, Brazil expanded its share of the global market from 15% to 22% (15.4 to 38.4 million t)
WALKER ET AL.
-
i
and exported 64%. of its 'production. In addition, domesti c soybean consumptiOn doubled, from 6.6 to 13.6 million 1. .A num.ber of stud~p6 suggest that global demand for soy Will contmue to]rw o.ver the next few decades. As with beef, many soy p ducts are considered "superior" goods and eve~ conse - ative predictions from the United Nation; PopulatIOn Division [2004] and RCW [2004] suggest that the . demand for soybeans will increase from 225 .6 ml'11'Ion t m 2001 to 385 million tin 2020. Future Brazilian dynamics have also been a~dressed, considering production and export figures ~bserved m th~ 1990s. In particular, Brazil's soybean productIon could easily - grow to 73 million t by 2020 ,M'th ~ore than 55 million t exported to global markets [Rodrigues, 2004].
71
ticularly Roraima, where production, nonexistent in 1998, reached 36,000 thy 2005. Only Acre and Amapa appear to ?e unaffected by the soy boom, although output remains low m Amazonas (5136 t). . T~ese spatial dynamics are shown for a longer time period m Figures 3 and 4 (1990 to 2005). Here can be observed the advance of the soy frontier, first in the central and southern cerrados of Mato Grosso, where it is highly concentrated in 1990: Although the some soy was produced in Tocantins and m Maranhao, as well as in the border areas between Mato Grosso and Rondonia, Amazonian soy fanning in the early 1990s. was largely a single-state phenomenon. This changes rad~cal~y b~ 2005 (Figure 4). Clearly, the majority ~f soy fa~mg IS stIll found in Mato Grosso, with productIon practIcally everywhere in the state excepting areas in 6.3. The Dynamics o/Soy Production the bordering Rondonia"Para and az oAm nas. D far. northwest . esplte ItS growth and concentration in Mato Grosso soy . The explosive growth of soybean production in Amazo' ma from 3 to 2.0 million t a-I between 1990 and 2005 has has taken root throughout the Amazon basin. As for Mato Grosso, the figure reveals a substantial westbeen accomp~med by increases in area planted from 16,000 ward ;n?vemen~of farming into the southeastern parts of to 70:000 ~ [IBGE, 2005]. Disaggregated state dynamics Ro.ndoma, c~eatIng, with Mato Grosso, a nearly continuous ~re given m Table 3 for the period from 1998 to 2005 durStriP of soy m southern Amazonia Legal, buffered only b mg which Pro~uctio~ became a region-wide pheJ1om~non. we~lands to the south. In addition, several widely disperse~ Soy had estabhshed Itself by 1998 in both Mato Grosso and fOCI hav_e emerged. ~rom the east in the states of Bahia and Rondonia, although Mato Grosso clearly dominated with an Maranhao, soy fa~mg now merges into the significantly output of about 7 million t. By 2005, the regional pattern had expan~ed pro~uctlOn zone of Tocantins. These areas, in ch~nged. Soy farmers in Mato Grosso significantly increased With th:!;: Mato Grosso croplands. The graphical tum, Imk up their output to nearly 18 million t, but Rondonia and Para t~e soy frontier could overrun the native data suggest that both ex.ceeded 200,000 t of production, with output in Para cerrados of the basip., which occupy its southern and eastern expand.mg al~ost ten times over the 7-year period. Further, flanks [see Mueller, 2003]. Tocantms, With a reasonable production in 1998 (123,085 t), Besid~s .Mato Grosso, Rondonia, and Tocantins, Para bec~~e the second largest Amazonian producer, with nearly sho~s slgmficant emergent production, with rather high out1 milhon.tons (905,328). States producing less than 100,000 t puts m Paragominas and Santarem. Incipient soy farming is by 2005 mclude Acre, Amapa, Amazonas and Roraima Of observable, along the Transamazon Highway (e.g., Altamira these, Roraima and Amazonas showed s;ong growfu, ~ar~nd Uruara) an~ now forms a corridor along BR-163 linkmg the ~rod~ctlOn areas in Mato Grosso with Santarem. Soy, farmmg IS also found in five municipios in the south of Table 3. Growth in Soy Production Amazonian States' Pa.ra_(Santana do ~raguaia, Santa Maria das Barreiras, ConAve Annual ce19ao do AragUa/a, Redenyao, and Floresta do Araguaia) 1998 States Growth (% a-I) a~d across the Amazon River from Santarem, in Alenquer 2005 300 Acre Fmally, the cerrados to the far north in Roraima now sup~ 114 -0.0 Amapa 0 port a. crop, as do the Amazonas municipios across the Rio Amazonas 0 0.0 68 796 5136 Madeira from Porto Velho (Humaita) and in close proximity Mato Grosso 7228052 17 ,761,444 " 18 to Manaus.
Para Rondonia Roraima Tocantins
2438 , 15 ,970 0 123,085
204,302 233,281 36,400 905,328
1034 172 NAb 79
'V~I~es given in tons. Data source: IBGE, ProdUl,;ao Agricola MUlliclpal (PAM). ~A, not available.
7. IMPLICATIONS FOR THE AMAZONIAN LANDSCAPE .Agri.cultural development is of great importance to Brazil gIVen ItS comparative natural advantages. A long growing season and cheap land have conspired to make it a world
72
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
WALKER ET AL.
73
Soybean in the Legal Amazon Planted Area - 2005 "
Legend (In Hectares)
Legend (In Hectares)
CJ No data or No Production
CJ No data or No Production
_
1 - 250 251-500
_
1 - 250 251 - 500
_
501 -1,000
_
501 -1,000
_
1,001 - 5,000
_
5,001 - 10,000
_
10,001 - 25,000
_ _
25,001 - 50,000 >50,000
source: IBGE, Producao Agricola Municipal (PAM)
o I
255
510 I
1.020 Kilometers
I
Figure 3. Planted soy area by municipio, 1990.
powerhouse in the production of agricultural commodities. But agriculture possesses a land-demanding production function, which means that when it expands into forested areas, trees give way to crops and pasture grasses over large areas. The expansion of pasture into Amazonia has long generated controversy in this regard, and it is a fact that pasture constitutes the lion's share of cleared lands in the north region [Walker et al., 2008]. The question is: Can we expect such agriculturally driven clearance to continue, and if so, by how much? Earlier sections of this chapter addressed the demand picture for both cattle products and soy. Projections of impacts on the Amazonian landscape require that this picture be distilled into an estimate of the demand for land.
7.1. Agricultural Expansion and Von Thiinen
In tackling such a challenge, it is useful to situate Amazonian agriculture in a conceptual framework that links commodity demands to the input of land for production. To this end , we consider the model of von Thiinen, who pointed out that (l) agricultural activity occurs so long as rents are positive, and that (2) rents are functions of prices for the products of land and inputs to their production. Von Thiinen also noted that landscapes reveal spatial patterns of crop locations with intensive forms of land use found in nearer areas of ~opulation concentration, and extensive ones, farther away.
_
1,001 - 5,000
_
5,001 - 10,000
-
10,001 - 25,000
_
25,001 - 50,000 > 50,000
source: IBGE, Producao Agricola Municipal (PAM)
o I
255
510
I
1,020 Kilometers
I
Figure 4. Planted soy area by municipio, 2005.
With the Thunian framework, we conceptualize the expansion of the agricultural frontier into Amazonia as being driven by increasing rents [Walker, 2004; Walker et al., 2008]. These rents are bolstered by prices for farm and ranch products stemming from globalizing demand, and by production cost reductions due to improvements in the transportation system [Mueller, 2003]. At the lead edge is ranching, found far from population centers because it generates rents far from market centers. Behind ranching comes the complementary advance of soy and mechanized agriculture. more generally. Under a Thunian formulation, deforestation is the manifestation of an advancing agricultural frontier, occurring when potential rents, previously nonexistent by virtue of market or infrastructure conditions, become positive [Walker and Solecki, 2004].
Two issues must be addressed before we consider Amazonian land cover dynamics within this conceptual framework. The first is that of agricultural intensification. Simply put, intensification is the adoption of new farming practices or technologies that raise output per unit land. Consequently, intensification leads to reduced demands for land, ceteris paribus, and for this reason, many have appealed to it as a solution to the problem of deforestation. The second issue involves the mechanisms of forest loss under a multicrop Thunian system and specifically the role of soy expansion in driving Amazonian deforestation, given that cattle ranching is an active partner. As for intensification, it is often imposed on a farmer or farming group by virtue of land scarcity, as has been extensively observed in the historic record [Boserup, 1969]. For
74
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
the Amazonian case, fanners and ranchers are unlikely to intensify production by much, given the abundance of cheap land, even with new technologies. Moreover, with abundant land, inten~ve systems can promote deforestation if they generate more rents than the nonintensive system [White et al., 2001; Arima et al., 2005]. Thus, in the following discussion, we do not consider intensification per se and address the case of technologically static systems for both ranching and soy production. In addition, we do not regard the replacement of pasture by say as a fonn of technological intensification, strictly speaking, and reserve the tenn from this point on to describe increasing efficiency ofland use for the production of specific crops, such as soy itself. Thus, soy fanning intensifies if newly implemented technologies yield higher, unit-area soy production. The second issue involves the identification of the underlying forces driving an advanced edge of agriculture or ranching into so-called "uncultivated wilderness," to use the original tenninology of von Thiinen [Walker, 1999]. In a single commodity world, as for example with simply ranching, such an exercise devolves to describing the circumstances that would increase rents for ranching outputs such as beef. If prices for meat rose, for example, areas would be brought into production that previously did not generate positive rents. A similar result is obtained with a reduction in transportation costs. In a two-commodity world, the picture grows more complicated. Assuming that, to begin with, soy is found "behind" the cattle frontier; several possibilities arise. The first is that market conditions change for only one ofthe products, soy or beef. If the price for beef rises but not for soy, then we are in the situation first described, with an advancing cattle frontier as has been historically observed over the past several decades. On the other hand, if the price for soy rises but not for beef, soy advances into areas that were previously pastured, until such time as available pastures are exhausted, and only forest land remains available. At that point, soy directly replaces forest. . The circumstances as described are unrealistic for the Amazonian case, as it appears that the market situation of the recent past has generally favored both soy and ranching [Brandao et al., 2005]; in addition, the benefits of infrastructure do not discriminate by crop, and all agricultural activities on a frontier receive a rent windfall with new investments. Thus, soy potentially affects Amazonian forest cover by two mechanisms. In the first instance, soy may be "pushing" the cattle frontier deeper into forest [Sawyer, 2008; Vera-Cruz et al., 2008]. This occurs if soy occupies productive pasture land, due to rising rents, and if rents continue to rise for cattle products. In the second instance, market conditions may be stronger for soy, in which case soy production "leapfrogs" into areas of primary forest, in advance of ranching.
Land use leapfrogging, driven by the same mechanism, is observed when sprawl swallows up agriculture on the urban periphery and converts natural areas into residential land use [Walker and Solecki, 2004].
WALKERET AL.
!ab!e 4. Deforested Area Converted to Mechanized Agriculture and Area Converted to Mechanized Agriculture From Forest Reported III Literature for Three Study Areas
,
~)1
"
Study Location
Time Period Analyzed
Deforested Area" (k:m2) (a)
Mato Grosso state
2001-2004
38,097
Santarem and Be1terra, Para municipios
1975-1986 1986-1997 1997-1999 1999-2004 2004-2005 1996-2001
7.2. The Greater Impact: Pasture or Soy?
As has been discussed, the expansion ofpasture has been a prime driver of deforestation in the Amazon basin. The question that now arises is, "What impact will the Amazon's new big crop, soy, have on the forest?" We speculate that soy has exerted both "pushing" and "leapfrogging" effects, alluded to in the conceptual discussion above, but that market conditions favoring soy are currently emerging. We base this on remote-sensing analysis of forest conversion, resolved into type of land use, and size of clearing. Data on the fate of deforested lands are not available for the entire Amazon basin, but studies have used remote sensing to distinguish among pasture, mechanized cropland, mainly soy, and other land uses in three areas, the state of Mato Grosso [Morton et ai., 2006, 2009a], Santarem and Belterra municipalities in the state of Para [Venturieri et ai., 2007], and Vilhena municipality, in the southeastern part of Rondonia [Brown et ai., 2005]. In this regard, Morton et al. [2006] report that 12% to 14% of deforested area in Mato Grosso was converted directly to cropland between 2001 and 2004 (Table 4). The percentage peaks at 23% in 2003 when both deforestation rates and the price of soybeans were relatively high, which is consistent with a Thunian "leapfrog" of soy into the primary forests of Mato Grosso. In addition, land deforested for cropland accounted for 28% to 33% of all lands converted to mechanized agriculture. This counters the claim that all cropland expansion occurs only on previously cleared lands, mainly pasture [Morton et al., 2009a, 2009b]. A similar pattern is observed in Vilhena in eastern Rondonia. Although the majority of cropland expansion takes place on previously cleared land, a substantial portion (22% from dense forest and 200/0 from less dense forest) directly consumed forested land between 1996 and 2001 [Brown et ai., 2005]. Taking a longer-tenn perspective, Venturieri et al. [2007] document the emergence of mechanized agriculture as a driver of deforestation since 1975 in the municipalities of Santarem and Belterra, Para. No mechanized soy production was present in the study area prior to 1999, but in the periods 1999-2004 and 2004-2005,8% and 2.7% of new croplands were created from forests, respectively. The study area also exhibits a smaller proportion of direct conversion to mechanized cropland than in Mato Grosso or Rondonia (8.2% and 10.7% for the two periods). In Para, denser forest may favor the use of already-cleared pasture by expanding croplands.
75
Vilhena, Rondonia municipio
. Deforested Area Conve11ed to Area Converted Mechanized to Mechanized Agriculture Agriculture b Area (krn2) (b) (c) 16,370
4670-5463
Deforested Area Converted to Mechanized Agriculture (% of total) (c)/(a)
Area Converted to Mechanized Agriculture From Forest (% of total) (c)/(b)
12.25-14.34
28.53-33.37
Source
Morton
et at. 821 739 419 527 140 not reported
0.0 0.0 0.0 544 560 70.36
0.00 0.00 0.00 44 15 15.71 (dense) 14.21 (less dense)
0.00 0.00 0.00 8.35 10.71 not reported
0.00 0.00 0.00 8.09 2.68 22 (dense) 20 (less dense)
[2006] Venturieri et ai. [2007]
Brown et ai. [2005]
:Includes parcels deforested for all use~ includ~ng mechanized agriculture, pasture, and not yet in production. Includes parcels converted to mechanIzed agnculture from all land covers including forests, pastures and successional vegetation.
Data on size of clearings associated with individual deforestation events also points to the growing importance of soy and mechanized agriculture, more generally, as a direct cause of forest loss in the Amazon basin. In particular, mechanized operations involving soy typically make large clearings quickly in order to hasten the start-up of production, given the presumed degree of capitalization and risks associated with soy fanning. By way of contrast, a relatively larger component of pasture creation is linked to small and medium producers, with low levels of technology, who proceed in a piecemeal fashion in fonning their pastures over longer periods of time [Walker, 2003].
Table 5 indicates that the vast majority of annual clearings were less than 100 ha in 2001-2005, which reflects the large populations of smallholders in the basin who mainly clear the forest to make Way for pastures; in which case, they represent the advanceq edge of the cattle frontier [see Walker, 2003; Walker et al.,:2008]. Medium (100-1000 ha) and large (> 1000 ha) clearings, dedicated to both pasture and soy, are relatively few in number, but they contribute disproportionately to area cleared. Thus, medium-sized clearings account for only 3% of all clearings, but 37% of the region's deforestation; large clearings, at less than 1%, account for 13% of forest cleared. These proportions vary widely by state, with
TableS.. Proportion of Number of Deforested Polygons and Deforested Area in Small «100 ha), Medium (100-1 OOOha) and Lar e g Categones (> 1000 ha) by State for 2001-2005 Number of Deforested Polygons in Size Category (%)
Deforested Area in Size Category (%)
State
Deforested Area (2001-2005)" (krn 2)
Small
Medium
Large
Small
Medium
Large
Acre Amapa Amazonas Maranh1io b Mato Grosso Para Rondonia Roraima Tocantins Total
3,344 111 5,120 4,642 44,959 33,840 16,427 1,312 985 110,743
0.99 0.99 0.99 0.97 0.92 0.96 0.97 0.99 0.98 0.94
0.01 0.01 0.01 0.03 0.08 0.03 0.03 0.01 0.02 0.03
<0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01
0.82 0.78 0.70 0.69 0.37 0.48 0.70 0.72 0.78 0.50
0.16 0.20 0.24 0.27 0.49 0.34 0.28 0.22 0.23 0.37
0.02 0.02 0.06 0.04 0.14 0.18 0.02 0.05 0.01 0.13
"Data from http://www.obtinpe.br/prodes. bThe 2001 data are excluded from analysis of deforestation polygons because they contain spurious large clearings.
76
WALKER ET AL.
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
the greatest proportions of total deforestation accounted for by medium and large clearings found in Mato Grosso and Para, both of which show a significant development of mechanize~ agriculture and soy. Examples from Mato Grosso, Para, and Rondonia illustrate varying deforestation dynamics related to capital-intensive clearings for mechanized cropland (assuming that many of the large clearings are accounted for by crops, especially soy) in 2001-2005 (Figure 5). In Mato Grosso, interannual variability in deforestation from 2001 to 2005 was driven mainly by variability in the deforested area from large clearings; the high deforestation years of 2003 and 2004 have the highest proportion of large clearings and lowest propor-
tion of small clearings. Evidently, mechanized agriculture, leading to large clearings, rather than smallholder deforestation, controls recent land cover dynamics in Mato Grosso. By way of contrast, in Para, the high deforestation year of 2004 has the highest proportion of deforestation area in small clearings. Deforestation in this state appears to be more tightly linked, in the aggregate, to smallholder production, with deforested land primarily devoted to pastures [Walker et al., 2000]. A similar pattern holds for Rondonia, with an even higher proportion of deforestation occurring in small clearings [Browder et al., 2008]. The spatially variable attribution of deforestation by size of holding has been observed before [Walker et al., 2000]. In temporal terms at
MATOGROSSO
I
Q)
N
o medium
14000 , - - - - - - - - - - - ,
'en .!:
'"~
0.8
.. ..
'" 5' 0.6
'tJ
..
Q)
1fj
..
..
~~
~rl 0.4
..
..
..
'5 c
'e
E
~
10000
* '"
'tJ
8000 6000
'~*
2000
~
0.2
0
N 12000
0
4000
0-
e
o
0
"-
2001
2002
2003
2004
2005
2001 2002 2003 2004 2005
PARA 14000 - r - - - - - - - - - - - - ,
Q)
.!:O!
~
'"
.!:
'~"
'"
'tJ
0.8 ~
* Q)
.Q Q)
:
l
'tJ
'5 c 0
0
0.6
.l'!
rl 0.4
..
..
.. ..
0)
12000
~ 10000 ~ 8000
~
6000
~
4000 2000
Cl
0
~
0.2
'e 0
0-
e
2001 2002 2003 2004 2005
0
"-
2001
2002
2003
2004
"'"
..
2005
RONDONIA
14000 - r - - - - - - - - - - - - ,
0.8
..
0.6 0.4
~ 12000 ";10000 ~ 8000 2 6000 ~ 4000
'*
0.2
o
0 2001
2002 2003 2004 2005
2000
0 2001
2002
2003
2004
2005
Figure 5. Dynamics of clearing sizes in three states in the Legal Amazon, 2001-2005.
basin-scale, however, the role of largeholders has probably grown [Alves, 2002] .. In general, the res.mrch suggests that pasture is the primary "proximate" caust{of forest clearance in the Amazon basin, and that most c~land expansion is occurring on previously deforested lands. Nevertheless, a substantial and growing proportion is occurring in forested areas via a "leapfrogging" of land use. This is especially so in Mato Grosso, where as much as 23% of new deforestation occurred in response to land clearance for mechanized agriculture, presumably soy. As soy continues its advance elsewhere in the Amazon basin, the percentage of direct conversion of forests can be expected to rise. That pasture remains the primary proximate cause of forest loss does not mean that soy is an insignificant factor in Amazonian deforestation. Even if soy were to show no direct conversion whatsoever, its demand for already-cleared land would be likely to "push" compensatory deforestation by the displacement of ranching. Given that market conditions for Amazonian cattle products have been mainly good, and are likely to remain so for some time, it is reasonable to assume that much of the productive pasture occupied by soy has been recovered elsewhere in the Amazon)asin, via deforestation. Thus, a sizeable fraction of the current deforestation rate may be accounted for by the Thunian "pushing" effect, as described [Sawyer, 2008]. This observation has implications for interpretation of the Associayao Brasileira das Industrias Oleos Vegetais soy moratorium on expansion in the Amazonian biome. Specifically, soy may not in-and-of itself directly precipitate deforestation, at the same time as it displaces pastures at the expense of forest.
77
necessary to estimate (1) the amount of growing national and global demand that will be met by Amazonian producers, and (2) the amount of land necessary to meet it. To this end, we develop projections for intensive and extensive regimes of both products. For soy, EMBRAPA [2002] indicates that 650,000 km2 of virgin cerrado, mostly in Amazonia, remains unexploited, capable of supporting mechanized commercial grain and oilseed production. If Brazilian soy production grows 23 million t by 2020 [Rodrigues, 2004], another 57,000 km2 ofland will be needed under an intensive production regime (4 t ha- I) and 76,666 km2 under an extensive one (3 t ha- I ). These lands will most likely come from Amazonia, given recent movements of the soy frontier [Mueller, 2003]. As for cattle, a demand projection to the year 2020 requires the addition of 393,000,000 animals to the global herd. If one tenth ofthese are added to Amazonian herds, the amount of new land required ranges from 196,500 to 393,500 km2 , depending on assumed stocking densities. Despite our documentation of soy's leapfrogging into primary forest, most soy expansion in the near term will probably occur on either (1) existing pastures, both productive and degraded, or on (2) the abundant cerrados still remaining [BrandQo et al., 2005; Mueller, 2003]. In contrast, most additions to the regional herd will range on newly deforested lands. Thus, in th~ absence of any "pushing" by soy onto productive cattle range, the deforestation increment could reach 393,000 km4, the area of the new cattle pastures, under low stocking d~nsity. With low-productivity soy farming and ranching, Thunian "pushing" could add 76,666 km2 of new deforestation: This transpires if soy expands only on preexisting pasture, which is then displaced in equal meas7.3. Projected Expansion o/Demand/or Agricultural ure at the cattle frontier. This scenario gives an upper bound Products and Land to the deforestation increment by 2020 of 469,666 km2 which, annualized over 15 Years (2005-2020), yields a high The preceding analysis has described both conceptually annual deforestation rate of ~3l,000 km2 . Such a number and empirically the current face ofAmazonian deforestation, is improbable and depends on strong assumptions regarding which presents itself mainly as a conversion of forest for productivity and the extent of pasture displacements by soy. pasture and soybean fields producing for expanding markets. Over time, soy production will intensify, and herd manageThus, loss of forest in the north region is best interpreted as ment will lead to higher densities. In addition, the Brazilian a response to domestic and global demands for the region's cerrado retains considerable scope for agricultural expanagricultural products, and future losses will similarly result sion, although forest conservation at the expense of cerrado from expansions of the same demands. In the individual sec- is ecologically costly given the biome's diversity and raptions on cattle and soy, we discussed the scope for such ex- idly dwindling expanse [Klink and Machado, 2005; Jepson, pansions, and they are considerable. It remains to translate 2005]. Despite these caveats, the calculation is nevertheless these consumer demand figures into estimates of demands suggestive that over the short- to midrun, strong pressures on for the prime factor of agricultural production, namely, land.. the Amazonian forest are likely to continue. At this point, we must obviously enter the realm of speculation, but we do so on the basis of informed judgment. 8. CONCLUSIONS The chapter has focused much attention on the demand situation for both Amazonia cattle products and soy. To deThe LBA program has sought to illuminate how Amazonia functions as a large-scale, ecological system, partly in termine possible impacts on the Amazonian landscape, it is
i 78
I
,
'i,1
!
l
I
~
i
"
l'
! ~. ~i
,, ;,.
f
!
,\
I
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
the interest of conserving its unique ecological resources for future generations. This chapter addresses the region's primary disturbance, the expansion of mechanized farming, mainly sol and of cattle ranching into the Amazon basin. To this end, it summarized the policy process that led to agricultural development of the region, outlined the current market situations facing regional producers, and described the dynamics of both sectors in spatial terms. We suggested at the outset that effective policies for conserving the Amazonian environment would require an understanding of the region's agriculture, and we have attempted to provide this understanding. Thus, we end the chapter with a brief discussion of the implications of our exposition of Amazonian agriculture, the face of the region's environmental change for four decades, for policy formulation. That said, Brazil, a sovereign nation, is certainly entitled to exploit the natural wealth within its borders to better the social welfare. The trick is to do so sustainably. In environmental terms, the inexorable growth of Amazonian pastures, pushed on by soy and possibly cane in the years to come [Sawyer, 2008], appears to have all the appearances of a manifest destiny [Walker et al., 2008]. The Brazilian government, together with the states of Amazonia Legal, has erected environmental safeguards with the National Systems of Conservation Units and its designation of vast stretch of forest as protected areas, off-limits to farmers and ranchers [Law 9.985, 2000; Decree 4340, 2002]. The Indigenous Reserves, also dedicated to forest conservation by the forestry code and the National Plan for Protected Areas [Decree 5758, 2006], add substantial area to the green wall offorest defense (http://www.funai.gov.br/). In fact, the SNUC, Indigenous Reserves, and state set-asides of conservation lands covers an estimated 2.3 million km 2 or 43% of Amazonia Legal. On paper, this would seem potentially adequate. Combined with the requirements for large forest reserves on private holdings, up to 80%, a large fraction of the Amazonian forest would be preserved if the boundaries ofthe green wall maintain their integrity. Approaches to achieving the laudable goal of the protected areas program cross a wide spectrum from regulatory interventions to market-based tools. As for state intervention, the reach of the law has been dramatically extended by satellite-based technologies that enable rapid assessment of protected area boundaries and forest reserves on private property. However, maintaining these boundaries is potentially costly and sometimes problematic. The degradation of protected areas and indigenous reserves by predatory loggers and ranchers seeking cheap pasture points to the difficulty of transforming paper parks into sustainable green space [Euler et al., 2008]. As for reserves on private property, a satellite can identify violations, but the imposition of fines on pos-
sibly thousands of properties over the continental reaches of the largest tropical forest in the world presents a daunting task. Nevertheless, measures focused on production chains and the transport of products from illegally deforested lands [i.e., Decreto Federal 6321, 2007], presumably monitored via certification, hold new promise for enforcement. Market-based tools have traditionally been promoted as alternatives to the efficiency shortcomings ofregulation, and the Amazonian case is no different. In this regard, the idea of a carbon market is an idea whose time may have come given Amazonia's massive carbon riches. Although the creation of such markets presents challenges, one alternative is the implementation of tradable development rights (TDRs) in the use of forest reserves on private property. TDRs do not necessarily conserve carbon in the aggregate, but they are capable of spatially directing agricultural expansion in a manner consistent with ecological zoning and the maintenance of protected areas, thereby conserving biodiversity [Chomitz, 2004]. Market interventions that reduce the demand for land outright include green loans, a legal innovation that enables banks to act as environmental stewards in providing credit to agricultural interests [Federal Decree 6321, 2007; Resolution 3545, 2008 of the Central Bank]. Green loans are potentially important in managing the expansion of soy farms, given their capital intensity. As has been suggested, soy's "pushing" of pasture potentially conceals its ultimate forest impact, and must be addressed by any policy intervention aimed specifically at soy. As for cattle, growing consumer demands for products with environmental bona fides may soon place price premiums on beef not originating from newly deforested lands. Government assistance in safeguarding the credibility of certification programs will be needed, in order to shape incentives in a manner consistent with conservation. To this point in the chapter, no mention has been made of the issue of land distribution, and social welfare more generally, and we have treated agriculture mostly as a social aggregate of farm and ranching enterprises. This conceals the tremendous variety in types of farms and ranches found in Amazonia, ranging from household operations on the very threshold of subsistence, to highly capitalized ventures covering tens of thousands of hectares.· Such diversity points to complex social processes, including a movement for land reform now impacting the region, with the creation of hundreds of assentamentos, or land reform settlements, and conflict that sometimes turns bloody [Simmons et al., 2007]. Long-run conservation of the Amazonian forest will ultimately depend on increases in agricultural productivity, an outcome requiring private investments by the men and women who have risked much to stake their claims on the
WALKER ET AL.
frontier. To make such ihvestments, these landowners will need to be secure in their property rights, in which case, social policy will al,,:)6 be necessary to guarantee the envir~nmental futur~o the region, policy that resolves the conflict over land t everyone's satisfaction, and establishes land-tenure sec rity at the same time that it improves the livelihoods of the rural poor. Environmental sustainability in Amazonia cannot be achieved without attention to social justice and the development of an agricultural economy that benefits all. Acknowledgments. The authors would like to acknowledge support from NASA project (NNG06GD96A) "Spatially Explicit Land Cover Econometrics and Integration with Climate Prediction: Scenarios of Future Landscapes and Land-Climate Interactions" and from NSF project (BCS-0620384) "Collaborative Research: Globalization, Deforestation, and the Livestock Sector in the Brazilian Amazon." The views expressed are those of the authors and do not necessarily reflect those of the supporting agencies. We would like to thank Marcellus Caldas, Eugenio Arima, Cynthia Simmons, and Peter Richards for providing comments on an early version of the manuscript. We would also like to thank Ritaumaria Pereira and Dante Vergara for research support and graphics production.
REFERENCES Adamoli, 1., 1. Macedo, L. M. Azervedo, and 1. Madeira Netto (1985), Caracterizacao da regiao dos cerrados, in Solos dos Cerrados: Tecnologias e Estrategias de Manejo, EMBRAPA, edited by W. J. Goedert, pp. 40-73, Brasilia, Brazil. AGRIANUAL (2000), Anuario da Agricultura Brasileira, FNP Consultoria and Comercio. Almeida, L. A., R. A. S. Kiihl, M. Miranda, and G. Campelo (1999), Melhoramento da soja para regi6es de baixas latitudes, in Recursos Geneticos e Melhoramento de Plantas para 0 Nordeste Brasileiro, Versao 1.0, edited by M. A. de Queiroz, C. O. Goedert, and S. R. R. Ramos, Embrapa Semi-Arido, PetrolinaPEfEmbrapa Recursos Geneticos e Biotecnologia, Brasilia-DF. (Available at http://www.cpatsa.embrapa.br) Alves, D (2002), Space-time dynamics of the deforestation in Brazilian Amazonia, Int. J. Remote Sens., 23(14), 2903-2908. Anualpec (2003), Anuario da Pecuaria Brasileira 2003, FNP Consultoria, Sao Paulo. Arima, E. Y., and C. Uhl (1997), Ranching in the Brazilian Amazon in a national context: Economics, policy, practice, Soc. Nat. Resour., 10,433-451. Arima,E. Y., P. Barreto, and M. Brito (2005), Pecuraria na Amazonia: Tendencias e implicac6es para a conservacao, Imazon, 75, Belem. Barros, G. S. D. C. (Coordenador) (2002), Economia da pecuaria de corte na regiao Norte do Brasil, Indicadores Pecuarios, Cepea, Piracicaba, SP. Boserup, E. (1965), The Conditions ofAgricultural Growth, Allen and Unwin, London.
79
Brandao, A., G. Rezende, and R. Marquest (2005), Crescimento Agricola no Brasil no Periodo 1999-2004: Explosao da Soja e da Pecuaria Bovina e seu Impacto sobre 0 Meio Ambiente, Working Paper 1103, IPEA, Rio de Janeiro. Browder, J. (1988), The social costs of rain forest destruction: A critique and economic analysis of the hamburger debate, Interciencia, 13, 115-120. Browder, J. 0., M. A. Pedlowski, R. Walker, R. H. Wynne, P. M. Summers, A. Abad, N. Becerra Cordoba, and J. Mil-Homens (2008), Patterns of development in the Brazilian Amazon: Deforestation, land-use, and socio-economic stratification of the rural work-force in Rondonia, 1992-2002, World Dev., 36, 1469-1492. Brown, 1. C., M. Koeppe, B. Coles, and K. Price (2005), Soybean production and conversion of tropical forest in the Brazilian Amazon in the case of Viihena, Rondonia, Ambio, 34(6), 462-469. Castro, A. M. G. De, S. M. V. Lima, A. Freitas Filho, H. R. De Souza, A. R. De Souza, C. N. De Castro (2001), Competitividade da Cadeia Produtiva da Soja na Amazonia Legal, Convenio SUDAMIFundacao do Desenvolvimento da UFPE. Cattaneo, A. (2001), Deforestation in the Brazilian Amazon: Comparing the impacts of macroeconomic shocks, land tenure, and technological change, Land Econ., 77(2),219-240. Cattaneo, A. (2005), Inter-regional Innovation in Brazilian agriculture and deforestation in the Amazon: Income and environment in the Balance, Environ. Dev. Econ., 10,485-511. Chomitz, K. M. (2004), Transferable development rights and forest protection: An exploratory analysis, Int. Reg. Sci. Rev., 27(3), 348-373. I CONAB (2003), Esta,tisticas Agricolas. (Available at www.conab. gov.br) I
EMBRAPA (2002), Tecnologias deProducao de Soja-Regiao Central do Brazil, 2003, 199 pp., SojafEmbrapa CerradosfEmbrapa Agropecuaria Oeste, ESALQ, Londrina, Brazil. EMBRAPA-SOJA (2002), Tecnologias de Produr;iio de Soja - Regiiio Central do Brazil, 2003, 199 pp., Embrapa SojafEmbrapa CerradosfEmbrapa Agropecuaria Oeste, ESALQ, Londrina. Euler, A., et al. (2008), 0 Fim da Floresta? A Devastacao das Unidades de Conservacao e Terra Indigenas no Estado de Rondonia, Grupo de Trabalho Amazonica, Porto Velho, Brazil. Falesi, I. (1976), Ecossistema da Pastagem Cultivada na Amazonia Brasileira, Empresa Brasileira de Pesquisa AgropecuariaiCentro de Pesquisa Agroflorestal do Tropico Umido, Belem, Brazil. Faminow, M. (1997), Spatial economics oflocal demand for cattle products in Amazon development, Agric. Ecosyst. Environ., 62, 1-11. Fearnside, P. (1980), The effects of cattle pasture on soil fertility in the Brazilian Amazon: Consequences for beef production sustainability, Trop. Ecol., 21, 125-137. Goulding, M., N. Smith, and D. J. Mahar (1995), Floods of Fortune: Ecology and Economy Along the Amazon, Columbia Univ. Press, New York. Hall, A. L. (1987), Agrarian crisis in Brazilian Amazonia: The Grande Carajas programme, J. Dev. Stud., 23(4), 522-552. Hall, A. L. (1989), Developing Amazonia: Deforestation and Social Conflict in Brazil's Carajas Programme, Manchester Univ. Press, Manchester, N. H.
80
WALKER ET AL.
EXPANSION OF INTENSIVE AGRICULTURE AND RANCHING
Hartwig, E. E., and R. A. S. Kiihl (1979), Identification and utilization of a delayed flowering character in soybean for short-day conditions, Field Crops Res., 2, 145-151. Hecht, S. B. 9985), Environment, development and politics: Capital accumulation and the livestock sector in Eastern Amazonia, World Dev., 13, 663-684 Hecht, S. B., and A. Cockburn (1989), The Fate of the Forest: Developers, Destroyers, and Defenders of the Amazon, Verso, London. Hecht, S. B., R. B Norgaard, and G. Possio. (1988), The economics of cattle ranching in East~rn Amazonia, Interciencia, 13, 233-240. Helfand, S. M., and G. Castro de Rezende (2001a), The impact of sector-specific and economy-wide policy reform in agriculture: The case of Brazil 1980-1998, Working Paper 01-34, Department of Economics, University of California, Riverside, California. Helfand, S. M., and G. Castro de Rezende (2001b), Brazilian agriculture in the 1990s: Impact of the policy reforms, Working Paper 785, IPEA, Rio de Janeiro, Brazil. IBGE (2005), Municipal Agricultural Production (PAM), Instituto Brasileiro de Geografia eEstatistica. (Available at http://www.sidra. ibge.gov.br/bda/acervo/acerv02.asp?e=v&p=PA&z=t&o=10) IPAM (n.d.), Desmatamento na Amazonia: Medidas e efeitos do Decreto Federal 6.321/07, IPAM, Belem, Brazil. Jepson, W. (2005), A disappearing biome? Reconsidering landcover change in the Brazilian savanna, Geogr. J., 171, 99-111. Jordan, C. F. (1982), The nutrient balance of an Amazonian rain forest, Ecology, 61,14-18. Kaimowitz, D., B. Mertens, S. Wunder, and P. Balanza (2004), Hamburger Connections Fuels Amazon Destruction: Cattle Ranching and Destruction in Brazil's Amazon, Research report, CIFOR: Center for International Forestry Research, Jakarta, Indonesia. Kiihl, R. A. S., L. A. Almeida, and A. Dall'agnol (1985), Strategies for cultivar development in the tropics, in World Soybean Research Conference III Proceedings, pp. 301-304, Westview, Boulder, Colo. Klink, C., and R. B. Machado (2005), Conservation ofthe Brazilian cerrado, Conserv. BioI., 19(3), 707-713. Liverman, D. M., and S. Vila (2006), Neoliberalism and the environment in Latin America, Annu. Rev. Environ. Resour., 31, 327-363. Mahar, D. J. (1979), Frontier Development Policy in Brazil: A Study ofAmazonia, Praeger, New York. Mahar, D. J. (1989), Government policies and deforestation in the Brazilian Amazon, in Environmental Management and Economic Development, edited by G. Schramm and J. J. Warford, pp. 87-116, Johns Hopkins Univ. Press, Baltimore, Md. Margulis, S. (2004), Causes ofDeforestation ofthe Brazilian Amazon, World Bank Working Paper, no. 22, Washington, D. C. McGrath, D., and M. D. C. Vera-Diaz (2006), Soja na Amazonia: Impactos ambientais e estrategias de mitiga9ao, Cienc. Ambiente, 32,151-165. Mollo, M. L. R., and A. Saad Filho (2004), The neoliberal decade: Reviewing the Brazilian economic transition, in Marx Inter-
national IV, Paris. (Available at http://netx.u-paris10.fr/actuelmarx/indexa.htm) Morton, D., R. DeFries, Y. Shimabukuro, L. Anderson, E. Arai, F. Espirito-Santo, R. Freitas, and J. Morisette (2006), Cropland exc pansion changes deforestation dynamics in the southern Brazilian Amazon, Froc. Natl. Acad. Sci. US. A., 103(39),14,637-14,641. Morton, D., R. DeFries, and Y. Shimabukuro (2009a), Cropland expansion in cerrado and transition forest ecosystems: Quantifying habitat loss from satellite-based vegetation phenology, in Advances in Applied Biodiversity Science, Cerrado Land Use and Conservation, edited by C. Klink, R. Cavalcanti, and R. DeFries, Conserv. IntI., Washington, D. C., in press. Morton, D., Y. Shimabukuro, B. Rudorff, A. Lima, R. M. Freitas, and R. DeFries (2009b), Conservation challenge at the agricultural frontier: Deforestation, fire, and land use dynamics in Mato Grosso, Agua Ambiente, in press. Mueller, C. C. (2003), Expansion and modernization ofagriculture in the Cerrado: The case of soybeans in Brazil's Center- West, Working Paper 306, Department of Economics, Universidade de Brasilia, Brasilia, Brazil. Nehmi Filho, V. A. N., and J. Pusch (2003), Avancos dos graos nas terras de pastagens do Centro-Oeste, Annualpec, FNP Consultoria, 333-336. Owen, W. (1987), Transportation and World Development, Johns Hopkins Univ. Press, Baltimore, Md. Pacheco, P. (2005), Populist and capitalist frontiers in the Amazon: Diverging dynamic of agrarian and land-use change, Ph.D. dissertation, Clark University, Worchester, MA. Pfaff, A. S. P. (1999), What drives deforestation in the Brazilian Amazon?, J. Environ. Econ. Manage., 37, 26-43. Piketty, M. G., J. Bastos da Veiga, J. F. Tourrand, A. M. N. Alves, R. Poccard-Chapuis, and M. Thales (2005), Determinantes da expansao da pecuaria na Amazonia Oriental: Consequencias para as politicas publicas, Cadernos de Ciencia e Tecnologia, Brasilia, 22(1), 221-234. Poccard-Chapuis, R., M. Thales, A. Venturieri, M. G. Piketty, B. Mertens, J. Bastos da Veiga, and J. F. Tourrand (2005), A Cadeia Produtiva da Carne: Uma Ferramenta para Monitorar as Dinamicas nas Frentes Pioneiras na Amazonia Brasileira, Cadernos de Ciencia e Tecnologia, Brasilia, 22(1), 125-138. RCW (2004), Mundo: Balam;o de Oferta e Demanda da Soja. (Available at http://www.rcwconsultores.com.br/radarsoja/) Rodrigues, R. L. V. (2004), Analise dos fatores determinantes do desflorestamento na Amazonia Legal, tese de Doutorado apresentada a coordena9ao dos Programas de P6s-Gradua9ao de Engenharia da Universidade Federal Do Rio De Janeiro. Rohter, L. (2003), Mad cow disease in the United States: Exports; Brazil and Argentina expect rising beef sales, The New York Times, December 27, 2003. (Available at http://query.nytimes. comlgstifullpage.html?res=9FOCE5DAI43EF934AI5751C1A9 659C8B63&sec=health&spon=&pagewanted=print) Santana, A. C., et al. (1997), Reestruturar;iio Produtiva e Desenvolvimento na Amazonia: Condicionantes e Perspectivas, BASA, Belem, Brazil. Santos, R. (1980), Historia Economica da Amazonia (1800-1920), T.A. Queiroz, Sao Paulo, Brazil.
Sawyer, D. (2008), Climate I change, biofuels and eco-social impacts in the Brazilian ~azon and Cerrado, Philos. Trans. R. Soc. Ser. B,363, 1747r 1752. Simmons, C. S. (2002),.Development spaces: The local articulation of conflicting development, Amerindian rights, and environmental policy in ea~tefu Amazonia, Prof Geogr., 54, 241-258. Simmons, C. S., R. Walker, E. Arima, S. Aldrich, and M. Caldas (2007), Amazon land wars in the south of Para, Ann. Assoc. Am. Geogr., 97(3), 567-592. Sinclair, T. R., N. Neumaier, J. R. B. Farias, and A. L. Nepomuceno (2005), Comparison of vegetative development in soybean cultivars for low-latitude environments, Field Crops Res., 92(1), 53-59. Sioli, H. (1973), Recent human activities in the Brazilian Amazon region and their ecological effects, in Tropical Forest Ecosystems in Africa and South America: A Comparative Review, edited by B. J. Meggers, E. S. Ayensu, and W. D. Duckworth, pp. 321-324, Smithsonian Institution Press, Washington, D. C. Souza, P. I. De M., C. T. Moreira, and C. R. Spehar (2000a), BRS Milena: A mid-cycle high productivity soybean cultivar for the Brazilian Savannah, Pesq. Agropec. Bras., 35(8), 1695-1699, ISSN 0100-204X. Souza, P. I. De M., et al. (2000b), BRS Celeste: New soybean cultivar to the grain production system of the Brazilian Savannah, Pesq. Agropec. Bras., 35(2), 467--470, ISSN 0100-204X. Topall, O. (1991), Sisterma de cria9ao de bovinos nos lotes da coloniza9ao oficial da Transamazonica, regiao de Maraba, in Agriculture Paysannes et Developpement: Carai'be-Amerique Tropicale, pp. 203-228. United Nations Population Division (2004), World Population Prospects: The 2002 Revision Population Database. (Available at http://esa.un.org/unpp/p2kOdata.asp) U.S. Department of Agriculture (1997), Agricultural Statistics, U.S. Gov. Print. Off., Washington, D. C. USDA-FAS (2004), Soybean: World Supply and Distribution, Foreign Agricultural Service, Official, USDA. (Available at http:// www.fas.usda.gov/psdlcomplete tables) Va1derde, 0., and C. V. Dias (1967), A Rodovia Belem-Brasilia, IGBE, Rio de Janeiro, Brazil. Vance, J. E. (1986), Capturing the Horizon: The Historical Geography ofTransportation, Harper and Row, New York. Venturieri, A., A. Coelho, M. C. Thales, and M. D. R. Bacelar (2007), Analise da expansao da agricultura de graos na regiao de Santarem e Belterra, Oeste do estado do Para, in Simposio Brasileiro de Sensoriomento Remoteo, vol. 13, 21-26 Abril 2007, Florian6polis, pp. 7003-7010, Instituto Nacional de Pesquisas Espaciais (INPE), Sao Jose dos Campos, Sao Paulo, Brazil. Vera-Diaz, M. D. C., R. Kaufmann, D. Nepstad, and P. Schlesinger (2008), An interdisciplinary model of soybean yield in the Amazon basin: The climatic, edaphic, and economic determinants, Ecol. Econ., 65(2), 420--431.
81
Walker, R. T. (1999), The structure of uncultivated wilderness: Land use beyond the extensive margin, J. Reg. Sci., 39(2), 387410. Walker, R. T. (2003), Mapping process to pattern in the landscape change of the Amazonian frontier, Ann. Assoc. Am. Geogr., 93(2), 376-398. Walker, R. T. (2004), Theorizing land cover and land use change: The case of tropical deforestation, Int. Reg. Sci. Rev., 27, 247270. Walker, R., and E. Reis (2007), A basin-scale econometric model for projecting future Amazonian landscaped, final report, NASA, Greenbelt, Md. Walker, R. T., and B. Solecki (2004), Theorizing land cover and land use change: The case of the Florida everglades and its degradation, Ann. Assoc. Am. Geogr., 94(2), 311-328. Wallcer, R., E. Moran, and L. Anselin (2000), Deforestation and cattle ranching in the Brazilian Amazon: External capital and household process, World Dev., 8, 683-699. Walker, R., J. Browder, E. Arima, C. Simmons, R. Pereira, M. Caldas, R. Shirota, and S. Zen (2008), Ranching and the new global range: Amazonia in the 21st century, Geoforum, doi:lO.1016/ j.geoforum.2008.10.009, in press. Weinstein, B. (1983), The Amazon Rubber Boom: 1850-1920, Stanford Univ. Press, Palo Alto, Calif. White, D., F. Holmann, S. Fujisaka, K. Reategui, and C. Lascano (2001), Will intensifying pasture management in Latin America protect forests?-Or is it the other way round?, in Agricultural Technologies and TJ.opical Deforestation, edited by A. Angelsen and D. Kaimowitz, ,CABI, Oxford, U. K. Wiebelt, M. (1995), Stopping deforestation in the Amazon: Tradeoff between ecological and economic targets, Weltwirtschaflliches Archiv., 131(3), 542-562. Woodward, H. D. (1988), The Amazon alliance and the PMACI: Possibilities for Indian participation in Brazilian planned development, Lat. Am. Anthropol. Rev., 1(1), 24-25. Zen, S. (2002), Diversifica9ao como Forma de Gerenciamento de Risco na Agricultura, Tese de Doutorado, Escola Superior de Agricultura Luiz de Queiroz; ESALQ-Universidade de Sao Paulo---USP, Piracicaba, Brazil.
R. DeFries, Department ofEcology, Evolution, and Environmental Biology, Columbia University, New York, NY 10027, USA. M. del C. Vera-Diaz, Instituto de Pesquisa Ambiental da Amazonia, Avenida Nazare 669, Belem, CEP 66035-17 PA, Brazil. Y. Shimabukuro, Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos, SP CEP 12227-010, Brazil. A. Venturieri, CPATU-Embrapa, Tv. Eneas Pinheiro, sin, Belem, CEP 66095-100 PA, Brazil. R. Walker, Department of Geography, Michigan State University, 234 Geography Building, East Lansing, MI 48823, USA. ([email protected])
t··
Scenarios of Future Amazonian Landscapes: Econometric and Dynamic Simulation Models Stephen Perz, l Joseph P. Messina,2 Eustaquio Reis,3 Robert Walker,2 and Stephen J. Walsh4 This chapter addresses two broad classes of models frequently used in the land use/land cover change (LULCC) literature, namely, econometric and dynamic simulation approaches. We discuss both in light of analyses of LULCC in the Amazon, highlighting contributions of the Large-Scale Biosphere-Atmosphere Experiment in the Amazon program. We first discuss LULCC scenarios, a key approach to evaluating future LULCC in the presence of uncertainty that requires input from models. The bulk of the chapter then pursues a description of the basic elements ofeconometric and simulation models for LULCC scenario development, where we highlight the strengths and weaknesses of each modeling approach. We conclude by returning to issues involved in the process of scenario development, highlighting opportunities for engaging stakeholders with models for the sake of improving LULCC outcomes.
1. INTRODUCTION
,
li ri
i
are important in the context of the Large-Scale BiosphereAtmosphere (LBA) Experiment in Amazonia program and other Amazonian research as key human dimensions of climate change [see also Alves etal., this volume; Walker et al., this volume; Nobre et al., this volume]. Further, models and simulations both support construction of scenarios of future change in landscapes of Amazonia, as well as other regions important to global climate. There are numerous reviews of models and simulations of LULCC, many of them recent [e.g., Brown et al., 2004a; Evans et al., 2005; Verberg et al., 2006], and our goal here is not to provide another broad overview. Instead, this chapter seeks to emphasize modeling and simulation efforts of LULCC in the context of the LBA program. Similarly, there are many substantive reviews of the numerous causes of deforestation and other types ofland cover change [e.g., Geist and Lambin, 2002; Gutman et al., 2004; Moran and Ostrom, 2005; Lambin and Geist, 2006], so we focus on the factors established as important in Amazonia by scientists in LBA and other research initiatives focused on this region [see Alves et al., this volume; Asner et al., this volume; Walker et al., this volume; Pfaffet al., this volume; Brondizio et al., this volume].
Models are representations of how processes operate to generate observable patterns; simulations amount to models that examine changes over time given specific assumptions. Models and simulations are thus key analytical tools for the study of land use and land cover change (LULCC), which
f.'\ .•. i.'
'f i
L l' .
'
I ·1
'I
lDepartment of Sociology and Criminology and Law, University of Florida, Gainesville, Florida, USA. 2Department of Geography, Michigan State University, East Lansing, Michigan, USA. 3Instituto de Pesquisa Economica Aplicada, Rio de Janeiro, Brazil. 4Department of Geography, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
:··i.~ ~
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.102912008GM000736 83
84
age desirables and discourage undesirables. Scenario development has consequently been employed in many contexts by diverse social actors, including governments evaluating future revenue and spending, corporations trying to anticipate price fluctuations, or the Intergovernmental Panel on Climate Change in its effort to evaluate future climate change. Scenarios can thus support planning procedures that often involve quantitative models and simulations, including for the case ofLULCC [Alcamo et al., 2006]. Initially, scenarios developed as narratives are generally qualitative; the challenge then is to render them in quantitative terms with data input into empirical models and/or parameters as inputs for simulations. While some LULCC scenarios are aspatial and use accounting techniques [e.g., Evans et al., 2001], most LULCC scenarios are spatially explicit, which allows for mapping and thus visualization of the spatial distribution of LULC and their modeled changes over time [Alcamo et al., 2006]. This is especially the case for regional and local scenarios, which generally have higher spatial resolutions than global models. A key issue in quantitative scenario development concerns the temporal scale of the scenarios; longer-term scenarios must be considered more 2. SCENARIOS OF FUTURE speculative than probable due to the increasing uncertainties AMAZONIAN LANDSCAPES of envisioning change over longer time periods. Feedbacks Scenarios amount to stories about change that make spe- from initial changes, surprises ("shocks"), and other sources cific, explicit assumptions about the underlying forces that of nonlinear changes make long-term scenarios harder to arinfluence the outcome of interest [Peterson et al., 2003; ticulate plausibly and in detail. An additional issue concerns Baker et al., 2004; Soares-Filho et al., 2004; Alcamo et al., the spatial scale of the driving forces and processes underly2006]. The immediate objective of scenario development ing LULCC scenarios. A general contention in the LULCC is to envision multiple plausible paths of future change, community is that proximate causes such as land use decigiven uncertainty about the future. This envisioning is often sions by specific social actors are the primary determinants done via collaboration among scientists, policymakers, and of localized LULCC, whereas distant causes such as ecostakeholders knowledgeable about the region and issues of nomic growth and integration influence large-scale LULCC concern. Key to development of scenarios is to articulate a [Wood, 2002; Lambin and Geist, 2006]. This is likely to afhandful of narratives, usually three to five, which are evalu- fect selection of the key drivers of LULCC, depending on ated to be not just possible, but credibly probable, by knowl- the spatial resolution and extent of LULCC scenarios. Scenario development is already well established for the edgeable parties. Envisioning multiple sc~narios, in tum, affords an appraisal of which are more likely to occur and case of future Amazonian landscapes. Laurance et al. [2001] provide an example of alternative scenarios of future land which are preferable in a normative sense. The logic of creating multiple scenarios provides a means cover in 'Amazonia, based on Brazilian plans for new infraof addressing uncertainty. Rather than seek accurate results structure. They evaluated past impacts of infrastructure to by obtaining a single point estimate with error bars, develop- inform assumptions supporting simulations offuture impacts ment of multiple scenarios allows consideration of several of new infrastructure. With spatial data for road corridors possible outcomes and a range of contrasting possibilities. and road surface types, along with information about variScenario development is based on the recognition that in- ous resource uses in Amazonia, and assumptions about the complete information about the present, the future, and al- distance and degree to which those uses would affect forests ternatives in assumptions about process drivers influence along roads, they projected future forest cover and degree of degradation for Brazilian Amazonia. Based on simulations model and scenario design and implementation. With scenarios in hand, we can identify drivers of desir- run to 2020, they presented "optimistic" and "nonoptimisable and undesirable change outcomes, which can, in tum, tic" scenarios and showed that the former yielded less land support formulation of policies and social action to encour- area with cleared or degraded forest than the latter.
We center our attention on relevant methodological as well as substantive issues, since both are necessary to understand the strengths and limitations of various approaches to modeling for slttnario development. More specifically, we will emphasize econometric (statistical) models and dynamic simulation models, with examples from the Brazilian and Ecuadorian Amazonia. Our discussion will feature issues of working with spatial data and the problem of scale as it manifests itself in modeling decisions conceming the role of human agency in its biophysical and socioeconomic contexts. We begin with a discussion of scenarios of future landscapes, emphasizing the logic of creating alternative scenarios and the role of modeling in scenario development. We then pursue discussions of econometric modeling and dynamic simulation modeling, noting their strengths and limitations, as well as their roles in constructing scenarios of future Amazonian landscapes. The chapter concludes with observations regarding the challenges and opportunities for improving scenario development via improvements in LULCC modeling.
,,
!
.• ·.• 1
I
[
I
,~
I
PERZ ET AL.
SCENARIOS OF FUTURE AMAZONIAN LANDSCAPES
Another prominent example comes from Soares-Filho et al. [2002, 2004, 2006],_ who have developed dynamic simulation models to supp¢t scenario development for road corridors within Brazi~n Amazonia. Soares-Filho et al. [2002, 2004] began wi~takeholder interviews and participatory workshops in towns along the BR-163 highway corridor that runs north-south through central Brazilian Amazonia. These activities allowed development of initial scenarios about future landscape change along the corridor, in planning for paving by the government. Qualitative scenarios, in tum, led to identification of decision rules about land use, based on land quality, market prices, law enforcement, and other considerations. They articulated two main scenarios, "business-as-usual" (BAU) that followed historical patterns of economic expansion, and "governance" that assumed compliance with land use laws and respect for boundaries and use restrictions in parks and other protected areas. Their simulation model includes a set of assumptions for parameters that define a given scenario, which results in transition probabilities from one land cover to another. Those probabilities are then applied to a cellular automata model that calculates changes in land cover for spatially explicit cells along the BR-163 road corridor in I-year time steps from 2001 to 2030. Soares-Filho et al. [2004] show substantial differences in land cover change over time between the BAU and governance scenarios. Soares-Filho et al. [2006] have broadened their application of this modeling approach to Brazilian Amazonia, as a whole, and extended their scenarios to 2050. These efforts have influenced other work on modeling to produce scenarios of future Amazonian landscapes in the context of the LBA program. Such efforts have incorporated somewhat different modeling strategies with additional assumptions and data. The next sections of this chapter will focus on modeling, especially econometric models and dynamic simulation models in LBA projects that underlie development of other quantitative scenario models of future Amazonian landscapes.
85
relationships among LULCC drivers and outcomes. Alternatively, abstract models, such as dynamic simulations, often rely more heavily on assumptions in the design of the models themselves in order to observe complexity in dynamics ofLULCC under controlled conditions. Dynamic simulation models need not be accurate to be useful; simulations can deviate from empirical data and reveal important properties about how processes in complex systems operate [Evans et al., 2005]. That said, dynamic simulations can, of course, also incorporate empirical data and findings from empirical models, an issue to which we will return. For these and other types of models, there are unavoidable trade-offs as regards their complexity [Evans et al., 2005]. A key trade-off is the complexity of the data and/or assumptions in a model. It is by now clear that LULCC involves numerous driving forces and processes. There is great cache in inclusion of more variables in empirical models, though that depends on data availability, and more complicated econometric models can encounter difficulties with empirical data, such as multicolinearity and endogeneity. Dynamic simulations also face the trade-off in terms of complexity, for building in more model components allows one to capture more nuances in the processes involved, but can also make it harder to interpret the dynamics at play. In the end, LULCC models often exhibit greater detail in those aspects best understood by the researchers fuvolved, whereas the components least understood are nece~sarily those most simplified. For both statisti¢al and dynamic simulation modeling, making clear the a,ssumptions of the decisions underlying the models is cruoial. Clear assumptions facilitate model evaluation via calibration and validation, such as through diagnostic tests of econometric models and sensitivity analysis of dynamic simulation models. Whereas calibration refers to the fit of a model to the data it was intended to model, validation refers to the fit of a model to "independent" data, those beyond data originally used to design the model. There is great emphasis on calibration in econometric models, which are often specified using a particular data set and study case, and evaluated with a battery of diagnostic tests that quantify 3. SELECTED ISSUES IN LULCC MODELS the degree to which assumptions of the model are violated by the data. Conversely, there is considerable emphasis in There are many types of models that researchers have em- dynamic simulation modeling on validation due to the freployed in the study ofLULCC [Lambin, 1994], and they vary quently deductive nature of simulation model design, which in many different respects [Verberg et al., 2006]. A funda- can then be applied to various study cases (or data obtained mental distinction is between empirically based and abstract later) to see how well the simulations match the empirics. Of models. With an empirical foundation, LULCC modeling course, ideally, there should be calibration and validation of amounts to estimating parameters for drivers ofLULCC by . all models [Verberg et al., 2006]; moreover, in dynamic simfitting data to the assumptions of a model, as in econometric ulation models, some analysts have argued that there should models. There, great emphasis is placed on specifying as- be validation not only of outcomes but also of the processes sumptions that are appropriate for the data one intends to fit, thought to generate those outcomes [Brown et al., 2004a; resulting in accurate estimates of parameters that quantify Evans et al., 2005].
t; it 86 SCENARIOS OF FUTURE AMAZONIAN LANDSCAPES
.... .,
Beyond the distinction between empirical and abstract models, a key contrast in LULCC models concerns whether they are spatially explicit or aspatial. Given that LULCC is an eminlllltly spatial phenomenon, most LULCC models are spatially explicit, which raises issues specific to the handling of spatial data and modeling of spatial processes. The spatial resolution of models can influence the results, as more refined spatial resolution can result in weaker statistical models (per the modifiable areal unit problem) and require more processing time. There are also issues of spatial mismatch. In empirical models, socioeconomic and biophysical data are often only available for different units of observation, which makes data integration awkward. Further, in any model, when data are available for the same units, or if we simulate processes for a single set of units (such as pixels or cells), they may not reflect processes operating on that spatial scale. Spatial mismatches in data and processes require attention in models, so that they are able to capture spatial processes adequately. 4. ECONOMETRIC MODELS OF LULCC
4.1. Econometric Models in Theory
, i
i· I
~;
,! 'I
. .'
'
1
I
.: .•••.. •. !.
[
,~
".
II
An important class of models used in assessing LULC, and in predicting future LULC scenarios, belongs to those referred to as econometric models [Kaimowitz and Angelson, 1998; Barbier, 2001]. These are based on economic theory, and on the empirical estimation of relationships derived from theory, using statistical methods such as regression. Relevant economic theories informing application to LULC (1) focus on how the factors land, labor, and capital are organized into production functions, in order to realize produced outputs for consumers, and (2) represent the manner in which economies evolve through time, what is referred to as the dynamics of the economic system [Walker, 2004]. The production function is central to any effort to econometrically model LULC, given that it represents the means by which an economic agent links his or her objectives, such as maximizing profits, to the use of land. For example, if an assessment of a forested parcel of land indicates that farming would be profitable there, then the factors of production, including land, combine under the farmer's guidance to produce the goods of interest, causing forest to be cleared in the process. In this rendering, the production function is the technical matrix linking inputs to outputs for the activity in question, namely, farming [see Asner et al., this volume; Walker et al., this volume; Brondizio et al., this volume]. The production function describes only technological relationships, and in order to transform it into an operational concept for modeling, it is necessary to introduce likely hu-
PERZ ET AL. man behaviors that reflect hypotheses about the way people will act under certain conditions and about the institutional structure of the economy. Although the real world shows tremendous variation in this regard, in published research on deforestation in Amazonia where a theoretical model is explicitly stated, the assumptions typically made are that individuals maximize profits, and they do so in competitive markets for both inputs and outputs [e.g., Walker, 2004]. These hypotheses enable the derivation of demand curves for all factors of production, assuming the producer faces perfectly elastic markets or fixed prices. In the analysis of deforestation, the "derived" demand for land is of fundamental importance, and what is sought analytically is its extent and location in terms of resource quality characteristics, as well as important features of the economic environment such as transportation costs, affecting the farmgate prices for both inputs and outputs. If possible, it is informative to also distinguish differences of land use, as between land dedicated to crops or to pasture, or to fallow, although the prime environmental impact, namely, loss offorest, is the same independently of how land is used after forest is cleared. To ignore the production function and its ancillary concept, the derived demand for land, poses risks for the analyst, given the high likelihood of estimating an incorrect model. This is to say, the derived demand curve, a mathematical function, provides a list of variables that one must consider in using statistics to fit econometric models. Other potential issues arise in fitting any such model beyond specifying the right list of variables. Geographic contexts can change in subtle ways, introducing bias into results, as can the heterogeneity of agents active in the landscape. For example, the profit maximizing impulses of a small-scale shifting cultivator are likely to be different than a well-capitalized rancher. Finally, pitfalls await careless application of any econometric model that does not pay attention to underlying relationships between variables. To this point, the discussion has been of a static nature, in which LULC is a demand for land derived from production needs arising from a production function, and the economic behavior of the agents involved, in this case the Amazonian farmer or rancher. Such a simplistic statement conforms to an the equation Y=f(X),
where Y is the "dependent variable," in this case a LULC measure such as deforestation, and X is a vector of "independent variables" affecting the derived demand for land, such as resource quality characteristics and distance measures affecting prices for inputs and outputs, the so-called market or von Thiinen factors. More often than not, the f(X)
87
function is assumed to be linear, although functional trans- where Xl is a vector containing independent variables that formations of the independent variables allow for flexibility . change over time, and X is a vector of those that do not; the in representing the a~al relationships between the depend- b values in this case are coefficient vectors whose elements ent and independen)'variables. correspond to specific variables. The distingui~g characteristic of econometric models Econometric work on deforestation in Amazonia has a compared to other approaches to land cover representation long history, stemming from the seminal model of Reis and is that functionf(X) is based on statistical theory, namely, colleagues [Reis and Guzman, 1994; Anderson and Reis, estimation theory and is derived from actual observations on 1997], who first implemented the derived demand concept the empirical values of Yand X; moreover, the mathemati- in the present context. Early studies tended to be conducted cal relationship so determined possesses certain optimality at relatively aggregate scales (e.g., counties), and they were properties. For example, if the function is linear in X, the tasked with gaining insight into deforestation drivers rather mathematical relationships devolves to than in predicting possible future landscapes [Pfaff, 1999]. With the advent of GIScience technologies and remote sensing, a considerable amount of disaggregation has occurred in the statistical work, and equations are now estimated using for k independent variables, where X = (Xl, X 2,.. .xk). In fact, observations at the census tract and even the pixel scale [e.g., there exist an infinite number of possible values for the b in Chomitz and Gray, 1996; Mertens et al., 2000; Cropper et this equation, and the challenge is to make an informed deci- al., 2001; Nelson et ai., 2001; Moore et al., 2007], includsion in this regard. The econometric approach finds a unique ing for Amazonia [e.g., Chomitz and Thomas, 2003; Pfaffet set of values linking empirical observation on Y and the X al., 2007]. With so much concern about loss of Amazonian values, a set that reduces the squared error between the ob- forest, researchers have turned their attention to forecasting, served values of Y, and its predicted value, determined by the using econometric models such as those described here as sum of the products on the right-hand side of the equation. tools for prediction [Anderson et al., 2002]. As suggested, the model statement as presented is inherently static. This is a notable abstraction, given that economic 4.2. Econometric Mpdels in Practice: systems are dynamic, and deforestation is a dynamic change Deforestation in Amazonia in land cover. Thus, econometric predictions of LULC of necessity take into account the dynamics of the situation, An example of ad econometric model developed under the which imposes substantial data requirements on the analysis. auspices of the LBA program is described by Walker and Of obvious note, the dependent variable must be based on at Reis [2007] and Pfaff et al. [2007]. Results as well as data least two different time periods showing extent of forest for inputs from that work are presented here in order to provide observational units, be they states, counties, or highly dis- a recent example of econometric forecasting relevant to the aggregate pixels. The same point holds for the independent Amazonian case. variables, such as population growth, and the extension of The intent of the model was to project future Amazonian key elements of cost-reducing infrastructure such as roads landscapes out to 2020, given alternative scenarios deter[see Pfaff et al., this volume]. Of course, certain so-called mined by infrastructure investments, environmental gov"fixed" effects remain constant through time, such as soil ernance regimes, and population growth. The econometric quality, topographic relief, and aspects of the microclimate, model was first estimated, at census tract level, using deneglecting, obviously, global warming or climate-vegetation forestation data for periods 1976-1987 (Antropismo map, feedbacks (see chapters in section 2). Thus, the Yvariable Diagn6stico Ambiental (Instituto Brasileiro de Geografia e representing deforestation in the equation above is actually Estatistica (IBGE», 1:2,500,000 scale), 1986-1992 (TRFICthe outcome of measurement, or MSU Land Cover data, pixel size reduced to 200 m), and 1992-2000 (Tropical Rain Forest Information CenterMichigan State University (TRFIC-MSU) and Projeto de Monitoramento da Floresta Amazonica Brasileira por Satelwhere deforestation up to time t is taken to be the amount ite of INPE (PRODES-INPE) (2000) land cover digital occurring over the interval, t - 1. to t, with the extent of forest . maps, pixel size reduced to 200 m). The road data, which cover, F, measured at the beginning and end ofthe period. A provided an important exogenous variable, covered the pemodel with dynamics may be stated as riods 1968-1975, 1975-1983, and 1987-1993. These data were obtained by using Departamento Nacional de EstraYt = bo + blXI + bX, das de Rodagem paper maps (Mapa Rodoviario, Estados da
88
SCENARIOS OF FUTURE AMAZONIAN LANDSCAPES
Amazonia Legal) in combination with a digital map from IBGE (2004: Mapas Interativos de Transportes). The projections then considered scenarios defined on three factors, including trends in population growth, the nature of anticipated infrastructure investments, and efforts at governance, reflected by intensity of control over forest conservation in protected areas and on private holdings. Only the non-cerrado parts of the Amazon basin were considered. The demographic projections considered two possible futures, one with current trerlds in fertility, mortality, and migration projected out into the future (2020), and the other with increased out-migration (which implies reduced population growth). The demographic analyses accessed microdata files of Brazil's 2000 demographic census and used age structures for males and females, along with age-specific fertility rates (based on children ever born to women of reproductive ages), child survival figures (to estimate life ex-
PERZ ET AL.
pectancies and mOliality curves based on model life tables) and in-migrants and out-migrants during 1995-2000 (yielding net migration, which we annualized for purposes of projections beyond 2000). Two scenarios were considered for infrastructure investment, as well. Here, the model focused on the road projects of AVaTu;a Brasil and successor programs, using government documents to create a data coverage yielding changes in measures for the roads variables that would result if the projects were implemented. The two infi'astructure scenarios entertained were for settings in 2020, with and without Aval1l;a Brasil improvements in the transportation system. For governance, three degrees of effectiveness of development restraints on public and private lands were assessed. For the case of "no governance," development (and LULCC) occuned without limitation, while for the case of "perfect governance," no changes happened on public lands,
and private holdings were only deforested to 20% of their extent at maximum (excepting, of course, those areas where that amount had alreaCly been exceeded). For "medium governance," 100% protection occurred on areas protected by the federal gov9~1l11ent, mostly national parks and indigenous reserves (that is, no deforestation occUlTed), whereas 75% protection (or up to 25% deforestation) occurred on federal "sustainable use" lands (e.g., national forests) and state protected areas, and 50% protection occUlTed on state sustainable use lands (e.g., state forests). FUliher, a 50% rule applied to private holdings (i.e., disallowing land clearance exceeding 50%). The model was fit using empirical data for past time periods, and then the projected changes in independent variables (where relevant) were applied with the estimated parameters to project the magnitude of deforestation in 2020. Here we present results that afford systematic comparisons among the scenarios. For the scenario of expected population
growth/Avanya Brasil investments/no governance (henceforth scenario 1), deforestation reaches 31 % in 2020 (Figure 1); in the expected population growth/Avanya Brasil investments/medium governance scenario (scenario 2), deforestation rises to 19% (Figure 2); in the expected population growthlNO Avanya Brasil investments/medium governance scenario (scenario 3), deforestation is 19% in 2020 (Figure 3); and with low population growthINO Avanya Brasil investments/complete governance (scenario 4), deforestation reaches 16% (Figure 4). Of importance to the debate about road improvements in Amazonia are the results for scenarios 2 and 3, with and without Avanya Brasil, for expected levels of population growth and medium governance. Note that hardly any difference in deforestation in 2020 is observable between scenarios 2 and 3, which indicates the minimal impact of the infrastructure projects as defined. It is important to keep in mind, however, that these projections only consider the road projects of Avanya Brasil and that these
Expected Population Growth, Road Investments, and Low Governance 70'O'OW
GO'O'O'W
GO'O'O'W
Expected Population Growth, Road Investments, and Partial Governance 70 (fOVI
0'0'0'
0'0'0"
10'0'0"8
10'0'0'8
Prob Deforestation •
0,000000 - 0,200000
•
N
0,200001 - 0.400000 0.400001 - 0,600000 0,600001 - 1,000000
2020 31% Deforestation
cerrado
0,000000 - 0,200000 0,200001 - 0.400000
2020
0.400001 - 0,600000 0,600001 - 1.000000
19% Deforestation
cerrado
no data
no data
N
Note: Population growth "'lith historical levels of regional out-migration
Avan98 Brasil projeots implemented (200Q...2010) Enforcement of codes In protected areas as follows:
Indigenous lands and Federel protected areas;
100%
Federal sustainable Use ereas and State protected arees: 75%
2()'O'O'$
70'O'OW
GO'O'O"W
GO'O'O'W
Figure 1. Deforestation by 2020, scenario I: Expected population growth, Avanya Brasil investments, no governance.
L
iist~al~.s~u~sta~/n~ab~'.~uEs.~8~re8~s~:~~~~ Enforcement of a 50'Jj;> rule on private holdIngs _ __=50:%_ __..----5J5~O~~~~~~~~~~~~K~1I0~m~e~teJrs 70'fJO'W
89
f;lYffO"",,,
Figure 2. Deforestation by 2020, scenario 2: Expected population growth, Avanya Brasil investments, partial governance.
:<'\)'0'0-5
90
SCENARIOS OF FUTURE AMAZONIAN LANDSCAPES
PERZ ET AL.
low
Investments, and Partial Governance
EXiPelc:te'd Population Growth, No
High 700'[r'l/
7QOOVI
4000W
000
100('-S
Prob Deforestation . . 0.000000 - 0.200000
000'
woos
100'OS
. . 0.000000 - 0.200000
0.200001 - 0.400000
0.200001 - 0.400000
0.400001 - 0.600000
N
0.400001 • 0.600000 0.600001 - 1.000000
0.600001 - 1.000000 cerrado
no data
20mJS
L
Note:
Cerrado
Population growth with hlstoricallevals of regional out-migration
No AV81Wa BrasU projects Implemented Enforcement of codes in protected areas as follows: Itldegl;l/1ous fends 8nd Federal protected arees' Federal susteinable use areas end Federal protected areas' State sustainable use areas:
~En~fo~rc~.m~e~nt~o~fa~50~~~~r~ul:.o~n~p~riv~at:.h~O~ldi~ng~S 70'(JOW
91
~=
ti}'OTfW
no
data
I"ote:
PopUlation gro'v'lth with increased rates of out-migration
Indigenous lands and federal protected areas:
530 Kilometers
.J~!iS~~~;!!!!!!!!!!!!!!!!~__=~. WO'iJW
41)"O'OW
Figure 3. Deforestation by 2020, scenm10 3: Expected Population growth, no Avanya Brasil investments, partial governance.
4.3. Econometric Models ofLULCC: Strengths and Limitations Econometric models can thus fit empirical data and apply the estimated parameters to projected data in order to construct scenarios of future LULCC. A strength of econometric models is their emphasis on calibration, i.e., model specification and diagnosis of estimation problems due to bias, inconsistency, and other potential estimation pitfalls, as well as use ofparameters not estimated in the model directly, such as when they are taken from other studies. Regressions provide point estimates of parameters, which are typically combined with inferential statistics to provide estimates of probable error in the parameter estimates. Consequently,
N
1"0 Avanca Brasil Projects implemented Enforcement of codes in protected areas as follows:
econometric models are very useful for empirically verifying or falsifying theoretical hypotheses about the variables most important for LULCC. A key issue for modeling ofLULCC, and another strength of econometric models, concerns handling of spatial processes.. The example provided here drew on econometric models that took account of spatial autocorrelation, that is, statistical association in measurement among spatial units of observation [Anselin, 1988,2003]. This was necessary because deforestation in Amazonia occurs largely along road corridors, which stretch across pixels, census tracts, municipalities, and other units of analysis. Spatial econometric tests can evaluate the statistical significance of autocolTe1ation as well as identify the appropriate econometric model that can best account for the spatial autocorrelation in order to estimate unbiased parameters [cf. Paz and Skole, 2003; Pfaff et al., 2007].
100%
Federal Sustainable use areas and federal protected areas: 100% 2000S ' - -
State Sustainable use areas: ;;;.En.:;,fo,;crc:;:e;;.;m;;,:en:.:..1;;,:of.::a.::20::.;~,;;,;b '.::UI.::e..:::on:..!p:::;ri::;;va::.:te~h::::ol;;:,:din:,:,g.::s ro~w
projects are almost entirely of paving preexisting highways where LULCC has been occurring for decades.
2020 16% Deforestation
540
270
o
Km
100%
-.l ffiaow
<':o'o'os
sj'(j(rw
Figure 4. Deforestation by 2020, scenario 4: Low population growth, no Avanya Brasil investments, complete governance.
The foregoing examples also point out limitations, or at least the difficulties, of working with econometric models. For dynamic processes such as LULCC, multitemporal data are necessary, preferably for three or more time points in order to determine whether there are nonlinearities in changes over time. This imposes serious data requirements. Similarly, econometric models such as linear regression, by default, assume linearity in the relationships between variables, which is not necessarily the case. While the transformations available to account for nonlinearity are generally easy to do, they can require time to fully assess. Another imporiant issue in econometric models for sce~ nario development is endogeneity. In many LULCC models, there is multiway causation and simultaneity among the variables at hand, and if this is ignored and simple one-way causal relationships are categorically assumed in model
specification, the results can be biased and velY likely incorrect. Static models do not easily incorporate multiway causation. Moreover, dynamic models with panel data mayor may not capture feedbacks occurring over time, depending on whether the time steps afforded by available data correspond to the actual feedback processes operating. Econometrics, however, includes diagnostic tests for the presence of endogeneity, as well as specific methods including instrumental variable models and others that can account for simultaneity among variables. The choice of the most appropriate specification depends on the range of endogenous variables and the availability of instrumental variables (i.e., variables related to one but not all endogenous variables that can be used to estimate values of the related endogenous variable). A final preoccupation with econometric models is the issue of scale. Economic theory recognizes micro- and macrolevel
92
SCENARIOS OF FUTURE AMAZONIAN LANDSCAPES
processes, and econometric models, including those presented here, account for scale-specific mechanisms that can influence LULCC. These are useful simplifications for they afford madgeable models. But limitations in data availability often mean that modeling of LULCC over a large area, such as Brazilian Amazonia, requires use of, e.g., aggregate data, rather than data for individuals or entire economies, where economic theory places its emphasis. This is resolved theoretically in economics by assuming that individual preferences additively sum to thdse that manifest in aggregate counts, a simplification that may gloss over important distinctions among interest groups, local communities, and other social actors that can also affect LULCC [Wood, 2002). The existence of driving forces of LULCC that operate on one or another of many levels of scale, and the interactions among them via cross-scale processes, complicate understanding of mechanisms that affect LULCC and are often addressed in a limited fashion in econometric models [e.g., Brown et al., 2004a: 399--400; Evans et al., 2005]. Multilevel econometric modeling offers a way to address these concems, but multilevel models are still constrained by limitations of available data.
PERZ ET AL.
landscape in which they interact, using lUles that describe agent behaviors and environmental feedbacks that influence subsequent agent land use decisions. Drawing on these general discussions of CA and ABMs in modeling LULCC, we provide examples of dynamic simulation models applied to LULCC in Ecuadorian Amazonia. Finally, we note how econometric models can be used to inform dynamic simulation models. We conclude with a discussion of the strengths and limitations of dynamic simulation modeling as applied to LULCC. 5.1. Complexity Them)! and Spatial Models
A complex system is one in which its multiple components interact in ways that link pattems and processes across scales. The complex nature of such systems is seen as emerging from nonlinearities due to large numbers of interactions involving feedbacks occurring within the system [Cilliers, 1998; Malanson, 1999; Malanson et al., 2006; Crawford et al., 2005). Complexity theory holds that systems cannot be suitably understood without a focus on the feedbacks and nonlinearities that lead to emergent multiscale phenomena [Matthews et al., 1999; Manson, 2001]. Complexity theories are being used to examine the causes 5. DYNAMIC SIMULATION MODELS and consequences of LULCC dynamics in Amazonia and In many respects, dynamic simulation models comple- beyond. Studies in LULCC dynamics that look for universal ment econometric models by avoiding many of the limita- prope1iies in spatially extended systems highlight feedback tions of statistical estimation. Whereas econometric models mechanisms and critical thresholds to link people, place, and generally proceed inductively, working from available data the environment across space-time scales and, fUliher, seek which may be limited in certain ways, dynamic simulation to understand system behaviors and trajectories influenced models can proceed more deductively, taking as their point by pattem-process relations and system dynamics [Matof departure what is known about a system and using that tlwws et al., 1999; Manson, 2001). Properties emerging from to inform assumptions in a model in order to evaluate the local nonlinear feedbacks constrain the evolving patterns of system's behavior. This also frees up dynamic simulation LULCC and can produce a system with identifiable future modeling from requirements to work only with the spatial altemative states in which instabilities can "flip" a system units at hand in available data. Dynamic simulations can also into another behavior regime by changing the pattems and build in assumptions about spatial contagion (which results processes that control land use change [Wolfi'am, 1984; in spatial autocorrelation) and cross-scale processes (across Blaclanan, 2000). Spatially explicit modeling approaches such as CA and any of several levels of scale). ABMs are highly suited to the study of landscape dynamics In this section, we discuss the foundations of dynamic and complexity theOly and how landscape pattems form and simulation modeling in complexity (systems) theory, focusevolve. These modeling approaches allow LULCC researching on the case of LULCC. We pay particular attention to ers to study landscape pattems using spatial simulations, the linkage of pattern to process, consideration of exogenous identify candidate explanations for specific landscape patand endogenous factors, and cross-scale interactions. We tems, assess endogenous and exogenous factors in specific then focus on the use of cellular automata (CA) and agenttrajectories of landscape change, and develop altemative based models (ABMs) as key elements in dynamic simulasystem states as scenarios of LULCC. Dynamic simulation tion modeling. We begin by discussing spatial organization models thus allow deductive development of scenarios, that and how CA models are equipped to examine spatial StlUCis, identification of probable LULCC trajectories based on ture using standardized grid cells and a set of lUles to guide theoretical assumptions about the operation of a complex simulation. We then discuss how ABMs are used to incorporate the interactions among human agents and the dynamic system.
5.2. Spatial Olganization and Cellular Automata Models
The spatial organization of landscapes manifests itself in various forms acrg'ss scales. The nature of the form is likely to change with s)ile changes, leading to changes in landscape homogeneity 01' heterogeneity [Turner et al., 1989; Atlanson and Tate, 2000]. Understanding the nature and genesis of these respective spatial characteristics is a computational ecology question [Levin, 1992; Stoms, 1994; Tumer et al., 1989). As ecological variables are inherently spatial, the spatial variation of those variables in a landscape is a function of endogenous or exogenous processes occurring in that landscape [Tumer, 1990). Assessment of landscape variability can feature pattem metrics [Li and Wu, 2004], where changes in spatial pattem indicate changes in underlying processes or initial conditions [Meen temeyer, 1989; Levin, 1992; Qi and Wu, 1996]. Assumptions regarding relationships between pattem and process have been used to characterize landscape trajectories and to make comparisons across multiple landscapes over space and time [Rindfilss et al., 2007). CA are spatially explicit, grid-based models that allow assessment of pattem generation in landscapes based on assumptions about the processes presumed to generate certain pattems. CA models generally operate on cells that are contiguous and homogeneous in the sense that they can take one of a common set of possible states (i.e., land cover categories). CA models incorporate lUles for changes in cell states. Rules are based on the modeler's understanding ofprocesses at work in the landscape and expectations about the resulting landscape pattem. More specifically, the lUles may reflect the ecological relationships presumed among variables in the model, the characteristics of neighboring cells (spatial processes that reflect expected pattems), and past changes in the cell (endogeneity and feedbacks). CA models are iterative, calculating cell changes over time. This allows for observation of complex change on multiple scales under controlled circumstances. CA models penuit the testing of processes leading to specific pattems theorized to occur on one or another among several possible scales. In addition, CA models can be repeated with varying initial conditions or assumptions about the change processes to see whether the expected pattem metrics in the landscape still emerge. Pattern trajectories from almost all initial states tend to converge over time toward specific "attractor" states. These attractor states include a very small fraction of the total list of all possible states, demonstrating the process of' irreversible evolution, the self-organization often referred to in the complexity literature. By searching for repeating patterns on multiple spatial and temporal scales, it becomes possible to characterize the
93
complexity of a system, at least in limited terms. There are four classes of CA models in te1ms of the complexity of the pattems in the dynamics they generate [Wolfi'am, 1984]: (1) evolution into a homogeneous arrangement, (2) evolution into endlessly cycling periodic stmctures, (3) evolution producing random pattems, and (4) complex pattems with localized stlUctures changing through space and time. In the last of these classes, the type desired, a particular final pattem may evolve fi'om many different initial conditions. These types of systems require computational or algorithmic complexity equal to the explicit simulation in place [Manson, 2001). These systems are, in effect, unpredictable in a purely dete1ministic sense and must be resolved by explicit simulation. Self-organization and the nature of the attractor states determine the form and extent of the final, emergent pattems. Some systems contain multiple variables with apparent complexity; however, a system based on multiple variables does not, in itself, guarantee complexity. In essence, multiple interactions impossible to characterize using traditional mathematical fOlms give the appearance of "complexity," while actually being more precisely defined as "complicated." This distinction is fundamental in any modeling endeavor purporting to be complex systems-based. Frequently, modeling environn}.ents found in the literature confuse the two. Tightly specifying models creates a situation where the results are preordained, and the "complexity" is controlled in a manner that violates the underlying assumptions of complex systems. 5.3. Agent-Based Models
ABMs incorporate human agents who make decisions that affect the cells in a modeled landscape [Parker et al., 2003]. ABMs thus go beyond CA models via incorporation of human action by relating human decisions to landscape characteristics. Agents have their own characteristics (various types of assets, preferences about goals, etc.) that complement the characteristics of the landscape. Human action is incorporated via the addition of a human decision model linked to the CA model. Generally, in the context of ABMs for the case ofLULCC, agents are assumed to act rationally, to maximize their utility per economic action. This raises issues of incomplete information and other aspects ofbounded rationality [Parker et al., 2003; Evans et al., 2005]. Like CA models, ABMs are iterative models that lend themselves to dynamic simulations. Agents and landscapes receive initial values, agents make decisions during the first time step, the model generates the results at the cellular level based on agent decisions to modify the landscape and based on ecological processes, agents and landscape cells
94
SCENARlOS OF FUTURE AMAZONIAN LANDSCAPES
receive modified values, and the cycle repeats. ABMs can incorporate agent learning, such that outcomes that do not maximize utility yield different decisions later. Across many time stepsr ABMs thus generate spatially explicit changes in landscapes as well as the characteristics of agents (e.g., changes in assets), which may be nonlinear. As their name suggests, ABMs emphasize proximate determinants of LULCC via the focus on human decisions that directly alter the landscape, but ABMs can serve as the foundation for multiagent-ba'sed models as well [Parker et al., 2003). There, multiple agents, with varying characteristics, make decisions about different land parcels, each with their own ecological characteristics. Agents may interact and influence each other, and together, agents participate in markets. Agent interactions and market price changes, along with biophysical factors, constitute contextual opportunities and constraints for individual agents that may modify their decision making. ABMs with multiple agents thus incorporate dynamic elements for agents and landscapes, allowing for observation of space-time dependency in decision making and landscape change, observation of exogenous and endogenous factors, and feedback mechanisms and critical thresholds. By simulating the individual actions of many diverse agents, and characterizing the resulting system behavior and outcomes over time, ABMs can be useful tools for studying the effects of processes on patterns, and patterns on processes, that operate at multiple scales. Experimentation is the key to understanding similar and different envirolli11ents, environments shaped by similar and different processes, and environments influenced by the context of place where the synthesis or distillation of patternprocess relations across sites may be difficult to discern. By taking into account their commonalities and differences in structure, function, and evolution across time and space, ABMs offer the possibility of considerable insights into system dynamics. Agent-based experiments provide flexibility and considerable analytical power to examine, for example, the effects of implementing a proposed policy. ABMs have recently been used to explore complex systems in the cases of LULCC [Deadman et al., 2004; Evans and Kelley, 2004; Brown et al., 2004b, 2005; An et al., 2005], rural to urban migration [Jaylson et al., 2006], ecosystem management [Nute, 2004; Bousquet et at., 2001], and agricultural economics [Berger, 2001). ABMs use a bottomup approach to allow the testing of alternative theories to explain specific observable patterns [Grimm and Railsback, 2005). The generation of different types of landscape patterns over space and time based on different theoretical approaches yield a set of future scenarios of change that can include endogenous changes and exogenous shocks, which can alter trajectories of landscape change. Emergent be-
PERZ ET AL.
havior is seen at a regional scale as an outcome of actions and patterns at local settings [Malanson, 1999; Walsh etal., 2006,2008a). 5.4. Applications ofComplexity Them}' and Simulation Models to LULCC in Ecuadorian Amazonia
Tropical forest frontiers have been studied as complex, self-organized systems. For instance, Malanson et al. [2006] used power-law distributions of advancing deforestation to suggest the emergence of complexity in northern Ecuadorian Amazonia. They reported the existence of a "development front" that occurs as a consequence of the collective actions of individual farm households and their decisions about deforestation and agricultural extensification. The development front manifested as a land conversion wave in the landscape at the regional scale [Walsh et al., 2008a). Messina and Walsh [2001, 2005] developed spatially explicit simulations using CA approaches to examine LULCC, i.e., deforestation, urbanization, and agricultural extensification, in Ecuadorian Amazonia, as well as the rates and spatial patterns of land conversion relative to a set of initial conditions, spatial processes, and growth or transition rules. Their findings suggested that a more homogeneous landscape evolved over time as a consequence of household decision making, primarily influenced by household demographics, resource endowments of farms, and geographic accessibility of households to roads and markets and administrative towns, a scenario that fits the theoretical understanding of how in-migration of farmers onto existing fanTIs through purchase and subdivision of plots, along with the growth of services and population, transform the natural landscape. Subsequent CA research [Walsh et al., 2008b] has examined the linkages between people and the environment by explicitly considering pattern-process relationships and the nahlre of feedback mechanisms among many factors that influence LULCC. In this research, the CA modeling approach emphasizes the human dimensions of LULCC by includi~g socioeconomic and demographic characteristics at the household level along, biophysical data that describe the resource endowments of farms, the geographic accessibility of farms to roads and communities, and the evolving nature of human-environment interactions over time and space in response to a host of exogenous and endogenous factors. The work is informed by Ecuadorian population census data, a longitudinal household survey administered by the Ecuador Project Team in 1990 and 1999, a cross-sectional survey of communities administered in 2000, GIS coverages indicating geographic access of farms to roads and communities, and a satellite image-based LULCC time series used
to initialize the CA model and create landscape transition probabilities to guide spatial simulations. LULCC scenarios were examined by;domparing model outcomes generated for a base CA mO'del and an alternative CA model to explore the effect§,.tbf increases in household income on land use change patferns at the farm level, achieved as a consequence of improved geographic accessibility to roads and communities and increased off-fan11 employment. Findings indicate that increases in household income are associated with more land in pasture and more land being cultivated for crops. In addition, more land in secondmy forest succession occurs as a consequence of greater access to roads and communities, thereby, affording a better opportunity for off-farm employment and greater levels of household income.
95
Messina et at. [2006] explored the varying trajectories of landscape change as a result of different degrees of protection in and around the Cuyabeno Wildlife Reserve in northeastern Ecuador, using ecological pattern metrics, a satellite image time series, and neutral models. Findings indicate that LULCC occurred within the context of conflicts attributed to the emergent land tenure systems surrounding the Cuyabeno Wildlife Reserve. Patrimony forest, a restricted land use categOly on lands sunendered to colonists who had settled in the reserve, facilitated colonization and communal land titles. Changes in the degree of protection and the implementation of protection buffers have accelerated the process of fragmentation and exacerbated the conflict between development and protection, resulting in a pronounced deviation in the spatial structure of the landscape from what was theoretically expected. Changes in land tenancy and the im5.5. Econometric Models, Simulation Models, and LULCC plementation of protective buffers exacerbated the process of deforestation and forest fragmentation. Thus far in this chapter, we have treated econometric Mena et al. [2006] analyzed the drivers of deforestation models and dynamic simulation models as separate ap- and secondaly forest succession within the buffer area of the proaches to modeling LULCC for scenario development. Cuyabeno Wildlife Reserve using a socioeconomic and deAs we have noted, however, the two types of models have mographic survey collected by the MinistlY of Environment certain complementarities and may be used to inform each of Ecuador and a spatial lag model to account for spatial other. Here we report econometric analyses of LULCC in autoconelation. Findings indicated that different combinanorthern Ecuadorian Amazonia. Each of the following ex- tions of factors coptributed to the generation of secondmy amples draws on empirical data, focuses on pattern metrics, forest and fallowed lands in 1990 and 1999. Off-farm emand analyzes those metrics using econometric models with ployment, household assets, and male adults on the farm data for households. The results indicate which variables are were consistently significant in the models. Other important important for various LULC outcomes and can serve as the factors included the percentage of the farm under legal title basis for agent decision lUles, incorporation of specific types and distance to water. Factors contributing to a change in the of agent interactions, and consideration of processes that op- area of secondary forest and fallow between 1986-1996 and erate on larger scales such as the community or landscape 1996-2002 were vehicle access to farms and hired labor. levels, to infOlm ABMs and dynamic simulations. Household decisions about LULCC are shaped by their Pan et at. [2004] studied the composition and spatial association with neighboring farms, development sectors organization of deforestation, agriculture, and secondaly (clusters of farms), and local communities. As such, an implant succession at the farm level in Ecuadorian Amazo- pOltant issue in the fonTIulation of econometric models is the nia, through the integration of data from the 1990 and 1999 estimation of zero-sum outcome variables (land use shares) household survey and an assembled satellite image time se- through the specification of effects at not only the household ries. Pattern metrics for LULC were calculated at the farm level, but also the community level. To do this, a multilevel level through the generation of a hybrid digital classification model can be used to assess the covariance structure in an efof Landsat Thematic Mapper data. Generalized linear mixed fort to account for within-area heterogeneity (i.e., subdivided models were then used to study pattern-process relations in farms within development sectors and/or within the "hinter1990 and 1999 using various measures of spatial structure land" of communities), hierarchical effects (i.e., viewing (i.e., ecological pattern metrics) at the landscape level as the small areas as part of larger areas), and spatial effects (i.e., dependent variables. Results indicated that rapid popula- spatial linkages offarms to other farms and between commution growth caused substantial subdivision of plots, which, nities). Pan and Bilsborrow [2005] used a multilevel model in turn, created a more complex landscape in 1999 than in' to assess C0ll1111unity effects on LULC patterns at the fmm 1990. Key factors predicting the spatial organization of the level, focusing on the number of patches in each LULCC landscape at the farm level were population size and compo- class on the farm. Findings indicate that communities with a sition, plot ownership fragmentation, expansion of the road civil registrar appear to playa larger role in influencing the and electricity networks, age of plot, and topography. number of patches on a farm, and less fallow land exists on
96
SCENARIOS OF FUTURE AMAZONIAN LANDSCAPES
these more connected farms because of market opportunities at nearby towns. Contextual factors at the community level significantly affect LULC at the farm level. Fmiher, months of hired labor, size of the fatID, year of farm establishment, and change in population density are important factors in the pattem ofLULCC at the farm level. 5.6. Dynamic Simulation Models ofLULCC: Strengths and Limitations
Dynamic simulation models can thus support development of scenarios for LULCC by linking lUles about agent decisions and cell changes to iterative models that yield landscape change over time. A strength of dynamic simulations is that CA models allow for visualization of landscape changes at high spatial and temporal resolutions (i.e., pixels and individual years). Visualizations can be interpreted by researchers, policymakers and stakeholders, which facilitates science-to-policy and science-to-action transfers of lmowledge about LULCC. Another strength of dynamic simulations is that ABMs emphasize the linkage of human decision making by land managers to LULCC, allowing for dynamic modeling of key proximate determinants of LULCC. This can be based on theoretical expectations about human action (e.g., utility maximization, bounded rationality, etc.) or on empirical data from past time points. Finally, dynamic simulations are also well suited to producing output pertaining to the future, looking beyond analyses based on available data. This makes dynamic simulations well suited to scenario development, via alterations in model assumptions, especially those that concem the context in which agents operate, such as changes in policy envirol11l1ents. The ability to simulate change under controlled circumstances in dynamic simulation models allows for a deeper exploration of how complex systems work. This makes calibration and validation of dynamic simulations an important enterprise. To the extent that data for past time points are available, we can calibrate simulation models by basing them on past pattems of LULCC and human decisions about land use and compare the resulting model to the historical data. We can also conduct sensitivity analyses on the model by altering assumptions based on past data to see how such alterations affect the model results, vis-a-vis the data, Sensitivity analysis can proceed by modifying the lUles for cell changes (i.e" testing for sensitivity of parameters) or by modifying the cell sizes (i.e., testing for scale dependency), among other possibilities. Validation of simulation models has also received considerable attention and has frequently been the focus of calls for more inquily [e.g., Brown et al., 2004a; Evans et al., 2005; Verberg et aZ., 2006]. Validation generally refers to
PERZ ET AL.
assessment of pattems in outcomes, but it can also refer to assessment of processes generating the pattems (i.e., model stlUcture), Whereas dynamic simulation models emphasize pattem generation, as in CA models, there is a need to conduct validation not only on resulting pattems, especially for longer-run simulations, but also on the processes generating those pattems, especially in multiagent ABMs. Validation of simulations via comparisons of model output to data for time points beyond those available when the model was built usually emphasizes comparisons to satellite imagelY. Validation also needs to consider changes in the drivers of land use, especially when empirical LULC trajectories change. Beyond the challenges of calibration and validation, dynamic simulation models bear limitations. One issue is the question of whether to build simulation models deductively (based on theory) or inductively (based on empirical data). As has been noted here, there are advantages to proceeding inductively, for the resulting model will have an empirical foundation, but available data may impose limitations on what is feasible. Proceeding deductively allows for somewhat greater control over model constlUction, which can allow for more straightfOlward exploration of the dynamics of the system under study, though that begs other questions, One concems the issue of how simple or complicated a dynamic simulation model should be. While a key decision concems the number of variables and parameters to be incorporated, a related issue is the number of feedbacks and spatial relationships to be included. As noted earlier, building in more complicated assumptions in the model can obscure the possibility of observing complex dynamics arising from relatively simple assumptions. Related to these issues in dynamic simulation modeling is another impOliant limitation, namely, that these models do not account for unexpected shocks ("surprises") such as major weather events, regime changes, wars, and the like. There also remain challenges pertaining to spatial scale when attempting to use results fi'om econometric models to inform dynamic simulation models [cf. Evans et aI" 2005: 211-213]. In principle, the two types of models can be compleme~taty, such as when econometric output yields estimates of parameters to support design of decision lUles in ABMs. However, the spatial scale (both spatial resolution and spatial extent) as well as the temporal depth of empirical models rarely matches those desired in dynamic simulations. If econometric results come from aggregate data, there is the risk of ecological fallacies, i.e" imputing individual behaviors from data based on aggregates or assuming that the same pattems emerge on different spatial scales. Similarly, if econometric results pertain to one site among many diverse places in a region as large as Amazonia, simulations can also be found wanting.
6. CONCLUSION
97
envisioning, modeling, and scenario development, followed by stakeholder evaluation and further envisioning. Econometric and simulation models (especially when it includes stakeholders), also have applications beyond modeling exercises, such as post hoc validation to see if model projections are in fact accurate to describe events later in time, or by affecting public policies or social action that changes LULCC outcomes. Both applications are, however, difficult to directly evaluate, As noted above, post hoc validation is difficult due to data requirements. We repeat that the whole point of simulation modeling as applied to scenarios is not so much to gain an accuracy assessment of model fit to future change as it is to serve as a planning instlUment for decision makers and stakeholders to act on the basis of information about the consequences of LULCC under specific assumptions. Conceming econometric models, the standard practice for evaluation is to report goodness of fit and to lUn diagnostics for potential problems such as endogeneity. In general, spatially explicit LULCC models rarely report on spatial goodness of fit [e.g" Walker, 2003], though this type of assessment is becoming a priority for LULCC modelers [e.g., Verberg et al., 2006]. It is also rare for a given LULCC model to have direct, concrete policy impacts. While there are arguably examples of high-profile p'1blications that have stimulated policy debate conceming deforestation in Amazonia [cf. Laurance et al., 2001], we are,not aware of systematic studies ofLULCC models and their! policy ramifications. It is probably safer to say that broader initiatives such as the LBA program, or lNPE's ongoing monitoring program, have had impacts in a more diffuse sense of making data and findings available in the public domain, which, in tum, can serve as a stimulus for state policies and social action to mitigate deforestation, improve land use, and othelwise yield better LULCC outcomes.
Methods for s9~tJario development vary, but they are generally demat;ping and yet remain wanting in certain respects, Alcamo let al. [2006] list several needs that coneem LULCC scenario development, including more detail about the driving forces (processes) behind land use, an expanded suite of LULCC outcomes of interest (beyond deforestation for agricultural land), incorporation of extreme events, and two-way dialogue with stakeholders. Conversely, Verberg et aI, [2006] emphasize improvements in modeling that stem from improvements in scenario development, notably the need for more systematic methods of translating qualitative scenarios into quantitative models, These parallels feahlre the role of models in scenario development, which reveals oppOliunities to improve LULCC scenarios via further work on LULCC modeling and vice versa, To the extent that scenario development relies on models, scenarios are limited by the limitations in the models chosen. In pati, development of scenarios can thus be improved via attention to the limitations in the underlying models, whether by incorporation of better or newer data, and a greater focus on calibration and validation. Incorporation or cross-scale processes and feedback mechanisms is also difficult and not yet widespread in LULCC models. One key issue related to validation in the context of Amazonia, given the considerable diversity in pattems and processes of LULCC there, is the ability to "scale up" from models with localized scope to broader portions of the region. Highly spatially explicit models, or models based on considerable local data, may not generalize neatly to the region as a whole, The role of stakeholders in scenario development is also key [cf. Peterson et aI" 2003], but that also raises important issues. In places with social conflicts, the envisioning process is not straightfOlward, especially if there are significant inequalities, which can result in dominance of discussion by more powerful stakeholders [cf. Edmunds and Wollenberg, REFERENCES 2001]. One possibility to address that problem is to incorporate model output as a focus for dialogue among stake- Alcamo, 1., K. Kok, G. Busch, 1. A. Priess, B. Eickhout, M. Rounholders, so that the envisioning process can build on input sevell, D. S, Rothman, and M, Heisterman (2006), Searching for from researchers and data along with the perspectives and the future of land: Scenarios fi'om the local to the global scale, experiences of stakeholders. This could open a larger space in Land-lise and Land-cover Change: Local Processes and Global Impacts, edited by E, F. Lambin and H. Geist, pp, 138-155, for dialogue beyond one focused on the grievances and poSpringer, Berlin. litical preferences of interest groups, This also takes advantage of the ability of models to reveal nonlinear dynamics Alves, D. S., D, C. Morton, M. Batistella, D. A. Roberts, and C. Souza Jr. (2009), The changing rates and patterns of deforestain complex systems, which may not be apparent to staketion and land use in Brazilian Amazonia, Geophys, Monogr, holders, while also incorporating the grounded experiences Ser" doi: 10.1 029/2008GM000722, this volume, of stakeholders, which can reveal incolTect assumptions in An, L., M. Lindelman, J, Qi, A. Shortridge, and 1. Liu (2005), models. Input from stakeholders about the credibility of the ·Exploring complexity in a human-environment system: An agentmodel scenarios can thus inform downstream adjustments based spatial model for multidisciplinaly and multiscale integrain the scenarios, affording a more iterative process between tion, Ann, Assoc. Alii, Geogr" 95(1), 54--79, ~j
98
SCENARIOS OF FUTURE AMAZONIAN LANDSCAPES
Anderson, L. E., and E. J. Reis (1997), Deforestation, development and government policy in the Brazilian Amazon: An econometric analysis, IPEA Working Paper No. 513, Rio de Janeiro, IPEA. Anderson, LitE., C. W. 1. Granger, E. 1. Reis, D. Weinhold, and S. Wunder (2002), The Dvnamics o(Dejorestation and Economic Growth in the Brazilian Amazon, Cambridge Vniv. Press, Cambridge, V. K. Anselin, L. (1988), Spatial Econometrics: Methods and Models, Springer, Dordrecht. Anselin, L. (2003), Spatial externalities, spatial multipliers and spatial econometrics, Int. Reg. 'Sci. Rev., 23(2), 153-166. Asner, G. P., M. Keller, M. Lentini, F. Merry, and C. Souza Jr. (2009), Selective logging and its relation to deforestation, Geophys. Monogr. Ser., doi: 1O.1029/2008GM000723, this volume. Atkinson, P. M. and Tate, N. J. (2000), Spatial scale problems and geostatistical solutions: A review, Prof Geogr., 52, 607-623. Baker, 1. P., D. W. Hulse, S. V. Gregory, D. White, 1. Van Sickle, P. A Berger, D. Dole, N. H. Schumaker (2004), Altemative futures for the Willamette River Basin, Oregon, Ecol. Appl., 14(2), 313-324. Barbier, E. (2001), The economics of tropical deforestation and land use: An introduction to the special issue, Land Econ., 77(2), 155-171. Berger, T. (2001), Agent-based spatial models applied to agriculture: A simulation tool for teclmology diffusion, resource use changes and policy analysis, Agric. Econ., 25(2-3), 245-260. Blackman, T. (2000), Complexity theOlY, in Understanding ContempoI'm)! Society: Theories ofthe Present, edited by G. Browning, A Halcli, and F. Webster, pp. 139-151, Sage Publications, London. Bousquet, F., C. Le Page, 1. Bakam, and A TakfOlyan (2001), Multi-agent simulations of hunting wild meat in a village in eastem Cameroon, Ecol. Modell., 13, 331-346. Brondizio, E. S., A Cak, M. M. Caldas, C. Mena, R. Bilsborrow, C. T. Futemma, T. Ludewigs, E. F. Moran, and M. Batistella (2009), Small farmers and deforestation in Amazonia, Geophys. Monogr. Sa., doi:lO.l029/2008GM000716, this volume. Brown, D. G., R. Walker, S. Manson, and K. Seto (2004a), Modeling land use and land cover change, in Land Change Science: Observing, Monitoring and Understanding Trajectories of Change on the Em!" 's Sl/Ijace, edited by G. 'Gutman et aI., pp. 395-409, Springe~;, Dordrecht. Brown, D. G., S. Eiage, R. L., Riolo, and W. Rand (2004b), Agent based and analrical modeling to evaluate the effectiveness of greenbelts, Environ. Modell. Software, 19(12), 1097-1109. Brown, D. G., S. E. Page, R. L. Riolo, M. Zellner, and W. Rand (2005), Path dependence and the validation of agent-based spatialmodels ofland-use, Int. J. Geogr. b1f. Sci., 19(2), 153-174. Chomitz, K. M., and D. A Gray (1996), Roads, land use, and deforestation: A spatial model applied to Belize, World Bank Econ. Rev., 10(3), 487-512. Chomitz, K. M., and T. S. Thomas (2003), Determinants ofland use in Amazonia: A fine-scale spatial analysis, Am. J. Agric. Econ., 85(4), 1016-1028. Cilliers P. (1998), Complexity and Postmodemislll, Routledge, New York.
PERZ ET AL. Crawford, T., Messina, 1., Manson, S. and D. O'Sullivan (2005), Complexity science, complexity systems, and land use research, Environ. Plann. B Plann. Des., 32, 792-798. Cropper, M, 1. Puri, and C. Griffiths (2001), Predicting the location of deforestation: The role of roads and protected areas in north Thailand, Land Econ., 77(2), 172-186. Deadman, P., D. Robinson, E. Moran, and E. Brondizio (2004), Colonist household decision-making and land use change in the Amazon rainforest: An agent-based simulation. Environ. Plann. B Plann. Des., 31, 693-709. Edmunds, D., and E. Wollenberg (2001), A strategic approach to multistakeholder negotiations, Dev. Change, 32, 231-253. Evans, T. P., and H. Kelley (2004), Multi-scale analysis ofa household level agent-based model of land cover change, Environ. Manage., 72(1-2),57-72. Evans, T. P., E. Manire, F. de Castro, E. Brondizio, and S. D. McCracken (200 I), A dynamic model of household decision making and parcel level land cover change in the Eastern Amazon, Ecol. Modell., 143,95-113. Evans, T. P., L. D. K. Munroe, and D. C. Parker (2005), Modeling land-use/land-cover change: Exploring the dynamics of humanenvironment relationships, in Seeing the Forest and the Trees: Human-Environment 1nteractions in Forest Ecosystems, edited by E. F. Moran and E. Ostrom, pp. 187-213, MIT Press, Cambridge, Mass. Geist, H. 1., and E. F. Lambin (2002), Proximate causes and underlying driving forces of tropical deforestation, BioScience, 52(2), 143-150. Grimm, V., and S. F. Railsback (2005), 1ndividual-Based Modeling and Ecology, Princeton Univ. Press, Princeton, N. 1. Gutman, G., A. C. Janetos, C. O. Justice, E. F. Moran, 1. F. Mustard, R. R. Rindfuss, D. Skole, B. L. Turner II, and M. A. Cochrane (Eds.) (2004), Land Change Science: Observing, Monitoring and Understanding Trajectories of Change on the Earth's Surface, Springer, Dordrecht. Jaylson 1. S., A. L. Espindola, and 1. T. P. Penna (2006), Agent-based model to rural-urban migration analysis, Physica A, 364,445-456. Kaimowitz, D., and A. Angelson (1998), Economic Models of Tropical Deforestation: A Review, CIFOR, Jakarta, Indonesia. Lambin, E. F. (1994), Modeling Deforestation: A Revie'w, Office for Official Publications of the European Community, Luxembourg. Lambin, E. F., and H. Geist (Eds.) (2006), Land- Use and Land-Covel' Change: Local Processes and Global Impacts, Springer, Berlin. Laurance, W. F., M. A Cochrane, S. Bergen, P. Fearnside, P. Delamonica, C. Barber, S. D' Angelo, and T. Fernandes (2001), The future of the Brazilian Amazon, Science, 291, 438-439. Levin, S. A. (1992), The problem ofpattem and scale in ecology, Ecology, 73, 1943-1967. Li, H., and 1. Wu (2004), Use and misuse of landscape indices, Landscape Ecol., 19, 389-399. Malanson, G. P. (1999), Considering complexity, Ann. Assoc. Alii. Geogr., 89(4), 746-753. Malanson, G. R., Y. Zeng, and S. J. Walsh (2006), Complexity at advancing ecotones and frontiers, Environ. Plan. A, 38, 619632.
99
Manson, S. M. (2001), Sirnplifying complexity: A review of com- Peterson, G. D., G. S. Cumming, and S. R. Carpenter (2003), Sceplexity theOlY, GeofD/'um, 32, 404-414. nario planning: A tool for conservation in an uncertain future Matthews, K. B., A ~: Subaald, and S. Craw (1999), ImplemenConserI'. Bio!., 17,358-366. ' tation of a spatij!)' decision support system for rural land use Pfaff, A., et al. (2007), Road investments, spatial spillovers, and planning: Integl;ating geographic information systems and endeforestation in the Brazilian Amazon, J. Reg. Sci., 47(1), 109vironmental 1)J6dels with search and optimization algorithms, 123. Comput. Electron. Agric., 23, 9-26. Pfaff, A., A. Barbieri, T. Ludewigs, F. Meny, S. Perz, and E. Meentemeyer, V. (1989), Geographical perspectives of space, time, Reis (2009), The impact of roads in the process of deforestaand scale, Land~'cape Ecol., 3,163-173. tion, Geophys. Monogr. Ser., doi: I 0.1029/2008GM000737, this Mena, C. F., A. Barbieri, S. 1. Walsh, C. M. Erlien, F. L. Holt, and volume. R. E. Bilsborrow (2006), Pressure on the Cuyabeno Wildlife RePfaff, A S. P. (1999), What drives deforestation in the Brazilian serve: Development and land use/cover change in the Northern Amazon? Evidence from satellite and socioeconomic data, J. Ecuadorian Amazon, World Dev., 34(10),1831-1849. Environ. Econ. Manage., 37, 26-43. Meliens, B., W. Sunderlin, O. Ndoye, and E. F. Lambin (2000), Qi, Y., and 1. Wu (1996), Effects of changing spatial resolution on Impact of macroeconomic change on deforestation in southern the results of landscape pattern analysis using spatial autocorCameroon: Integration of household survey and remotely-sensed relation indices, Landscape Ecol., 11,39-49. data, World Dev., 28(6), 983-999. Reis, E., and R. M. Guzman (1994), An econometric model of Messina, 1. P., and S. 1. Walsh (2001), 2.5D morphogenesis: ModAmazon deforestation, in The Causes o,(Tropical Deforestation, eling land use and land cover dynamics in the Ecuadorian Amaedited by K. Brown and D. W. Pearce, pp. 172-191, UCL Press, zon, Plant Bcol., 156(1), 75-88. Vancouver. Messina, 1. P., and S. 1. Walsh (2005), Dynamic spatial simulation Rindfuss, R. R., B. Entwisle, S. J. Walsh, C. F. Mena, C. M. Ermodeling of the popUlation-environment matrix in the Ecuadolien, and C. L. Gray (2007), Frontier land use change: Syntherian Amazon, Environ. Plann. B Plann. Des., 32, 835-856. sis, challenges, and next steps, Ann. Assoc. Am. Geogr., 97(4), Messina, 1. P., S. 1. Walsh, C. F. Mena, and P. L. Delamater (2006), 739-754. Land tenure and deforestation patterns in the Ecuadorian AmaSoares-Filho, B., C. Pennachin, and G. C. Cerqueira (2002), DIzon: Conflicts in land conservation in frontier settings, App!. NAMICA-A stochastic cellular automata model designed to Geogr., 26(2),113-128. simulate the landscape dynamics in an Amazonian colonization Moore, N., E. Arima, R. Walker, and R. Ramos da Silva (2007), frontier., Eco!. MOdell., 154, 217-235. Uncertainty and the changing hydroclimatology of the Amazon, Soares-Filho, B., A Alencar, D. Nepstad, G. Cerqueira, M. del C. Geophys. Res. Lett., 34, Ll4707, doi: 10. 1029/2007GL0301 57. Vera Diaz, S. Rivpro, L. Sol6rzano, and E. Voll (2004), SimulatMoran, E. F., and E. Ostrom (Eds.) (2005), Seeing the Forest and ing the response Of land cover changes to road paving and govthe Trees: Human-Environment Interactions in Forest Ecosysernance along a major Amazon highway: The Santarem-Cuiaba tems, MIT Press, Cambridge, Mass. cOlTidor, Global Change BioI., 10, 745-764. Nelson, G. C., V. Harris, and S. W. Stone (2001), Deforestation, Soares-Filho, B. S., D. C. Nepstad, L. M. Curran, C. Coutinho Cerland use, and property rights: Empirical evidence from Darien, queira, R. A Garcia, C. Azevedo Ramos, E. Voll, A McDonald, Panama, Land Econ., 77(2), 187-205. P. Lefebvre, and P. Schlesinger (2006), Modelling conservation Nobre, C. A, G. O. Obreg6n, 1. A Marengo, R. Fu, and G. Poveda in the Amazon basin, Nature, 440, 520-523. (2009), Characteristics ofAmazonian climate: Main features, GeoStoms, D. M. (1994), Scale dependence of species richness maps, phys. Monogr. Ser., doi: 10.1 029/2008GM000720, this volume. Pro,f. Geogr., 46, 346-358. Nute, D. (2004), NED-2: An agent-based decision support system Turner, M. G. (1990), Spatial and temporal analysis of landscape for forest ecosystem management, Environ. Modell. Software, patterns, Landscape Eco!., 4(1), 21-30. 19(9), 831-841. Turner, M. G., R. V. O'Neill, R. H. Gardner, and B. T. Milne Pan, W. K. Y., and R. E. Bilsborrow (2005), The use of a multilevel (1989), Effects of changing spatial scale on the analysis oflandstatistical model to analyze factors influencing land use: A study scape pattern, Landscape Ecol., 3, 153-162. of the Ecuadorian Amazon, Global Planet. Change, 47(2-4), Verberg, P. H., K. Kok, R. G. Pontius Jr., and A Veldkamp (2006), 232-252. Modeling land-use and land-cover change, in Land-use and Pan, W. K. Y., S. 1. Walsh, R. E. Bilsborrow, B. G. Frizzelle, C. M. Land-covel' Change: Local Processes and Global Impacts, edErlien, and F. D. Baquero (2004), Farm-level models of spatial ited by E. F. Lambin and H. Geist, pp. 117-135, Springer, Berpatterns of land use and land cover dynamics in the Ecuadorian lin. Amazon, Agric. Ecosyst. Environ., 101, 117-134. Walker, R., and E. Reis (2007), A Basin-Scale Econometric Model Parker, D. S., S. M. Manson, M. Janssen, M. P. Hoffmmrn, and P.' for Projecting Future Amazonian Landscapes, Final Report, Deadman (2003), Multi-agent systems for the simulation of land NASA-LBA Project L-24, East Lansing, Mich. use and land cover change: A review, Ann. Assoc. Am. Geogr., Walker, R., R. DeFries, M. del C. Vera-Diaz, Y. Shimabukuro, and 93(2),314-337. A. Venturieri (2009), The expansion of intensive agriculture and Perz, S. G., and Skole, D. (2003). Social determinants of secondmy ranching in Amazonia, Geophys. Monogr. Ser., doi:IO.1029/ forest in the Brazilian Amazon, Soc. Sci. Res., 32(1), 25-60. 2008GM000735, this volume.
100
SCENARIOS OF FUTURE AMAZONIAN LANDSCAPES
Walker, R. T. (2003), Evaluating the performance of spatially explicit models, Photogramm. Eng. Remote Sens., 69(11), 12711278. Walker, R. T. (2004), Theorizing land use and land cover change: The case lI-o f tropical deforestation, Jnt. Reg. Sci. Rev., 27, 247270. Walsh, S. J., B. Entwisle, R. R. Rindfuss, and P. H. Page (2006), Spatial simulation modeling of land use/land cover change scenarios in N011heastern Thailand: A cellular automata approach, J. Land Sci., 1(1), 5-28. Walsh, S. J., Y. Shao, C. F. Mena, and A. L. McClealY (~008a), Integration of Hyperion satellite data and a household social survey to characterize the causes and consequences of reforestation patterns in the Northern Ecuadorian Amazon, Photogramm. Eng. Remote Sens., 74(6),725-735. Walsh, S. J., J. P. Messina, C. F. Mena, G. P. Malanson, and P. H. Page (2008b), Complexity theory, spatial simulation models, and land use dynamics in the northern Ecuadorian Amazon, GeoForum, 39(2), 867-878.
Wolfram, S. (1984), Cellular automata as models of complexity, Nature, 311, 419-424. Wood, C. H. (2002), Introduction, in Deforestation and Land Use in the Amazon, edited by C. H. Wood and R. Porro, pp. 1-38, Univ. of Florida Press, Gainesville, Fla.
Road Impacts in Brazilian An1azonia J. P. Messina and R. Walker, Department of Geography, Michigan State University, East Lansing, MI 48824, USA. S. Perz, Department of Sociology and Criminology and Law, University of Florida, Gainesville, FL 32611, USA. (sperz@soc. ufl.edu) E. Reis, Instituto de Pesquisa Econ6mica Aplicada, Rio de Janeiro RJ 20020-010, Brazil. S. Walsh, Depar1ment ofGeography, University ofN011h Carolina at Chapel Hill, Chapel Hill, NC 27599, USA.
J.
Alexander Pfaff, I Alisson Barbieri,2 Thomas Ludewigs,3 Frank Meny,4 Stephen Perz,5 and Eustaquio Reis 6 We examine the evidence on Amazonian road impacts with a strong emphasis on context. Impacts of a new road, on either deforestation or socioeconomic outcomes, depend upon the conditions into which roads are placed. Conditions that matter include the biophysical setting, such as slope, rainfall, and soil quality, plus externally determined socioeconomic factors like national policies, exchange rates, and the global prices of beef and soybeans. Influential conditions also include all prior infrastructural investments and clearing rates. Where development has already anived, with significant economic activity and clearing, roads may decrease forest less and raise output more than where development is aniving, while in pristine areas, short-run clearing may be lower than immense long-nm impacts. Such differences suggest careful consideration of where to invest further in transport.
1. INTRODUCTION The roads of Brazilian Amazonia are often portrayed in a melodramatic fashion. For decades, pictures have shown dirt paths, smoldering forest remains, poor people, and perhaps a message: these are unpaved roads; imagine what paving and capital investment could do! Such "visual cost-benefit
lSanford School of Public Policy, Duke University, Durham, North Carolina, USA. 2Cedeplar, Federal University of Minas Gerais, Belo Horizonte, Brazil. 3The World Bank, Brasilia, Brazil. 4Woods Hole Research Center, Falmouth, Massachusetts, USA. 5Department of Sociology and Criminology and Law, University of Florida, Gainesville, Florida, USA. 6Instituto de Pesquisa Economica Aplicada, Rio de Janeiro, Brazil. Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM00073 7
analysis," not surprisingly, conveys some tmths but not all. When deforestation occurs, it does indeed affect ecosystems. Roads increase access to forest and clearing follows, with ecological impacts: providing suitable habitat for some species but reducing and fragmenting other habitats, degrading streams and water quality, fostering the spread of exotic invasive species, causing wildlife mortality and species loss, and even bringing about local climate change [Trombulak and Frissell, 2000; Forman et al., 2003; Fearnside, 2007]. All are important potential impacts of forest loss discussed expertly in other patis of this book. This chapter is based on people-focused research from across Amazonia and over time. We present the view that roads differ in their forest impacts, and forest loss is not their only impact. While on average new road investments increase deforestation, it must be recognized that a road's forest impact depends on the context in which the investment in lowering transport cost occurs. Further, loss of forest is not the only consideration, as there are numerous impacts of roads on the Amazonian ecosystem and on human welfare. We refer to work at different scales. This chapter's authors, and celiainly the larger group whose research we cite, analyze scales from the household to the village, county, state, Brazilian Amazonia Legal, country of Brazil, and the 101
102
PFAFF ET AL.
ROAD IMPACTS IN BRAZILIAN AMAZONIA
made travel and settlement of the Central Highlands unattractive, until gold was discovered in the eighte~nthce.n tury. The few navigable rive~'s fl.owed w~st, ~reatll1.g qmte a long detour and correspondmg mcrease.m tlanspOlt costs. Moving to the northeast region of BrazIl, t~le BorboreI~a Mountain Range along with both the poor sOlis and the and climate combined to make economic settlement beyond t~le narrow coastal strip unsustainable. Finally, in Amazol1la, although river navigation was and is the main ~orm of . port, the impenetrable vegetation for a long tnne restncted human settlements to the riverine strips of land [Goulart, . 1959; Dean, 1995]. During the sixteenth and seventeenth centunes, tec~nology for the long-distance transport of goods was essentIally restricted to Indian and African slaves' shoulders. Horses and carts were inadequate for the steep slopes of the Sen:a do Mar for instance. Mules were first introduced as subStItutes fo~' slaves during the gold discoveries in the eighte~nth century. The railroads, introducing wheels for the first tnne in Brazilian histOly, together with cultivation of coffee, wel:e one of the key driving factors behind expansion of the agncultural frontier as well as ,the industrialization of the central-southern region of the country during the last ~umier of the nineteenth and first half of the twentieth centunes. Trucks and the development of the road network pushed the agricultural frontier in a northwesterly direction to,,:ard the Amazon basin during the second half of the twentl~th century. The main factors behind thes~ sl:i~ts were the. r~se of the domestic auto industry and the pnontles and subsldl~s to road transportation within government budgets and public tariffs [Banlt, 1978]. From 1960 t~ 1975, there was a massive expansion of roads in all regIOns of the countly. The Brazilian road network expanded from 440,000 to 1,418,000 lGn (Instituto Brasileiro de Geogra~a ~ Estatistica, va.rious issues). Following this trend, road bmldmg was ~ ~ornelstone of the regional development strategy for BraZilian Amaz~ nia. First came the Belem-Brasilia Highway, completed m 2. TRANSPORT COSTS AFFECT FOREST 1964 before Operayao Amazonia. This highway ~reated the first time overland connection between AmazOl1la a.nd the 2.1. Across Basins and Decades rest of the countly. An immediate consequence of thiS roa? We begin with a long view in space a~d time: i.e., looking building was the migration to northern Goia~, ~outheast P~ra, across all of Brazil, over several centunes. It IS not always and southern Maranhao. Extensive cattle rmsmg spread 111 a disorganized fashion, despite official efforts to regulate rural recognized that transport costs have always been a central factor in defining patterns of develo~ment: For ~xam- settlements [see, e.g., Mahar, 1989; Mueller,.1983~. ·' l the 1970s DU lll g , massive investments III aXial routes e Brazilian hinterland's adverse sOli, relIef, clImate, p1e, th b l' (Cuiaba-Porto Velho complet~d i.n 1970,. Tr~nsamaz011lca vegetation, and hydrology limited development y ma (l~lg with 2200 km in 1974, and Cmaba-Santarem m 1976) gave transport prohibitively costly [Silva, 1949; Summerlnll, greatly improved access to the hinterlands. Almeida [1992] 2003]. From Rio Grande do Sui to Bahia, the dense forests .of estimates that, in this period, road investn~ent. reache~ $4 the Serra do Mar, intense rains in summer, and slopes, wIth billion. This was complemented by col011lZatIOn pro~ects almost 1000 m of altitude change in 100 lGn from the coast, and agricultural research. From 1974 to 1986, National
multicountly Amazon basin. Although each scale m~y highlight different points, it is always important to consIder the impact of roads within a wider context. We staft with the long view in time and space. Remote sensing can be used at the temporal scale of decad.es and t~Ie spatial scales of Brazil and the entire Amazon basm (Brazilian Amazonia is a large component of both ofthese) to stud.y the change in broad forest coverage [see Alves et al., t~IS volume]. On the people side, census data of good qualIty and reasonable spatial and temporal coverage and fr~que~c'y exist at these large scales. At this level of aggregatIOn, It IS clear that transport costs influence land use. Th~s, because road investments impact the cost of transport, m gen~ral, lowering transport costs leads to increased deforestatIOn. However, the context in which road investme~ts are ma~e is important and even at this broad scal~ some dIfferences m new road impact arise. We see that pnor forest access an.d prior forest clearing have a powerful influence on the estlmated impacts of new roads on forests. New roads a~pe~r to have lower immediate impacts if prior development IS eI.ther velY high or vely low, while between .those o ~ndpomts, a fall in transport cost has its greatest nnmedlate Impact on deforestation. . This variation in broad patterns invites us down III scale to learn more about the contexts and the types of roads placed in them. The larger-scale analysis suggests tl:at as the roadgenerating processes differ, at regio~lal or vllla~e scale, so will the relationship of roads to cleanng. That sm.d, analyses at these smaller scales also confirm the general Importance of transport cost for land use. Finally, we, st.ep back to consider evidence of road impacts on people s I~co~nes, to .flag the importance of nonlocal influences, includmg.mteractIOns within Amazonian ecosystems, and to emphasIze that empirical analyses of roads such as those discussed here are relevant for future roads policies.
t:v
A
•
Integration Plan and Land Redistribution and North/Northeast Stimulus ProgranJ together invested approximately $13 billion [Diniz, 1995J,As the agricultural frontier reached the flatter Central Highlands, opportunities for mechanization opened. The maiJibeneficiaries of this were soybean farmers and cattle ranch~rs. In 1984, the highway linking Porto Velho to Brasilia was paved as part of the World Bank-funded Polonoroeste program. The Polonoroeste included three new settlement projects, with the largest one in Machadinho, in the northeastern corner of the state of Rondonia. The occupation of Rondonia proceeded through the 1970s into the 1980s despite the Brazilian Institute of Colonization and Agrarian Reform's (INCRA's) loss of control over the colonization process [Monte-Mol', 2004]. INCRA's roles were limited to the selection of settlers, land distribution, and design and construction of urban and rural areas in the new projects: roads, rural parcels, urban nuclei, and public buildings [Monte-Mal', 2004; Barbieri et al., 2009]. Eventually, these road investments slowed down, and the responsibilities were decentralized, with states and municipalities taking more active roles. Overall, these investments in components of the Amazonian transportation system, enabled by initial federal investments, lowered transportation costs between Amazonia and the rest of the countly [Simmons et al., 2007; Walker et al., 2009]. For example, Walker et al. [2009] document that by 1995, about a third of the Amazon basin (in the south and east) could be reached from Sao Paulo by ground transpOliation in less than 50 h, whereas in 1968, only a tiny strip of the southeastern margin had this degree of accessibility. When road building is financed by federal agencies, the decisions by states and municipalities are subordinate. However, once major axial roads are built by the federal government, there is political pressure for the expansion of smaller roads. Such pressures can be practically impossible to resist at the state and local levels.
103
remaining forest, with clearing occurring closer to the roads [see, e.g., Alves, 2002; Arima et af., 2005, 2008]. Thus, different kinds of networks produce different deforestation patterns. Landscapes often exhibit a patchwork mosaic of land cover types, with the urban and agricultural land uses tending to be closer to roads, while primary and secondary forests are farther away from roads. 2.3. ~Multiple 1/'onsmission Mechanisms
Some local stories or models suggest that the impacts of roads are straightforward and intuitive. New investments lower transport costs and thereby raise farm-gate prices received by producers for outputs. At the same time, input prices are reduced. Both ofthose changes increase the profit from agriculture. However, more than one type ofoutput can be produced by a given household, and some outputs are better produced on forest land than on cleared land (see Andersen et al. [2002] for discussion on the value of standing forest in Brazilian Amazonia). The drop in transport costs could also increase the profits from production of those forest outputs. Thus, for a new road to increase deforestation, the gain in profits for the outputs from the cleared land must be greater than the gain in profits from the forest products. Other types of orttputs are produced in cities, on land that is already deforested [see Andersen et af., 2002]. For example, in a city, significapt outputs of services could be produced in an area whose value in agricultural production might be considerably lower. Given this spatial variation in types of production, individuals choose not only what to consume but also to which types ofproduction they will dedicate their time. New roads could lead to urban migration, potentially counteracting other effects on agriculture in terms of total forest impact. Analogously but within agriculture, new roads could, in principle, yield spatial concentration with more intensive production arising on less land. These mechanisms are all important given the regional importance of 2.2. Transport Costs 1l1atter Locally Too urbanization. Yet the impacts of roads on urbanization are complex Broad patterns or associations across space may not al- and have many linkages to deforestation. Analyzing them ways hold as we look more closely. Yet even at the "micro" requires integrating across scales, spatially and temporally. or household level, transport costs are critical in many dif- Barbieri et al. [2009] cast Amazonian rural-urban migration ferent theories about land use. Thus, we might expect trans- as actors maximizing opportunities given many constraints pOli to be locally influential, too. Several case studies are and opportunities, which are determined in part by infraconsidered below, and they all confirm this view. structure development: "in both Brazilian and Ecuadorian In fact, Brazilian Amazonia is the source of famous im- . cases, urbanization (... ) may be a typical response not only ages linking local transportation investments to local defor- to socioeconomic, demographic and land use changes in the estation patterns, e.g., the "fishbone" pattern of clearing in frontier, but also to structural changes in the national and colonization projects. Generally, the form of "spatial archi- global economy." Many possibilities for road impacts can tecture" of the roads will influence the "geometry" of the arise within such dynamics.
104
ROAD IMPACTS IN BRAZILIAN AMAZONIA
quenced road investments and clearing ofthe Mayan forests, manuscript in preparation, 2007) show, for instance, that the Both the large- and small-scale summaries above suggest earlier investments in new roads significantly influence dethat whet'e there are new roads, there will also be more clear- forestation in the future, controlling for later investments. It is also worth noting that the total impact of a road will ing. That is indeed what empirical study at basin level finds, extend beyond its census tract or even its municipio or at least on average given the variation in the impacts, over Pfaff[1999], for instance, suggests that higher road county. space and time, which is discussed more below. For discusdensity in a county affects rates of deforestation in adjacent sion of differences across countries in the settings in which counties. Census-tract data permit examination of such spadeforestation occurs, see, for instance, Geist and Lambin tial spillovers within a county. In principle, a new road could [2001]. Consistent with macroscale and microscale anecdotes spatially concentrate economic activity within a county, above, the empirical evidence at various scales across Ama- perhaps even lowering deforestation in some tracts. Pfaff zonia suggests that greater access due to roads raises rates of et al. [2007] extend the previous analysis from Pfaff et a1. deforestation. Systematic simple organization of basin-scale, (submitted manuscript, 2006) of 1976-1987 deforestation municipio-level data on forest, roads, and other factors driv- following lagged road investments. The robust result is that ing land use, from Reis and Guzman [1992] andf'faff[1999] deforestation increases in census tracts within 100 km of any census tract receiving new roads. This supports prior results on, suggests that more roads yield more clearing. Chomitz and Thomas [2003] considerably increase the and indicates a higher total impact. number of observations by using census tract data instead of 3. ROAD LOCATION AFFECTS ROAD IMPACT counties. There are 10 to 20 times more observations in the census tract data than in the county data set. They find results consistent with earlier works, though they note that their and 3.1. Pristine, Highly Developed, and In-Benveen others' average estimated road impacts seem relatively low While average results are informative, few new road in(see also G. D. de Luca, World Banle Development Research vestments are average in all dimensions. Unpaved roads difGroup, Development and deforestation: A review, manufer from paved: paving from scratch differs from paving over script in preparation, 2007) on this, and for a comparison a previously unpaved road. As important as anything, the with perhaps typical assumptions, see, for instance, Laurfirst road offering access to an area differs from additional ance et al. [2001]). roads. Chomitz [2006] spans many locations and policy isFollowing Chomitz and Thomas [2003], A. R. Pfaff et al. of setting. sues in emphasizing the impOltance (Roads and deforestation in the Brazilian Amazon, submitted Andersen et al. [2002] make this explicit by considering to the B. E. Journal ofEconomic Analysis and Policy, 2006, the effects of prior development, measured as prior deforAvailable at: www.duke.eduJ~asp9; hereinafterrefened to as estation. With around 250 observations (conglomerates of Pfaff et aI., submitted manuscript, 2006) revisit this issue uscounties), they estimate an interaction between roads and ing census tracts, and observations of road and forest changes over time. The change over time, and census-tract resolu- prior clearing, assuming that higher prior deforestation will tion, permit county-specific statistical controls to improve always raise or always lower road impact. They find it lowestimates. Focusing on 1976-1987 deforestation rates, using ers the impact, i.e., that when prior deforestation is higher, lagged 1968-1975 new road investments, they confirm that the forest impacts of a new road will always be lower. Furnew roads increase deforestation on average. Fmther, while ther, they extrapolate from this estimated constant trend [~ee allowing that new road impacts can vary with prior develop- also' Weinhold and Reis, 2008], concluding that for high ment (measured as prior deforestation), by breaking this large enough prior deforestation, reducing transpOlt cost will resample into prior-development categories, they find that all duce clearing. Pfaff et al. (submitted manuscript, 2006) use 6000 data significant impacts are increases in clearing. points to find support for this idea, but not for that result. It is wOlth noting that the total impact of a road will extend More observations, for census tracts, permit a split of the beyond the first decade after the investment has been made. data into prior-deforestation categories. As do Andersen et Actors will respond to a road investment over time, with the al. [2002], they find significant increases in clearing when rate of response, in pmt, a function of the availability of other roads are placed into locations where less than half of the production inputs. Some actors may respond with additional original forest has been cleared (which represent the great investments, such as in health or educational infrastructure majority of census tracts). Unlike Andersen et aI., they find that complements the road in attracting migrants. Pfaffet al. road impacts are higher for the next highest level of prior [2006] and D. A. Conde and A. Pfaff (Duke University, Se-
2.4. Average Forest Impacts at Basin Scale
PFAFF ET AL.
clearing, (50-75% of odginal forest has been cleared). From 75% up, impact is insjgnificant. Thus, road impact is smaller with highest or low9st prior development and is largest in between. This supppi·ts the contention that prior development matters. Yet it r~futes a simple trend and is never negative. As just noted, ixtrapolating that trend can yield negative estimated impacts otherwise not found. The findings of Conde and Pfaff (manuscript in preparation, 2007) for Mayan forest in Mexico, Guatemala, and Belize support these results. Deforestation rates never fall after investments in new roads and refuting the simple linear trend, clearing increases ar~ lower in the pristine areas than in the areas of middling prior development. Pfaff et al. [2006] then used road data over time to measure prior development using prior roads. To examine empirically whether this could matter, Pfaffet al. [2006] split new paved road investments into those following prior unpaved roads in a census tract and those in census tracts where new investment was not paving previously unpaved roads. They find that new paved road investments in places with prior unpaved do not have significant impacts on clearing, controlling for the impacts of prior paved roads too. With the same controls, new paved road investments without prior unpaved raise deforestation. Thus, while again the new roads never lower clearing, these results, using prior roads to measure prior development, support the idea that a road can have lower impacts given high development. C. I. Delgado et al. (Duke University, New roads are not all made equal: REDD-relevant evidence on the influence of prior development from a deforestation frontier in the tri-border region of Brazil, Peru and Bolivia, manuscript in preparation, 2008) find the same result for the Interoceanic Highway in westem Brazilian Amazonia and Peru, which is also near to and influences deforestation in Bolivia. When first constructed, as an unpaved route, the highway clearly raises deforestation rates in all three countries. However, its paving in Brazil after 2000 does not appear to influence local clearing. Using three categories of prior development, from Pfaff et al. (submitted manuscript, 2006), and using prior roads to measure prior development, from f'fc{/f et al. [2006], f'faff and Roba/ino [2009] apply pixel data for the entirety of Brazilian Amazonia. This provides another significant jump in obselvations, allowing the distance to the closest prior road to be measured as a prior-development control in studying a new road's impacts and allowing matching methods to be applied for better causal inference. Their results support the idea by Pfaff et al. (submitted manuscript, 2006): highest initial road impact in middle ranges of prior road distance; lower initial impacts when the closest prior roads are either velY far, in isolated frontiers without conditions that support
105
production, or vety close, where the existing development given prior access can generate clearing dynamics with impact distinct from road effects. This pattem of impacts vatying across space, as a ftmction of levels of prior development, suggests that the spatial pattem of emerging networks will affect development's impacts on forest. For instance, if the shortest path for a development goal is through middling prior development, then considering longer altemative paths along existing transport routes can involve a trade-off between higher road costs and lower deforestation. Whether that trade-off is attractive may be affected by carbon payments. Based on the first-decade impacts cited here, one might posit the same trade-off for longer paths through pristine areas. However, while the first-decade increases in clearing for those areas may be lower than increases in regions of middling prior development, recall that total impact of a new road includes the long-run shifts brought on if additional investments follow the new road (and Pfaffet al. [2006] show that in Brazilian Amazonia, many new roads follow on old roads). Thus, the longer-run impact of opening up access to a pristine area could be considerably larger. In addition, if species habitat (as opposed to carbon storage) were a leading forest concem, then the fragmenting impacts of cutting through pristine habitat could dominate even in the shOlt run (consider, for insthnce, the estimated impacts of new roads on jaguar habitat by Conde Ovando [2008]). 3.2. Where Are Rqads, Why, and So What?
Roads are not located by blindfolded planners randomly throwing darts at maps. Rather, such significant infi'astructural investments are usually driven by specific motivations. Those may be identifiable from historical documents. However, often such documents may not exist, or may not represent the key details of meetings, or even intentionally may not reveal all. Here we must substitute statistical associations and our own interpretations for such documents. The reason for doing so is that motivation matters. Specifically, a lack of understanding of why a road was located can confound the estimation of the causal forest impact of new roads. We assume that road planners had more infonnation at their disposal than we have as analysts. Thus, we are likely to not be able to account statistically for some factors that mattered. For instance, if planners aim a road at the place most likely to boom in agricultural production, and we do not obselve all of the factors that made it so, we might assign such factors' impacts to roads. Simplifying and formalizing this, D. Weinhold et al. (London School of Economics, Land use and transportation costs in the Brazilian Amazon, manuscript in preparation, 2006)
106 ROAD IMPACTS IN BRAZILIAN AMAZONIA
extend the work of Andersen et al. [2002] by examining statistically whether some roads appear to follow upon prior deforestation. They find this is the case. Thus, a significant portion ofllbad investments may be following on some prior development, which may signal to road planners that various factors will facilitate more output. Along these lines, see that a significant fraction of all the roads in Brazilian Amazonia are near a city. In several cases, this is linked to INCRA's sponsored colonization pattem known as urbanismo lUral, which established the city as the "loci" of social and economic organization in several parts of Amazonia [Barbieri et al., 2009]. A stated rationale was "bringing the town to the country" through linking lUral parcels to urban areas via a network of local roads. This was a central component of a strategy to effectively occupy Amazonia [Barbieri et al., 2009]. Browder and Godfrey [1997], considering settlement in Rondonia, suggest that this reproduced in the forest an infrastlUcture and abandonment found in slums of more-developed areas in Brazil. For our purposes, given the results above that the impacts of roads on forest vary with context, where in fact roads are placed clearly affects their actual and estimated impacts. Thus, if past new roads were often placed in locations where their impacts actually were relatively low, analysis at the regional level might find, erroneously, that impacts are always low. Yet commentators could still be correct to claim that a particular new road could lead to a significant rise in deforestation. 4. ROAD TYPES AND PROCESSES 4.1. Official Versus Unofficial
Roads are varied in Amazonia. One might distinguish by constlUction (e.g., paved versus unpaved) or other engineering specifications, but another useful distinction is between the "official" and the other "unofficial" roads [Brandao and Souza, 2006; Perz et al., 2005, 2007a, 2007b]. Official roads, or "primalY" or "development" roads, are interregional highways built or financed by national or state govemments. They appear on official maps and run for hundreds of kilometers connecting cities in different palis of the country or even in different countries. Examples in Brazilian Amazonia include well-known highways such as the Transamazon, the CuiabaSantarem (BR-163), and the BR-364 through Rondonia. The main reasons for building official highways include regional integration, to facilitate economic development by facilitating access to global markets, as well as geopolitical objectives, such as securing national borders. Official roads have received considerable attention as large-scale infrastlUcture projects that were often initiated with velY little public debate. Past official road projects in
PFAFF ET AL.
Amazonia in the 1960s and 1970s are linked to considerable deforestation and lUral violence in conflicts over land [Goodland and Irwin, 1975; Schmink and Wood, 1992]. Thus, it is not surprising that new official road projects in the region have been substantially criticized conceming their prospective impact [see, e.g., Laurance et al., 2001], including specific debate about road paving [Nepstad et al., 2002]. In terms of their total impacts, it has been noted that official roads form sparse networks [Arima et al., 2005]. They are few in number and lUn in parallel hundreds of kilometers apart with few intersections (though the intersections and the effectiveness of the network would increase under current new road plans). At least with the current network, that could leave large blocks of forest intact. However, as was stated above, road investments and access tend to lead to follow-on investments including other new roads. Thus, indirect impacts of official roads accessing pristine areas may be considerable. In particular, official roads may stimulate constlUction of unofficial roads. "Secondmy" or "settlement" or "logging" roads, depending on who constlUcts them, are created by nonstate actors such as loggers and colonists. They are local in extent and often do not appear on official maps. The main purpose for building them is to access natural resources. This can be simply for local livelihoods, though complexities and conflicts may arise, and we discuss govemance below. Unofficial roads are thus specifically for resource exploitation and underlie economies of natural-resource-based communities in frontier areas. Indeed, another term for unofficial roads is "endogenous roads" as their location directly follows from local opportunities for greater output. They may even be directly funded by profitable local extraction and end when profitability does. Many official road projects involve upgrading rather than new constlUction (e.g., the well-known Avan<;a Brasil project includes significant paving over of unpaved roads). Unofficial roads, in contrast, are now being rapidly constructed. Interpretation of satellite images allows maps of such roads [Brandao and Souza, 2006]. Analysis of centralwestern Para, a frontier area, reveals rapid expansion. From 1990 to 2001, unofficial roads grew from 5042 to 20,769 km. During that time, the extent of official roads was fixed. By 2001, unofficial roads comprised over 80% of the total road network in the area. Such mapping elsewhere across Brazilian Amazonia also reveals extensive unofficial road networks [Lentini et al., 2005, pp. 78-79]. 4.2. Logging and Unofficial Roads
Perhaps the key actor in unofficial road building in Amazonia is the logging film [Grogan et al., 2002; Arima et al.,
107
2005,2008]. Loggers build roads to reach high-value timber, ever, such official road recognition has only further motiraising profits by minin}izing distances, while avoiding steep vated road extensions and additional informal colonization hillsides and rivers, )yhich require expensive bridges. Thus, [Perz et al., 2005, 2007a]. loggers build road~} through landscapes along topographic This strategy's success has relied on colonists' ability contours [see Arinla et al., 2005]. to form political pressure groups. State decentralization in / Loggers build roads ofvmying quality. Roads to large areas Brazil during the 1990s made municipal governments more for multiyear exploitation are often high quality for reliable important for many functions including road maintenance. passage, whereas short-term-access roads are of low quality. Consequently, promises of road maintenance arose in political There are temporal cost trade-offs here: more expense in road campaigns in many parts of Amazonia, with votes-f~r-roads building can lower maintenance. This has important ramifi- deals being cut. While the political ecology literature emcations. Dense logging road networks tend to be tempormy, phasizes inequalities, contestation, and conflict over natuabandoned when profitable timber is exhausted, while roads ral resources [e.g., Schmink and Wood, 1992; Hall, 1989], to larger timber stands remain for many years. some cases of unofficial roads illustrate the potential for A related investment by those loggers with scale in mind cooperation and negotiation in building roads for access to is in larger tlucks [Stone, 1998]. So-called "Romeo and Ju- resources. liets," i.e., double-trailer large trucks, are often early adoptions by loggers regardless of logging quality. Govemment 4.4. Colonist-Logger-Indigenous Interactions investments affect such decisions [Bauch et al., 2007]. Paving investments that lower transports costs pennit logging Cooperative political ecology is also evident in relations to expand further into the forest. Government's role may between colonists and loggers with respect to road building. even affect who makes decisions. Lima and Merl)) [2003] A challenge for logging road building in Brazil is that land and Bauch et al. [2007] note a trend to subcontract transport, claims must be for permanent productive land use, while possibly to avoid bureaucracy for movement of logs. loggers often extract timber and move on. The presence of Exhaustion of the timber near markets impels road exten- colonists offers a solution. Colonists with established land sions. Profits fi'om initial timber extraction help to fund these claims but poor roaps can make deals with loggers to sell extensions. Thus, to the extent they yield profitable timber timber on their lots' to loggers in return for road building extl'action, unofficial roads will be built [Perz et a1., 2007a, and/or maintenance . Migrants seeking to acquire land make 2007b]. The result is an extensive logging road network in such deals for unofficial roads that reach unclaimed land. many areas, up to 200 km from urban centers. In terms of im- Colonists gain road access or maintenance, while loggers pact, even if logging is selective, the areas with such networks receive legal cover. exhibit extensive forest degradation [Nepstad et a1., 1999]. Yet unofficial roads can also generate social tensions and reveal conflicts, such as over the route for a new road. For 4.3. Colonization and Unofficial Roads instance, while loggers seek to minimize costs and avoid rivers, colonists seek roads alongside the front of their lots, usuFrontier colonization areas also exhibit unofficial road ally in straight lines. Loggers seeking to build roads through building, though following a logic somewhat different from a lot held by a colonist in order to reach timber in a more that of the loggers. Along such official roads as the Transa- remote area may thus face difficult negotiations [Perz et a1., mazon and BR-364, the state built feeder roads, which per- 2007a]. The extension of unofficial roads also sometimes pendicularly intersected the highway evelY 5 km, forming breaches indigenous reserves and state forests, creating conthe "fishbone" architecture. The feeders ran up to 10 km in flicts over land tenure. Demarcation of the Cachoeira Seca both directions from the primmy roads but then in-migration do Iriri indigenous territOly to the south ofthe Transamazon led to demand for land beyond this. Thus, colonists began to Highway is controversial because unofficial roads built by unofficially extend the feeder roads [Walker, 2003]. Their loggers into the indigenous territory have allowed colonists focus was land tenure and rights to provide a basis for fam- to informally settle there [see Perz et al., 2005, 2007a]. ily livelihoods. To facilitate state recognition of the land claims, colonists 5. SPECIFIC ROAD CASES chose routes following the state design for the official road network. This worked, and it also motivated additional road 5.1. Transport and Land Use in Acre extension [Walker, 2003]. Recently, the state has sought to recognize new colonization areas as a way of prohibiting 5.1.1. Projeto de Colonizar;iio Humaitri. This project was fmiher road building that would enter protected areas. How- established in 1981 in Porto Acre in the state of Acre on
108
ROAD IMPACTS IN BRAZILIAN AMAZONIA
60,334 ha, which straddle the Rio Acre. The state capital of Rio Branco is accessible along 35 km of paved road (AC-O 10). Unlike most government-sponsored colonization projetts in Amazonia, which were on public lands, Humaita was a re-distributive land reform project on private land. Owners of the rubber farms were willing to sell off some land, in part due to the decline of the rubber economy. Its design is a radial road network around a central village, and originally, it was divided in 951 lots (ZEE 2000:38), all implemented by the INCRA. The radial design provides a good connection between the secondary roads and the main paved road to Rio Branco, but like "fishbones," the design did not take account of the watershed drainage system. The land use strategies of agriculture and mixed husbandly depend on all-weather roads in the Humaita region. Most agriculhlral products are taken to market during the rainy season, when the condition of unpaved roads is often poor. Mixed husbandlY is composed mainly of daily products and fish. Several farmers reported changing land use strategy after losing their crops or milk production because of poor access to urban centers and markets. Both types of outputs are also dependent on access to electricity, which is likely to be correlated with road network conditions. Lack of market access is probably the biggest factor keeping families in "subsistence fanning," in keeping with von Thiinen's link between income per area and urban proximity [Dunn, 1970; Walker et al., 2009]. Paving ofBR-3l7 will eventually link Acre to the Pacific Ocean as part of Avam;a Brasil, to facilitate shipping of central Brazil's agricultural production to international markets (Asia in particular) and to enhance Amazonian economic development [Nepstad et al., 1999]. This connection is expected to boost the economy and to change land cover along BR-3l7 [Brown et al., 2002]. 5.1.2. Road impacts. TranspOliation co~ts are highly significant in explaining the land use choices of farmers in Humaita settlement. At greater distances and transport costs, one finds more subsistence and extractivism; with lesser transpOli costs, more land is in agriculture and in mixed husbandlY. A multinomial logistic regression for 63 farmers shows a significant effect on land use strategies. Roads also matter for lot turnover and land consolidation. During the early stages ofthe frontier's development, settlers who lack infrastructure are isolated from markets and services. They may abandon their lots or may sell them at low prices to investors. During more advanced stages of frontier development, with infrastructure and land markets, fanners face an incentive to sell, as rising land demand raises the
PFAFF ET AL.
price of land. An examination of transpOli costs (in time) and total land area (a proxy for lot consolidation) finds a highly significant positive link. Thus, more land consolidation appears to have occurred further from urban centers [L udewigs, 2006]. As the settlement's life continued, difficulties in access to urban centers in remote areas of Humaita led to a number of farmers being willing to sell their lots at lower prices than would be obtained in more accessible areas. Note that returns to investments in land purchases in these areas have been favorable with cattle ranching or land speculation, providing a shelter against inflation. Land investors favored contiguous lots that could be consolidated into larger properties, which facilitate cattle ranching. Yet lots served by better quality roads and closer to urban centers had sufficiently higher profits in agriculture that colonist farmers were willing to move to Humaita. 5.1.3. Road demand and supply. What socioeconomic and environmental conditions reinforce the importance of roads? The agricultural calendar, based on the seasonal rainfall cycle through the year, leads the harvest time of most agriculhlral crops to fall during the rainy season. This is also the time of the year when road quality is worst, given the negative impact of rainfall on the maintenance of road conditions. Thus, road maintenance is important [Nelson, 1973]. It has been shown that the frequency of heavy traffic plus factors such as soil type, relief, and the geography of the drainage system all influence the road conditions that maintenance needs to address. However, the allocation of road maintenance expenditure is linked not only to the condition of roads but most importantly to the political influence of farmers and rural producer associations. Land investors connected to the government can attract maintenance. Well-organized producer associations can, too. This is a clear case in which the potential for output can affect road investments, which can then confound impact estimates; at the least, evidence from the places that have more roads may not apply to where there are few. Part{cipation of both public and private stakeholders in various land programs may help improve road construction and maintenance. Promising examples of public-private partnerships in roads have been reported for the Santarem region (Nepstad et al. [2004] and Lima et al. [2006] note that a single logging company can create and maintain over 700 km of agriculhlral access roads in INCRA settlements). These may more effectively attend to farmers' demands for better road infrastructure and land titling. They also have increased the legal commercialization of timber. An increased efficiency of resource use raises the chance that settlers will remain in their lots.
5.2. Roads and Transallllizon Smallholders Meny et at. [200q]Js how that for smallholder settlements on the Transamazgl1 Highway, the greater the distance to a city, the lower th~lsettlers' land value (following the observation of Walker ei al. [2002] that high-value systems tend to be found closer to the Transamazon Highway than low-value, or subsistence, systems). For each additional kilometer from the city, pel' hectare values decline by R$2.l9 (or approximately $1.00). This result is roughly consistent over three settlement regimes studied-purchased, formally settled by INCRA, and informally settled. These results support the contention that distance plays an important role inland values. In the difficult conditions along the Transamazon Highway, one might well expect that the quality of the road is paramount for the outcomes generated. Poor roads could soon disappear or become impassable in the wet season. The portion ofthe road that is "improved," i.e., covered with gravel, varies considerably in this area, and thus, such a hypothesis can be tested. Meny et al. [2006] find in fact that the quality of the road, in terms of just dirt versus with gravel, does not significantly affect land values. Such "all-weather dirt highways," with gravel, currently comprise only 22% of total average distance, perhaps too little to affect land values in these settlements. Men)! et al. [2006] also consider the relationship of these settlement regimes to the roads. For instance, on average, individuals attain lots 3 years after the road was built. However, Men)! et al. break this down by regime to reveal significant differences. Formal settlers arrived almost simultaneously with the roads. Informal settlers arrived on average 2.3 years later, while those who purchased their lots did so on average 4.8 years after the establishment ofthe road. These clear differences convey the typical trend in this area of initial roads with the formal settlement, then informal settlement alongside informal road extension, be that by the settlers or by loggers, and then a formalization of the land market as more formal buyers of the land entered the area. 5.3. Roads, Population Mobility, and Deforestation in Northern Ecuadorian Amazonia 5.3.1. Oil exploitation and initial settlement. Northern Ecuadorian Amazonia (NEA) is a sparsely populated tropical lowland rainforest and an area of extraordinmy biodiversity [k~}lers et al., 2000; Bilsborrow et al., 2004]. Altihlde varies from the Andean foothills to about 200 m above sea level at the eastern border with Peru. Soil is more fertile than in most lower Amazonian areas of Peru or Brazil, with pockets of volcanic (black) soils, though quality is highly variable, and there is much poor red soil. Unlike Brazilian Amazonia,
109
NEA has a year-round growing season with rain occurring in all 12 months. This leads to slash-and-mulch clearing (leaving cut trees to decompose) with little burning. Also in this area, unlike Brazilian Amazonia, large-scale commercial agriculture, ranching, and logging have never played major roles, with most forest cleaning occurring at the hands of small farmers [see also Brondizio et al., this volume]. Initial colonization, starting in the 1960s and tlu'ough the 1980s, is closely linked with the discovely and exploitation of oil in Sucumbios province, near what would become Lago Agrio (Nueva Loja). Oil has become the most important source of export earnings in the countly. Smce the 1960s, oil companies have constll1cted roads into the forested lowlands in order to lay pipelines and to connect wells to the network, which pipes oil over the Andes to Esmeraldas for export. A major consequence of oil exploitation was the establishment of a network of roads by the oil companies. This greatly facilitated settlement. Previous research in Ecuador [Rudel, 1983; SOllthgate et al., 1991] and elsewhere (e.g., Almeida [1992] on Brazil, Heckandon [1983] on Panama, Kaimmvitz [1997] on Bolivia) has found road access to be a dominant factor in land clearing. The road infi"astructure in NEA was improved as a result of some expansion and some paving of primmy roads, as well as the construction or expansion of secon~my roads. In thinking about the impacts of roads on forests, we note that Bilsborrow et al. [2004] and Barbieri et al. [2006] show that in contrast to i Brazil, Indonesia, and other countl'ies, NEA settlement was spontaneous, not government-sponsored. Migrants settled along oil roads, with successive arrivals claiming land plots behind the farms along the roads. Most came from the Siena region ofrural poveliy, extremely concentrated landholdings, and "minifundia," which acted as a strong population push (versus any pull from abundant, cheap land in the NEA). The colonists were poor and arrived without capital to invest in their plots. Thus, oil roads opened vast forest areas to settlers, facilitating deforestation in NEA. Forest cover in the study area fell, from essentially 100% in 1970 to 59% in 1990 and 45% in 1999 [Bilsborrow et al., 2004]. Some deforestation was due to the creation of farm subdivisions during the 1990s along roads. But road expansion also occurred. In 1999, 51 % of households reported some road construction or improvement near their houses since 1990. 5.3.2. Second-generation colonists and urbanization. The 1990s brought a second impOliant wave of deforestation linked to an out-migration, mostly of sons and daughters, from pioneer migrant settler households to other rural areas of Ecuadorian Amazonia [Barbieri, 2006; Barbieri et al., 2006]. A 1999 field survey found many new land
110
ROAD IMPACTS IN BRAZILIAN AMAZONIA
subdivisions associated with pioneer colonists' children demanding land upon reaching adult age. They occupied and cleared intact forest in the plot or moved away to another forest arealtin Amazonia or to the incipient urban areas in NEA. Overall, population mobility and redistribution are now dominant demographic factors in regions such as NEA, as fertility and mOliality have fallen considerably, as has natural 'population growth [Barbieri, 2006]. Given this demographic dynamic, the relatively extensive road network was a key factor in deforestation, facilitating migration to other rural areas though also to urban areas [Barbieri, 2006]. For instance, the 1999 survey found that subdivisions arose along roads and near towns, given the impOliance of labor mobility. Note that even the urban migration could be pali of an ongoing process of agricultural production causing deforestation, if families diversify income to address multiple sources of risk and thus keep fanning [see, e.g., Barbieri, 2006; Barbieri et al., 2006; Barbieri and Carr, 2005; Barbieri et al., 2005]. In keeping with the subdivisions developing close to the roads, Barbieri [2006] shows that longer walking distance from the household to the nearest road decreases the odds of rural-urban migration and of local off-farm employment, for both the old and the newer cohorts of colonists. These local transport costs appear to significantly hinder interactions between various locations. These deforestation pressures continue to the present. High fertility, new immigrants, the expectation of further expansion of the oil industly (with recent discoveries of large new deposits and the beginning of construction of a second trans-Andean oil pipeline) and consequently the expansion of road networks all point to increasing pressures on forest in NEA. Mena et al. [2006] find that the Cuyabeno Wildlife Reserve, one of the most important protected areas in Ecuador, and within NEA, is threatened following the permission to exploit oil given to the Brazilian oil company Petrobras. Existing and planned roads to the Cuyabepo Reserve have recently become a major facilitator of settler colonization in the areas within or nearby the reserve. 6. BROADER SOCIOECONOMIC CONTEXTS 6.1. Local Road and Development Benefits
To this point, our almost exclusive focus has been the impacts of new roads upon deforestation. Yet one must suspect that there is a reason why people put in all the effort it takes to build roads other than the widespread clearing of forest, which per se is rarely, if ever, mentioned as the goal (though land titling policies rewarding land "improvement" came close to such a direct incentive). Speaking plainly, while cer-
PFAFF ET AL.
tainly the motivations for roads have differed across space and time, it is likely that some actors associated a new road's construction with some form of local and regional benefits. For instance, the notion that a road's socioeconomic impacts can be positive holds at the level of the household and of the village. Both official and unofficial roads can provide access to natural resources, facilitate market access for rural producers plus the integration of economic sectors, and reduce costs of spatial mobility for people, capital, and also inf0l1l1ation [Owen, 1987; Vance, 1986]. Thus, roads clearly can be central to economic development and to social wellbeing. Unofficial roads are instrumental in local economies. Logging firms provide work for many from frontier communities. Large sawmills employ up to 300 employees per firm. The sector also indirectly generates employment in tractor repairs, trucking of sawn wood, and sales of wood products. Beyond jobs, roads built by loggers improve the access of rural populations to local markets, where they can sell produce, and to urban services such as education and health care. Rural communities thus view logging roads as crucial for improving quality of life [Perz et al., 2005]. That said, the socioeconomic impacts are not all rosy. Changes in access may well yield social conflicts over land and other natural resources, as well as debate concerning their impacts on preexisting livelihood strategies, including, for instance, those based on threatened resources [Brown et al., 2002; Reid and Bowles, 1997; Schmink and Wood, 1992]. These outcomes are certainly not caused exclusively by roads per se. However, at least we can say that the development outcomes in new frontiers have both highs and lows, not only over time but also in terms of distribution across groups. Returning to a broad view across space for such empirical evidence, consider Andersen et al. 's [2002] analysis of census data for Amazonia at county level. In short, they find that over decades, the gains from agricultural production and ranching in Amazonia have risen. Their results suggest both learning-by-doing by individual fanners and that an impOltant .factor was the adaptation of cultivars and the tech~ nologies provided by the agricultural research of Empresa Brasileira de Pesquisa Agropecuaria (EMBRAPA) (Cattaneo [2001, 2005] also suppOlis the importance of such innovations in agriculture). Andersen et al. [2002] estimate the value of cleared land in Amazonia using three methods: observed land prices, site studies of agriculture in both Para [Almeida and Uhl, 1995] and Acre [Vosti et al., 2001], and simulations from their own county-level deforestation regression modeling that relies upon the census data on past land usage. One useful perspective on all such estimates, they add, is that in a region with abundant land and valying scarcity of labor and capital,
might expect the per-hectare returns to vary widely and might also expect that overtime labor and capital would become less scarce, whjfe land would become scarcer. Vosti et al. [2001'1 found that typical traditional-technology falms in ACIJ yield a discounted net present value of ha- 1 (usinga discount rate of9%). That could be doubled with more intensive technologies, but many cannot afford the required outlays for labor and capital. Almeida and Uhl [1995] find that in Para, the net present value of land with sustainably grown perelmial crops is about $5000 ha- I (though they are using a discount rate of6%). The simulations based on county census data by Andersen et al. [2002], of course, also ValY with the discount rate chosen. Adding the rural and urban changes in gross domestic product (GDP) associated with the new roads constructed during the period 1970-1995, and using a 2% rate of discount, they find a per-hectare net present value of cleared land to be $5676 (similar to Almeida and Uhl, though they note that there are questions about how best to add the rural and urban outcomes). They also note that urban gains are relatively constant, while lural benefits increased over time. Tying into the issue of context, Andersen et al. [2002] note that the gains in GDP from paved roads are higher when there is prior economic activity. Recalling from above that the clearing impacts of new roads in very highly cleared areas may be lower (perhaps especially when adding up the impacts that are likely to unfold over time with all responses to the roads, including new roads), these results for GDP impact may well suggest that the ratio of the GDP gain to the loss of forest could be maximized with a form of spatial zoning in which new roads intensify the development that has already been occurring instead of opening up relatively pristine areas for deforestation. 6.2. Nonlocal Influences on Transport Investments and Their Impacts
Almost all of the above concerned actions and consequences in Amazonia itself. Even more spatially specific, it concerned the local conditions in which a road investment occurs, and the resulting local impacts upon forest as well as employment and output. The exception, which was nonlocal, although still within the basin, was spatial deforestation spillovers to neighboring areas. Yet spatial interaction can go in both directions. Much as a road in one locality can affect forest in other areas, socioeconomic shifts in other, nonlocal areas can affect the forest in places where local road investments occurred. They can affect the total amount of forest, taking prior road investments as fixed, and can affect the impacts of the road investments upon the forest.
III
Other countries provide examples of such interactions across space and time [Pfc{ff, 2000]. In the New England region of the northeast of the United States of America, as population spread and agricultrlral production increased, forests steadily vanished until the mid~·1800s. Then, railroads linked the region to the Midwest, where the flatter and high quality agricultural land put steady downward pressure on New England agriculture returns. Local agricultural prices then fell. This external influence dominated forest outcomes in the region. Despite increasing population and output, New England forest increased. That might not be surprising had no roads been built in New England. However, that was the case despite roads being built. While all else equal new roads can increase agricultural profits and push a few more parcels over the threshold to deforestation, a major downward shock to those profits implies that one could build many roads, and previously cleared land still would reforest. Thus, the direction of trade is critical. Pfaffand Walker [2009] apply this thinking to Amazonia. Turning back to Amazonia, then, analogous to Midwest soil quality are externally determined prices in Amazonia: the Brazilian real's exchange rate and the prices of beef and soybeans. Also, as noted above, an analog to the introduction of rail in the United, States of America that increased forest in the nOltheast is the introduction of agricultural innovation in Brazil that facilitated production and thus decreased Amazonian forest. Suchifactors can be dominant. Cattaneo [2001] considers such innovation (which is the focus of his 2005 work cited above) as well as the exchange rate. His computable general equilibrium modeling suggested that a 40% devaluation of the "real" in real terms, at that time, in the long run would have led to a decrease of 12% in deforestation. More generally, changes in relative prices of impOlis and exports matter. Whatever the level of the exchange rate, another important price is that of soy. Though Brazil is a large producer, for any level of Brazilian soy supply, many global factors such as the consumption of soy in China will significantly influence that price (for relevant discussion see, for instance, Nepstad et al. [2006] whose Figure 1 conveys temporal trends for the whole of Amazonia as well as M. del Vera Diaz et al. (An interdisciplinary model of soybean yield in the Amazon Basin, manuscript in preparation, 2007) whose Figure 1 locates soy centers). This key factor in the incentive to produce can affect not only deforestation given the transport infrastructure but also the plans to develop further infrastructure, such as for getting soy to travel through Belem headed eastward (Fearnside [2001] and Nepstad et al. [2002], for instance, provide further discussion of transport plans).
112 ROAD IMPACTS IN BRAZILIAN AMAZONIA The importance of such influences makes predicting forest cleating rates a challenge. Even with perfect knowledge oflocal conditions, and further even with perfect control over local transport iniitastructure and other policies, to perfectly predict both levels of deforestation and the impacts of new roads, one needs to consider the effects of such external influences. 7'. BROADER ECOSYSTEMIC CONTEXTS So far, we have examined all of the following: the local conditions, both biophysical and socioeconomic with a focus on roads, which affect local land uses and deforestation; the spillovers from those local roads to deforestation in nonlocal (e.g., neighboring) areas; and the effects of nonlocal socioeconomic changes on local forest outcomes. That leaves out the impacts of nonlocal biophysical changes. Without question, shifts in hydrological and climate systems elsewhere in Amazonia and outside can bring shifts in hydrology, climate in any given Amazonian location [see Silva Dias et al., this volume; Marengo et al., this volume]. This spatial interaction can go in both directions, i.e., many Amazonian changes matter elsewhere too. Stepping back, first, we should state explicitly that, in general, biophysical context strongly influences land use and deforestation. While this was not emphasized above, within the analyses of deforestation at all scales cited above, biophysical factors such as slope, rainfall total and distribution, and soil quality have repeatedly been shown to matter empirically, K. Anderson et al. (The effects of climate change on profitability and land use in Brazilian agriculture: A municipal cross-section analysis, Presentation at the XV Congresso Brasileiro de Agrometeorologia, 2007.) for some very recent work focused on climate impact, and see Chomitz and Thomas [2003] for evidence of such constraints. Given such effects, and given further that deforestation can affect ecosystem function, the potential for iterative feedback between land use choices and ecosystem adaptations exists. That allows, in principle, for multiple equilibria in land use and ecosystem function. While such models are not yet well developed, an example of one piece is the exploration of land clearing's impact on the climate by Moore et al. [2007]. When such connections are empirically established, alongside further work on land use impacts of particular elements such as rainfall distributions, perhaps integrated modeling of land use and ecosystems could enhance predictions of impacts. Finally, we must mention spatial pattern. To this point, we have focused explicitly only on the total amount of deforestation and the temporal pattern of clearing. Once some clearing and development have occurred, we suggested, the new road impacts on both deforestation and socioeconomic gains may be quite considerably altered.
PFAFF ET AL. Spatial pattern matters too (see, for instance, the works of Laurance and Bierregaard [1997] and Bierregaard etal. [2001]). For a given level of total forest loss, the general assertion is that when there are smaller, more irregular, and more isolated forest fragments remaining, then ecological function has been further impaired [Laurance and Biaregaard, 1997; Bierregaard et al., 2001; Laurance et aI., 2002]. Such impairment is said to potentially generate not only biodiversity loss but also a biomass collapse and carbon emissions that would contribute significantly to climate change [Gash et al., 1996]. Forest fi'agmentation links not only to such broad ecosystem changes but also to more local spatial interactions through fire. Degradation via fi'agmentation can raise ground surface temperatures and reduce precipitation, thereby elevating the risks of drought. Along with the increased litter fall fi'om dying trees, this raises the likelihood of fires, which fulther increase forest vulnerability by modifying vegetation structure [Cochrane et al., 1999; Nepstad et al., 2001; see also Meir et al., this volume]. As do results above, these ecological points suggest careful attention to spatial targeting of new road location. 8. CONCLUDING REMARKS 8.1. Road Context and Road Impact
Our greatest emphasis has been on context. The impacts of new roads on forest loss and social gains depend on the conditions into which roads are placed. Conditions that matter include the various biophysical factors that affect land use, such as slope, rainfall, and soil quality. They also include external socioeconomic factors like national policies, exchange rates, and the price of soybeans. Further, we emphasized, in particular, that those influential conditions include prior roads and deforestation. Where development has already come with investments and access for people, new roads may decrease forest less and raise production more than when entering pristine areas. Belief in such differences suggests careful consideration of where to invest further in transport. There are limits to the precision of temporal predictions of new road impacts based on the past. Yet even relatively rough robust differences in the impacts of new roads across space could matter for planning over time. Additional firstorder rationales derived from ecosystem science, such as about impacts of spatially fragmenting any given amount of forest, also seem relevant for road policy. Socioeconomic and ecological evidence both may support, for instance, leaving the standing forest in large tracts. As to how to do that, certainly one step could be targeting investments along routes that are already established. Paving
unpaved roads that have already generated clearing could, it seems, generate additional contributions to welfare with relatively lower marginal ,~ncreases in deforestation. Yet we have seen that road invesJinents may be followed by other investments including fu)!ther new roads. Thus, steps to limit sprouting or spreading d~velopment could have value in addressing trade-offs between development and conservation goals. It has been suggested; for instance, that placing protected areas alongside roads (in a "road park") might limit road impacts. While little evidence exists to comment on this possibility, C. Delgado et al. (Might protected areas constrain road deforestation impacts? Chico Mendes Extractive Reserve and the Inter-Oceanic Highway, paper presented at Amazon in Perspective Conference, Inst. Nac. de Pesqui. da Amazonia, Manaus, Brazil, 2008.) analyze the Chico Mendes Extractive Reserve and find that, while it has been cleared more than other reserves in Acre, it is actually blocking considerable clearing given its location near the Interoceanic Highway. 8.2. Frontier Governance and Relative Impacts
The issue of predicted impacts has been raised in several useful illustrations of scenarios concerning future'rates of deforestation in Amazonia [see, for instance, Laurance et al., 2001; Soares-Filho et al., 2006 and Walker et aI., 2007). Taken as a whole, the Amazonian forest scenarios appear to suggest that some form of enhancement of reserves or zoning or governance could have much larger impacts in, e.g., lowering forest loss, than would potential changes due to altering the planning of new roads. This raises a few questions, including how best to count road impacts for spatial transport plamling. When a new road penetrates a pristine area, as noted, there will be immense pressure for federal, state, and local complementary investments to improve local quality of life over time and space. Thus, however one delineates the marginal impact of each investment (which could be complicated), the long-run effect ofthe new road accessing a pristine area is greater, by far, than the initial years' local forest losses. It also raises the question of what frontier governance is feasible. Modeling that generates glowing potential impacts of governance assumes, whatever level of prevention of deforestation is desired, can be attained and, implicitly, at reasonable cost. That may be the case for Costa Rica, but it may not be for the gigantic frontier of Amazonia. Chomitz [2006] emphasizes varied governance challenges. Blending governance and roads, our discussions concerning the pressures driving unofficial roads highlight critical issues. On the one hand, these roads are crucial to livelihoods and communities' development in frontier areas. On the other, unofficial roads generate ecological losses as well
113
as some social problems and, if local resource extraction is unsustainable, may only delay inevitable declines in frontier communities. This means that for an optimal path, unofficial roads require effective governance. Environmental governance generally has received considerable attention in recent years, emphasizing state- and community-based models [Paz et al., 2007b]. State-based models emphasize parks, tax breaks and incentives for sustainable resource use, and punishment of violators. They have had only mixed effectiveness within Amazonia, however, due to the large areas to be monitored and limited enforcement capacities. For unofficial roads, the lack of state presence in Amazonia has produced a generation oflocal players used to relative autonomy and resistant to state-imposed regulations. This led to discussion of community-based approaches to enviromnental governance. They highlight the fact that people in Amazonia have managed natural resource for generations and are increasingly demonstrating an ability to mobilize and form local organizations. Yet community governance is hampered by local inequalities and the potential for capture by powerful families, as well as by a limited capacity to respond to external threats or large-scale processes [Paz et al., 2007b]. Hybrid governance models might combine state capacity and oversight with yommunity-based participation and responsiveness to local exigencies. The Instituto de Pesquisa Ambiental da Amazonia has sought to work with communities along BR-163 to engage in zoning along the corridor [Nepstad et al., 2002]. Stakeholder workshops have sought to link data on past land use and projections of future scenarios to planning [Alencar et al., 2004]. Another example is the "Madre de Dios, Acre, Pando (MAP) Initiative" in southwestern Amazonia where the Interoceanic Highway is being paved (see Iniciativa MAP, 2007, available at www. map-amazonia.net and Brown et al. [2002]). This trinational effort has focused on cross-border exchanges among stakeholders on roads, climate, and other prospective changes in the region. It has led to grassroots activities with some state support to plan for and to mitigate the impacts of paving official roads and expansions of unofficial roads. In thinking about future road impact, we must understand not only official spatial development plans and the broader contexts in which new road investments may occur but also the critical local contexts. Aclmowledgments. A. Pfaff thanks numerous researchers who . are not authors but whose work and suggestions contributed to this chapter, including many at LBA Science Meetings. He also gratefully acknowledges all the financial support provided by the Tinker Foundation Inc. and NASA's "LBA" program (i.e., NASA projects NCC5-694 and NNG06GD96A). Funding for the doctoral training of Thomas Ludewigs at Indiana University (ill) was provided by
114
ROAD IMPACTS IN BRAZILIAN AMAZONIA
the National Council of Technological and Scientific Development (CNPq/Brazilian government), by the Anthropological Center for Training and Research on Global Environmental Change (ACTIU) as partflf a NASA research grant, by the School of Public and Environmental Affairs (SPEA-IU), by the Center for Institutions, Population and Global Environmental Change (CIPEC-IU), and by the WWF Prince Bernhard Scholarship Fund for Nature Conservation.
REFERENCES Alencar, A., D. Nepstad, D. McGrath, P. Moutinho, P. Pacheco, M.del C. Vera Diaz, and B. Soares Filho (2004), Desmatamento na Amazonia: Indo Alem da 'Emergencia Cronica', Belem, IPAM. Almeida, A. L. O. (1992), The Colonization of the Amazon, Univ. of Texas Press, Austin, TX. Almeida, O.T. d., and Uhl, C. (1995), Developing a quantitative framework for sustainable resource-use plmming in the Brazilian Amazon, FVorld Del'. , 23,1745-1764. Alves, D. S. (2002), Space-time dynamics of deforestation in Brazilian Amazonia, Int. J. Remote Sens., 23, 2903-2908. Alves, D. S., D. C. Morton, M. Batistella, D. A. Roberts, and C. Souza Jr. (2009), The changing rates and patterns of deforestation and land use in Brazilian Amazonia, Geophys. Monogr. Ser., doi: 10.1029/2008GM000722, this volume. Andersen, L. E., C. W. J. Granger, E. J. Reis, D. Weinhold, and S. Wunder (2002), The Dynamics of Deforestation and Economic Growth in the Brazilian Amazon, Cambridge Univ. Press. Arima, E. Y., R. T. Walker, S. G. Perz, and M. M. Caldas (2005), Loggers and forest fi'agmentation: Behavioral models of road building in the Amazon basin, Ann. Assoc. Am. Geogr., 95, 525541. Arima, E. Y., R. T. Walker, M. Sales, C. Souza Jr., and S. G. Perz (2008), Emergent road networks and the fi'agmentation of space in the Amazon Basin, Photogramm. Eng. Remote Sens., 74(6), 699-709. Barat,1. (1978), A Evolucao dos Transportes no Brasil, IBGE, Rio do Janeiro. Barbieri, A. F. (2006), People, Land, and Context: Multi-scale Dimensions of Population Mobility in the Ecuadorian Amazon, ProQuestlUMI, Ann Arbor, Mich. Barbieri, A. F., and D. L. Carr (2005), Gender-specific out-migration, deforestation and urbanization in the Ecuadorian Amazon, Global Planet. Change, 47(2-4), 99-110. Barbieri, A. F., D. O. Sawyer, and B. S. Soares-Filho (2005), Population and land use effects on malaria prevalence in the Southern Brazilian Amazon, Human Ecol., 33(6), 847-874. Barbieri, A. F., R. E. Bilsborrow, and W. K. Pan (2006), Farm household lifecycles and land use in the Ecuadorian Amazon, Populo Environ., 27(1),1-27. Barbieri, A. F., R. L. M. Montemor, and R. E. Bilsborrow (2009), Towns in the jungle: Exploring linkages between rural-urban mobility, urbanization and development in the Amazon (no prelo), in Urban Population and Environment Dynamics in the
PFAFF ET AL.
Developing World: Case Studies and Lessons Leamed, edited by A. de Sherbiniin et aI., CICRED, Paris, France. Bauch, S. C., G. S. Amacher, and F. D. Merry (2007), Cost ofharvest, transport, and milling in the Brazilian Amazon: Estimation and policy implications, For. Policy Econ., 9, 903-915. Bierregaard, R. 0., Jr., C. Gasc6n, T. E. Lovejoy, and R. C. G. Mesquita (Eds.) (2001), Lessons fi'om Amazonia: The Ecology and Conservation of a Fragmented Forest, Yale Univ. Press, New Haven, Conn. Bilsborrow, R. E., A. F. Barbieri, and W. Pan (2004), Changes in population and land use over time in the Ecuadorian Amazon, Acta Amazonica, 34(4), 635-647. Brandao, A. 0., Jr., and C. M. Souza Jr. (2006), Mapping unofficial roads with Landsat images: A new tool to improve the monitoring ofthe Brazilian Amazon rainforest, Int. J. Remote Sens., 27, 177-189. Brondizio, E. S., A. Cak, M. M. Caldas, C. Mena, R. Bilsborrow, C. T. Futemma, T. Ludewigs, E. F. Moran, and M. Batistella (2009), Small farmers and deforestation in Amazonia, Geophys. Monogr. Ser., doi: 1O.1029/2008GM000716, this volume. Browder, 1. 0., and B. 1. Godfrey (1997), Rai11!'orest Cities: Ur-
banization, Development and Globalization of the Brazilian Amazon, Columbia Univ. Press. Brown, 1. F., S. H. C. Brilhante, E. Mendoza, 1. R. de Oliveira (2002), Estrada de Rio Branco, Acre, Brasil aos Portos do Pacifico: Como maximizar os beneficios e minimizar os prejuizos para 0 desenvolvimento sustentavel da Amazonia Sul-Ocidental, in La Integracion Regional entre Bolivia, Brasil y Peru, Serie Seminurios, Mesas Redondas y COI(j'erencias, no. 25, edited by A. Wagner Tiz6n, and R. Santa Gadea Duarte, pp. 281-296, CEPEI (Centro Peruano de Estudios Internacionales), Lima, Peru. Cattaneo, A. (2001), Deforestation in the Brazilian Amazon: Comparing the impacts of macroeconomic shocks, land tenure and technological change, Land Econ., 77(2), 219-240. Cattaneo, A. (2005), Inter-regional innovation in Brazilian agriculture and deforestation in the Amazon: Income and environment in the balance, Environ. Dev. Econ., 10, 485-511. Chomitz, K. M. (2006), At Loggerheads? Agricultural E\pansion, Poverty Reduction, and Environment in the Tropical Forests, World Bank Policy Research Report, World Bank, Washington, D. C. Chomitz, K. M., and T. S. Thomas (2003), Determinants of land use in Amazonia: A fine-scale spatial analysis, Am. J. Agric. Econ., 85(4),1016-1028. Cochrane, M., A. Alencar, M. D. Schulze, C. M. Souza Jr., D. C. Nepstad, P. Lefebvre, and E. A. Davidson (1999), Positive feedbacks in the fire dynamic of closed canopy tropical forests, Science, 284, 1832-1835. Conde Ovando, D. A. (2008), Road impacts on jaguar habitat in the Mayan forest, Ph.D. thesis, Duke Univ., Durham, N. C. Dean, W. (1995), With Broadx and Firebrand-The Destruction of the Brazilian Atlantic Forest, Univ. of California Press, Los Angeles. Diniz, C. C. (1995), A Dinamica Regional Recente da Economia Brasileir e Suas Perspectivas, Texto para Discussao, No. 375, IPEA, Rio de Janeiro, Brazil.
Dunn, E. S., Jr. (1970), Thb equilibrium of land-use patterns in agriculture, in Spatial E;conomic Theo/)" R. D. Dean et aI., pp. 233-249, The Free Press, New York. Fearnside, P. M. (200 ~!): Soybean cultivation as a threat to the environment in Brazil,jiEnviron. Conserv., 28(1), 23-38. Fearnside, P. M. (2007), Brazil's Cuiba-Santarem (BR-163) Highway: The environmental cost of paving a soybean corridor through the Amazon, Environ. Manage., 39, 601-614. Forman, R. T. T., et' al. (2003), Road Ecology: Science and Solutions, Island Press, Washington, D. C. Gash, J. C. H., C. A. Nobre, 1. M. Roberts, and R. L. Victoria (Eds.) (1996), Amazonian Deforestation and Climate, John Wiley, Chichester. Geist, H. J., and E. F. Lambin (2001), What Drives Tropical Defor-
estation? A Meta-analysis ofProximate and Underlying Causes of Deforestation Based on Subnational Case Study Evidence, LUCC Report Series, no. 4, LUCC International Project Office, Univ. of Louvain, Belgium. Goodland, R. J. A., and H. S. Irwin (1975), Amazon Jungle: Green
Hell to Red Desert? A Discussion of the Environmental Impact ofthe Highway Construction Program in the Amazon Basin, Elsevier, New York. Goulart, 1. A. (1959), Meios de Transpore e Instrumentos de Transporte no Interior do Brasil, Minesterio da Educacao e Culture, Servico de Documentacao, Rio de Janeiro, Brazil. Grogan, 1., P. Barreto, and A. Verissimo (2002), Mogno na Amazonia Brasileira: Ecologia e Manejo, IMAZON, Belem, Brazil. Hall, A. L. (1989), Developing Amazonia: Deforestation and Social COIiflict in Brazil's Carajas Programme, Manchester Univ. Press, Manchester, UK. Heckandon, S. (1983), Cuando se Acaban los Montes, Impretex (for Smithsonian Tropical Research Institute), Panama City, Panama. Kaimowitz, D. (1997), Factors determining low deforestation: The Bolivian Amazon, Ambio, 26(8), 537-540. Laurance, W. F., andR. O. BierregaardJr. (Eds.) (1997), Tropical For-
est Remnants: Ecology, Management and Conservation of Fragmented Communities, University of Chicago Press, Chicago, Ill. Laurance, W. F., M. A. Cochrane, S. Bergen, P. M. Fearnside, P. Delamonica, C. Barber, S. D'Angelo, and T. Fernandes (2001), The future of the Brazilian Amazon, Science, 291, 438-439. Laurance, W. F., T. E. Lovejoy, H. L. Vasconcelos, E. M. Bruna, R. K. Didham, P. C. Stouffer, C. Gascon, R. O. Bierregaard, S. G. Laurance, and E. Sampaio (2002), Ecosystem decay of Amazonian forest fragments: A 22-year investigation, ConserI'. BioI., 16, 605-618. Lentini, M., D. Pereira, D. Celentano, and R. Pereira (2005), Fatos Florestais da Amazonia 2005, IMAZON, Belem, Brazil. Lima, E., and F. Meny (2003), The views of Brazilian producersIncreasing and sustaining expOlis, in Growing Timber Exports:
The Brazilian Tropical Timber 1ndustly and Intel'l1ational Markets, IIED Small and Medium Ente/prise Series, no. 1., edited by D. Macqueen, pp. 82-102, IIED, London, UK. Lima, E., F. Meny, D. Nepstad, G. Amacher, C. Azevedo-Ramos, F. Resque, and P. Lefebvre (2006), Searching for sustainability: Forest policies, smallholders, and the Trans-Amazon Highway, Environment, 48, 26-37.
115
Ludewigs, T. (2006), Land-use decision making, uncertainty and effectiveness of land reform in Acre, Brazil, Ph.D. dissertation, Indiana University, Bloomington, Ind. Mahar, D. (1989), Govel'l1ment Policies and Deforestation in Brazil's Amazon Region, Technical report, World Bank (World Wildlife Fund, Conservation Foundation), Washington, D.C. Marengo, 1., C. A. Nobre, R. Betts, P. Cox, G. Sampaio, and L. Salazar (2009), Global warming and climate change in Amazonia: Climate-vegetation feedback and impacts on water resources, Geophys. Monogr. Ser., doi:IO.102912008GM000743, this volume. Meir, P., et al. (2009), The effects of drought on Amazonian rain forests, Geophys. Monogr. Ser., doi:1O.1029/2008GM000882, this volume. Mena, C. F., A. F. Barbieri, S. 1. Walsh, C. M. Erlien, F. L. Holt, and R. E. Bilsborrow (2006), Pressure on the Cuyabeno Wildlife Reserve: Development and land use/cover change in the Northern Ecuadorian Amazon, World Dev., 34,1831-1849. Meny, F., G. Amacher, D. Nepstad, P. Lefebvre, E. Lima, and S. Bauch (2006), Industrial development on logging frontiers in the Brazilian Amazon, Int. J. Sustain. Dev., 9, 277-296. Monte-M6r, R. L. (2004), Model'l1ities in the Jungle: Extended Urbanization in the Brazilian Amazonia, Calif. unpublished Ph.D. disseliation, University of California, Los Angeles. Moore, H., E. Arima, R. Walker, and R. Ramos da Silva (2007). Uncertainty and the changing hydroclimatology of the Amazon, Geophys. Res. Lett., 34, L14707, doi:1O.1029/2007GL030157. Mueller, C. C. (1983)10 Estado e a Expansao da Fronteiro Agropecuaria na Amazonia Brasileira, Estud. Econ., 13(3), 657-679. Myers, N., R. A. Mitt~rmeier, C. G. Mittenneier, G. A. da Fonseca, and 1. Kent (2000), Biodiversity hotspots for conservation priorities, Nature, 403, 853-858. Nelson, M. (1973), The Development of Tropical Lands, Johns Hopkins Univ. Press, Baltimore, Md. Nepstad, D. C., et al. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505-508. Nepstad, D., G. Carvalho, A. C. Barros, A. Alencar, 1. P. Capobianco, 1. Bishop, P. Moutinho; P. Lefebvre, U. Lopes Silva Jr., and E. Prins (2001), Road paving, fire regime feedbacks, and the future of Amazon forests, For. Eco!. Manage., 154, 395-407. Nepstad, D. C., D. McGrath, A. Alencar, A. C. Barros, G. Carvalho, M. Santilli, and M. del C. Vera Diaz (2002), Frontier governance in Amazonia, Science, 295, 629-630. Nepstad, D., A. Alencar, A. C. Barros, E. Lima, E. Mendoza, C. A. Ramos, and P. Lefevre (2004), Goveming the Amazon timber industry, in Working Forests in the Neotropics: Consen/ation Through Sustainable Management?, edited by D. Zarin et aI., Columbia Univ. Press, New York. Nepstad, D. C., C. M. Stickler, and O. T. Almeida (2006), Globalization of the Amazon soy and beef industries: Opportunities for conservation, ConserI'. Bioi., 20,1595-1603. Owen, W. (1987), Transportation and World Development, Johns Hopkins Univ. Press, Baltimore, Md. Peri, S. G., C. Souza Jr., E. Arima, M. Caldas, A. O. Brandao Jr., F. K. A. Souza, and R. Walker (2005),0 dilemma das estradas nao-oficiais na Amazonia, Cienc. Hoje, 37, 56-58.
116
ROAD IMPACTS IN BRAZILIAN AMAZONIA
Perz, S. G., M. M. Caldas, E. Y. Arima, and R. T. Walker (2007a), Socio-spatia1 processes of unofficial road-building in the Amazon: Socioeconomic and biophysical explanations, Dev. Change, 38, 529-5~1. Perz, S. G., C. Overdevest, E. Y. Arima, M. M. Caldas, and R. T. Walker (2007b), Unofficial road building in the Brazilian Amazon: Dilemmas and models of road governance, Environ. Conserv., 34, 112-121. Pfaff, A.' (1999), What drives deforestation in the Brazilian Amazon? Evidence from satellite and socioeconomic data, J. Environ. Econ. Manage., 37(1), 26-43. Pfaff, A. (2000), From deforestation to reforestation in New England, USA, in Global Prospects ofDeforestation and Forest Transition, edited by M. Palo and H. Vanhanen, Springer, New York. Pfaff, A., and 1 Robalino (2009), REDD Roads? Spatial frontier dynamics and spatial variation in causal impacts in the Brazilian Amazon, Presentation and draft for EPA Spatial Land-use Frontiers of Economics Workshop, Washington, D. C., 26 June. Pfaff, A., and R. Walker (2009). Regional Interdependence and Forest Transitions: Forest loss elsewhere limits local transitions' global relevance, Land Use Policy, in press. Pfaff, A., et al. (2006), Econometric estimation of deforestation impacts from roads and other drivers, paper presented at LBA 10th Science Team Meeting, Natl. Inst. of Amazonian Res., Brasilia, 6 Oct. Pfaff, A., et al. (2007), Road investments, spatial spillovers, and deforestation in the Brazilian Amazon, J. Reg. Sci., 47, 109-123. Reid, 1 W., and I. A. Bowles (1997), Reducing the impacts of roads on tropical forests, Environment, 39,10-17. Reis, E. 1, and R. Guzman (1992), An econometric model of Amazon deforestation, IPEAIRio Worldng Papers. Rudel, T. (1983), Roads, speculators, and colonization in the Ecuadorian Amazon, Human Ecol., 11,385-403. Schmink, M., and C. H. Wood (1992), Contested Frontiers in Amazonia, Columbia Univ. Press, New York. Silva Dias, M. A., R. Avissar, and P. Silva Dias (2009), Modeling the regional and remote climatic impact of deforestation, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000817, this volume. Silva, M. M. F. (1949), Geograjia dos Transportes no Brasil, lEGE, Rio de Janeiro, Brazil. Sitmuons, C. S., R. T. Walker, E. Arima, S. Aldrich, and M. Caldas (2007), The Amazon Land War in the South ofPal'll, Ann. Assoc. Am. Geogr., 97(3),567-592. Soares-Filho, B. S., D. C. Nepstad, L. M. Curran, G. C. Cerqueira, R. A. Garcia, C. A. Ramos, E. Voll, A. McDonald, P. Lefebvre, and P. Schlesinger (2006), Modeling conservation in the Amazon Basin, Nature, 440, 520-523.
Southgate, D., R. Sierra, and L. Brown (1991), The causes oftropical deforestation in Ecuador: A statistical analysis, World Dev., 19(9),1145-1151. Stone, S. W. (1998), Evolution of the timber industry along an aging frontier: The case of Paragominas (1990-1995), World Dev., 26,433-448. Summerhill, W. R. (2003), Order Against Progress, Stanford Univ. Press, Stanford, Calif. Trombulak, S. C., and C. A. Frissell (2000), Review of ecological effects of roads on terrestrial and aquatic ecosystems, Conserv. Bioi., 14,18-30. Vance,l E. (1986), Capturing the Horizon: The Historical Geography o.lTransportation, HarperCollins, New York. Vosti, S., C. Carpentier, 1 Witcover, and 1 Valentim (2001), Intensified small-scale livestock systems in the Western Brazilian Amazon, in Agricultural Technologies and Tropical Deforestation, edited by A. Angelsen and D. Kaimowitz, CABI, Oxfordshire, UK. Walker, R. T. (2003), Mapping process to pattern in the landscape change of the Amazonian Frontier, Ann. Assoc. Am. Geogr., 93(2), 376-398. Walker,R. T., S. Perz,M. Caldas, andL. G. Texeira da Silva (2002), Land use and land cover change in forest frontiers: The role of household life cycles, Int. Reg. 'Sci. Rev., 25(2), 169-199. Walker, R. T., E. Reis, A. Pfaff, and S. Perz (2007), A Basin-Scale Econometric l'Vlodel for Projecting Future Amazonian Landscapes, Final Report, NASA NCC5-694. Walker, R., et al. (2009), Ranching and the new global range: Amazonia in the 21st century, Geo.lorum, doi:10.10161 j.geoforum.2008.l0.009, in press. Weinhold, D., and E. 1 Reis (2008), Transportation costs and the spatial distribution of land use in the Brazilian Amazon, Global Environ. Change, 18, 54-68. A. Barbieri, Cedeplar, Federal University of Minas Gerais, Belo Horizonte, MG 30170-120, Brazil. ([email protected]) T. Ludewigs, World Bank, Brasilia, DF 70712-900, Brazil. ([email protected]) F. Meny, Woods Hole Research Center, Falmouth, MA 025401644, USA. ([email protected]) S. Perz, Department of Sociology and Criminology and Law, University of Florida, Gainesville, FL 32611, USA. (sperz@soc. ufl.edu) A. Pfaff, Sanford School of Public Policy, Duke University, Durham, NC 27708, USA. ([email protected]) E. Reis, Instituto de Pesquisa Economica Aplicada, Rio de Janeiro, RJ 20020-010, Brazil. ([email protected])
Small Fanners and Deforestation in Amazonia Eduardo S. Brondizio,l Anthony Cak,2 Marcellus M. Caldas,3 Carlos Mena,4,5 Richard Bilsborrow,6 Celia T. Futemma,7 Thomas Ludewigs,8 Emilio F. Moran, l and Mateus Batistella 9 This chapter discusses the relationship between small farmers' land use and deforestation, with particular attention paid to the past 30 years of Amazonian colonization in Brazil and Ecuador. Our analysis calls attention to common features uniting different social groups as small farmers (e.g., social identity, access to land and resources, technology, market, and credit), as well as the variability between small farmers in terms of time in the region (from native populations to recent colonists), contribution to regional deforestation, types of land use systems. At a regional level, small farmers contribute to the majority of deforestation events, but are responsible for only a fraction of the total deforested area in Amazonia. We discuss three misconceptions thflt have been used to define small fanners and their contribution to the regional economy, development, and deforestation: (1) small farmers have backward land use systems associated with low productivity and extensive deforestation and subsistence production, (2) sma'll farmers contribute to Amazonian deforestation as much as large farmers, and (3) small farmers, particularly colonist farmers, follow an inexorable path of: deforestation unless curbed by government action. We conclude the chapter discussing their growing regional importance and the need for more inclusive public policies concerning infrastructure and services and valorization of resources produced in rural areas ofAmazonia.
I Department of Anthropology and Anthropological Center for Training and Research on Global Environmental Change, Indiana University, Bloomington, Indiana, USA. 2School of Public and Environmental Affairs and Anthropological Center for Training and Research on Global Enviromuental Change, Indiana University, Bloomington, Indiana, USA.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10. 1029/2008GM0007 16
3Department of Geography, Kansas State University, Manhattan, Kansas, USA. 4Colegio de Ciencias Biologicas y Ambientales, Universidad San Francisco de Quito, Quito, Ecuador. 5Also at Department of Geography, University of North Carolina, Chapel Hill, NOlih Carolina, USA. 6Biostatistics Department and Carolina Population Center, Uni, versity of North Carolina, Chapel Hill, North Carolina, USA. 7Sorocaba Campus, Universidade Federal de Sao Carlos, Sao Paulo, Brazil. 8Center for Sustainable Development, University of Brasilia, Brasilia, Brazil. 9Embrapa Monitoriamento por Satelite, Campinas, Brazil. 117
118
BRONDIZIO ET AL.
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
1. INTRODUCTION Small fanners represent diverse social groups in Amazonia. (Om~lse of the term small farmer in this chapter also includes social groups often designated under categories such as extractivists, colonists, agroextractivists, quilombolas, "traditional populations," family agriculture, collectors, fishers who practice agriculture, and other regional-cultural designations. In some cases, these categories involve a combination of collective and private use of resources. It also includes those who, old or new to the region, may not have formal land title or proof of use rights but who are directly involved with land use and resource management. For these reasons, in some instances we use the term small holders interchangeably with small farmers.) As such, this encompassing term has been used broadly by researchers and policy makers to describe diverse populations with cultural, historic, demographic, and economic variability. In different ways, they have been significant to the understanding of the dynamics of land use and deforestation, resource management and governance, and urbanization in the region. The histOly of small fanners in Amazonia spans fi'om historic populations (i.e., historical peasantly), usually lumped together under the terms Caboclo or Riberen5, and migrant colonists dating back to colonization schemes of the late nineteenth centmy (e.g., the region near Bragantina, in the northeastern part of the state of Para, Brazil) to successive migrations of colonists throughout the twentieth centmy, particularly following government-sponsored projects and spontaneous migrations beginning in the late 1960s. Each of these successive waves of colonists arrived in Amazonia fi'om different areas, with different backgrounds, and with different reasons for settling in the region. Even today, small farmers continue to arrive to and migrate within the region; for example, the number of colonization settlements (for agrarian reform projects) in the region has reportedly tripled between 1994 and 2002, with over 500,000 families settled during this period [Barreto et al., 2005; INCRA, 2000, 2002]. In the north region of Brazil, which encompasses the area defined as Amazonia Legal, family-based farms represent 85.4% of all rural propeliies and cover 37.5% of the total area (ha) [Glianziroli et al., 2001]. This is comparable to other areas of Brazil, where, according to preliminaty data from the 2006 Brazilian Agricultural Census [IBGE, 1998, 2009a], in Brazil there are 5,204,130 rural properties (estabelecimentos rurais) covering an area of354.9 million ha, of which around 85% represent small holders (estabelecimentos rurais familiares) and occupy 107.8 million ha (30.5% of the area of all rural properties). In other regions of the Amazon Basin outside Brazil, however, colonization by small farmers has been largely
spontaneolls. This is the case, for example, in Northern Ecuadorian Amazonia (NEA), where only three small colonization projects were executed with limited success [Uqllillas, 1984; Tamariz and Villaverde, 1997; Pichon and Bilsborrow, 1999]. Thus, virtually all agricultural colonization in NEA has been unstructured, with the role ofthe government limited to passive granting of temporary or provisional land titles (certificados de posesion) to groups (precooperativas) of colonists (colonos) after they initially settle on the land in a certain area, with later granting of permanent land titles (escrituras) [Barsky, 1984]. (Colonist families pay small amounts for each, but nearly half of the farm families establishing the initial farms or fincas in the 1970s and 1980s never made the final payments for escrituras before the government land titling institute (Ecuadorian Institute of Agrarian Refonn and Colonization (IERAC)) was eliminated in 1993 as part of public sector shrinking neoliberal policies.) Much of the current debate about deforestation in Amazonia has been defined by contrasting land use and land cover change resulting fi'om small- and large-scale fanning activities [Walker et al., 2000; Aldrich et al., 2006; OESP,2008a, 2008b]. Yet, in aggregated'tenns (i.e., at the state level), small farmers contribute to only a small portion of the areal extent of regional deforestation, although these smaller, deforested areas represent the majority of deforestation events in the region. The ambiguity in defining small farmers, especially given their diverse backgrounds and motivations, and stereotypes about small-scale production systems as destructive and backward have played a role in the interpretation of their contribution to regional deforestation. For these reasons, this is a discussion which requires both an examination of figures and rates concerning land use change and considerations about the political ecology of regional development. Small-scale production systems are often lumped together as a single category and regarded as lacking technology and resembling abandoned fields while disregarded in terms of their productivity, contribution for food production, and agro-diversity. Furthermore, the lack of data on the contributicm of small-scale production systems to the regional economy, to export, and food provisioning to urban centers has added to the lack of understanding about their regional importance. In this context, some aspects of small farmers are rendered visible, while others invisible depending on the interlocutors [Brondizio, 2004; Costa, 2006]. Although much remains to be learned about small-scale land use systems in Amazonia, since the 1980s, there has been a growing body of research shedding light on certain segments of small farmers, especially colonists, who represent an impOliant group driving land cover change in the region [e.g.,Moran, 1981, 1990; Smith, 1982; Fea1'l1side, 1986;Lena and Oliveira, 1992; Arcu!jo, 1993; Caviglia, 1999; Jones
et al., 1995; McCracken et al., 1999,2002; Brondizio et al., 2002; Moran et al., 2092; Walsh et al., 2002; Flitemma and Brondizio, 2003; CalGas et al., 2007; Caldas, 2008; Tura and Costa, 2000; NJwphy, 2001; Walker, 2003; Perz, 2001; Perz and Walker,l2002; Pichon et al., 2002; Castallenet and Jordan, 2002; (;osta et al., 2006; Browder et al., 2008, among many others]. However, proportional to their demographic and economic importance, the contribution of small fanners to regional land use and food production continues to be invisible and stigmatized in the eyes of policy makers and some sectors of the regional population. While large-scale producers continue to be responsible for driving regional deforestation, small farmers are playing an ever-increasing role in the region through the fonnation of a complex socialenvironmental mosaic in the region. They connect rural and urban areas through social and economic networks [Padoch et al., 2008]; they are present in virtually allnonindigenous reserves and increasingly involved with regional conservation issues [Campos and Nepstad, 2006]; they are present across consolidated and new areas of agropastoral expansion [Costa, 2008; Moran et al., 2008]; they are the basis of a large economy involving river and forest resources [Smith et al., 2007; Brondizio, 2008]; their production systems include a gradient from velY intensive, diverse and agronomically sophisticated to extensive, opportunistic and unproductive systems [Pinedo-Vasquez et al., 2002, 2003; Silva-Forsberg and Fea1'l1side, 1997; Marquardt, 2008; Brondizio and Siqlleira, 1997; Peroni et al., 2007; Smith et al., 1996]; they experience violence and conflict associated with land and resources [Simons, 2005]; they represent an expanding regional political movements [Campos, 2006]; and they serve as an emblem of the region's challenges to face climate change and prospects of sustainable development [Oz6rio de Almeida and Campari, 1995; Zarin et al., 2004; Brondizio and Moran, 2008]. This chapter discusses the relationship between small fanners (and small holders in general) and land use and deforestation, with particular attention paid to the past 30 years of Amazonian colonization. Our analysis calls attention to common features uniting different social groups as small fanners or small holders (e.g., social identity, access to land and resources, technology, market, and credit), as well as the variability between small farmers in terms of time in the region (fi'om native populations to recent colonists) and other variables. These features have been important factors influencing land use behavior and deforestation in the region. We stmi by exploring different definitions of what constitutes' small farmers in the region, based on categories such as historical groups, income, and farm size. (While our analysis includes as small farmers those who may be landless and work as sharecroppers, we do not examine the landless movement
119
in the region as the complexity of the topic would require a full treatment in an article.) Following an examination of deforestation trajectories at the levels ofthe region, settlement (groups of farm lots), and individual farm lots, we focus our discussion on three misconceptions that have been used to define small farmers and their contribution to the regional economy, development, and deforestation: (1) small farmers have backward land use systems associated with low productivity and extensive deforestation and subsistence production, (2) small farmers contribute to Amazonian deforestation as much as large farmers, and (3) small farmers, particularly colonist fanners, follow an inexorable path of deforestation unless curbed by government action. We base our discussion on long-term research sites where coauthors have worked as part of the broader Large-Scale BiosphereAtmosphere (LBA) Experiment in Amazonia Program [Batistella and Moran, 2005; Costa et al., 2007; Batistella et al., 2008], including colonization settlements in the states ofPara, Acre, and Rondonia in Brazil, and colonization areas in the Ecuadorian Amazon (Figure 1). (Our constraint coverage of Brazil and Ecuador, based on LBA-related projects, leaves out important dynamics associated with small fanners in Bolivia, Peru, Colombia, and Venezuela which involves similar but also velY different processes associated with public policies, economy and demography changes, production of illegal crops and drugs, and various forms of conflicts with large-scale logging concessions, large-scale farmers, oil companies, and government programs. However, we believe that most of the issues discussed here from the perspectives of Brazil and Ecuador are also relevant to other Amazonian countries.) Supporting data and statistical analysis, much of which have been published elsewhere, involve household and farm surveys, long-term ethnographic investigation, as well as multitemporal remote sensing and spatial analysis used to assess the contribution of fanners of different sizes to regional deforestation. 2. DEFINING SMALL FARMERS IN AMAZONIA Although small farmers in Amazonia have remained analytically ill defined, there has been increasing attention to understanding sociocultural and political differences and similarities among different social groups and categories lumped together as the Amazonian peasantly [Brondizio, 2004; Adams et al., 2008]. The use of this term has greatly varied according to the interlocutor, purpose and political context, or region of interest. Three common typological definitions of small farmers and small holders have been employed in the region, based on (1) historical and social categories, (2) economic and income classes, or (3) landbased classes, such as farm size. Table 1 provides a com-
BRONDIZIO ET AL. 800'(J"W
121
500'(J"W
600'(J"W
700'0"W
parative illustration of different typologies to characterize groups of farms and fatmers based on economic and farmsize criteria. An altetnative categorization, based on the organization of fann;':Jabor and broad class structure, has been proposed by tlHJ.A3razilian National Institute of Colonization and Agrarian Reform (INCRA) [INCRA/FAO, 2000] and defines family-based, in contrast to patron-based (i.e., proxy for large farmers) agriculture and farm operations. Family agriculture, in this case, is defined by (1) the main decision maker as the producer, (2) the number of paid laborers as lower or equal to the amount offamily labor, and (3) the size of property as lower or equal to the regional pattern [Guanziroli et al., 2001]. (Family labor includes individuals who are 14 years old or older; individuals younger than 14 years old are considered but to be part-time (50%) workers. Each region in Brazil maintains a different maximum property size criteria that is established by the Brazilian Federal government; thus the size of family prope1ties cannot exceed regional requirements. In the case of the north region, the maximum property size is 1122 ha [Guanziroli et al., 2001].)
2.1. Historical Categories Historical and social categories have generally been used to contrast groups related to the colonial history ofthe region with recent migrants, but has been less clear about those in between. For example, a variety of social groups have been lumped together under the tenn Caboclo, a te1m that has been used to describe nonindigenous historical or "traditional" populations and, until recently, has canied a strong stigma and negative connotation [Brondizio, 2004, 2008; Adams et al., 2008; Hiraoka, 1992]. During the past decade, the concept of "traditional population" to refer to historical peasant groups unde1taking small-scale, forest-based land use systems and often occupying areas of interest for forest conservation has become an increasingly popular and significant legal instrument to guarantee land rights to numerous communities and families of small holders characterized as small falmers, extractivists, quilombolas, fishers, and/or a number of other historical and regional denominations. It, in tum, has also become a marker of cultural identity and ancestry throughout the region. However, and arguably, the
Table 1. Property Size Class Categories by Income and Area From Fundo Constitucional de Financiamento do Norte Program (FNO), 800'(J"W
60 O'(J"W
700'(J"W
800'(J"W
Legend Brazilian Rural Population Density (persons*kn1
Brazilian National Institute of Colonization and Agrarian Reform (INCRA), and Brazilian Ipstitute of Geography and Statistics (lEGE) . for Brazil and for the Northern Ecuadorian Amazona
500'(J"W
700'(J"W
600'(J"W
Income in the NOlihern Ecuadorian Amazon
500'(J"W
2
PRONAF
)
0,0 - 2,5 2,5 - 5,0 5,0 - 10,0
Fmm Size Classes (ha)
p 0 0
10.0 - 50.0 50.0 - 100,0 >100,0 -Rivers -Highways # Cities State Capitals ! A Country Capitals ~ Study Areas E'Za Rural Settlements I::::: :1 Indigenous, Conservation, and State and Federal Protected Areas
V)
p
V)
p
0
0
S
S
V) V)
P0 p,
P 0 P,
V) V)
P
p
Gross Income (R$ (% From Farming))
lEGE Size Classes
FNO
Gross Income (R$)
Total Area (ha) <5
Total Income (R$)
Total Area (ha)
1-1.9
4S2.2
<2,000 (30%)
SO,OOO
2--4.9 5-9.9
77S.7 1,017.2
2,000-14,000 (60%)
SO,000-160,000
5-20
14,000-40,000 (70%)
160,000-1,000,000
10-19.9 20-29.9 30-39.9 40-49.9 50-59.9
995.S 1,275.4 1,402.1 1,640.6
40,000-60,000 (SO%)
>1,000,000
20-50 50-100 100-"15 Regional Modules"
2,161.7
60-90
4,013.7
200-500
>90
2,126.6
500-1,000
0
'"
<1 1-2
3,000-S,000
2-5
S,000-15,000 15,000-27,500 >27,500
5-10 10-20 20-50 50-100
<10 10-100 100-200 200-500
100-200
500-2,000
1,000-2,000 2,000-5,000 5,000-10,000
fl 800'(J"W
700'(J"W
60 O'(J"W
500'(J"W
Figure 1. Rural population density (persons*km-2 ) by municipality, as counted by lEGE (2000 Cens~s: available at http://www.sidra.ibge.gov.br/cd/defaultcd2000.as~:o=2&i=P) for the st~tes bordering the Legal Braz1han A1~~ZO~; location of study areas in the Ecuadorian and Braz1han Amazon; and locatiOn ofmral settlements, protected areas, Hvers,
Total Area of Settlement (ha)
<0 0-3,000
0
0
iii
highways, capitals, and cities.
Mean Household Income (US$)
INCRA Size) Classes
FNO Size Classes
10,000-100,000 >100,000 aNorthem Ecuadorian Amazon data are fromBilsborrow et at. [2004]. Note that columns are independent of each other.
>2,000
122
BRONDIZIO ET AL.
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
academic-political use of the tenn "traditional populations," when associated with small-scale land use systems, has created more ambiguity and problems than it has resolved; this topic, while relevant to discussion of Amazonian small farmers and deforestation, is beyond the scope of this paper [DeCastro et al., 2006; Barreto-Filho, 2006]. Perhaps even more encompassing is the tenn colonist, or migranr fanner, which has been commonly used in the land use literature to describe more recent migrant groups, whose arrival often has been spontaneous or through planned government settlement projects [Brondizio, 2004; Caldas et al., 2007]. In the Brazilian Amazon, for example, the history of planned settlement and government-sponsored colonization dates back to the late nineteenth century with the fonnation of settlements along the Belem-Bragantina railroad, intended to serve as food production areas for a region dedicated to the expansion of the extractivist rubber economy. Following the decline of rubber as a significant export commodity around 1910, small waves of migration have influenced the occupation ofthe region around cities such as Santarem and Belem, and to smaller extent, every other state in the region, though this migration was largely sparse compared to later waves of migration. Beginning in the 1930s, Japanese colonists in the state of Para spearheaded a new wave of small-scale land use production systems, and subsequently, Japanese settlers and their descendants experimented new fonns of intensive small-scale land use systems. Their experimentation with various crops and means of cultivation led to their status, by the 1960s, as one ofthe top world producers ofblack pepper. Even with a sharp decline in black pepper production during the late 1960s as a result from both Fusarium disease and decreasing world market prices, Japanese colonists, such as the community of Tome-Ayu continued to innovate by ushering in new fonns of intensive agroforestry systems that focused on high-value fruit production, coupled with fruit processing plants for export. Cooperatives such as Cooperativa Agricola Mista de Tome Ayu (CAMJA) have shown the potential of intensive small-scale production coupled with transformation industries and export networks to aggregate value to local production [Yamada, 1999]. Besides black pepper and native fruits, in the Lower Amazon, the Japanese community also cultivated jute (Corchorus capsularis), a vegetal fiber that was brought to Amazonia by a Japanese agronomist. Jute production in Amazonia quickly grew, and Amazonia became one of the main international producers of this crop for more than 40 years until the end of the 1970s when jute production in Brazil started to decline due to competition from Asian producers [Gentil, 1988; Winkerprins, 2006]. Following the decline in production, most of the Japanese families left the lower Amazon region and migrated to other regions of Brazil, mainly to capital
cities, though jute still continues to be cultivated by local populations. While some regions of the Amazon, such as around Santarem in the lower Amazon, experienced significant waves of migration during the 1950s and 1960s, particularly from the northeast of Brazil, the rate of migration of small fanners dramatically increased after 1970 as a result of gov-. ernment-sponsored programs that selected families on the basis of origin, age, and composition to settle in various parts of Amazonia [Moran, 1981]. This process was particularly marked along road systems that were constructed as part of different national integration and colonization schemes, including areas along the Transamazon Highway in the state of Para, in the state of Rondonia, and later in the state of Acre; these changes were dramatic, in some cases, doubling the population moving to rural areas on a yearly basis. However, a significant part of these populations that moved to rural areas and colonization settlements in the 1970s soon became disregarded and abandoned without adequate support, passable roads, and service infrastructure in rural areas. Many ended up moving to urban areas and emerging nearby urban centers [Browder and Godfrey, 1997], eventually leading to an increased process of lot turnover that has characterized much of the failure and disregard of colonization areas of the region and the paradox of land aggregation in areas of agrarian refonn [Ludewigs et al., 2009; Campari, 2002]. As in the aforementioned frontier areas of Brazilian Amazonia, in NEA, land clearing (deforestation) and land use change by migrant fanners generally followed the construction of infrastructure (especially roads) laid out for oil extraction. Oil companies, starting in the early 1970s, built roads to lay pipelines to extract petroleum. Families seeking land then poured into the region, mostly from the Sierra or Highlands regions, which were characterized by considerable landlessness and land concentration along the roads. Farms of about 50 ha (250 by 2000 m) were established along roads. When all the land was taken in an area or sector along the primary roads, the next wave of settler families then settled beqind the first settlers on fanns parallel to the roads, on so-called lineas (generally 2 km) behind the farms along the roads. Eventually, fanns were established and ratified by the IERAC commonly up to the fifth or 6th linea, though some have reached a distance of up to the 14th linea. Continuing through today, the oil industry and related services have remained as a pull factor attracting migrants to the NEA. 2.2. Economic Categories
Economic and income categories also have been used to differentiate small fanners from larger landowners involved in regional agropastoral economies. (It is important to note
that social movements, shch as those represented by Pesagri, rural labor unions, "Grito da Terra," and a variety of local and regional organi,i'ations, tend to use historical class distinction to diffi?:retiate patronage (larger landowners, local elites) and the al labor force (including small land holders). Some oft e strongest social movements, however, are associated with ancestry and historical land use systems, such as the case of the rubber tappers movement, which emerged under the leadership of rural union leader Chico Mendes but evolved to represent a way of life and economy based on the historical association between fanner, or extractivist, and forest. The church, particularly through movements associated with pastoral committees for land, has played an important role contributing to the organization of small fanners in cooperatives and producers' associations. The region today houses a variety of producers' associations and confederations aiming at uniting different sectors ofrural populations.) Overall, 72.3% of fanns in Brazilian Amazonia have a total income of less than US$1500, 22% are between US$1500 and US$4000, and less than 2% is above US$7500 [INCRAI FAO, 2000] (based on exchange rate ofUS$l/R$2). In the Ecuadorian Amazon, mean income ranges from US$482 (fann lots <2 ha) to US$4000 (fann lots 60-90 ha). In particular, credit programs have typically defined producers on the basis of income as a proxy for assigning different fanners and group of fanners to different credit programs (Table 1). For example, the Fundo Constitucional de Financiamento do Norte program (FNO), controlled by the Amazonian Bank (BASA), defines categories of fanners according to income, despite using a questionable range of income groups to represent, or potentially misrepresent, the reality ofrural households in the region and the way different groups benefit from credit. Income, specifically gross annual production (GAP), is used to aggregate fanners into four categories: (1) minifanners, (2) small-scale fanners, (3) medium-scale fanners, and (4) large fanners. However, the GAP used to define each category seems distant from the reality of most families in rural areas. Mini-scale fanners, for instance, have a GAP corresponding to less than US$13,800 per year, while small-scale fanners are defined between US$13,800 and US$27,600 per year. Medium-scale fanners are defined between US$27,600 and US$I72,400 per year and largescale above that [BASA, 2002, 2004]. According to INCRA and Brazilian Institute of Geography and Statistics (lEGE) data [INCRAIFAO, 2000], the average family income (renda total) for the North region is around R$2904 (approxim!,ltely US$ 1540), while the fann income (renda do estabelecimentoY averagesR$1935 (approximately US$1025). For "large" owners (patronagem), income is R$11,883 and R$9691, respectively (approximately US$6300 and US$5137). It is not surprising, then, that BASA has characterized the distribu-
123
tion of FNO credit as overwhelmingly beneficial for miniand small-scale fanners (62% of these categories receiving funds), while in reality, groups receiving credit also include medium and large fanners. Field surveys [Brondizio, 2004] indicate that the rate of credit acquisition by colonists varies significantly according to the availability and conditions of government credit programs. However, over the course of 30 years of settlement along the Transamazon Highway, we found that around 56% of households interviewed received credit at least once. Annual rate of acquisition, however, is less than 10%. These figures are significantly lower for the Santarem-Belterra region. A recent survey (2007/2008) of riverine fanners in the Amazon estuary shows that over 90% of fanners have never received credit for land use activities, though they represent the most important sector producing ayai fruit for regional and external markets [Brondizio, 2009]. For all of these regions, however, since 2006, there has been a slight increase in small credit loans, typically less than US$500, from the FNO program. These loans have been given for activities such as weeding or maintenance of existing fields. However, these programs have lacked any sort of assistance or monitoring and have functioned more as small aid grants than an agricultural credit involving monitoring and assistance. 2.3. Land Holding/and Land Clearing Categories
Fann size categbries have been used in a variety of ways to represent groups of fanners. For categorizing small fanners, the range of fann sizes varies widely in different parts of the region and should thus be viewed in relative tenns. While peri-urban lots and areas with a long history of settlement have tended to contain fann lots varying from 1 to 50 ha, lots granted within colonization areas generally have ranged from 50 and 150 ha~ though in some cases have included lots larger than 400 ha usually designated as "glebas." Overall, the average area of family-based fanns in the Brazilian Amazon is 57 ha, while the average area of large fanns is 1009 ha (compared to the Brazilian average of 26 and 433 ha, respectively) [Guanziroli et al., 2001]. In some cases, however, families settled in extractive reserves may have access to significant larger areas usually combining private and cornmonly held land. Sizes of agricultural clearings, derived from spatial assessments of deforestation, such as Instituto Nacional de Pesquisas Espaciais, Sao Paulo, Brazil's (INPE's) Programa de Calculo do Desflorestamento da Amazonia (PRODES) project [INPEIPRODES, 2003], have also been used as a proxy to defIne categories of fanners. Although summary data from INPE-PRODES aggregates deforested areas from 1 to 15 ha and from 15 to 50 ha, these intervals have been
124
I
!Ij
I
j
I
II
j·i
II
II !
BRONDIZIO ET AL.
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
used, by the media for instance, to describe clearing done al., 2002; McCracken et al., 1999; Deadman et al., 2004; by "small farmers." Field data, however, have indicated that Siqueira et al., 2007; Lim et al., 2002; Evans et al., 2001]. most small farmers tend to deforest areas (i.e., annual defor- For the Uruara study area, we built upon a study of small estation e*nts) smaller than 5 ha and, on average, lack the properties, typically one or more 100 ha lots (the average capacity (i.e., labor and capital) to clear larger areas, usually property size in the sample is 133 ha), located an appreciclearing 0.5 to 3 ha per year [Brondizio et al., 2002]. Yet for able distance (approximately 18 km) from the Transamazon the purpose of our analysis, particularly at a regional, ag- Highway. This study used household surveys to analyze gregated level, size of clearings provides an acceptable com- market access and land process (i.e., distance from the main parative indicator to understand the composition of events highway), household dependency (i.e., the number of indimaking up regional deforestation. In lieu ofbetter data, these viduals not engaged in farm work, namely, children, women, assessments provide a proxy to compare the contribution of and elderly individuals), family household structure (i.e., counts of individuals in age-sex cohorts), amount of hired "small" and "large" farmers to regional deforestation. labor employed (in person-days or diarias), age of the household head in years, access and use of agricultural credit for 3. METHODS farm activities, and length of time (in years) the household Our analysis is based on field observations and bib- has been living on the property. Overall, farming systems liographic data from different areas of the Amazon Ba- are highly diversified across annuals crops (53% of the studsin, including colonization settlements along the BR-230 ied properties adopted some type of annual system, such as Transamazon Highway (Altamira, Brasil Novo, Medicililn- rice, beans, and com), perennial crops (72% of the studied dia, Uruani) and the BR-163 (Santarem, Belterra); rural properties adopted some type of perennial system, including communities in the Amazon estuary (Ponta de Pedras) in the coffee, cocoa, and pepper), and pasture (95% of the studied state of Para; colonization areas in the Brazilian states of properties raised cattle). Pasture is the dominant land use Acre (Porto Acre) and Rondonia (Machadinho, Anari); and and averages about 23 ha per property, as is common for colonization areas in NEA. Besides variation in soil qual- Amazonian colonists [see Walker, 2003]. For the Santarem and Belterra sites, we studied 5953 ity and forest types, these sites have different colonization farm-lots using a stratified random sample of 244 and 401 and demographic histories. Our analysis also includes field in 2001 and 2003, respectively, and community-level studobservations and bibliography concerning small holders livies (n = 409) [D 'Antona et al., 2008; Vanwey et al., 2007]. ing within extractivist reserves [such as the Tapaj6s National Santarem region had until recently the largest proportion of Forest (FIona Tapajos) near Santarem] and peri-urban farmsecondary vegetation, and lower adoption of perennial crops ers such as around Altamira and Santarem, where we have (36%) and cattle ranching (46%), but 77% of the farms prodeveloped field research. The Altamira, Brasil Novo, Medicililndia, and Uruara duced annual crops. Located approximately 400 km from the state capital Porto study sites were part of a large colonization and settlement Velho, the research sites of Machadinho and Vale do Anari program enacted by the INCRA, beginning in the early have different histories and settlement designs. Comparative 1970s with the opening of the Transamazon Highway. In studies of these sites included longitudinal socioeconomic the Altamira, Brasil Novo, and Medicililndia study areas, we surveys, settlement level institutional analysis, ecological built upon studies that comprised of 371~ farm-lots along field studies and vegetation inventories, multitemporal rethe Transamazon Highway and feeder roads (travess5es) usmote s~nsing, and landscape fragmentation analysis [Baing a stratified random sample of 402 and 399 farms dur2001]. The Machadinho settlement started with an tistella, ing 1997 and 2005, respectively, with a subsample of 171 in 2 with 2934 plots designated to receive small area of2090 km 2001. These studies include detailed sociodemographic and economic surveys of households stratified by time in the re- farmer colonists from other states. In 1988, Machadinho begion. The research team collected a set of data related to land came a municipality and expanded to incorporate four other use allocation over time, labor, technology, market, credit, settlements and small towns. In 1989, the rural population and use of forest resources, and a set related to family re- represented two thirds of Machadinho's total population. productive and migratory histories, socioeconomic and labor Less than 10 years later, the rural population diminished to arrangements. On average, farm activities are quite diverse one third of the total. Vale do Anari started as a spontaneous in Altamira, with farmers producing annual crops (50% of settlement, but was later established as a planned colonizastudied properties), cacao and other perennials (35%), cat- tion area by INCRA in the early 1980s and, in 1994, betle (95%), and horticultural activities (40%) [Siqueira et al., came a municipality. As other colonization areas, it lacked 2003; Moran et al., 2005; Vanwey et al., 2007; Brondizio et assistance, urban infrastructure, or administrative autonomy.
These dynamics led the tWo settlements to have different institutional arrangements, rules of forest use, and outcomes in terms of the int~ctions between the colonists and the environment [Bati/tella et al., 2003]. The majority of the colonists in thisytea came from the south of Brazil, mainly the State of Parana, bringing with them specific production systems. The result, in terms of spatial organization of farming plots, is 11 mosaic of pasturelands, perennial crops, mainly coffee and cocoa, and annual crops (com, rice, and beans). Land cover characteristics are also defined by different stages of land occupation and secondary succession contrasting with the native rain forest [Batistella et al., 2003]. For the Amazon estuary site in the municipality of Ponta de Pedras (State of Para), we have conducted longitudinal ethnography and surveys of households (n = 143) and communities (n = 6) differentiated by economic and institutional histories and land tenure [Brondizio, 2008]. Estuarine communities studied here have over 95% adoption of as;ai agroforestry and a decreasing rate of annual crops such as manioc, and abandonment of pasture areas and mechanized agriculture implemented during the 1970s and 1980s by external development projects. The Humaita study site was part ofthe Humaita settlement in the State of Acre that was implemented in 1981 to subdivide rubber extraction estate into 948 lots. This site was surveyed in 2003 and 2004 (n = 98 farms) [Ludewigs, 2006]. The Acre site resembles the Transamazon Highway study region, in that, it has experienced a high rate of adoption of pasture (95%) but lower rates ofperennial (42%) and annual (48%) crops. The NEA study site is located in a settlement area that also has been considered an area of high biological and cultural diversity [Myers, 1990; Orme et al., 2005]. Similar to many settlements in Brazilian Amazonia, Ecuadorian farms approved by the IERAC were mostly homogenous in size, each approximately 50 ha, and were rapidly acquired, in part, to circumvent the creation of two large national parks and protected areas in the region and the provision of communal land to indigenous communities, most of whom had lived in the region for centuries in a seminomadic situation. This meant that the continuation of in-migration after 1990, when most of the nontitled areas had already been occupied by colonist families, led to a process of farm subdivision of the original fincas, resulting in even smaller farms. Thus, the average farm size decreased from 46.5 ha in 1990 to only 25.5 ha in 1999 [Bilsborrow et al., 2004]. We used data from longitudinal household surveys administered in 1990 and' 1999. The survey is based on a statistically representative probability sample of 470 farm plots selected in 1990 using two-stage sampling, with lists of settlement areas with the number of farms and the total area in each constituting the
125
sample frame. In the first stage, 64 cooperatives or areas of colonization (sectors) were selected with probabilities proportional to size. In the second stage, the number of farms randomly selected from each sample sector was based on the size of the sector (called in-sampling, probabilities proportional to estimated size). Questionnaires were administered to each household head and spouse covering, together, their economic situation in the previous residence, land tenure, and acquisition in the NEA, land use, agricultural production and technology, work on and off the farm, credit, household composition, migration, fertility, health, dwelling quality, household assets, contacts with local communities for services, etc. The number of households increased greatly between 1990 and 1999, due to both subdivisions among heirs and sales to new in-migrants continuing to come to the region in search of land. To derive deforestation estimates at the state, regional, and farm levels for each of these areas, we used multitemporal remote sensing data and analyses. For settlement and farm level analysis, research groups provided estimates for their respective study area. Data for Brazilian Amazonia was based on 2003 PRODES data from INPE and was collected for the states of Acre, Para, and Rondonia [INPEIPRODES, 2003], all of which have relatively large populations ofsmall farmers. PRODES ?ata was aggregated into four classes: deforestation, forest,water, and other (e.g., clouds and older cleared areas) and ianalyzed using ArcGIS (Environmental Systems Research 1nstitute (ESRI), Redlands, CA) to assess the number and size of clearings (i.e., polygons of deforestation in the PRODES data) and the amount and percent of deforested area (i.e., area of each clearing event). Each of the cases reported in this chapter benefit from a long list of publications, part of which is cited here. It is not our intention to develop new statistical analysis of primary data, but to refer to published work as we seek to understand similarities and differences among small farmers across the region. 4. EXAMINING FACTORS AND TEMPORAL PATTERNS ASSOCIATED WITH SMALL FARMER DEFORESTATION IN AMAZONIA 4.1. Size ofClearings as Indicator ofDeforestation at State and Regional Levels
Comparing state level deforestation in Brazil for 2003 using data on size of clearings [INPE-PRODES, 2003], we found that for the states of Para, Acre, and Rondonia, small clearings (e.g., up to 20 ha) are the most frequent size classes of deforestation, comprising approximately 88.1 % of the total number of forest clearings in Acre, 74.0% of the total
126
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
Bro~
,",<0
(4):::R 4)
p.,
~
OJ)
'g i:::l
~ "Cl
......
«l
'0
"Cl
i:::l
~
ilu
<
r-r-<')'
» u
OO .......
t-t.n
r-:
....... t- N
.............. , 0\ 0
g.~ <')O\onon'
0
...... o o 0
......
....
»
~
'0000\ ...... '
4)
......
u
1il
p.,
g.
4)"
....'"'
u
~(Ij~ (4):::R
,",<0 4)
~
.......
~~ o\N . ,........'O<')<')'
."j'
~
V("'f')("t')""=:tV)
..................... vvr-r-oo or) 0\ ..0 N" M' r-" 0\ M'
~ p., ~
§
-<'"'
oqt--;C?oq"'1"":"'!lI")
IOf--O"'IN ..............
U
'S 'S '0 0
0\ 0
ero 4)
~
"Cl
~""';""';odo~o
'Or-on<')N'
'4"
ro
Nr-'
'-'
'-'
p.,
If)"' l£)"'
...q'" ...........
~
0\"'
N
NlOt-O\ .............. VO
~lI")t--;O')oqC?",!C?
.............. N O O ............... O
0\ 0
......
[/)
~
<')r-r-'
~
~
~ p., <') 0 0 N
'g
.S
U
'" .!::'l
ero ~
OJ)
'ro
[/)
p.,
1il
i:::l
......
» u
'0<') 00 0\0 '0 r-o ~ ~~ ...... '
u
~
0 &~ oOv-lNv-lV)""';OO ...... V)
~ "0 p.,
»
u
i:::l '.0
~
'
4)
......
g.
ro ......
....'"'
'" 4)
5cd~
4)
~
,",<0
(4):::R
"Cl
§
4)
'-'
p.,
'"i:::l0
OJ)
0
OJ)
.§ 'g ~ '" u '
4)
u
'"'
4)
p.,
"Cl
§
» u i:::l
4)
g. 4)
....'"'
N QI ......
~
... ~
('f")"'
...f'
rr) ......."' ...........
N
\O.....-tON ........ ('f)OOO
r-oor-oooO\O\o
..tC"ir.-i""':ddvid 00 0
......
<
......
........ O\lOOO("t')t-~~
e~ro
oo'
<
~~~~~::eVloo",
4)
0
E I1.l
..q-"'
...... on
.Q
0
.......
....
0
4)
.....-t
p.,~
0
~ '+-<
\O("'f')
M'M'
4)
4)
p., i:::l
VlOV ....... O\ .......
-DC"ivir.-ioO'-:vio; <')r-<')o<')'
» U
i:::l 4) ~~ u
'"' &~ p.,~ .... 4)
» u i:::l 4)
g. ....'"' 4)
ON.....-4ffl ........ V)O.... o::::t...
On <')" 'On 'On OOn 0\"
f:!
~
............
0\01.000\.0-00
oq"'!"'1"'1~"":"C?
onNr-N ...... OOO r- ...... 0
......
("'f')\C)l()(""I("f")('"l')
........
(1')
ooo\NNON<')oo O\'ONr-on ...... N
NM'N N
0 <')
e~ '" '" 4)
~
4)
.!::' l [/)
0 ...... ooog~
:3~~:3~~N"Fl
number of clearings in Para, and 73.2% of the total number of clearings in Rondonia. However, when considering the area deforested, we found that these same clearings contributed only 7.6%, 3.2%, and 2.1 %, respectively, of the total deforestation for these states (Table 2). The larger clearings (i.e., greater than 2000 ha) comprised only approximately 0.10%, 0.47%, and 0.40% of the total number/events of clearings in Acre, Para, and Rondonia, respectively, but they accounted for 86.0%, 91.2%, and 94.5% of the total deforested area, respectively. This result indicates that large clearings are the most predominant features of deforestation in each state. The inclusion of the state of Mato Grosso would certainly reinforce this pattern. These large clearing~ may consist of clearings oflarge areas on one or several adjacent properties creating one large area of deforestation. Across all three Brazilian states, we found that the total number of clearings below 20 ha contributed to a small proportion of the total area deforested. Acre had the largest number of small clearings (e.g., number of clearings less than 20 ha), which contributed to a larger amount of cleared area relative to Para and Rondonia for similar clearing size classes. However, both Para and Rondonia had a greater number of large clearings (e.g., greater than 2000 ha), which contributed to an overall larger, cleared area than in Acre. While the contribution of small farmers to total regional deforestation may be relatively small, their contribution varies across states and within states and, as such, has different environmental consequences. The large number of clearing events, albeit relatively small in total area, has important implications for environmental changes, depending on their landscape context. In colonization areas for instance, settlement design and institutional arrangement contribute to the cumulative spatial pattern of forest cover and distribution of deforestation [Batistella, 2001; Batistella et al., 2003). Depending on the property design of particular settlements, isolated small clearings can add up to represent large-scale environmental change. Furthermore, the diversity of land use systems and the role and intensity of external pressures makes it difficult to generalize the contribution ofsmall farmers to the regional environment as a whole. In many cases, one finds productive interactions between the agro-ecology and the spatial-temporal arrangements of local production systems and their landscapes. At the same token, under particular contexts, they can impact wildlife habitat and population, contribute to resource depletion and soil erosion, and the spread of accidental fires to forests [Toniolo, 2004; Sorrensen, 2004]. According to the FAO, Ecuador had the highest rate of deforestation in South America in the last two decades [FAG, 2001,2005], and within Ecuador, the NEA is the second most active deforestation front after the Choc6 region in the coastal
BRONDIZIO ET AL.
province of Esmeraldas [Sierra, 2000). Although deforestation rates are decreasing in NEA, from 2.5% cleared per year between 1986 ~1\.d 1996 and 1.8% per annum in 19962002, these rat:1,are still comparably high [Mena et al., 2006b). While;ne most important agent of deforestation in the Ecuadotian Amazon has been the small farmer, and smallholding agriculture has been the main process affecting forests, the impact of other processes on forest ecosystems in NEA should not be overlooked; for example, two large agro-industrial projects, starting in 1973 when two corporations were given land titles to establish African palm plantations, have deforested a combined area of roughly 20,000 ha [Santos and Messina, 2008], the equivalent of 400 farms or about 1.8% of the colonization area [colonization area calculated from the areas of precooperatives of settlement in the Aguarico Zone and Coca Zone of Instituto Ecuatoriana de Reforma Agraria y Colonizaci6n (IERAC)] in the NEA. When comparing deforestation across colonization settlements in different parts of the region, one finds that interregional differences in forest clearing rates are closely related to age and history of the settlement (Figure 2). For example, the Santarem-Belterra region already experienced deforestation before the 1970s, but experienced hIgh rates of conversion of forests to agropastoral uses during the period between1973 and 1979, followed by widespread secondary regrowth. After this period, migration to the area and rates of forest conversion stayed relatively lower until recent expansion of soybeans, starting around 1999. In Altamira, one observes pulses of deforestation coinciding with rates of migration and lot occupation (1973-1979) followed by periods
Porto Acre, AC
of decline (1985-1993) and subsequent expansion of cattle ranching and deforestation such as during the 1997-2003 periods. The Acre site experienced spikes in deforestation both during the late 1980s and after the mid-1990s. In all cases, it is important to observe the variability of deforestation rates among farm lots within the same settlement. In the Amazon estuary, however, one sees an opposite trend in deforestation associated with the intensification of small-scale agroforestry systems by riverine farmers and the existence of a diversified forest and river-based resource economies closely engaged with regional and global markets [PinedoVasquez and Padoch, 2009; Pinedo-Vasquez et al., 2001; Smith et al., 2007; Brondizio, 2008]. For more than two decades, the estuary has been undergoing a "forest transition" associated with the decline of annual crops and the rise of forest products, a process which has simultaneously led to land use intensification and population increase in urban and rural areas [Winklerprins, 2002; Padoch et al., 2008; Costa and Brondizio, 2009]. 4.2. Farm Size as a Variable FaIm size, in particular, has played an important role in the land use allocation strategies of farmers and the distribution of deforestatipn events over time. (Classes of private property size are controversially treated across the literature and within goyernment agencies and programs. Several levels of details ahd typologies are available and informed our organization of size and categories. We consulted websites and publications from government agencies such as the
Santarem, PA
g 100
Altamira, PA
100
100
80
80
80
~
60
60
60
Q
40
40
40
20
20
'0
>-l >.
on
'tl
<S .,
] >-l
~ ~ Po<
20 0 I/')
<0
0)
0)
~
~
....
M .... ~
<0
,
co ~
~ ~ ~ N
0) 0)
~
<0
0) ~
<0
0) 0)
,
~
'"~ 0)
'" 0 0
~
0) 0)
0 N ....
0
<0
0)
0)
~
~
....
,
0
0)
'0)
~
127
<0
<0
0)
,
co
N
0) 0) ~
~
<0
0)
0)
~
~
0
'"00
~
0) 0)
~o
~~
1II~
....
....
,
~
I/')
0)
r:::
0) ~
0) 0)
,
<0 ....
~
I/')
<0
0)
a <0
~
2~ ~ c;;
0)
<0
0) 0) ~
N 0)
0)
~
<0 ~
Time Period
Figure 2. Percentage of land (%) deforested by farm propecty lots in agricultural settlements in Porto Acre, Santarem, and Altamira, Brazil (boxes correspond to average values, comprehended within quartiles for each time period).
'" 0 0
....~ 0) 0) ~
128
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
IBGE and INCRA, government programs such as FNO, and publications from the Brazilian nongovernment organization IMAZON.) While the percentage of deforestation corresponds n~gatively with farm size (i.e., the larger the farm, the smaller the percentage deforested), absolute deforestation (i.e., total deforested area in hectares) is positively correlated to farm size. In absolute terms, the contribution by small farmers to regional deforestation pales in relation to the contribution by large farmers (Table 2). However, this relationship varies for differel'J.t parts of the region. 4.2.1. Santarem example. The estimated deforestation rate and the factors explaining this rate are sensitive to the unit of measurement, particularly when they are assessed in relation to property size. Figure 3 illustrates these relationships for the Santarem region of Brazilian Amazonia. Whereas small farmers (e.g., farm properties that are 200 ha or less) tend to have higher proportions of areas in use, the absolute size of deforested area is small when compared to large farmers (e.g., farm properties larger than 200 ha), despite the fact that the deforested area represents a smaller percentage of a largeholder's property. Analyzing private properties of different sizes in the Santarem-Belterra region, we found that, in absolute terms, while small farmer deforestation events tend not to exceed 5 ha, largeholders deforested areas ranging from 10 to 500 ha. From 1986 to 1999 in the area examined, smallholders together deforested approximately 1641 ha of land, while largeholders together deforested 6064 ha. Similar patterns can also be found in terms of areas maintained in use and clearing of secondary succession. Smallholders maintained up to 82% of their property in use, while largeholders had less than 5% in use. In absolute terms, smallholders had between 0 and 50 ha of area in production, while largeholders had up to 100 ha of area, which was not a substantially different absolute amount of land in production relative to property size. However, this relationship is changing with the consolidation and expansion, of soybean production and other large-scale mechanized crops [D 'Antona et al., 2006]. 4.2.2. Uruara example. Table 3 illustrates. deforestation by year and by aggregated property size classes. Because of the distribution of properties in this particular site, the research group made a distinction only between large glebas of 3000 ha (n = 9) and all other property sizes (n = 3263), the vast majority of which were 100 ha. The few intermediatesized farms (glebas), those with approximately 400 ha, are included here in the small farmer count, since there are cases where a family own multiple adjacent lots of 100 ha. They found that, in general, the amount of deforested land in the two size classes increased over time. The degree of
increment, however, was different, with large glebas adding about 5 km2 of cleared land over the 13-year period, from 33.7 to 38.6 km2 • Deforestation associated with these highly capitalized interests occurred early in the colonization period during the 1970s and has been rather static since that time. Smallholder deforestation has steadily increased, more than doubling from 447.1 to 1048.1 km2 of cleared area over the same period. This difference, however, represents the disproportional number of small farmers analyzed (n = 3263) vis-a-vis large land holdings (n = 9). Because the recent immigration and settlement of smallholders are not significant in the study area within the bounds of the cadastral map used for this analysis, large amounts of deforestation have not been caused by further settlement. 4.2.3. Ecuadorian example. In Ecuadorian Amazonia, patterns of land use were clearly visible according to the duration (e.g., years of settlement) and size of farm (Figure 4) [Barbieri et al., 2005]. The research group, which includes some of the authors here, found a decrease of forest through time across all farm sizes, though among the smaller farms (0-25 ha), the decrease offorest cover is highest in the early years of settlement and clearing. The latter mostly refers to the secondary wave of deforestation that occurred in the study region in the 1990s linked to property subdivision. Meanwhile, the proportion of the cleared area in pasture increases with duration of settlement, while the cleared area in perennial and annual crops together increases only slightly over time, then decreases slightly, reflecting some replacement of crops by pasture over time on medium and larger farm sizes. As mentioned before, the extensive subdivision of the original farms or fincas madres since 1990 has led to a second wave of deforestation [Bi/sborrow et al., 2004; Barbieri et al., 2005; Pan and Bi/sborrow, 2005]. Thus, the original farms containing no subdivision still had, in average, 56.1% of their total area covered by forests in 1999, while farms with two and three or more subdivisions had only 47% and 32% in, forests, respectively [Pan and Bi/sborrow, 2005]. Because of the processes of settlement that occurred in the Ecuadorian Amazon, virtually all of the patches of deforestation are small compared to those in the Brazilian Amazon, and with the subdivision process since 1990, cleared patches are even smaller. Figure 5 shows how most deforestation in the NEA has occurred in very small patches of 1 to 5 ha. 4.3. Summarizing Variables Explaining Deforestation Among Small Farmers
The complexity of factors underlying land use decisions among small farmers defies any simplistic or linear expla-
BRONDIZIO ET AL. lOO,-----------------_~
'1
,)80 60
*
* * * * * *
* * * * I *
'"' ** ***
I
*
*
40
*
* land-Cover Classes
~
*
*#
20
_Areos in Use •
c
"
~
~
.,,2....I..rr.....~U. .!.L--.J •
0 .L--'-+"'~4.-.-l...+.::Jc...J...II::e.__ up to 10
20 up 1050
10 up to 20
100 up to 200
50 up to 100
Deforestation 55 Deforestation Forest
500 up to 2000
200 up to 500
2000 above
closs of size property
500 450 400 350 300 250 200 150
Land-Cover Classes
*
100 0'
50
~
0
,£
«
_Areos In Use •
up to 10
20 up to 50
10 up ta 20
100 up to 200
50 up to 100
-=--'--l-L--l • 500 up 10 2000
200 up la 500
Deforestation 55 Deforestation Forest
2000 above
Figure 3. Land cover change (area and percentage offami lots) according to farm-lot size classes for the BR-163-Santarem-Belterra Region (State of PanI) between 1986 and 1999. Land cover includes area maintained in use areas deforested from secondary succession vegetation (SS), and areas defo~ested from mature forest. '
129
130
BRONDIZIO ET AL.
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
Table 3. Deforested Land and Annual Deforestation Rates in Uruani, Para, From 1986 to 1998 by Size Class and Year Property Size
1986 Area (km2)
1988 Area (km2)
Large fami'S (Glebas) (n = 9)::::: 3,000 ha
33.7
29.9
All other (n = 3,263)
447.1
578.5
1991 Area (km2 )
Deforestation Ratea (% a-I)
-3.2 3.3
Deforestation Rate" (% a-I)
1999 Area (km2 )
Deforestation Rate a (% a-I)
36.4
7.2
38.6
0.7
750.6
9.9
1,048.1
4.9
apercentage rates are annualized for the periods 1986-1988, 1988-1991, and 1991-1999. Source is Aldrich et al. [2006].
I
nation of deforestation, which,' nevertheless is unfortunately common with this topic. Although we highlight the importance of specific variables, one should remember that small farmers make land use decisions in a multidimensional way. In other words, a decision to deforest may represent at the same time a reaction to a market opportunity (e.g., price of beef or a crop), a way of increasing land value and property legitimacy (e.g., deforestation as a proof of "use"), and/or a step toward forming a farm. The rate, timing, and form of deforestation, thus, will be influenced by different conditions and needs of farm families, such as their economic and social expectations, household size and composition, knowledge of forest resources, previous experiences and preferences, technology and capital available, and location of the farm in terms of distance and accessibility. Differ-
ent processes, demographic, socioeconomic, cultural, and environmental, are at work and interact at different spatial and temporal scales to promote different land use strategies [Kaimowitz and Angelsen, 1998; Wood and Porro, 2002; Perz, 2001; Brondizio, 2006]. The determinants of deforestation include both household factors and exogenous factors, many of which are regionand country-specific, including the pressure of commodity markets and national policies, such as incentives for agropastoral expansion aiming at export, oil related investment, and conservation efforts. Table 4 illustrates the importance of different variables explaining deforestation among small farmers, particularly those in colonization areas, and the studies examining the importance of these variables. In sum, simplistic analysis of causality provides an ill picture of these
F.rms 28 to 80 h. 80.0,--------------------,
Fanne 0 to 28 ha 80,0,--------------------,
............. _------_ .. _-----------
-g
... 80,0
60.0
~
40,0
'5
-~-
If. 20.0
.................. 0,0
........ _-- ...-..
~
------- ..
4-8
...................................~~.~.~~~~~.~.~.~~::.:.:.:==--
7-9
10-12
13-15
16-18
40.0
'l;
11-
20,0
. .....................
.............-
__~-_,____--,------~~~----1
+-,----=--,-----~--~--,-------,------~--1
0-3
----------------
§
-
.!!
0.0 +-_~
>19
7·9
4-6
0-3
Years of settlement
...:.F..::.nn:::.::.8:::.mo::.re~lh::.:.:.:::.n..::50:..:.h=-.
80,0 ,---
- ......
------ ... -
~
40.0 'l; If. 20,0 0.0
.
.,
•••••
+.,---,------,-------,------~-~--~-~
0-3
4-6
7-9
10-12
13-15
.".
... 60.0
.... _-- ........... _- .....
......................
....... n"
13-15
A:..::II:..:.r.::.rm:::...:::siZ::.8.=---
80,0,-
---,
--------- .. _-- .. _---- .... _------
.".
... 60,0
10-12
16-18
>19
Yea.. or eelliement
16-16
>19
~
40,0 'l; If. 20.0
_- ---------
-.
- ------ -- ------ -- -- --- - - - ----- - - ----- - - ----
-
0,0 L--~~_ _~-_,____-~--~-~----I 7-9 10-12 13-15 16-18 >19 4-6
v.... or ..tU8m8nl
Years of settlement
1- cropslperenllials ..... _. pasture -
- rarest
1
Figure 4. Patterns of land use according to duration (years of settlement) by size of farm in the Ecuadorian Amazon. Reprinted from Barbieri et al. [2005], with kind permission from Springer Science and Business Media.
131
economic reasons), farmers, large and small, may choose to invest in land clearing and deforestation to increase the value of their property or to speculate in areas expected to be targeted for agropastoral expansion. In the process, they may ad~pt a combination of annual crops followed by pasture, WhICh can help to maximize short-term return and minimize risks (i.e., receive return from crops, open new opportunities with pasture, and increase the market value of the farm) [Vosti et aI., 2003]. For instance, for more than two decades cattle ranching has been widely adopted by small farmer~ in Brazil as a strategy to secure fast returns, to facilitate cooperation with neighbors on raising and expanding herds, to minimize risks associated with storage and dependency on transportation (of perishable crops), and to improve their ability to negotiate and sell their livestock to a wide and diverse group of buyers [Hecht, 1993]. Other attributes are also important, such as level of wealth [Pichon, 1997; Alston et al., 2000; Walker et al., 2002; Alston et aI., 1993] and length of residence on the property [Pichon, 1997; Walker et al., 2002; Vanwey et al., 2007]. For example, the time of settlement in the NEA, as in Brazilian Amazonia, is an important factor in land clearing, with less clearing on more recently settled land [Pan et al., 2004,2007]. Both Pan et al. [2007] and Barbieri et al. [2005] note that the relationship between time since settlement and deforestation is not1a strictly causal relationship, since it also o indicates the locati,on in time and space of different cohorts 2 4 5 6 7 8 9 16 30 of migrant colonists. Thus, more recent cohorts have to setDeforested Patch Size (ha) tle farther from rOflds and towns. In the NEA, for instance, older farms, closer: to roads and towns, experienced the most Figure 5. Forest patch size frequency (a) and percent of total deforpopulation growth in 1990-1999 (Figure 6), but less deforestation (%) (b) within sampled farms. estation than the more recent cohorts, reflecting the faster pace of deforestation on farms settled more recently, since relationships. Thus, we do not want to convene the idea that they are in the early stages of settlement. a single variable can explain deforestation, but illustrate the Incorporating demographic variables to explain the derelative role of specific factors influencing the process. forestation in the region, we note that family size [Pichon, For example, according to different studies, besides age 1997; Pichon et al., 2002], number of men in the household of the farm and time of settlement, distance to market repre- [Walker et al., 2002; Pan and Bilsborrow, 2005; Caldas sents one of the most important variables explaining defor- et al., 2007; Sydenstricker Neto and Vosti, 1993], and level estation among small farmers, particularly those settled in of dependency [Walker et al., 2002] have an impact on either colonization areas. Several studies have shown that distance agricultural systems or in the amount of land to be deforto markets have a negative impact on deforestation [Pichon, ested. Marquette [1998] and Barbieri et al. [2005] also note 1997; Walker et aI., 2002; Caldas et al., 2007]. In many the important effects of the family life cycle and household studies, market factors are used as a proxy for distance to ~pe in land use change, particularly according to the Chayomarket in explaining deforestation; however, these studies, VIan consumer/labor ratio, which was found to be important in attempting to model individual agents, may fail to capture in the colonization area [Marquette, 1998], including in the the use of natural resources on the landscape and the role of' buffer area of the Cuyabeno Wildlife Reserve [Mena et al., local markets (e.g., cattle commercialization between neigh- 2006a]. bors). Land market is also a key factor explaining deforestaran and Bilsborrow [2005] studied the determinants of tion,i in some cases, independent of distance and location. land use in 1999 (shares of each farm in four different forms Since cleared land has higher market value (for legal and of land use, in forests, in perennial crops such as coffee, in
132
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
BRONDIZIO ET AL.
301.===:;::::;:;:::::::::::::;---------------,80
Table 4. Summary of Variables Influencing Deforestation Among Small Farmers Type of Correlation"
Examples From the Literature
Age, length of residency, and famil~ife cycle
+
Access to credit and number of credit acts Househ91d labor and number of men
+
McCracken et al. [1999], Moran et al. [2002], Qrondizio et al. [2002], Walker et al. [2000], Pichon [1997], Caldas et al. [2007], Vanwey et al. [2007], Perz [2001], Perz and Walker [2002] Walker et al. [2000], Pichon et al. [2002], Ludewigs et al. [2009], Tura and Costa [2000] McCracken et al. [2002], Futemma and Brondizio [2003], PichOn [1997], Sydenstricker Neto and Vosti [1993], Walker et al. [2002], Pan and Bilsborrow [2005], Caldas et al. [2007] Pichon et al. [2002], Walker et al. [2000, 2002] Pichon et al. [2002], PichOn [1997] Walker et al. [2000], D 'Antona et al. [2006]
Independent Variable
_Pop 1990 ..... Pop 1999 -
Hired labor Mechanization and chainsaw Lot size Pasture (%)
+ + + Relative deforestation Absolute deforestation +
+
Deforested (%) Lot turn over (number of owners) and lot aggregation Distance to markets
+ +
Topography (flatness) Soil quality and water availability Level of wealth and income
+
Land security
±
+
+
Moran et al. [2002], McCracken et al. [1999], Walker et al. [2000], Pichon [1997], Ludewigs [2006], Murphy [2001] McCracken et al. [1999], Vanwey et al. [2007] Ludewigs [2006], Ludewigs et al. [2009], Vanwey et al. [2007], D 'Antona et al. [2006] Pichon [1997], Moran et al. [2002], McCracken et al. [1999], Caldas et al. [2007], Walker et al. [2002] McCracken et al. [1999], PichOn [1997], Pan et al. [2004] Caldas et al. [2007], Moran et al. [:2002], Pichon [1997] Alston et al. [1993], Jones et al. [1995], PichOn [1997], Murphy [2001] Futemma and Brondizio [2003], Alston et al. [2000], PichOn [1997], Toniolo [2004]
"Plus sign indicates positive correlation; minus sign indicates negative correlation.
annual crops, and in pasture) at the household level, based upon a multiresponse linear model. Their results reveal the most powerful determinants to be plot size, plot access to road and the nearest community, years living on plot, household labor availability, especially males, and population density on the plot. It is striking that population density is a powerful factor even when plot size and all demographic variables are included, which provides strOl,lg support for the important, independent effects of population pressure. Yet, data from the Amazon estuary shows that afforestation can occur simultaneously with population increase in urban and rural areas. The estuarine region (over 20 municipalities in Brazil) has witnessed a forest transition and rates of deforestation close to zero, due to the expansion of agroforestry-based a9ai fruit production and a forest economy, which involves a variety of timber and nontimber resources. In addition to expanding national and global markets for these products, estuarine farmers are culturally familiar and knowledgeable about forest management and resources, which has allowed them to respond to market opportunities using local management technologies and multicropping agroforestry systems [Brondizio, 2008; Jarvis et a!., 2007;
133
Rerkasem and Pinedo-Vasquez, 2007; Pinedo-Vasquez and Padoch, 2009; Brookfield, 2001; Padoch and PinedoVasquez, 2006]. Ironically, these same systems are often regarded as backward and unproductive. Some studies have observed that large deforested areas often appear on properties that have families with. substantial family labor resources, including hired labor [Pichon, 1997; Walker et al., 2002; Pan and Bilsborrow, 2005]. In Ecuador, as in Brazil, deforestation within colonization areas also relates to the duration of residence on the property, education level, a~d age ofhousehold head [Pichon, 1997; Alston et aI., presented paper, 1993]. Families with longer periods of residency have deforested larger areas; however, the type of forest used also varies with duration of residence, in that, families with longer settlement histories tend to eventually use and clear secondary forests, compared to more recent settlers who clear remaining primary forest areas [Brondizio et al., 2002; Perz and Walker, 2002]. Environmental and resource constraints also structure the amount and type of deforestation, particularly incombination with each other. For example, Pan et al., [2004] show that landscape complexity and fragmentation, two important
25
. --Cily(km)
Forest 1990
/~
-...... Forest 1999
...
,/ ".".
"" ""
60
.~;"""
_ _ '/"7..-.... ~_&-.
--...."...:.. ... --;;.--..-........ - .. ,....", ,'" "...
0'"
............................... -- ....
-----_..
". "
'
..
,'.
I------I._---_._~
5
20
O+--------.------+-----t------+-----+O <1970
1970-74
1975-79
1980-84
1985-89
11me (years)
Figure 6. Pop 1990, finca population in 1990; Pop 1999, Finca population in 1999; Forest 1990, the percent of forest cover on a finca in 1990; Forest 1999, the percent offorest cover on a finca in 1999; Nearest City (km), Euclidean distance from thefinca to the nearest ofthe four major communities in 1999. Reprinted from Pan et al. [2007], with kind permission from Springer Science and Business Media.
measures within Landscape Ecology that have implications for the flux of matter and energy in ecosystems, are associated with household size and composition, expansion of the road and electrical networks (increasing access of farm households), year of plot settlement, and topography. This indicates that in the Northern Ecuadorian Amazon, the spatial arrangement of agricultural plots and forested patches within the landscape have strong connections with changing socioeconomic processes. Soil quality also directly affects land allocation [Moran et al., 2002] and, in combination with other factors such as topography and water, can influence type of crop adoption and deforestation in different ways [McCracken et al., 1999]. Pan et al. [2004], for instance, state that flat land is important to annual crops and pastures and may be preferred for clearing. Caldas et al. [2007] show that soils suitable for pasture and with water can positively affect the total area deforested. However, their results are not sustained when spatial autocorrelation analyses are implemented. In short, several factors related to the rapid population growth, plot subdivision, plot location/accessibility, and resource availability all have contributed to the conversion of forest to crops and pasture for cattle, which in tum has created a more complex and fragmented landscape. Even though demographic and environmental factor~ are implicated as drivers of deforestation, credit and land secu-' rity, are also important. That is, forest conservation is positively associated with land security [Pichon, 1997; Alston et al., 2000; Fearnside, 2001]. Nevertheless, there are disagreements regarding this view. Walker et al. [2000] argue
that land security can be a facilitator in credit acquisition, and consequently, can be used for pasture formation. Finally, it is infreasingly recognized, but little studied, that rapid urbanization associated with adjacent processes of lot turnover an~ land speculation in agrarian settlements shape the spatial pattern and rate of deforestation for years to come. For example, the main urban centers in the NEA have had high rates of population growth and constitute key centerpoints or poles of development. Several studies have found proximity to these towns and their markets' important factors contributing to deforestation aiming at increasing land value [Mena, 2001; Bilsborrow et al., 2004; Pan et al., 2004; Barbieri and Carr, 2005; Pan and Bilsborrow, 2005]. However, as the estuary example above illustrates, this relationship depends on the perceived value (e.g., water protection, recreation, symbolic meaning, storage ofresources) and market for forest resources. In some cases, market forces can promote a relationship between farmers and forests inverse to that described by the Ecuadorian case above or other colonization areas in Brazil [Brondizio, 2008, 2009]. It is also important to note that colonist farmers understand better the importance of forests over time. Most farmers tend to set aside forest areas dedicated to protecting water sources and hunting grounds and to have access to various raw materials needed for daily farm operations, such as wood, fibers, and roof covers [Muchagata, 1997; Brondizio et al., 2002; Campos, 2006]. Furthermore, recent studies have called attention to the growing interdependency between rural and urban populations that rely on forest resources for their production
134
SMALL FARMERS AND DEFORESTATION IN AMAZONIA BRONDIZIO ET AL.
and consumption needs, but also rely on city services for health, education, and commerce. This pattern of rur~l-urban connections and interdependency represents a growmg r~al ity for thell"Amazon, within which small farmers playa vItal role [Padoch et at., 2008].
farmers, especially in the Brazilian Amazon have minimum access to extension service and technology: 5.7% of ~arm ers use extension services, 9.3% of farmers have electrIC en3 70/ offarmers use mechanized implements and other " d 87 lo/c ergy, . /0 forms of technologies (such as ammal tractIOn), an. ... ° of farmers depend on manual labor for land use actiVIties. 5. SMALL FARMERS, LAND USE, These regional numbers corroborate household surveys"c~r AND DEFORESTATION: MOVING ried out in the sites illustrated here. Furthe~ore, the d:sBEYOND MISCONCEPTIONS mantling" ofthe Empresa de Assi.ste~cia Te~mca e Extensao Rural (EMATER) extension servIce m BraZIl after 1990 has As we stated in the beginning of this chapter, small farm- contributed to the lack of assistance and support to sm~ll ers form diverse social groups in Amazonia, but share m~ny farmers. Although its support varies across states, and ~n widespread misconceptions about themselves and the regIOn, spite of the effort of many agricultural extension agents,. I~ as described by Schmink and Wood [1992, p. 6]: (1) small ost of Amazonia they do not have conditions even to VISIt m , . , rt d producers are not efficient; (2) peasant people are culturall'y local farmers, or as one extension agent m Para repo e , retrograde; (3) extractive activities are ~ackward; (4) tra.dI- "EMATER now-a-days is a 'dead-alive' (vivo morto), that tional knowledge is worthless; (5) troplc~l ~orests prOVIde I'S here but without a working phone line, gasoline and trans., and technicians to attend to farmers ., requests" few useful economic goods, with only a hmlted num?er ~f portation hardwoods; and (6) community property rights a~e antIthetI- [Brondizto, 2004, field notes]. Yet, it ~s usually among cal to private property. Such develop.ment paradIgms cre.ate small farmers that we find the most creatIve land use so.luinherent barriers to protecting the enVIronment,.to prese~mg tions, such as planting consortiums, bett~r land preparation the boundaries of Indian lands, and to defendmg the. nghts methods, land use diversification, and a hIgh ~egree o.f agr?of small farmers [Schmink and Wood, 1992:6]. In thI~ co~ b·10 d'IVersi'ty . Colonist farmers tend to expenment WIth. dlf. t t an interpretation of small farmers and deforestatIon m ferent methods of land use management, usually combmmg ex, I" f . 1 Amazonia should be concerned with the po ItICS 0 regIOna techniques brought from other regions. development, interest groups and the distribution of ecoDespite having smaller property sizes, limited access to nomic incentives, the role of external for~es, a~~, not le.ast, technology, and technical assistance, small f~rmers \smallthe views of development and what constItu~es modernIza- holders) compare positively to lar~eholders m BraZIl. Action" put forward by different sectors of soc~ety. B~low, we cording to some studies, in Amazoma, small farmers have an discuss three common misconceptions assOCIated WIth ~mall average annual income ofR$52/ha (approximat~ly.uS$29), farming land use systems and deforestation in Amazoma. almost five times more than the largeholders, whIle I~ southern Brazil, smallholders have an average annual mcome f R$2411ha (approximately US$128) and largeholders an 5.1. Misconception 1: Small Farmers Have Bac":'~rd and annual income of R$991ha (approximately US$53) L an d Use S:ystems Associated With Low ProductIVIty d . Extensive Deforestation and Subsistence Pro uctlOn [Guanziroli et at., 2001]. Dif~er~nt exam~les of small-scale, highly productive systems eXIst m the regIOn, such as among Small farm land use varies from int~nsive to extensive agroforestry farmers of Tome-Ac;;u, cocoa farmers of the methods, including sophisticated agricultural. sys~ems com- Transamazon, ac;;ai farmers throughout the Amazon estua.ry, bining indigenous technology, as well as hIgh mput pro- horticultural farmers around large urban centers, and ma~IOc duction systems [Brondizio, 2004; Cos!a et at., 2006]. In farmers in many areas who use multicropping and multIvacolonization areas, land use evolves WIth the. age and ex- riety systems. Small farmers throughout the region are highly engaged periences of farmers in the region. Sma~l f~rmmg la~d uses include various forms of swidden cultivation, hortIculture in market dynamics, responding to price changes and new and polyculture, intensive agrofore~try, forest management market opportunities, while combining household consumpand extractivism, and cattle ranchmg. Access to technol- tion and commercialization. Though many crops are ~ro ogy is a recurrent problem among small farmers who o~te~ duced for familial consumption, such as manioc, beans, nce, have to rely on the use of fire and manual ~ools that h~It com, and ac;;ai fruit, depending on the region, th.es~ cr.ops their ability to change their land use strategIes, even a~Id also have been produced exclusively for commerclahzatI~n. perceived problems such as extended drought [BrondlzlO, Important commodity products for small farmers have m2004; Brondizio and Moran, 2008; Costa, 2006]. Data from cluded not only manioc, rice, beans, and com, ?ut also black INCRA and IBGE [Guanziroli et at., 2001] show that small pepper, coffee, cocoa, a great diversity of fruItS and seeds,
~verage
•
.~.
.-
--§-
•
135
horticultural products, timber and nontimber forest products, husbandry, and fish f~ing. Yet, in Brazil, and according of small farmers with shifting agriculture, which, in tum, to some studies, although small farmers have 25% of the is considered by nature as destructive and unproductive. It overall agriculturll,l'credit, family properties are responsi- is common place to find interpretations of small-scale proble for approximttely 37.9% of total national production, at duction systems using shifting cultivation as the main conleast part of w~h is for export. For example, almost 20% tributors to global deforestation and the main threaten to of smallholders in Brazil sell more than 90% of their total biodiversity [Primack and Corlett, 2005; Palm et at., 2005; production, while close to another 40% sell more than 50% Hartshorn, 2006]. While the contribution of shifting cultiof their production. At the national level, smallholders are vation to deforestation varies enormously across the world, responsible for the production of 52% of milk from cattle, Amazonian small farmers are commonly lumped together 58% of pork, 40% of chicken and eggs, and the majority of as part of this global category [Netting, 1993]. This kind of fresh produce used for daily consumption [Guanziroli et at., analysis, while raising some important issues and relevant to 2001]. However, small farmers and small farming groups particular parts of the world, rely mostly on a general stereoface significant constraints in transportation and the abil- type applied to shifting cultivation systems [Dove, 1983], ity to negotiate prices, despite their engagement in global rather than examining how they vary in time and space, in market systems and organization in various forms of coop- some cases, actually contributing to increased habitats and eratives and associations. They tend to pay higher transpor- biodiversity [Pinedo-Vasquez et al., 2002]. Another comtation prices to supply food to local urban markets and, in mon example of this type of discourse has been illustrated most cases, have to carry their products on their own backs by statements from researchers and technicians from federal or in buses, on truck rides, on mules, and on bicycles. Many, research agencies who argue that the lack of technology if not most, depend on negotiations with middleman to sell makes small farmers the main agent of deforestation, due to their products, which make their economic return per unit low productivity. For instance, "smallholders deforest to eat, produced lower, independent oftheir success and productiv- that is, to grow com, rice and beans, and afterwards tum the ity as farmers [Brondizio et at., 2003]. Successful local pro- land into pasture," while they view that "it is wrong" to say grams to facilitate producers to sell directly in urban areas, that largeholders are responsible for deforestation because such as the case of Altamira and Santarem, are limited and they have technolo,gy that allows them to have higher prolack wider support. Furthermore, the lack of transformation duction in a smallet area. They point out that low technology industries to process agricultural goods and natural resources means low level of occupation, in terms of number of cows within the region condemns farmers to sell their products as per hectare (1.2 h~adlha with low production, and up to 2 unprocessed or semiprocessed raw material, which perpetu- headslha with higqer technology input), but fail to consider the variability of level of productivity, the diversity of land ates value added concentrated outside the region. use systems employed by small farmers, the level of support they receive, the history of conflicts associated with these re5.2. Misconception 2: Small Farmers Contribute to gions, and the role of speculative deforestation among large Amazonian Deforestation as Much as Large Farmers farmers and loggers [Costa, 2004]. Some studies have addressed the relative role offarm size While the contribution of small farmers to Amazonian and economic scale upon regional deforestation and demondeforestation varies across countries, states, subregions, strated a lesser role of smallholders in driving deforestation and periods, in aggregated terms, small farmers contribute in Amazonia relative to large holders, contrary to common to a small proportion of regional deforestation area, when generalizations about small farmers. For example, in a study compared to their large-scale counterparts (Table 2). Durof the Brazilian portion of the Amazon basin, Fearnside ing the 1990s, INPE's deforestation assessments, including [1993] estimated that large-scale holders were responsible frequency of clearing by size class, were often used in the for 70% of all deforestation in 1990 and 1991, while Walker media to finger-point the causes of deforestation toward et al. [2000] showed that the relative degree of accountabilsmall farmers. Even today, this has continued to be the case, ity is spatially variable. In a microregion in the south ofPani, although small farmers are currently being treated as part where the influence of the Agency for the Development of of an aggregate category under INCRA [among numerous the Amazon (SUDAM) was strong in the 1970s and 1980s, examples see OESP, 2008a, 2008b; Fotha Online, 2008]. to all of the land cleared large enterprises accounted for close Several authors point out smallholders as the main defor(in 1986), while along the Transamazon Highway, in the viesters, while not considering their relative contribution viscini,ty of the Uruani area, the proportion was only 8% (in a-vis other sectors and the units of analysis used to make 1992). Location and spatial organization of settlement areas comparisons. A common problem is the general association also are significant in explaining patterns of deforestation.
BRONDIZIO ET AL.
136 SMALL FARMERS AND DEFORESTATION IN AMAZONIA
Analysis of deforestation in relation to farm size thus needs to account for both absolute (e.g., ha) and relativ~ (e.g., %) measures of deforestation. In general, we see an mverse relationship lJJetween these two measures, in ~elation ~o farm size, though these relationships also v~ry wIth locatIon and context history of occupation, and social group. A gr~at technological and economic divide e~ists .between groups -of small and large farmers in Amazon~a. FIel~ da~a collected among rural families in the Santarem regIOn m 2001 and 2002 indicate that more than 90% ~f the fa~ers depend on axes, shovels, and machetes f~r theIr pro~uctIOn, whereas large-scale producers use machmes, chemIcal fertilizers, and pesticides, mostly subsidized by bank loan.s. The percentage of small farmers who were able to obtam (applied for and received) credit from 1998 to 2002 w~s lower than 5%, compared to 80% among lar~e ~armers m this study area. The INCRA data confirm that l~mIted acce~s to technology technical assistance, and finanCIal support IS widely found ~mong small farmers all over.Br~zil, especially in the north and northern regions [Guanzzroll et al., 2001]. On the other hand, riverine farmers of the Amazon .estuary have been able to use their knowledge of multicroppmg systems to reach levels of intensification higher than any ,oth~r regional production system and hav~ transforme~ altaI fruIt agroforestry systems into the mos~ Important land use and economy activity in the region WIthout a~y help ~om .research and government agencies [BrondizlO and Szquezra, 1997' Brondizio, 2008]. De~pite their economic and te~hnological. disadvantages, hich limit their ability to intenSIfy productIOn on the land, :mall farmers have been able to keep some portio~ ofmature forest within their properties (rather than extensIvely clearing properties to increase production area). The research team reporting on the Santarem region found that on small private properties ranging in size from 0.9 to 200 ha, forest cover is approximately 41.89 ha and, on.average, ~m~ll farmers maintain approximately 60% of ~heir propertIes m
forest. However, by aggregating small farmers, we found that 85.5% of their land is kept in forest, as opposed to l&rge farmers that maintain only 70.7% in forest (Table 5). The latter is a result of higher deforestation rates among large farmers, making their relative contribution to total deforestation significantly higher in this region. 5.3. Misconception 3: Small Farmers, Particularly. Colonist Farmers, Follow an Inexorable Path ofDeforestation Unless Curbed by Government Actions
The extent, amount, and trajectory of deforestation va.ry significantly at the farm lot level, depending on charactenstics such as time in the region, knowledge of forest resources and views of forest as productive land, stage of farm formation household demographics, capital, short- and longterm ~oals, and market opportunities for diffe~ent crops (see Table 4 for a detailed list of variables). The hIgh rate of adoption of cattle ranching among small .farmers correlates to risk minimization strategies, aggregatIOn of land value, and economic incentives to adopt pasture and cattle ranching as land use strategies. At the same time, small farmers adopt diverse land use strategies, including ann~al and perennial crops, different forms of husbandry, fis?mg and hunting, and a variety of off-farm and sharecroppmg labor arrangements. . . Secondary forest areas are widely used m farmmg syste~s among small farmers. Data from the Tra~samazon reg!on of Altamira-MediciHindia and BR-163 regIOn of SantaremBelterra indicate that among colonists, phases of lot formation lead to increased use of fallow land, and older, settled farmers tend to use more secondary forests than mature forests [Brondizio et al., 2002]. These results also show that between 1986 and 1999, older, settled farmers cleared more secondary forests than mature forests, compared to recent-settled smallholders and largeholders (Figure 3)..Ev~n small farmers with less than 10 ha were still able to mamtam
Table 5. Absolute and Relative Distribution of Land Cover Changes in the Santarem-Belterra Region, Para, From 1986 to 1999 According to Different Property Sizes: Small and Large Property Size <200ha >200 ha <200 ha >200 ha
N
Areas in Use
Clearance of Secondary Forest
1,823.00 37.00
683.64 4,829.31
610.11 4,745.25
1,823.00 37.00
2.60 6.30
Deforestation of Mature Forest
Regeneration
Mature Forest
Total
Absolute Area, has 1,641.33 6,063.93
907.38 6,363.18
22,823.73 53,800.65
26,700.03 76.068.36
Relative Distribution Area, % 2.30 6.10 6.20 8.00
3.40 8.40
"Aggregated for all properties within each class.
85.50 70.70
100.00 100.00
areas of forest, albeit smaller than areas of secondary vegetation. The majority of small farmers with lot sizes up to 10 ha maintain at least ~5% of the land in forest, while those with 10 to 20 ha l1}iiintain approximately 40% or more, and most farmers of 2/J to 50 ha lots maintain more than 50% in forest. HoweveP(as not~d above, one finds significant variation within settlements and between subregions of Brazilian Amazonia. Time in the region also relates to a great appreciation and knowledge of the economic and ecosystem services of forest areas and to forest conservation. However, market opportunities can be strong enough to motivate deforestation or reforestation independent of time in the lot and available technology. During the 1990s, for instance, farmers along the Transamazon increased deforestation to form pastureland during a period of high prices for beef and declining prices for cocoa. On the other hand, farmers in the Amazon estuary have virtually abandoned deforestation and annual crops during the past two decades in favor of forest management and agroforestry systems for regional, national, and international markets. However, along the Transamazon, besides shifting prices for beef and cocoa, rates ofpasture expansion have been influenced by an active land market formally or informally sanctioned by INCRA. In summary, small farmers tend to decide their deforestation strategies based on external and internal conditions during different phases oftheir farm operation and family life cycle, but particularly market opportunities for land and commodities.
137
in addition to the lack of access to services such as education and health,. and lack of support for local entrepreneurship. All these factors lead to a high rate oflot tum over and increase the impoverishment of small-scale farm~rs. One of our comparative studies involving three settlement in the states ofPani and Acre indicates a rate oflot turnover around 75% over the life of these settlements, which illustrates not only the active land market within agrarian settlements in Brazil, which returned to land concentration under the eyes of agencies such as INCRA, but also the scale of challenges, economic, cultural and social, infrastructural, faced by small farmers [Ludewigs et al., 2009]. Small farmers have been disregarded already very early in the developmental project of Amazonia [Moran, 1981; Wood and Schmink, 1979] and continue to be so today amid new settlement projects, even by policies targeted at small-scale production systems. Perhaps, the current precarious situation of EMATER offices throughout Amazonia illustrates also the situation of the population they aim at serving. Even in areas of active economy involving small farmers, such as the altai fruit economy of the Amazon estuary, rural households depend mostly on retirement income of family members and government aid such as the bolsa familia (family-aid) [Brondizio, 2009]. Lack of governance in the Amazon frontier has been cited as one of the main problems regarding deforestation [Nepstad et al., 2002]. The disregard of small farmers ih relation to credit, extension services, technology, transpprtation, and access to markets tends to create and promote;: land speculation and a vicious cycle of 6. CONCLUSION selling small lots to largeholders and moving to new frontier forest land and urban areas. This is an active and ongoing It is important to understand small farmers in the context situation throughout Amazonia. More integrated policies are of the political ecology of deforestation and conservation needed, which incorporate effective agrarian programs (e.g., in Amazonia, hence, breaking old assumptions that rural technology, extension services, credit, and support for comcommunities are homogeneous and adapted or are failed mercialization), legal and institutional infrastructure (e.g., adaptations to external environments. This is an old but re- land titling, legal definition of rights of resource use, and current discussion in Latin America [Durham, 1988; Rose- monitoring and sanctions on forest clearing), and socioculberry, 1993]. The development paradigm guiding public tural recognition (e.g., valorization of forests as productive policy since the 1960s has led policy makers to condemn land; valorization of small-scale production systems). Simismall-scale production systems as transitory and inefficient larly, conservation policies should include and promote diwithout addressing the problems faced by rural popula- verse local systems of production without trying to "freeze" tions. Most public policy programs have a tendency to frame and "essentialize" smallholders as "traditional populations" small-scale production as only related to family consump- expected to protect forests with disregard of their economic tion needs, while failing to consider their contribution and needs. Perhaps most important, small farmers, and the region wider economic potential, thus missing the opportunity to as a whole, would benefit from policies aiming at promoting support more inclusive forms of regional development. In value aggregation of agricultural and forest resources, which spite of political discourse and numerous public policy pro- . could facilitate commercialization to the advantage of programs, this is still the case in Amazonia. ducers, generate employment in urban and rural areas, and Amazonia parallels other regions of Latin America where gen~rate revenue to be reinvested at the municipal scale. In a development model intersects with rural impoverishment, other words, small farmers would benefit from policies fosenvironmental problems, and inequality in land distribution tering the development of transformative industries within
138
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
the region, which involves the participation of producers. External subsidies to promote conservation-development programs in the region have proved at best to be tra~sitory in the past. C~dits for carbon or subsidies for protectI~n of environmental services will run into similar problems If empty of mechanisms for economic participation and valorization of regional resources. Small farmers form a sizable population in Amazonia and represent an important form of employment for a lar~e c.ontingent of people that otherwise have little or no op~lOn ill a region with limited transformative industries .and h~le e.mployment outside the informal economy. ~helf contrlbutIo.n to food production for regional consumptlOn and export IS undeniable and growing. Their presence in rural and urban areas through various forms ofsocial and economic networks, and their presence in virtually all nonindigenous reserves of Brazilian Amazonia indicate their central role in the overall development and governance of the region. The quali~ and seriousness of public policies concerning the economIC and social needs of small farmers will continue to influence their land use systems and role on regional land use and deforestation. More attention and fewer stereotypes will contribute to improve their condition and reconcile their eco~omic contribution and environmental footprint in Amazoma. Note added in proof. As this paper goes to press, IBGE [2009b] released a report (2 October 2009) confirming the overwhelming importance of small farmers for food production and security in Brazil (e.g., 70% beans, 87% manioc, and 58% milk consumed nationally) and rural employment (employing 75% of the rural labor force). Confirming the analysis presented in this paper, the report shows small f~ ers producing more in less area. Collected for the first tIme as part of a national-level census (2006), these dat~ c~nfirm our arguments in favor of the social and economic I~por tance of small-scale production systems at local and natIOnal levels and reaffirm our call to overcome m~sconceptions and the invisibility of small farmers in Brazil. . Acknowledgments. We thank the U.S. National Aeronautics and Space Administration (NASA) LBA-ECO Program (grants NCC5-334, NCC-695, NNG06GD86A), NSF HSD Program (grant BCS0527578), and NICHD (grant HD 358llO-0~/07) for supporting research at Indiana University's A~thropologlcal Center for Training and Research on Global EnVironmental Change (ACT). We also thank NASA LBA-ECO for supporting the proje~ts (NNG06GD96A) '''Spatially Explicit Land Cover Econometrics and Integration with Climate Prediction: Scenarios of Future Landscapes and Land-Climate Interactions" an~ th~ project (N~C5-694) "ABasin-Scale Econometric Model for ProJectmg Amazoman Landscapes," and the National Science Foundation for s~pporting t~e project "Patterns and Processes ofLandscape Change In the BrazIl-
BRONDIZIO ET AL.
ian Amazon: A Longitudinal, Comparative Analysis of Smallholder Land Use Decision-Making-BCS137020 both at Michigan S.tate University and which provided support to Mar~ellus M. Cald~s. The Ecuador Project acknowledges the U.S. NatIOnal Aeronautics and Space Administration (Grants NCC5-295 and Earth Science Fellowship NNG04GR12H), the National Institutes ofHealt~ (R~ 1HD38777-0l), and the Carolina Population Center, at the Un~verslty of North Carolina at Chapel Hill, as well as the collaboratIOn and support of the Ministries of Environment and Agriculture and the former National Planning Agency (CONADE). The GTZ Profors program and the Compton and Summit foundations also supported the 1999 survey fieldwork. We thank FAPESP for providing PostDoctoral Scholarship (n. 01/02578-2 and 03/01933-9) and FAPESP Research Financial Support (n. 01/11473-0) to Celia Futemma. We thank EMBRAPA-CPATU in Belem, EMBRAPA-NMA in Campinas and EMBRAPA offices in Santarem, Altamira, and Belterra, the'LBA Program, particularly the LBA office in Santare~, Para State Brazil EMATER offices in different parts of the regIOn, the office'in Belem. We also thank colleagues at Indiana UniverIP sity's Department of Anthropology and ACT particularly Andrea D. Siqueira, Vonnie Peischl, Scott Hetrick, Linda B~rchet, and a~l ACT graduate students and interviewers involved With these ~roJ ects. A special thank you goes to all Amazonian farmers who kin~ly received us on many occasions. Finally, we are grateful for the msightful comments and suggestions of two external reviewers and the editors of this volume.
AM
REFERENCES Adams, C., R. S. S. Murrieta, W. A. Neves, and M. Harris (Eds.) (2008), Amazonian Historical Peasants: Invisibility in a Changing Environment, Springer, Dordretch, The Netherlands. Aldrich, S., R. T. Walker, E. Y. Arima, M. Caldas, J. ~rowder a~d S. G. Perz (2006), Land-cover and land-use change In the BraZilian Amazon: Smallholders, ranchers, and frontier stratification, Econ. Geog~,82,265-288. . Alston, L. 1., G. D. Libecap, and B. Mueller (2000), Property nghts to land and land reform: Legal inconsistencies and the sources of violent conflict in the Brazilian Amazon, J. Environ. Econ. Manage.,39,162-l88. . Araujo, R. (1993), La cite domestique: Strategies familiale.s et I~~ ginaire social sur un front de colonisation en Amazome bresIlienne these de Doctorat, Nanterre-Paris X. Barbieri: A. F., and D. L. Carr (2005), Gender-specific outcmigration, deforestation and urbanization in the Ecuadorian Amazon, Global Planet. Change, 47, 99-110. Barbieri, A. F., R. E. Bilsborrow, and W. K. Pan (2005), Farm household lifecycles and land use in the Ecuadorian Amazon, Populo Environ., 27,1-27. . Barreto, P., C. Souza Jr., A. Anderson, R. Salomao, and 1. Wiles (2005), Pressao Humana no Bioma Amazonia, in 0 Estado da Amazonia, No.3, 6 pp, Imazon, Belem. (Available at http:// www.imazon.org.br)
Barreto-Filho, H. (2006), Popula~6es tradicionais: introdu~ao a critica da ecologia politica de uma na~ao, in Sociedades Caboclas Amazonicas: Moaernidade e Invisibilidade, edited by. C. Adams, R. S. S. M/I;;'ieta, and W. A. Neves, pp. 125-144, AnaBlume, Sao Paule{ Barsky, O. (1984)"ta Reforma Agraria Ecuatoriana, Corporaci6n Editora Nacional, Quito, Ecuador. BASA (Bank of Amazonia) (2002), 0 Fundo Constitucional de Financiamento do Norte e 0 Desenvolvimento da Amazonia, M & S Editora, Belem, Para, Brazil. BASA (Bank of AmazoniaY (2004), Relatorio das Atividades Desenvolvidas e dos Resultados Obtidos no Exercicio 2003, with Fundo Nacional de Financiamento do Norte, Gerencia de Estudos Economicos e Rela~6es Institucionais, and Coordenadoria de Planejamento, Belem, Para, Brazil. Batistella, M. (2001), Landscape Change and Land-UseILandCover Dynamics in Rondonia, Brazilian Amazon, CIPEC Dissertation Series, No.7, Center for the Study of Institutions, Population, and Environmental Change, Indiana University, Bloomington, IN. Batistella, M., and E. F. Moran (2005), Human dimensions of land use and land cover in the Amazon: A contribution for LBA, Acta Amazonica, 35(2), 249-257. Batistella, M., S. Robeson and E. F. Moran (2003), Settlement design, forest fragmentation, and landscape change in Rondonia, Amazonia, Photogramm. Eng. Remote Sens., 69(7), 850-812. Batistella, M., E. F. Moran, and D. S. Alves (Orgs.) (2008), Amazonia: Natureza e Sociedade em Transformar;Go, 1st ed., Editora da Universidade de Sao Paulo, Sao Paulo. Bilsborrow, R. E., A. F. Barbieri, and W. K. Pan (2004), Changes in population and land use over time in the Ecuadorian Amazon, Acta Amazonica, 34, 635-647. Brondizio, E. S. (2004), Agricultural intensification, economic identity, and shared invisibility in Amazonian peasantry: Caboclos and colonists in comparative perspective, Cult. Agric., 26, 1-24. Brondizio, E. S. (2006), Landscapes of the past, footprints of the future: Historical ecology and the analysis of land use change in the Amazon, in Time and Complexity in Historical Ecology: Studies in the Neotropical Lowlands, edited by W. Balee and C. Erikson, pp. 365-405, Columbia Univ. Press, New York. Brondizio, E. S. (2008), The Amazonian Caboclo and the Ar;ai palm: Forest Farmers in the Global Market, Advances in Economic Botany Monograph Series, 403 pp., N. Y. Bot. Garden Press, Bronx, N. Y. Brondizio, E. S. (2009), Forest resources, family networks and the municipal disconnect: Examining recurrent underdevelopment in the Amazon estuary, in Development and Conservation of the Amazonian Floodplains: The Decade Past and the Decade Ahead, edited by M. Pinedo-Vasquez et al., N. Y. Bot. Garden Press, Bronx, N. Y., in press. Brondizio, E. S., and E. F. Moran (2008), Human dimensions of . climate change: the vulnerability of small farmers in the Amazon, Philos. Trans. R. Soc. Ser. B, 363, l803-l809doi:1O.l098/ rstb.2007.0025. Brondizio, E. S., and A. D. Siqueira (1997), From extractivist to forest farmers: Changing concepts of agricultural intensificatio'n
139
and peasantry in the Amazon estuary, Res. Econ. Anthropol., 18, 233-279. Brondizio, E. S., S. D. McCracken, E. F. Moran, A. D. Siqueira, D. R. Nelson, and C. Rodriguez-Pedraza (2002), The colonist footprint: Toward a conceptual framework of land use and deforestation trajectories among small farmers in the Amazonian frontier, in Deforestation and Land Use in the Amazon, edited by C. H. Wood and R. Porro, pp. 133-161, University Press of Florida, Gainesville, FL. Brondizio, E. S., C. C. M. Safar, and A. D. Siqueira (2003), The urban market of A~ai fruit (Euterpe oleracea Mart.) and rural land use change: Ethnographic insights into the role of price and land tenure constraining agricultural choices in the Amazon estuary, Urban Ecosyst., 6(1/2), 67-98. Brookfield, H. (Ed.) (2001), Exploring Agrodiversity, Columbia Univ. Press, New York. Browder, J. 0., and B. J. Godfrey (1997), Rainforest Cities: Urbanization, Development, and Globalization of the Brazilian Amazon, Columbia Univ. Press, New York. Browder, J. 0., M. A. Pedlowski, R. Walker, R. H. Wynne, P. M. Summers, A. Abad, N. Becerra-Cordoba, and J. Mil-Homens (2008), Revisiting theories of frontier expansion in the Brazilian Amazon: A survey of the colonist farming population in Rondonia's post-frontier, 1992-2002, World Dev., 36(8), 1469-1492. Caldas, M., R. T. Walker, E. Arima, S. Perz, C. Wood, S. Aldrich, and C. Simmons (2007), Theorizing land use and land cover change: The peasant economy of Amazonian deforestation, Ann. Assoc. Am. Geogr.,,97, 86-100. Caldas, M. M. (2008~, Settlement formation and land cover and land use change: Pi. case study in the Brazilian Amazon, Ph.D. dissertation, Michigan State University, East Lansing, MI. Campari, 1. S. (2002), Challenging the turnover hypothesis of Amazon deforestation: Evidence from colonization projects in Brazil, Doctoral dissertation, The University of Texas at Austin, Austin. Campos, M. T. (2006), From villains and victims to environmental activists: The case of Amazonian colonos, in Frontier Encounters: Indigenous Communities and Settlers in Asia and Latin America, edited by D. Geiger, International Work Group for Indigenous Affairs (IWGIA), Copenhagen, Denmark. Campos M. T., and D. C. Nepstad (2006), Smallholders, the Amazon's new conservationists, Conserv. Bioi., 20, 1553-1556. Castellanet, c., and C. Jordan (2002), Participatory Action Research in Natural Resource Management: A Critique of the Method Based on Five Years' Experience in the Transamazonica Region ofBrazil, Taylor and Francis, New York. Caviglia, 1. L. (1999), Sustainable Agriculture in Brazil: Economic Development and Deforestation, New Horizons in Environmental Economics Series, Edward Elgar, Cheltenham, U. K. Costa, F. A. (2008), Heterogeneidade Estrutural e Trajet6rias Tecno16gicas na Produ9ao Rural da Amazonia: Delineamentos para Orientar Politicas de Desenvolvimento, in M. Batistella; E. F. Moran, D. S. Alves (Org.), Amazonia: Natureza e Sociedade em Transformar;Go, 1st ed., vol. 1, pp. 137-180, Editora da Universidade de Sao Paulo, Sao Paulo.
140
SMALL FARMERS AND DEFORESTATION IN AMAZONIA
Costa, F. A., R. Hurtienne, and K. Kahwage (Orgs.) (2006), Inovar;iio e Difusiio Tecnologica da Agricultura Familiar na Amazonia, 1st ed, vol. 1,278 pp., NAEA, Belem. Costa, L. ~ (2006), Comunica9ao & Meio ambiente: A analise das campanhas de preven9ao a incendios florestais na Amazonia, Serie Teses do Nucleo de Altos Estudos Amazonicos, Editora da Universidade Federal do Para, Belem, Brazil. Costa, N., (2004), 0 desmatamento na Amazonia e os pequenos agricultores, 0 Estado de Siio Paulo, Agencia Estado, 6 May. Costa, S. M., and E. S. Brondizio (2009), Cities along the floodplains of the Brazilian Amazob, in Development and Conservation of the Amazonian Floodplains: The Decade Past and the Decade Ahead, edited by M. Pinedo-Vasquez et aI., N. Y. Bot. Garden Press, Bronx, N. Y., in press. Costa, W. M., B. Becker, and D. S. Alves (Orgs.) (2007), Dimensoes Humanas da Biosfera-Atmosfera da Amazonia, Sao Paulo: Editora da Universidade de Sao Paulo-Edusp. D'Antona, A. 0., L. K Vanwey, and C. Hayashi (2006), Property size and land cover change in the Brazilian Amazon, Populo Environ., 27(5-6), 373-396. D'Antona, A. 0., A. D. Cak, and L. K VanWey (2008), Collecting sketch maps to understand property land use and land cover in large surveys, Field Methods, 20, 66-84. Deadman, P., D. Robinson, E. Moran, and E. Brondizio (2004), Colonists household decision making and land use change in the Amazon rainforest: An agent-based simulation, Environ. Plann. B Plann. Des., 31, 693-709. DeCastro, F., A. D. Siqueira, E. S. Brondizio, and L. C. Ferreira (2006), Use and misuse of the concepts of tradition and property rights in the conservation ofnatural resources in the Atlantic Forest (Brazil), Ambient. Soc., 9(1), 23-39. Dove, M. (1983), Theories of swidden agriculture, and the political economy of ignorance, Agroforestry Systems, 1, 85-99. Durham, W. H. (1988), Political ecology and environmental destruction in Latin America, in The Social Causes ofEnvironmental Destruction in Latin America, edited by M. Painter and W. H. Durham, pp. 249-264, Univ. of Michigan Press, Ann Arbor, MI. Evans, T. P., A. Manire, F. de Castro, E. Brondizio, and S. D. McCracken (2001), A dynamic model of household decisionmaking and parcel-levelland cover change in the Eastern Amazon,Eco!. Modell., 143, 95-113. : FAO (2001), Global Forest Resource Assessment 2000, Food and Agriculture Organization, Rome, Italy. FAO (2005), Global Forest Resource Assessment 2005, Food and Agriculture Organization, Rome, Italy. Fearnside, P. M. (1986), Human Carrying Capacity ofthe Brazilian Rainforest, Columbia Univ. Press, New York. Fearnside, P. M. (1993), Deforestation in the Brazilian Amazon: The effect of population and land tenure, Ambia, 22, 537-545. Fearnside, P. M. (2001), Land-tenure issues as factors in environmental destruction in Brazilian Amazonia: The case of southern Para, World Dev., 29(8), 1361-1372. Folha Online (2008), Incra lidera lista dos 100 maiores desmatadores da Amazonia Legal, diz Meio Ambiente, By Renata Giraldi. (Available at http://wwwl.folha.uol.com.br/folha/brasiU ult96u360405.shtml)
Futemma, C., and E. S. Brondizio (2003), Land reform and land use changes in the Lower Amazon: Implications to agricultural intensification, Hum. Ecol., 31(3),369-402. Gentil, 1. (1988), A juta na agricultura de varzea na area de Santarem-Medio Amazonas, Boletim do Museu ParaenseEmilio Goeldi, 4(2), Serie Antropologia, 118-199. Guanziroli, C., A. Romeiro, A. M. Buainain, A. Sabbato, and G. Bittencourt (2001), Agricultura Familiar e Reforma Agraria no Seculo XXI, Garamond, Rio de Janeiro. Hartshorn, G. S. (2006), Understanding tropical forests, BioScience, 56(3), 264-265. Hecht, S. (1993), The logic of livestock and deforestation in Amazonia, Bioscience, 43(10), 687-695. Hiraoka, M (1992), Caboclo and Riberefio resource management in Amazonia: A review, in Conservation ofNeotropical Forests. Workingfrom Traditional Resource Use, edited by K Redford, and C. Padoch, pp. 134-157, Columbia Univ. Press, New York. IBGE (1998), Censo Agropecuario 199511996-Brasil, 94 pp., Instituto Brasileiro de Geografia e Estatistica, Rio de Janeiro. IBGE (2009a), Censo Agropecuario 2006-Brasil, Sistema SIDRA, Instituto Brasileiro de Geografia e Estatistica, Rio de Janeiro. (Available at http://www.sidra.ibge.gov.br) IBGE (2009b), Censo agropecuario 2006: Agricultura Familiar, Primeiros Resultados, Brasil, Grandes Regioes e Unidades da Federa9ao, report, 267 pp., Instituto Brasileiro de Geografi a e Estatistica, Rio de Janeiro. (Available at http://www.ibge. gov. br/home/es tatistica/economia/agropecuaria/censoagro/ agrijamiliar_2006/familia_censoagr02006.pdf) INCRA (2000), Relatorio de Atividades INCRA 30 Anos, Instituto Nacional de Colonizacao e Reforma Agraria. (Available at http://www.incra.gov.br/publicacoes/relatorios.html) INCRA (2002), Balanr;o da Reforma Agraria e da Agricultura Familiar, Instituto Nacional de Colonizacao e Reforma Agraria. (Available at http://www.incra.gov.brl) INCRAIFAO (2000), Relatorio Projeto de Cooperar;iio Tecnica INCRAIFAO "Novo retrato da agricultura familiar: 0 Brasil redescoberto," G. A. Bittencourt, and Alberto Di Sabbatto (Coordena9ao), Instituto Nacional de Colonizacao e Reforma Agraria, Brasilia. INPEIPRODES (2003), Projeto PRODES e Coordenar;iio-Geral de Observar;iio da Terra, Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos, Brazil. (Available at http://www. obt.inpe.br/prodes) Jarvis, D., C. Padoch, and H. D. Cooper (Eds.) (2007), Managing Biodiversity in Agricultural Ecosystems, Columbia Univ. Press, New York. Jones, D. W., V. H. Dale, 1. 1. Beauchamp, M. A. Pedlowski, and R. V. O'Neill (1995), Farming in Rondonia, Resour. Energy Beon.,17,155-l88. Kaimowitz, D., and A. Angelsen (1998), Economic Models of Tropical Deforestation: A Review, Center for International Forestry Research, Bogor, Indonesia. Lena, P., and A. Oliveira (Eds.) (1992), Amazonia: A Fronteira Agricola 20 Anos Depois, Edi90es CEJUP, Belem. Lim, K, P. Deadman, E. Moran, E. Brondizio, and S. McCracken (2002), Agent-based simulations of household decision making
BRONDIZIO ET AL. and land use change in: Altamira, Brazil, in Integrating Geographic Information Systems and Agent-Based Modeling Techniques for Simulatii'JC Social and Ecological Processes edited " 277-310, Oxford Univ. Press. ' by H. R. Gimblet"pp. Ludewigs, T. (200q'j, Lal}d-use decision making, uncertainty and effectiveness 9i1~nd reform in acre, Brazilian Amazon, Doctoral dissertation, Indiana University, Bloomington. Ludewigs, T., A. de O. D'Antona, E. S. Brondizio, and S. Hetrick (2009), Agrarian'structure and land use change along the lifespan of three colonization areas in the Brazilian Amazon, World Dev., 37(9), 1348-1359. Marquardt, K A. (2008), Burning Changes: Action Research with Farmers and Swidden Agriculture in the Upper Amazon, Doctoral thesis No. 2008:42, Faculty of Natural Resources and Agricultural Sciences, Swedish Agricultural University (SLU), Uppsala, Sweden. Marquette, C. (1998), Land use patterns among small farmer settlers in the northeastern Ecuadorian Amazon, Human Ecol., 26, 573-598. McCracken, S. D., E. S. Brondizio, D. Nelson, E. F. Moran, A. D. Siqueira, and C. Rodriguez-Pedraza (1999), Remote sensing and GIS at farm property level: Demography and deforestation in the Brazilian Amazon, Photogramm. Eng. Remote Sens., 65, 1311-1320. McCracken, S., A. D. Siqueira, E. F. Moran, and E. S: Brondizio (2002), Land use patterns on an agricultural frontier in Brazil; Insights and examples from a demographic perspective, in Deforestation and Land Use in the Amazon, edited by C. Wood and R. Porro, pp. 162-192, Univ. Press of Florida, Gainesville. Mena, C. (2001), Deforestation in the Napa Basin: Socioeconomic Factors, Spatial Patterns, and Metrics, Florida International University, Miami, FL. Mena, C., A. F. Barbieri, S. 1. Walsh, C. M. Erlien, F. L. Holt, and R. E. Bilsborrow (2006a), Pressure on the Cuyabeno Wildlife Reserve: Development and land use/cover change in the Northern Ecuadorian Amazon, World Dev., 34, 1831-1849. Mena, C., R. E. Bilsborrow, and M. E. McClain (2006b), Deforestation in the Napo Basin: Socioeconomic factors, spatial patterns, and metrics, Environ. Manage., 37, 802-815. Moran, E. F. (1981), Developing the Amazon, Indiana Univ. Press, Bloomington, IN. Moran, E. F. (1990), Private and public colonization schemes in Amazonia, in The Future ofAmazonia: Destruction or Sustainable Development?, edited by D. Goodman, and A. Hall, St. Martin's Press, New York. Moran, E. F., E. Brondizio, and S. McCracken (2002), Trajectories of land use: Soils, succession, and crop choice, in Deforestation and Land Use in the Amazon, edited by C. H. Wood, and R. Porro, pp. 193-217, Univ. of Florida Press, Gainesville, FL. Moran, E., E. Brondizio, and L. VanWey (2005), Population and Environment in Amazonia: Landscape and Household Dynam- ' ics, Population, Land Use, and Environment, edited by B. Entwisle and P. Stem, pp. 106-134, The National Academies Press, Washington, D. C. Moran, E. F., E. S. Brondizio, and M. Batistella (2008), Trajetorias de Desmatamento e Uso da Terra na Amazonia Brasileira: Uma
141
Analise Multiescalar, in Amazonia: Natureza e Sociedade em Transformar;iio, edited by M. Batistella, E. F. Moran, and D. S. Alves, pp. 137-180, EDUSP, Sao Paulo. Muchagata, M. (1997), Forests and people: the role of forest production in frontier farming systems in Eastern Amazonia, Development Studies Occasional Paper 36,79 pp., University of East Anglia, School of Development Studies, Norwich, England. Murphy, L. L. (2001), Colonist farm income, off-farm work, cattle, and differentiation in Ecuador's northern Amazon, Hum. Organ., 60(1), 67-79. Myers, N. (1990), The biodiversity challenge: Expanded hot-spots analysis, Environmentalist, 10, 243-256. Nepstad, D., D. McGrath, A. Alencar, A. C. Barros, M. Carvalho, and M. d. C. Vera Diaz (2002), Frontier governance in Amazonia, Science, 295, 629-631. Netting, R. M. (1993), Smallholders, Householders: Farm Families and the Ecology ofIntensive, Sustainable Agriculture, Stanford Univ. Press, Stanford. OESP (2008a), Credito Facil do Governo Contribui com a Desmantamento na Amazonia, 0 Estado de Sao Paulo. OESP (2008b), TCU diz que Pequenos Produtores Respondem par 18% do Desmatamento, 0 Estado de Sao Paulo. Orrne, C. D. L., et al. (2005), Global hotspots of species richness are not congruent with endemism or threat, Nature, 436, 1016-1019. Ozorio de Almeida, A. L., and 1. S. Campari (1995), Sustainable Settlement in the Brazilian Amazon, Oxford Univ. Press, New York. I Padoch, C., and M. Pinedo-Vasquez (2006), Concurrent activities ~nd invi~ib!e ~ec~ologies: An example of timber management III Amazoma, III Human Impacts on Amazonia: The Role ofTraditional Ecological Knowledge in Conservation and Development, edited by D. Posey, and M. Ballick, Columbia Univ. Press, New York. Padoch, C., E. S. Brondizio, S. Costa, M. Pinedo-Vasquez, R. Sears, and A. Siqueira (2008), Urban forest and rural cities: Multi-sited households, consumption patterns, and forest resources in Amazonia, Ecol. Soc., 13(2), 2. (Available at http://www.ecologyandsociety.org/vo113/iss2/art21) Palm, C., S. Vosti, P. Sanchez, and P. Erickson (Eds.) (2005), Slash-and-Burn Agriculture: The Search for Alternatives, Columbia Univ. Press, New York. Pan, W. K, and R. E. Bilsborrow (2005), The use of a multilevel statistical model to analyze factors influencing land use: A study of the Ecuadorian Amazon, Global Planet. Change, 47, 232-252. Pan, W. K, S. J. Walsh, R. E. Bilsborrow, B. G. Frizzelle, C. M. Erlien, and F. Baquero (2004), Farm-level models of spatial patterns of land use and land cover dynamics in the Ecuadorian Amazon, Agric. Ecosyst. Environ., 101,117-134. Pan, W. K., D. Carr, A. Barbieri, R. Bilsborrow, and C. Suchindran (2007), Forest clearing in the Ecuadorian Amazon: A study of p~tterns over space and time, Populo Res. Policy Rev., 26(5-6), 635-659, doi:lO.1007/s11 113-007-9045-6. Peroni, N., P. Y. Kageyama, and A. Begossi (2007), Molecular differentiation, diversity, and folk classification of "sweet" and
142 SMALL FARMERS AND DEFORESTATION IN AMAZONIA "bitter" cassava in Caif;ara and Caboclo management systems (Brazil), Genetic Resources and Crop Evolution, 54(6), 1333-1349. Perz, S. G. (2001), Household demographic factors as life cycle determin~ts ofland use in the Amazon, Populo Res. Policy Rev., 20, 159-186. Perz, S. G., and R. Walker (2002), Household life cycles and secondary forest cover among smllll farm colonists in the Amazon, World Dev., 30,1009-1027. Pichon, F. (1997), Settler households and land-use patterns in the Amazon frontier: Farm-level evidence ·from Ecuador, World ' Dev., 25, 67-91. Pichon, F., and R. E. Bilsborrow (1999), Land use systems, deforestation and demographic factors in the humid tropics: Farmlevel evidence from Ecuador, in Population and Deforestation in the Humid Tropics, edited by R. Bilsborrow and D. Hogan, pp. 175-207, International Union for the Scientific Study of Population, Liege, Belgium. Pichon, F., C. Marquette, L. Murphy, and R. E. Bilsborrow (2002), Endogenous patterns and processes of settler land use and forest change in the Ecuadorian Amazon, in Deforestation and Land Use in the Amazon, edited by C. Wood and R. Porro, Univ. Press of Florida, Gainesville, FL. Pinedo-Vasquez, M., and C. Padoch (2009), Urban, rural and inbetween: Multi-sited households, mobility and resource management in the Amazon floodplain, in Mobility and Migration in Indigenous Amazonia: Contemporary Ethnoecological Perspectives, edited by M. N. Aiexiades, Berghahn, Oxford, U. K., in press. Pinedo-Vasquez, M., D, J. Zarin, K Coffey, C. Padoch, and F. Rabe10 (2001), Post-boom logging in Amazonia, Hum. Ecol., 29(2),219-239. Pinedo-Vasquez, M., C. Padoch, D. McGrath, and T. XimenesPonte (2002), Biodiversity as a product of smallholder response to change in Amazonia, in Cultivating Biodiversity: Understanding, Analysing and Using Agricultural Diversity, edited by H. Brookfield et aI., pp. 167-178, ITDG, London, U. K Pinedo-Vasquez, M., D. McGrath, and T. Ximenes (2003), Brazil (Amazonia), in Agrodiversity: Learning from Farmers Across the World, edited by Fl. Brookfield, H. Parsons, and M. Brookfield, pp. 43-78, UNU Press, Tokyo, Japan. Primack, R. B., and R. T. Corlett (2005), Tropical Rain Forests: An Ecological and Biogeographical Comparison, Blackwell, New York. Rerkasem, K, and M. Pinedo-Vasquez (2007), Diversity and innovation in smallholder systems in response to environmental and economic changes, in Managing Biodiversity in Agricultural Ecosystems, edited by D. Jarvis, C. Padoch, and H. D. Cooper, Columbia Univ. Press, New York. Roseberry, W. (1993), Beyond the agrarian question in Latin America, in Confronting Historical Paradigms, edited by F. Cooper et aI., pp. 318-370, Univ. of Wisconsin Press, Madison. Santos, C., and J. P. Messina (2008), Multi-sensor data fusion for modeling African palm in the Ecuadorian Amazon, Photogramm. Eng. Remote Sens., 74(2), Special Issue on Remote Sensing Data Fusion, 711-723.
Schmink, M., and C. Wood (1992), Contested Frontiers in Amazonia, Columbia Univ. Press, New York. Sierra, R. (2000), Dynamics and patterns of deforestation in the Western Amazon: The Napo deforestation front, 1986-1996, Appl. Geogr., 20,1-16. Silva-Forsberg, M. C., and P. Fearnside (1997), Brazilian Amazonian caboclo agriculture: Effect of fallow period on maize yield, For. Ecol. Manage., 97(3), 283-291. doi:1O.1016/S03781127(97)00070-4. Simons, C. S. (2005), Territorializing land conflict: Space, place, and contentious politics in the Brazilian Amazon, Geojournal, 64,307-317, doi 10.l007/s10708-005-5809-x. Siqueira, A D., S. D. McCracken, E. Brondizio, and E. F. Moran (2003), Women and work in a Brazilian agricultural frontier, in Gender at Work in Economic Life, edited by G. Clark, pp. 243267, Altimira Press, Lanham, Md. Siqueira, A D., A O. D'Antona, M. F. Dantona, and E. F. Moran (2007), Embodied decisions: Reversible and irreversible contraceptive methods among rural women in the Brazilian Amazon, Hum. Organ., 66(2),185-195. Smith, N. (1982), Rainforest Corridors: The Transamazon Colonization Scheme, Univ. of California Press, Berkeley. Smith, N., R. Vasquez, and W. Wust (2007), Amazon River Fruits: Flavors for Conservation, Amazon Conservation Association (ACA), Miss. Bot. Garden Press, S1. Louis, Miss. Smith, N. J. H., 1. C. Falesi, P. T. Alvin, and E. A S. Serrao (1996), Agroforestry trajectories among smallholders in the Brazilian Amazon: Innovations and resiliency in pioneer and older settled areas, Eco!. Econ., 18,15-27. Sorrensen, C. (2004), Contributions of fire use study to land use/ cover change frameworks: understanding landscape change in agricultural frontiers, Hum. Ecol., 32(4), 395--419. Sydenstricker Neto, J., and S. A Vosti (1993), Household size, sex composition, and land use in tropic moist forests: Evidence from the Machadinho Colonization Project, Rondonia, Brazil, paper presented at Annual Meeting, Popu1. Soc. of Am., Cincinnati, Ohio. Tamariz, M. E., and X. Villaverde (1997), Diagnostico de la Tenencia de la Tierra en las Provincias de Sucumbios y Napo, FEPP, Quito, Ecuador. Tonio10, A (2004), The role of land tenure in the occurrence of accidental fires in the Amazon Region: Case studies from the National Forest of Tapajos, Para, Brazil, Doctoral dissertation, Indiana University, Bloomington. Tura, L. R., and F. de A Costa (Eds.) (2000), Campesinato e Estado na Amazonia: Impactos do FNO no Para, Brasilia Juridical FASE, Brasilia, Brazil. Uquillas, J. (1984), Colonization and spontaneous settlement in the Ecuadorian Amazon, in Frontier Expansion in Amazonia, edited by M. Schmink and C. Wood, pp. 261-284, Univ. of Florida Press, Gainesville, FL. VanWey, L., A. O. D'Antona, and E. S. Brondizio (2007), Household demographic change and land use/land cover change in the Brazilian Amazon, Populo Environ., 28,163-185.
BRONDIZIO ET AL.
143
Vosti, S. A, E. M. Braz, 2. L. Carpentier, and M. V. N. Oliveira the tropics, Doctoral dissertation, Univ. of Florida, Gainesville, (2003), Rights to forest products, deforestation and smallholder FL. income: Evidence fmin western Brazilian Amazon, World Dev., Zarin, D., J. R. A1avalapati, F. E. Putz, and M. Schmink (Eds.) 31(11), 1889-190!l'~ (2004), Working Forests in the Neotropics: Conservation Walker, R. T. (200ii, Mapping process to pattern in the landscape through Sustainable Management?, Columbia Univ. Press, New change of the"A(/m"azonian frontier, Ann. Assoc. Am. Geogr., 93, York. 376-398. Walker, R. T., E. F. Moran, and L. Anselin (2000), Deforestation and cattle ranching in the Brazilian Amazon: External capital and household process, World Dev., 8, 683-699. Walker, R. T., S. Perz, M. Caldas, and L. G. T. Silva (2002), Land M. Batistella, Embrapa Monitotiamento por Satelite, Av. Soluse and land cover change in forest frontier: The role of house- dado Passarinho 303, Fazenda Chapadllo, Campinas, SP CEP hold life cycles,Int. Reg. Sci. Rev., 25,169-199. 13070-115, Brasil. Walsh, S. J., J. P. Messina, K Crews-Meyers, R. E. Bilsborrow, R. Bilsborrow, Biostatistics Department and Carolina Populaand W. K Pan (2002), Characterizing and modeling patterns of tion Center, University of North Carolina, Chapel Hill, NC 27516deforestation and agricultural extensification in the Ecuadorian 2524, USA Amazon, in Linking People, Place, and Policy: A GIScience ApE. S. Brondizio and E. Moran, Department of Anthropology, Inproach, edited by S. J. Walsh and K Crews-Meyers, pp. 187- diana University, 701 E. Kirkwood, Student Building 130, Bloom214, Springer, Boston, MA. ington, IN 47405, USA ([email protected]) Winkerprins, A (2006), Jute in the Lower Amazon, J. Hist. Geogr., AD. Cak, School of Public and Environmental Affairs, Indiana 32(4),818-838. University, Student Building 331, Bloomington, IN 47405, USA WinklerPrins, A M. G. A. (2002), Recent seasonal floodplainM. Caldas, Department of Geography, Kansas State University, upland migration along the lower Amazon River, Brazil, Geogr. 118 Seaton Hall, Manhattan, KS 66502, USA Rev., 92(3), 415--431. C. T. Futemma, Campus de Sorocaba, Rodovia JOllo Leme dos Wood, C., and M. Schmink (1979), Blaming the vi~tim: Small Santos, krn-llO, SP-264, Bairro Itinga, Sorocaba, SP CEP 18052farmer production in an Amazon Colonization Project, Stud. 780, Brazil. Third World Soc., 7,77-93. T. Ludewigs, Centro de Desenuolvimento Sustentavel, UnBWood, C. H., and R. Porro (Eds.) (2002), Deforestation and Campus, Universitarlo Darch Ribeiro-Gleba A, Bloco C-Av. L3 Land Use in the Amazon, Univ. of Florida Press, Gainsville, Norte, Asa Norte, Brasilia DF, CEP 70904-970, Brazil. FL. C. Mena, Colegio! de Ciencias Biologicas y Ambientales, UniYamada, M. (1999), Japanese immigrant agroforestry in the Brazil- versidad San Francisco de Quito, Campus Cumyaba Diego de ian Amazon: A case study of sustainable rural development in Robles SIN, Quito, Ecuador.
Understanding the Climate of Amazonia: Progress From LBA Carlos A. Nobre and Jose A. Marengo Center for Earth System Science, National Institute for Space Research sao Jose dos Campos, Brazil
Paulo Artaxo Institute ofPhysics, University ofsao Paulo, sao Paulo, Brazil
The Amazon plays an important role in the functioning of the Earth's climate. It acts as one of the critical heat sources for the global atmosphere via evaporation of water vapor at the surface and release of heat in the middle and upper troposphere by latent heat of condensation in tropical convective clouds. Forest evaporation year-round is about 3 to 3.5 mID d- 1. In contrast, for savanna regions bordering the Amazon forest, evaporation is reduced during the dry season because of the limitation of soil moisture. Deforestation causes a large reduction in dry season evaporation. The cloudiness-rainfall regime of the undisturbed forest/is similar to that of a tropical ocean, which led us to call the Amazon Basin a "green ocean." Biogenic volatile organic compounds are released by the forest into the atmosphere, where some play an important role in providing aerosols and cloudl condensation nuclei. On the other hand, aerosols emitted by biomass burning may cause reduction of rainfall. Together the chapters in this section reveal a complex mix of interacting processes that acting in concert control the movement and composition of the atmosphere above Amazonia. Each separate study reveals some new insight into one facet, but together they reveal an integrated system in which change in one component will produce impacts in another. The most important message is that deforestation is not just a change in land use, but it impacts the functioning of the Amazonian ecosystem itself. Changes in the evaporation and the chemical composition of the atmosphere produce changes in the cloud physics and in the dynamics and thermodynamics of the atmospheric circulation. These, in tum, impact the rainfall and the hydrological cycle.
Its immense size and its position close to the equator en- tent heat; water evaporated at the surface is transported into sure that Amazonia plays a critical role in the functioning of the upper troposphere by intense tropical convection, from the Earth's climate. We know that in general terms Amazo- there it contributes energy to driving the global atmospheric nia is a strong global source of water vapor and therefore la- . circulation. Nobre et al. [this volume] give a more complete analysis, but in simple terms, water vapor is transported in Amazonia and Global Change westward moving trade winds from the Atlantic through Geophysical Monograph Series 186 the Amazon Basin, evaporation recycling the rainfall as it Copyright 2009 by the American Geophysical Union. moves. The barrier of the Andes Mountains then diverts the 1O.1029/2009GM000903 low-level air southward, which transports moisture to the La 145
146 UNDERSTANDING THE CLIMATE OF AMAZONIA Plata Basin. However, superimposed on this simple analysis, there are marked temporal, interannual to interdecadal, and spatial variations. Much of the temporal variation is caused by fluctuadbns in surface temperatures in the Atlantic and Pacific oceans. But how much results from forcing through climate change and deforestation? Identifying these anthropogenic changes in the processes driving the land-atmosphere interactions has been the central task for atmospheric scientists within LBA. Analysis of flux tower data '[ da Rocha et ai., this volume] has shown that high-rainfall sites with short dry seasons have evaporation rates of typically 3 to 3.5 mm d- I , with dry season evaporation being about 10% greater than that in the wet season. On the savanna margins, there is an opposite response, and dry season evaporation falls in response to reduced soil moisture. Deforestation reduces the dry season evaporation further, with the atmospheric boundary layer, which connects the surface to the atmosphere above, responding to that drying [Betts et ai., this volume]. Each day the atmospheric boundary layer typically grows to over a kilometer in depth, driven by the convective flux from the surface. Deforestation increases the fraction of incoming solar radiation that is transferred into sensible heat flux, with the height of the boundary layer, and the consequent cloud base, responding. The dry season cloud base over deforested pasture may be nearly 4 times that found over the forest in the wet season, where the low height is more typical of the tropical oceans. This analogy between wet season forests and the oceans, the Amazon Basin as the "green ocean," is a recurring theme in the results from LBA. Biogenic volatile organic compounds are released by the forest into the atmosphere, where some play an important role in providing aerosols and cloud condensation nuclei [Kesseimeier et ai., this volume]. There are a large number of these compounds, and the individual role in rainfall generation of each chemical species is far from understood. Nevertheless, new measurements have resulted in the fa.ctors controlling emission becoming characterized, with variables such as light, temperature, photosynthesis, and phenology providing a new basis for modeling emission and its seasonal variation. As land is deforested, natural biogenic compounds are replaced by aerosol particles released during burning, either during land clearance or maintenance burning of pastures [Longo et ai., this volume]. These aerosols have been sampled from aircraft and tracked by remote sensing. Remote sensing combined with meteorological models can now be used to forecast smoke impacts on weather and air quality. A plume tracked by satellite as it moved from Amazonia and across southern Brazil was successfully modeled. In the wet season the concentrations of naturally emitted organic condensation nuclei have been found to be very low, again,
NOBRE ET AL. as found over the oceans, but these lead to the formation of large cloud droplets [Artaxo et ai., this volume]. In contrast, in the dry season the high concentration of aerosols released during burning produces a large population of small droplets. These small droplets inhibit precipitation, leading to a direct impact of land use on climate. The dry season aerosol concentration has also been found to increase the amount of diffuse sunlight; this initially increases the penetration of light into the canopy and increases photosynthesis, but at high concentrations, photosynthesis shuts down completely. Many experiments with global climate models have compared a totally deforested Amazonia with a control run with the present forest. Almost all predict that deforestation results in decreased evaporation and moisture convergence, leading to reduced rainfall. However, high-resolution regional meteorological models now suggest that partial deforestation may promote preferential cloud formation and increased rainfall over deforested areas [Silva Dias et ai., this volume]. Clearly, there must be a tipping point at some critical level of clearance where the smaller-scale processes are overwhelmed by the large-scale processes. Such tipping points have been identified in other studies [Marengo et ai., this volume]: increasing greenhouse gas concentration, deforestation, widespread biomass burning, more EI Niiio-like conditions in the Pacific, and changes in the tropical Atlantic are all drivers leading to changes in the Amazonian water cycle. A permanent switch from forest to savanna-like vegetation as the climate tips from quasi-stable state to another has been identified as an unlikely event for small perturbations but one with a finite possibility of potentially enormous impact of significant magnitude for global warming or deforestation. Together the chapters in this section reveal a complex mix of interacting processes that acting in concert control the movement and composition of the atmosphere above Amazonia. Each separate study reveals some new insight into one facet, but together they reveal an integrated system in which change in one component will produce impacts in another. The most important message is that deforestation is not just a change in land use, but it impacts the functioning ofthe Amazonian ecosystem itself. Changes in the evaporation and the chemical composition ofthe atmosphere produce changes in the cloud physics and in the dynamics and thermodynamics of the atmospheric circulation. These, in tum, impact the rainfall and the hydrological cycle. REFERENCES Artaxo, P., et al. (2009), Aerosol particles in Amazonia: Their composition, role in the radiation balance, cloud formation, and nutrient cycles, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000778, this volume.
Betts, A. K., G. Fisch, c. voh Randow, M. A. F. Silva Dias, l C. P. Cohen, R. da Silva, andD. R. Fitzjarrald (2009), The Amazonian boundary layer and,.mesoscale circulations; Geophys. Monogr. Ser., doi:lO.l029/J,008GMOOOn5, this volume. da Rocha, H. R, A. g: Mafi.?:i, and l Shuttleworth (2009), Evapotranspiration, Geophfs. Monogr. Ser., doi: 1O.1029/2008GM000744, this volume. Kesselmeier, J., A.. Guenther, T. Hoffmann, M. T. Piedade, and J. Warnke (2009), Natural volatile organic compound emissions from plants and their roles in oxidant balance and particle formation, Geophys. Monogr. Ser., doi:l0.l029/2008GM000717, this volume. Longo, K. M., S. R. Freitas, M. O. Andreae, R Yokelson, and P. Artaxo (2009), Biomass buming in Amazonia: Emissions, long-range transport of smoke and its regional and remote impacts, Geophys. Monogr. Ser., doi:10.1029/2008GM000847, this volume. Marengo, l, C. A. Nobre, R. A. Betts, P. M. Cox, G. Sampaio, and L. Salazar (2009), Global warming and climate change in Ama-
147
zonia: Climate-vegetation feedbacks and impacts on water resources, Geophys. Monogr. Ser., doi:lO.l02912008GM000743, this volume. Nobre, C. A., G. O. Obregon, J. A. Marengo, R Fu, and G. Poveda (2009), Characteristics of Amazonian climate: Main features, Geophys. Monogr. Ser., doi: 10.1 029/2008GMOOOnO, this volume. Silva Dias, M. A., R Avissar, and P. Silva Dias (2009), Modeling the regional and remote climatic impact of deforestation, Geophys. Monogr. Ser., doi: 10.1 02912008GM000817, this volume.
P. Artaxo, Institute of Physics, University of Sao Paulo, Sao Paulo, SP 05508-900, Brazil. J. A. Marengo and C. A. Nobre, Center for Earth System Science, National Institute for Space Research, Sao Jose dos Campos, SP 515 12201, Brazil. ([email protected])
II "
J~
Characteristics of Amazonian Climate: Main Features Carlos A. Nobie, Guillermo O. Obregon, and Jose A. Marengo Centro de Ciencias do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, Brasil
RongFu Jackson School o/Geosciences, University o/Texas at Austin, Austin, Texas, USA
German Poveda Escuela de Geociencias y Medio Ambiente, Universidad Nacional de Colombia, Medellin, Colombia
This chapter summarizes our current knowledge on the mean climatological features ofAmazonia. Significant uncertainties remain in our understanding of the complex dynamics of climate and climate variability in that region, which are due, in part, to the lack of observational data. The strong seasonality of the rainfall and the relatively rapid transition between the wet and dry season associated with onset of the rainy season is related to the establishment M the South America Monsoon System (SAMS). The SAMS is controlled by large~scale thermodynamic conditions influenced by the near-equatorial sea surface temperature (SST). It has been suggested that land-surface dryness in the dry season is the main cause of the delay in the onset of the subsequent wet season. The 30- to 60-day oscillation is the major mode ofintraseasonal variability. Interannual variability ofthe hydroclimatic system is strongly related to El Niiio-Southern Oscillation. More generally, tropical Pacific and Atlantic SSTs control rainfall variability in Amazonia, and SWAtlantic SST anomalies influence the variability of the South Atlantic Convergence Zone (SACZ). Land surface-atmosphere interactions have been proposed as a possible dynamical mechanism for the unexplained variance at the annual and interannual timescales. At decadal and interdecadal timescales, rainfall variability is related to the Pacific Decadal Oscillation mainly over the southern portions, and linked to the North Atlantic Oscillation. At paleoclimate timescales, there is large uncertainty on major aspects of rainfall variability over tropical South America. For instance, there remains uncertainty on the basic character of rainfall anomalies over Amazonia, whether drier or wetter, during the Last Glacial Maximum, and paleoclimate reconstructions still suffer from lack of data. 1. INTRODUCTION Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1 029/2008GMOOOnO
The Amazon basin is one of the three quasi-pelmanent centers of intense convection embedded in the equatorial trough zone. It plays a pivotal role in the functioning of the 149
NOBRE ET AL. 150
151
CHARACTERISTICS OF AMAZONIAN CLIMATE
northwest-southeast orientation from Amazonia. to ~he s.ubglobal climate. The forests ofAmazonia playa critical role in tropics near the coast of southeast Brazil, proJectmg mto regulating climate at both regional and global levels. Through the adjacent South Atlantic Ocean [Kodama, 1992, 1993]. intense ~vapotranspiration, the tropical forests pump lat~nt In the lower troposphere, the trade winds from the equ~ heat into the atmosphere to balance the strong surface rad.la- torial Atlantic penetrate into Amazonia and then tum antltive heating. The strong and extensive trop.ical convectlon clockwise east of the Andes Cordillera to flow southward over the continent during the Southern HemIsphere su~er and southeastward to 15°S, where the flow then .becomes tranl'ports the latent heat to the upper troposphere and dIS- cyclonic in the central parts of the continent formmg a low tributes it to the temperate zones. In doing so, the for~s~ and 20 0 S. d convection together cool funazonia, while also provldmg.a near All of the above mentioned characteristics of the o?serve strong tropospheric heat source for the glo~al atmosphenc circulation over tropical and subtropical South ~menca durcirculation. Thus, the release of latent heat IS a large s~urce ing the austral summer form the South Amencan ~ummer of heating in the region and is responsible for t~e reg~onal Monsoon (SASM). Many of these features that.dommate the circulation characteristics in austral sum~er [Sl~va Dzas et regional circulation are recognized as the typIcal monsoon al., 1987]; it also can generate significan~ clrculatlOn anoma- characteristics [Zhou and Lau, 1998; Vera et.al., 2006]. The lies in the Northern and Southern HemIspheres as telecon- SASM main features are best developed dunng the Decem. nection patterns [Grimm and Silva Di~s, 1995].. ber-February summer months and include a large-sca~e la~d The complex interactions among climate vanables.m th.e ocean temperature gradient, low pressure over the 111t~r.lOr Amazon basin have important implications for potential cl~ of the continent (Chaco Low) and high pressur~ (BoliVian mate change, both globally and local~y. Because Amazoma High) with anticyclonic circulation aloft, a vertlc.ally .overis a data-sparse region, the climate va~ia~les are poorly qu~n turning circulation with a rising branch over the illt~nor of tified; significant uncertainties remal~ m our und~rstandmg the continent and sinking'motion over the ocean, and ~ntense of the different processes that underlIe the dyna~llc mecha- moisture influx to the continent at low levels responslb~e for nisms of the climate and its variability over a wIde range of strong seasonal precipitation changes, as well as .a mmsture temporal and spatial scales. ..' outflow from the Amazon region to La Plata basm, referred Tropical convection is the main charactenstlc of climate to as the South American low-level jet east of the Andes over the Amazon basin. It is mainly modulated ~y lar~e [Marengo et al., 2004]. scale atmospheric features including the Hadley clfculatlon and the Intertropical Convergence Zone (ITCZ),. the. Walker 2. SPATIAL DISTRIBUTIONS OF CLIMATIC . 1at'lOn, the 40 - to 60-day intraseasonal oscillatlon, and VARIABLES Clrcu . . waves, but also by meteorologIcal processes atmosp henc such as the penetration of extratropical cold fronts [Santos 2.1. Temperature de Oliveira and Nobre, 1986]. Also, the im~ortance of~and surface-atmosphere feedbacks on Amazoman hy~ochma At seasonal timescales, mean air temperature do~s not tology cannot be overstated. Regional patterns oframfal~ de- vary a great deal in most of the region with the e~ceptlon.of pend largely on the water and energy surfa~e budg~ts ~nven southern Amazonia (Rondonia, Mato Grosso): T~lS behavlOr by seasonal and diurnal cycles of solar heatl~g, which ill tum is due to the high values of incident solar radlatlOn throu~h modulate the recycling of precipitation WhICh can. account out the year. Mean air temperature values are between 24 to of rainfall in Amazoma [Elta- 26?C with annual amplitude of 1° to 2°C. In southern Ama. te1y 25 - 30°/ for approxlma /0 hir and Bras, 1996]. Important roles are .also playe.d by the zonia, the annual cycle of temperature is more .pronounced sea surface temperature (SST) of the tropIcal Atla~tlc Ocean due to solar forcing and also due to the pe~etratlOn of extra: [Dickinson, 1987] on the eastern part of the basill and the tropical cold fronts. Here, the annual amp.htud~ can re~ch 3 forcing of the Andes along the western p~rt .. to 4°C. More important is the large amplitude m .the d~urnal Two of the most distinctive charactenstlcs of th~ upper cycle of temperature and solar heating for the tnggenng of level atmospheric circulation in the Sou~hern HemIsphere convection and the development of intense storms over the er the tropical South Amenca are the well., h "B r . summer ov region [Fu et al., 1999]. defined anticyclone centered over Bohvla, t e o IVlan High" [Kreuels et al., 1975; Virji, 1981] and a trough near 2.2. Precipitation the coast of northeast Brazil [Kousky and ~an, 1981). The South Atlantic Convergence Zone ~SA~Z) IS another Imp?rA large number of analyses of the spatial and. temporal tant feature of the summer circulatlOn m the South Am~nca distribution ofrainfall over Brazil and South Amenca can be region. It is a wide and long convergence zone followmg a
used to describe the main features ofprecipitation [Schwerdt- scending motion and is part of the mechanism that leads to feger, 1976; Ratisbona/1976; Caviedes, 1981;Salati, 1987; _ ilie very low rainfall over that region. Southern Amazonia Horeletal., 1989;FipteroaandNobre, 1990;RaoandHada, and Bolivia Altiplano are strongly heated during the austral 1990; Marengo, 19J2, 1995; Rao et al., 1996; Marengo and warm season resulting in enhanced tropospheric zonal temperature gradients and enhanced upper tropospheric meridiNobre, 2001; Map!ngo, 2005]. In the nOlihel; pmi of the basin, the spatial and seasonal onal flows in the vicinity of both coasts of South America. In rainfall distribution shows a significant heterogeneity. The addition, the mechanical blocking of the cross-Andes flow southern part has distinct dry and wet seasons, with a maxi- and lee genesis also contribute to strong pressure gradients mum of precipitation occurring in the austral summer. To- on the east side of the Andes and lower level meridional tal annual rainfall shows two maxima located, respectively, flow, leading to strong episodes of low-level jets following around the mouth ofthe Amazon River and over the western strong zonal wind over the Andes [Campetella and Vera, part of the basin. The maximum of annual rainfall, located 2002; Wang and Fu, 2004]. over northwestern Amazonia with an annual total of over At the opposite extreme of the annual cycle during aus3000 mm is associated with low-level convergence of east- tral winter (Plate 1b), upper level westerly flow extends unerly moisture flow, likely a result of the presence and also impeded across South America. During both extremes of to the concave shape of the Andes to the west of that maxi- the annual cycle, easterly flow at 850 hPa extends into the mum [Nobre, 1983]. The high rainfall over this region may Amazon basin and then turns toward the south as it apbe understood as the response of the dynamical fluctuation proaches the Andes mountains. These features of both of the center of quasi-pennanent convection [Marengo and the upper and lower level circulation over tropical South Hastenrath, 1993], in combination with the large amount of America, which are closely linked to the maximum precipilocal evapotranspiration contributing to precipitation recy- tation over Amazonia and the SACZ, encompass the main cling. The second precipitation maximum located over the characteristics of other tropical monsoonal circulations and mouth of the Amazon River has been associated with the are currently known as South American Monsoon System ITZC [Hastenrarth and Heller, 1977] and local circulations (SAMS) [e.g., Zhou and Lau, 1998; Marengo et al., 2004; related to instability lines, which appear along the coast Vera et al., 2006]. IThe typical low-level wind reversal mainly during late afternoon forced by the sea breeze circu- toward prevailing westerlies of other monsoonal circulalation [Cohen et al., 1995]. tions is not seen in the SAMS because the Andes CordillDuring the austral spring, precipitation increases over the era prevents that from happening. However, the reversal of Amazon basin, and a NW-SE band of precipitation devel- meridional winds frbm southerly in winter to northerly in ops linking tropical convection in the west of the basin to summer is similar to other monsoon systems, and it strongly precipitation activity in the extratropics. In the austral sum- influences moisture transport and distribution of the rainfall mer season, there is a marked maximum centered at about [Wang and Fu, 2002]. 10 S. That maximum is prolonged to the SE to form SACZ. The main source of moisture over the Amazon basin is the Most of tropical and subtropical South America receives tropical Atlantic Ocean through a persistent northeasterly more than 50% of its total annual rainfall in austral sum- flow most of the year. To the south, the flow turns southmer [Figueroa and Nobre, 1990] in the form of convective ward, supplying moisture to the higher latitudes of South rainfall with strong diurnal variation. In southern Amazonia, America. A large part of the southward moisture transport daily rainfall amounts are of the order of 10 mm d- I on av- is carried out by a northerly low-level jet, with a maximal erage over vast regions, reaching over 30 mm d- I in heavy wind speed on the order of 15 m S-I at 850 hPa, at approxirainfall episodes. mately 17°S and 62°W. This low-level jet is responsible for the transport of water vapor and heat fi'om the Amazon to Paraguay, northern Argentina, and southern Brazil [Nogues2.3, Winds and Geopotential Height Paegle and Mo, 1997; Marengo et al., 2004]. The distinctive aspect of the upper level circulation over The upper level circulation over South America durtropical South America during the austral summer is the ing austral winter is characterized by weak winds over ilie Bolivian High (Plate 1a). Its genesis is strongly related to tropics, while the subtropical westerly jet is stronger and latent heat released over the regions of high precipitation in located equatorward in comparison to its summer position, Amazonia [Silva Dias et al., 1983; Figueroa et al., 1995; consistent with the descending branch of the Hadley-type Seluchi et al., 1998, among many others]. Downstream of circulation over that area. At lower levels, a northwardthe Bolivian High to the east, there is an upper level trough displaced near-equatorial low-pressure trough characterizes off the east cost of northeast Brazil, associated with de- the circulation. A northward cross-equatorial flow turning 0
152
NOBRE ET AL.
CHARACTERISTICS OF AMAZONIAN CLIMATE
producing high convective nebulosity during fall and spring [Oliveira, 1986]. In "dnter, cold front penetration can procooling in the rfegion, especially southern and western Amazonia, and terpperatures can drop to very low values, as during the strong'polal'outbreak of June 1994, where temperatures in Rio Branco, Acre, fell to 11°C. 00
2.4. Swface Pressure
The spatial distribution of the sea level pressure over the Amazonia is almost constant throughout the year, due to its tropical position. The maximum values are observed in wintertime and the minimum in summertime. During this latter season, low values of pressure south of Amazonia extend to subtropical areas east of the Andes. 2.5. Solar Radiation and Cloudiness
Wmls
15°N
b) JJA: 200 hPa WiJUI
..,
00
Wmls
Plate 1. Upper level circulations (200 hPa) over tropical South America: (a) austral SUiumer (December-Febmary) and (b) austral winter (June-August).
clockwise is found over the tropical Atlantic Ocean. Near the surface, during wintertime, surges of cold high-latitude air, known locally as "friagens," move across southeastern Brazil and Amazonia from the south, greatly modifying the atmospheric structure and climatic conditions. "Friagens"
can produce severe frost in areas of southern Brazil and substantial cooling in the Amazon basin [Hamilton and Tarifa, 1978; Marengo et al., 1997a, 1997b]. The cold air heading these thrusts of high-latitude air may reach as far north as the equator. The events are relatively common in Amazonia
Top-of-the-atmosphere solar radiation in Amazonia between 5°N and lOOS varies from a maximum value of 36.7 MJ m-2 d- 1 in December-January to a minimum value onO.7 MJ m-2 d- 1 in June-July [Salati and Marques, 19&4]. At the surface, incident solar radiation is about 16-18 MJ m-2 d- 1• The seasonal cycle of incident solar radiation at the surface in central Amazonia shows maximum values in September/ October and minimum values in DecemberlFebruary [Cu?! et al., 1996]. This temporal distribution is controlled mostly by cloudiness related to Amazonian convection [Hore! et al., 1989]. The outgoing longwave radiation (OLR), which describes upwelling infrared emission from the Earth, can also be used to estimate the depth of convective clouds. Low values of OLR denote areas of deep convection, abundant rainfall, and enhanced upper level divergence [Liebmann et al., 1998]. The Amazonia exhibits a strong annual cycle in OLR [Hore! etal., 1989; Kous/01, 1988], with a minimum during the rainy season of austral summer.
153
2003], except for some elevated zones in the Andes. The rainy season over most of the Amazon basin located in the SH is between November and March, with a peak in DecemberFebruary (DJF), and the dry season is fi'om May to September. On the northern portions of the basin, there is a reversal of this phase, whereby the rainy season occurs from May through October and the driest period from December through Februmy. These results have large-scale manifestations in the seasonal records of river discharge with a single broad annual maximum for the Amazon River [Richey et al., 1989; Amerasekera et al., 1997]. Hydroc1imatological feedbacks between the Andes and the Amazon River basin are discussed by Poveda et al. [2006]. Also, Vizy and Cook [2007] present evidence of the strong coupling of rainfall between the Andes and Amazonia ever since the Last Glacial Maximum (LGM). 3.2. Onset ofthe Rainy Season and ofthe South America Monsoon System
The transition between wet and dry seasons is short in Amazonia. The onset of the wet season occurs normally within the period of a single month. The transition from the wet to the dry season takes longer than a month. The onset of the rainy season in most of the Amazon basin, which is closely associated to the establishment of the South American Monsoon System, occurs as a rapid shift of the area of intense convection:between the northwestern extreme of the continent and latitudes south of the equator, around midOctober [Kousky, 1988; Hore! et al., 1989; Vera et al., 2006; Marengo et al., 2001; Liebman and Marengo, 2001]. The demise of the SAMS occurs typically from April to May. In the Amazon basin, Marengo et al. [2001], using pentad averages of gauge-based rainfall observations, did not find a large wind signal related to the onset, which suggests that the onset is controlled by large-scale thermodynamic conditions, especially in the southern Amazonia [Fu et al., 1999]. They did, however, find relationships between tropical SST and the onset and end of the rainy season in central Amazonia and near the mouth of the Amazon River. SST anoma3. TIME EVOLUTION lies were not found to be related to the timing of the onset in southern Amazonia. Both of these findings are consistent 3.1. Annual March ofTemperatures and Precipitation with the arguments of Fu et al. [1999] that, near the equaAs mentioned above, the values of the mean air tempera- tor, SST influence on onset may be important because the hIre over Amazonia are between 24° and 26°C, and the contrast between land and sea temperatures is small. On the yearly amplitude is between 1° to 2°C [Williams and SCltm'i, other hand, the relationship between tropical Pacific and At2004], using climatological data of mean air temperature lantic SSTs and rainfall confined to the equatorial region of fi'Olll INMET [1992], it was found that the annual cycle is Brazilian Amazonia are found during the transition season predominant, and only few stations within 5° of the equator between wet and dry regimes or, entirely, within the dry seashowed double peak in the annual march of temperahlre. son [Liebmann and Marengo, 2001]. This argument implies The armual march ofrainfall shows also only one peak [Hsu that the SST seems to control the seasonal totals through the and Wallace, 1976; Figueroa and Nobre, 1990; Obregon, timing of the rainy season onset or end.
154
NOBRE ET AL.
CHARACTERISTICS OF AMAZONIAN CLIMATE
In addition to the influence of tropical SST, Fu and Li [2004] suggest that the land surface dryness, represented by the Bowen ratio, during the dly season appears to be a main cause of th<e strong delay in subsequent wet season onset. This is because an increase of evapotranspiration during late dry season, associated with the seasonal increase ofleaf area index or greelmess of the rainforest is critical for initiating the transition from dry to wet season. Finally, changes of fre, quency and intensity of the "friagens" during the transition season can also affect the onset dates [Li and Fu, 2006]. Over much of tropical and subtropical South America, more than 50% of the annual precipitation falls during the summer months, associated with the establishment of the SASM. As a monsoon system, the SASM is dynamically and geographically different from the maritime ITCZ, although the latter is sometimes erroneously invoked to explain the seasonal march of precipitation over the South American continent [Vuille and Werner, 2005]. On interannual and longer timescales, summer precipitation shows significant variations in intensity and spatial extent, which are still not velY well understood. This variability is caused by a number of factors influencing the SASM during both the developing and mature stage, including tropical Atlantic SST [Mechoso et al., 1990; Hastenrath and Greischar, 1993; Marengo and Hastenrath, 1993; Vuille et al., 2000a], the El Nino-Southern Oscillation (ENSO) [Aceituno, 1988; Vuille, 1999; Garreaud and Aceituno, 2001; Paegle and Mo, 2002; Grimm, 2003, 2004; Lau and Zhou, 2003], land surface conditions such as soil moisture or vegetation cover through their influence on precipitation [e.g., Poveda et al., 2001; Oyama and Nobre, 2003; Koster et al., 2004; Poveda and Salazar, 2004; Xue et al., 2006], which, in turn, can feed back onto the ocean,at, mosphere dynamics of the tropical Atlantic Ocean [Poveda and Mesa, 1997; Wang and Fu, 2007] and interactions with the extratropical circulation [e.g., Garreaud and Wallace, 1998; Seluchi and Marengo, 2000; Chou and Neelin, 2001; Marengo et al., 2004]. The relative importance of the vari, ous factors contributing to SASM variability, however, is often difficult to determine, as many components such as tropical Pacific and Atlantic SST are dynamically coupled with each other [Elifield, 1996; Uvo et al., 1998; Vuille et aI., 2000b; Pezzi and Cavalcanti, 2001; Giannini et al., 2001; Ronchail et al., 2002]. 4. MECHANISMS OF AMAZONIAN CLIMATE VARIABILITY 4.1. Solar Forcing ofthe Seasonal Climate
While the annual cycle of solar radiation is undoubtedly the key factor in forcing convection in this region, there re,
main gaps in our understanding of the spatial variability of rainfall within the Amazon basin itself. The impact of local insolation on precipitation can be explained by balancing net energy input at the top of the atmospheric column with the export of energy by the diver, gent circulation that accompanies convection [Biasutti et al., 2004]. Also, increased insolation reduces the stability of the atmosphere in the convection centers, but not in mon, soon regions. The mmual cycle can be thought of as being forced locally by the direct action of the sun and remotely by circulations forced by regions of persistent precipitation organized primarily by SST and, secondarily, by land pro, cesses [Biasutti et al., 2003]. The annual cycle dominates convection over Amazonia [Hore! et al., 1989], with con, vection to a first approximation following (or lagging) solar insolation.
4.2. Interannual Climate Variability Although the annual cycle dominates the climate variabil, ity over Amazonia, interannual variability is quite remark, able, as revealed by historical records of the Amazonian rivers [Molion and de Moraes, 1987; Richey et al., 1989; Marengo, 1992, 1995; Guyot et al., 1998; Marengo et al., 1998] showing that interannual variability of precipitation in Amazonia is vety significant, which is dynamically linked with consistent anomalies in the whole set of variables ofthe surface water and energy balances over the Amazon basin. Most studies of interannual variability of Amazonian rain, fall have focused attention on anomalies associated with the ENSO phenomenon [Aceituno, 1988; Ropelevl'ski and Halp, ert, 1987,1989; Rao and Hada, 1990; Figueroa and Nobre, 1990; Obregon and Nobre, 1990; Marengo, 1992, 1995; Marengo and Hastenrath, 1993; Rao et al., 1996; Poveda and Mesa, 1997; Marengo and Nobre, 2001; Fu et al., 2001; Poveda and Salazar, 2004; Poveda et al., 2006]. The role of land surface,atmosphere interactions and, particularly, that of soil moisture and evapotranspiration have been proposed as possible dynamical mechanisms for the unexplained vari, ance of hydroclimatological processes at mmual and interannual timescales over Amazonia [Poveda and Mesa, 1997; Makarieva and Gorshkov, 2007]. Precipitation reduction in tropical South America dur, ing El Nino is also consistent with the development of an anomalous position and direction of the Hadley cell over the equatorial region. Implicit here is the existence of a positive feedback effect between the tropical precipitation and the Hadley circulation [Numaguti, 1993; Kiehl, 1994]. Nega, tive anomalies in tropical South American precipitation during El Nino are also associated with negative anomalies in soil moisture at interannual timescales [Nepstad et al.,
2004; Jipp et aI., 1998; Poveda and Mesa, 1997; Poveda et aI., 2001]. The work by Poveda et al. [2006] discusses the suite of hydrcJblimatic anomalies associated with the occurrence of EI1j{no in tropical South America, including Amazonia. / 4.3. Intraseasonal Variability
Over tropical South America, the principal mode of climatic fluctuations in the intra,almual spectral band, is in the 30, to 60-day oscillation, as manifested in the OLR anom, all' patterns at 250 hPa for the period 1979-1990 [Mo and Kousky, 1993]. 4.4. Tropical Pactfic and Atlantic Ocean
The correlation between precipitation in Brazilian Ama, zonia and SSTs over the Pacific and Atlantic has been doc, umented since the early 20th centUlY. The impact of each ocean on variability (frequency and intensity) of the wet/dly season over the Amazon basin and the underlying mecha, nisms are gradually being clarified. However, at the outset, it must be emphasized that the combined tropical Pacific and Atlantic SST variability explains little more than 50% of interannual precipitation variance over Amazonia and not much is known about other mechanisms, internal or external to the region, responsible for the remaining unexplained interannual variability. The influence of the tropical Pacific is mainly through perturbations of a Walker,like cell mechanism. El Niiio epi, sodes with warm Equatorial Pacific SSTs are associated with a weakening of the cell, with subsidence and reduced cloudi, ness and rainfall over nOlihern, central, and eastern Amazonia. A detailed discussion of mechanisms acting on hydroclima, tological anomalies over the region during El Nino is given by Poveda et al. [2006; see their Figure 8]. During La Nina, generally opposite conditions prevail, and rainfall and river discharge are above the average. ENSO could also influence rainfall in southeastern Amazonia through an anomalous atmospheric wave train £i'om the south tropical Pacific to subtropical South America forced by SST and atmospheric heating anomalies in the South tropical Pacific [Kalney et al., 1986; Liebmann et al., 1999; Fu et al., 2001]. Observational, conceptual, and atmospheric and coupled atmospheric,ocean models show evidence that the tropical Atlantic strongly influences interannual climate variability. of the Americas [Hastenrath and Heller, 1977; Moura and Shukla, 1981; Hastenrath et aI., 1984; Hastenrath, 1990; No, bre and Shukla, 1996]. Although the role ofthe tropical Pacific has been emphasized in studies of the association between SST and Amazonian rainfall, the Atlantic Ocean strongly
155
influences rainfall. Marengo [1992] and Rao et al. [1996] showed that increased rainfall in the Amazon basin is associ, ated with an increase of water vapor transport from the Atlan, tic. In particular, the eastern region of Amazonia is strongly influenced by the atmospheric and oceanic condition of the tropical Atlantic [Molion, 1993; Nobre and Shukla, 1996]. The influence of the tropical Atlantic SSTs over Amazo, nian rainfall is associated with Hadley,like cell perhlrba, tions. Positive rainfall anomalies in northern Amazonia are concomitant with: (a) anomalously warm waters in the tropi, cal North Atlantic, (b) cold surface waters in the equatorial South Atlantic, (c) weak northeast trades, which entails a reduced influx of moisture coming £i'om the Atlantic toward the Amazon basin. Consequently, the ITCZ is located anom, alously to the north of its average position. 5. MECHANISMS CONTROLLING DROUGHTS IN AMAZONIA: PACIFIC VERSUS ATLANTIC CONTROLS As mentioned above, droughts in Amazonia are usually associated with EI Nifio or warming in the tropical North At, lantic, and El Nino,caused droughts are most pronounced in the central and northern areas of the basin, as in 1926, 1983, and 1998. Previous ~tudies [Poveda and Mesa, 1997;Marengo et al., 1998, 2008a, 2008b; Ronchail et aI., 2002; Poveda and Salazar, 2004, among many others] have identified negative rainfall anomalies ih Amazonia associated with ENSO events and to SST anomalies in the tropical Atlantic as well. The studies have linked some ofthe major droughts in Amazonia to (a) the occurrence of intense El Nino events, (b) strong warming in the surface waters of the tropical North Atlantic during the NOlihern Hemisphere SUlmner,autumn season, or (c) both. VelY intense El Nino events have been associated with the extreme droughts in 1925-1926, 1982-1983, and 1997-1998, and the last two also experienced intense warming in the tropical North Atlantic along with warming in the equatorial Pacific. There is evidence of extensive droughts, and perhaps widespread fires, linked to paleo,ENSO events occurring in the Amazon basin in 1,500, 1000, 700, and 400 B.P., and these events might have been substantially more severe than the 1982-1983 and 1997-1998 ones [Meggers, 1994]. The best documented case of an earlier drought event in Amazonia linked to El Nino event was during 1925-1926 [Sternberg, 1987; Williams et aI., 2005]. Rainfall in central, northern Brazilian Amazonia and southern Venezuela in 1926 was about 50% lower than normal. Contrmy to the above droughts, the droughts of 2005 as well as those in 1963-1964 and in 1979-1981 did not occur associated with El Nino events. While several studies analyze the droughts of 1982-1983 [e.g., Marengo et al., 1998], and
156
CHARACTERISTICS OF AMAZONIAN CLIMATE
NOBRE ET AL.
1997-1998 [e.g., Nepstad et al., 1999] and 2005 [Marengo et al., 2008a, 2008b; Zeng et al., 2008] and their impacts on
displacement of the SACZ. According to Robertson and Mechoso [2000], interannual SACZ variability is accom-
climate, hydrology, and fires in Amazonia, there are only casual refet>ences to the drought event of 1963-1964. The combined effects of tropical Pacific and Atlantic SSTs explain 53% of the Amazon basin rainfall variability, with comparable contribution from the Pacific and the Atlantic [Uvo et al., 1998], suggesting that the effect of other sources of variability, such as land surface processes and variability of frequency of transients from the South Atlantic may be also important in the interatmual rainfall variability in the region. For example, these processes could contribute to the interannual rainfall anomalies through changing wet season onset dates [Fu and Li, 2004; Li and Fu, 2006].
panied by SST anomalies with atmospheric forcing in the southwest Atlantic, with a dipole structure at about 40 0 S. This variability is independent of the ENSO. Doyle alld Barros [2002] suggested a positive feedback in the interannual scale between positive (negative) SST anomalies in the western part of the subtropical Atlantic and weak (intense) intensity of the SACZ that intensifies the SACZ low-level circulation. To understand the coupling between the SACZ and the South Atlantic, Chaves and Nobre [2004] carried out a series of experiments with atmospheric and ocean models. These results suggest that negative SST anomalies, generally observed in the SACZ region, represent a response of the ocean to the atmospheric forcing.
atmospheric teleconnections or changes in the Hadley cell [Robertson et al., 2000]. There are some observational evidence linking Amaze11 River discharge and NAO index. The correlations are hi,ghest at interannual (5-6 years) and interdecada1 timescaJes, and the NAO leads river discharge by about 9 months [Obregon and Nobre, 2004]. Considering that deforested areas in Amazonia are located mostly over the southern portions of the basin and that this region presents a long-term, positive trend in rainfall, it is possible to conjecture that this trend is more likely to be associated with the interdecadal variability related to the PDO, to a lesser extent to the NAO and less to deforestation. 5.4. Possible Influence ofBiomass Burning Aerosols on the Monsoon Transition
5.1. Ocean Modulation ofCoastal Climate
Sea-breeze circulations and squall lines that propagate over Amazonia fi'om the Atlantic coast constitute a complex system where scale interactions range from the large-scale enviromnenta1 characteristics, to the mesoscale and cloudscale circulations. Such systems are the key rain-producing mechanisms explaining the precipitation maximum over the Atlantic coast. At the mesoscale, the propagating and nonpropagating squall lines are basically initiated by the sea breeze circulation [Kousky, 1980], while the cloud-scale circulations maintain the squall-line propagation in a quasisteady state. For example, the mesoscale circulation associated with the maritime breeze may organize the convection into coastal squall lines, which are responsible for a significant portion of the precipitation over the eastern Amazonia [Garstang et al., 1994; Cohen et al., 1995]. The interaction between the large-scale circulation and the maritime breeze circulation also determines the time of day for the precipitation events on the nOliheastern coast of South America [Kousky, 1980; Negri et al., 2000]. 5.2. Atmospheric and Oceanic Controls ofthe SACZ
The SACZ connects atmospheric processes at low latitudes with those of the subtropics and mid-latitudes and, thus, is a mechanism allowing variability due to mid-latitude internal atmospheric dynamics to influence directly Amazonian climate variability and vice versa. Tropical convection is a key factor on both the onset and the maintenance of the SACZ through the latent heat release in the Amazon region [Kodama, 1992; Liebmann et al., 1999]. Several studies have analyzed the influence ofthe ocean surface temperature anomalies on SACZ variability. Barros et al. [2000] indicated that warm (cold) SST in the region between 20-40 0 S and west of 300 W is followed by a southerly (northerly)
5.3. Decadal and Interdecadal Variability: Relationship to PDO and NAO
Historical records of rainfall in Amazonia show decadal and interdecadal variability [Dias de Paiva and Clarke, 1995; Chu et al., 1995; ZhOli and Lau, 2001; Matsuyama et al., 2002; Marengo, 2004; Botta et al., 2002; Chen et al., 2003]. At the regional scale, a slight decreasing trend is observed over northern Amazonia, while the southern part presents a positive trend, and apparent climate shifts were identified around the middle of the decades of the 1940s and 1970s [Marengo, 2004]. After 1975, the northern/southern part of Amazonia shows relatively less/more rainfall when it is compared with the former period [Obregon and Nobre, 2003, Marengo, 2004].
These climate shifts are related to the change in both atmospheric and oceanic circulation over the North Pacific which took place in 1975-1976 associated with the Pacific Decadal Oscillation (PDO). During the period after 1975, apparently associated with the positive phase ofPDO, there was less rainfall over northern Amazonia linked to more frequent and intense E1 Nifios (1982-1983, 1986--1987, 19901991, 19,97-1998). Another important potential mechanism of long-term climate variability in Amazonia is the North Atlantic Oscillation (NAO). Annual variation ofrainfall seems to be linked, at least indirectly, to the NAO because strong Atlantic trades bringing moisture into Amazonia are associated with southward-displaced ITCZ, which is in turn related to an anomalous distribution of Atlantic SST [Moura and Shukla, 1981; Nobre and Shukla, 1996]. Rajagopalan et al. [1999] present statistical evidence that SSTs in the subtropical South Atlantic are associated with variations in the NAO. This connection might work via the impact of South Atlantic SSTs on Amazonian rainfall; the latter influencing the NAO via
In addition to a large number of studies addressing the influences of biomass burning aerosols on surface radiation budget, clouds and rainfall [e.g., Artaxo et al., 2002; Andreae et al., 2004], summarized in chapters B5 and B6 of this book, several recent studies have explored the impacts of biomass burning aerosols on the climate variabiJity of the monsoon transition. For example, Zhang et al. [2008] and Liu [2005] have suggested that the radiative effect, including both direct and semidirect effects, of the biomass burning aerosols could delay and weaken the transition to the summer monsoon circulation based on regional climate model simulations forced by aerosol radiative forcing estimated from observations from the Smoke Aerosols, Clouds, Rainfall and Climate (SMOCC) field campaign and the Moderate Resolution Imaging Spectroradiometer (MODIS), respectively. An analysis of the MODIS aerosols and cloud data by Yu et al. [2007] also suggests that the aerosol-cloud relationship changes interannually with climate condition, such that it could amplify or reinforce the original climate anomalies. In patiicular, during an anomalously dty transition season, warm cloud fi'action decreases with aerosol optical depth. Whereas in a normal and relatively wet transition season, warm cloud fraction increases with aerosol optical depth. Although whether these aforementioned results could be generalized remains unclear, they nevertheless suggest a potentially significant impact of biomass burning aerosols on climate variabilities of the monsoon circulation transition.
6. PALEOCLIMATE IN AMAZONIA DURING THE LAST GLACIAL MAXIMUM: WAS THE AMAZON DRIER DURING THE LGM? The short duration of instrumental records and the paucity of proxy records, means that little is known about either the character or the causes of longer (millennial to orbital)
157
timescale variations in tropical climate and their possible global teleconnections. There is, however, a small variety of paleoclimatic time series available for tropical South America. Many of these exist for the Altiplano of Bolivia, Peru, and Colombia, where the most important archives include sediment cores fi'om fluvial deposits, lacustrine sediments and pollen, and salar (salt flat) environments, as well as ice cores fi'om tropical glaciers. These records reveal largeamplihlde climate changes with a range of periodicities. Based on the poshllate that the ocean-atmosphere interactions that influence modern interannual climatic variability in tropical South America also influenced climatic variability on millennial and orbital timescales, climate-sensitive tropical ecosystems can provide important infOlmation that may help us to fill the gaps in our knowledge concerning the evolution of rainforests during periods of full glaciation. Small changes in precipitation in the Amazon basin have immediate consequences for the survival of the Andean cloud forest because its dominant source of moisture today is the Atlantic Ocean. Research on plant communities [Colinvaux et al. 2000] and pollen records fi'om the Amazon fan [Haberle and Maslin, 1999] have concluded that Amazonia was drier during the LGM. Mourguiart and Ledl'u [2003], shldying a 40,000year lacustrine record from the Eastern Cordillera in Bolivia in an endemic speci~s-rich and ecologically threatened region found a dry LGM, indicating a drastic decrease of the Amazonian moisture squrce. To explain this aridity, they infer steep temperature ~radients between the pole and equator in both hemispheres that would have reduced considerably the size and displacement of the ITCZ and the austral summer precipitation. This major change in water supply induced a dramatic reduction in species diversity and suggests that the Andean cloud forest did not provide refugia for tropical lowland taxa during full glacial times. In contrast, sedimentary records of Lake Titicaca reveal that the Altiplano was wetter during the LGM than in modern climate [Bakel' et al., 2001] and that has led to the inference that the Amazon was also wetter. They make use of three physical mechanisms to explain LGM wetness in Amazonia. First, wet season insolation was at a maximum in the southern tropics at 20,000 years B.P. [Baker et al., 2001]; thus, the SASM was maximized. Second, during the LGM, zonal (cold in the east [Bard et al., 2000], warm in the west [Riihlermann et al., 1999]) and meridional (cold in the north, warm in the south [Mix et al., 1999]) SST gradients in the equatorial Atlantic were favorable for enhanced SST forcing of the northeast trades and atmospheric advection of water vapor into Amazonia [Bakel' et al., 2001]. Third, lower equatorial Atlantic SST favored increased gradients between land and sea-surface temperature during the austral summer, also enhancing water vapor transport into Amazonia.
158
NOBRE ET AL.
CHARACTERISTICS OF AMAZONIAN CLIMATE
There are still many unresolved issues to determine whether Amazonia was drier and may have seen a decrease in the areas covered by forests, or wetter and fully covered by foresd One fundamental question, of course, is to see whether paleoclimatic data from the Altiplano is relevant to reveal the paleoclimate of Amazonia. It is necessary to believe that large-amplitude precipitation changes are synchronous and of the same sign for the Altiplano and Amazonia [Baker et al., 2001]. For instance, it is possible that the LGM was wetter than today in Amazonia, whereas most recent studies have concluded that Amazonia (and even the global tropics) were drier [Betancourt et a1., 2000; Argollo and Mourguiart, 2000; Heine, 2000; Thompson et al., 2002]. Whereas such millennial variability ofthe North Atlantic region is synchronous with millennial changes in tropical South America [Bakel' et a1., 2001], recent studies have concluded that the Younger Dlyas, for example, was only a northern hemisphere event [Bennett et al., 2000]. Interestingly, a recent study concluded just the opposite: that Amazonia was drier during the Younger Dryas [Maslin et al., 2000]. To add more complexity to the supposed persistent relationship between Atlantic SST anomaly patterns and rainfall variability over the Altiplano, a detailed modern climatological study [Vuille et al., 2000b] concluded that rainfall variability in the central Andes is not correlated with Atlantic SSTs. Recently, the climate of the LGM over South America without remote influences was simulated using a Regional Climate Model [Cook and Vizy, 2006]. Results showed that in the Amazon basin during the LGM, the calculated rainfall was 25~ 35% lower than in the present day simulations. In that model simulation, the 2- to 4-month delay in the onset of the rainy season was due to a drier low-level inflow from the Atlantic onto South America in comparison to present day climate. We cmmot overlook the role of deforestation on Amazonian climate or the diverse roles ofland use/land cover change on precipitation [Pielke et al., 2007]. Most of these studies find that deforestation leads to significant reduction in rainfall over Amazonia [Werth and AVissar, 2002; Oyama and Nobre, 2003; Scunpaio et a1., 2007; Sampaio, 2008]. All these land use changes would weaken the hydrological cycle in Amazonia and globally. The detailed analysis of Amazonian deforestation and climate is presented in chapter B7 ofthis book.
Acknowledgment. The contribution of one of the authors (G. Poveda) is part of the GRECIA Program, funded by COLCIENCIAS of Colombia. REFERENCES Aceituno, P. F. (1988), On the functioning of the Southern Oscillation in the South American Sector. Part I: Smface climate, JY!on. Weather Rev., 116, 505~524.
Amerasekera, K. N., R. F. Lee, E. R. Williams, and E. A. B. Eltahir (1997), ENSO and the natural variability in the low of tropical rivers, J. Hvdrol., 2000, 24-39. Andreae, M. 0, D. Rosenfeld, P. Artaxo, A. A. Costa, G. P. Frank, K. M. Longo, and M. A. F. Silva Dias (2004), Smoking rain clouds over the Amazon, Science, 303, 1337-134 J. Argollo, 1., and P. Mourguiart (2000), Late Quaternary climate history ofthe Bolivian Altiplano, Quat. Int., 72,37-51. Artaxo, P., 1. V. Martins, M. A. Yamasoe, A. S. Proc6pio, T. M. Pauliquevis, M. O. Andreae, P. Guyon, L. V. Gatti, and A. M. Cordova Leal (2002), Physical and chemical properties of aerosols in the wet and dty season in Rondonia, Amazonia, J. Geophys. Res., I07(D20), 8081, doi: IO.l02912001JD000666. Baker, P. A., G. O. Seltzer, S. C. Fritz, R. B. Dunbar, M. 1. Grove, P. M. Tapia, S. L. Cross, H. D. Rowe, and J. P. Broda (2001), The histoty of South American tropical precipitation for the past 25,000 years, Science, 291, 640-643. Bard, E., R. Rostek, 1.-L. Turon, and S. Gendreau (2000), Hydrological Impact of Heinrich events in the Subtropical Northeast Atlantic, Science, 289, 1321-1324. Barros, V. R., M. Gonzalez, and 1. Camilloni (2000), Influence of the South Atlantic convergence zone and South Atlantic Sea surface temperature on interannual summer rainfall variability in Southeastern South America, The~r. Appl. Climatol., 67, 123-133. Bennett, D. K., S. G. Haberle, and S. H. Lumley (2000), The last Glacial-Holocene transition in Southern Chile, Science, 290, 325-328. Betancourt, J. L., C. Latorre, 1. A. Rech, 1. Quade, and K. A. Rylander (2000), A 22,000-year record ofmonsoonal precipitation from northern Chile's Atacama Desert, Science, 289, 1542-1546. Biasutti, M., D. S. Battisti, and E. S. Sarachick (2003), The annual cycle over the tropical Atlantic, South America, and Afhca, J. Clim., 16,2491-2508. Biasutti, M., D. S. Battisti, and E. S. Sarachick (2004), Mechanism controlling the annual cycle of precipitation in the Atlantic sector in an atmospheric GCM, J. Clim., 17,4708-4723. Botta, A., N. Ramankutty, and 1. A. Foley (2002), Long-term variations of climate and carbon fluxes over the Amazon basin, Geophys. Res. Lett., 29(9),1319, doi:lO.10291200IGLOI3607. Campetella, C. M., and C. S. Vera (2002), The influence of the Andes mountains on the South American low-level flow, Geophys. Res. Lett., 29(17), 1826, doi: 10.1 029/2002GLOI5451. Caviectes, C. (1981), Rainfall in South America, seasonal trends and spatial correlation, Erdkunde, 35, 107-118. Chaves, R. R., and P. Nobre (2004), Interactions between sea surface temperature over the South Atlantic Ocean and the South Atlantic Convergence Zone, Geophys. Res. Lett., 31, L03204, doi: 10.1029/2003GLOI8647. Chen, T. c., S. Takle, 1. H. Yoon, S. T. K. Croix, and P. Hsieh (2003), Impacts on tropical South America Rainfall due to changes in global circulation, paper presented at 7th International Conference on Southern Hemisphere Meteorology and Oceanography, Wellington, New Zealand. Chou, C, and 1. D. Neelin (2001), Mechanisms limiting the southward extent of the South American summer monsoon, Geophys. Res. Lett., 28, 2433-2436.
Chu, P. S., Z.-P. Yu, and S. Hastenrath (1995), Detecting climate change concurrent with deforestation in the Amazon basin: Which way has it gone?, 81m. Am. Meteorol. Soc., 75,579-583. Cohen, 1. C. P., M./fL F. Silva Dias, and C. A. Nobre (1995), Environmental condifions associated with Amazonian squall line: A case study, M611. Weather Rev., 123,3163-3174. Colinvaux, P. A., P. E. De Oliveira, and M. B. Bush (2000), Amazonian and neotropical plant communities on glacial time-scales: The failure of the aridity and refugc hypotheses, Quat. Sci. Rev., 19,141-169. Cook, K. H., and E. K. Vizy (2006), South American climate during the Last Glacial Maximum: Delayed onset of the South American monsoon, J. Geophys. Res., 111, D0211O, doi:10.l029/ 2005JD005980. Culf, A. D., 1. L. Esteves, A. O. Marques Filho, and H. ROcha (1996), Radiation, temperature and humidity over forest and pasture in Amazonia, in Amazonian Deforestation and Climate, edited by 1. H. C. Gash, C. A. Nobre, and 1. M. Roberts, pp. 175-191, Wiley, New York. Dias De Paiva, E. M. C., and R. Clarke (1995), Time trends in rainfall records in Amazonia, Bull Am. Meteorol. Soc., 75,579-583. Dickinson, R. (1987), The Geophysiology ojAmazonia: Vegetation and Climate Interactions, 526 pp., Jolm Wiley, New York. Doyle, M. E., and V. R Barros (2002), Midsummer low-level circulation and precipitation in subtropical South America and related sea surface temperature anomalies in the South Atlantic, J. Clim., 15, 3394-3410. Eltahir, E. A. B., and R. L. Bras (1996), Precipitation recycling, Rev. Geophys., 34, 367-378. Enfield, D. B. (1996), Relationships of inter-American rainfall to tropical Atlantic and Pacific SST variability, Geophys. Res. Lett., 23,3305-3308. Figueroa, N., and C. A. Nobre (1990), Precipitation distribution over Central and Western Tropical South America, Climanalise, 5,36-48. Figueroa, N., P. Satyammty, and P. L. Silva Dias (1995), Simulations of the summer circulation over the South America region with an eta coordinate model, J. Atmos. Sci., 52, 1573-1584. Fu, R., and W. H. Li (2004), Influence ofland surface on transition from dry to wet season over the Amazon, TheOl·. Appl. Climatol., 78,97-110. Fu, R, B. Zhu, and R. E. Dickinson (1999), How do atmosphere and land smface influence seasonal changes of convection in the tropical Amazon?, J. Clim., 12,1306-1321. Fu, R., M. Chen, W. Li, and R. E. Dickinson (2001), How do tropical sca surface temperatures influence the seasonal distribution of precipitation in the equatorial Amazon?, J. Clim., 14,40034026. Garreaud, R, and P. Aceituno (2001), Interannual rainfall variability over the South American Altiplano, J Clim., 14, 2779-2789. Garreaud, R., and J. M. Wallace (1998), Summertime incursions' of midlatitude air into subtropical and tropical South America, Mon. Weather Rev., 126,2713-2733. Garstang, K., L. Massier Jr., 1. Halverson, S. Greco, and J. Scala (1994), Amazon coastal squall lines. Part I: Stmcture and kinematics, Mon. Weather Rev., 122, 608-622.
"
159
Giannini, A., 1. C. H. Chiang, M. Cane, Y. Kushnir, and R. Seager (200 I), The ENSO teleconnection to the tropical Atlantic Ocean: Contributions of the remote and local SST's to rainfall variability in the tropical Americas, J. Clim., 14, 4530-4544. Grimm, A. M. (2003), The El Nifio impact on the summer monsoon in Brazil: Regional processes versus remote influences, J. Clim., 16, 263-280. Grimm, A. M. (2004), How do la Nifia events disturb the summer monsoon system in Brazil?, Clim Dyn., 22, 123~138. Grimm, A. M., and P. L. Silva Dias (1995), Analysis of tropicalextratropical interactions with influence functions of a barotropic model, J. Atmos. Sci., 52, 3538-3555. Guyot, J. L., 1. Callede, M. Molinier, W. Guimaraes, and E. de Oliveira (1998), La variability hydrologique actuelle dans Ie bassin amazonien, Bull. Inst. Fl'. Etudes Andines, 27, 779-788. Haberle, S. G., andM. A. Maslin (1999), Late Quatem31yvegetation and climate change in the Amazon basin based on a 50,000 year pollen record fi'om the Amazon Fan, ODP Site 932, Quat. Res., 51, 27-38. Hamilton M. G., and 1. R. Tarifa (1978), Synoptic aspects of polar outbreak leading to fi'ost in tropical Brazil, Mon. Weather Rev., 106,1545-1556. Hastemath, S. (1990), Prediction of northeast Brazil rainfall anomalies, J. Clim., 3, 893-904. Hastemath, S., and L. Greischar (1993), Further work on the prediction of northeast Brazil rainfall anomalies, J. Clim., 6, 743~758. Hastenrath, S, and L. Heller (1977), Dynamics of climatic hazards in Northeast Brazi~, Q. J. R. Meteorol. Soc., 103, 77-92. Hastenrath, S., M. C.'Wu, and P. S. Chu (1984), Towards the monitoring and predictjon of nOlih-east Brazil droughts, Q. J. R. Meteol'ol. Soc., 110,111-425. Heine, K. (2000), Tropical South America during the Last Glacial Maximum: Evidence from glacial, periglacial and fluvial records, Quat. Jnt., 72,7-21. Hore!, L. D., A. N. Hah111a11l1, and 1. E. Geisler (1989), An investigation of convective activity over the Tropical Americas, J. Clim.,2, 1388-1403. Hsu, C. P., and 1. M. Wallace n976), The global distribution of annual and semiannual cycles in precipitation, Mon. Weather Rev., 104, 1093-110 I. Instituto Nacional de Meteorologia~INMET (1992), NOl1nais Climatol6gicas (1961-1990), report, p. 84, Departamento Nacional de Meteorologia, Brasilia, Brazil. Jipp, P. H., D. C. Nepstad, D. K. Cassel, and C. Reis de Carvalho (1998), Deep soil moisture storage and h'anspiration in forests and pastures ofseasonally-chy Amazonia, Clim. Change, 39, 395-412. Kalney, E., K. C. Mo, and 1. Paegle (1986), Large-amplitude, Sh01iscale stationa1y Rossby waves in the Southern Hemisphere: Observations and mechanistic experiments to determine their origin, J. Atmos. Sci., 43, 252-275. Kiehl, 1. T. (1994), On the observed near cancellation between longwave and shortwave cloud forcing in tropical regions, J. Clim., 7,559-565. Kodama, Y. M. (1992), Large-scale COl111l1on features of subtropical precipitation zones (the Baiu fi'ontal zone, the SPCZ and the SACZ), Part I, Characteristics of sub-tropical frontal zones, J. Meteol'Ol. Soc. Jpn., 70,813-836.
160
CHARACTERISTICS OF AMAZONIAN CLIMATE
Kodama, Y. M. (1993), Large-scale common features of sub- Marengo, J. A, C. Nobre, and A Culf(1997b), Climatic impacts of Friagens in forested and deforested areas of the Amazon Basin, tropical convergence zones (the Baiu frontal zone, the SPCZ, J. Appl. Met., 36,1553-1566. and the SACZ). Part II: Conditions of the circulations for generMarengo, J. A, J. Tomasella, and C. R. Uvo (1998), Trends in streamating the S'fCZs, J. Meteorol. Soc. Jpn., 71,581-610. flow and rainfall in h'opical South America: Amazonia, eastern Brazil, Koster, R. D., et al. (2004), Regions of strong coupling between and northwestern Pem, J. Geophys. Res., 103(D2), 1775-1783. soil moisture and precipitation, Science, 305,1138-1140. Kousky, V. E. (1980), Diurnal rainfall variation in the northeast Marengo, J. A, B. Liebmann, V. E. Kousky, N. P. Filizola, and 1. C. Wainer (2001), Onset and end of the rainy season in the Brazil, Mon. Weather Rev., 108,488-498. Brazilian AnJazon basin, J. Clim., 14, 833-852. Kousky, E. (1988), Pentad outgoing longwave radiation climatology Marengo, J. A, W. Soares, C. Saulo, and M. Nicolini (2004), Clifor the South American sector, Rev. Bras. Meteorol., 3, 217-231. matology ofthe LLJ east ofthe Andes as derived from the NCEP Kousky, V. E., and M. A Gan (1981), Upper tropospheric cyclonic reanalyses, J. Clim., 17, 2261-2280. vOliices in the tropical south Atlantic, Tellus, 33A, 538-550. Kreuels, R., K. Fraedrich, and E. Ruprecht (1975), An aerological Marengo, J. A, C. A. Nobre, J. Tomasella, M. Oyama, G. Sampaio, R. de Oliveira, H. Camargo, I.. M. Muniz, and 1. F. Brown (2008a), climatology of South America, Meteor. Rundsch., 28, 17-24. The drought of Amazonia in 2005, J. Clim., 21, 495-516. Lau, K. M., and J. Zhou (2003), Anomalies of the South America summer monsoon associated with the 1997-1999 El Nifio- Marengo, J. A, C. A Nobre, J. Tomasella, M. F. Cardoso, and M. D. Oyama (2008b), Hydro-climatic and ecological behaviour Southern Oscillation, Jnt. J. Climatol., 23,529-539. of the drought of Amazonia in 2005, Phi/os. Trans. R. Soc. Ser. Li, W. H., and R. Fu (2006), Influence of cold air intmsions on the B, 363,1773-1778. wet season onset over Amazonia, J. Clim., 19, 257-275. Liebmann, B., and J. A Marengo (2001), Interannual variability of Maslin, M. A, E. Durham, S. J. Burns, E. Platzman, P. Grootes, S. E. Greig, M. Nadeau, M. Schleicher, U. pflaumann, and B. the rainy season and rainfall in the Brazilian Amazon basin, J. Lomax (2000), Palaeoreconstruction ofthe Amazon River freshClim., 14,4308-4318. water and sediment discharge using sediments recovered at Site Liebmann, B., J. Marengo, J. Glick, 1. Wainer, V. Kousky, and O. 942 on the Amazon Fan. J. Qua't. Sci., 15,419-434. Massambani (1998), Comparison of long wave radiation, divergence and rainfall in the Amazon basin on subseasonal scales, Matsuyama, H., J. Marengo, G. Obregon, and C. A. Nobre (2002), Spatial and temporal variability of rainfall in tropical South J. Clim., 7,2898-2909. America as derived from the Climate Prediction Center merged Liebmann, B., G. N. Kiladis, J. A Marengo, T. Ambrizzi, and J. D. analysis of precipitation, Jnt. J. Climatol., 22, 175-195, Glick (1999), Submonthly convective variability over South America Mechoso, C. R., S. W. Lyons, and J. A Spahr (1990), The impact and the South Atlantic Convergence Zone, J. Clim., 12, 1877-1891. of sea surface temperature anomalies on the rainfall over northLiu, Y. Q. (2005), Atmospheric response and feedback to radiative east Brazil, J. Clim., 3, 812-826. forcing from biomass burning in tropical South America, Agric. Meggers, B. (1994), Archeological evidence for the impact of For. Meteorol., 133, 40-53. Mega-El Nifio events on Amazonia during the past two millenMakarieva, A M., and V. G. Gorshkov (2007), Biotic pump of nia, Clim. Change, 28, 321-338. atmospheric moisture as driver ofthe hydrological cycle on land, Mix, A C., A E. Morey, N. G. Pisias, and S. W. Hostetler (1999), Hydro!. Earth Syst. Sci., 11,1013-1033. Foraminiferal faunal estimates of paleotemperature: CircumMarengo, J. A (1992), Interannual variability ofsmface climate in venting the no-analog problem yields cool ice age tropics, Palethe Amazon Basin, Jnt. J. Climatol., 12, 853-863. oceanography, 14, 350-359. Marengo, J. A (1995), Interannual variability of deep convection over the tropical South America sector as deduced from ISCCP Mo, K. C., and V. E. Kousky (1993), Further analysis of the relationship between circulation anomaly patterns and tropical conC2 data,Int. J. Climatol., 15,995-1010. vection,J. Geophys. Res., 98(D3), 5103-5113. Marengo, J. A (2004), Interdecadal variability and trends ofrainfall Molion, I.. C. B. (1993), Amazonian rainfall and its variability, in across the Amazon basin, Theor. Appl. Climatol., 78,79-96. Hydrology and Water Management in the Humid Tropics, edited Marengo, J. A (2005), The characteristics and variability of the by M. Bonel, M. M. Hufschmidt, and J. S. Gladwell, pp. 99-111, atmospheric water balance in the Amazon basin: Spatial and Cambridge Univ. Press, Cambridge. temporal variability, Clim. Dyn., 24,11-22. Marengo, J. A, and S. Hastenrath (1993), Case studies of extreme Molion, I.. C. B., andJ. C. de Moraes (1987), Osciladio SuI e descarga de rios na America do Sui Tropical, Rev. Bras. Eng., 5, 53-63. climatic events in the Amazon basin, J. Clim., 6, 617-627. Marengo, J. A, and C. Nobre (2001), General characteristics and Moura, A. D., and J. Shukla (1981), On the dynamics of droughts in NOliheast Brazil: Observations, theory and numerical experiments variability of climate in the Amazon basin and its links to the with a general circulation model, J. Atmos. Sci., 38, 2653-2675. global climate system, The Hydroclimatological Framework of Amazonia, Biogeochemistry of Amazonia, edited by J. Richey, Mourguiart, P., and M.-P. Ledru (2003), Last Glacial Maximum in an Andean cloud forest envirol1111ent (eastern Cordillera, BoM. MacClaine, and R. Victoria, pp. 17-41, Cambridge Univ. livia), Geology, 31,195-198. Press., New York. Marengo, J. A, A Cornejo, P. Satyamurty, C. A Nobre, and W. Sea Negri, A J., E. N. Anagnostou, and R. F. Adler (2000), A lO-yr climatology of Amazonian rainfall derived from passive micro(1997a), Cold waves in the South American continent. The strong wave satellite observations, J. Appl. Meteorol., 39, 42-56. event of June 1994, Moll. Weather Rev., 125,2759-2786.
V.
NOBRE ET AI.. Nepstad, D., A. Moreira, and A. Alencar (1999), A Floresta em Chamas: OrigellS, Jmpactos e Prevenr;iio de Fogo na Amazonia, Programa PilotP>para prote<;ao das Florestas Tropicales do Brasil, Brasilia, B)?asil, 202 pp. Nepstad, D., P. LeJebver, U. Lopes da Silva, J. Tomasella, P. Schlesinger, I.. Sdl6rzano, P. Moutinho, D. Ray, and J. Guerreira Benito (2004), Amazon drought and its implications for forest flammability and tree growth: A basin-wide amllisis, Global Change Bioi., 10, 704-717. Nobre, C. A (1983), The Amazon and climate, paper presented at Climate Conference for Latin America and The Caribbean, WMO, Paipa, Colombia. Nobre, P., and J. Shukla (1996), Variations of sea surface temperature, wind stress, and rainfall over the tropical Atlantic and South America, J. Clim., 9, 2464-2479. Nogues-Paegle, J., and K. C. Mo (1997), Altemating wet and dry conditions over South America during SUll'uner, Mon. Weather Rev., 125,279-291. Numaguti, A (1993), Dynamics and energy balance of the Hadley circulation and the tropical precipitation zones: Significance of the distribution of evaporation, J. Atmos. Sci., 50, 1874-1887. Obregon, G. O. (2003), Dinamica da Variabilidade climMica da precipita<;ao sobre a America do Sui, Ph.D. thesis, 172 pp., Instituto Nacional de Pesquisas Espaciais- INPE, Brazil. . Obregon, G., and C. A Nobre (1990), Principal Component Analysis applied to rainfall in Amazonia, Climanalise, 5, 35-46. Obregon, G., and C. A Nobre (2003), A climate shift in mid-l 970 in Northwest Amazonia and Southem Brazil, paper presented at 7th International Conference on Southern Hemisphere Meteorology and Oceanography, Wellington, New Zealand, edited by The American Meteorological Society, Boston, Mass., pp.88-89. Obregon, G. 0., and C. A. Nobre (2004), Hydrologic variability over the Amazon basin and its relationship with ENOS and NAO, paper presented at Conference of CLIVAR 2004, Baltimore, Mmyland USA, June, 2004. Oliveira, A. S. (1986), Intera<;5es entre sistemas frontais na America do Sui e a convec<;ao da Amazonia, M. S. thesis, 246 pp., Instituto Nacional de Pesquisas Espaciais - INPE, Brazil. Oyama, M. D., and C. A Nobre (2003), A new climate-vegetation equilibrium state for tropical South America, Geophys. Res. Lett., 30(23), 2199, doi:IO.l029/2003GLOI8600. Paegle, J. N., and K. C. Mo (2002), Linkages between summer rainfall variability over South America and sea surface temperature anomalies, J. Clim., 15, 1389-1407. Pezzi, I.. P., and 1. F. A. Cavalcanti (2001), The relative importance of ENSO and tropical Atlantic sea smface temperature anomalies for seasonal precipitation over South America: A numerical study. Clim. Dyn., 17,205-212. Pielke Sr., R. A, J. Adegoke, A Beltnin-Przekurat, C. A Hiemstra, J. Lin, U. S. Nair, D. Niyog, and T. E. Nobis (2007), An over- . view of regional land-use and land-cover impacts on rainfall, Tellus, 3B, 1-15. Poveda, G., and O. J. Mesa (1997), Feedbacks between hydrological processes in tropical South America and large scale oceanic atmospheric phenomena, J. Clim., 10, 2690-2702.
161
Poveda, G., and L. F. Salazar (2004), Arumal and interannual (ENSO) variability of spatial scaling properties ofa vegetation index (NOVI) in Amazonia, Remote Sens. Environ., 93, 391-401. Poveda, G., A Jaramillo, M. M. Gil, N. Quiceno, and R. I. Mantilla (2001), Seasonality in ENSO-related precipitation, river discharges, soil moisture, and vegetation index (NDVI) in Colombia, Water Resow'. Res., 37,2169-2178. Poveda, G., P. R. Waylen, and R. Pulwarty (2006), Annual and inter-annual variability of the present climate in northern South America and southern Mesoamerica, Palaleo, 234, 3-27. Rajagopalan, B., Y. Kushnir, and Y. M. Toun'e (1998), Obselved decadal midlatitude and tropical Atlantic climate variability, Geophys. Res. Lett., 25,3967-3970. Rao, V. B., and K. Hada (1990), Characteristics of rainfall over Brazil: Alillual variations and connections with the southem oscillation, Theor. Appl. Climatol., 42, 81-91. Rao, V. B., I. F. A Cavalcanti, and K. Hada (1996), Annual variations of rainfall over Brazil and water vapor characteristics over South America, J. Geophys. Res., 101(D21), 26,539-26,551. Ratisbona, C. R. (1976), The climate of Brazil. Climate of Central and South America, World Survey o.f Climatology, vol. 12, edited by W. Schwerdtfeger and H. E. Landsberg, pp. 219-293, Elsevier, Amsterdam. Richey, J. E., C. A Nobre, and C. Deser (1989), Amazon river discharge and climate variability: 1903 to 1985, Science, 246, 101-103. Robertson, A W., aqd C. R. Mechoso (2000), Interannual and interdecadal variability of the South Atlantic Convergence Zone, Mon. Weather Rev., 128, 2947-2957. Robertson, A W., c.iR. Mechoso, and Y.-J. Kim (2000), The influence of Atlantic sea surface temperature anomalies on the North Atlantic Oscillation, J. Clim., 13, 122-138. Ronchail, J., G. Cochoruleau, M. Molinier, J. L. Guyot, A G. D. M. Chaves, V. Guimaraes, and E. De Oliveira (2002), Interannual rainfall variability in the Amazon basin and sea surface temperatures in the equatorial Pacific and the tropical Atlantic Oceans, Jnt. J. Climatol., 22, 1663-1686. Ropelewski, C. F., and M. S. Halpert (1987), Global and regional scale precipitation patterns associated with the EI Nifio/Southem Oscillation, Mon. Weather Rev., 115, 1606-1626. Ropelewski, C. F., and M. S. Halpert (1989), Precipitation pattems associated with the high index phase of the Southern Oscillation, J. Clim., 2, 268-282. Ruhlemann, C., S. Mulitza, J. P. Muller, G. Wefer, and R. Zahn (1999), Warming of tropical Atlantic Ocean and slowdown of thermohaline circulation during the last glaciation, Nature, 402, 511-514. Salati, E. (1987), The forest and the hydrological cycle, in The Geophysiology o.f Amazonia, edited by R. E. Dickinson, pp. 273-293, John Wiley, New York. Salati, E., and J. Marques (1984), Climatology of the Amazon region, inAmazon: Limnology and Landscape Ecology o.fa Mighty Tropical River and its Basin, edited by H. Sioli, and W. Junk, Dordrecht, the Netherlands. Sampaio, G., C. Nobre, M. H. Costa, P. Satyamurty, B. S. SoaresFilI10, and M. Cardoso (2007), Regional climate change over eastern
162
CHARACTERISTICS OF AMAZONIAN CLIMATE
Amazonia caused by pasture and soybean cropland expansion, Geophys. Res. Lett. 34, Ll7709, dOl: 1O.1029/2007GL030612. Sampaio de Oliveira, G. (2008), ConseqLH~ncias climaticas da substitui<;ao graclttal da floresta Tropical Amazonica por pastagem degradada ou pOl' planta<;ao de soja: Um estudo de modelagem, Ph.D. thesis, 417 pp., Instituto Nacional de Pesquisas Espaciais, Brazil. Santos de Oliveira, A, and C. Nobre (1986), hlteractions betwen frontal syste;ms in South America and tropical convection over Amazon, paper presented at 2nd International Conference on Southern Hemispheric Meteorology, Wellington, New Zealand, 1-5 December. Schwerdtfeger, W. (1976), The atmospheric circulation over Central and South America, in Climates of Central and South America: World Survey of Climatology, vol. 12, edited by W. Schwerdtfeger, 532 pp., Elsevier, Amsterdam. Seluchi, M., and J. A. Marengo (2000), Tropical-midlatitude exchange of air masses during SUlmner and winter in South America: Climatic aspects and examples of intense events, Int. J. C!imatol., 20, 1167-1190. Seluchi, M., Y. V. Serafini, and H. Le Treut (1998), The impact ofthe Andes on transient atmospheric systems: A comparison between observation and GCM results, Mon. Weather Rev., 126, 890---912. Silva Dias, P. L., W. H. Schubert, and M. DeMaria (1983), Largescale response of the tropical atmosphere to transient convection, J. Atmos. Sci., 40, 2689-2707. Silva Dias, P. L., 1. P. Bonatti, and V. E. Kousky (1987), Diurnally forced tropical tropospheric circulation over South America, Mon. Weather Rev., 115, 1465-1478. Sternberg, H. (1987), Aggravation offloods in the Anlazon as a consequence of deforestation?, Geografiska Annalei', 69A, 201-219. Thompson, L. G., et al. (2002), Kilimanjaro ice core records: Evidence of Holocene climate change in tropical Africa, Science, 298, 589-593. Uvo, B. C., C. A Repelli, S. E. Zebiak, and Y. Kushnir (1998), The relationships between tropical Pacific and Atlantic SST and Northeast Brazil monthly precipitation, J. C!im., 11, 551-562. Vera, C., W. Higgins, 1. Amador, T. Ambrizzi, R. Garreaud, D. Gochis, D. Gutzler, D. Lettenmaier, 1. Marengo, C. R. Mechoso, et al. (2006), Toward an unified vision of the American Monsoon Systems, J. C!im., 19,4978-5000. Vilji, H. (1981), A preliminmy study of summertime tropospheric circulation patterns over South America estimated from cloud winds, Mon. Weather Rev., 109, 599--610. Vizy, E. K, and K H. Cook (2007), Relationship between Amazon and high Andes rainfall, J. Geophys. Res., 112, D07107, doi: 10.1029/2006JD007980. Yuille, M. (1999), Atmospheric circulation over the Bolivian Altiplano during dly and wet periods and extreme phases of the Southern Oscillation, Int. J. C!imatol., 19,1579-1600. Yuille, M., and M. Werner (2005), Stable isotopes in precipitation recording South American summer monsoon and ENSO variability~Observations and model results, C!im. Dyn., 25, 401-413. Yuille, M., R. S. Bradley, and F. Keimig (2000a), Climatic variability in the Andes of Ecuador and its relation to tropical Pa-
cific and Atlantic sea surface temperature anomalies, J. C!im., 13,2520-2535. Yuille, M., R. S. Bradley, and F. Keimig (2000b), Interannual climate variability in the central Andes and its relation to tropical Pacific and Atlantic forcing, J. Geophys. Res., 105(010), 12,447-12,460. Wang, H., and R. Fu (2002), Cross-equatorial flow and seasonal cycle of precipitation over South America, J. Clim., 15, 1591-1608. Wang, H., and R. Fu (2004), Influence of cross-Andes flow on the South American low-level jets, J. C!im., 17, 1247-1262. Wang, H., and R. Fu (2007), The influence of Amazon rainfall on the Atlantic ITCZ through convectively coupled Kelvin waves, J. C!im., 20,1188-1201. Werth, D., and R. Avissar (2002), The local and global effects of Amazon deforestation, J. Geophys. Res., 107(D20), 8087, doi:1 0.1 02912001 JD000717. Williams, E. R., and G. Satori (2004), Lightning, thermodynamic and hydrological comparison of the two tropical continental chimneys, J. Atmos. Sol. Terr. Phys., 66, 1213-1231. Williams, E., A Dall' Antonia, and V. Dall'Antonia (2005), The drought of the centUly in the Amazon Basin: An analysis of the regional variation of rainfall in South America in 1926, Acta Amaz., 35, 231-238. Xue, Y., F. De Sales, W. Li, C. R. Mechoso, C. Nobre, andH.-M. H. Juang (2006), Role of land surface processes in South American monsoon development, J. Clim., 19, 741-762. Yu, H., R. Fu, R. E. Dickinson, Y. Zhang, M. Chen, and H. Wang (2007), Interannual variability of smoke and warm cloud relationships in the Amazon as inferred from MODIS retrievals, Remote Sens. Environ., 111,435-449. Zeng, N., J. Yoon, and J. Marengo, A Subramaniam, C. Nobre, A Mariotti, and 1. D. Neelin (2008), Causes and impacts of the 2005 Amazon drought, Environ. Res. Lett., 3, 014002, doi: 10.1 088/1748-9326/3/1/014002. Zhang, Y., R. Fu, H. Yu, R. E. Dickinson, R. N. Juarez, M. Chin, and H. Wang (2008), A regional climate model study of how biomass burning aerosol impacts land-atmosphere interactions over the Amazon, J. Geophys. Res., 113, D14S15, doi:IO.l029/ 2007JD009449. Zhou, J., and K. M. Lau (1998), Does a monsoon climate exist over South America?, J Clim., 11, 1020---1040. Zhou, 1., and K. M. Lau (2001), Principal modes ofintermmual and decadaI variability of summer rainfall over South America, Int. J. Climatol., 21,1623-1644.
R. Fu, Jackson School of Geosciences, University of Texas at Austin, Austin, TX 78712, USA 1. A. Marengo, C. A Nobre, and G. O. Obregon, Centro de Ciencias do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, SP 12630, Brasil. ([email protected]) G. Poveda, Escuela de Geociencias y Medio Ambiente, Universidad Nacional de Colombia, Medellin, Colombia.
The Amazonian Boundmy Layer and Mesoscale Circulations A. K. Betts,l G. Fisch,2 C. von Randow,3 M. A. F. Silva Dias,4 1. C. P. Cohen,S R. da Silva,6 and D. R. Fitzjarrald7
The interactions between the Amazonian boundary layer, the surface, atmospheric convection, aerosols, and larger-scale circulations are complex. The field experiments inAmazonia have provided rich insights into the daytime and nighttime boundary layer in different regions and seasons over both forest and pasture and into the coupling between the surface fluxes, the boundary layer, precipitation, and cloud radiative forcing. We discuss the typical diurnal cycle of Amazonian convection, the self-organization into mesoscale systems in different synoptic regimes, and the role offorest and river breeze circulations. We review the coupling between aerosols, smoke, and convection in the dry season; ozone transpOlis by deep convection; and microphysical and electrical impacts on convection.
I. INTRODUCTION Understanding the complex interaction of clouds, rain, and the biosphere in the Amazon was the reason for the Large-Scale Atmosphere-Biosphere Experiment in Amazonia (LBA) experiment [Silva Dias et al., 2002a]. The focus of this chapter is the Amazonian boundary layer (BL) and its coupling with the surface, atmospheric convection, and larger-scale circulations. An early BL study using radiosoundings and tethered balloons was made during the
[Atmospheric Research, Pittsford, Vermont, USA. 2Instituto de Aeromiutica e Espa<;o, Sao Jose dos Campos, Brazil. 3Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos, Brazil. 4Department of Atmospheric Sciences, University of Sao Paulo, Sao Paulo, Brazil. 5Departamento de Meteorologia, Universidade Federal do Para, Belem, Brazil. 6UFPA-LBA Santarem, Alter do Chao, Brazil. 7Atmospheric Sciences Research Center, University at Albany, State University of New York, Albany, New York, USA. Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10. 1029/2008GM000725
Global Tropospheric Experiment Amazon Boundary Layer Experiment (ABL~ 2B) [Martin et al., 1988] at a tropical forest site near Ma,haus, followed by a longer study at contrasting forest and pasture sites during the Rondonia Boundary Layer Experiment (RBLE) [Nobre et al., 1996]. During the LBA experiment, two main field campaigns took place to characterize the structure and evolution of the BL [Fisch et al., 2004]. The first was during the wet season in early 1999: the Wet-Season Atmospheric Mesoscale Campaign (WETAMC) and the coincident Tropical Rainfall Measuring Mission (TRMM). The second field program was during the dry-to-wet transition season, September to November 2002; it also had two components, the study of Radiation, Cloud, and Climate Interactions (LBA-RACCI) and the study of the interaction of Smoke Aerosols, Clouds, Rainfall and Climate (LBA-SMOCC). The surface energy partition is controlled by the availability of water for evaporation, which depends on precipitation, soil moisture storage, and the vegetation cover [da Rocha et al., this volume]. Over the Amazon Basin, precipitation comes mostly from deep convection, which is coupled not only to the surface BL, surface topography, and heterogeneities (section 3.7) but also to circulations on continental and global scales [Nobre et al., this volume]. Deep convection is also self-organizing on the mesoscale along cold pool outflow boundaries (section 3). Horel et al. [1989] described, using reanalysis data and outgoing long-wave radiation 163
164
BETTS ET AL.
AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATIONS
165
\
from satellites, the movement of deep convection over South heterogeneity and mesoscale circulations must be important, America, which impacts the rainfall distribution over the both within the deforested area, where strips of forest remain tropical Americas. Machado et ai. [2004] showed the cli- (spacing of the order of 5 km), as well as between the forest matologicafaspects of the seasonal and diurnal variability and pasture regions, a distance of 90 km. of convection over Amazonia and its links to different vegetation types and large-scale forcing, using long-time series 2. WHAT IS DISTINCTIVE ABOUT THE AMAZONIAN BOUNDARY LAYER? of radiosonde and satellite data for cloud cover (from the International Satellite Cloud Climatology Project (ISCCP» Over land, the boundary layer goes through a strong diand the normalized difference vegetation index. Figure 1 shows that July is the driest U:onth for the majority of Bra- urnal cycle [e.g., Betts, 2003] as it links the heated surface zil (Figure la). The dry season duration in Figure Ib shows to the atmosphere during the day and uncouples from the two minima, one over west Amazonia, associated with the atmosphere at night. Since the BL couples the atmosphere to monsoon circulation and persistent convection, and another the surface on the daily timescale, its structure depends both in the south ofBrazil, associated with the penetration of cold on differences in the surface, the roughness and the availabilfronts. We shall give considerable emphasis to studies in SW ity of water for evaporation, and differences in the overlying Amazonia (specifically Rondonia and the Rio Madeira Ba- atmosphere. Over the moist forested Amazon Basin, the surface water storage is large (available soil water storage may sin) in section 2 because of the large seasonal cycle. Prior to LBA, the Rondonia Boundary Layer Experi- approach 800 mm, see Hodnett et ai. [1996] and Figure 4), ment (RBLE 3) collected data from representative regions and the surface vegetative conductance is correspondingly of forest and deforested pasture in the Ji-Paraml region of large (typical values are around 30 mm S-I [Wright et ai., Rondonia during the 1993 dry season [Nobre et ai., 1996]. 1996]). Consequently, the drop in relative humidity across Nobre et al. noted the much deeper daytime BL over the the leaf at the surface (which 'is related to the height of the pasture site, where the surface sensible heat flux is roughly lifting condensation level) is small, and so daytime cloud double that over the forest in the dry season. In contrast, the base heights are also small, typically reaching only 1000 m nighttime BL is deeper over the aerodynamically rougher [Betts et ai., 2002a], somewhat more in the dry season [Noforest. However, simple one-dimensional mixed layer mod- bre et ai., 1996]. In addition, the cooled nighttime stable BL els failed to capture the diurnal cycle of the growth of the generally saturates and is about 300 m deep over the aerodaytime unstable BL and underestimated the large differ- dynamically rough forest [Nobre et ai., 1996]. In southern ences in the afternoon BL depth between forest and pasture Amazonian regions, where there is a strong annual cycle of [Fisch et ai., 1996]. Fisch et al. suggested that the horizontal precipitation, there is sufficient available storage of water in
a) Driest Month
b) Dry season duration (Month)
.1-3 .4-6 0 7
.9 08
_10-12 •
Rainfall and NDVl observation box
Figure 1. (a) Driest month (month of minimum high cloud cover). (b) Dry season duration defined using high cloud cover. Adapted from Machado et af. [2004], reprinted with permission of Springer-Verlag.
the soil in forested areas that the drop in evaporation in the dry season and correstlbnding deepening of the dry adiabatic subcloud layer is sIOO11 (less than 20-30% [da Rocha et ai., this volume]). In c~tras.t, in deforested regions in Rondonia, where available~il moisture is much reduced, the evaporation falls sharply in the dry season, and the subcloud layer depth may reach nearly 2000 m in the afternoon. Over most of the Amazon Basin, the overlying tropical atmosphere is convectively coupled. In the rainy season, the entire troposphere is linked to a single moist adiabat, nearneutral stability for deep convection, with a minimum saturation equivalent potential temperature close to the freezing level in the middle troposphere. In the dry season in the southern Amazon, when there is subsidence aloft, a weak inversion caps a layer of shallow cumulus, and only the lower troposphere is convectively coupled on many days. However, in all seasons, a partly cloudy convective boundary layer, which is coupled to the surface, controls the surface radiative energy budget. We first look at atmospheric differences between seasons and then at the seasonal differences in the surface processes between forest and pasture in Rondonia. 2.1. Monthiy Mean Atmospheric Differences Between Wet and Dry Seasons
100
" 200
:S
,, , \
400
\ \ \
500
,, \
~
::J
(f) (f)
,,
,
\
600
" ....... ,
~
a.
700
~----------------
,,------- i
900 1000
,
Subsidence J
800
-0.03
"
ERA-40 1999 Madeira River basin
a ---''::'~~''''''----i
-0.05
-0.01
0.01
0.03
0.05
Omega (Pa s-l) 100
,,-------===:::~
,,
200 300
, ,, I
co
400
a. :S 500 ~
::J
The seasonal cycle over SW Amazonia is driven by the large-scale circulation [see Nobre et ai., this volume]. Figure 2a shows the monthly mean vertical omega field (vertical velocity in pressure units) over the Rio Madeira Basin, derived from hourly model data from the European Centre for Medium-Range Weather Forecasts (ECMWF) 40 Year Reanalysis (ERA-40) [Betts and Viterbo, 2005]. Both months show convergence and ascent near the surface, but in the rainy season, mean ascent is positive throughout the troposphere, peaking around 300 hPa. This destabilization is, of course, balanced by heavy convective precipitation. During the dry season, subsidence dominates, peaking at around 700 hPa, and this suppresses deep convection. The result is the dramatic difference in the mean relative humidity (RR) of the troposphere, shown in Figure 2b (as well as in precipitation and cloud cover, not shown). In February in the rainy season, RR is above 70% in the lower troposphere, while in August, mean RR falls steeply above 700 hPa to about 20% in the iniddle troposphere. As a result ofthis different tropospheric forcing, the stabil- . ity of the atmosphere with regard to moist convection is very different in the two seasons, as shown in Figure 3, which shows the mean vertical profiles of equivalent potential temperature, eE, and eES its saturation value, which represents temperature in this moist adiabatic reference system. The
_uu
" ,,
300
co a.
--Feb Aug
'" ,
(f) (f)
I
I I I I
,
,,
\
600
.......
I
'~r~~~~~~~~~
~
a.
I
--Feb Aug
___ u
I
700
~~
,
900 1000
,,
'"
800 I
\
b
I
I
I
I
I
I
... _-
+--.-.--.-r-,--,---r-,-----.---r----.--.-~~....--l
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
RH Figure 2. (a) Vertical profiles of monthly mean vertical velocity field for February and August 1999 for Rio Madeira Basin from ERA-40. (b) Same as Figure 2a but for RH profiles.
dotted lines represent moist adiabatic parcel ascent from the 30-hPa layer near the surface. In February, this lifted parcel is buoyant through much of the troposphere. In contrast, in August, the daily mean parcel is never buoyant if lifted. Indeed, in the dry season, the shallow cumulus convection occurs only in the daytime, when the surface heating is strong, and it is generally trapped below the temperature inversion from 700 to 600 hPa, which, in tum, is maintained by the subsidence shown in Figure 2a. Rain events in the dry season typically occur with the intrusion of cold fronts from the extratropics [Fisch et ai., 2004]. Note that the subsidence in Figure 2a peaks at the base of the inversion, where
166
AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATIONS 100
(?
200
ERA-40 1999 Madeira River basin
3c!O
---DeE (Feb)
400
Q.
::S 500 ~'
:::J (J) (J)
-eES(Feb) ----- e E (Aug) - - - e ES (Aug)
600
~
Q.
700 800 900 1000 +----.--,----,--..,----.----,-----,,.-r-.----,---.-.,-,....---j 320 330 340 350 360 370 380 390 e E, e ES (K)
Figure 3. Monthly mean vertical profiles of8E and 8ES for FeblUaty and August 1999 for Rio Madeira Basin fi'om ERA-40 reanalysis.
the shallow cumulus detrain their liquid water and cool the atmosphere. 2.2. SlI1.face Seasonal Differences Between Forest and Pasture
Forest and pasture respond differently to similar atmospheric forcing [Culfet al., 1996]. von Randall' et al. [2004] show seasonal comparisons between pasture (Fazenda Nossa Senhora) and forest (Rebio Jaru) sites in Rondonia. Monthly averages of air temperature and specific humidity and monthly totals of rainfall measured over the pasture and forest sites during the period Februaty 1999 to September 2002 are shown in Figure 4a. The range in monthly mean air temperature is small, and it is not easy to identify a clear seasonal pattern. Strong precipitation events in Janua1y-Februaty and penetrating cold fronts in June-July influence the averages in some years. On the other hand, a clear drop in specific humidity and a drastic reduction in rainfall during the dry seasons are observed at both sites. Rainfall amounts were higher at the forest site (2000~2400 nUll a-I) than at the pasture site (1400-2000 111111 a-I), but it is unclear whether these differences are truly representative oflarger areas [von Randall' et al., 2004]. However, Ferreira da Costa et al. [1998] compared rainfall fi'om contrasting sites (forest and pasture) at Rondonia during four rainy seasons (DecemberFebruary) from 1992 to 1995 and also found 28% greater rainfall over forest than over pasture. Fisch et al. [2007] analyzed the spatial variation of the convective rainfall in this
region, using the TRMM rain gauge network. They used a statistical model of a rainfall event and concluded that at a distance greater than 5 km, the measured rainfall from typical convective cells had a poor cOlTelation (0.2 ~ 0.4). The specific humidity is also always higher in the forest area, with average values ranging from 15.8 g kg-I in the d1y seasons to 17.5 g kg-I in the wet seasons, while at the pasture, the average values are 13.4 and 16.0 g kg- l in the d1y and wet seasons, respectively. Figures 4b and 4c show the storage of water for the layers fi'om 0 to 2 m and 2 to 3.4 m, respectively, in the soil profile. Both forest and pasture show a very pronounced seasonal cycle, with much larger d1y season decreases in the forest than the pasture, especially in the lower layer, where the deep root extraction is much greater in the forest [von Randow et al., 2004; Negron Juarez et al., 2007]. Consequently, over the forest, the surface Bowen ratio, given by ratio BR = HIAE (where Hand AE are the surface sensible and latent heat fluxes respectively) varies little, increasing slightly in the range 0.3-0.4 from wet to dry seasons. However, over the pasture, where there is little deep root extraction, BR increases from ab'out 0.4 in the rainy season [Betts et al., 2002a] to 0.65 in August [von Randall' et al., 2004]. Galviio and Fisch [2000] reported BR increases from 0.21 to 0.3 over the forest and from 0.32 to 0.76 over the pasture between the wet and dry seasons. There remains considerable unce1tainty in these surface flux estimates because of the lack of energy balance closure using eddy correlation methods [von Randall' et al., 2004]. In a later study combining eddy cOlTelation and scintillometly measurements, von Randall' et al. [2008] show that the eddy fluxes do not fully capture the spatial variability ofturbulence, especially at low frequencies. As this affects both Hand AE, it is likely that the percent unce1tainty in the Bowen ratio is smaller than in the individual fluxes. Over forest sites, where deep soil water is available even through a d1y season, evapotranspiration depends largely on the surface net radiation [Negron Juarez et al., 2007; Hasler and AVissar, 2007]. The surface radiative budget is thus a critical component of the surface forcing of the BL. Figure 5 (top) [von Randm\' et al., 2004] shows the mean annual cycle of surface albedo; Figure 5 (bottom) shows the net short-wave flux, Sn, net long-wave flux, L n, and net all-wave radiation flux, R n , over forest and pasture. The solar radiation reflected by the pasture is larger throughout the year and reaches a peak early in the dry season, much earlier than the lower forest peak in albedo. Several factors contribute to the seasonal variation in incident solar radiation (not shown): solar zenith angle, cloud cover, which peaks in the rainy season, and smoke from burning during the dry season. However, the albedo differ-
BETTS ET AL. '7 OJ
..>::
28
-9
£
u
'E :::J
.c u
t;:::
'0 Q)
Cl.
2 91"
1+
~
00
22 20
o
o
. '0 . ~ .0
700
'00'00'0,' 0 "
600
0
\;
500
(J)
u c ro
18
0
16
'0 ~
-
14
~
12
'l"V.~ "V'v. \ "\. I
Q)
NT
..v;~
l':'y.
....
\~ r:rf.
v
J
.rv~\VT..
\}? " V '. / ~
Cl.
E
2
~
E
g
fi
'. Vl
."\: ¥?.
'\
::J
167
400
"
.,
T
v\rr"" v \ 'trv.vv
E 5 c 0
~
300
:t=
200
0..
Cl.
'0 Q)
L-
100
10 0 Jan/99
Jul/99
Jan/OO
Jul/OO
Jan/01
Jul/01
Jan/02
Jul/02
Jan/03
800 600
'0 (f)
.S:
400
'--
())
co 200
+-'
:s:
'+--
0
())
800
(c)
2.0 - 3.4 m
-e..- Pasture
0>
~
0
-.0.-
600
Forest
C /)
400 200 Jan/99
Jul/99
Jan/OO
Jul/OO
Jan/01
Jul/01
Jan/02
Jul/02
Jan/03
Figure 4. (a) Monthly averages of air temperature (circles), specific humidity (inverted triangies), and monthly totals of precipitation (columns) measured over the forest (represented by solid symbols) and pasture (represented by open symbols) from FeblUaty 1999 until September 2002. Storage of water in the soil at forest and pasture sites for the layers fi'om (b) 0-2 m and (c) 2-3.4 m deep. Adapted from von Randow et at. [2004], reprinted with permission of Springer-Verlag.
ence dominates the difference in Sn between forest and pasture. Outgoing L n has a large seasonal variation, peaking in August when the atmosphere is d1y (Figure 2b) and cloud cover is a minimum. Outgoing L n is slightly larger over the pasture primarily because of warmer mean daytime temperatures. Throughout the year, Rn is greater over forest because of its lower albedo and slightly smaller outgoing long-wave flux. The mean al1l1ual reduction in Rn over the pasture is 13.3% [von Randow et al., 2004], smaller than the 20.4% given for the boreal forest by Betts et al. [2007], where snow cover also reduces R n over grass in late winter.
2.3. Diurnal Cycle ofthe Amazonian BL
The daytime convective BL over Amazonia is rarely cloud-free, so the depth of the mixed layer below cloud base, the lifting condensation level and the near-surface RH are all tightly coupled [see Betts et al., 2006]. Near-surface RH is strongly influenced by the availability of water for evaporation, so forest sites where rooting is deep (Figure 4) and pasture'sites show larger differences in the d1y season than in the wet season. Afternoon mixed layer heights range from 700 to 1100 m in the rainy season over forest and pasture, when
168
AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATIONS
Figure 5. Monthly averages of (top) surface albedo and (bottom) net short-wave radiation (Sn, circles), net long-wave radiation (L n, squares) and net all-wave radiation (R n, inverted triangles) over forest (solid symbols) and pasture (open symbols) during 1999-2002. Adapted from von Randow et al. [2004], reprinted with permission of Springer-Verlag,
Bowen ratios are low and RH is high [Fisch et al., 2004; von Randow et al., 2004]. In the dry season, strong subsidence (Figure 2a) brings dty air into the BL, and evaporation is reduced over the pasture in Rondonia, so mixed layer depths are much larger and can reach 2000 m over the pasture. Betts et al. [2002a] and Strong et al. [2005] discuss the surface diurnal cycles of temperature, humidity, lifting condensation level, equivalent potential temperature, surface fluxes, and BL cloud for easterly and westerly regimes (see section 3.3) at the Rondonia pasture site in the 1999 rainy season. They show that the downward solar radiation and the fluxes of sensible and latent heat are lower for the westerly wind regime, which has more stratiform cloud but has a higher water vapor mixing ratio with a weaker diurnal cycle. The easterly wind regime shows an early morning maximum of mixing ratio, followed by a fall to a minimum in the afternoon, as the cumulus clouds mix water vapor up and out
of the subcloud layer more rapidly than is provided by surface evaporation. As the rainy season progresses throughout January and February 1999, there is a steady transition toward cloudier conditions and lower surface fluxes. Daytime surface Bowen ratio for this pasture site is about 0.4 and falls slightly as the rainy season progresses. Typically, in the afternoon, evaporatively driven downdrafts from convective rainbands transfOlm the boundaty layer. The fall of equivalent potential temperature in the boundary layer is about 10K and is similar for both regimes, but the boundaty layer cooling by individual convective events during the westerly regimes is reduced because the subcloud layer is shallower on average, as rain events are weaker but more frequent. This boundary layer modification by rainbands is rather similar to that seen in other moist convection regimes in the tropics (e.g., in Venezuela [Betts, 1976]). Figure 6 compares the seasonal cycle of the diurnal cycle for the near-surface variables for the Rondonia pasture and forest sites during 2001, using data from von Randow et aI, [2004]. Note that the temperature, T, and relative humidity data RH are fi'om Vaisala HMP35A instruments, mounted at v~ry different heights: 8.3 m over the pasture and 60 m over the forest floor, well above the mean canopy height of 35 m [von Randow et al., 2004]. The "wet" season mean is January, February, and March; the "dry" season here is simply August, when mean subsidence is strongest over Rondonia; and the "dry-to-wet" transition is an average of September and October. Although the mean temperature in Figure 6a varies little over the year, the diurnal amplitude of temperature doubles between the wet and dty seasons, as the atmosphere gets drier and less cloudy, and the outgoing net long-wave radiation, which is a primaty driver of the diurnal temperature range [Betts, 2006], doubles (see Figure 5). The forest data (at 60 m) are warmer than the pasture (at 8.3 m) in the rainy season, but the pasture becomes watmer in the daytime in the dty season with a greater increase in diurnal range. The diurnal moisture structure in Figure 6b shows the seasonal fall of water vapor mixing ratio, Q, between wet and dty seasons and the rapid recovety by the transition season. The forest always has a greater mixing ratio than the pasture [von Randow et al., 2004]. For Figure 6, a low bias of 4.3% near saturation at the pasture site was corrected, but it is possible that part of the humidity difference between sites is still instrumental, as these instruments are only calibrated to a few percent in RH. Indeed, absolute calibration of humidity instruments is difficult in these very moist tropical environments [Betts et aI" 2002c]. Away from the rainy season, mean mixing ratio rises steeply from a morning minimum at sunrise, when the atmosphere is saturated at the surface (Figure 6d), as evaporation is trapped in the stable nocturnal BL.
BETTS ET AL. 34 - - r - - - - - - - - - - - - - - , Pasture: Season
19
-;f-Ory:8
32
~j>-
-transition: 9,10
,"'---Wet: 1,2,3
18
Foresl: Season
,
~
Ol
.><:
.3
I-
a
Rondonia 2001
-Ory:8 - - "Transition: 9,10 .............. -Wel:1,2,3
Ir ' -..... ~9I..P" /..... ,:_::::,: "__ - -. -. - _/
16
.,.. ....
~
............. --~....
17
169
_
-... '-
.......
""/ ...........-
.... _
/~
"
.. ...... ~Jot:-::T- ..... / -
15 14 13 12 Pasture: Season
22
-Ory:8
11 Rondonia 2001
o
3
6
9
12
15
18
21
- - -Transition: 9,10
o
24
3
6
UTe Forest: Season
Rondonia 2001
-Ory:8 -
"Transition: 9,10
- - -Wet: 1,2,3
350
g
345
UJ
340 335
9
12
15
18
21
24
UTe
360 - , - - - - - - - - - - - - - - , 355
b
----Wei: 1,2,3
10
240 220 200 180 160 ro c.. 140 .c --::., 120 ~ 100 c.. 80 60
-,--------------:=:----,
Rondonia 2001
30
Forest: Season
-Ory:8 -
-Transition: 9,10
- 40
---Wet: 1,2,3 Pasture: Season
-Ory.8
- 50
- - -Transition: 9,10
nnWel:1,2,3
70 80
Pasture: Season
-Ory:8
- - - Transition: 9,10
----Wei: 1,2,3
o
3
6
9
12
15
18
21
24
UTe
40 " " 20
90 d
o -hrr-,-,-....-.-...-r-.-r-,-,rr-,-,-....-.-.-.-.-+ o 3 6 9 12 15 18 21 24
100
UTe
Figure 6. Diurnal cycle of (a) T, (b) Q, (c) eE, and (d) PLCL for Rondonia forest and pasture sites in the wet and dry seasons and the transition from dry to wet in 1999.
Only a few hours after sunrise, the deepening mixed layer breaks through the noctumal inversion and mixes with the residual layer above, and Q stops rising, as BL clouds transport water vapor out of the mixed layer. In the dry season, this coupling with the convective cloud layer above leads to a strong fall of Q following the morning peak and a weaker fall in equivalent potential temperature eE , shown in Figure 6c. Over the forest, eE is larger than over the pasture, as it depends mainly on Q. In the wet season, mean eE rises during the daytime, until the onset of deep convection brings lower· eE air from higher levels into the subcloud layer in downdrafts [Betts et al., 2002a]. Note that the mean eE over. the forest reaches a slightly higher mean peak in 2001 in the transition season, when deep convection is strongest (see section 3.5). Finally, Figure 6d shows the pressure height from the surface to the lifting condensation level (LCL), PLCL (the
different measurement heights above the surface have been added). As discussed at the beginning of this section, the LCL over Amazonia is a good estimate of daytime cloud base and the depth of the mixed layer. The corresponding near-surface RH is shown on the right-hand scale (with slight approximation). In August 2001, mean afternoon cloud base reaches nearly 240 hPa (2100 m) above the pasture; while in the wet season, afternoon cloud base height is only 60 hPa (545 m) over the forest. Such a low cloud base is more typical of the tropical oceans, so this has led to the description of the Amazon Basin as a "green ocean." 2.4. Nocturnal BoundalY Layer
Nocturnal boundary layer (NBL) development is different between dry and wet seasons and between forest and pasture. The greater surface stress over the forest surface
170
BETTS ET AL.
AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATIONS
generates a deeper NBL through enhanced turbulent mixing [Nobre et at., 1996]. The long-wave radiative cooling of the surface is much larger in the dry season because of the lower humidity tf the overlying atmosphere and reduced nocturnal cloud cover [Betts, 2004, 2006], and this leads to a more stable NBL [R. M. N. dos Santos, 2005]. In addition, the NBL develops fi'om an afternoon BL, which typically has a deep "well':mixed" dry adiabatic structure in the dry season, but develops from a disturbed rain-cooled structure closer to a wet adiabatic temperature pr~file in the rainy season. Consequently, it is easier to define the depth of the NBL in the dry season, when the surface cooling is larger and the residual BL above is closer to a dry adiabat. A comparison between the NBL over the forest (Rebio Jaru) and pasture (Fazenda Nossa Senhora) in Rondonia for the August 1994 RBLE 3 dly season was given by R. M. N. dos Santos [2005]. Using primarily tethered balloon data to define the height and temperature profile of the NBL, she showed that the difference, ~e, between the surface and NBL top (a measure of the strength of the NBL) reached about 11K at 0500 local solar time (LST) (before sunrise) over both forest and pasture. However, at 0500 LST the NBL depth was typically 30% greater over the rougher forest (420 ± 84 m) than over the pasture (320 ± 46 m). This difference is consistent with the theoretical analysis ofNBL depth and strength, based on reanalysis model data by Betts [2006]. Another factor that may contribute to a deeper forest NBL is that the sensible heat flux reverses sign, and the NBL growth starts almost 1 hour earlier over the forest than over the pasture in Rondonia [Oliveira and Fisch, 2000]. R. M. N. dos Santos [2005] also analyzed three data sets collected during the TRMM-LBA rainy season in Rondonia at the same forest and pasture sites and a third transitional forest-pasture site at Rolim de Moura. In the rainy season, the NBL is similar at all three sites. Compared to the dly season, the NBL depth is shallower, typically about 240 m before sunrise, and much weaker in strength, with ~e typically in the range of 3.5--4K [R. M. N. dos Santos, 2005], because the nighttime long-wave cooling is much less in the humid wet season. The near-surface layer usually saturates at night in the rainy season. The NBLs dttring the wet season were mostly weakly stable over forest and pasture. The most intermittent turbulence, associated with the greatest MoninObukhov CJ/L) stability parameter, was seen at the transitional forest-pasture site [R. M. N. dos Santos, 2005]. Low-level nocturnal jets are commonly observed in Rondonia: in the RBLE 3 dry season on about 80% of the days and in the February 1999 wet season experiment on about 60% of the days [R. M. N. dos Santos, 2005]. Jet speeds are similar over forest and pasture and in both dry and wet seasons (mean 7.6 ± 1.9 m S-I, sample size N = 94
soundings); but the height of the jet maximum was typically lower in the dry season (470 ± 165 m, N = 29) than in the wet season (670 ± 145 m, N = 65). Cohen et al. [2006] report similar low-level jets in eastern Amazonia. The shear generated by the low-level jet is a top-down forcing of the NBL [R. M. N. dos Santos, 2005]. Acevedo et al.. [2004] estimated NBL depths during two campaigns, July and October 2001 near Santarem, from surface fluxes and local changes in temperature and humidity as measured by tether or captive balloons. For this site not far from the Tapaj6s River, their estimates ofNBL depth were smaller (less than 150 m), comparable to the thickness of the nocturnal fog layer before dawn. They suggest that the radiative flux divergence at the interface between fog and the clear air above leads to a strong temperature inversion. They used budget methods to compare the accumulation of C02 in the NBL with eddy correlation estimates of the surface C02 flux.
120
ERA-40 - 0.6 Madeira basin 1990-2001 - 0.7 ::0
::r:
0.8
..J
o ..J a. 40
80
- 0.7
----. PR<1mm day'1 -:E-- 15
20
o
0.8
..J
40
0.9 20
a
1.0
o
o
-30
~
~
0.1
0.4
~ --l-:r-Fi .
0.3 0.2
•0.4
LMT
0.9
<0.6
b
- - 0.6
1.0
l- -t__ }-.---------::~--·PR<1mm
0.1
.----
0.0
C
,rE ~ ...J
-50 -60
c:
-70
U
cloud --.;E--0.1 - - '0.2 -0.3 -+-0.4 --- 0.5
day'1
'15
n
ERA-40 Madeira basin 1990-2001
-40
Q.
0.5
0
-
EF
Q
<3 t3
<0.2 LMT
EF t-
::J
RH
0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95
0.3
'0
_n_.
-:E-- 0.2
0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 0.4
2.5. Local Thermodynamic BL Processes and Land-Swface Feedback
~
60
0
..J
a.
0.2 0c:
We will shift attention now to the coupling of surface, BL, and atmospheric processes and discuss some of the links between surface evaporation, cloud base, cloud cover, and surface cloud radiative forcing on the daily timescale [Betts, 2004; Betts and Viterbo, 2005]. For illustration, we again use the ERA-40 reanalysis for the period 1990-2001 averaged over the Rio Madeira Basin. We averaged the hourly model data to give daily means. When we average over the diurnal cycle, we see the balance on the daily timescale between surface processes, downward mixing into the BL, and the effect of falling precipitation evaporating into the BL. In the model system, soil water is a strong control on evaporation as well as BL and cloud processes [Betts and Viterbo, 2005]. However, the model does not represent well the Amazonian soils and deep rooting structures, and compared to Figure 4 for the Jam forest, the model has much less water available for evapotranspiration in the dly season. So instead of soil water storage, we will use the surface evaporative fi'action, EF = AE/(AE + H), to represent the range of availability of surface and subsurface water states. EF can also be thought of as the fraction of the surface available energy going into evapotranspiration. Figure 7 has four plots. Figure 7a links surface evaporative fi'action, EF, to near-surface RH and PLCL (the height of the LCL or cloud base) for different daily precipitation rates. A representative set of standard deviations of the daily mean data is shown. Not surprisingly, as EF increases, mean cloud base descends and RH increases; but RH also increases as precipitation increases. This is a highly coupled system.
100
ERA-40 - 0.6 Madeira basin 1990-2001
__ •
0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95
EF
171
-80
d -90 -r---r--,---,--.----r-.--.------,.---I 0.5 0.6 0.7 0.8 0.9
RH
Figure 7. (a) Near-surface PLCL and RH as a function ofEF and daily precipitation rate (PR). (b) Same as Figure 7a but for a function of lower middle tropospheric RH. (c) Cloud albedo and low cloud cover as a function of EF and PRo (d) Surface net long-wave flux as a function ofRH and cloud albedo.
When the LCL is lower, more precipitation is likely; but the converse is also true: the evaporation of precipitation as it falls through the subcloud layer will lower the LCL. Now the balance of the subcloud layer depends not only on the surface EF but also on the humidity of the air above the BL, which the BL entrains as it grows deeper during the daytime. Figure 7b is similar to Figure 7a, but the days have been stratified by RHLMT, which is the average daily RH for the lower middle tropospheric (LMT) layer between 500 and 700 hPa (typically above the BL). The RH of the sub- . cloud layer shifts toward a moister state as the lower middle troposphere gets moister. Figures 7a and 7b look rather similar because the RH of the lower middle troposphere and the precipitation are themselves coupled to the midtropospheric
vertical motion field on the daily timescale [Betts and Viterbo, 2005]. Figure 7c, upper curves, shows that the model low cloud fraction below 700 hPa increases weakly with EF and is coupled to increasing precipitation. A more quantitative measure of the cloud field is the surface short-wave cloud radiative forcing (SWCF), defined as SWdn - SWdn(clear), the reduction in the incoming clear-sky surface downward short-wave flux (SWdn) by the cloud field. From the SWCF, we can define (following Betts et al. [2006]) a nondimensional measure, the effective surface cloud albedo , UCIoud, as
Uelaud = -SWCF/SWdn(clear).
(1)
172
BETTS ET AL.
AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATIONS
The lower curves in Figure 7c show that Ucloud, related to the short-wave cloud forcing, increases with surface EF and quite strongly with precipitation rate. We used the ERA-40 clear-sky~flux to compute Ucloud in (1). Since ERA-40 has only a seasonal representation of aerosols, the strong variability in atmospheric aerosol absorption due to local fires (extensive in the dty-to-wet season) will be projected in this analysis onto the cloud forcing and Ucloud· Figure 7d shows the link between the mean surface RH and the surface net long-wave flux, LI/' stratified by Ucloud· The lowest curve is the clear-sky LI/: we see that the outgoing net clear-sky flux decreases sharply as the BL gets moister and cloud base lowers. The upward shift of LI/ with
0.7 0.6 0.5 -c 0.4 ::J 0
0.7
ERA-40 Rondonia 1999-2002
0.6
Months --JFM ---r---AMJ - - -JA
0.5 -c 0.4
- -so
::J 0
-i-ND
u
0.3
d
0.2
;"
"
'f-
......................... -::.. 0.1 0.0
-
-
,-
E
~
- -so
................... .......................
0.2
....-3' ..- ::,......
0.1
b 0.0
ERA-40 Rondonia 1999-2002
-10
H
-20
-3D
-30
":l
E
~ -40
AO
u cloud --.;E-- 0.1 - - -0.3 -0.5 -rD.?
,/ / /
-50
-50
-60
-60
-70
-70
-80
/0
Jaru Forest Rondonia 1999-2002
c
..J
1.0
0.9
0.8 EF
0.7
0.6
1.0
0.9
0.8 EF
0.7
c
..J
Months --JFM ---I"'AMJ - - -JA
0.3
/ ,-,-
-20 ~~
Jaru Forest Rondonia 1999-2002
-i-ND
U d
a 0.6
-10
increasing cloud is a measure of the long-wave cloud forcing on the daily timescale. Note that the standard deviations of the daily data are rather small, typically <5 W m-2 • Betts et al. [2006] showed that these long-wave relationships in ERA-40 were in close agreement with flux tower measurements over the boreal forest. Can we evaluate any of these model relationships using data from LBA? Figure 8 shows some similarities and ferences between the data (Figures 8b and 8d) from the Rebio Jam forest site [von RandOl-vet al., 2004] and the nearest grid point in ERA-40 (Figures 8a and 8c). Precipitation at a point is too noisy a field for stratification, and we have few upper air observations during these 4 years of near-surface
........ ........
'l
" ,t',
,- /
d
c
-80 0.6
0.7
0.8 RH
0.9
1.0
0.6
0.7
0.8
0.9
1.0
RH
Figure 8. (a and b) Cloud albedo as a function ofEF and season for ERA-40 and Jaru f?r~st data. (c and d) Surface net long-wave flux as a function ofRH and cloud albedo for ERA-40 and Jaru forest. AbbreVIatIOns are JFM, January, February, and March (the wet season); AMJ, April, May, and June; JA, July and August (the dry season); SO, September and October (dry-to-wet transition); and ND, onset of the wet season.
data. So, instead, we used a seasonal stratification, dividing the year into fiv9monthly groups: January, Febmaty, and March, the w~rseason; April, May, and June; July and August, the dty s~ason; September and October, dty-to-wet transition; and U1b onset of the wet season. Figures 8a and 8b compare the dependence of Ucloud on EF. ERA-40 has a larger seasonal range of EF: greater in the wet season and smaller in the dty season than the Jam forest. However, the lack of energy balance closure in the eddy correlation forest flux measurements, used here as daily means, introduces some uncertainty into the data. von Randow et al. [2004] showed that attributing the entire energy imbalance to an underestimate in the evaporation would increase EF. Figure 8a shows for ERA-40 the increase in cloud with surface evaporation and with the seasonal change in the large-scale forcing between dty and wet seasons. Figure 8b shows a similar seasonal change and a weak increase with EF, but the representative error bars shown are significantly larger for the data than the reanalysis. Figures 8c and 8d show the coupling of surface net longwave, L'l> with near-surface RH and derived Ucloud. ERA-40 has a dry bias with respect to the forest site in all seasons (not shown). The reanalysis data are at the lowest model level (about 10 m), and the forest measurements are at 60 m, and this could account for 2% of this bias. ERA-40 has a wider seasonal range ofRH. Since ERA-40 has much less soil water storage than is available to the Jam forest (Figure 4), this leads to lower values of both EF and RH in the dty season in the reanalysis. Figure 8c for a single grid point is similar to Figure 7d for the much larger Madeira Basin. It shows the fall in outgoing LI/ with increasing humidity and increasing cloud. Figure 8d, the corresponding plot derived from the forest observations (using the ERA-40 clear-sky flux calculation), is similar, but the range of dependence on cloud is reduced, and the standard deviations are again larger. In general terms, the Amazonian forest with deeper rooting and large soil moisture reserves has a smaller seasonal variation in the surface fluxes than ERA-40. Consequently, the Amazonian surface fluxes and BL are less strongly coupled to the atmospheric forcing than in ERA-40. The forest provides a local climate stability, which is lost with deforestation, when effective soil water storage is reduced. The daily mean fields and fluxes give a coherent picture of the local land-surface-atmosphere interaction, and from them we can draw inferences about the feedback of land surface memory on the seasonal evolution of the boundary . layer and convection. Fu and Li [2004] looked at the interannual variability of the onset of the wet season and its links to rainfall in the dty season using 15 years of the ERA-40 data. They found that a longer dty season with less rainfall, and therefore reduced evaporation, led to reduced convec-
173
tive available potential energy (CAPE), greater stability, and a seasonal delay·in the initiation of precipitation and the onset of the rainy season. This is fully consistent with Figure 7a, where we see that a lower evaporative fraction and precipitation gives a drier near-surface RH with a corresponding higher cloud base LCL. They suggested that if land use change in Amazonia increased the severity ofthe dry season, this would also delay the onset of the rainy season. However, local land-surface feedbacks interact with processes of larger scale in determining the circulation. Li and Fu [2006] show the impact of cold front intrusions on the wet season onset. Grimm et al. [2007] suggest that low spring precipitation leads to low spring soil moisture and high late spring surface temperature in central east Brazil. This induces a topographically enhanced low-level anomalous convergence and cyclonic circulation over southeast Brazil, which enhances the moisture flux from northern and central South America into central east Brazil, setting up favorable conditions for excess rainfall during the peak summer monsoon. This relation is especially strong during El Nino-Southern Oscillation years. Antecedent wet conditions in spring lead to opposite anomalies. They also suggest that the mountains of southeast Brazil may playa role in anchoring the patterns of intraseasonal variability. 3. NATURE OF AMAZONIAN CONVECTION I
In the rainy season, convection over Amazonia is sensitive to processes on many time and space scales because the BL is so close to moist neutrality. Unless suppressed by cloud cover, surface evaporation from the forest is generally high, typically around 4 mm d-1 [Shuttleworth, 1988], and this generates convective instability. Machado et al. [2004] used long time series of radiosonde data and the ISCCP satellite data for cloud cover to discuss the climatological aspects of the seasonal and diurnal variability of convection over Amazonia. Equatorial forest sites, Manaus and Bel6m (Figure 1), have smaller seasonal amplitudes of cloud cover and precipitation than sites within the deforested arc of southern Amazonia (Vilhena) and savanna (Brasilia). Equatorial and forest sites show significant seasonal cycles in precipitation and cloud cover, but these are linked to rather small seasonal changes in CAPE. During the rainy season, the differences among the sites are quite small. Greater differences in precipitation, cloud cover, vegetation stress, and thermodynamics appear during the dry season. The strongest thunderstorms occur during the transition dty-to-wet season and at the beginning of the wet season, when atmospheric CAPE is largest for nearly all sites. During the rainy season, the atmosphere is close to the saturated adiabatic lapse rate because of the large area covered by convective cloud
BETTS ET AL. 174
175
AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATIONS
systems. The high cloud fraction, closely associated with the extensive midlevel clouds, which reduce the morning incomconvective clouds, has nearly the same mean diurnal phase ing solar radiation and evaporation [Betts and Jakob, 2002b; for the wet and dry seasons for all regions. A high cloud Rickenbach,2004]. minimum~in the morning (0% and 15% in the dly and wet seasons, respectively) is followed by a rapid increase in the 3.2, Modeling the Diurnal Cycle of Convection early afternoon, reaching a maximum (20% and 45% in dry Over Amazonia and wet seasons, respectively) at the end of the afternoon. Modeling the diurnal cycle of convection over Amazonia The maximum total cloud cover during the wet season is in large-scale models has, however, proved challenging, just during the night. The majority of the rain events during the because the BL is so close to moist neutrality in the rainy wet season occur around 1400 LST in Belem; Manaus has season and responds so rapidly to both large-scale forcing a secondary maximum around 2000 LST, and Brasilia also and to radiative feedbacks on the diurnal cycle of the surface has the majority of rain events at this time [Machado et ai" fluxes. Global forecast and climate models use convection 2004]. The synoptic-scale convective cloud organization, parameterizations, and these typically produce deep conmainly associated with the South Atlantic Convergence vection soon after sumise [Betts and Jakob, 2002a, 2002b], Zone (SACZ), and Amazonian squall lines are important as soon as the convection scheme "sees" the deep unstable features in central Amazonia. The savanna and deforested atmosphere, characteristic of the rainy season (Figure 3). regions have a larger frequency of thunderstol1TIS, associto the convection scheme in the ECMWF Modifications ated with larger CAPE, than the forest sites. Near Belem, sea model [Bechtold et al., 2004], in which thicker layers are breeze and river breeze effects influence local convection. In lifted and tested for instability, produced some improvement Rondonia during the rainy seasons of 1992 to 1995, Ferreira of deep convection a few hours. but only delayed the onset da Costa et ai, [1998] found that the time of the maximum This appears to be an intrinsic problem because the paramrainfall was later (1600 LST) at the forest than at the pasture do not handle the development time scale of the eterizations site (1400 LST). highly turbulent shallow convective boundary layer as it grows through the morning hours until the first precipitating 3.1. Amazonian Convection and the Diurnal Cycle congestus form around 1100 LST. These generate the cold The daytime convective BL typically goes through a rapid pools that organize the subsequent deep convection [Tompdeepening phase, which last several hours until the first kins, 2001], a process that also takes a few hours. Models cumulus congestus form around 1100 LST [Pereira et al., that resolve the explicit development of deep convection 2000; Betts et al., 2002a; Silva Dias et ai" 2002a]. The cold have proved more successful in realistically simulating the pools fOl1TIed by the precipitation-driven downdrafts fi'om diurnal cycle of precipitation [Silva Dias et ai" 2002b]. Rathese cumulus congestus then start the upscale organization mos da Silva and Avissar [2006] using a l-km resolution of precipitating convective lines, which subsequently sta- model for Rondonia show that the diurnal cycle of domainbilize the BL and troposphere in the afternoon. However, averaged accumulated rainfall can be properly simulated the convective response to small perturbations in stability with suitable initial profiles of atmospheric relative humidity is rapid, and large-scale forcing can organize propagating and soil moisture. Khairoutdinov and Randall [2006] using squall lines of larger scale and can also force precipitation at a velY high resolution model (with a lOO-m horizontal grid) night. Rickenbach [2004] examines the origins of a second- and specified surface fluxes have successfully simulated the ary nocturnal maximum in cloudiness and precipitation in shallow to deep transition for a sample day from the TRMMsouthwestern Amazonia using satellite, radar, sounding, and LBA experiment (23 Febmary 1999). They explicitly show profileI' observations from the TRMM-LBA and WETAMC the role of cold pools from the first precipitating congestus in field campaigns in early 1999. They found that following organizing the development of deep convection. However, it locally generated afternoon convection, organized deep con- is fair to say that the successful simulation of all the interactvection often conh'ibutes to a postmidnight maximum in ing scales in global models has not yet been accomplished. the rainy area and high cloudiness. Many of these nocturnal convective events can be traced to large-scale squall lines, which propagate westward thousands of kilometers from their origin along the northeast coast of Brazil [Cohen et al., 1995]. Nocturnal stratiform drizzle and cloudiness typically weaken and delay the onset of the following afternoon's convection because they stabilize the atmosphere and leave
3.3, Synoptic Links: Mesoscale Analyses ofAmazonian Convection
Prior to the LBA field campaigns, Cohen et al. [1989] established the climatology of squall line systems originating on the north coast of Brazil. They found the largest fre-
quency between April and August. ABLE 2B [Harriss et al., 1990], which took place from 13 April to 13 May 1987, further characterize<;i'the environmental conditions associated with Amazo~~n convection and squall lines [Greco et ai" 1990; Garstql1g etal" 1994; Cohen et ai" 1995]. The initiation of sqcfull lines near the Brazilian northeast coast involves an interaction of easterly waves, sea breeze circulations, and the continental heat source [Cohen et al., 1995]. Once initiated in this favorable synoptic environment with a lower tropospheric easterly jet, squall lines can propagate in a quasi-steady fashion toward the southwest, covering 1000 km in 24 hours (a propagation speed of 12.8 m S-I). Similar fOl1TIation and westward propagation also occur in the summer rainy season (July-August) over Venezuela ifthere is a similar easterly flow [Betts et al., 1976; Fernandez, 1980], although system lifetimes are typically only a few hours. During the South American summer, an upper tropospheric anticyclonic circulation known as the Bolivian High is followed in the east by a trough, which extends over the western Atlantic Ocean, known as the nOliheast Brazilian trough [Kouslry and Gan, 1981; Albrecht and Silva Dias, 2005]. At low levels, a continental heat low develops over Gran Chaco in Argentina. This low-level atmospheric circulation pattern has a nOlihwesterly flow along the eastern Andes in the tropics and subtropics and a predominant east-northeasterly flow over the Amazon Basin, accompanied by intense convective activity and precipitation. Kousky [1979] observed that the slow moving cold fronts and subtropical upper tropospheric cyclonic vortices play an important role in characterizing the precipitation over Brazil. These quasi-stationary fronts are referred to as the South Atlantic Convergence Zone [Nobre, 1988], and they are recognized as one of the main features of the wet season [Silva Dias and Marengo, 1999]. Seluchi and Marengo [2000] found that meridional transport of air between the tropics and midlatitudes in South America is the most intense in the entire Southern Hemisphere and is a function of the position of the SACZ. As previous Shldies suggested that the characterization of the low-level jets in South America is due to these baroclinic systems, many authors have studied the connection between the SACZ and convection over the Amazon Basin, using data collected during the LBA WETAMC and TRMM field experiments in Rondonia [Cal1 Jalho et al., 2002; Halverson et ai" 2002; Herdies et al., 2002; Pereira Filho et al., 2002; Rickenbach et al., 2002; Silva Dias et al., 2002a; Tokay et al., 2002]. Wet season (January and FebmalY) precipitation over the Amazon Basin is clearly linked to the synoptic pattern over' South America. Two distinct regimes of lower h'opospheric winds (westerlies and easterlies) were observed in Rondonia [Herdies et al., 2002; Rickenbach et al., 2002] during the WETAMC and TRMM components ofLBA [Silva Dias
et al., 2002a]. The westerly (easterly) winds were associated with sh'ong (weak) convective activity over the SACZ. Variations in the strength of the SACZ regime is in hnn associated with the intraseasonal oscillation [Carvalho et al., 2002; Albrecht and Silva Dias, 2005]. A stronger SACZ regime is more efficient in transporting tropical moisture from the Amazon Basin into the extratropics. Observations from surface-based radar [Rickenbach et al., 2002] suggested that mesoscale convective systems in the strong SACZ regime were significantly larger in areal coverage, with weaker rainfall intensity and weaker vertical development of the convective cells. The diurnal variation of rain intensity and rain areal coverage generally showed afternoon maxima for both regimes but with important differences, suggesting more explosive convective cell growth in the non-SACZ regime and the dominance of nocturnal stratiform rain processes in the SACZ regime. Individual case studies support this picture of propagating squall line systems in the low-level easterly flow regime with stronger updrafts and more precipitation ice in the middle and upper troposphere [Cifelli et al., 2002; Pereira Filho et al., 2002] and greater elech'ification [Halverson et al., 2002; Peterson et ai" 2002]. The westerly regimes are characterized by reduced CAPE and weaker stratiform rain systems with convective elements embedded in a moister envir~nment [Cifelli et al., 2002; Halverson et al.,2002]. . Petersen et al. (2006] reformulated this classification of flow regimes in terms ofpelWI'bations to the cross-equatorial flow. They showeq, using TRMM lightning and precipitation data from four wet seasons, that it is the southeasterly crossequatorial flow that has the greatest rain intensities, thermodynamic instability, and electrification. In contrast, the regime associated with a northerly low-level cross-equatorial flow (which gives low-level westerly shear) has more stratifonn rain in a moister troposphere but has less convective instability and electrification. 3.4. Aerosols, Smoke, and Convective Transports
Aerosols, coming fi'om natural sources and fires, playa key role over Amazonia, both radiatively, impacting the surface fluxes and BL stability, and in the nucleation of moist convection, Aerosol pariicle number size distributions and hygroscopic properties were measured [Rissler at al., 2006] at a pasture site in Rondonia. The measurements were performed from 11 September to 14 November 2002 as part of LBA SMOCC and cover the latter part of the dry season (with heavy biomass burning), a transition period, and the ow;;et of the wet season. These show the diurnal cycle of the aerosol siz~ distributions and hygroscopic properties, coupled to the diurnal evolution of the boundary layer, decreasing
BETTS ET AL.
176 AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATlONS during the morning with vertical mixing and increasing in the afternoon with peak fire activity (in the lat.ter part ?f t.he dry season). ;I'he radiative impact of aero.sols 1S to red1stnbute heat within the atmospheric BL. Durmg ~~e d:r season, aerosol absorption produces a thermal stab1hzatlOn of the atmosphere, which inhibits convection [Freit~s et.al., 2007; 1. A. R. dos Santos, 2005] by both the reductlOn m the surface net ;adiative flux and the warming of the upper aerosol layer. \ d . Aerosol concentrations significantly alter clou .m1crophysics. Andreae et al. [2004] suggest that t~e heavy smoke observed from forest fires in the Amazon B~Sl~ r~duces cloud droplet size and so delays the onset ofprec1p1tatlOn from 1.5 km above cloud base in pristine clouds to more than 5 km in polluted clouds and more than 7 km in pyroclouds above fires. Suppression of low-level rainout and aerosol washout allows the transport of water and smoke to upper levels, where the clouds appear "smoking" as they detram much of the pollution. Elevating the onset of pr~cipitationmeans ~he latent heat of freezing is released at h1gher le:els~ causmg intense thunderstorms, large hail, and greater hkehhood for overshooting cloud tops into the stratosphere. Pollutants and water vapor detrained in the lower stratosphere may have profound radiative impacts on the climate system.
and little if any lightning. During the wet season .easte.rlY wind regimes, aerosol concentration, C~P~, and hg~tnmg yield per unit of rainfall are all larger, so 1t 1S ~ot poss1bl~ to determine whether greater aerosol or CAPE 1S respons1~le for greater electrification. The highly polluted aero~ol-nch October when biomass burning is at a peak [Fuzzl et al., 2007] a; the end of the d1y season in Rondonia, .is char~c~er ized by large values of CAPE and high ele~tncal actlv~ty. Evidence for a substantial role for aerosol m suppressmg warm rain coalescence was seen in this most highly p.olluted period. However, the lack of distin.ction. in the ele~tnc.al parameters (peak flash rate and lightmng y1eld per umt ramfall) between aerosol-rich October and aerosol-poor Nov~mber in the premonsoon regime again casts d?ubt .on a pnmaty role for aerosol in enhancing cloud electnficatlOn.
3.5. Impact ojMicrophysical and Electrical Differences on Convection
Over Amazonia, the traditional distincti~n between "maritime" clouds containing small concentratlOns (about 50 to 100 cm-3) of large droplets and "continental" clouds co~ taining tenfold-larger concentrations of s~naller droplets 1S blurred. Maritime clouds precipitate easl1y by. warm. processes whereas coalescence is often suppressed m contmental clo~ds, which often have to grow vertica.lly to s.upercoo~ed levels to precipitate by cold processe~ mvolvmg the. 1ce This distinction between convective' clouds remamed ph ase. . . h ' fr a microphysical one until the observations of hg tn1.n g ~m space revealed a dramatic contrast between the hghtmng over land and ocean [Orville and Henderson, 1986]. The differences in updraft strength above land and ocean surf ces related to the greater CAPE over land [Rutledge et al., 1~92; Williams et al., 1992], were t~en ident~fied as anoth~r cause for differences in the cloud mlCrophys1cal and e~e~tr1 cal properties. Williams et al. [2002] examined four d1stmct meteorological regimes in Amazonia to try ~o separat~ ~he roles of boundary layer aerosol and CAPE m determmmg continental cloud strucmre and electrification. The "gr~en cean" westerly regime of the Amazonian wet season 1S a ~istinctly maritime-like regime with minimum aerosol C~? centration, minimum CAPE (of the order of 1000 J kg ),
3.6. Coupling ojOzone Transport to Deep Convection
Vertical transports by moist convection are the prima~ process for mixing in the tropical atmosphere.. Atmosph~nc constituents are mixed throughout the convective BL dunng the period of morning growth, and with the onset ~f deep convection, convective updrafts and downdrafts, dnven by condensation and evaporation, overtum the whole trop.osphere. Near-surface measurements of equivalent potential temperature and ozone at night, when background levels of ozone are low, clearly show that convective downdr~fts rapidly transport air with higher ozone and lower eqUlvalent potential temperature down to the surface from around 800 hPa [Betts et al., 2002b]. This downward transpo~t of ozone may play a significant role in the photochem1stry of the atmosphere boundary layer and increase the surfac.e deposition of ozone. The upward transport of 10\~-ozone atr by deep convection in tum affects the photochem1st:r of the upper troposphere and gives insight into the convective mass transports [Kley et al., 1997, 2007] 3.7. Forest Breeze and River Breeze Circulations
Give~ the instability of the Amazonian atmosph~re, diffe~ ences in the surface forcing can affect the convective .organ~ zation. The challenge of understanding and modehng ~h1S over Amazonia has been the subject of several papers. Szlva Dias and Regnier [1996] addressed the mod~lin~ of mes?scale circulations driven by a deforested reglOn m Rondonia using a mesoscale model. Souza et al. [2000] proposed a simple theory for how temperature differe~ces f~rced by surface heterogeneities can force mesoscale c1rcu.latlOns and applied it to the forest-pasture difference~ ~een .m RB~E 3 data collected in the dry season in Rondoma. Szlva Dzas et al. [2002b] examined the impact of topography and defor-
estation on the development of organized precipitating lines in Rondonia using data from dual Doppler radar analysis, radiosondes, and suUace and boundary layer observations. They complemented the observational study with a series of regional high-r,>J0lution model simulations at 2-km resolution over a 300 km x 300 km area initialized by a moming radiosonde profile. They concluded that during a period of very weak large-'scale forcing, topography and deforestation played a significant role, although the discrete propagation of individual cells and their coupling with upper atmosphere circulations were responsible for the development of multiple lines. River breeze circulations also play an important role in organizing Amazonian convection [de Oliveira and Fitzjan'ald, 1993; Silva Dias et al., 2004; Lu et al., 2005]. The Santarem Mesoscale Campaign was conducted in August 2001 close to two major rivers of the Amazon Basin, the Tapaj6s and the Amazon. The observations indicate that during weak trade wind episodes, the Tapaj6s River breeze actually induces a westerly flow at the eastern margin with an associated line of shallow cumulus. With strong tl'ade winds, the river breeze still slows the easterly flow over the river and organizes cloud lines east of the river. -The atmospheric circulation induced by the river has been interpreted with the help of a high-resolution numerical simulation. A single cell forms during late morning over the Rio Tapaj6s, and it evolves in the afternoon with ascending motion in the eastern margin and a descending branch in the western margin, which suppresses cloud formation. During the night, convergence is seen along the center of the Tapaj6s. The atmospheric C02 concentration variations were simulated using the Colorado State University Regional Atmospheric Modeling System with four nested grids that included a I-kin finest grid centered on the Tapaj6s National Forest [Lu et al., 2005]. The results also suggested that the topography, the differences in roughness length between water and land, the "T" shape juxtaposition of Amazon and Tapaj6s rivers, and the resulting horizontal and vertical wind shears all facilitated the generation oflocal mesoscale circulations. Thus the particular geometiy of the rivers with respect to the trade winds has implications for the generalization of the surface measurements of turbulent fluxes of heat, moisture, and C02 in the Tapaj6s region. 4. CONCLUSIONS The field experiments in Amazonia have given us a ricli database of information on the BL in different regions and seasons over both forest and pasmre, which show the interaction of many processes. The surface radiative fluxes are modulated by the solar seasonal cycle and the daily and sea-
177
sonal variation in cloud cover and atmospheric aerosols. The partition of the surface net radiation into sensible and latent heat fluxes depends largely on the availability of water for evapotranspiration. Over the forest, where rooting is deep and the soil moisture reservoir is large, the annual cycle of evapotranspiration is small, even where there is a long dry season (as in Rondonia in SW Amazonia). Large daytime surface evaporation generates moist convective instability and a strong diumal cycle of convection, which is sensitive to dynamic forcing on larger scales. When forcing is weak, the unstable convective BL of shallow cumulus grows rapidly until about 1100 LST, when the BL is deep enough that the first cumulus congestus form and precipitate. Evaporating precipitation drives downdrafts, and the cold outflow boundaries start the self-organization of the convection into lines of precipitating cumulonimbus. When dynamic forcing is strong, precipitation and cloudiness may peak at night. The nighttime BL over the forest typically saturates and forms a layer of fog, some 150 m deep with strong radiative cooling at the top. The seasonal cycle is controlled by processes of much larger scale. In August, large-scale subsidence dominates over Rondonia, d1ying the troposphere and suppressing the convective BL sufficiently that precipitation is greatly reduced. Over defo~ested pastures, where access to deep soil water is greatly rbduced, EF drops substantially in the dry season, and afternoon cloud base may rise to 2000 m with reduced shallow dumulus cloud cover. The diurnal range of temperature doubles as the surface net long-wave cooling also doubles. The dry-to-wet transition season is marked by the largest atmospheric CAPE, high levels of smoke aerosols from widespread fires, the most energetic deep convection with the highest tops, and the greatest lightning flash rates. The prima1y vertical transport is by deep convection, exchanging gases, aerosols, and smoke between the BL and the upper troposphere. In addition to self-organization on outflow boundaries, mesoscale patterns of convection are organized by heterogeneities in topography or surface heat fluxes, associated with different vegetation and soil water availability. The large Amazonian rivers exert their influence on the low-level flow. The shift in lower tropospheric wind patterns over Rondonia from northeast to northwest, associated with an increase in the intensity ofthe SACZ, alters the convective organization structure and propagation as they change both the vertical shear and the BL aerosols feeding into clouds. In the lowlevel easterly flow regime, propagating squall line systems predominate, with stronger updrafts and more precipitation ice in the middle and upper troposphere and greater electrification. The westerly regimes are characterized by reduced CAPE and weaker stratiform rain systems with convective
----------
178
BETTS ET AL.
AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATIONS
elements embedded in a moister environment. Squall lines of still larger scale originate from the nOltheast coast, ~nd their influence propagates for days, organizing convectlOn and particullrly driving precipitation at night. , Stepping back, we have made considera?le pro~ress m describing and understanding the complex mterac~lOns between the continental-scale meteorology and climate of South America and more local processes: such as the co~ piing of biosphere, aerosols, and microphysics; clouds, ram and mesoscale processes over the diurnal cycle; and the vertical coupling between the surface, boundary layer, ~loud radiative forcing, and troposphere. From an observ.atlOnal perspective, continuous monitoring of BL depth usmg sodars or lidars could substantially help further analyses. A few regions have not yet been studied, such as NW Amazonia, which has an intact forest and no dry sea.son, and t~e areas of savannah with tropical forest in Roraima State m . ,benoith Amazonia. Advective processes at the transitions . tween forest and deforested areas need more mvestigat~on, specifically in their energy, water, and car~on budgets; highresolution modeling could be useful for this. Yet substantial modeling problems remain because our computational capability is still insufficient to represent the whole range of scales from the microphysical t~ th~ global and the development of satisfactOlY parametenzatlOns for the smaller scales has been slow. Amazonia is a pmticularly challenging modeling environment because. during the rainy season the atmosphere is so close to saturation and hence so unstable to moist processes on a range of scales. As a result, forecast and climate models differ in their land-surface-BLcloud coupling, and many still have errors in the diurnal.cycleo Consequently, the careful evaluation of models a~amst the rich database of Amazonian observations still prOVides a critical path forward. Aclmowledgments, Alan Betts acknowledges support from NSF under grant ATM0529797 and ii-om Nj~.sA u,J:lder ~E':S ~rant NNG05GQ88A. Maria Assunyao F. da Sliva Dtas, Gllbelto Fls~h, and Otavio Acevedo acknowledge the support of Conselho N~ClO nal de Desenvolvimento Cientifico e Tecnol6gico through thelr research grant (PQ). FAPESP and MCTlMilenio also funded part of the research.
REFERENCES A ceve do , 0 . C ., 0 . L .L. Moraes , R da Silv.a, D, R. Fitzjarrald, R, K, Sakai, R. M. Staebler, and M. 1. Czikowsky (2004), Infe:ring noctumal surface fluxes fi'om vertical profiles of, scalars m an Amazon pasture, Global Change Bioi" 10, 1-9, dOl: 10.11111 j.1529-8817,2003.00755.x. . , . Albrecht, R. I., and M, A. F. Silva Dias (2005), Mlcroph~slCal ev~ dence of the transition between predominant convectlve/stratl-
form rainfall associated with the intraseasonal oscillation in the southwest Amazon, Acta Amazonica, 35(2), 175-184, Andreae, M. 0., D, Rosenfeld, P. Aliaxo, A. A. Costa, G. P: Fra~, K. M. Longo, and M. A. F. Silva-Dias (2004), Smokmg ram clouds over the Amazon, Science, 303, 1337-1342. Bechtold, P., 1.-P, Chaboureau, A. Beljaars, A. K. Betts, M. M~ller, M. Kohler, M. Miller, and J.-L. Redelsperger (2004), The Sll11Ulation of the diumal cycle of convective precipitation over land in a global model, Q. J. R, Meteorol, Soc" 130, 3119-3137, Betts, A. K. (1976), The thermodynamic transformation of the tropical subcloud layer by precipitation and downdrafts, J. Atmos, Sci., 33, 1008-1020. . Betts, A. K. (2003), Diurnal cycle, in Encyclopedia ofAtmosphenc Sciences, edited by J. R, Holton, 1. Pyle, and 1. A. Curry, pp. 640-643 Academic, London. Betts, A. K: (2004), Understanding hydrometeorology using global models, Bull. Am, Meteorol. Soc., 85, 1673-1688, Betts, A. K. (2006), Radiative scaling of the nocturnal boundmy layer and the diurnal temperature range, J. Geophys. Res., 111, D07105, doi: 10.1 029/2005JD006560. Betts, A. K., and C. Jakob (2002a), Evaluation of the diurnal cyc~e of precipitation, surface thermodynamics, and surface fluxes m the ECMWF model using LBA data, J. Geophys, Res., 107(D20), 8045, doi: 10.1 029/200 IJD000427. Betts, A. K., and C. Jakob (2002b), Study ofdiurnal cycle ofconvective precipitation over Amazonia using a single column model, J. Geophys. Res" 107(023),4732, doi:1O.1029/2002JD002264. Betts, A. K., and P. Viterbo (2005), Land-surface, boundary layer, and cloud-field coupling over the southwest~l1l Amazon in ERA-40, J. Geophys, Res., 110, D14108, dOl:l0.l029/ 2004JD005702, Betts, A. K., R. W. Grover, and M. W, Moncrieff(1976), Stlucture and motion of tropical squall lines over Venezuela, Q. J. R, Meteorol. Soc" 102,395-404. Betts, A. K., J, D, Fuentes, M. Garstang, and 1. H. Ball (2002a:' Surface diumal cycle and boundary layer structure over Rondonia during the rainy season, J. Geophys. Res., 107(D20), 8065, doi: 10.1029/200IJD000356. . Betts, A. K., L. V. Gatti, A. M. Cordova, M. A. F. Silva Dms, and J. D. Fuentes (2002b), Transport of ozone to the surface by convective downdrafts at night, J. Geophys. Res" 107(D20), 8046, doi' 10.1 029/2000JDOOO 158. Betts,' A: K., J. H. Ball, and 1. Fuentes (2002c), Calibration and cOlTection ofLBA/TRMM Abracos pasture site merged dat~set, ftp://daac.ornl.gov/data/lba/physical_climatelBetts, Oak Rldge Natl. Lab., Oak Ridge, Tenn. Betts, A. K., J. Ball, A. Barr, T. A. Black, J. H. McCaughey, ~nd~. Viterbo (2006), Assessing land-surface-atmosphere couplmg m the ERA-40 reanalysis with boreal forest data, Agric. For. Mete01'01.,140,355-382, doi:1O.10l6/j.agrfonnet.2006.08.009. . Betts, A. K., R. Desjardins, and D. Worth (2007), Impact of agnculture forest and cloud feedback on the surface energy balance in BOllliAS, Agric, For, Meteorol" 142, 156-169, doi:l0.1016/ j.agrformet.2006,08.020. " Carvalho, L. M, V., C. Jones, and M. A, F. Sliva Dlas (200.1)' Int~a seasonal large-scale circulations and mesoscale convectiVe actlv-
ity in tropical South America during the TRMM-LBA campaign, J. Geophys, Res" 107(D20), 8042, doi:l0,1029/2001JD000745, Cifelli, R" W. A. Petersen, L. D, Carey, S. A. Rutledge, andM. A. F, da Silva Dias (2002), Radar observations of the kinematic, microphysical, and}precipitation characteristics of two MCSs in TRMM LBA, Geophys. Res., 107(020), 8077, doi:1O.1029/ 2000JD000264, Cohen, 1., L. Sa, D, Nogueira, and A. Gandu (2006), High resolution simulation the eastern Amazonia, 87(36), Jt. Assem, Suppl., Abstract IN33A-06, Cohen,1. C.P., M, A. F. Silva Dias, and C. A. Nobre (1989), Aspectos climatologico das linhas de instabilidade na Amazonia, Climanalise Bol, Monit, Anal. Clim, , 4, 34-40. Cohen, 1. C. P., M. A. F. Silva Dias, and C. A. Nobre (1995), Environmental conditions associated with Amazonian squall lines: A case study, Mon. Weather Rev" 123, 3163-3174, Culf, A. D" 1. L. Esteves, A. de 0, Marques Filho, and H. R da Rocha (1996), Radiation, temperature and humidity over forest and pasture in Amazonia, in Amazonian Deforestation and Climate, edited by 1. H. C. Gash et al., pp. 175-191, John Wiley, New York. daRocha, H. R, A. O. Manzi, and 1. Shuttleworth (2009), Evapotl'anspiration, Geophys. Monogr. Ser., doi: 1O.1029/2008GM000744, this volume. de Oliveira, A. P., and D, R. Fitzjarrald (1993), The Amazon River breeze and the local boundmy layer. 1, Observations, Bound LayerMeteorol" 63,141-162. dos Santos, L. A. R, (2005), Analise e caracterizayao da camada limite convectiva em area de pastagem, durante 0 periodo de transiyao entre a estayao seca e chuvosa na Amazonia (experimento RACCI-LBAIRondonia), INPE-14049-TDI/1 064, 118 pp., Inst. Nac. de Pesqui. Espaciais, Sao Jose dos Campos, Brazil. dos Santos, R. M. N. (2005), Study ofthe nocturnal boundary layer in Amazonia (in Portuguese), Ph.D. thesis, Inst. Nac, de Pesqui, Espaciais, Sao Jose dos Campos, Brazil. Fernandez, W. (1980), Environmental conditions and stmcture of some types of convective mesoscales observed over Venezuela, Arch, Meteorol. Geophys, Bioklimatol., Ser. A, 31, 71-89. Ferreira da Costa, R" 1. R. P. Feitosa, G. Fisch, S. S. de Souza, and C. A. Nobre (1998), Variabilidade diaria da precipitayao em regi5es de floresta e pastagem na Amazonia, Acta Amazonica, 28, 395-408. Fisch, G., 1. Tota, L. A. T. Machado, M. A. F. Silva Dias, R. F. da F. Lyra, C. A. Nobre, A. 1. Dolman, and 1. H. C. Gash (2004), The convective boundmy layer over pasture and forest in Amazonia, TheOl·. Appl. Climatol" 78,47-59, doi:1O.1007/s00704004-0043-x, Fisch, G" I. F. Vendrame, and P. C. de M. Hanaoka (2007), Variabilidade espacial da chuva durante 0 experimento LBA/TRMM 1999 na Amazonia, Acta Amazonica, 37(4), 583-590. Fisch, H. R, A. D, Culf, and C. A. Nobre (1996), Modelingcon-. vective boundaly layer growth in Rondonia, in Amazonian Deforestation and Climate, edited by 1. H. C. Gash et al., pp, 425-435, John Wiley, New York. Freitas, S, R., et al. (2007), The Coupled Aerosol and Tracer Transport model to the Brazilian developments on the Regional At-
l
179
mospheric Modeling System (CATT-BRAMS). Part 1: Model description and evaluation, Atmos. Chem. Phys. Discuss" 7, 8525-8569. Fu, R., and W. Li (2004), The influence of the land-surface on the transition from dry to wet season in Amazonia, Theor, Appl. Climatol., 78,97-110, Fuzzi, S" et al. (2007), Overview ofthe inorganic and organic composition of size-segregated aerosol in Rondonia, Brazil, from the biomass-burning period to the onset of the wet season, J. Geophys. Res., 112, D01201, doi:10. 1029/2005JD006741. Galvao, 1. A. da C., and G, Fisch (2000), Balanyo de energia em area de floresta e pastagem na Amazonia (Ji-Parana, RO), Rev. Bras, Meteorol., 15, 25-37, Garstang, M., H. L. Massie, 1. Halverson, S. Greco, and 1. Scala (1994), An;tazon coastal squall lines. Part 1: Structure and kinematics, Mon, Weather Rev., 122, 608-622. Greco, S., R. Swap, M. Garstang, S. Ulanski, M. Shipman, R C, Harriss, R Talbot, M. O. Andreae, and P. Artaxo (1990), Rainfall and surface kinematic conditions over central Amazonia during ABLE 2B, J. Geophys. Res" 95, 17,001-17,014, Grimm, A. M., 1. S. Pal, and F, Giorgi (2007), Connection between spring conditions and peak summer monsoon rainfall in South America: Role of soil moisture, surface temperature, and topography in eastern Brazil, J. Clim., 20, 5929-5945. Halverson, 1. B" T. M. Rickenbach, B. Roy, H. Pierce, and E. Williams (2002), Environmental characteristics of convective systems during TRMM-LBA, Mon, Weather Rev., 130, 14931509. Harriss, R c., et al.(1990), The Amazon Boundmy Layer Experiment: Wet season, 1987, J. Geophys, Res., 95,16,721-16,736. Hasler, N., and R. Avissar (2007), What controls evapotranspiration in the Amazon Basin?, J. Hydrometeorol., 8, 380-395, Herdies, D, L., A. da Silva, M. A. F. Silva Dias, and R Nieto Ferreira (2002), Moisture budget of the bimodal pattem of the smnmer circulation over South America, J. Geophys, Res., 107(D20), 8075, doi: 10.1 029/200 1JD000997. Hodnett, M. G" M, D. Oyama, 1. Tomasella, and A. de O. Marques Filho (1996), Comparisons of long-term soil water storage behaviour under pasture and forest in three areas of Amazonia, in Amazonian Deforestation and Climate, edited by 1. H: C. Gash et al., pp. 57-78, John Wiley, New York. Horel,1. H., A. N. Hahmann, and 1. E. Geisler (1989), An investigation of the annual cycle of convective activity over the tropical America, J. Clim" 2, 1388-1403, Khairoutdinov, M., and D. Randall (2006), High-resolution simulation of shallow-to-deep convection transition over land, J. Atmos. Sci" 63, 3421-3436. Kley, D" H. G. 1. Smit, H. Vomel, H, Grassl, V. Ramanathan, P. 1. Cmtzen, S. Williams, 1. Meywerk, and S. 1. Oltlnans (1997), Tropospheric water vapour and ozone cross sections in a zonal plane over the central equatorial Pacific, Q. J. R. Meteorol. Soc., 123,2009-2040. Kley, D" H. G. 1. Smit, S. Nawrath, Z, Luo, P. Nedelec, and R. H, Johnson (2007), Tropical Atlantic convection as revealed by ozone and relative humidity measurements, J. Geophys, Res" 112, D23109, doi: 1O.102912007JD008599.
.
---- -
180
AMAZONIAN BOUNDARY LAYER AND MESOSCALE CIRCULATIONS
Kousky, V. E. (1979), Frontal influences on n011heast Brazil, Mon. Weather Rev., 107, 1140-1153. Kousky, V. E., andM. A. Gan (1981), Upper tropospheric cyclonic vortices in the tropical South Atlantic, Tellus, 33(6), 538-551. Li, W., and Fu (2006), Influence of cold air intrusions on the wet season onset over Amazonia, J. Clim., 19, 257-275. Lu, L., A. S. Denning, M. A. da Silva-Dias, P. da Silva-Dias, M. Longo, S. R. Freitas, and S. Saatchi (2005), Mesoscale circulations and atmospheric C02 variations in the Tapajos Region, Para, Brazil, J. Geophys. Res., 110, D21102, doi:1O.1029/ 2004JD005757. Machado, L. A. T., H. Laurent, N. Dessay, and 1. Miranda (2004), Seasonal and diurnal variability of convection over the Amazonia: A comparison of different vegetation types and large scale forcing, Theor. Appl. Climatol., 78,61-77. Martin, C. L., D. Fitzjarrald, M. Garstang, A. P. Oliveira, S. Greco, and E. Browell (1988), Structure and growth of the mixing layer over the Amazonian rain forest, J. Geophys. Res., 93, 13611375. Negron Juarez, R 1., M. G. Hodnett, R. Fu, M. L. Goulden, and C. von Randow (2007), Control of dry season evapotranspiration over the Amazonian forest as infelTed from observations at a southem Amazon forest site, J. Clim., 20, 2827-2839. Nobre, C. (1988), Ainda sobre a Zona de Convergencia do Atlantico SuI: A importancia do Oceano Atlantico, Climanalise, 3(4), 30-33. Nobre, C. A., G. Fisch, H. R. da Rocha, R. F. F. Lyra, E. P. da Rocha, A. C. L. da Costa, and V. N. Ubarana (1996), Observations of the atmospheric boundary layer in Rondonia, in Amazonian Deforestation and Climate, edited by 1. H. C. Gash et aI., pp. 413-424, John Wiley, New York. Nobre, C. A., G. O. Obregon, 1. A. Marengo, R. Fu, and G. Poveda (2009), Characteristics of Amazonian climate: Main features, Geophys. Monogr. Ser., doi:1O.1029/2008GMOOOnO, this volume. Oliveira, P. J., and G. Fisch (2000), Efeito da turbulencia na camada limite atmosferica em areas de floresta e pastagem na Amazonia, Acta Amazonica, 15, 39-44. Orville, R. E., and R W. Henderson (1986), Global distribution of midnight lightning: December 1977 to August 1978, Mon. Weather Rev., 114, 2640-2653. Pereira, L. G. P., M. A. F. Silva Dias, A. 1. Pereira Fi1ho, and P. T. Matsuo (2000), Timing of convection initiation during the WETAMC-LBA, paper presented at First LBA Scientific Conference, Ministry of Sci. and Technol., Belem, Brazil. Pereira Filho, A. J., M. A. F. Silva Dias, R. 1. Albrecht, L. G. P. Pereira, A. W. Gandu, O. Massambani, A. Tokay, and S. Rutledge (2002), Multisensor analysis of a squall line in the Amazon region, J. Geophys. Res., 107(D20), 8084, doi: 10.1029/ 2000JD000305. Petersen, W. A., S. W. Nesbitt, R. 1. Blakeslee, R. Cifelli, P. Hein, and S. A. Rutledge (2002), TRMM observations of intraseasonal variability in convective regimes over the Amazon, J. Clim., 15, 1278-1294. Petersen, W. A., R Fu, M. Chen, andR. Blakeslee (2006), Intraseasonal forcing of convection and lightning activity in the south-
i.
ern Amazon as a function of cross-equatorial flow, J. Clim., 19, 3180-3196. Ramos da Silva, R., and R. Avissar (2006), The hydrometeorology of a deforested region of the Amazon Basin, J. Hydrometeorol., 7,1028-1042. Rickenbach, T. M. (2004), Nocturnal cloud systems and the diurnal variation of clouds and rainfall in southwestem Amazonia, Mon. Weather Rev., 132, 1201-1219. Rickenbach, T. M., R. N. Ferreira, J. B. Halverson, D. L. Herdies, and M. A. F. Silva Dias (2002), Modulation of convection in the southwestem Amazon Basin by extratropical stationary fronts, J. Geophys. Res., 107(D20), 8040, doi:10.1029/2000JD000263. Rissler, 1., A. Vestin, E. Swietlicki, G. Fisch, J. Zhou, P. A11axo, and M. O. Andreae (2006), Size distribution and hygroscopic properties of aerosol particles from dry-season biomass buming in Amazonia, Atmos. Chem. Phys., 6, 471-491. Rutledge, S. A., E. R. Williams, and T. D. Keenan (1992), The Down Under Doppler and Electricity Experiment (DUNDEE): Overview and preliminmy results, Bull. Am. Meteorol. Soc., 73, 3-16. Seluchi, M. E., and 1. A. Marengo (2000), Tropical-midlatitude exchange of air masses during summer and winter in South America: Climatic aspects and examples of intense events, Int. J. Climatol., 20, 1167-1190. Shuttleworth, W. 1. (1988), Evaporation from Amazonian rainforest, Proc. R. Soc. London, Ser. B, 233,321-346. Silva Dias, M. A. F., and P. Regnier (1996), Simulation of mesoscale circulations in a deforested area of Rondonia in the dly season, in Amazonian Deforestation and Climate, edited by J. H. C. Gash et aI., pp. 531-547, John Wiley, New York. Silva Dias, M. A. F., et al. (2002a), Cloud and rain processes in a biosphere-atmosphere interaction context in the Amazon Region, J. Geophys. Res., 107(D20), 8071, doi:10.1029/2001JD000335. Silva Dias, M. A. F., et al. (2002b), A case study of convective organization into precipitating lines in the southwest Amazon during the WETAMC and TRMM-LBA, J. Geophys. Res., 107(D20), 8078, doi: 1O.10291200lJD000375. Silva Dias, M. A. F., P. L. Silva Dias, M. Longo, D. R. Fitzjarrald, and A. S. Denning (2004), River breeze circulation in eastern Amazonia: Observations and modelling results, Theor. Appl. Climatol., 78, 111-121. Silva Dias, P. L., and 1. Marengo (1999), Aguas atmosfericas, in Aguas Doces do Brasil-Capital Ecol6gico Usos Multiplos, Explora9GO Racional e ConservacGo da Cunha Rebour;as, edited by A. Rebouyas, B. Braga Jr., and J. G. Tundizi, pp. 65-116, Inst. de Estud. Avanyados da Univ. de Sao Paulo, Sao Paulo, Brazil. Souza, E. P., N. O. Renno, and M. A. F. Silva Dias (2000), Convective circulations induced by surface heterogeneities, J. Atmos. Sci., 57, 2915-2922. Strong, C., 1. D. Fuentes, M. Garstang, and A. K. Betts (2005), Daytime cycle of low-level clouds and the tropical convective boundaly layer in southwestern Amazonia, J. Appl. Meteorol., 44,1607-1619. Tokay, A., A. Kruger, W. F. Krajewski, P. A. Kucera, and A. 1. P. Fi1ho (2002), Measurements of drop size distribution in the
BETTS ET AL. southwestern Amazon basin, J. Geophys. Res., 107(D20), 8052, doi: 1O.10291200IJD000355. Tompkins, A. M. (2091), Organization of tropical convection in low vertical wind shears: The role of cold pools, J. Atmos. Sci., 58, 1650-1671. .)1 von Randow, c.iet al.(2004), Comparative measurements and seasonal vari~tions in energy and carbon exchange over forest and pasture in south west Amazonia, Theor. Appl. Climatol., 78, 5-26. von Randow, C., B. Kruijt, A. A. M. Holtslag, and M. B. L. de Oliveira (2008), Exploring eddy-covariance and large-aperture scintillometer measurements in an Amazonian rain forest, Agric. For. Meteorol., 148, 680-690. Williams, E., et al. (2002), Contrasting convective regimes over the Amazon: Implications for cloud electrification, J. Geophys. Res., 107(D20), 8082, doi: 1O.10291200lJD000380. Williams, E. R, S. G. Geotis, N. Renno, S. A. Rutledge, E. Rasmussen, and T. Rickenbach (1992), A radar and electrical study of tropical "hot towers," J. Atmos. Sci., 49, 1386-1395. Wright, I. R., C. A. Nobre, J. Tomasella, H. R. da Rocha, 1. M. Robe11s, E. Vertamatti, A. Cu1f, R. C. S. Alva1a, M. G. Hodnett,
181
and V. N. Ubarana (1996), Towards a GCM surface parameterization of Amazonia, in Amazonian Deforestation and Climate, edited by 1. H. C. Gash et aI., pp. 473-504, John Wiley, New York.
A. K. Betts, Atmospheric Research, 58 Hendee Lane, Pittsford, VT 05763, USA. ([email protected]) 1. C. P. Cohen, Departamento de Meteorologia, Universidade Federal do Para, Belem, PA 66075-900, Brazil. R. da Silva, UFPA-LBA Santarem, Rua Lauro Sodre, 1330, Alter do Chao, PA 68109-000, Brazil. G. Fisch, Instituto de Aeronautica e Espayo, Sao Jose dos Campos, SP 12228-904, Brazil. D. R. Fitzjal1'ald, Atmospheric Sciences Research Center, University at Albany, State University of New York, Albany, NY 12203, USA. M. A. F. Silva Dias, Depm1ment of Atmospheric Sciences, University of Sao Paulo, Sao Paulo, SP 05508-900, Brazil. C. von Randow, Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos, SP 12227-010, Brazil.
Natural Volatile Organic Compound Emissions From Plants and Their Roles in Oxidant Balance and Particle Formation Jiirgen Kesselmeier, l Alex Guenther, 2 Thorsten Hoffmann,3 Maria Teresa Piedade,4 and Jorg Warnke 3 Numerous biogenic volatile organic compounds (VOC) species are released into the atmosphere from tropical forests. Measuring all those which are relevant for atmospheric chemistly or for the carbon budget is challenging. Large-Scale Biosphere-Atmosphere (LBA) Experiment field campaigns substantially increased the number of field studies of isoprene and monoterpene emissions, as well as of the exchange of several other VOC species. This chapter reports about the progress made within LBA from primmy emission measurements at the plant species level up to discussions of the oxidative capacity of the atmosphere and formation of secondary organic aerosol particles and cloud condensation nuclei from biogenic hydrocarbons. VOC emission from Amazonian ecotypes has strong effects on atmospheric chemistry, which are obviously not fully understood in the case of the tropical atmosphere. Atmospheric flux studies within numerous field experiments resulted in new knowledge about local to regional scale biqgenic VOC exchange and improved modeling. New data obtained from field as Well as from laboratory studies helped to characterize VOC emissions from the Amazonian forest underlying seasonality within dry and wet seasons. Furthermore, first insight was obtained into the potential of floodplain areas affected by long-lasting flooding periods which can cause special emission adaptation.
IMax Planck Institute for Chemistry, Mainz, Germany. 2National Center for Atmospheric Research, Boulder, Colorado, USA. 3Institute of Inorganic and Analytical Chemistry, Johannes Gutenberg University of Mainz, Mainz, Germany. 4Instituto Nacional de Pesquisas da Amazonia, Manaus, Brasil.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM0007l7
1. PROCESSES CONTROLLING TRACE GAS EXCHANGE BETWEEN VEGETATION AND THE ATMOSPHERE
1.1. Ecological Biodiversity in Amazonia and Volatile Organic Compounds Exchange Among the many challenges in studying the exchange of trace gases between forests and the atmosphere, how to extrapolate emission rates and fluxes is of highest importance. The emission of reactive volatile organic compounds (VOC) is highly plant species dependent [Kesselmeier and Staudt, 1999]. FUlthermore, because we need to go to the level of a branch or leaf in a given plant species to identify the release of trace gases, results from measurements made above the forest will be biased by chemical reactions occUlTing 183
184
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS KESSELMEIER ET AL.
during the time of transpOli. A reasonable approach for modeling VOC release for European forests is to take into account the main forest tree species and their emission rate estimates which have been identified at a branch or leaflevel. However, this approach is impossible to apply to a tropical rainforest with its huge biodiversity of species and phytogeographic regions, for which knowledge on the biotic and abiotic components and relations is still scarce [Junk et al., 2000]. Brazilian Amazonia comprises an area of 5 million km2, 80% of which is classified as so-called terra firme [Pires and Prance, 1985; Pires, 1973]. The remaining 20% can be divided into six phytogeographic formations according to their floristic and stlUctural characteristics, namely, the mixed and transitional forests, the savannas, secondary forests, agricultural areas, and floodplains [Braga, 1979; Junk, 1993]. In floodplains, the seasonality imposed by the regular flooding of the big rivers is the most conspicuous feature inducing environmental constraints on the biota [Junk et al., 1989; Ayres, 1986, 1993; Junk, 1997]. Water level fluctuation exhibits amplitudes of7 to 13 m between the high and low water conditions with flooding periods of 1 to 8 months per year [Junk, 1989; Kubitzld, 1989]. Different plant communities become established depending on the length ofinundation. Thus, changes in the big river's water levels and duration of inundation will drive species composition, diversity, and physiognomy of the floodplain forests [Junk, 1989; Worbes, 1997; Ayres, 1993; Wittmann et al., 2006; Piedade et al., 2001]. Plants exhibit different capacities for tolerating periods ofhypoxia or anoxia as a consequence of the inundation, which can last from hours, to days, weeks, or months [Parolin et al., 2004]. One of these adaptations might be energy production by fermentation and detoxification of compounds such as ethanol by transport into the canopy leaves with a subsequent re-metabolization and/or release into the atmosphere (see section 1.4) [Rottenberger et al., 2008]. Such a strategy, by the plants, could increase the significance of floodplain areas for atmospheric chemistry. However, the impact of fermentation processes on VOC release under long-term flooding is not known. These examples highlight the importance of studying VOC emissions, adaptation, and physiology of key species in each one of these six environments, as well as the importance of comparing a key species which grows in all the environments. Different key species as well as different adaptation strategies may have a significant influence on the release of volatile organic compounds that we should know about and understand. 1.2. Volatile O/ganic Compounds in the Atmosphere
Numerous biogenic VOC species are released into the atmosphere, and measuring all those which are relevant to
atmospheric chemistry or the carbon budget is challenging. Measurement techniques used are always limited to special groups of compounds and can only give a view into parts of the story. Proton transfer reaction mass spectromeuy (PTR-MS) [Lindinger et al., 1998], which is now broadly used [see Karl and Guenther, 2004] can simultaneously detect a large number of VOC species. Crutzen et al. [2000] repOli on a large number of organic species measured using this instmment over the rain forest in Suriname during an aircraft campaign and also discuss a weakness of the PTRMS measurements, which is detecting ion masses without clear identification. Neveliheless, the PTR-MS has become a valuable tool to look for emissions of trace gases into the atmosphere. Measurements made with more conservative tools have nevertheless provided interesting data sets for non-methane VOC species in the atmosphere over Amazonia. During the Cooperative Large-Scale Biosphere-Atmosphere (LBA) Airborne Regional Experiment 1998 (LBA-CLAlRE-98) campaign, Kesselmeier et al. [2000] measured the atmospheric mixing ratios of different species ofVOC at a ground station in an open forest site at Balbina, 100 km north of Manaus. The most prominent VOC species present in air during this wet season campaign (March-April) were formaldehyde and isoprene, each up to several ppb. Concenu'ations of methyl vinyl ketone as well as methacrolein, both oxidation products of isoprene, were significantly below 1 ppb, indicating a velY low oxidation capacity in the lower atmospheric boundary layer, which is in agreement with a daily ozone maximum of <20 ppb. Totalmonoterpene concentration was below 1 ppb. These data can be compared with a more comprehensive investigation during the LBA-EUSTACH campaigns (LBAEUSTACH-l, ApriVMay 1999, and LBA-EUSTACH-2, September/October 1999) in the Rebio Jam, an ecological reserve 100 km north of Ji-Parana in the state of Rondonia, in southwest Amazonia, dming the "wet-to-dry season transition" and "dry-to-wet season u'ansition" periods [Kesselmeier et al., 2002b]. Here, samples were obtained at the canopy top close to the potential sources/sinks for these compounds, as well as above the forest. The most prominent VOCs identified in the air during April/May were isoprene, formaldehyde, and formic acid, with mixing ratios of each ranging up to several palis per billion (Ppb), velY similar to the Balbina site. Oxidation products of isoprene such as methyl vinyl ketone as well as methacrolein ranged around 1 ppb. Total monoterpene concentration was below 1 ppb. This changed at the transition phase from dly to wet in September/October. CI-C2 organic acids and Ct-C2 aldehydes exhibited a significant increase up to 17 and 25 ppb, respectively, which is thought to result from vegetation fires, as are
185
the high methanol concentr'ations. Of high interest however
~2002], t~ey. reported 44 out of95 examined plant species to I~llea:~ sl?lllficant amounts of isoprene of more than 20 flg
neal the Clown l;yglOn and was well above 10 ppb at 10-20 m above the f~,est. Interestingly, monoterpene species dec:eased. Suc;lVat~os?~eric measurements give an impresSIOn ~bout the vanabilIty and concentration ofVOCs during the dlff~rent s~as?ns which are due to the seasonality of the veg~t~tIOn emiSSIOns, climatologic factors, and anthropogelllc mfluences such as fires. Such results are in close accordance with Guenther et al [19.99] who predict~d that dry season isoprene emissions: mamly because of higher leaf temperatures might be higher than t~ose in the wet season. Reports fro~ other rain forest regIOns also support this scenario, i.e., isoprene flux data ~'om central Africa [Ser~Yl et al., 2001]. Furthermore, emisSIOn measurements on the branch level demonstr'ated a d _ bl' f' ou mg 0 Isoprene emission rates fi'om Hymenaea courbaril and monoterpene emissions from Apeiba tibourbou also increased nearly twofold [Kuhn et al., 2002a, 2004a].
g h (gIVen as carbon on a dly weight basis). However, it should be noted that the species studied were not randomly selected or selected based solely on their local dominance FOl: example, a large number of Ficus species (all isopren~ el~~tt~rs) we.re. measured in order to investigate emission vall~tIOns wlthm the Ficus genera. Three of the 21 Ficus species pI:obed .during both the wet and dly season periods showed higher Isoprene emission rates in the dly season b factor~ of 2-10 compared to the wet season values reporte~ b~ KIlI1ger ~t al. [20~2].. No~e of the 21 species showed siglllfi.ca~ltly gleater emiSSIOn m the wet season. Monoterpene emls:~o~~ ?n a carbon and leaf dly weight basis exceeded 1.0 mg g h m only 4 of38 species surveyed. However, the authors assumed that the emission factors were approximatel ~n order of magnitude too low due to sparse foliation resulZ I~~ fi'o~ dly season senescence. The lUbber tree Hevea braslhensls was con~rmed as a significant and light-dependent mo~o~erpe~e eIllitter. Wilske et al. [2007] also studied VOC e~ll1SSIOns.ln Yunnan Province, China where he investigated eight tropICal tree species of SE Asia using dynamic Teflon bag bra.nch ~nclosure~ a~d enclosed the branches for longer adap~atIOn tUll~S. EmiSSIOn potentials of four species were cons~derab~y dl~f~rent from those previously reported. Two species el?ltted ISoprene. Six species emitted monoterpenes though With low standard emission factors between
wer~ sUbs~antial inc~'eases in the atmospheric mi~ing ratio~ of ~Iogen~c comP9unds. Isoprene increased up to 30 ppb
1.3. Primaly Emission Quality and Quantity in Amazonia at the Plant-Species Level .
De~pite our. in~reasing knowledge during the last decade, we ~~111 face slglllfic~nt open questions concerning the composluon of VOC mixtures, quantity, or seasonality, which can only be answered by using enclosures to monitor u'ace gas exchange as it relates to primaly plant physiology. Such measurements are typically perfonlled with enclosures at th~ leaf or ~ra~ch level. In the course of the LBA decade, ~nma~y emiSSIOns from tropical rainforest vegetation were mvestlgated under natural field conditions as well as within pl~nt chan:bers under controlled conditions. The results obtamed ~unng these investigations were interesting. OutSide the Amazonian region, Geron et al. [2002] s~re~ned 20 plant species at a lowland tropical wet forest site m Costa Rica and repOlied that 50% of them emitted isoprene. As these species cover some 35-50% ofthe total basal area, these results clearly indicate that a high proportion of the canopy leaf area can be regarded as a source of isoprene a result which was also supported by flux measurement~ above the canopy. Methanol and especially acetone fluxes we~'e ~lso fou~d to be significant, in contrast to monoterpene emiSSIOns WhICh were nondetectable or very low. The most ._ recent and widest overview of biogenic VOCs in Ch' 1 d' Ina, In c u mg some tropical regions, is by Klingel' et al. [2002J, who screened for more than 500 species with short-term m~asurements: Geron et al. [2006] continued the studies of KlInger et al. m the southem subtropical Yunnan Province Peoples' Republic of China. In agreement with Geron et a/
The most elaborate screening of Amazonian tree species is reported?y Harley et al. [2004]. In this study, more than 100 tree. specI.es were investigated for their isoprene emission ca~aclty usmg different enclosure systems. For initial screenmg of VOC emissions with no separation between isoprene and other VOC species, a handheld photoionization detector (!hermo Environmental InstlUments, Inc.) was used. AdditlO~ally, cartridge samples were analyzed for a better resolutl~n ~nd identi~cation of VOC species. In these studies, ~PPlOxI~r:ately 38 Yo of 125 tree species examined at six sites m Br.azlhan Amazonia were found to emit isoprene. The quest.IOn of h~w many trees emit monoterpenes and other v.olatI1es re~a~ned open. But even with this high number of smgle species mvesugated, any upscaling of isoprene can be
186
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
problematic as far as biodiversity is concerned. The authors therefore developed a method to assign emission rates to tree species which were never measured. Such an approach is not without pr&lems; it clearly reveals the gaps of our knowledge and underlines the necessity to continue primaly emission measurements to improve our emission-rate estimates on a larger scale. This interpretation is supported by recent findings [Wilske et al., 2007] demonstrating that there can be significant discrepancies between results obtained by a fast screening and data from a more intensive enclosure type investigation of a few plant species. Enclosure measurements in Amazonia with dynamic branch enclosures and adaptation time of more than one day per plant species coupled to the simultaneous determination of climatological variables and physiological parameters such as assimilation, transpiration, and stomatal conductance contributed significantly to a better understanding of the emission processes and controls. Kuhn et al. [2002a] investigated tree species in a secondary forest during the wet season of 1999 at a remote field site in Rondonia; H. courbaril was found to be a strong isoprene emitter, and A. tibourbou was identified to exclusively emit substantial amounts of monoterpenes in a light-dependent manner, but no isoprene. The diel emission pattern of both tree species was similar in regard to controlling environmental parameters, such as light fluctuation, and temperature. This is a clear demonstration of the now well-accepted, closely related metabolisms of isoprene and monoterpenes. A strong light dependence of biogenic monoterpene emissions may have a strong impact on estimated global flux rates for tropical regions. Whereas extratropical regions are characterized by strong annual growth cycles, tropical regions are often regarded as exhibiting only little climatic and growth variability. However, though we accept tropical regions as a dominant trace gas source all year round, it is now well accepted that there are strong seasonal cycles which can be clearly detected by dendrochronology [Worbes, 1999; Worbe~ and Junk, 1999; Schongart et al., 2002, 2004]. Recently, Myneni et al. [2007] used remote sensing to observe a strong leaf area fluctuation in Amazonian rainforest. Similarly, changes in trace gas exchange can be expected. Kuhn et al. [2004a] observed leaf age-dependent seasonal differences comparing wet-todry and dry-to-wet transition phases. Strong seasonal differences in emission capacity were observed in these cases. The isoprene standard emission factor in the caSe of H. courbaril was about twofold at the end of the dly season with freshly developed new leaves, compared to the end of the wet season. In contrast, standardized monoterpene emission rates of A. tibourbou exhibited a decrease at the end of the season. Such branch emission data are in close accordance with atmospheric mixing ratios [Kesselmeier et al., 2002b].
Hence, using a single standard emission factor to represent an entire seasonal cycle is no longer adequate. Furthermore, in addition to the variability of isoprene emission, considerable amounts of monoterpene emissions were detected in the period between bud break and leaf maturity of this tree at the end of the dry season [Kuhn et al., 2004b]. Apparently, in addition to light and temperature, we need to invest more research into the roles of other potential factors, such as leaf developmental stage, water and nutrient status, and stresses like the oxidative capacity of the ambient air. Obviously, such factors contribute significantly to the current emission capacity during the relevant season. Within this context, it is interesting to see that a strong linear correlation between the isoprene emission capacity and the gross photosynthetic capacity was observed covering all developmental stages and seasons [Kuhn et al., 2004b]. Such findings may represent a valuable basis on which to model the seasonal variation of isoprenoid emission capacity. Though we leamed much in the course of the LBA studies, there is still a need for a more basic understanding of the regulation of primary emissions from important plant species. This becomes evident especially in view of a growing number of publications reporting about surprising biogenic emission qualities, amounts, and regulation for vegetation species thought to have been sufficiently investigated as, for example, in the very recent case of European Beech [Moukhtar et al., 2005; Dindmfet al., 2006]. 1.4. Plant Species Gro,ving in Floodplain Areas Adapt to the Fluctuating Water Table
More than 300,000 km2 in size, the Central Amazon floodplain represents one of the largest inundation areas in the world; there are up to 210 days of continuous flooding per year with an average flood amplitude of several meters [Junk, 1997]. Inundation of trees in these areas causes drastic changes in soil chemistry and oxygen availability to plant roots [Parolin et al., 2004]. This stress affects primary plant physiology as well as secondary metabolism, for example; photosynthesis, and potentially also the isoprenoid metabolism, respectively [Rotten berger, 2003; Rottenberger et al., 2008]. Furthermore, studies on European tree species demonstrated that leaves emit ethanol and acetaldehyde as a physiological response to anaerobic conditions in the roots [MacDonald and Kimmerer, 1993; Kreuzwieser et al., 1999; Holzinger et al., 2000], which in tum initializes alcoholic felmentation in the roots with ethanol production. The major portion of the root ethanol is transported with the transpiration stream to the leaves, where it can be re-metabolized by stepwise oxidation to acetaldehyde and acetate, mediated by the leaf enzymes ADH and aldehyde dehydrogenase
KESSELMEIER ET AL.
(ALDH). Through this mechanism, the plant recaptures carbon and energy whicll;was invested into ethanol and avoids an ~cct~mulation ~f,jphytotoxic ethanol and acetaldehyde. A ftactlOn of the~e compounds can be emitted into the atmosphere, a pr~~ss which can be considered to be a "leak" between prodtiction and consumption of these compounds. In the atmosphere, all three C2-compounds are of high impmiance for tropospheric chemistry and influence the oxidati."e capacity of the atmosphere, the production of organic mtrates, as well as atmospheric acidity, especially in remote areas [Carlier et al., 1986; Keene et al., 1983; Kesselmeier 2001; Singh et al., 1995; Talbot et al. 1990' Thom'P sm/ 1992]. '" In contrast to our knowledge of terra fitme tree species [Rottenberger et al., 2004, 2005], floodplain area trees are poorly investigated; however, there have been intensive investigations on the anaerobic production and metabolism of ethanol (fermentation) in roots and its transportation to the leaves (see Figure 1). Only a limited number of studies of the flooding-induced release of compounds fi'om leaves such as ethanol and acetaldehyde, and only one pilot study for plants fi'om Amazonia have been repmied [Rottenberger, 20~3; Rottenberger et al., 2008]. Considering the vast area of lllundated forest as well as the duration of flooding, the
Emission into the atmosphere
187
Amaz.onian floodplain forests represent potentially one ofthe most Important vegetative sources of atmospheric ethanol, acetalde~yde, and acetic acid, which may have an impact on b~th regIOnal as well as global atmospheric chemistry and chmate. I?espite our knowledge of plant physiological processes
whlc~ can be initialized by flooding, important questions
re~alll to be investigated. Our experience is based on expenments under controlled conditions; to our knowledge, there are no field experiments. Hence, we do not know how adult trees behave. Furthermore, we do not know how trees deal with the long flood pulse. If some species switch over to fennentation, will the emissions continue over the whole ~ooding period? Another important question is whether the ~sOprenOl? ~etabolism is affected. Can we expect changes III the emiSSIOn rates of isoprene or monoterpene emitters? If photosynthesis decreases and VOC emission is not affected, or even increases, the carbon budget of the plant would be affected. Stress effects, such as high temperatures, are known to cause such a shift. How do floodplain area adapted plants react? Do only trees with an excellent gas. supply through aerenchyma or adventivous roots keep their l~av~s, whereas others drop all leaf material to survive? Investi~atlOns made over several seasons of the seasonal fluctuatIOns ofVO~s, including acetaldehyde as a potential marker for floodplain emissions, will help to determine the relevance for atmo~pheric chemistiy and climate.
l.~. Exchange ofShort-Chain Aldehydes and Acids in Terra FU'llle and Floodplain Areas: Emission and Deposition
Within the LBA project, special attention has been given to the release of isoprenoids (isoprene and monoterpenes). Howev~r, n:an~ other VOCs can be released by the biosp~e~e III Significant quantities [Fehsenfeld et al., 1992; Komg et al., 1995; Guenther et al., 1995; Kesselmeier and Transport by the Staudt, 1999]. Furthermore, it is not only trace gas emission transpiration that can be fo~nd, but also deposition which may conti'ibstream ute to a r~cychng of carbon. Among these VOC species, the sh?li-challl oxygenated organic acids, fmmie acid and acetic aCid, as well as their homologous aldehydes are of interest The~e .tw~ acids contribute significantly to the acidity ofth~ preCIpitatIOn as discovered by Andreae et al. [1988] and Talbot e~ al. [1990] during the ABLE-2 campaign in central F.i~ure 1. Fermentation processes in the roots under aerobic con- ~mazon~a. The sholi-chain aldehydes play an impmiant role dltlons. allows the trees to overcome oxygen limitation which . III the o~ldative chemistly of the troposphere [Carlier et al., othelWlse would prevent energy production. The ethanol can be 1986; ~lI1gh et al., 1~95, 2001; Chebbi and Carlier, 1996]. transported into the canopy leaves where it is metabolized to acetalAccordlllg to an eat'her publication [Glasius et al., 2000b] ~ehyde and acetic acid (acetate) and recaptured by plant metabo80~.1 00% of fmm.ic aci.d, in particular, stems from biogeni~ hsm. Due to their volatility, a pali of these compounds can be lost sou~~es, whether It del1ves fi'om the photochemical decomto the atmosphere. posItIon of other biogenic trace gases (indirect emission), or
KESSELMEIERET AL.
189
188 NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
Studies on European tree species demonstrated that leaves is produced and released directly by plants (direct emission) emit ethanol and acetaldehyde as a physiological response to has not been determined. A dominating uptake as well as anaerobic conditions in the roots [MacDonald et al., 1989; emission was found in the LBA studies. Kreuzwieser et al., 1999; Holzinger et al., 2000], which iniKuhn et S/. [2002b] reported about the exchange of fOlmic tializes alcoholic fermentation with ethanol production. The acid and acetic acid between vegetation and the atmosphere major portion ofthe root ethanol is transpOlted with the tranin the wet-to-dty season transition and the dry-to-wet season spiration stream to the leaves, where it can be re-metabolized transition periods in 1999 in Rondonia. Enclosure measureby stepwise oxidation to acetaldehyde and acetate, mediated ments on individual branches mainly exhibited uptake of by the leaf enzymes ADH and aldehyde dehydrogenase formic acid and acetic acid for all plant species in both sea(ALDH). This mechanism allows the plant to recapture carsons. The uptake of organic acids was found to be closely bon and energy invested into ethanol and to avoid an accurelated to the ambient atmospheric mixing ratios. A bidimulation of phytotoxic ethanol and acetaldehyde. A fraction rectional exchange could be detected, but with a very low of these compounds can be emitted into the atmosphere, a compensation point, thus nearly excluding any emission. process which can be considered to be a "leak" between Generally, the forest could be regarded rather as a sink than a production and consumption of these compounds. In the atsource for organic acids, a conclusion which was supported mosphere, all three C2-compounds are of high impOltance by concentration gradients above the canopy. Strong diel for tropospheric chemistry influencing the oxidative capacvariations were found in both seasons and very high mixing ity of the atmosphere, the production of organic nitrates, ratios, caused by vegetation fires, occurred in the dry season. as well as the atmospheric acidity, especially in remote arA substantial contribution to the atmospheric burden by dieas [Carlier et al., 1986; Keene et al., 1983; Kesselmeier, rect emission was excluded. However, the chemical produc2001; Singh et al., 1995; Talbot et al., 1990; Thompson, tion by oxidation of primary biogenic reactive compounds such as isoprene and monoterpenes may be regarded as a 1992]. In contrast to investigations on the production and metabdominant source for these acids during the wet season. Dry olism of ethanol in roots and its transportation to the leaves, season data were dominated by vegetation fire emissions. only a limited number of studies of the flooding-induced reVety similar behavior was reported for form- and acetallease of compounds, such as ethanol and acetaldehyde, have dehyde [Rotten berger et al., 2004, 2005]. Branch enclosure been reported, and only one pilot study for plants from Amameasurements of several tree species showed more clearly zonia is reported [Ro ttenberger, 2003; Rottenbel'ger et al., a bidirectional exchange with a compensation point of 0.6 2008]. In contrast to the slight deposition of acetaldehyde ppb. Thus emission as well as deposition was observed, and acetic acid under nonflooded conditions, flooding of the though exchange was clearly dominated by deposition. roots caused leaf emissions of ethanol and acetaldehyde by Deposition velocities and the compensation point increased all tree species investigated. Those plant species exhibiting during the dty season. Uptake occUlTed through the stomata the highest ethanol and acetaldehyde emission also emitted under clean-air conditions in the wet-to-dry transition peacetic acid. Of special interest were emission bursts of all riod, whereas deposition to the leaf surface was estimated three compounds in the morning, which can be understood to be substantial for the dty-to-wet period when biomass as an accumulation during the night and a sudden release as burning activities and atmospheric mixing ratios were high. soon as the stomata opened in the morning. Emission rates Diel courses and concentration gradients ,above the canopy varied substantially among tree species, with maxima dif-2 support the conclusion of a photochemical oxidation ofbiofering ,by up to two orders of magnitude (3-200 nmol mgenically or pyrogenically emitted precursor compounds min-I for ethanol and 5 and 500 nmol m-2 min-I for acetalacting as sources for these aldehydes which can be taken up 2 dehyde). Acetic acid emissions reached 12 nmol m- min-I. by vegetation again. Differences between tree species can be understood as a However, the exchange of trace gases as mentioned above consequence of different kinds of adaptation strategies to is not always limited to deposition with some emission deovercome oxygen deficiency in the roots, i.e., development pending on the CUlTent ambient mixing ratios and the comof morphological root anatomy and morphology allowing pensation point. Special environmental conditions such as for enhanced root aeration. The pronounced differences in flooding and inundation of plants, which is a typical feature the relative emissions of ethanol to acetaldehyde and acetic of large areas of the Amazonian floodplains might change acid indicate that not only the ethanol production in the roots the picture [Junk, 1997; Parolin et al., 2004]. This caused but also the metabolic conversion in the leaf is an imporsome concern about trace gas emissions from partly flooded tant factor determining the release of these compounds to tree species which thus suffered from anoxic conditions of the atmosphere. In most cases, emissions initially increased their root system.
during the first 3 days of the flooding period, and then declined after 3 to 7 days. Some plant species which were not dramatically affectrd by the flooding obviously adapted to the anoxic situati?n. Others with strong physiological symptoms of injmy d,generated. The effects eft root anoxia repOlted above were only investigated under controlled conditions using young tree species. But the results indicate that there may be an important trace gas source to be better understood. The Amazonian floodplain forests thus potentially represent one of the important vegetative sources of atmospheric ethanol, acetaldehyde, and acetic acid, which may have an impact on both regional as well as global atmospheric chemistty and climate. Nothing is known about the behavior of adult tt'ees under field conditions and about how trace gas emissions adapt to longterm flooding, which is characteristic of the Amazonian floodplains. We therefore recommend studies ofthe effect of flooding on the exchange of the metabolically related compounds, ethanol and acetaldehyde, between the forests and the atmosphere in the Amazonian floodplain areas. 1.6. Primal)! Emissions Can Be Affected by Plant-Plant or Plant-Insect Relations Plants attacked by herbivores or pathogens may react by producing volatile compounds used for direct or indirect defense. Directly affecting toxic volatiles may cause a fast defense, whereas signaling compounds may indirectly contribute by activating a symbiotic chain by, for example, calling herbivore predators. Thus, plants may be alatmed by volatile organic compounds released by herbivore-attacked neighbors and activate defenses to avoid being attacked themselves. There are numerous repOlts in the literature [Baldwin et al., 2006; Dicke et al., 2003; Schulze et al., 2006], but investigations in tropical forests are rare.
1. 7. PrimalY Emissions Affect the Carbon Budget The carbon balance ofthe world's terrestrial ecosystems is uncertain and the carbon balance ofAmazonia especially is a matter of ongoing debate [see Houghton et al., this volume; see also Chou et al., 2002; Araujo et al., 2002; Carswell et al., 2002; Houghton, 2003; Baker et al., 2004]. However, it has become well accepted that the tropics appear to be a nearly neutral, or small net source of carbon. As reported very recently by Lloyd et al. [2007], estimates based on regional surface fluxes using ABL budgeting techniques fot' evaluating airborne data clearly suggested a close to neutral Amazon carbon balance. It is then of special interest to carefully check the contribution of volatile organic carbon to sinks and sources, especially as reactive carbon fluxes are
v.ery. sensitive to land cover and climate change and may vary stgmficantly dite to future perturbations [Guenther, 2002]. The role of vegetation as a source of volatile carbon was discussed by Frits Went and coworkers for the first time 50 years ago [Went, 1955, 1960a]. The amounts of these compounds were even regarded to be one of the sources of petroleum [Went, 1960b]. Today, our estimates of biogenic carbon released from terrestrial vegetation is above 1000 Tg a-I [Guenther et al., 1995]. Compared to the gross primaty ~roductivity of vegetation in the range of 120 Pg a-I, this tS only a small contt'ibution equivalent to one percent and consequently VOC emissions from vegetation are rarely included in estimates of global carbon fluxes. However, this picture changes as soon as we look a little closer and relate the VOC release to net primary productivity (NPP), net ecosystem productivity (NEP), or net biome productivity (NBP) [Geron et al., 2002; Guenther, 2002; Kesselmeier et al., 2002a]. Based on such studies, it can be assumed that with regard to the carbon budget of the terrestrial biosphere, a release ofVOC carbon is a significant loss ofphotosynthetically fixed carbon. Kesselmeier et al. [2002a] estimated the amount ofVOC carbon emitted in relation to the CO 2 taken up at a leaf or branch level to range up to a few percent of the photosynthetically assimilated CO 2. Based on numerous enclosure and micrometeorological flux measurements of simultaneous VOC 'emission and CO 2 exchange in the Mediterranean area, and the tropical rainforest in Amazonia, they demonstrated tha~ VOC flux estimates are small in relation to NPP and gross primary productivity (GPP). However, the amount of carbon lost as VOC emissions can be highly significant relative to NEP. Carbon losses on a GPP basis were estimated to be equivalent to 0.45% for temperate and 0.54% for tropical forests. As pointed out by Kesselmeier et al. [2002a], the VOC loss of 0.45% of a GPP of 120 Pg a-I accounted for 0.54 Pg C for the total isoprenoids, a number vety close to the modeled estimate of the global isoprene emission of about 0.5 Pg C [Guenther et al., 1995]. Compared to the range of current estimates of NEP, a loss of 3.5-39% of the NEP carbon was calculated. Ranges for global and tropical scenarios were very similar with 3.5-36% and 3.4-27%, respectively. Compared to NBP, the VOC carbon loss was even found to be equivalent, a clear demonstration of the necessity to take VOC fluxes into account for discussions on carbon exchange. Furthermore, the authors noted that such a contribution may increase if data for other VOC species are used in addition to the better-described isoprenoid emissions. The main uncertainty with these estimations was the fate of such reactive compounds. How much is converted to CO2 (and CO) bypassing terrestrial respiration. How much is returned to the terrestrial biosphere, with or without chemical tt'ansfOlmation into other organic compounds? Is this simply
190
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS KESSELMEIER ET AL.
an internal recycling of organic carbon within the terrestrial carbon pool? And how much is finally escaping the forest ecosystems by a transfer from the terrestrial to the marine biosphere, 'for example? As we know about deposition of some potential oxidation products (see section 1.5.), such as organic acids and aldehydes, we can assume that substantial amounts of these carbon emissions are recycled within the biosphere. Nevertheless, a SUbstantial part can be assumed to be lost into longer-lived oxidation products that are lost from the terrestrial biosphere by transport. Such a conclusion was recently confirmed by Naik et al. [2004] who integrated surface emission algorithms using a dynamic global ecosystem model, the integrated biospheric simulator (IBIS), to simulate biogenic fluxes of isoprenoids as a component of the climate-vegetation dynamics. In this simulation, increasing CO 2 levels caused an increase of VOC emissions as a result of increases in foliar biomass. The authors came to the conclusion that the increases in biogenic emissions could have significant impacts not only on regional and global atmospheric chemistry, but also on the global carbon budget. Suntharalingam et al. [2005] have recently demonstrated that the importance of accounting for biogenic VOC and other reduced carbon emissions in CO 2 inversion analyses is far greater than that expected based solely on its small contribution to the total carbon flux. This is because the usual assumption that the CO 2 source from atmospheric oxidation of these compounds is released at the surface produces a bias in the inversion modeling technique used for estimating regional carbon fluxes. They found that including reduced carbon emissions as a CO 2 source distributed in the atmosphere, rather than emitted at the surface, resulted in considerable differences in estimates of regional C02 fluxes. This included a reduction in the northern hemisphere carbon sink and the net carbon emission from tropical landscapes. 2. LOCAL TO REGIONAL SCALE BIOGENIC VOC FLUX MEASUREMENTS AND MODELING The first measurements of biogenic VOC (BVOC) in the Amazonian atmospheric boundary layer were made during a 1979/1980 aircraft research program in Brazil [Greenberg and Zimmerman, 1984; Crutzen et al., 1985]. This research effort was focused on biomass burning but included measurements of background conditions that characterized BVOC emitted from the tropical forest. Greenberg and Zimmerman [1984] report an average isoprene mixing ratio of 2.27 ppb and monoterpene mixing ratio of >5 ppb for heights between treetop and 2 km. Crutzen et al. [1985] concluded that there are large biogenic organic emissions from the tropical forest and that the emissions were responsible for both an enhancement of CO and depletion of ozone in
the boundary layer. They assumed an OH concentration of at least 5 x 105 molecules cm-3 and estimated that an isoprene flux of at least 2 mg m-2 h- 1 was required to maintain the observed isoprene mixing ratio. During aircraft flights over Guyana in 1984 and Brazil in 1985, widespread boundary layer isoprene mixing ratios of ~2 ppb were observed confirming that isoprene was emitted from Amazonian tropical forests in large quantities [Gregory et al., 1986; Rasmussen and Khalil, 1988]. Landscape average BVOC fluxes have been estimated from above canopy observations in Peru and in the Brazilian states of Para, Amazonas, and Rondonia. Studies conducted at six tower sites, ten tethered balloon launching locations, and three aircraft studies are listed in Table 1. The seven different flux estimation approaches used for estimating BVOC fluxes include indirect approaches, based on budgets and concentration gradients, and eddy flux techniques including direct eddy covariance measurements. An additional approach, combining enclosure BVOC emission measurements with tree biomass and species composition data, has been used to characterize fluxes at four ofthe sites in Table 1 and at additional sites in Ecuador [Harley et al., 2004]. The 1985 GTE ABLE2A study at the Reserva Ducke near Manaus included measurements of mixed layer height and vertical profiles of trace gas concentrations within and above the boundary layer [Zimmerman et al., 1988, Jacob and Wofty, 1988]. These observations provided an improved data set for characterizing the surface fluxes required to balance losses from boundaly layer oxidant concentrations. Zimmerman etal. [1988] used an estimate for daytime OHof8.3 x 105 molecules cm-3 [Jacob and Wofty, 1988] and observed isoprene (median = 2.03 ppb) and total monoterpene (median = 0.23 ppb) mixing ratios to calculate daily total emissions of 25 mg m-2 of isoprene and 5.6 mg m-2 of monotel'penes. Estimated isoprene emission peaked at 4 mg m-2 h- 1 around noon and averaged about 3.1 mg m-2 h- 1 between the hours of 0800 to 1600. Jacob and WqfSy [1988] using the same observations but a different analysis approach reported a daily total emission that is ~ 10% lower and a maximum emission that is ~50% higher than the Zimmerman et al. estimates. Davis et al. [1994] used these same tethered balloon data to estimate fluxes from the vertical concentration gradient and arrived at isoprene fluxes that were about 30% higher than the Zimmerman et al. estimates. Veliical profiles of isoprene and monoterpenes have been measured at ten additional Amazonian locations in Peru and the Brazilian states of Rondonia, Para, and Amazonas using tethered balloon [Helmig et al., 1998; Greenberg et al., 2004] and aircraft samplers [Kuhn et al., 2007; Karl et al., 2007]. While Zimmerman et al. [1988] collected 25 vertical profiles (18 daytime and 7 at night) consisting of 106 sam-
191
Table 1. Amazonian Fi~ld Measurements That Have Been Used to Estimate Above Canopy Isoprene and Monoterpene Fluxes Approach Budget
Season and Year
Year
aircraft
dly
Budget; mixed layer gradient
tethered balloon
early dry
1979, 1980 1985
Budget; mixed layer gradient Budget
tethered balloon tethered balloon
early dly
1996
wet
1998
tethered balloon
wet
1999
tethered balloon tethered balloon tower
wet
1999
wet
2000
early dry
2000
tower
wet, early dry
2000
REA
tower
early wet, early dry, wet
EC
tower tethered balloon aircraft
wet, early dry, late dry early dry
2000, 2001, 2002 2001 2001
early dry
2001
tower
early dry
2001
tower
late dry
2004
aircraft
late dry
2004
Budget
Budget Budget DEA;EC REA
Budget Mixed layer gradient REA; Surface layer gradient EC Mixed layer gradient; variance
Platform
pIes at different heights, the data sets for most of the other sites are considerably smaller and consist of less than ten pr~files. The median daytime mixed layer isoprene mixing ratio.s from the 11 sites ranges from 0.6 to 6.7 ppb. Median daytime total monoterpene mixing ratios varied from 80 to 69~ ppt a~d were positively correlated with isoprene with ~ typical ratio of ~ 0.1. Note that this is nearly 20% when expressed on a mass basis due to the higher molecular weight of the monoterpenes. The dominant monoterpene at all sites was a-pinene, which typically comprised about half of the
Location between Manaus and Humaita, Brazil Reserva Ducke: Forest 10 km nOlth of Manaus, Brazil forest 500 km west of Iquitos, Pem Balbina: four forest locations about 150 km NE of Manaus Brazil Fazenda Nossa Senhora: pasture 280 km SE Porto Velho, Brazil Jam Reserve: forest 270 km SE POltO Velho, Brazil Tapajos: forest 50 km south of Santarem, Brazil Tapajos: forest 50 km south of Santarem, Brazil Caxiuana: forest 330 km west of Belem, Brazil Cuieiras Reserve: forest 60 km north of Manaus, Brazil Tapajos: fordst 50 km south of Santarem, Brazil Balbina: for~st 150 km NE of Manaus Brazil Cuieiras Reserve: forest 60 km north of Manaus, Brazil Cuieiras Reserve: forest 60 km north of Manaus, Brazil Cuieiras Reserve: forest 60 km nOlth of Manaus, Brazil forest, wetlands and croplands regions between 10 and ISO km, north and east of Manaus, Brazil
References Crutzen et al. [1985] Jacob and Wofty [1988] Zimmerman et al. [1988] Davis et al. [1994] Helmig et al. [1998] Greenberg et al. [2004]
Greenberg et al. [2004]
Greenberg et al. [2004] Greenbelg et al. [2004] Rinne et al. [2002] Pegoraro et at. (unpublished data) Stefani et at. (unpublished data) Potosnak et at. (unpublished data) Greenberg et at. (unpublished data) Kuhn et al. [2007] Kuhn et al. [2007] Karl et al. [2007] Karl et al. [2007]
total mo~oterpenes. Helmig et al. [1998] observed relatively low median Isoprene mixing ratios (1.39 ppb), but report on~ ~fthe highe~t isoprene flux estimates (8.1 mg m-2 h- 1). ThiS IS due to their choice of a relatively high OH concentra6 tion (4.5 x 10 molecules cm-3). Helmig et al. also estimated fluxes from veliical concentration gradients, but due to the small sample size (n = 5) and the large uncertainties associa~ed with individual estimates from this technique, these estimates cannot be used to validate the higher OR estimate. However, recent observations by Kuhn et al. [2007] and
192
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
Karl et al. [2007] suggest that earlier studies [i.e., Crutzen et al., 1985; Jacob and Wofsy, 1988; Greenberg et al., 2004] may have ul1derestimated OH by a factor of 5 or more. Such interpretations were supported by the recent reports about an OH recycling process by isoprene oxidation which may explain such high OH levels [Lelieveld et al., 2008]. These findings ,give rise to new discussions about the interactions ofbiogenic emissions and atmospheric chemistry [Guenther, 2008] and to isoprene and monoterpene flux estimates which may be underestimated by at least a factor offive. The tower-based measurement approaches listed in Table 1 include surface layer gradients (SLG), relaxed eddy accumulation (REA), disjunct eddy accumulation (DEA), and eddy covariance (EC). The labor-intensive SLG, REA, and DEA techniques used in these studies have resulted in small data sets, typically less than 20 measurements, and because the samples were transported to laboratories in Europe and the United States, uncertainties associated with sample storage were introduced. Additional uncertainties are associated with the assumptions that are required with these techniques. The continuous measurements of an EC system can provide much larger data sets and more accurate results. Midday average fluxes from all of the tower campaigns listed in Table 1 range from 1.5 to 8.3 mg m-2 h-1 for isoprene and from 0.2 to 1.7 mg m-2 h- I for total monoterpenes. The mean isoprene and monoterpene fluxes from these studies are positively correlated, and total monoterpene emissions tend to be 10% to 20% of the isoprene mass flux. Tower measurement systems are particularly useful for measuring diurnal and seasonal variations in BVOC fluxes. Karl et al. 's [2007] eddy covariance measurements show that emissions of isoprene, total monoterpenes, methanol, and acetone are all controlled by both light and temperature. There have been no attempts to continuously measure BVOC fluxes throughout a year, but several of the data sets shown in Table 1 include some information on seasonal variations. These limited data suggest that emissions in the dly season, especially in the later part, are considerably higher than in the wet season. This is in agreement with the ambient isoprene concentration measurements of Kesselmeier et al. [2002b] and TrostdOlf et al. [2004]. An increase in emissions during the dly season has been reported, due to higher sunlight and leaf temperature, but phenology [Kuhn et al., 2002a] may also have an important role. Karl et al. [2007] measured BVOC concentrations above various land cover types (wetlands, soybean fields, mixed agriculture, primmy forest) notih and east of Manaus using an airborne fast response proton transfer reaction mass spectrometer (PTRMS). These measurements were made on only two flights because the main focus of the Shldy was on biomass burning, which was primarily occun'ing south
of Manaus. Their aircraft measurements did not include fast response vertical wind fluctuations, required for eddy covariance measurements, but they were able to estimate fluxes from concentration variations. This limited study has provided the first data for characterizing BVOC emission variations over relatively large scales (25 to 10,000 km2). Isoprene emissions varied over about an order of magnitude (1 to 10 mg n12 h- I) after emissions were adjusted to account for variations in temperature and light. The observed emission variations conesponded with differences in land cover types. Rasmussen and Khalil [1988] scaled up a constant emission rate, similar to that repOlied by Zimmerman et al. [1988], to estimate annual isoprene fluxes of 60 Tg of isoprene for a 5 x 106 km2 area of Amazonia. Guenther et al. [1995] extrapolated Zimmerman et al.'s [1988] emission rate using variable temperature, light, and leaf area and estimated an annual emission of96 Tg isoprene and 14 Tg monoterpenes for a 4.33 x 106 km2 area representing all tropical rainforests. The Guenther et al. estimate used a land cover data set that classified much of the Amazon basin as seasonal tropical forest and other land cove~ types. Guenther et al. [2006] incorporated some of the additional data shown in Table 1 and estimated isoprene emissions that ranged both higher and lower than Guenther et al. 's [1995] estimates depending on the weather and land cover data used to drive emissions. Shim et al. [2005] used satellite CH20 observations and a global chemishy and transpoti model to constrain global isoprene distributions. Their estimates of global isoprene emissions were about 13% higher than Guenther et al. [1995], but their estimate for South America was about 35% lower. 3. ATMOSPHERIC CHEMISTRY OVER AMAZONIA: PRODUCTION OF OZONE FROM BIOGENIC HYDROCARBONS, NO,. AND THE ROLE OF OH In the absence of anthropogenically induced pollution, intensive tropical atmospheric chemistry and physics in Amazonia is'dominated by the biosphere. Biogenic emissions of volatile organic compounds contribute to particle formation (see section 4) as well as oxidative chemistry in the rainforest environment. This is a complex chemistty driven by light, OH radicals, ozone, and NOx. Within this context, reactive volatile organic compounds can be involved in both ozone production, as well as ozone consumption. The net effect of VOC on ozone is governed by the amount of available NOx (NO and N0 2). For a more detailed discussion, see relevant literature [i.e., Carter and Atkinson, 1996; Sanhueza et al., 1996; Jacob and WqfSy, 1988; Jacob et al., 2002; Neeb et al., 1997a, 1997b; Williams, 2004]. In the following section, we will concentrate on the oxidation chemistry of isoprene,
KESSELMEIER ET AL.
since it is thought to be the main compound released from tropical vegetation. Simplified schemes/are used here to help to understand the basic reactions. T~e atmospheric oxidation ofVOC initiated by OH as well ap'slb 3 is described by the following reaction scheme. A subsequent sequence of reactions of the resulting peroxy radical (R0 2) leads to stable products such as alcohols, carbony1s, and acids. Oxidant consumption VOCs + OH, 0 3 -7 oxidation products (i.e., alcohols, aldehydes, acids, ketones). Ozone, N02, and NO coexist in the atmosphere in a photostationary equilibrium. NO destroys ozone in reaction 3, but the ozone reforms through reactions 1 and 2, which also regenerates NO. In the presence ofVOCs, the R0 2 can convert NO to N0 2 without destroying ozone, leading to production. Photostationary equilibrium N02+ hv -7 NO + 0 (A ~ 420 nm), 0+ 02 (+M) -7 0 3 (+M), NO + 0 3 -7 N0 2 + O2, R02 + NO -7 RO + N0 2.
(1) (2) (3) (4)
An OH-initiated and NO.--cata1yzed oxidation of a VOC species leads to the formation of many intermediates and is tightly coupled to the conversion of NO to N0 2 [see Jenkin et al., 2002; Delwent et al., 2007]. Ozone production potentials have been related to NO.,! VOC ratios for different sites, ranging from remote to polluted areas [Chameides et al., 1992]. The relation between 0 3, NOx, and VOC has been shown to be driven by complex nonlinear photochemistry. These relations dramatically change in Amazonia from the wet to the dly season. Dry season conditions with higher temperatures lead to substantial higher emission rates of biogenic VOC species from the forest, even though leaf area indices can decrease due to the adaptation strategies of some tree species of dropping their leaves under drought conditions. The most prominent changes are caused by anthropogenic influences such as slash and burn techniques to change forest ecosystems into pasture. Not only is NO.,. released by these vegetation fires and further additional anthropogenic sources, but VOC . emission is also significantly increased by fires [see also Longo et al., this volume]. This is best reflected by emission ratios ofVOC species to other fire emissions as reported by Andreae and Merlet [2001]. On the other hand, natural
193
climate changes from the wet to dly season can also have a significant impact on VOC emissions [Guenther et al., 1999; Serr;a et al., 2001; Kesselmeier et al., 2002b; Kuhn et al., 2002a, 2004a]. The impact of climatic factors has been also clearly demonstrated by Sanderson et al. [2003] who illustrated the impact of climate change on both isoprene emissions and ozone levels. Global isoprene emissions are predicted to increase on a carbon basis from 484 Tg a-I for the 1990s to 615 Tg a-I for the 2090s, as a result of climate and vegetation distt'ibution changes; the authors calculated an ozone increase of 10-20 ppb in some locations. They a.1so noted that these changes in ozone levels were closely hnked to changes in isoprene surface fluxes in regions such as the eastern United States or southern China. However, this effect was much less marked over Amazonia and Africa because of lower levels of nitrogen oxides. The VOCINO x ratios change due to anthropogenically induced land us~ change. A closer look into the OR/Ozone chemistry directs our view to the oxidation products methyl vinyl ketone (MVK) and methacro1ein (MACR) [Kesselmeier et al., 2002b; Kuhn et al., 2007]. Under remote environmental conditions over Amazonia during wet season conditions, the only known source of MVK and MACR is isoprene oxidation. This is a nice tool :for evaluating the oxidation conditions as the ratio (MVK + MACR)/isoprene can provide an indication of the exttlnt of oxidation. The instantaneous ratio (MVK + MACR)/lsoprene is not purely photochemically driven, but is also influenced by the isoprene emission rate, the proximity to eniission sources, as well as by atmospheric mixing [Montzka et al., 1995]. For NOx levels between 0.2 and 0.5 ppb as reported for the different seasons [Andreae et al., 2002], the (MVK + MACR)/isoprene ratios agreed extremely well with the relationship shown in the work of Biesenthal et al. [1998]. The ratio of MVKlMACR is also useful in examining the oxidant conditions during the day. This ratio depends on the yields of MVK and MACR from isoprene oxidation and on the relative reaction rates of isoprene, MVK, and MACR with both OH and ozone. OH-induced isoprene oxidation leads to MVK and MACR with respective yields of 32% and 23% [Tuazon and Atkinson, 1990]. But MACR is removed more quickly through further OH oxidation than MVK, thus shifting the MVKlMACR ratio to higher values. In the dry season, the daytime ratios were typically in the range of 1.3-1.5 [Kesselmeier et al., 2002b], which is quite close to the relative OH production yield of 1.4 reflecting a relatively high 03/0H ratio [Starn et al., 1998], i.e., a relatively low oxidation capacity due to OH concentrations. Interestingly, the conesponding wet season data showed higher MVKlMACR values, but were difficult to discuss
194
KESSELMEIER ET AL.
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
4. FORMATION OF AEROSOL PARTICLES due to higher individual uncertainties of the data as stated AND CLOUD CONDENSATION NUCLEI (CCN) by the authors. In a very recent paper, Kuhn et al. [2007] FROM BIOGENIC HYDROCARBONS discussed the oxidation capacity in the tropical convective boundaIy tayer (CBL) in an early dly season (July 2001). On the basis of (MVK + MACR)/isoprene and MVKlMACR 4.1. Introduction ratios, they concluded that OH must be much higher than Atmospheric aerosols interact both directly and indirectly previously assumed. Based on observed vertical gradients with the Earth's radiation budget and climate. As a direct of isoprene and its primalY degradation products MVK and effect, the aerosols scatter or absorb sunlight. As an indirect MACR, they estimated a range of OH during the daytime of 3-8 x 106 molecules cm 3 . This number is an order of effect, aerosols in the lower atmosphere can modify number magnitude higher than estimates by current state-of-the-art and size of cloud droplets, changing how the clouds reflect atmospheric chemistry/transport models. Only such high and absorb sunlight, thereby affecting the Earth's radiation estimates were able to reconcile VOC fluxes derived from budget. Aerosols can also act as sites for chemical reactions an airbome mixed layer gradient (MLG) approach. These to take place (heterogeneous chemistry). Hence, they play an results clearly demonstrate that the influence of VOCs on important role in global climate and atmospheric chemistry. The formation of aerosol particles from the oxidation of the oxidation capacity in models has high uncertainties hydrocarbons is only one but an important contribution to [Lelieveld et al., 2004] and that we need many more field measurements in order to understand these chemical proc- the overall composition of atmospheric aerosols. It is well esses in the atmosphere. This becomes even more clear known that biogenic VOCs especially have a high aerosol when comparing model estimates and measurements of fOlmation potential. Consequently, tropical regions in parother oxygenated species (acetaldehyde, methanol, formic, ticular are believed to be globally relevant source regions for and acetic acid), indicating severe gaps in our understanding these kinds of aerosol particles. In general, the volatile aeroof the present budgets of these species [von Kuhlmann et al., sol precursors (e.g., terpenes) are first decomposed in the gas 2003]. The chemical processes within the oxidation scheme phase by bimolecular reactions followed by the fOlmation obviously are not fully understood. Recently, Lelieveld et al. of products with a lower volatility. Higher functionalized [2008] reported about an OH recycling process by isoprene compounds with hydroxyl, carbonyl, carboxyl groups, or which may explain high OH levels in a pristine atmosphere. groups containing heteroatoms are fOlmed in an oxidizing The authors propose that natural VOC oxidation, notably environment, which will either condense on existing parof isoprene, contributes to an efficient recycling of OH in ticles or even fOlm new aerosol particles (gas-to-particle 10w-NOx air through reactions of organic peroxy radicals. conversion). To distinguish this fraction of tropospheric Thus emission of such reactive trace gases might enhance aerosols from the direct input of particulate organics into the cleansing capacity of the atmosphere in remote areas like the atmosphere, it is specified as secondaIy organic aerosol (SOA). However, although biogenic SOA is the major focus the Amazonian rainforest. The above mentioned new reports on higher OH concen- of this chapter, sources and results of field measurements of trations in the Amazonian atmosphere than earlier expected primary organic aerosol constituents will also be discussed illuminates severe gaps in our knowledge in a region, which here, especially those primaIy organic contr'ibutions which has a particularly active photochemistry dl.,le to high solar ra- are linked to tropical regions. diation and atmospheric water vapor. Additionally, due to the strong convection in Amazonia surface emissions influence 4.2. Historical Background a higher part of the atmosphere than in most other regions. The first relationship between volatile organic compounds This tropical reactor with its biogenic sources of VOCs and and the formation of atmospheric particles was probably nitrogen oxides is characterized by its very sensitive atmoproposed by Arie Haagen-Smit at the Califomia Institute spheric chemistry. Any changes in land use will significantly of Technology (Caltech) in 1952 for a strongly anthropoaffect these sources. Furthermore, additional new anthropogenically influenced environment. Studying various aspects genic sources will occur. We have experienced such changes of the Los Angeles smog formation, he not only explained already when comparing wet season (undisturbed) and dly ozone and peroxide fOlmation by the photochemistry of the season conditions in view of ozone production as well as released hydrocarbons and nitrogen oxides, but also linked increase of aerosol numbers and cloud effects. Such effects the decrease in visibility during smog episodes to the condenon atmospheric chemistry and physics can be expected to sation of aldehydes and acids formed by the oxidation of orbecome even stronger with increasing disturbance in the reganic volatiles. In 1960, F.W. Went, director of the Missouri lation between biology, chemistry, and physics. T
195
Therefore, it is useful to characterize the different sources (or source types) of organic aerosols in terms of their individual chemical composition. Selected source specific compounds can then be used as tracers for the origin of aerosols or to estimate the contribution of the different source types to the measured aerosol. There are a variety of biogenic and anthropogenic sources of primary organic aerosol constituents. Although exact numbers are missing, globally the input fi'om natural sources is believed to be prominent. However, for the aerosol com4.3. General Source Processes jor Atmospheric Particles position on the local or regional scale, even weak sources can Particles in the atmosphere are often divided into the cat- be impOliant [Schauer et al., 1996]. The following sources egories ofprimary and secondaly particles according to their belong to the most important sources of primary organic formation processes. Primary particles are released directly aerosols that have been characterized in tropical regions: (1) into the atmosphere, whereas secondary particles are pro- biomass buming, (2) plant abrasion, and (3) suspension and duced within the atmosphere as a consequence ofthe conver- release of bioaerosols. Biomass buming is a strong source for atmospheric aerosion ofvolatile precursors into low or nonvolatile substances. Formation processes of primary pariicles are basically me- sols, producing about four times more than fossil fuel bumchanical production (abrasion, suspension) and production ing [Kuhlbusch, 1998]. A main product and a general tracer during combustion processes (condensation of hot vapors used for biomass buming is levoglucosan (l,6-anhydro-~-D or fOlmation inside flames as described for soot particles) glucopyranose), an anhydro-sugar derived from the thermal [Seinjeld and Pm1dis, 1998]. In general, mechanical pro- degradation of cellulose during the combustion process [Sicesses create coarse particles (>2.5 flm), whereas combus- moneit, 1999]. Anhydro-sugars seem to be the most abuntion processes create fine particles which might coagulate dant compounds produced during the combustion process right after production (e.g., chain aggregates of soot parti- of plant material. I Other major compound groups identicles). Secondary atmospheric particles also belong to the fied in smoke particles from biomass buming are alkanes, fine particle fraction. The low volatile compounds formed alkenes, alkanoic ,acids, di- and triterpenoids, monosacchafrom the oxidation of the precursor compounds (e.g., VOCs, rides, methoxyphenols, and PAHs [Simoneit, 2002]. Some reduced sulfur compounds) can either form new atmospheric ofthese constituents are derived fi'om thermally altered plant pariicles (homogeneous nucleation), or they can condense material (like anhydro-sugars); others are unchanged ingreonto preexisting pariicles leading to increased particle size dients (like some wax alkanes). Especially lignin pyrolysis products like vanillic acid can be used as tracers for certain and mass and to an alteration of the chemical composition. Aerosol pariicles from biogenic VOCs contribute to the plant species. Plant abrasion is mainly induced by wind driven mecarbonaceous aerosol fraction. The term "carbonaceous aerosols" includes all aerosol constituents which are based chanical force, like the tUbbing of leaves against each other on carbon, i.e., the variety of different organic compounds [Rogge et al., 1993]. Identified substances in aerosols from produced from VOC oxidation or combustion processes; plant abrasion (green and dead leaves) are mainly constituhowever, elemental carbon, bioaerosols, and a few inorganic ents of the epicuticular plant waxes: n-alkanes, n-alkanals, carbon constituents are also included. The concentration of n-alkanols, n-alkanoic acids (fatty acids). These compound inorganic carbon, essentially as carbonate, is on average neg- groups are not velY specific for biological sources, but due ligible, at least when considering submicrometer particles. to their biosynthesis, specific patterns in carbon numbers of plant-derived wax components can be observed. Leaf wax alkanes have a strong odd carbon number predominance 4.4. Molecular Composition ojG/ganic Aerosols with dominant carbon numbers C29, C3l, and C33, whereas The organic fi'action of atmospheric particulate material alkanoic acids, alkanals, and alkanols are dominated by even can contain a large number of diverse molecular species. The. carbon numbers. Fossil fuel constituents show no predomicomposition is mainly dependant on the aerosol source with nance in carbon numbers. Therefore, it is possible to use possible modifications during atmospheric transport. The or- these specific patterns to identify contributions of plants to ganic mixture ranges from nonpolar hydrocarbons (alkanes) atmospheric aerosols [Simoneit et al., 1988]. to highly polar and water-soluble components, such as short Bioaerosols are primary organic aerosols with diameters dicarboxylic acids or sugars, to macromolecular organics. from ~ 10 nm to 100 flm that are either alive, cany living
Botanical Garden and fOlmer colleague of Haagen-Smit at Caltech, published allextensive article in Nature titled "Blue hazes in the atmosphere." Based on his observations when staying at a couptlyside site and everyday experiences, as well as his kn9wledge about secondaIy plant products, he also connected the occunence of the natural phenomena with the volatilization and gas phase oxidation of terpenes released from tel:restrial vegetation.
196
KESSELMEIER ET AL.
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
organisms or are released from living organisms like bacteria, fungi, algae, viruses, pollen, spores (e.g., from ferns), cell debristtbiofilms, etc. [Ariya and Amyot, 2004]. Generally, the size of bacteria is around 1 flm, pollen grains are mostly larger than 10 flm, and viruses are in the nanometer range. Each of these "particles" is usually itself a complex mixtur~ of various molecules. Bacteria may spread diseases, can act as cloud condensation nuclei and ice nuclei. They have been found even at high i;1ltitudes in the atmosphere and remote regions. Bacteria and fungi can be suspended from soil or plants by wind and from water surfaces by bubblebursting processes or sea spray. They can also be released by anthropogenic sources like farming, waste, and wastewater treatment. Bacteria can even live and grow in atmospheric water droplets like fog [Fuzzi et al., 1997] or even super cooled cloud droplets [Sattler et al., 2001]. The importance ofbioaerosols for atmospheric aerosol content is very unclear. Some authors repmi bioaerosols as being major components, whereas other studies report only insignificant contributions from bacteria to the atmospheric particulate material. In Amazonian aerosols, the nocturnal increase of coarse size particle mass (PMlO-PM2) was attributed to fungi [Graham et al., 2003a, 2003b]. During the wet season, biogenic particles accounted for 55-92% of the fine particle mass and for 65-95% of the coarse particle mass sampled in the Amazonian basin [Artaxo et al., 1988, 1990]. For Russia (Lake Baikal) and Germany (Mainz), contributions of primary biological aerosol particles (including plant fragments, pollen, etc.) to total atmospheric particles (>0.2 flm) were reported to be in the range of20% to 30%, respectively [Jaenicke, 2005]. 4.5. Sources and Composition ofSecondmJ! O/ganic Aerosols
As introduced above, secondaty organic aerosols (SOAs) are produced (1) by gas-phase oxidation of volatile organic compounds that can either form new particles, or condense onto preexisting particles, (2) by heterogeneous reactions on particle surfaces, or (3) by in-cloud processing. Precursors of organic SOA are mostly volatile reactive biogenic (e.g., terpenes), or anthropogenic (e.g., aromatics) hydrocarbons. Products formed can be relatively low volatile organics, which convert almost completely to the particle phase, or semivolatile organics, which partition between the gas and particle phase. This gas-particle patiitioning of semivolatile (and also low volatile) compounds can be described by gasparticle partitioning models [Pankow, 1994; Odum et al., 1996], in which the dependence of the concentration of an individual organic compound, i, in the patiicle phase, on the available absorbing organic aerosol mass (MO), the patii-
tioning coefficient of compound i and the concentration of i in the gas phase has the relationship: caef = cgas x Kom x MO,
(5)
where Kom is the partitioning coefficient of i (m3 flg-l) (temperature dependent), Caef is the concentration of compound i in the absorbing organic particle phase (ng m-3), c gas is the concentration of i in the gas phase (ng m-3), and MO is the concentration of the absorbing organic phase in the aerosol (flg ill3). Although the underlying equations are rather simple, the estimation of SOA mass is complicated for ce1iain locations and atmospheric conditions, as well as for use in regional and global SOA modeling, the strong temperature dependence of the partitioning coefficient also adds to the complication [Takekawa et al., 2003]. During the first years of SOA-research, attention was paid just to the rather low volatile or semivolatile oxidation products, which directly contribute to the patiicle phase by gasto-patiicle conversion. More recent studies also show that volatile carbonylic products formed in the gas phase oxidation of organics may contribute over a longer period of time to the SOA mass by the formation oflow volatile oligomers, for example, via acid catalyzed reactions of aldehydes or ketones on the patiicle surfaces or inside the particles (aldol reaction/condensation, acetal formation). These processes result in increased particle mass and a lower volatility. This might be the case for both biogenic and anthropogenic precursors [Jang et al., 2002, 2004; Gao et al., 2004; Iinuma et al., 2004; Kalberer et al., 2004]. There is also evidence for the direct formation of oligomeric products by heterogeneous reactions of unsaturated gas phase compounds (e.g., isoprene) on patiicle surfaces [Limbeck et al., 2003]. Several groups speculate that these oligomeric products formed from gaseous precursors could represent a substantial fraction of the so-called "humic-like substances" (HULlS) often identified in atmospheric aerosols. HULlS is a collective term for a group of patiicle phase compounds unidentified at the molecular level, which add to the water-soluble organic carbon. This generation of new particle phase products from gas phase constituents during the atmospheric lifetime of aerosols is part of the so-called atmospheric ageing of organic particles, a process that is currently not well characterized. Besides the incorporation of reactive gas phase species into the organic aerosol fraction by oligomer formation, ageing also includes the degradation or chemical modification of patiicle phase constituents by atmospheric oxidants. Since these chemical modifications will result in alterations of the physical (volatility, light absorption, light scattering) and
physicochemical properties (water solubility, CCN-activity) of atmospheric aerosols, the investigation of these processes has to be addressed ill future research on organic aerosols. The incorporatid~h of SOA formation into atmospheric models is not a!!)Sasy task, since a variety of chemical and physicochemical processes influence the SOA particle mass in the ambient atmosphere. A sensitivity analysis of SOA production and transport modeling [Tsigaridis and Kanakidou, 2003] showed an uncertainty factor of about 20 in predicting SOA production when considering the different influences of partitioning, ageing, and MO, excluding the uncertainties of precursor emissions and individual oxidation pathways. This results in an annual global production of SOA of 2.55 to 47.12 Tg of organic matter per year. Another study showed SOA production using the partitioning method and the bulk yield method (ignoring the partitioning mechanism) to be 15.3 and 24.6 Tg per year [Lack et al., 2004], respectively. Precursors of biogenic SOA in the continental environment are mainly unsaturated hydrocarbons, namely, (mono-) terpenes, sesquiterpenes, and isoprene. The SOA-forming potential of terpenes is well known and has been intensively investigated, e.g., [Went, 1960a; Yokouchi and AlI;be 1985' Zhang et al., 1992; Hoffinann et al., 1997, 1998; Ka~OlI1'a; et al., 1998; Griffin et al., 1999; Yu et al., 1999; O'Dowd et al., 2002], whereas isoprene was only recently found to form low volatile secondaty products [Claeys et al., 2004a, 2004b]. Known products of atmospheric isoprene oxidation are polyols and acidic compounds like 2-methyltetrols and 2,3-dihydroxymethacrylic acid. It was estimated that isoprene might add about 2 Tg of polyoIs to atmospheric SOA.
00H
~OqH
-tV
Pinicacid
(0:
4.6. Field Measllre/Ilents
Very few measurements of the aerosol composition in forested tropical regions exist, especially those focusing on the organic aerosol fraction. Nevertheless, some dedicated studies, partly in the framework of LBA, investigated the
~OOH ~HO Pinonaldehyde
o
~ ~OOH Carie acid
This is a substantial amount, although terpenes may add 10 times more to SOA. The most frequently studied and most impmiant SOA-fonning reactions of terpenes are gas-phase oxidations by ozone, OH- and NOr radicals. The oxidation of terpenes by ozone, OH radical, and photosmog generates a variety of oxygenated gas phase [Calogirou et al., 1999] and particle phase products which have been identified in chamber experiments [Christoffersen et al., 1998; Hoffinann et al., 1998; Yll et al., 1999; Glasius et al., 2000a; Koch et al., 2000; Larsen et al., 2001; Jaolli and Kamens, 2003a, 2003b, 2003c; Winterhalter et al., 2003]. Known terpene oxidation products relevant for SOA production mainly contain carbonyl, alcohol, and carboxylic acid functional groups. Products bearing carboxylic acid functional groups have low volatility and are therefore especially interesting for SOA formation. Figure 2 shows some ~mpmiant products from monoterpene oxidation. Recently, 1t was proposed that peroxides could also represent a major part of the SOA formed by terpene ozonolysis [Bonn et al., 2004], a suggestion which was recently confirmed by chamber studies [Docherty et al., 2005]. As mentioned above, oligomer formation from (semi-) volatile oxygenated terpene oxidation products might also contribute to SOA formation from biogenic precursors.
H
Pinonic acid
OOH
Caronic acid
197
OOH
1()..Hydroxypinonie acid
2~OH 02~H 0 Sabinic acid
Ketolimonic acid Ketolimononic acid
Figure 2. Important monoterpene oxidation products.
198
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS KESSELMEIER ET AL.
Table 2. Concentration Ranges of Some Sugars and Levoglucosan in the Atmospheric Particle Phase Measured at Different Locations" Concentration (ng m-3) Locati5n
Description
PM
Glucose
Sucrose
CI
Levogh\c.
Mycose
rural/marine
TSP
11-111 (50)
TSP 0--4 2.5 1.88-16.3 (7.9) >2.5 1.7-124 (48) rain-forest/biomass 2.5 14-62 (37) burning TSP 10-2210 (940) urban marine/remote rain-forest/remote
urbanlbiomass buming
6--444 (86) 0-1.8 <0.06-1.9 (1.5) <0.06-77 (37) 0.8-26 (7) 15-3060 (1108)
TSP
2.5-30 (13)
8-74
0-0.2 4.9-12 (6.9) 21-90 (48) 5-18 (9)
0.2-1.3 7.73-32.9 (15) <0.04--4.79 (1.79) 1182-6900 (2460)
8-1660 (477)
12-2452 (839)
Simoneit et al. [2004b] Graham et al: [2003a, 2003b] Graham et al. [2002] Simoneil et al. [2004b]
1162-33,400 (14,460)
"Mean concentration is shown within parentheses.
chemical composition of organics in the particle phase. In the following paragraphs, some results of these measurements are presented, and they are put into perspective with measurement results from other regions. Table 2 shows atmospheric concentrations of selected sugars and levoglucosan in different regions. Levoglucosan is a degradation product of cellulose and almost exclusively produced by combustion of plant material. Therefore, it is not surprising that concentrations are low in remote areas (around 1 to about 10 ng m-3), higher in urban areas due to the use of wood as fuel (roughly 100 to more than 1000 ng m-3), and reach highest concentrations in aerosols from massive biomass burning in tropical areas (around 1000 to more than 10,000 ng m-3). Sugar concentrations also show vely different concentrations from less than 1 ng m- 3 above the oceans, but they were also observed in substantial amounts, e.g., around 1000 ng m-3 in aerosol samples from Chile. Usually, concentrations of glucose, sucrose, and mycose are lower than 100 ng m-3 . The source of sugars is believed to be mainly soil dust (including suspended microorganisms) [Simonei( et al., 2004a]. Other observations in the Amazonian rainforest point to primaty contributions from living plants, e.g., glucose and sucrose may derive from pollen and fern spores, or mycose may derive from fungal spores [Graham et al., 2003a, 2003b]. Consequently, sugar concentrations for glucose, sucrose, and fructose were higher during the daytime, and levels ofmycose (trehalose), arabitol, and mannitol were higher during the night when a strong release of fungal spores occurred. Mycose, arabitol, and mannitol are well-known constituents of fungal spores and sucrose, glucose, and fmctose are known to be present in pollen grains. Levoglucosan, sugars and the lipid constituents represent the primary fraction of organic aerosols, whereas short-chain dicarboxylic acids have primary and secondary sources.
Oxocarboxylic acids derive mostly from secondary processes, either the oxidation ofVOCs or the further oxidation of carbonyls, and mono- or dicarboxylic acids. Oxalic acid (and other di- and oxocarboxylic acids) levels are rather low in marine and remote areas, although secondaty production can be observed in remote arctic regions [Kawamura et al., 2005]. Concentrations of oxalic acid are around 100 ng m- 3 in natural or remote regions as shown in Table 3. Urban concentrations are in the order of a few hundred ng m-3, and the highest concentrations are observed in biomass burning aerosols (more than 1000 ng m-3). The same trend can be observed for malonic and succinic acid, although concentrations are substantially lower. Particle phase constituents from the oxidation of biogenic VOCs (e.g., terpenes) are obviously of secondary origin. Therefore, it is mainly forested regions that are influenced by oxidation products from biogenic VOC. The acidic products of monoterpene oxidation add to the water-soluble OC, and therefore, they might be impOliant for the fonnation of CCN. Moreover, the oxidation ofterpenes has been linked to the formation of new particles above forests [0 'Dowd et al., 2002]. Although SOA from monoterpene oxidation seems to be very important for the global SOA budget [Chung and Seinfeld, 2002; Tsigaridis and Kanakidou, 2003], there exist only few measurements of paIiicle phase constituents derived from monoterpene oxidation. Mostly low volatile products of 0.- and ~-pinene, namely, pinic and pinonic acid have been measured. Table 4 shows a selection of reported concentrations. Measurements of ambient concentrations of products from other important monoterpenes, such as from limonene, 3-carene, or sabinene, are even less frequent. The concentrations of these terpene oxidation products in the paIiicle phase vary strongly, depending on time and location. Pinic and pinonic acid concentrations can range from
d
I
.Concentration (ng mg 3)
Reference
Oxalic Acid
Malonic Acid
Succinic Glyoxylic Acid Acid
340
244
117
\}
Gosnan, Jeju Island, Korea N-Pacific Amazonia, Brazil Rondonia, Brazil Santiago, Chile Kuala Lumpur, Malaysia
199
. S an G yoxalm the Atmospheric Pm1icle Phase Sampled in Distinct Areas"
Table3.ConcentrationsofDifferentA'·d
Desc~iption PM
Location
/'
Vienna, Austria NW-Pacific
urban
Tokyo, Japan Canada Amazonia (2001) Amazonia (1999)
TSP
marine
TSP
3.6--430
0.1-53
0.1-37
urban
TSP
89-815
16-163
13-168
Arctic
2
6.5-59 (26) 8.8-148 (57) 51-691 (329)
1.2-20 (7.6) 4.4--48 (22) 7.4-149 (56)
1.6-19 (7.7) 2.1-8 (4.9) 3.5-76 (31)
natural rainforest rainforest (burning season)
2.5 2.5
22
Maleic Acid
Malic Acid
Reference
63
Limbeck and Puxbaum [1999] 0.1-2.5 Mochida et al. [2003] 5.3-105 6.6-138 2.5-31 0-99 Kawamura and Yaslli [2005] <0.01-39 0.37-2.76 007 16 <0.01-6.5 Kawamura et al. . -. (1.5) (11.9) (1.3) (0.59) [2005] 0.19-0.55 4.2-24 Graham et al. (14.5) [2003a, 2003b] (0.37) 12-146 1.6-66 1.2-28 0.6-13 Graham et al. (25) (11) (6.3) (67) [2002]
"Values are mean concentrations or concentration ranges, sometimes given with
~elow 1 ~g m- to about 100 ng m-3 even at the same loca-
Pymvic Acid
. . mean concentrations m parentheses.
3
tIon. ~gam, veIY few results are repOlied about tlieir atmosphenc concentrations in forested tropical regions. 4.7. SummaJy and Research Needs
Organic contri~utions to the atmospheric particle phase
~re lrn.0wn .to denve ~'om primmy and secondary sources.
~ow-:olatile products, which formed immediately during the ' leachon of the precursor hydrocarbon and the oXI'dant 'd d ' ,wele con,sl .ere to ~ontnbute to aerosol formation. However, there IS. a growmg: awareness that chemical processes are also takmg place within the particle phase. The atmospheric :elevance of these I;eactions, for example, in connection with lsop~'ene oxidationlproducts, have to be evaluated in future stud.les. Moreover" there still exists a poor understanding of the lllfluence h 'of biogenically derived SOA components on t e atmosphenc water cycle (CCN or IN activity) Whl' h . t t' 11 '. , C IS po en Ja y a slgmficant linle between the terresh'ial biosphere and the atmosphere in forested tropical regions.
s~ecJally ~n the tropICS, these sources can account for the maJor fra~tlOn of submicrometer aerosols. Despite a series (e g LBA) , smog chamber ofmternahonal research proiects . ~. ., ~tudIes, and model development, a quantitative understandl~g of the processes linking emissions and their contributlO~ ~o the tl:opospheric patiicle phase still does not exist. 5. LBA CONTRIBUTIONS TO UNDERSTANDING ThIS IS .es~eclally h1.le for tropical forests, since not only are BVOC EMISSIONS AND THEIR ROLE ~he emlSSlOns of aerosol precursor gases poorly characterIN THE EARTH SYSTEM I~ed, but velY ~ew measurements of the chemical compositIon ofthe tropIcal organic aerosol have been made. Another Investigations of Amazonian BVOC began in the late l~rgely unknown area is the destiny of biogenic VOC oxida1970s " IS a . and early 1980s and demonstrated that A mazoma hon products within the particle phase. Up to now, only the major source of atmospheric BVOC and that th ese emlSSlOns " Table 4. Concentrations (Range .
'M
Ol
ean
) fL 0
. ow Volatile Products From the Oxidation of
a,-
an
Concentration (ng m 3) Location Tabua, Portugal
Time Aug 1996
Nova Scotia, Canada Jul1996 Pertouli, Greece Aug 1997 Amazonia, Brazil Jul2001
PinicAcid 0.39 83 0.48-0.59 0.4--4.4 1.1 (fine and coarse)
. Pinonic Acid 7.1 98 (cis-) 1.6--43 (trans-) 0.13-0.39 . 1-25.7
Norpinonic Acid 0.14-38 (cis-)
0-14 (trans-) 0.04-0.24
d
. ,rp'mene m. Ambient Aerosols A
Reference Kavollras et al. [1999] Yu et al. [1999] [Kavouras et al. [1999] Graham et al. [2003a, 2003b]
200
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
have a substantial impact on the chemical composition ofthe atmosphere. Global emission, chemistry, and transport modeling studies in the 1990s provided additional evidence that Amazoniafi BVOC emissions are an important component of the emih system. However, they also introduced some controversy associated with reconciling the large estimates of isoprene emissions with the observed concentrations of oxidation products. Since the emission estimates were based on relatively few measurements, and because very little was known about the factors controlling these emissions, many modeling studies implemented emission rates that were considerably lower than estimates based on field observations, but were within the large range of uncertainty associated with these emission estimates. LBA field campaigns dramatically increased the number of field studies of isoprene and monoterpenes emissions. Some LBA studies included direct flux measurements, in contrast to previous observations that only provided an indirect means of estimating terpenoid fluxes. The LBA results generally agree with earlier studies, but they also showed that there is substantial temporal and spatial variability. Additional LBA measurements provided an initial characterization of biogenic oxygenated VOC emissions. LBA studies also included improved efforts to integrate observations of chemistry, transport, and emissions. These studies provide evidence that an improved understanding of Amazonian atmospheric chemistty and cloud processes is needed and that this may explain at least some of the inconsistencies between estimates of isoprene emissions and atmospheric distributions of oxidation products. The improved logistical capabilities associated with the LBA research program provide a convenient oppOliunity for future investigations of BVOC emissions and their role in the emih system. Research priorities include extending efforts to integrate investigations of emissions, chemistry, transpOli, and cloud processes. Future studies should target a wide range of biogenic VOCs, including sesquiterpenes and oxygenated VOCs, and their products, including formaldehyde and CO (which can be observed with satellite-based remote sensing), and should investigate the impact ofBVOC on oxidant and aerosol distributions and on the carbon and water cycles. Additional research priorities include inv~s tigations of the processes contt'olling seasonal and spatl~l variations and the response to a changing earth system. This can be accomplished through a combination of tower-based multiyear continuous flux measurements of temporal ~ari ations, aircraft-based direct flux measurements of regIOnal variations and enclosure-based process studies. Significant advances in our understanding of the processes controlling Amazonian BVOC emissions and their impact on the emih system require a continued commitment and enhancement
KESSELMEIER ET AL.
of the intemational and multidisciplinary collaborations established by the LBA research program. Acknowledgments. We thank Jonathan Williams, Max Planck Institute for Chemistty, Mainz, Germany, for his help with the section about atmospheric chemistry. We greatly acknowledge the work of Tracey Andreae in the final editing of the manuscript.
REFERENCES Andreae, M. 0., and P. Merlet (2001), Emission oftt'ace gases and aerosols from biomass burning, Global Biogeochem. Cycles, 15(4), 955-966. . Andreae, M. 0., R. W. Talbot, T. W. Andreae, and R. C. Harnss (1988), Formic and acetic acid over the central Amazon region, Brazil: 1. Dry season, J. Geophys. Res., 93, 1616-1624. Andreae, M. 0., et al. (2002), Biogeochemical cycling of carbon, water, energy, trace gases, and aerosols in Amazonia: The LBAEUSTACH experiments, J. Geophys. Res., 107(D20), 8066, doi: 10.1029/200 IJD000524. Araujo, A. C., et al. (2002), Comparative measurements of car~on dioxide fluxes from two nearby towers in a central Amazoman rainforest: The Manaus LBA site, J. Geophys. Res., 107(D20), 8090, doi: 10.1 029/200IJD000676. Ariya, P. A., and M. Amyot (2004), New directions: The role of bioaerosols in atmospheric chemistry and physics, Atmos. Environ.,38,1231-1232. Artaxo, P., H. Storms, F. Bmynseels, R. Van Grieken, and W. Maenhaut (1988), Composition and sources of aerosols from the Amazon Basin, J. Geophys. Res., 93,1605-1615. Artaxo, P., W. Maenhaut, H. Storms, and R. Van Grieken (1990), Aerosol characteristics and sources for the Amazon Basin during the wet season, J. Geophys. Res., 95,16,971-16,985. Ayres, 1. M. (1986), White Uakaris and flooded forests, Ph.D. thesis, 338 pp., Cambridge Univ., Cambridge. Ayres, J. M. (1993), As Matas da VGrzea do MamirauG. MCT/ CNPq, 90 pp., Sociedade Civil Mamiraua, Brasilia.. ., Baker, T. R., et al. (2004), Are Amazonian forest plots mcreasmg m biomass?, Phi/os. Trans. R. Soc. London, Ser. B, 359, 353-365. Baldwin, 1. T., R. Halitschke, A. Paschold, C. C. von Dahl, and C. A.. Preston (2006), Volatile signaling in plant-plant interactions: "Talking trees" in the genomics era, Science, 311(5762), 812-815. Biesenthal, T. A., 1. W. Bottenheim, P. B. Shepson, S.-M. Li, and P. C. Brickell (1998), The chemistIy of biogenic hydrocarbons at a IUral site in eastern Canada, J. Geophys. Res., 103(DI9), 25,487-25,498. Bonn, B., R. von Kuhlmann, and M. G. Lawrence (2004), Hi~h contribution of biogenic hydroperoxides to secondary orgamc aerosol formation, Geophys. Res. Lett., 31, LI0I08, doi: 10.1029/ 2003GL019172. Braga, P. 1. S. (1979), Subdivisao fitogeogrMica, tipos de ;re.getayao, conservayao e inventario floristico da floresta amazomca, Acta Amazonica, Suppl. 9,53-80.
Calogirou, A., 8. R. Larsen, and D. Kotzias (1999), Gas-phase terpene oxidation products: J\l'eview, Atll/os. Environ., 33, 1423-1439. Carlier, P., H. Hannachi, and G. Mouvier (1986), The chemistry of carbonyl-cor~pounds in the atmosphere-A review, Atmos. Environ., 20(~4), 2079-2099. Carswell, F. E.;'et al. (2002), Seasonality in CO 2 and H 2 0 flux at an eastern Amazonian rainforest, J. Geophys. Res., 107(D20), 8076, doi: 10.1 029/2000JD000284. Carter, W. P. L., and R. Atkinson (1996), Development and evaluation of a detailed mechanism for the atmospheric reactions of isoprene and N0.r, Int. J. Chem. Kinet., 28, 497-530. Chameides, W. L., et al. (1992), Ozone precursor relationships in the ambient atmosphere, J. Geophys. Res., 97(D5), 6037-6055. Chebbi, A., and P. Car'lier (1996), Carboxylic acids in the troposphere, occurrence, sources, and sinks: A review, Atmos. Environ., 30(24), 4233-4249. Chou, W. W., S. C. Wofsy, R. C. Harriss, J. C. Lin, C. Gerbig, and G. W. Sachse (2002), Net fluxes of CO 2 in Amazonia derived from aircraft observations, J. Geophys. Res., 107(D22), 4614, doi: 1O.1029/200IJDOOI295. Christoffersen, T. S., et al. (1998), cis-Pinic acid, a possible precursor for organic aerosol fonnation from ozonolysis of alphapinene, Atmos. Environ., 32,1657-1661. Chung, S. H., and J. H. Seinfeld (2002), Global distribution and climate forcing of carbonaceous aerosols, J. Geophys. Res., 107(DI9), 4407, doi: 1O.1029/200IJD001397. Claeys, M., et al. (2004a), Fonnation of secondary organic aerosols through photooxidation of isoprene, Science, 303, 1173-1176. Claeys, M., W. Wang, A. C. Ion, 1. Kourtchev, A. Gelencser, and W. Maenhaut (2004b), Formation of secondary organic aerosols from isoprene and its gas-phase oxidation products through reaction with hydrogen peroxide, Atmos. Environ., 38, 4093-4098. Cmtzen, P. 1., A. C. Delany, J. Greenberg, P. Haagenson, L. Heidt, R. Lueb, W. Pollock, W. Seiler, A. Wartburg, and P. Zimmerman (1985), Tropospheric chemical composition measurements in Brazil during the dry season, J. Atmos. Chem., 3, 233-256. Cmtzen, P. J., et al. (2000), High spatial and temporal resolution measurements of primary organics and their oxidation products over the tropical forests of Surinam, Atmos. Environ., 34, 1161-1165. Davis, K. 1., D. H. Lenschow, and P. R. Zimmerman (1994), Biogenic norllllethane hydrocarbon emissions fi'om tethered balloon observations, J. Geophys. Res., 99, 25,587-25,598. Derwent, R. G., M. E. Jenkin, N. R. Passant, and M. 1. Pilling (2007), Photochemical ozone creation potentials (POCPs) for different emission sources of organic compounds under European conditions estimated with a master chemical mechanism, Atmos. Environ., 41, 257Q.-2579. Dicke~ M., A. A. Agrawal, and 1. Bruin (2003), Plants talk, but are they deaf?, Trends Plant Sci., 8, 403-405. Dindorf, T., U. Kuhn, L. Ganzeveld, G. Schebeske, C. Ciccioli, C. Holzke, R. Kiible, G. Seuferi, and J. Kesselmeier (2006), Significant light and temperature dependent monoterpene emissions from European beech (Fagus sylvatica L.) and their potential impact on the European volatile organic compound budget, J. Geophys. Res., 111, D 16305, doi: 1O.1029/2005JD006751.
201
Docherty, K. S., W. Wu, Y. B. Lim, and P. 1. Ziemann (2005), Contributions of organic peroxides to secondary aerosol formed from reactions of monoterpenes with 0-3, Environ. Sci. Technol., 39, 4049. Fehsenfeld, F., et al. (1992), Emissions of volatile organic compounds from vegetation and the implications for atmospheric chemistty, Global Biogeochem. Cycles, 6, 389-430. Fuzzi, S., P. Mandrioli, and A. Perfetto (1997), Fog droplets-An atmospheric source of secondary biological aerosol particles, Atmos. Environ., 31, 287-290. Gao, S., et al. (2004), Particle phase acidity and oligomer formation in secondary organic aerosol, Environ. Sci. Technol., 38, 6582-6589. Geron, C., A. Guenther, 1. Greenberg, H. W. Loescher, D. Clark, and B. Baker (2002), Biogenic volatile organic compound emissions from a lowland tropical wet forest in Costa Rica, Atmos. Environ., 36, 3793-3802. Geron, C., S. Owen, A. Guenther, 1. Greenberg, R. Rasmussen, 1. H. Bai, Q. 1. Li, and B. Baker (2006), Volatile organic compounds from vegetation in southern Yunnan Province, China: Emission rates and some potential regional implications, Atmos. Environ., 40, 1759-1773. Glasius, M., M. Lahaniati, A. Calogirou, D. Oi Bella, N. R. Jensen, 1. Hjorth, D. Kotzias, and B. R. Larsen (2000a), Carboxylic acids in secondary aerosols from oxidation of cyclic monoterpenes by ozone, Environ. Sci. Tee/mol., 34,1001-1010. Glasius, M., et al. (7000b), Sources to formic acid studied by carbon isotopic analysis and air mass characterization, Atmos. Environ., 34, 2471-2479. Graham, B., O. L. !Mayol-Bracero, P. Guyon, G. C. Roberis, M. O. Andreae, S. Qecesari, M. C. Facchini, P. Artaxo, W. Maenhaut, and P. Kiill (2002), Water-soluble organic compounds in biomass burning aerosols over Amazonia: I. Characterization by NMR and GC-MS, J. Geophys. Res., 107(D20), 8047, doi: 10.1 029/200110000336. Graham, B., et al. (2003a), Composition and diurnal variability of the namral Amazonian aerosol, J. Geophys. Res., 108(D24), 4765, doi: 10.1 029/2003JD004049. Graham, B., P. Guyon, P. E. Taylor, P. Artaxo, W. Maenhaut, M. M. Glovsky, R. C. Flagan, and M. O. Andreae (2003b), Organic compounds present in the natural Amazonian aerosol: Characterization by gas chromatography-mass spectrometry, J. Geophys. Res., 108(024), 4766, doi:10.1029/2003JD003990. Greenberg, 1. P., and P. R. Zimmerman (1984), Nomnethane hydrocarbons in remote tropical, continental, and marine atmospheres, J. Geophys. Res., 89(03), 4767-4778. Greenberg, J. P., A. B. Guenther, G. PetI'on, C. Wiedinmyer, O. Vega, L. V. Gatti, 1. Tota, and G. Fisch (2004), Biogenic VOC emissions fi'om forested Amazonian landscapes, Global Change Bioi., 10(5),651-662. Gregory, G. L., et al. (1986), Air chemistry over the tropical forest of Guyana, J. Geophys. Res., 91, 8603-8612. Griffin, R. 1., O. R. Cocker III, 1. H. Seinfeld, and D. Oabdub (1999), Estimate of global atmospheric organic aerosol fi'om oxidation of biogenic hydrocarbons, Geophys. Res. Lett., 26, 2721-2724.
202
KESSELMEIER ET AL.
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
Guenther, A. (2002), The contribution of reactive carbon emissions from vegetation to the carbon balance of terrestrial ecosystems, Chemosphere, 49, 837-844. Guenther, A. ~2008), Atmospheric chemistiy: Are plant emissions green?, Nature, 452, 701-702. Guenther, A., et al. (1995), A global model of natural volatile organic compound emissions, 1. Geophys. Res., 100, 8873-8892. Guenther, A., B. Baugh, G. Brasseur, l Greenberg, P. Harley, L. Klinger, D. Serva, and L. Vierling (1999), Isoprene emission estimates and uncertainties for ~he Central African EXPRESSO study domain, 1. Geophys. Res., 104, 30,625-30,639. Guenther, A., T. Karl, P. Harley, C. Wiedinmyer, P.1. Palmer, and C. Geron (2006), Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols fi'om Nature), Atmos. Chem. Phys., 6, 3181-3210. Harley, P., et al. (2004), Variation in potential for isoprene emissions among neotropical forest sites, Global Change BioI., 10(5), 630--650. Helmig, D., B. Balsley, K. Davis, L. R. Kuck, M. Jensen, l Bognar, T. Smith Jr., R. Vasquez Arrieta, R. Rodriguet, and l W. Birks (1998), Vertical profiling and determination of landscape fluxes of biogenic nonmethane hydrocarbons within the planetaty boundary layer in the Peruvian Amazon, 1. Geophys. Res., 103(Dl9),25,519-25,532. Hoffmann, T., l R Odum, F. Bowman, D. Collins, D. Klockow, R C. Flagan, and J. H. Seinfeld (1997), Formation of organic aerosols from the oxidation of biogenic hydrocarbons, 1. Atmos. Chem., 26, 189-222. Hoffmann, T., R. Bandur, U. Marggraf, and M. Linscheid (1998), Molecular composition of organic aerosols formed in the a,pinene/03 reaction: Implications for new particle formation processes, 1. Geophys. Res., 103, 25,569-25,578. Holzinger, R., L. Sandoval-Soto, S. Rottenberger, P. l Crutzen, and l Kesselmeier (2000), Emissions of volatile organic compounds from Quercus ilex L. measured by proton transfer reaction mass spectrometiy under different environmental conditions, 1. Geophys. Res., /05(Dl6), 20,573-20,579. Houghton, R A. (2003), Why are estimates ofthe terrestrial carbon balance so different?, Global Change BioI., 9(4), 500-509. Houghton, R. A., M. Gloor, l Lloyd, and C. P.otter (2009), The regional carbon budget, Geophys. Monogr. Ser., doi: 10.10291 2008GM000718, this volume. Iinuma, Y., O. Boge, T. Gnauk, and H. Herrmann (2004), Aerosolchamber study of the alpha-pinene/O-3 reaction: Influence of patiicle acidity on aerosol yields and products, Atmos. Environ., 38,761-773. Jacob, D. l, and S. C. Wofsy (1988), Photochemistty ofbiogenic emissions over the Amazon forest, 1. Geophys. Res., 93, 1477-1486. Jacob, D. J., B. D. Field, E. M. Jin, 1. Bey, Q. Li, l A. Logan, R. M. Yantosca, and H. B. Singh (2002), Atmospheric budget ofacetone, 1. Geophys. Res., 107(DlO), 4100, doi:l0.1029/2001JD000694. Jaenicke, R. (2005), Abundance of cellular material and proteins in the atmosphere, Science, 308, 73. Jang, M., N. M. Czoschke, and A. L. Northcross (2004), Atmospheric organic aerosol production by heterogeneous acidcatalyzed reactions, Chemphyschem, 5, 1647-1661.
Jang, M. S., N. M. Czoschke, S. Lee, and R. M. Kamens (2002), Heterogeneous atmospheric aerosol production by acidcatalyzed particle-phase reactions, Science, 298, 814-817. Jaoui, M., and R. M. Kamens (2003a), Gas and particulate products distribution from the photooxidation of alpha-humulene in the presence of NOx, natural atmospheric air and sunlight, J. Atmos. Chem., 46, 29-54. Jaoui, M., and R. M. Kamens (2003b), Gaseous and particulate oxidation products analysis of a mixture of alpha-pinene plus beta-pinene/O-3/air in the absence of light and alpha-pinene plus beta-pineneINO)air in the presence ofnatural sunlight, 1. Atmos. Chem., 44, 259-297. Jaoui, M., and R. M. Kamens (2003c), Mass balance of gaseous and patiiculate products from beta-pinene/O-3/air in the absence of light and beta-pineneINO.iair in the presence of natural sunlight, J. Atmos. Chem., 45, 101-141. Jenkin, M. E., S. M. Saunders, R. G. Derwent, and M. l Pilling (2002), Development of a reduced speciated VOC degradation mechanism for use in ozone models, Atmos. Environ., 36, 4725-4734. Junk, W. l (1989), Flood tolerance and tree distribution in central Amazonian floodplains, in Tropical Forests: Botanical Dynamics, Speciation and Diversity, edited by L. B. Holm-Nielsen, 1. C. Nielsen, and H. Balslev, prJ. 47-64, Academic, San Diego, Calif. Junk, W. l (1993), Wetlands of tropical South America, in Wetlands of the World, edited by D. Whigham, S. Hejny, and D. Dykyjova, pp. 679-739, Springer, Boston. Junk, W. J. (1997), The Central Amazon Floodplain: Ecology of a Pulsing System, Ecol. Stud., vol. 126, 525 pp., Springer, New York. Junk, W. l, P. B. Bayley, and R. E. Sparks (1989), The flood pulse concept in river-floodplain systems, in Proceedings ofthe International Large River Symposium (LARS), Can. Spec. Publ. Fish. Aquat. Sci., edited by D. P. Dodge, /06, 110-127. Junk, W. l, l l Ohly, M. T. F. Piedade, and M. G. M. Soares (2000), The Central Amazon Floodplain: Actual Use and Options for a Sustainable Management, 584 pp., Backhuys, Leiden, Netherlands. Kalberer, M., et al. (2004), Identification of polymers as major components of atmospheric organic aerosols, Science, 303, 1659-1662. Karl, T.,' and A. Guenther (2004), Atmospheric variability of biogenic VOCs in the surface layer measured by proton-transferreaction mass spectrometly, Int. J. Mass Spectrom., 239, 77-86. Karl, T., A. Guenther, R. J. Yokelson, l Greenberg, M. Potosnak, D. R Blake, and P. Artaxo (2007), The ti'opical forest and fire emissions experiment: Emission, chemistry, and transport of biogenic volatile organic compounds in the lower atmosphere over Amazonia, 1. Geophys. Res., 112, D18302 doi:10.10291 2007JD008539. Kavouras, 1. G., N. Mihalopoulos, and E. G. Stephanou (1998), Fonnation of atmospheric particles from organic acids produced by forests, Nature, 395, 683-686. Kavouras, 1. G., N. Mihalopoulos, and E. G. Stephanou (1999), Secondmy organic aerosol formation vs primaty organic aerosol
emission: In situ evidence for the chemical coupling between monoterpene acidic photooxidation products and new particle formation over for~ists, Environ. Sci. Technol., 33,1028-1037. Kawamura, K, and)iO. Yasui (2005), Diurnal changes in the distribution of dic¥,boxylic acids, ketocarboxylic acids and dicarbonyls in the·tlrban Tokyo atmosphere, Atmos. Environ., 39, 1945-1960. Kawamura, K, Y..Imai, and L. A. Ban'ie (2005), Photochemical production and loss of organic acids in high Arctic aerosols during long-range transport and polar sunrise ozone depletion events, Atmos. Environ., 39, 599--614. Keene, W. C., l N. Galloway, and l D. Holden Jr. (1983), Measurement of weak organic acidity in precipitation fi'om remote areas of the world, 1. Geophys. Res., 88(C9), 5122-5130. Kesselmeier, l (2001), Exchange of shOli-chain oxygenated volatile organic compounds (VOCs) between plants and the atmosphere: A compilation of field and laboratOly studies, 1. Atmos. Chem., 39(3), 219-233. Kesselmeier, l, and M. Staudt (1999), Biogenic volatile organic compounds (VOC): An overview on emission, physiology and ecology, 1. Annos. Chem., 33, 23-88. Kesselmeier, l, et al. (2000), Atmospheric volatile organic compounds (VOC) at a remote tropical forest site in central Amazonia, Atmos. Environ., 34, 4063-4072. Kesselmeier, l, et al. (2002a), Volatile organic compound emissions in relation to plant carbon fixation and the tel1'estrial carbon budget, Global Biogeochem. Cycles, 16(4), 1126, doi:l0.10291 200IGBOOI813. Kesselmeier, l, et al. (2002b), Concentrations and species composition of atmospheric volatile organic compounds (YOCs) as observed during the wet and dry season in Rondonia (Amazonia), 1. Geophys. Res., 107(D20), 8053, doi:IO.1029/2000JD000267. Klinger, L. F., Q.-l Li, A. B. Guenther, l P. Greenberg, B. Baker, and l-H. Bai (2002), Assessment of volatile organic compound emissions from ecosystems of China, 1. Geophys. Res., 107(ml), 4603, doi:l0.1029/2001JDOOI076. Koch, S., R Winterhalter, E. Uherek, A. Kolloff, P. Neeb, and G. K. Moortgat (2000), Formation of new particles in the gasphase ozonolysis of monotel'penes, Atmos. Environ., 34, 40314042. Konig, G., M. Brunda, H. Puxbaum, C. N. Hewitt, S. C. Duckham, and l Rudolph (1995), Relative contribution of oxygenated hydrocarbons to the total biogenic VOC emissions of selected mid-European agricultural and natural plant species, Atmos. Environ., 29,861-874. Kreuzwieser, l, U. Scheerer, and H. Rennenberg (1999), Metabolic origin of acetaldehyde emitted by poplar (Populus tremula x P. alba) trees, J. Exp. Bot., 50(335), 757-765. Kubitzki, K (1989), The ecogeographical differentiation of Amazonian inundation forests, Plant Syst. Evo!., 163, 285-304. Kuhlbusch, T. A. l (1998), Black carbon and the carbon cycle, Science,280, 1903-1904. Kuhn, U., S. Rottenberger, T. Biesenthal, A. Wolf, G. Schebeske, P. Ciccioli, E. Brancaleoni, M. Frattoni, T. M. Tavares, and l Kesselmeier (2002a), Isoprene and monoterpene emissions of Amazonian tree species during the wet season: Direct and in-
203
direct investigations on conti'olling environmental functions, 1. Geophys. Res. ,/07(mO), 8071, doi: I 0.1029/200 IJD000978. Kuhn, U., S. Rottenberger, C. Ammann, A. Wolf, G. Schebeske, T. Biesenthal, S. T. Oliva, T. M. Tavares, and l Kesselmeier (2002b), Exchange of short-chain monocarboxylic acids by vegetation at a remote tropical forest site in Amazonia, J. Geophys. Res., 107(D20), 8069, doi:10.1029/2000JD000303. Kuhn, U., S. Rottenberger, T. Biesenthal, A. Wolf, G. Schebeske, P. Ciccioli, E. Brancaleoni, M. Frattoni, T. M. Tavares, and l Kesselmeier (2004a), Seasonal differences in isoprene and light-dependent monoterpene emission by Amazonian tree species, Global Change Bio!., 10, 663--682, doi:10.11111j.15298817.2003.0077I.x. Kuhn, U., S. Rottenberger, T. Biesenthal, A. Wolf, G. Schebeske, P. Ciccioli, and l Kesselmeier (2004b), Strong correlation between isoprene emission and gross photosynthetic capacity during leaf phenology ofthe tropical tree species Hymenaea courbaril with fundamental changes in VOC emission composition during early leaf development, Plant Cell Environ., 27(12), 1469-1485, doi:10.llll1j.1365-3040.2004.01252.x. Kuhn, u., et al. (2007), Isoprene and monoterpene fluxes from Central Amazonian rainforest inferred from tower-based and airborne measurements, and implications on the atmospheric chemistlY and the local carbon budget, Annos. Chem. Phys., 7,2855-2879. Lack, D. A., X. X. Tie, N. D. Bofinger, A. N. Wiegand, and S. Madronich (2004), Seasonal variability ofsecondaly organic aerosol: A global m01eling study, 1. Geophys. Res., 109, D03203, doi: 10.102912003JD003418. Larsen, B. R, D. Di Bella, M. Glasius, R. Winterhalter, N. R. Jensen, and l Hjorth (2091), Gas-phase OH oxidation of monoterpenes: Gaseous and particulate products, 1. Annos. Chem., 38, 231-276. Lelieveld, l, F. J. Dentener, W. Peters, and M. C. Krol (2004), On the role of hydroxyl radicals in the self-cleansing capacity of the troposphere, Atmos. Chem. Phys., 4, 2337-2344. Lelieveld, J., et al. (2008), Atmospheric oxidation capacity sustained by a tropical forest, Nature, 452, 737-740. Limbeck, A., and H. Puxbaum (1999), Organic acids in continental background aerosols, Atmos. Environ., 33, 1847-1852. Limbeck, A., M. Kulmala and H. Puxbaum (2003), Secondaty organic aerosol formation in the atmosphere via heterogeneous reaction of gaseous isoprene on acidic particles, Geophys. Res. Lett., 30(19),1996, doi:IO.1029/2003GLOI7738. Lindinger, W., A. Hansel, and A. Jordan (1998), On-line monitoring of volatile organic compounds at pptv levels by means of proton transfer mass spectrometry (PTR-MS) medical applications, food control and environmental research [review], Int. 1. Mass Spectrom. Ion Process., 173,191-241. Lloyd, l, et al. (2007), An airborne regional carbon balance for Central Amazonia, Biogeosciences, 4, 759-768. Longo, K. M., S. R. Freitas, M. O. Andreae, R. Yokelson, and P. Artaxo (2009), Biomass burning in Amazonia: Emissions, long-range transport of smoke and its regional and remote impacts, Geophys. Monogr. Ser., doi:IO.1029/2008GM000847, this volume. MacDonald, R. C., and T. W. Kimmerer (1993), Metabolism of ti'anspired ethanol by eastem cottonwood (Populus deltoides Bartr.), Plant Physiol., /02(1), 173-179.
204
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
MacDonald, R. C., T. W. Kimmerer, and M. Razzaghi (1989), Aerobic ethanol-production by leaves-Evidence for air-pOllution stress in trees of the Ohio River Valley, USA, Environ. Pollut., 62(4),$37-351. Mochida, M., A. Kawabata, K. Kawamura, H. Hatsushika, and K. Yamazaki (2003), Seasonal variation and origins of dicarboxylic acids in the marine atmosphere over the western North Pacific, 1. Geophys. Res., 108(D6), 4193, doi: 1O.1029/2002JD002355. Montzka,' S. A, M. Trainer, W. M. Angewine, and F. C. Fehsenfeld (1995), Measurements of 3-methyl furan, methyl vinyl ketone, and methacrolein at a rural forested site in the southeastern United States, 1. Geophys. Res., 100(D6), 11 ,39311,401. Moukhtar, S., B. Bessagnet, L. Rouil, and V. Simon (2005), Monoterpene emissions from beech (Fagus sylvatica) in a French forest and impact on secondary pollutants formation at regional scale, Atmos. Environ., 39, 3535-3547. Myneni, R. B., et a!. (2007), Large seasonal swings in leaf area of Amazon rainforests, Proc. Nat!. Acad. Sci. U. S. A., 104(12), 4820-4823. Naik, V., C. Delire, and D. l Wuebbles (2004), Sensitivity of global biogenic isoprenoid emissions to climate variability and atmospheric CO 2 , 1. Geophys. Res. 109, D06301, doi:IO.l029/ 2003JD004236. Neeb, P., F. Sauer, O. Horie, and G. K. Moortgat (1997a), Formation of hydroxymethyl hydroperoxide and formic acid in alkene ozonolysis in the presence of water vapour, Atmos. Environ., 31, 1417-1423. Neeb, P., K. Bode, l Beck, L. Schafer, l Kesselmeier, and G. K. Moortgat (1997b), Influence of gas-phase oxidation on estimated emission rates of biogenic hydrocarbons, in Proceedings of the 7th European Symposium on Physico-Chemical Behaviour of Atmospheric Pollutants: The Oxidizing Capacity of the Troposphere, pp. 295-299, Office for Official Publications of the European Communities, Luxembourg (EUR 17482) ISBN 92-828-0158-6. O'Dowd, C. D., P. Aalto, K Hameri, M. Kulmala, and T. Hoffmann (2002), Aerosol formation-Atmospheric particles from organic vapours, Nature, 416, 497-498. Odum, l R., T. Hoffmann, F. Bowman, D. Collins, R. C. Flagan, and l H. Seinfeld (1996), Gas/pmiicle partitioning and secondary aerosol formation, Environ. Sci. Technol., 30, 2580-2585. Pankow, l F. (1994), An absorption model of the gas/aerosol partitioning involved in the formation of secondaty organic aerosol, Atmos. Environ., 28,189-193. Parolin, P., et a!. (2004), Central Amazon floodplain forests: Tree adaptation in a pulsing system, Bot. Rev., 70(3), 357-380, doi: 10.1663/0006-81 01(2004)070[0357:CAFFTA]2.0.CO;2. Piedade, M. T. F., M. Worbes, and W. l Junk (2001), Geoecological controls on elemental fluxes in communities of higher plants in Amazonian floodplains, in The Biogeochemist!), of the Amazon Basin, edited by M. E. McClain, R. L. Victoria, and l E. Richey, pp. 209-234, Oxford Univ. Press, New York. Pires, l M. (1973), Tipos de vegetac;ao da Amazonia, Publ. Av. Mus. Pal'. Emil Goeldi, 20, 179-202.
Pires, J. M., and G. T. Prance (1985), The vegetation types of the Brazilian Amazon, in Amazon: Key Environments, edited by G. T. Prance and T. E. Lovejoy, pp. 109-145, Elsevier, London. Rasmussen, R. A., and M. A K. Khalil (1988), Isoprene over the Amazon Basin, 1. Geophys. Res., 93(D2), 1417-1421. Rinne, H. J. I., A B. Guenther, J. P. Greenberg, and P. C. Harley (2002), Isoprene and monoterpene fluxes measured above Amazonian rainforest and their dependence on light and temperature, Atmos. Environ., 36(14),2421-2426. Rogge, W. F., L. M. Hildemann, M. A Mazurek, G. R. Cass, and B. R. T. Simoneit (1993), Sources of fine organic aerosol. 4. Particulate abrasion products from leaf surfaces of urban plants, Environ. Sci. Technol., 27, 2700-2711. Rottenberger, S. (2003), Exchange of oxygenated volatile organic compounds between Amazonian and European vegetation and the atmosphere, Ph.D. thesis, Dept. Biology, University Mainz. Rottenberger, S., U. Kuhn, A. Wolf, G. Schebeske, S. T. Oliva, T. M. Tavares, and l Kesselmeier (2004), Exchange of short-chain aldehydes between Amazonian vegetation and the atmosphere at a remote forest site in Brazil, Ecol. Appl., 14(4), Suppl., 247262. Rottenberger, S., U. Kuhn, A Wolf, G. Schebeske, S. T. Oliva, T. M. Tavares, and l Kesseln'leier (2005), Formaldehyde and acetaldehyde exchange during leaf development of the Amazonian deciduous tree species Hymenaea courbaril, Atmos. Environ., 39, 2275-2279. Rottenberger, S., B. Kleiss, U. Kulm, A Wolf, M. T. F. Piedade, W. Junk, and l Kesselmeier (2008), The effect of flooding on the exchange of the volatile C2-compounds ethanol, acetaldehyde and acetic acid between leaves of Amazonian floodplain tree species and the atmosphere, Biogeosciences, 5, 10851100. Sanderson, M. G., C. D. Jones, W. l Collins, C. E. Johnson, and R. G. Detwent (2003), Effect of climate change on isoprene emissions and surface ozone levels, Geophys. Res. Lett., 30(18), 1936, doi: 10.1029/2003GLOI7642. Sanhueza, E., M. Santana, D. Trapp, C. de Serves, L. Figueroa, R. Romero, A Rondon, and L. Donoso (1996), Field measurement evidence for an atmospheric chemical source of formic and acetic acids in the tropic, Geophys. Res. Lett., 23(9), 1045-1048. Sattler, B.', H. Puxbaum, and R. Psenner (200 I), Bacterial growth in supercooled cloud droplets, Geophys. Res. Lett., 28, 239-242. Schauer, l J., W. F. Rogge, L. M. Hildemann, M. A Mazurek, and G. R. Cass (1996), Source apportionment of airborne particulate matter using organic compounds as tracers, Atmos. Environ., 30, 3837-3855. Sch6ngart, l, M. T. F. Piedade, S. Ludwigshausen, V. Horna, and M. Worbes (2002), Phenology and stem-growth periodicity of tree species in Amazonian floodplain forests, 1. Trop. Ecol., 18, 581-597. SchOngart, l, W. J. Junk, M. T. F. Piedade, l M. Ayres, A Huttennann, and M. Worbes (2004), Teleconnection between h'ee growth in the Amazonian floodplains and the El Nino-Southern Oscillation effect, Global Change BioI., 10(5),683-692.
KESSELMEIER ET AL. Schulze, B., C. Kost, G. I. Arimura, and W. Boland (2006), Signalrezeption, biosyntheseund6kologie, Duftstoffe: Die sprache del' pflanzen, Chem. Unljerer Zeit, 40, 366-377. Seinfeld, l H., andl N. Pandis (1998), Atmospheric Chemist!)! and Physics, Johj!Wiley, Hoboken, N. l, ISBN 0-471-17815-2. Serc;a, D., et al. {WOI), EXPRESSO flux measurements at upland and lowland Congo tropical forest site, Tellus, Ser. B, 53, 220-34. Shim, C., Y. Wang, Y. Choi, P. I. Palmer, D. S. Abbot, and K. Chance (2005),' Constraining global isoprene emissions with Global Ozone Monitoring Experiment (GOME) formaldehyde column measurements, 1. Geophys. Res., 110, D24301, doi: 10.1 029/2004JD005629. Simoneit, B. R. T. (1999), A review of biomarker compounds as source indicators and tracers for air pollution, Environ. Sci. Pollut. Res., 6, 159-169. Simoneit, B. R. T. (2002), Biomass burning-A review of organic tracers for smoke from incomplete combustion, Appl. Geochem., 17, 129-162. Simoneit, B. R. T., R. E. Cox, and L. l Standley (1988), Organic matter of the troposphere-IV: Lipids in Hatmattan aerosols of Nigeria, Atmos. Environ., 22, 983-1004. Simoneit, B. R. T., V. O. Elias, M. Kobayashi, K Kawamura, A I. Rushdi, P. M. Medeiros, W. F. Rogge, andB. M. Didyk (2004a), Sugars-Dominant water-soluble organic compounds in soils and characterization as tracers in atmospheric particulate matter, Environ. Sci. Technol., 38, 5939-5949. Simoneit, B. R. T., M. Kobayashi, M. Mochida, K Kawamura, M. Lee, H.-J. Lim, B. l Turpin, and Y. Komazaki (2004b), Composition and major sources of organic compounds of aerosol particulate matter sampled during the ACE-Asia campaign, 1. Geophys. Res., 109, DI9SIO, doi:10.1029/2004JD004598. Singh, H. B., M. Kanakidou, P. J. Cmtzen, and D. J. Jacob (1995), High-concenh'ations and photochemical fate of oxygenated hydrocarbons in the global troposphere, Nature, 378(6552), 50-54. Singh, H. B., Y. Chen, A. Staudt, D. Jacob, D. Blake, B. Heikes and J. Snow (2001), Evidence from the pacific troposphere for large global sources of oxygenated organic compounds, Nature, 410(6832), 1078-1081. Starn, T. K, P. B. Shepson, S. B. Bertman, l S. White, B. G. Splawn, D. D. Riemer, R. G. Zika, and K Olszyna (1998), Observations of isoprene chemistry and its role in ozone production at a semitural site during the 1995 Southern Oxidants Shidy, 1. Geophys. Res., 103(Dl7), 22,425-22,435. Suntharalingam, P., l T. Randerson, N. Krakauer, J. A Logan, and D. J. Jacob (2005), Influence of reduced carbon emissions and oxidation on the distribution of atmospheric CO 2 : Implications for inversion analyses, Global Biogeochem. Cycles, 19, GB4003, doi: 10.1 029/2005GB002466. Takekawa, H., H. Minoura, and S. Yamazaki (2003), Temperature dependence of secondmy organic aerosol fonnation by photo-oxidation of hydrocarbons, Atmos. Environ., 37, 3413-' 3424. Talbot, R. W., M. O. Andreae, H. Berresheim, D. J. Jacob, and K. M. Beecher (1990), Sources and sinks of fonnic, acetic, and pymvic acids over central Amazonia: 2, Wet season, 1. Geophys. Res., 95,16,799-16,811.
205
Thompson, A M. (1992), The oxidizing capacity of the Earths atmosphere-Probable past and fuhire changes, Science, 256(5060), 1157-1165. Trostdorf, C. R., L. V. Gatti, A Yamazaki, M. l Potosnak, A Guenther, W. C. Matiins, and l W. Munger (2004), Seasonal cycles of isoprene concentrations in the Amazonian rainforest, Atmos. Chell/, Phys. Discuss., 4, 1291-1310. Tsigaridis, K, and M. Kanakidou (2003), Global modelling of secondary organic aerosol in the troposphere: A sensitivity analysis, Atmos. Chem. Phys., 3, 1849-1869. Tuazon, E. C., and R. Atkinson (1990), A product study ofthe gasphase reaction of isoprene with the OH radical in the presence of NO,H Jnt. 1. Chem. Kinet., 22, 1221-1236. von Kuhlmann, R., M. G. Lawrence, P. l Cmtzen, and P. l Rasch (2003), A model for shidies ofh'opospheric ozone and nonmethane hydrocarbons: Model evaluation of ozone-related species, J. Geophys. Res., 108(D23), 4729, doi:IO.1029/2002JD003348. Went, F. W. (1955), Air pollution, Sci. Am., 192, 63-72. Went, F. W. (1960a), Blue hazes in the atmosphere, Nature, 187, 641-643. Went, F. W. (1960b), Organic matter in the ahnosphere and its possible relation to petroleum fonnation, Proc. Natl. Acad. Sci. U. S. A., 46, 212-221. Williams, l (2004), Organic trace gases in the atmosphere: An overview, Environ. Chem., 1,125-136. Wilske, 8., K. F. Cao, G. Schebeske, J. W. Chen, A Wang, and J. Kesselmeier (2007), Isoprenoid emissions of h'ees in a tropical rainforest in Xishhangbanna, SW-China, Atmos. Enviroll., 41, 3748-3757. Winterhalter, R., R. }'an Dingenen, B. R. Larsen, N. R. Jensen and l Hjorth (2003), LC-MS analysis of aerosol particles frOl~ the oxidation of a-pinene by ozone and OH-radicals, Atmos. Chem. Phys. Discuss., 3, 1-39. Wittmaml, F., l Sch6ngart, J. C. Montero, T. Motzer, W. J. Junk, M. T. F. Piedade, H. L. Queiroz, and M. Worbes (2006), Tree species composition and diversity gradients in white-water forests across the Amazon Basin, 1. Biogeogr., 33, 1334-1347. Worbes, M. (1997), The forest ecosystem ofthe floodplains, in The Central Amazon Foodplain: Ecology ofa Pulsing System, Ecol. Stud., vol. 126, edited by W. l Junk, pp. 223-266, Springer, New York. Worbes, M. (1999), Annual growth rings, rainfall-dependent growth and long-term growth patterns of h'opical trees from the Caparo Forest Reserve in Venezuela, 1. Ecol., 87(3), 391403. Worbes, M., and W. l Junk (1999), How old are h'opical trees? The persistence of a myth, JA WA 1.,20(3),255-260. Yokouchi, Y., and Y. Ambe (1985), Aerosols formed from the chemical reaction of monotel'penes and ozone, Atmos. Environ., 19, 1271-1276. Yu, l Z., D. R. Cocker, R. l Griffin, R. C. Flagan, and l H. Seinfeld (1999), Gas-phase ozone oxidation of monoterpenes: Gaseous and particulate products, 1. Atmos. Chem., 34, 207-258. Zhimg, S.-H., M. Shaw, l H. Seinfeld, and R. C. Flagan (1992), Photochemical aerosol formation fi'om a-pinene- and l3-pinene, 1. Geophys. Res. 97,20,717-20,729.
206
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
Zimmerman, P. R., J. P. Greenberg, and C. E. Westberg (1988), Measurements of atmospheric hydrocarbons and biogenic emission fluxes in the Amazon boundary layer, J. Geophys. Res., 93(D2), 14J)7-1416.
A. Guenther, National Center for Atmospheric Research, Boulder, CO 8,0307-3000, USA. ([email protected])
T. Hoffmalill and J. Warnke, Institute of Inorganic and Analytical Chemistry, Johannes Gutenberg University of Mainz, D55128 Mainz, Germany. ([email protected]; warnke@ uni-mainz.de) 1. [(esselmeier, Max Planck Institute for Chemistry, D-55020 Mainz, Germany. [email protected]) M. T. Piedade, Instituto Nacional de Pesquisas da Amazonia, CEP 69060-001 Manaus, AM, Brazil. ([email protected])
Biomass Burning in Amazonia: Emissions, Long-Range Transport of Smoke and Its Regional and Renl0te I1npacts K. M. Longo!, S. R. Freitas 2 , M. O. Andreae 3 , R. Yokelson4 , and P. Artaxo 5 Every year, biomass burning in Amazonia continues to release large amounts of trace gases and aerosol patiicles into the atmosphere. The consequent change from low to velY high atmospheric concentrations of oxidants and aerosols therefore affects the radiative, cloud physical, and chemical properties of the atmosphere over Amazonia. This represents a dramatic peliurbation to the regional climate, ecology, water cycle, and human activities. Given the magnitude of burning in Amazonia and the efficiency of the atmospheric transpOli processes of fire emissions, these perturbations can affect the climate system even on a global scale. This chapter summarizes the knowledge acquired in the ambit of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia program about vegetation fire as a driving force of atmospheric disturbances over Amazonia. We describe the different fire behaviors for the region and present an updated review of emission and combustion factors for Amazonia. We discuss some of the available biomassburning emission inventories for the Amazonian region, dis¢ussing their assets and limitations. We further discuss atmospheric transport processes that are the main drivers ofthe dispersion offire emissions, introduce the most relevant concepts for numerical modeling of smoke transport, and show the general pattern of smoke transpOli over the South American continent. Finally, we present the current status of the understanding of local and remote impacts of smoke trace gases and aerosol particles, discussing the oxidizing power of the Amazonian atmosphere, as well as the radiation and heat budgets and consequences on cloud properties and distribution.
lCenter for Space and Atmospheric Sciences, National Institute for Space Research, Sao Jose dos Campos, Brazil. 2Center for Weather Forecast and Climate Studies, National Institute for Space Research, Cachoeira Paulista, Brazil. 3Max Planck Institute, Mainz, Germany. 4Department of Chemistly, University of Montana, Missoula, Montana, USA. 5Institute of Physics, University of Sao Paulo, Sao P1julo, Brazil. . Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1 029/2008GM00084 7
1. INTRODUCTION: BIOMASS BURNING IN AMAZONIA The composition of the atmosphere is controlled by several natural and anthropogenic processes, and emissions from biomass buming are one of its strongest drivers in the Southem Hemisphere [Crufzen and Andreae, 1990]. Agricultural residues have been bumt for millennia, and the reduction in forest area in North America and Europe over the last centuries has evidently contributed to the changes in atmo'spheric composition. More recently, during the last four to five decades, the rapid and intensive land use change in the tropics has led to more attention being paid to this issue.
207
206
NATURAL VOLATILE ORGANIC COMPOUND EMISSIONS FROM PLANTS
Zimmerman, P. R., J. P. Greenberg, and C. E. Westberg (1988), Measurements of atmospheric hydrocarbons and biogenic emission fluxes in the Amazon boundaty layer, 1. Geophys. Res., 93(D2), 1407-1416.
A. Guenther, National Center for Atmospheric Research, Boulder, CO 89307-3000, USA. ([email protected])
T. Hoffmmill and 1. Warnke, Institute of Inorganic and Analytical Chemistry, Johannes Gutenberg University of Mainz, D55128 Mainz, Germany. ([email protected]; warnke@ uni-mainz.de) J. Kesselmeier, Max Planck Institute for Chemistty, D-55020 Mainz, Germany. ([email protected]) M. T. Piedade, Instituto Nacional de Pesquisas da Amazonia, CEP 69060-001 Manaus, AM, Brazil. ([email protected])
Biomass Burning in Amazonia: Emissions, Long-Range Transport of Snloke and Its Regional and Rel110te Itllpacts K. M. Longo l , S. R. Freitas2 , M. O. Andreae 3 , R. Yokelson4 , and P. Artax0 5 EvelY year, biomass burning in Amazonia continues to release large amounts of trace gases and aerosol particles into the atmosphere. The consequent change from low to velY high atmospheric concentrations of oxidants and aerosols therefore affects the radiative, cloud physical, and chemical properties of the atmosphere over Amazonia. This represents a dramatic perturbation to the regional climate, ecology, water cycle, and human activities. Given the magnitude of burning in Amazonia and the efficiency of the atmospheric transport processes of fire emissions, these pelturbations can affect the climate system even on a global scale. This chapter summarizes the knowledge acquired in the ambit of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia program about vegetation fire as a driving force of atmospheric disturbances over Amazonia. We describe the different fire behaviors for the region and present an updatbd review of emission and combustion factors for Amazonia. We discuss some of the available biomassburning emission inventories for the Amazonian region, disdussing their assets and limitations. We further discuss atmospheric transport processes that are the main drivers ofthe dispersion offire emissions, introduce the most relevant concepts for numerical modeling of smoke transpOli, and show the general pattern of smoke transport over the South American continent. Finally, we present the current status of the understanding of local and remote impacts of smoke trace gases and aerosol particles, discussing the oxidizing power of the Amazonian atmosphere, as well as the radiation and heat budgets and consequences on cloud properties and distribution.
ICenter for Space and Atmospheric Sciences, National Institute for Space Research, Sao Jose dos Campos, Brazil. 2Center for Weather Forecast and Climate Studies, National Institute for Space Research, Cachoeira Paulista, Brazil. 3Max Planck Institute, Mainz, Gelmany. 4Department of Chemistty, University of Montana, Missoula, Montatia, USA. 5Institute of Physics, University of Sao Paulo, Sao Paulo, Brazil. . Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000847
1. INTRODUCTION: BIOMASS BURNING IN AMAZONIA
The composition of the atmosphere is controlled by several natural and anthropogenic processes, and emissions from biomass burning are one of its strongest drivers in the Southern Hemisphere [Crufzen and Andreae, 1990]. Agricultural residues have been burnt for millennia, and the reduction in forest area in North America and Europe over the last centuries has evidently contributed to the changes in atmospheric composition. More recently, during the last four to five decades, the rapid and intensive land use change in the tropics has led to more attention being paid to this issue.
207
208
BIOMASS BURNING IN AMAZONIA
It is important to emphasize that biomass burning as a major
atmospheric driver is not only restricted to the tropics. The high concentration of aerosol particles and trace gases observed in tlre Amazonian and Central Brazilian atmosphere during the dry season is associated with intense anthropogenic biomass-burning activity. Ozone, carbon monoxide, nitrogen oxides, and aerosol pariicle concentrations over South America and the sUlTounding ocean areas are regulated by biomass-burning emissions from savannah and forest fires. About 9200 Teragram (dry weight) is burned annually [Andreae and Merlet, 2001; Bergamaschi et al., 2000], contributing significantly to the atmospheric burden of pollutants. In South America, during the biomass-burning season, a regional smoke plume covering an area of about 4 to 5 million km2 has been frequently observed through remote sensing of South America [Prins et al., 1998). Inhalable aerosol particles with concentration as high as 400 Ilg m-3 have been measured close to the surface, and the colurnnintegrated aerosol optical thickness reaches 4.0 to 5.0 (440 nm) over large areas of central Brazil [Artaxo et al., 1998). Ozone concentrations in excess of 100 ppb are fi'equently observed thousands ofkilometers away from forest fires, and the ozone phytotoxicity certainly affects the unburned forest. On a regional and global scale, the persistent and heavy smoke layer over an extensive tropical region may alter the radiation balance and the hydrological cycling. Carbon uptake by the forest, expressed by the net ecosystem exchange (NEE) is heavily affected by the aerosol layer over the forest; where, at low aerosol levels, an increase of 30% to 40% in NEE was observed for aerosol optical depth (AOD) up to 1.2 at 550 nm [Oliveira et al., 2007). This effect happens because the aerosol particles in the atmosphere increase the diffuse solar radiation, and the forest canopy geometry leads to enhanced photosynthesis. But when the AOD exceeds about 1.5, the effect of the reduction in total solar flux staris to predominate, and NEE stari to decrease; for values of AOD near 4 or 5, it shuts down almost yompletely. This effect of changing the ratio of diffuse to direct radiation has strong implication for the carbon balance over tropical forests [see Artaxo et al., this volume, and references therein]. A second strong effect of aerosol particles emitted through biomass burning is the resultant changes in cloud microphysics, development, and stmcture. Clouds are a critical ingredient of the radiation balance and the hydrological cycle. The presence of biomass-burning pariicles in the atmosphere also modifies the solar radiative balance by changing the cloud microphysics. These particles act as cloud condensation and ice nuclei, promoting changes in the cloud drop spectmm and, consequently, altering the cloud albedo and precipitation [Rosenfeld et al., 2006]. This suggests that the biomass-burning effects may extrapolate the regional scale
LONGO ET AL.
and influence the pattern of planetary redistribution of energy from the tropics to medium and high latitudes via convective transpori processes. Changes in cloud cover due to the presence of large amounts of black carbon pariicles are well documented in the work of Koren et al. [2004, 2008] and Kat!finan and Koren [2006]. See also Artaxo et al. [this volume] for an overview of this issue. Emissions fi'om the combustion of any type of fuel depend directly on the chemical composition of that fuel and the combustion conditions. For biomass burning, most data are available for wood combustion. Different tree species develop markedly different woody constituents during growth, and typically, all wood consists of various forms of lignin, celluloses, and fillers. Emission factors (EFs) are important because they are used in regional and global models to study the influence of the biomass-burning emissions on regional and global climate. Deforestation in Brazilian Amazonia has been studied using remote sensing techniques [Camara et al., 2005; Morton et al., 2005]. The average deforestation rate for the 1990s was 17,000 km2 per year, increasing to approximately 25,000 km2 in 2002 and 2003 [Instituto Nacional de Pesquisas Espaciais (INPE), 2008], going down to 10,000 km2 in 2007 [see Schroeder et al., this volume). Up to 2005, an estimated area of 16% of the total Brazilian Amazonia, i.e., an area of 5.8 million km2 was deforested [INPE, 2008]. Deforestation mainly occurs in the southern and eastern parts of Amazonia, while the central areas that are less accessible are relatively well preserved. Deforestation affects the ecosystem in several ways: First, there is a change in the energy and water balance when forest is replaced by pasture, and this change has the potential to alter the atmospheric water content and precipitation patterns [Silva Dias et al., 2002]. Second, a large amount of aerosol particles is released into the atmosphere, as forests are cut and burned in the course of managing pastures and fields, leading to profound changes in the atmospheric composition [Artaxo et al., 1998, 2002] and surface radiation balance [Schafer et al., 2002a, 2002b; Procopio et al., 2003, 2004]. Amazonia deforestation and biomass burning can tr'igger a positive feedback cycle of increased fire disturbance and local drought conditions, amplifying droughts linked to both anthropogenic global climate change and natural climate variability [Nobre et al., 1991; Marengo et al., 2008). This chapter presents an overview of the knowledge acquired in the ambit ofthe Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) program about emissions to the atmosphere fi'om vegetation fires in Amazonia. The atmospheric transport of smoke and their impacts on the atmospheric composition, weather, and climate at local, regional, and global scales will be addressed.
2. BIOMASS-BURNING EMISSION ESTIMATES Biomass combusti6n is a complex mix of chemical and physical processe~( A detailed description and references can be found in)'he work of Yokelson et al. [1996, 1997], which is summa'rized below. The processes involved in biomass combustion can be visualized by considering the effects of increasing temperature on fi'esh biomass. The first effect of rising temperature is the distillation of absorbed species with low boiling points, mostly water. At higher temperatures (~50o--700 K) bonds begin to break in the macromolecules that comprise biomass. This process is known as low-temperature pyrolysis, and itreleases a white smoke and small molecules with sufficient vapor pressure to enter the gas phase. Most ofthe smoke particle species and gases are oxygenated organic compounds such as methanol and acetic acid. The selective release ofoxygenated compounds leaves the remaining "parent" biomass enriched in carbon, a product known as "low-temperature char." At higher temperatures (700-900 K), the low-temperature char begins to emit aliphatic compounds (containing mostly C and H) further enriching the biomass in carbon and producing high-temperature char (high in aromati~ components). The chemisorptions of O 2 on high-temperature char is exothermic, and it provides energy for gasification reactions that convert carbon in the solid char to products such as CO and CO 2 . Intense gasification is commonly known as "glowing" combustion. A glowing front (typically 1000 K) can propagate across a fuel element pyrolyzing much or all of the biomass ahead of the front and producing a mix of all the products noted above [Bertschi et al., 2003). In the absence of flames, the pariicles and gases emitted by pyrolysis and glowing directly enter the atmosphere as pollutants. When the concentration of volatile gases and their temperature is above a threshold, they can react rapidly with oxygen to produce turbulent diffusion flames with temperatures typically near 1400 K. The flames efficiently oxidize the entrained volatile gases to species such as H 20, CO 2 , and NO x . Incompletely oxidized species such as CO are also generated in comparatively small amounts. Black smoke that is high in elemental carbon is formed by condensation just above the flames. The flames are also important as a heat source to drive furiher pyrolysis of fresh biomass, which (along with glowing) generates more volatiles to feed continued flaming. In practice, fires are usually ignited by applying sllfficient heat to initiate flaming/glowing at a point. A mixture of flaming and glowing then propagates across the fuel bed and pyrolyzes much of the available fuel. During this time, most (but not all) of the pyrolysis and glowing products are oxidized by entrainment into the flames. Once the flames
209
have passed across the whole fuel bed, the rate of volatile production and flaming begins to drop, along with the concentrations of most of the emitted species. At this point the probability of flame oxidation of the emitted volatiles also decreases, and the smoke begins to increasingly reflect the products of smoldering combustion. The flaming and mixed flaming/smoldering phases of the fire normally account for 50-95% of the total fuel consumption. A final smolderingonly phase continues as long as 5-10% of the heat generated by the fire is transferred to fresh fuel. The amount of fuel consumption by the smoldering phase is heavily dependent on fuel geometry with closely packed fuels being consumed much more efficiently [Bertschi et al., 2003). Ultimately, the biomass-burning emissions depend upon many controlling factors. In this section, we provide an updated review of the measured emission and combustion factors for Amazonia with an assessment of their accuracy and regional representativeness.
2. I. How Fire Behavior Affects Emission Measurements for Different Fire Types Most anthropogenic fires in the tropics usually begin with ignition along one edge or two opposing edges of the treatment area. At the start of the fire, all the emissions are from flaming combustion and entrained in the flame-induced convection columll. As the flame front propagates inward, the convection cohlmn also entrains the emissions from any smoldering combustion that continues in the area just vacated by the flames. In a homogeneous fuel bed, a steady mixture of flaming and smoldering emissions can be produced from much of the fuel. These emissions are best sampled from the air. When smoldering continues after the convection envelope is too far away to entrain the emissions, or after convection fi'om the entire site has ceased, the fire emissions are produced by what we define as residual smoldering combustion (RSC). RSC emissions must be sampled fi'om the ground. Since flaming and smoldering combustion produce smoke with different chemical composition, the fire behavior described above is an imporiant consideration for representative sampling of the emissions from different types of fires. 2.1.1. Major types of.fires that occur in Brazilian Amazonia. Next is a brief summary ofthe fire types that are important in Brazilian Amazonia and relevant features for their associated emissions. 2:1.1.1. Savanna (Cerrado)/ires. Part of the southern Amazon basin is covered by savanna [Coutinho, 1990] that is burned every 1-3 years to improve grazing. These fires rapidly
210
LONGO ET AL.
BIOMASS BURNING IN AMAZONIA
consume 5-10 tha- I of mostly grass [Coutinho, 1990; Ward et al., 1992; Kauffman et al., 1994; Andrade et al., 1999]. We expect that RSC normally accounts for <10% of fuel consumptiog [Bertschi et al., 2003]. 2.1.1.2. Primm)! deforestation fires. Evergreen tropical forest is the dominant ecosystem in the Amazon basin. Deforestation rates measured since 1978 have ranged from 11 to 29 X 103 kmZ a-I (~2 x 106 ha annually (http://www.obt. inpe.br/prodesl)) [see also ScJlroeder et al., this volume]. Deforestation fires feature much greater total aboveground biomass (TAGB) loading than savanna fires: e.g., 288, 402, 265, 349 ± 21 (n = 7), and 292 t ha-I reported by Carvalho et al. [1998,2001], Fearnside et al. [1993], Guild et al. [1998], and Ward et al. [1992]. In these studies, the percentage of the TAGB consumed by the fire was 50, 21, 29, 48 (n = 7), and 53. RSC likely accounts for less than 10% of the total fuel consumption [Christian et al., 2007]. 2.1.1.3. Pasture maintenance fires. Pasture fires are intermediate in TAGB and fuel characteristics between savanna and primaty deforestation fires because residual wood debris (RWD) fi'om the first deforestation fire on the site usually persists for many years. RepOlied TAGB ranges from 119 t ha- I (with 87% of TAGB being RWD in a 4-year-old pasture) to 53 t ha- I (47% RWD, in a 20-year-old pasture) [Barbosa and Fearnside, 1996; Guild et al., 1998; Kauffman et at., 1998]. Large-diameter RWD that bumed mostly by RSC was reported to account for 38-49% ofthe fuel consumption in the above studies. Fearnside [1990] reported that ~75% of the bumed forest was converted to pasture. Pastures are usually subjected to maintenance bums every 2-3 years for 10-20 years [Guild et al., 1998] before they are abandoned or converted to other uses. As a result, pastures occupy most of the deforested land, and pasture buming is the most common type offire in Amazonia on an area basis. For Brazilian Amazonia, the total biomass bumed in pasture fires is comparable to the total bumed in deforestation fires: ~240 Tg a-I each [Kauffman et al., 1998]. 2.1.2. Recent land use trends affecting fire emissions.
In Brazil, the above picture is now being modified by the rapid growth in large-scale, mechanized soybean and sugar cane production, especially in the state ofMato Grosso. The croplands for soy are provided both by conversion ofpastures and direct conversion of primaty or secondary forest. To enable mechanized agriculture, all the large-diameter wood must be removed, which is only practical using fire, often assisted by mechanical piling of the fuel. This could imply larger fuel loadings and larger, more intense fires. Morton et al. [2006] found that within Iviato Grosso from 2001 to 2004,
pasture was still the main use following deforestation, but that fraction was decreasing (to 66%), and direct transition to large (>25 ha) areas of cropland accounted for up to 23% of deforestation. Deforestation for cropland accounted for 28% of clearings larger than 200 ha in 2003. We speculate that the expansion of mechanized agriculture may be associated with a regional increase in the area of individual fires, the fuel consumption per unit area, and fire intensity.
2.2. Biomass-Burning Emissions Measurements Relevant to Amazonia
In 1979 and 1980, Crutzen [1995] made the first airbome measurements of CO, CH4, total nonmethane hydrocarbons, and other species emitted by Amazonian fires. A groundbased component speciated selected nonmethane organic compound (NMOC) emissions [Greenberg et al., 1984]. As part of the Atmospheric Boundary Layer Experiment (ABLE 2A) in 1985, Andreae et al. [1988] added emissions measurements for COz, CO, NO." SOz, and major particle constituents and also characterized some postemission transformations. In 1990, Ward et al. [1992] made tower-based measurements that closely related the EFs for the main trace gases and PM Z.5 to vegetation type as part of the Biomass Buming Airbome and Spacebome Brazil experiment. Blake et al. [1996] speciated additional selected hydrocarbons in slightly aged biomass burning plumes in 1992. In 1995, the most complete biomass-buming experiment up to that time was carried out in Amazonia: Smoke, Clouds, and RadiationBrazil [Katifrnan et al., 1998]. Ferek et al. [1998] reported detailed measurements of both the trace gas and patiicle species and particle optical properties. The data from all the above campaigns was synthesized in a review paper by Andreae and Merlet [2001]. They recommended EFs (grams of compound emitted per kilogram of dry fuel bumed) for the main global fire types based on data available at the time. The above work include only a small amount of data for oxygenated volatile organic compounds (OVOC), which are difficult to measure, yet critical in tropospheric chemistry [Trentmann et al., 2005], accounting for ~80% ofthe NMOC emitted by fires [Yokelson et al., 2008]. In 2000 and 2001, field and laboratory measurements of savanna fire emissions were made for the first time with instrumentation capable of quantifying both hydrocarbons and OVOC [Yokelson et al., 2003; Christian et al., 2003]. Thus, to estimate the emissions from Amazonian savanna fires, an up-to-date source is the table for savanna fires in the work of Christian et al. [2003] with Andreae and Merlet [2001] for additional species. The EF can be adjusted for RSC by referring to Bertschi et al. [2003] and Christian et al. [2007].
In the 2002 Amazonian dly season, the Smoke, Aerosol, Clouds, Rainfall, and c;limate Campaign (SMOCC) detailed the chemistlY and physics of the biomass-buming pat'ticles and their inte~Jiction with clouds [Chand et al., 2006; Rissler et al., 2006; Vestin et al., 2007; Fuzzi et al., 2007]. The number of particles emitted per unit amount of biomass bumed was quantified to improve assessments of the affects on cloud physics [Guyon et al., 2005]. During the 2004 dry season, the Tropical Forest and Fire Emissions Experiment (TROFFEE) took place in Brazilian Amazonia as pat't of LBA. There were two major fire-related goals in TROFFEE. One was to employ both airbome sampling of lofted plumes and ground-based sampling of RSC so that improved fire-integrated EF could be estimated for both the main fire types: deforestation and pasture maintenance fires [Yokelson et al., 2007; Christian et al., 2007]. The second objective was to employ instrumentation capable ofmeasuring all the major types of organic emissions [Karl et al., 2007]. Nineteen fires were sampled from the air, and five were sampled from the ground. The results were synthesized as described by Yokelson et al. [2008] to produce recommended EF for the two main fire types in Brazilian Amazonia (Table 1). 2.3. Natural Variation in Emission Factors
In Figure 1, we plot the fire-average EF for selected compounds versus modified combustion efficiency (MCE = !lCOz/(!lCO z + !lCO), which serves as an indicator of the relative amount of flaming and smoldering combustion for biomass buming. This shows the natural variation in EF resulting from deforestation fires buming under a range of vegetative/environmental conditions and different mixtures of flaming and smoldering combustion. Figure 1a shows the NOx EF, which increased as MCE (and thus flaming combustion) increased. Figures Ib-1d shows the pattem typical for NMOC, the EF for these "smoldering compounds" increased with decreasing MCE. Figure Ie shows that EFPMlO also increased with decreasing MCE. The range in EF (with MCE) for the data shown is about a factor of two. In theory, capturing the variation in EF with MCE would significantly enhance the accuracy of emission estimates and the input for local-global models. For instance, if we include the RSC measurements of Christian et al. [2007], the EFC~ varies by about a factor of 20 over the MCE range sampled during TROFFEE. Unfortunately, it is not possible to measure the MCE of fires fi'om space as they occur. One cannot even be confident of seasonal trends in average MCE for fires in the major, global biomass-buming areas for reasons discussed in the work of Yokelson et al. [2007]. For example, in TROFFEE, there was evidence that the MCE oflofted plumes increased as the dly season progressed, but it is suspected that
211
the amount of low-MCE RSC may also increase as the large diameter fuels dry out [Yokelson et al., 2007]. Thus, for now, one MCE and a set of associated EF for all the detected emissions was estimated that are intended for application to the whole dly season in Table 1. The EF in Table I represents only fi'esh, minutes-old smoke. This is because ShOl'tly after emission, large, rapid changes in trace gas and particle chemisuy and particle mass can occur as documented elsewhere [Hobbs et al., 2003; Yokelson et al., 2007, 2009]. 2.4. Higher Particle Emission Factors Measured During TROFFEE
The average patiicle EFs measured druing TROFFEE are significantly larger than in previous work or recommendations. Ferek et al. [1998] repOlied a range of EFP~ from 2 to 21 g kg- I and a study average of ~ 11 g kg-I for Brazilian deforestation fires. The tower-based measurements of Ward et al. [1992] retumed values for EFPMz.5 ranging from 6.8 to 10,4 g kg- 1 with an average of ~9 g kg-I for forest fuels. The TROFFEE average value for PM lO is significantly higher at 17.8 ± 4.1 g kg-i. For most types ofbiomass burning, the PM lO values are about 20% higher than the PMz.5 or P~ values [Artaxo et al., 1998]. Applying this factor to the study average of Ferek et al. [1998] gives a projected PMlO of ~13 g kg-I, lower than the TROFFEE average, but within the unceliainty. One reason why the TROFFEE EFPM lO is higher than the projected EFPM lO based on Ferek et al. [1998] could be related to fire size and intensity. Ferek et al. [1998] noted that their largest and most intense fire in Brazil had a much higher EFPM4 or P~CO ratio than the other Brazilian fires they sampled. They proposed that EFPM increases with fire size and intensity and cited EFPM3.5 measurements from 15 to 25 gkg- I (implying an average PM lO of ~25 g kg-I) for large, intense North American fires [Radke et al., 1991; Hobbs et al., 1997]. In the TROFFEE data, the lowest EFPMlO (12-14 g kg-I) are from the smallest fires sampled [Yokelson et al., 2007]. The largest EFPMlO (26,4 g kg-I) was obtained on the largest and most intense plume encountered. Thus, we speculate that the larger TROFFEE EFPM for Brazil could be due to sampling larger, more-intense fires (on average) than in previous studies in Brazil. If cOlTect, this suggests two topics deserving futiher research: (1) what size and intensity of fires conu'ibute to what fi'action ofthe regional biomass bruning, and (2) is there a trend in fire size related to trends in land-use (discussed in section 2.1.1). 2.5. Regional Biomass-Burning Emissions Inventories for South America
Bottom-up, biomass-buming emission inventories are essentially the product of the amount of biomass bumed times
-----'"
212
BIOMASS BURNING IN AMAZONIA
LONGOET AL. Table 1. (continued)
Table 1. Emission Factors (EF) for Primaty Tropical Deforestation and Pasture Maintenance Fires Recommended EF ~
Species (Tropical Forest and Fire Emissions Experiment (TROFFEE) data) COz CO Modified combustion efficiency NOxasNO CH4 CZH4 CzHz C3H 6 HCOOH CH3COOH HCHO CH 30H Phenol Acetol + methyl acetate Furan NH3 HCN Species with no ground data CZH6 Acetonitrile Acetaldehyde ACly10nitrile Acrolein Acetone Propana1 Pyrro1e Isoprene Methyl vinyl ketone Methacro1ein Crotona1dehyde Methyl ethyl ketone Methyl propanal Benzene C6 Carbony1s 3-Methy1furan 2-Methylfuran Hexanal 2,3-Butanedione 2-Pentanone 3-Pentanone Toluene Other substituted furans Fura1dehydes Xy1enes Ethy1benzene
Ground Averagea (g kg-I)
Air Averageb (g kg-I)
1343 228.8
1615 101.4
0.788 0.33 17.12 1.42 0.09 1.43 0.26 19.73 1.88 10.30 2.42 8.89 2.08 1.64 0.35
0.910 1.77 5.68 0.95 0.28 0.45 0.59 3.43 1.66 2.57 0.34 0.72 0.33 1.08 0.68 0.90 0.37 1.38 0.04 0.58 0.57 0.09 0.11 0.37 0.35 0.14 0.21 0.45 0.16 0.26 0.21 0.53 0.08 0.01 0.66 0.07 0.03 0.20 1.08 0.26 0.13 0.08
Lab Average C (g kg-I) 1677 57.5 0.949 1.67 3.82 1.83 0.33 0.56 0.58 2.84 0.66 2.29 0.81 1.81 0.45 3.39 0.39
0.5 1.71 0.29 1.34 0.99 0.16 0.42 0.46 0.46 0.18 0.28 0.78 0.28 0.65 0.61 0.77 0.11 1.29 0.14 0.06 0.56 1.55 0.41 0.34 0.18
Primaty Deforestation Methodd (g kg-I) 1601 107.8
Pasture Maintenance Methode (g kg-I) 1506 152.4
0.904 1.70 6.25 0.98 0.27 0.50 0.57 4.25 1.67 2.95 0.45 1.13 0.41 1.10 0.66
0.861 1.19 10.26 1.14 0.20 0.84 0.46 9.95 1.75 5.66 1.17 3.99 1.03, 1.30 0.54
Methodh
Method i
1.01 0.41 1.55 0.04 0.65 0.63 0.10 0.12 0.42 0.39 0.15 0.24 0.50 0.18 0.30 0.24 0.59 0.08 0.01 0.73 0.08 0.03 0.22 1.21 0.29 0.14 0.08
213
1.80 0.74 2.77 0.08 1.16 1.13 0.18 0.22 0.75 0.70 0.28 0.42 0.90 0.32 0.53 0.42 1.05 0.15 0.03 1.31 0.14 0.06 0.39 2.17 0.51 0.26 0.15
Annual Averages
Recommended EF ;!
Amazon Regionf (Tg) 746 62.4
0.69 3.96 0.51 0.11 0.32 0.25 3.41 0.82 2.07 0.39 1.23 0.35 0.58 0.29 0.67 0.28 1.04 0.03 0.43 0.42 0.07 0.08 0.28 0.26 0.10 0.16 0.34 0.12 0.20 0.16 0.39 0.06 0.01 0.49 0.05 0.02 0.15 0.81 0.19 0.10 0.06
Global Tropical Deforestation g (Tg) 2130 143.4
2.25 8.32 1.30 0.36 0.66 0.76 5.65 2.23 3.93 0.60 1.50 0.55 1.47 0.88 1.34 0.55 2.06 0.06 0.86 0.84 0.13 0.17 0.56 0.52 0.21 0.31 0.67 0.24 0.39 0.32 0.79 0.11 0.02 0.98 0.10 0.05 0.29 1.61 0.38 0.19 0.11
Species (Tropical Ferest and Fire Emissions E4eriment (TROFFEE) data) Other TROFFEE speciesi PMJO PM2.5 Glycolaldehyde Propanenitri1e OCSk DMS k CFC 12k MeONO} EtONO zk i-PrONO} n-PrONO} 2-BuONO} 1-Butenek k TI'allS- 2-Butene k Cis-2-Butene
Ground Average" (g kg-I)
Air Averageb (g kg-I)
Lab AverageC (g kg-I)
Primary Deforestation Method d (g kg-I)
Pasture Maintenance Methode (g kg-I)
9.93 0.87 0.61
18.5 14.8 1.32 0.09
25.77
17.83
Other major species Hz Nz SOz
Amazon Region f (Tg)
Global Tropical Deforestation g (Tg)
23.4 18.7 3.09 0.17
10.06 8.04 1.06 0.06 0.0119 0.0011 0.0014 0.0078 0.0027 0.0005 0.0002 0.0003 0.0096 0.0077 0.0097
24.61 19.68 1.75 0.12 0.0329 0.0030 0.0037 0.0217 0.0076 0.0013 0.0005 0.0008 0.0266 0.0214 0.0268
48.70
17.87
34.28
1.82 1.49 0.27
5.05 4.12 0.76
0.0247 0.0022 0.0028 0.0163 0.0057 0.0010 0.0003 0.0006 0.0200 0.0161 0.0202
Total identified nonmethane organic compound
Annual Averages
l
3.8 3.1 0.57
aFrom Christian et al. [2007]. bFrom Yokelson et al. [ 2007]. CAverage of Fourier transform infrared spectrometers and proton transfer reaction-mass spectrometers if measured by both. dAssuming 5% of ground average and 95% of airborne average [Christian et al., 2007]. eAssuming 40% of ground average and 60% of airborne average [Christian et al., 2007]. fAssuming 240 Tg biomass burned in each fire type [Yokelson et al., 2007]. gAssuming 1330 Tg biomass burned [Andreae and Merlet, 2001] coupled with TROFFEE primary deforestationEFs. hComputed from 1.12 times air average [Yokelson et al., 2008]. iComputed from 2.00 times air average [Yokelson et al., 2008]. iSee Yokelson et al. [2008] for computation method. kBased on one canister sample of smoke fi'om Yokelson et al. [2007]. IFrom Andreae and Merlet [2001].
an EF. The estimation of the biomass burned can be accomplished if the aboveground biomass density, the combustion factor (the fraction of the fuel load actually combusted), and the burned area are available. In the late 1970s, Hao and Liu [1994] built a database for the spatial (5°) monthly distribution of the amount of biomass burned in tropical America, described in a paper which also includes estimates for African and Asian continents. The biomass-burning inventory that is commonly used by global models, part of the global emission source database called Emission Database for 010bal Atmospheric Research (EDOAR) [Olivier et al., 1999],
with 2.5° resolution and monthly time variation, is based on the work of Hao and Liu [1994]. Duncan et al. [2003] (hereafter D2003) combined fire-count data from the Along Track Scanning Radiometer and the advanced velY high resolution radiometer (AVHRR) World Fire Atlases to determine the typical seasonal and interannual variability of biomassburning emissions with a 1° x 1° spatial resolution. Using the Total Ozone Mapping Spectrometer Aerosol Index as a proxy to estimate the strength of emissions, the authors estimated the mean variability of CO emissions from biomass bmning. More recently, Giglio et al. [2006] and Van del' We,! et al.
---
214
LONGO ET AL.
BIOMASS BURNING IN AMAZONIA 8.0 a)
5.0
0' z
y =56.006x - 49.208 R2 = 0.759
4.0
/I)
T" ~----'-----'--'-~-'I
7.0 6.0
:i
.!! 30
5.0 .
~ 4.0 w
cS'
~ 2.0
3.0
w
y =-47.105x + 48.555 R2 =0.521
2.0
1.0
1.0 0.0 +--,---,---,---,----1 0.86 0.88 0.90 0.92 0.94 0.96
0.0 + - - , - - , - - , - - , - - - - 1 0.86 0.88 0.90 0.92 0.94 0.96
MCE
1.6
MCE
..
c)
1.4
""
LL
w 0.6
0.2
::r: 2.5
o
:f 2.0 ()
•
LL
w 1.5
•
y
d)
3.0
"
... 1.0 ::r: c3 0.8
..
3.5
.. "~~
1.2
0.4
4.0 .,----',.---'----------,
~----------,
y
1.0
=-11.338x + 11.273
=-14.654x + 15.903 R2
0.5
2
R = 0.521
0.0 +--,--,--,---,-----1 QOO QW QOO Q~ Q~ QOO
=0.485
0.0 +--,---,,---,---,---1 0.86 0.88 0.90 0.92 0.94 0.96
MCE
30 25 c:>
:ED..
MCE
,------.---.-~.--.---~--.-
..
..__,
0.7
~----------,
e)
f)
0.6 0.5
20
'"
z
15
~ 0.4
10
w
~ 0.3
LL
w
y = -113.437x + 121.076
5
.
R2 = 0.339
0+--.---,---,---,----1 QOO QW QOO Q~ Q~ QOO
~
•
0.2
I
y = 0.778x ·0.339
R2
0.1
=0.026
0.0 + - - , - - , - - - , - - - - , - - - - - 1
0.86 0.88 0.90 0.92 0.94 0.96
MCE
MCE
Figure 1. Fire-average emission factors (EF) plotted versus fire-average modified combustion efficiency for selected compounds [Yokelsoll et al., 2007]: (a) NO.n (b) CH 4, (c) EFC zH4, (d) EFCH30H, (e) PMIO, and (f) CH3CN. The range
ofEF shows that significant variability is an inherent feature of biomass buming.
[2006] using burned area estimates from remote sensing, a biogeochemical model, and EFs from the literature, estimated fire emissions during the 8-year period fi'om 1997 to 2004. This dataset, called Global Fire Emissions Database (GFED), has lOx 10 spatial resolution with 8-day and I-month time
steps. The Global Wildland Fire Emission Model (GWEM) provides emissions for several species based on the data fi'om the European Space Agency's monthly Global Burnt Scar satellite product (GLOBSCAR) and more recently GBA2000 of the Joint Research Centre of the European Commission
and results from the Lund-Potsdam-Jena Dynamic-global vegetation model for the year 2000. GWEM yields over five times less carbon monoxide emissions than GFED estimation for South America ~hd presents an early maximum emission in August, again~.Vfhe agreement of a maximum in September of all the inventories cited above. The relatively poor result of GWEM for South America was attributed mainly to the insufficient performance of the global burnt area products GLOBSCAR [Hoe1zemann et a1., 2004] and GBA2000 [Hoe1zemann, 2007] over this region. As a result of an effort motivated by the necessity of biomass-burning emission estimates with daily resolution for operational chemical weather forecasting over the South American (SA) continent, the Brazilian Biomass Burning Emission Model (BBBEM) uses a daily hybrid remote sensing fire product in order to minimize missing remote sensing fire observations [Freitas et a1., 2005; Longo et a1., 2007]. The fire database presently utilized is a combination of the Geostationary Operational Environmental Satellite (GOES)Wildfire Automated-Biomass Burning Algorithm product [Prins et a1., 1998], the Brazilian National Institute for Space Research fire product, which is based on the AVHRR aboard the NOAA polar orbiting satellites series [Setzer and Pereira, 1991; Setzer and Malingreau, 1996], and the Moderate Resolution Imaging Spectroradiometer (MODIS). The most recent GWEM version (1.4) including a correction for South America based on BBBEM methodology enhanced estimated CO emissions by 30% and improved the seasonality, shifting the emission maximum to September [Hoe1zemann, 2007]. An intercomparison between the four inventories described above for CO fi'om vegetation fires in South America was carried out by Longo et a1. [2007]. The 3-monthly mean (August, September, and October 2002) of CO flux (mg m-z day-I) is shown in Plate 1, according to (a) BBBEM, (b) GFED, (c) D2003, and (d) EDGAR. D2003 refers to the mean seasonal estimation with 10 resolution. For GFED, the 10 and 8-day time resolution data conesponding to the above mentioned time period were used. The inventories BBBEM, GFED, and that of D2003 show general agreement with respect to the emission locations and have strong gradient of the emission field. On the other hand, EDGAR prescribes a too wide and smooth emission field with values less than 150 mg m-2 day-I. BBBEM shows general agreement with the GFED in terms of patterns and estimation. D2003 shows also similar location patterns; however, over central Brazil and Mato Grosso state (fi'om 20 0 S to 12°S and fi-mn60 o W to 40 0 W) the emissions are much higher than BBBEM and GFED. The inventories all show the maximum emission over the so-called arc of deforestation [see Schroeder et a1., this volume], as expected. From all the three inventories, BBBEM yields the finest scale, since its spatial resolution can be as fine as the
215
pixel size of the satellite sensor used for fire detection. Also, as it is based on fire count detection, BBBEM emissions are velY well correlated with the number of detected fires within South America, but they are not directly proportional due to the different biomes and associated EFs attributed to each fire location. GFED and BBBEM are comparable during August, but GFED becomes much lower in September; this behavior is unexpected, since this month conesponds to the peak of the burning season. In October, fires started to be inhibited by rainfall and presented a sharp reduction in number during the last week. In this case, BBBEM, GFED, and D2003 showed the expected decrease, while EDGAR prescribed a small increase for October. It is wOlth noting that, because of its finest spatial and temporal resolution, BBBEM is able to prescribe emissions only where and when fires were in fact detected, an impOltant feature for regional chemical weather forecasting [Longo et a1., 2007] . However, all these inventories are undeniably highly sensitive to the limitations and inherent unceltainties of the EFs and input data sets used for biomass-burned estimates. Newer and promising methodologies use the fire radiative energy to estimate emission rates [Kaufinan et a1., 2003; Riggan et a1., 2004; Ichoku and Kaufinan, 2005; Smith and Wooster, 2005; Pereira, 2008]. Also, recent studies have been showing the importance of impn,wing spatial and temporal resolution of emission inventories for regional and even global modeling purposes. Regarding EFs, the addition of many reactive OVOC compounds to the list of quantified species emission (section 2.2) represents a valuable piece of information for atmospheric chemistly modelers. Incorporating the TROFFEE EF for RSC into bottom-up estimates of fire emissions fi'om Amazonia increases the estimated annual regional fire emissions for several important VOC within the range of 10-50%. Photochemical box models show that one important effect of increased VOC is to speed up the initial smoke photochemisuy [Mason et a1., 2001, 2006; Trentmann et a1., 2005]. Higher VOC emissions also imply greater potential for secondaty aerosol formation [Yoke1son et a1., 2008, 2009]. 3. LONG-RANGE TRANSPORT OF BIOMASSBURNING PRODUCTS IN AMAZONIA Atmospheric transport is driven mainly by the wind speed at the large scale and turbulence at the local scale. The typical transpOlt time scales for atmospheric constituents are 1-2 years for interhemispheric exchange, 2 weeks for meridional transport throughout latiulde belts, and about a month for the vertical transport in the troposphere [K1ey, 1997]. However, the' atmospheric u'ansport of biomass-burning emissions over tropical regions is strongly associated both with the typical intense deep moist convection and the potent updrafts
LONGOETAL.
217
CO BBBEM
related with the initial buoyancy provided by vegetation fires. These vety effiqient mechanisms of vertical transport tend to boost the lar;ge-scale atmospheric transport and significantly reduce ~~e global mixing time of biomass-burning emissions. This)§ related to higher wind speeds in the free troposphere, where the pollutants are more rapidly advected away from the source regions. Also, when the pollutants are transported to the free troposphere, their residence time increases because removal processes are much less efficient than in the planetary boundary layer (PBL). These processes altogether define the vegetation fires that are widely spread over tropical regions as a key agent on the regional and global distribution of trace gases and aerosol particles and their consequent impacts on regional and global climate.
10N
;t
EQ
lOS
20S
30S
D2003
aSn =>v(-) 1 ( ---r=>i) -Q -+v. s11 =--V. Posnv + n' at Po
3.1. Numerical Modeling ofthe Atmospheric Transport ofBiomass-Burning Emissions
10N
Numerical modeling of atmospheric biomass-burning emission transport requires the solution of the continuity equation for trace gases and aerosol particles mixing ratios sn = PI/PO:
EQ
aS
lOS
~
Q
n -+v.Ll(sl)= -n, at 1 Po
20S
30S BOW
70W
70W
Plate 1. The 3-monthly mean CO distribution offour biomass-burning inventories (a) Brazilian Biomass Burning Emission Model, (b) Global Fire Emissions Database, (c) Duncan et al. [2003], and (d) Emission Database for Global Atmospheric Research for August-November 2002 [Longo el al., 2007]. The color scale refers to the mean amount of CO emitted in mg m-2 day-I. 10N
EQ
lOS
20S
30S
40S
70W
60W
50W
40W
BOW
70W
60W
50W
40W
Plate 2. Aerosol optical depth at 500 mll (color scale) for 27 September 2002 from (a) CCATT-BRAMS model and
(b) Moderate Resolution Imaging Spectroradiometer retrieval. The streamlines at 2 km height from the model are also shown in Plate 2a.
(1)
where Pl1 and Po are the densities of the tracer 11 and the air, respectively, and is the wind velocity. The term Qn' normally called forcing, contains all the physical and chemical processes for production and loss of the species, 11, which are mainly emissions (already described in section 2 for the biomass-burning sources), chemical reactions, particle transformations, or dty and wet removal processes. This set of equations, together with appropriate initial and boundary conditions (in the case oflimited-area models), provides the evolution in space and time of emitted trace gases and aerosols mixing ratios. In general, this set of equations has no analytical solution. It requires numerical methods, parameterizations, and computer resources for an approximated solution through a discretization methodology (e.g., finite differences). The computational limitation implies the use of the so-called "scales separation" of all possible atmospheric motions, which is detennined by the chosen space-time discretization. This basically means that the discretization will necessarily separate all the existent atmospheric motion scales in two families: the processes that are explicitly solved (grid scale) and those that are not solved (subgrid scale). However, the nonlinear aspect of the equations involved allows energy exchange between scales and so subgrid processes do generally have a net effect on the grid-scale variables. The accounting for the net effect of subgrid fluxes on the
v
grid scale is achieved by so-called parameterizations, which are undoubtedly simple compared to the highly complex real physical processes they mean to represent. They are usually based on limited observational data sets and on the presently still incomplete level of understanding of interscales exchanges. Thus, physical parameterizations are recognized as an important source of uncertainties in numerical modeling of the atmosphere in general. The numerical solution for the mass conservation equation (equation (1» can be achieved through the spatial and temporal discretization and decomposition of the tracer mixing ratio and wind speed into their mean values and fluctuation components (the Reynolds decomposition) [Stull, 1988]. Following this approach, equation (1) can be rewritten as (2)
The second term on the left side of equation (2) refers to the advection in the grid scale, and Ql1 is the mean net production in the grid cell by all processes not described as transpOli. The first term on the right side should include all the subgrid or nonresolved transpOli mechanisms. Moreover, CUlTent computational power does not allow equation (2) to be solved at once, considering all tetms simultaneously. The Splitting Operator is a popular technique to do this: instead of solving the full equation at:once, it solves each process independently and then couples tne various changes resulting from the separate pariial solutions [Yanenko, 1971; Seinfeld and Pandis, 1998; Lanser and Verwer, 1998]. It is worth highlighting that in this framework, the solution of equation (2) represents the mean tracer mixing ratio sn within the grid volume of finite spatial dimensions (.6..," Lly, Llz). Then, model results must be compared with observational data, taking into account the scale and representativeness ofthe latter. Several atmospheric pollutants transport models on regional and global scales have been proposed in the literature. Chatfield et al. [1996] used the Global-Regional Atmospheric Chemistry Event Simulator to introduce a conceptual model of fire emissions and chemical production of the African/Oceanic plumes. Grell et al. [2000] described a multiscale complex chemistry model coupled to the Penn StatelNational Center for Atmospheric Research nonhydrostatic mesoscale model (MM5). The Georgia Tech/Goddard Global Ozone Chemistry Aerosol Radiation and Transport (GOCART) model is an example of a global transport model. Chin et al. [2000] employed GOCART to simulate the atmospheric global sulfur cycle. Model of Ozone and Related Tracers is an "off-line" global chemical transport model appropriate for simulating the three-dimensional (3-D) distribution of chemical species in the atmosphere
218
LONGO ET AL.
BIOMASS BURNING IN AMAZONIA
[Brasseur et aI., 1998; Horowitz et al., 2003]. More recently, fully coupled "online" regional transport models based on atmospheric models are becoming more common, such as the Coupled C~emistry-Aerosol-Tracer Transport model coupled to Brazilian Regional Atmospheric Modeling System (CCATT-BRAMS) [Freitas et al., 2009; Longo et al., 2007] and the Weather Research and Forecasting Model [Grell et al., 2005; Fast et al., 2006], to name but a few. CCATT-BRAMS, developed in the context of the LBA program, has been designed to provide a suitable tool to study the atmospheric transport of biomass-burning emissions and their impacts on weather and air quality. It is an Eulerian transport model fully coupled to the BRAMS regional model. The tracer transport simulation is made simultaneously, or "online", with the atmospheric state evolution using exactly the same time step, as well as same dynamics and physical parameterizations. The general mass continuity equation for tracers (in a form of tendency equation and in the context of the Splitting Operator) solved in the CCATTBRAMS model is:
as (as)
at = . at
(as) adv
at
+
~~f +
(as)
~
II
at
~~~~'II
III
(3)
+(as) +(as) +W +R +Qpr' at at sh,t!
chem reac
CDIlV
~
V
'-v-I
"-v-"
VI
VII
'-y-'
VJJI
where s is the grid box mean tracer mixing ratio, term (I) represents the 3-D resolved transport term (advection by the mean wind), term (II) is the subgrid-scale diffusion in the PBL, terms (III) and (IV) are the subgrid transport by deep and shallow convection, respectively. Term (V) is the net production or loss associated to chemical reactions. Term (VI) is the wet removal, term (VII) refers to the dry deposition applied to gases and aerosols particles, and finally, term (VIII) is the source term that includes the plume rise mechanism associated with vegetations fires. Figure 2 illustrates the main subgrid-scale processes involved in biomass-burning smoke trace gases and aerosols transport simulated by the CCATT-BRAMS system. A detailed description of the parameterizations for each one of these processes can be found in the work of Freitas et al. [2005,2007,2009] and Longo et al. [2007]. 3.2. Main Processes Related To Smoke Atmospheric Transport
Vegetation fires emit trace gases and aerosol particles to the atmosphere with temperatUres much higher than the ambient air and with positive buoyancy, which favors vertical transport. Due to the radiative cooling and the efficient heat transport by convection, there is a rapid decay of temperature above the fire area. Also, the interaction between smoke and the environment produces eddies that entrain colder environmental air into the smoke plume, which dilutes the plume and reduces buoyancy. The dominant characteristic is a strong upward flow with an only moderate temperature
Model top 20-30 km
mass
inflow
1',1.
N
100
1000 m /',)<
N
10-100 km
Figure 2. Several subgrid processes involved in gases/aerosols transport and simulated by Coupled Chemistry-AerosolTracer Transport model coupled to Brazilian Regional Atmospheric Modeling System (CCATT-BRAMS) system. Extracted from the work of Freitas et al. [2007].
excess above ambient [Riggan et a1., 2004]. The final plume height is controlled 1}Y the thermodynamic stability of the atmospheric environtl1ent and the surface heat flux released fi'om the fire. Mor16ver, if water vapor reaches condensation, the additional bj,lDyancy gained from the latent heat release plays an important role in determining the effective injection height of the plume [Freitas et al., 2007]. However, the occurrence of strong horizontal winds might enhance the lateral entraimnent and even prevent the plume reaching the condensation level, particularly for small fires, impacting on the injection height. Low-density biomass fires, such as burning of cerrado and pastures, typically release smoke into the PBL. On the other hand, forest fires, with high-density vegetation and heat release rate of about 10 GW that typically last for a few hours can inject smoke directly into the low and medium troposphere (3 to 10 km high) and even into the stratosphere developing pyrocumulus [Fromm et al., 2000; Fromm and Servranckx, 2003; Jost et al., 2004; Rosenfeld et al., 2006]. Including the plume rise fi'om vegetation fires driven by their own initial buoyancy into regional and global models is a difficult task. In the absence of this mechanism, the pyrogenic emissions often are released at the surface It;vel in the model, or vertically distributed in an arbitrary way [Turquety et al., 2007], or using some empirical relationship between the injection height and fire intensity [Lavoue et al., 2000; Wang et al., 2006]. Freitas et al. [2006, 2007] introduced this subgrid process by embedding a I-D cloud-resolving model with appropriate lower boundary conditions in each column of the 3-D atmospheric model. This I-D plume model is driven by remote sensing fire location and size, a look-up-table of typical range of heat fluxes, and the updated atmospheric conditions provided by the 3-D host model. This allows the plume rise to be simulated explicitly within each model column with fires, which provides the effective injection height of material emitted during the flaming phase. On the other hand, the smoke fi'action released into the PBL is mixed and vertically transported by turbulence, producing a homogenous mixing layer 1 to 3 Ian deep during the day. However, dense smoke haze layers can produce a net cooling of the air near the surface and a weakening of the mixing layer hU'bulence due to solar radiation attenuation, which inhibits the smoke mixing [Longo et aI., 2006]. AOD values of 1-3 (500 mn channel) correspond to a negative radiative forcing range of 120-250 Wm-2 [Procopio et aI., 2004; Schafer et aI., 2002b; Artaxo et al., this volume, and references therein]. In fact, AOD observations of the Aero-. sol Robotic Network sites in Amazonia, with high smoke influence,frequently yield AOD values of up to 3 at 500 mll channel [Hoelzemannet aI., 2009]. Shallow and nonprecipitating convective systems over the Amazon basin grow normally on the top of the PBL and,
219
typically, transport gases and particles to the low troposphere enhancing their atmospheric dispersion. The deep convective and precipitating systems, however, act differently, depending on the hygroscopicity properties of the atmospheric constituents. For example, CO 2 and CO, which have low hygroscopicity, are efficiently transported by the ascending stream to the cloud top and detrained into the medium and high troposphere, while carbonaceous aerosol particles are more efficiently absorbed into cloud droplets and scavenged with precipitation. Convective systems also induce the development of descending streams, which bring air parcels fi'om the midtroposphere to dilute and cool the PBL. Several authors [e.g., Chatfield and Crutzen, 1984; Dickerson et al., 1987; Pickering et al., 1988; Thompson et aI., 1996; Chatfield et al., 1996; Longo et al., 1999; Andreae et al., 2001; Freitas et al., 2000, 2005] have been shldying the transport of trace gases and aerosols fi'om biomass burning, with special attention to the atmospheric transport by circulations associated with deep and moist convection. They showed the relevance of these mechanisms on the distribution of pollutants in the medium and high troposphere. The cloud venting is taken into account in regional- or global-scale transport models through cumulus parameterizations normally using the mass flux approach. The effectivene~s of plume rise is comparable with cloud venting by deep moist convection as a mechanism for transporting sJ;l1oke from the PBL to the upper troposphere, and both! are much more effective than shallow convection. A detCliled discussion about the relative role of these three smoke vertical transport mechanisms is given by Freitas et al. [2007], who compared model results with CO data retrieved by the "Measurements of Pollution in the Troposphere" (MOPITT) instrument, on board the EOS/Terra satellite [Em1l10ns et al., 2004]. Basically, the total absence of any subgrid-scale convective transport in the model results in a heavily polluted PBL and a very clean free troposphere. When only shallow convection is considered, it gives a minor gain in model performance. Even though deep convection allows a better representation of the transport to the upper troposphere, it alone is not enough to describe proper venting from lower to middle levels. The plume rise mechanism alone provides much better results for CO in the PBL and the lower and middle troposphere, but does not allow the upper troposphere to be correctly populated by CO. CO transport models that include all the main vertical transport mechanisms, shallow and deep moist convection, and the pyroconvection induced by vegetation fires, show the best agreement with the MOPITT CO retrieval. CCATT-BRAMS model simulations of biomass-burning emissions were also evaluated with airborne measurements
220
BIOMASS BURNING IN AMAZONIA
of CO within the 5-km colunm covered by the aircraft [Freitas et al., 2009; Longo et al., 2007] during the LBA field campaigns ~oke, Aerosols, Clouds, Rainfall, and Climate (SMOCC) and Radiation, Cloud, and Climate Interactions in the Amazon (RACCI) that took place in the Amazon basin between mid-September and early November 2002 [Fuzzi et al., 2007]. These model results show that the inclusion of the transport terms described above and represented in equation (3) are sufficient to capture the general pattern ofsmoke transpOli either regarding veliical profiling in the PBL and lower troposphere and regional distribution. Although, the model resolution of 35 km did not allow the point-by-point reproduction of the subgrid phenomena effects in the profiling, it did succeed in representing the mean pattern of each airborne profile, with the model results falling within the standard deviation of observations in most of the cases. See the work of Freitas et al. [2009] and Longo et al. [2007] for details. During the SMOCC/RACCI campaign, high values of CO and PM 2 .5 were observed near surface level in an Amazonian site under strong influence of fire emissions, Fazenda Nossa Senhora Aparecida (l0045'44"S, 62°21 '27"W) near the town Ouro Preto do Oeste in the State of Rondonia. Maximum values of CO and PM 2 .5 observed there were as high as 4000 ppb and 210 Ilg m-3 , respectively. The time series of CO and PM2 .5 were characterized by strong vmiability, associated either with the transport of aged smoke and fresh emissions from local fires in the vicinity ofthe measurement site. Longo et al. [2007] have demonstrated that to be able to simulate the observed strong time variability of CO or PM2 .5 near surface level, it is clitical to use daily remote sensing fire counts to correctly ascertain emissions in space and time. The use of climatological or monthly variating emissions results in simulation errors of smoke tracer concentrations concerning both time variability and magnitude. HOlizontally, the atmospheric transpOli is dominated by advection, which drives the smoke toward the atmospheric flow either in the PBL or free troposphere. To illustrate the long-range transpOli of biomass-burning emissions, in Plate 2, we show a regional smoke plume covering a considerable pmi ofthe SA continent revealed by AOD (channel 550 nm) (a) simulated by CCATT-BRAMS model on 27 August 2002 and (b) retrieved by MODIS-Terra. Smoke emitted from vegetation fires in the Amazon Basin and central Brazil was transpOlied southward following the atmospheric flow in the PBL (see the streamlines at 2 km height in Plate 2a). The approach of a cold front system (not shown) sloped up the low-level polluted air (to typically around 6-10 Ian high), which was then transpOlied toward the Atlantic Ocean driven by a midlatitude wave train. Model dynamic was able to fairly reproduce the general shape and intensity of this continental smoke plume.
LONGO ET AL.
221
3.3. The General Pattern ofAtmospheric Transport of Biomass-Burning Emissions Over South America The burning season of the SA continent occurs during austral winter. The westward displacement of the South Atlantic Subtropical High (SASH) pressure system and the northward motion of the Intertropical Convergence Zone (ITCZ) establish a high pressure area with little precipitation and light winds in the lower troposphere over the central region of the continent [Satyamurty et al., 1998], synchronized with a shift of the convection in the Amazon basin to the nOlihwestern part of SA. This climatology propitiates the spreading of fires all over SA, and a dense regional plume of smoke covers an area of about 4-5 millions of square kilometers that persists for about 3 months. The smoke transport mean pattern indeed may be explained in terms ofthe trade winds, the SASH, and the barrier effect of the Andes Mountains. The position of the SASH determines the inflow of clean maritime air into the biomassburning area, playing an important role in defining the shape of the regional smoke plume as it is the primary mechanism responsible for the dilution ot' polluted air. In the nOliheast region, in spite ofthe typical huge number offires, the smoke loading is relatively low due to the continuous venting of clean oceanic air carried by the trade winds, besides the typical low vegetation fuel load. The Andes Mountains on the west side of SA, together with the SASH, impose a longrange transport of smoke from its source areas to the south and southeast of SA, thus disturbing larger areas downwind in the subtropics. Most ofthe smoke in the lower troposphere is exported to the Atlantic Ocean throughout the southeastern part of the continent driven by the South American Low Level Jet (SALLJ) on the east side of the Andes. The SALLJ is a wind maximum immersed in a poleward warm and moist flow in the low troposphere [see Nobre et al., this volume; Marengo et al., this volume; Vera et al., 2006]. The episodic intelTuptions of the SALLJ by cold fronts aniving on subtropical SA are responsible for disturbances in atmospheric stability 'and in the wind fields defining the latitude of the southeastward smoke flow. These events also periodically cause a phenomenon called friagem [Marengo et al., 1997a, 1997b] that generates frost in southern and southeastern Brazil as well as changes in the wind speed and direction and surface temperature and humidity deep into the north of Amazonia. The episodes of friagems allow the smoke to invade pristine areas ofthe Amazon basin, with implications for the atmospheric chemistry. The transport of the smoke to the nOlihwestern pmi of Amazonia toward the convective zone enhances the transport of smoke products to the upper troposphere. In fact, a well-defined regional layer of smoke tracers at the upper levels (~500 hPa) over SA has been
Plate 3. Atmospheric Infrared Sounder 500 hPa CO retrievals (ppbv, color scale) for 22 September 2002 (adapted from McMillan et al. [2005]).
observed by airborne measurements as well as by remote sensing [Andreae et al., 2001; McMillan et al., 2005]. A typical pattern for this upper level smoke layer distribution is shown in Plate 3. Modeling studies indicate the deep moist convection and pyroconvection as the key mechanisms act-
A
B CO(ppbv)
....e
--
CO
E
20N
~
CI)
";::..
ing on this transport [Freitas et al., 2000, 2007; Andreae et al., 2001; Gevaerd et al., 2006]. During the LBA-Cooperative LBA Airborne Regional Experiment (CLAIRE) 1998 field campaign [Andreae et al., 2001], airborne measurements over Suriname sampled a strong polluted layer with
15N
to
:t
JOO
ION 5N
<3:'<:1'
EQ 5S
175 150 140
lOS 15S
130
120
110
o 360
100 365
200 370
300 375
PI~te 4. (a) Airborne measurements of (red) CO and (green) CO 2 over Suriname obtained during CLAIRE-08 field campaign [Andreae et al., 2001], and (b) model simulation of biomass-burning CO (ppbv, color scale) around 11 Ian height on26 March 1998 from Roraima fires [Gevaerd et al., 2006].
222
BIOMASS BURNING IN AMAZONIA
chemical composition characteristic of aged biomass-burning smoke over a clean air column at high altitudes above 9 km on 26 March 1998 (Plate 4a). Back-trajectOlY analysis indicated th:t this layer was associated with the emissions from severe wild fires in cerrado and forest areas in Roraima State in the north of Amazonia [Freitas et al., 2000; Andreae et al., 2001]. The fire emissions were advected southwestward in the low troposphere until they got to a deep convection area. The smoke was then entrained into these deep clouds, which transported its lo~ hygroscopic fi'action to the upper troposphere (Plate 4b). In general, the wet deposition resulting from the smoke/ cloud interaction processes tends to be associated with local precipitation over Amazonia, but mainly with low level jets or South Atlantic anticyclones, connecting Amazonia and the southern part of South America via biogeochemical cycling of nutrients (Plate 5a). On the other side, the dry deposition of smoke aerosol particles coincides mostly with the biomass-burning emissions area (Plate 5b). Biomass-burning emissions include ozone precursors that together with natural VOC and plenty ofUV radiation in Amazonia efficiently form tropospheric ozone (see section 5.1). The 0 3 is produced downwind in the vicinity of fire regions, which typically defines two main corridors of 0 3 deposition, following the edge of the Andes Mountains southward and northward (Plate 6). This pattern is associated with events of 0 3 and its precursors transport to the north by the cold front approach and to the south by the anticyclone circulation and SALLJ. Over Sao Paulo state, a corridor is also formed starting from Sao Paulo metropolitan area, involving mainly reactions of NOx and VOCs from urban (mainly vehicles source emissions) and rural areas (such as sugar cane burning). These transport and deposition patterns might induce degradation of forest and agricultural areas (such as sugar cane within Sao Paulo State and soya bean within Mato Grosso State). 4. REGIONAL AND REMOTE IMPACTS OF BIOMASS-BURNING PRODUCTS In its unperturbed state, the Amazonian atmosphere is characterized by velY low concentrations of aerosols and oxidants (Figure 3) [Andreae et aI., 2002; Artaxo et al., 2002; Andreae, 2008]. The emission of smoke from biomass burning therefore causes dramatic changes in the radiative, cloud physical and chemical properties of the atmosphere over Amazonia, which affect regional climate, ecology, water cycle, and human activities. These changes are summarized in Figure 4, which shows the processes in the atmosphere over the perturbed and smoke-polluted Amazonia. Given the magnitude of burning activity in Amazonia, these perturbations can affect the climate system even on a global scale.
LONGO ET AL.
223
4.1. Impacts on Atmospheric Chemist/)! Vast amounts of biogenic VOC are continually emitted from the rainforest into the atmosphere [see Kesselmeier et al., this volume, and references therein]. These compounds are constantly being removed from the atmosphere by oxidation into water-soluble compounds (e.g., polar organics or CO 2) and subsequent surface dly deposition or uptake by cloud drops, snow or ice, followed by precipitation. The most important initial step in the chemical removal mechanisms is the reaction with the hydroxyl radical OR, the atmospheric "detergent" [Crutzen, 1995]. The primaty hydroxyl radical source is the photodissociation of ozone and subsequent reaction of oxygen atoms with water. OR concentrations are highest in the tropics because of its regime with high levels of UV radiation and water vapor. Most of the oxidation of methane, CO, and other trace gases occurs in the "Great Tropical Reactor," the region of high hydroxyl radical concentrations in the tropical troposphere (Figure 3). The tropical region, and patiicularly Amazonia, thus plays a key role not only in regulating physical climate, but also in maintaining the chemical cOlnposition of the atmosphere. The reaction with OR radicals is also the dominant sink for methane; therefore, changing OR concentrations also affect the lifetime and thus the atmospheric concentration of this important greenhouse gas. The relative amounts of hydrocarbons and NO x play crucial roles in the photochemical oxidation of hydrocarbons. At velY low levels ofNOx, a characteristic ofthe unperhll'bed Amazon, hydrocarbon oxidation removes ozone and consumes hydroxyl radicals, while at higher NOx levels, more ozone and reactive radicals are produced [Butler et al., 2008]. Fires emit a huge variety of trace gases (summarized in section 2), comprising the main ingredients of smog chemistry, VOC (including OVOC) and NO.1' The addition ofpyrogenic NOx thus transforms the Amazonian atmosphere from an oxidant-consuming into an oxidant-producing enviromnent and set up the same processes that are active in urban smog. This includes the development of high ozone concentrations, irritant gases such as peroxyacyl (PANs) and acidic components, such as nitric acid and a variety of organic acids [Browe!! et al., 1990; Jacob and Wofty, 1990; Kirchhoffet al., 1990; Richardson et al., 1991; Mauzera!! et al., 1998; Thompson et al., 2001]. In addition to the effects of pyrogenic trace gases, the interaction of smoke aerosols with solar radiation also changes the photolysis rates of key components of the photochemical reaction chains [Albuquerque et al., 2005] and thereby affects atmospheric chemical processes. Oxidant chemistry, including 0 3 formation, begins within the fire plumes from biomass burning [Andreae et aI., 1988; Mauzera!! et al., 1998] and continues in the regional atmo-
250
750 500
100 75 50 25 15 10 70W
60W
50W
40W
Plate 5. Accumulated (a) wet and (b) dly deposition of smoke aerosol as simulated by CCATT-BRAMS model. The color scale refers to the total amount of aerosol deposited throughout August and September 2002 in mg m -2.
sphere [Kirchhoffet al., 1989,1990; Richardson et aI., 1991]. Ultimately, air masses containing elevated ozone concentrations are expOlied from the SA continent over the Pacific and Atlantic oceans, and even to other continents, especially Southern Africa. This leads to seasonally very high ozone concentrations, especially over the central South Atlantic [Fishman et al., 1996; Thompson et aI., 1996,2001]. As a result of the smog chemistly caused by pyrogenic emissions, the Amazonian forest is subjected to substantial deposition of nutrients, but also plant-toxic compounds, especially 0 3 [Gut et al., 2002; Kirlanan et aI., 2002; Rummel et al., 2002, 2007]. The deposition ofOVOC species, such as organic acids and aldehydes, is also elevated during the fire season [Kesselmeier et al., 2002; Kuhn et aI., 2002]. Ozone concentrations over forests during the burning season are sufficiently high that they must be expected to reduce plant primaty productivity. On the other hand, nitrogen deposition may have some feliilizing effect to the remaining rainforest, albeit at the expense of the forest that has been burned elsewhere. In general, extremely bad air quality conditions persist during 90% of the burning season pe110d, causing health problems in the exposed communities [Ignotti et al., 2007,2009].
[Talbot et al., 1988; Echalal' et al., 1998; Artaxo et al., 2002] results in a sharp increase in scattering and absorption of incoming sunlight.! This is evident in an increase of aerosol optical thiclmess (a measure of the extinction of sunlight by aerosols) from Val}leS around 0.05-0.08 in the wet season to
4.2. Impacts on Atmospheric Radiation, Photosynthesis, and Radiative Forcing The dramatically elevated concentrations of aerosol particles in the Amazonian atmosphere during the fire season
Phlte 6. Accumulated deposition of 0 3 as simulated by CCATTBRAMS model. The color scale refers to the total amount of ozone deposited throughout August and September 2002 in 10-3 kg m-2 .
224
LONGO ET AL.
BIOMASS BURNING IN AMAZONIA
THE GREAT TROPICAL REACTOR as the bio:sptllere
Figure 3. The great tropical reactor as operated by the biosphere. Copyright M. O. Andreae; 2004. Reprinted with
permission.
0.9 or more in the fire season [Andreae, 2008; Schqfer et al., 2008]. The perturbation of solar radiation flux by pyrogenic aerosols affects vegetation by changing the light climate to which plants are exposed and thereby the carbon budget of the Amazon Basin. It also affects the energy budgets of the
surface and troposphere, and thus causes direct radiative forcing of climate and modification of cloud processes and precipitation. Finally, aerosols also influence atmospheric photochemistry by changing the radiative flux and thus the photolysis rates of impOliant chemical species such as 0 3
Figure 4. The great tropical reactor perturbed by deforestation and pollution. Copyright M. O. Andreae, 2004. Reprinted
with permission.
and N02 [Dickerson et al., 1997; Castro et al., 2001; Albuquerque et al., 2005]. By scattering li~Jtt' back to space and by absorbing light, aerosols reduce th,e amount of direct solar radiation available for plant photosy6thesis. On the other hand, some of the scattered light is scattered in the fOlward direction and anives on the canopy in the form of diffuse radiation. Overall, the canopy thus receives less light, but at a higher ratio of diffuse to direct radiation [Schafer et al., 2002a, 2002b, 2008). This results in a complex response of photosynthesis to increasing aerosol levels, as less light becomes available to the leaves at the top of the canopy, but more light reaches the "shade leaves" that only receive diffuse radiation. As a result, net primmy production initially increases with increasing aerosol load, but then decreases again at even higher aerosol burdens [Yamasoe et al., 2006; Oliveira et al., 2007). The presence of an aerosol layer reduces the amount of solar energy arriving at the surface and thereby produces a negative (cooling) radiative forcing at the surface. Values of -20 to -70 W m-2 have been reported for this forcing in Amazonia [Ross et al., 1998; Procopio et al., 2004]. On the other hand, the absorption of light by the light-.absorbing carbon (LAC) component of the smoke aerosol [Andreae and Gelencser, 2006] leads to a warming of the tropospheric layers in which the smoke resides. This results in a stabilization of the atmosphere and consequently a reduction of cloudiness [Feingold et al., 2005; Longo et al., 2006; Zhang et al., 2008). Because of the high reflectivity of the smoke aerosols, they reflect more light back to space than the unpolluted Amazonian atmosphere, provoking a net cooling forcing to the radiation budget measured at the top of the atmosphere. During the dly season, this forcing is of the order of-5 to -12 W m-2 [Procopio et al., 2004). Thus, the net effect of smoke aerosols is a cooling forcing that is quite pronounced at the local and regional scale, and even significant for global climate [Robock, 1991].
225
wet season, and changes in the basin-scale patterns of wind divergence and convergence [Silva Dias et al., 2002; Zhang et al., 2008). As aerosols also act as cloud condensation nuclei (CCN), they are able to change the microphysical behavior of clouds and, consequently, also their dynamics and precipitation efficiency [Rosenfeld et al., 2008; Martins et aI, 2009]. Over Amazonia, the large differences in CCN concentration between wet and dly season lead to pronounced changes in cloud microphysical properties, especially the droplet effective radius [Roberts et al., 2003; Kaufinan and Nakajima, 1993; Feingold et al., 2001). This increases the reflectivity of the clouds and has a cooling effect on climate. It also reduces the rate at which cloud droplets can grow to raindrops in those parts of the cloud that are below the freezing level. The change in microphysical propeliies also induces a reduction or complete suppression of rainfall from relatively shallow ("warm") clouds [Andreae et al., 2004; Rosenfeld et al., 2008; Silva Dias et al., 2002). The suppression of early rain from the "warm" part of the clouds allows more of the water vapor to ascend to the freezing level and above, where more water can condense because of the lower temperature of condensation. Furthermore, the latent heat of freezing is released in addition to the latent heat of condensation. Both of these effects result in an invigoration of cloud dynamics and an intensification of precipitation [Rosenfeld et al., 2008; Martins et al., 2009]. At even higher aerbso1 concentrations, the fonnation of precipitation is suppressed even in cold clouds, and the cooling radiative effect of the aerosol reduces the energy available for convection. Consequently, the invigoration of convection and precipitation by aerosols has a maximum at intermediate aerosol concentrations around 1000-3000 cm-3 . Observational suppOli for this conceptual model has been found over Amazonia by remote sensing studies [Lin et al., 2006; Koren et al., 2008). The increase in the role of the mixed-phase region in clouds (i.e., the region where water and ice phase coexist) 4.3. Impacts on Clouds and Precipitation by aerosol microphysical effects also has consequences on The effect of pyrogenic aerosols on the surface and at- the type and amount of lightning activity. Studies in Amamospheric radiation budget has already been mentioned in zonia have shown an increased lightning activity under the the previous section. The resulting suppression of cloudiness presence of biomass smoke [Williams et al., 2002; Andreae is further enhanced by the "cloud burning" effect of LAC et al., 2004]. Due to their chemical composition, smoke aeroparticles inside cloud air and cloud droplets, which leads to sols also enhance the frequency of positive cloud-to-ground a warming inside the cloud, in general, and the droplets, in lightning strokes [Lyons et al., 1998; Fernandes et al., patiicular. This causes clouds to evaporate even when they. 2006). have formed in spite of the reduction of surface heating, an effect thaI has been observed by remote sensing over Ama- 4.4. Global EfJects zonia [Koren et al., 2004). Overall, the radiative effect of smoke aerosols on clouds leads to reduced cloudiness (parThe implications of biomass burning in Amazonia for gloticularly for small clouds), a delayed transition from dry to bal climate and atmospheric composition remain to be fully
226
BIOMASS BURNING IN AMAZONIA
LONGO ET AL.
explored. Obviously, during years of very extensive burning Andreae, M. 0., et a1. (1988), Biomass-burning emissions and associated haze layers over Amazonia, J Geophys. Res., 93, due to climatic variability (El Nino) or extreme deforestation 1509-1527. rates , the enhanced emissions of greenhouse gases are seen It Andreae, M. 0., et a1. (2001), Transport of biomass burning smoke as interarmmll variations in the growth rate of CO 2 , CH4 , etc. to the upper troposphere by deep convection in the equatorial [Langenjelds et al., 2002]. Teleconnections resulting from region, Geophys. Res. Lett., 28, 951-954. perturbation of convection dynamics are difficult to explore Andreae, M. 0., et a1. (2002), Biogeochemical cycling of carbon, at this time because of the inadequate parameterization of water, energy, trace gases, and aerosols in Amazonia: The LBAaerosol effects 011 clouds and precipitation in global models, EUSTACH experiments, J Geophys. Res., 107(D20), 8066, but initial studies have shown significant effects [Nobel' et doi:l0.1029/2001JD000524. Andreae, M. 0., D. Rosenfeld, P. Artaxo, A. A Costa, G. P. Frank, a I., 2003]. Due to the fact that burning predominantly takes K. M. Longo, and M. A F. Silva-Dias (2004), Smoking rain place in the trade wind region, a substantial part of the emisclouds over the Amazon, Science, 303, 1337-1342. sions are transported toward the ITCZ, where they can beArtaxo, P., E. T. Fernandes, J. V. Martins, M. A. Yamasoe, P. V. come subjected to deep convection and transported into the Hobbs, W. Maenhaut, K. M. Longo, and A Castanho (1998), upper troposphere and the tropical transition layer [Freitas Large-scale aerosol source apportionment in Amazonia, J Geoet al., 2000; Andreae et al., 2001]. Enhancement of convecphys. Res., 103, 31,837-31,847. tion by the mechanisms discussed above and the suppression Artaxo, P., 1. V. Martins, M. A Yamasoe, A S. Procopio, T. M. of scavenging at very high aerosol levels make the vertical Pauliquevis, M. O. Andreae, P. Guyon, L. V. Gatti, and AM. C. transport of smoke particularly effective. While a large fracLeal (2002), Physical and chemical properties of aerosols in the tion of aerosols is removed by scavenging during these conwet and dry seasons in Rondonia, Amazonia, J Geophys. Res., vection events, even a modest fraction of surviving smoke 107(020), 8081, doi: 10.1029/200 lJD000666. aerosols can make an important contribution to the aerosol Artaxo, P., et a1. (2009), Aerosol particles in Amazonia: Their composition, role in the radiation bal~nce, cloud formation, and nutribudget of the very clean upper troposphere. The same apentcycles, Geophys. Monogr. Ser., doi: 10.1029/2008GM000778, plies to reactive pyrogenic trace gases, e.g., acetone and forthis volume. maldehyde, which can play important roles in the chemistry Barbosa, R. 1., and P. M. Fearnside (1996), Pasture burning in of the upper troposphere. Acknowledgment. The authors would like to thank Judith Hoelzemann for the final revision of the manuscript.
REFERENCES Albuquerque, L. M. M., K. M. Longo, S. R. Freitas, T. Tarasova, A Plana Fattori, C. Nobre, and L. V. Gatti (2005), Sensitivity studies on the photolysis rates calculation in Amazonian atmospheric chemistry-Part I: The impact of the direct radiative effect of biomass burning aerosol particles, Atmos. Chem. Phys. Disc., 5, 9325-9353. Andrade, S. M. A, W. N. Neto, and H. S. Miranda (1999), The dynamics of components of the fine fuel after recurrent prescribed fires in Central Brazil savannas, Proceedings of the Bushfire 99 Conference, Album, Australia. Andreae, M. O. (2008), Correlation between cloud condensation nuclei concentration and aerosol optical thickness in remote and polluted regions, Atmos. Chem. Phys. Disc., 8,11,293-11, 320. Andreae, M. 0., and A Gelencser (2006), Black carbon or brown carbon? The nature of light-absorbing carbonaceous aerosols, Atmos. Chem. Phys., 6,3131-3148. Andreae, M. 0., and P. Merlet (2001), Emission of trace gases and aerosols from biomass burning, Global Biogeochem. Cycles, 15(4), 955-966.
Amazonia: Dynamics of residual biomass and the storage and release of aboveground carbon, J Geophys. Res., 101(D20), 25,847-25,857. Bergamaschi, P., R. Hein, C. A M. Brenninkmeijer, andP. 1. Crutzen (2000), Inverse modeling ofthe global CO cycle 2. Inversion of 13C/l2C and 18 0/ 160 isotope ratios, J Geophys. Res., 105(D2), 1929-1945. Bertschi,1. T., R. 1. Yokelson, D. E. Ward, R. E. Babbitt, R. A Susott, 1. G. Goode, and W. M. Hao (2003), Trace gas and particle emissions from fires in large diameter and belowground biomass fuels, J Geophys. Res., 108(D13), 8472, doi: 10. 10291 2002JD002100. Blake, N. 1., D. R. Blake, B. C. Sive, T.-Y. Chen, F. S. Rowland, 1. E. Collins, G. W. Sachse, and B. E. Anderson (1996), Biomass burning emissions and vertical distribution of atmospheric methyl halides and other reduced carbon gases in the South Atlantic region,J Geophys. Res., 101, 24,151-24,164. Brasseur, G. P., D. A Hauglustaine, S. Walters, R. 1. Rasch, 1.-F. Miiller, C. Granier, and X. X. Tie (1998), MOZART, a global chemical transport model for ozone and related chemical tracers 1: Model description, J Geophys. Res., 103(D2l), 28,265-28,289. Browell, E. V., G. L. GregOlY, R. C. Harriss, and V. W. 1. H. Kirchhoff (1990), Ozone and aerosol distributions over the Amazon Basin during the wet season, J. Geophys. Res., 95(DlO), 16,88716,901. Butler, T. M., D. Tarabonelli, C. Bruehl, H. Fischer, H. Harder, M. Martinez, 1. Williams, M. G. Lawrence, and 1. Lelieveld (2008), Improved simulation of isoprene oxidation chemistry with the ECHAM5/MESSy chemistry-climate model: Lessons from the
227
GABRIEL airborne field campaign, A/mos. Chem. Phys., 8, Dickerson, R. R., S. Kondragunta, G. Stenchikov, K. L. Civerolo, 4529-4546. B. G. Doddridge, and B. N. Holben (1997), The impact ofaeroCamara, G., A. P. D. M.1. Escada, S. Amaral, T. Carneiro, sols on solar ultraviolet radiation and photochemical smog, SciA. M. Monteiro, ~)iAralljo, 1. Vieira, and B. Becker (2005), Amence, 278, 827-830. azonian defores~fion models, Science, 307(5712), 1043-1044. Duncan, B. N., R. V. Martin, A C. Staudt, R. Yevich, and J. A Carvalho, J. A, Ii'., N. Higuchi, T. M. Aralljo, and 1. C. Santos Logan (2003), Intermillual and seasonal variability of biomass (1998), Combustion completeness in a rainforest clearing experburning emissions constrained by satellite observations, J Geoiment in Manaus, Brazil, J Geophys. Res., 103(D11), 13,195phys. Res., 108(02),4100, doi:10.1029/2002JD002378. 13,199. Echalar, F., P. Artaxo, 1. V. Martins, M. A. Yamasoe, F. Gerab, Carvalho,1. A., Jr., F. S. Costa, C. A. Gurgel Veras, D. V. Sandberg, W. Maenhaut, and B. Holben (1998), Long-tennmonitoring of E. C. Alvarado, R. Gielow, A M. Serra Jr., and 1. C. Santos atmospheric aerosols in the Amazon Basin: Somce identification (2001), Biomass fire consumption and carbon release rates of and apportionment, J Geophys. Res., 103, 3 I,849-3 I ,864. rainforest-clearing experiments conducted in northern Mato Ennnons, L. K., et a1. (2004), Validation of Measmements of PolGrosso, Brazil, J Geophys. Res., 106(D16), 17,877-17,887. 1ution in the Troposphere (MOPITT) CO retrievals with aircraft Castro, T., S. Madronich, S. Rivale, A Muhlia, and B. Mar (2001), in situ profiles, J Geophys. Res., 109, D03309, doi:1O.10291 The influence of aerosols on photochemical smog in Mexico 2003JD0041 01. City, Atmos. Environ., 35, 1765-1772. Fast, 1. D., W. 1. Gustafson Jr., R. C. Easter, R. A Zaveri, Chand, D., P. Guyon, P. Artaxo, O. Schmid, G. P. Frank, L. V. Rizzo, 1. C. Barnard, E. G. Chapman, G. A Grell, and S. E. Peckham O. L. Mayol-Bracero, L. V. Gatti, and M. O. Andreae (2006), Opti(2006), Evolution of ozone, particulates, and aerosol direct racal and physical propeliies of aerosols in the bonndary layer and diative forcing in the vicinity of Houston using a fully coupled fi'ee h'oposphere over the Amazon Basin dming the biomass burnmeteorology-chemistly-aerosol model, J Geophys. Res., 111, ing season, Atmos. Chem. Phys., 6, 2911-2925. D2l305, doi:10.1029/2005JD006721. Chatfield, R. B., and P. 1. Crutzen (1984), Sulfur dioxide in remote Fearnside, P. M. (1990), Fire in the tropical rain forest ofthe Amaoceanic air: Cloud transport of reactive precursors, J< Geophys. zon basin, in Fire in the Tropical Biota: Ecosystem Processes Res., 89(D5), 7111-7132. and Global Challenges, edited by 1. G. Goldammer, pp. 106Chatfield, R. B., 1. A Vastano, H. B. Singh, and G. Sachse (1996), 116, Springer, Berlin. A general model of how fire emissions and chemistry produce Fearnside, P. M., N. ],-eal Jr., and F. M. Fernandes (1993), RainAfrican/oceanic plumes (0 3 , CO, PAN, smoke), J Geophys. forest burning and the global budget: Biomass, combustion efRes., 101(019), 24,279-24,306. ficiency, and charc;oal formation in the Brazilian Amazon, J Chin, M., R. B. Rood, S.-1. Lin, J.-F. Miiller, and A. M. Thompson Atmos. Chem., 98, '733-743. (2000), Atmospheric sulfur cycle simulated in the global model Feingold, G., L. A Remer, 1. Ramaprasad, and Y. 1. Kaufman GOCART: Model description and global propeliies, J Geophys. (2001), Analysis of smoke impact on clouds in Brazilian bioRes., 105, 24,671-24,687. mass burning regions: An extension of Twomey's approach, J Christian, T., B. Kleiss, R. 1. Yokelson, R. Holzinger, P. 1. Crutzen, Geophys. Res., 106,22,907-22,922. W. M. Hao, B. H. Saharjo, and D. E. Ward (2003), Comprehen- Feingold, G., H. Jiang, and 1. Y. Harrington (2005), On smoke sive laboratOly measurements of biomass-burning emissions: 1. suppression of clouds in Amazonia, Geophys. Res. Lett., 32, Emissions from Indonesian, African, and other fuels, J Geophys. L02804, doi: 10.1 029/2004GL021369. Res., 108(023),4719, doi:1O.1029/2003JD003704. Ferek, R. J., 1. S. Reid, P. V. Hobbs, D. R. Blake, and C. Liousse Christian, T. 1., R. J. Yokelson, 1. A. Carvalho Jr., D. W. T. Grif(1998), Emission factors of hydrocarbons, halocarbons, trace fith, E. C. Alvarado, 1. C. Santos, T. G. S. Neto, C. A. G. Veras, gases, and particles fi'om biomass burning in Brazil, J Geophys. and W. M. Hao (2007), The tropical forest and fire emissions Res., 103(024), 32,107-32,118. experiment: Trace gases emitted by smoldering logs and dung Fernandes, W. A, 1. R. C. A Pinto, O. Pinto Jr., K. M. Longo, and from deforestation and pasture fires in Brazil, J Geophys. Res., S. R. Freitas (2006), New findings about the influence of smoke 112, D18308, doi: 10.l02912006JD008147. from fires on the cloud-to-ground lightning characteristics in the Coutinho, L. M. (1990), Fire in the ecology of the Brazilian cerAmazon region, Geophys. Res. Lett., 33, L20810, doi: 10. 10291 rado, in Fire in the Tropical Biota: Ecosystem Processes and 2006GL027744. Global Challenges, edited by J. G. Goldammer, pp. 82-105, Fishman, 1., V. G. Brackett, E. V. Browell, and W. B. Grant (1996), Springer, Berlin. Tropospheric ozone derived from TOMS/SBVV measurements Crutzen, P. 1. (1995), Overview of tropospheric chemishy: Develduring TRACE A, J Geophys. Res., 101,24,069-24,082. opments during the past quarter century and a look ahead, Fara- Freitas, S. R., M. A F. S. Dias, P. L. S. Dias, K. M. Longo, day Discuss., 100, 1-21. P. Aliaxo, M. O. Andreae, and H. Fischer (2000), A convective Crutzen, P. 1., and M. O. Andreae (1990), Biomass burning in the kinematic trajectOly technique for low-resolution atmospheric tropics: Impact on atmospheric chemistJy and biogeochemical models, J Geophys. Res., 105(D19), 24,375-24,386. cycles, Science, 250, 1669-1678. Freitas, S. R., K. M. Longo, M. Silva Dias, P. Silva Dias, R. Dickerson, R. R., et a1. (1987), Thunderstorms: An important mechChatfield, E. Prins, P. Artaxo, G. Grell, and F. Recuero (2005), anism in the transport of air pollutants, Science, 235, 460-465. Monitoring the transport of biomass burning emissions in South
228
LONGO ET AL.
BIOMASS BURNING IN AMAZONIA
America, Environ. Fluid Mech., 5(1-2),135-167, doi:1O.1007/ s 10652-005-0243-7. Freitas, S. R., K. M. Longo, and M. O. Andreae (2006), Impact of including the plume rise of vegetation fires in numerical simulations of associated atmospheric pollutants, Geophys. Res. Lett., 33, Ll7808, doi: 10.1 029/2006GL026608. Freitas, S. R., K. M. Longo, R. Chatfield, D. Latham, M. A F. Silva Dias, MO. Andreae, E. Prins, J. C. Santos, R. Gielow, and J. A Carvalho Jr. (2007), Including the sub-grid scale plume rise of vegetation fires in low resolutiolf atmospheric transport models, At/nos. Chem. Phys., 7,3385-3398. Freitas, S. R., et al. (2009), The Coupled Aerosol and Tracer Transport model to the Brazilian developments on the Regional Atmospheric Modeling System. Part I: Model description and evaluation, Atmos. Chem. Phys., 9, 2843-2861. Fromm, M. D., and R. Servranckx (2003), Transport of forest fire smoke above the tropopause by supercell convection, Geophys. Res. Lett., 30(10),1542, doi:10.1029/2002GL016820. Fromm, M., J. Alfred, K. Hoppel, J. Hornstein, R. Bevilacqua, E. Shettle, R. Servranckx, Z. Li, and B. Stocks (2000), Observations of boreal forest fire smoke in the stratosphere by POAM III, SAGE II, and lidar in 1998, Geophys. Res. Lett., 27, 14071410. Fuzzi, S., et al. (2007), Overview ofthe inorganic and organic composition of size-segregated aerosol in Rondonia, Brazil, from the biomass burning period to the onset of the wet season, J. Geophys. Res., 112, D01201, doi:1 0.1029/2005JD006741. Gevaerd, R.; S. R. Freitas, and K. M. Longo (2006), Numerical simulation of biomass burning emission and transportation during 1998 Roraima fires [CD-ROM], in Proceedings of 8th International Conference on Southem Hemisphere Meteorology and Oceanography (ICSHMO), Foz do Iguar;u, Brazil, 24-28 April, pp. 883-889, INPE, Sao Jose dos Campos, ISBN 85-17-00023-4. Giglio, L., G. R. van del' Werf, J. T. Randerson, G. J. Collatz, and P. Kasibhatla (2006), Global estimation of burned area using MODIS active fire observations, Atmos. Chem. Phys., 6, 957974. Greenberg, J. P., P. R. Zimmerman, L. Heidt, and W. Pollock (1984), Hydrocarbon and carbon monoxide emissions from biomass burning in Brazil, J. Geophys. Res., 89, 1~5O-1354. Grell, G., S. Emeis, W. Stockwell, T. Schoenemeyer, R. Forkel, J. Michalakes, R. Knoche, and W. Seidl (2000), Application of a multiscale, coupled MM5/chemistry model to the complex terrain of the VOTALP valley campaign, Atmos. Environ., 34, 1435-1453. Grell, G., S. Peckham, R. Schmitz, S. McKeen, G. Frost, W. Skamarock, and B. Eder (2005), Fully coupled "online" chemistry within the WRF model, Atmos. Environ., 39(37), 6957-6975. Guild, L. S., J. B. Kaufmann, L. J. Ellingson, D. L. Cummings, E. A Castro, R. E. Babbitt, and D. E. Ward (1998), Dynamics associated with total above ground biomass, C, nutrient pools, and biomass burning of primary forest and pasture in Rondonia, Brazil during SCAR-B, J. Geophys. Res., 103, 32,091-32,100. Gut, A, et al. (2002), Exchange fluxes of N02 and 0 3 at soil and leaf surfaces in an Amazonian rain forest, J. Geophys. Res., 107(020), 8060, doi: 10.1 029/200lJD000654.
Guyon, P., et al. (2005), Airborne measurements of trace gases and aerosol particle emissions from biomass bmning in Amazonia, Atmos. Chem. Phys., 5, 2989-3002. Hao, W. M., and M.-H. Liu (1994), Spatial and temporal distribution of tropical biomass burning, Global Biogeochem. Cycles, 8(4),495-503. Hobbs, P. V., J. S. Reid, R. A Kotchenruther, R. J. Ferek, and R. Weiss (1997), Direct radiative forcing by smoke from biomass burning, Science, 275,1777-1778. Hobbs, P. V., P. Sinha, R. J. Yokelson, T. J. Christian, D. R. Blake, S. Gao, T. W. Kirchstetter, T. Novakov, and P. Pilewskie (2003), Evolution of gases and particles from a savanna fire in South Africa, J. Geophys. Res., 108(Dl3), 8485, doi:10.1029/ 2002JD002352. Hoelzemam1, J. J. (2007), Global Wildland Fires Impact on Atmospheric Chem istl)l, 200 pp., VDM-Verlag Dr. Mi.i11er, Saarbruecken, Germany. Hoelzemann, J. J., M. G. Schultz, G. P. Brasseur, C. Granier, andM. Simon (2004), Global Wildland Fire Emission Model (GWEM): Evaluating the use of global area burnt satellite data, J. Geophys. Res., 109, Dl4S04, doi:10. 1029/2003JD003666. Hoelzemann, J. J., K. M. Longo, R. M. Fonseca, N. M. E. do Rosario, H. Elbem, S. R. Freitas, and C. Pires (2009), Regional representativity of AERONET observation sites during the biomass burning season in South America determined by couelation studies with MODIS Aerosol Optical Depth, J. Geophys. Res., 114, D13301, doi:10.102912008JD010369. Horowitz, L. W., et al. (2003), A global simulation of tropospheric ozone and related tracers: Description and evaluation of MOZART, version 2, J. Geophys. Res., 108(D24), 4784, doi:10.1029/ 2002JD002853. Ichoku, C., and Y. J. Kaufman (2005), A method to derive smoke emission rates from MODIS fire radiative energy measurements, IEEE Trans. Geosci. Remote Sens., 43(11), 2636-2649. Igpotti, E., S. Hacon, A M. Silva, W. L. Junger, and H. A Castro (2007), Effects of biomass burning in Amazon: Method to select municipalities using health indicators, Rev. Bras. Epidemiol., 10, 453-464. Ignotti, E., 1. Valente, S. Hacon, K. M. Longo, S. R. Freitas, and P. Artaxo (2009), Impacts of pmiiculate matter (PM2.s) emitted from biomass burning in the Amazon regarding hospital admissions by respiratory diseases: Building up enviromnental indicators anel a new methodological approach (online), Rev. Saude Publica, in press. Instituto Nacional de Pesquisas Espaciais (INPE) (2008), Monitoramento da cobertura florestal da Amazonia pOl' satelites Sistemas PRODES, DETER, DEGRAD e QUEIMADAS 2007-2008, report, Coord. Geral de Obs. da Terra, Minist. da Ciencia e Tecnol., Sao Jose dos Campos, Brazil. (Available at http://www.obt. inpe.briprodeslRelatorio]rodes2008.pdf). Jacob, D. J., and S. C. Wofsy (1990), Budgets ofreactive nitrogen, hydrocarbons, and ozone over the Amazon forest during the wet season, J. Geophys. Res., 95, 16,737-16,754. Jost, H.-J., et al. (2004), In-situ observations of mid-latitude forest fire plumes deep in the stratosphere, Geophys. Res. Lett., 31, LlI101, doi:10.1029/2003GLOI9253.
229
Karl, T. G., T. J. Clu'istian, R. .T. Yokelson, P. Artaxo, W. M. Hao, (2002), Exchange of short-chain monocarboxyclic organic acand A Guenther (2007), The tropical forest and fire emissions ids by vegetation at a remote tropical forest site in Amazonia, J. experiment: MethOd/evaluation of volatile organic compound Geophys. Res., 107(D20), 8069, doi: 10.1 029/2000JD000303. emissions measur~d by PTR-MS, FTIR, and GC from tropical Langenfelds, R. L., R. J. Francey, B. C. Pak, L. P. Steele, J. Lloyd, biomass burning,;/Atmos, Chem. Phys., 7,5883-5897. C. M. Trudinger, and C. E. Allison (2002), Interannual growth Kauffman, J. B., n. L. Cummings, and D. E. Ward (1994), Relarate variations of atmospheric CO2 and its o13 C, H 2, CH4 , and tionships of fire, biomass and nutrient dynamics along a vegetaCO between 1992 and 1999 linked to biomass burning, Global tion gradient in the Brazilian cerrado, J. Ecol., 82, 519-531. Biogeochem. Cycles, 16(3), 1048, doi: 10.1029/200IGB001466. Kauffman, J. B., D.L. Cummings, and D.E. Ward (1998), Fire Lanser, D., and J. G. Verwer (1998), Analysis of Operator Splitting in the Brazilian Amazon 2. Biomass, nutrient pools and losses for Advection-Diffusion-Reaction Problems from Air Pollution in cattle pastures, Oecologia, 113, 415-427, doi:1O.1007/ Modeling, CWI Report MAS-R9805. s004420050394. Lavoue, D., C. Liousse, H. Cachier, B. J. Stocks, and J. G. Kaufman, Y., and 1. Koren (2006), Smoke and pollution aerosol Goldammer (2000), Modeling of carbonaceous particles emitted effect on cloud cover, Science, 313, 655-658, doi:1O.1126/sciby boreal and temperate wildfires at northern latitudes, J. Geoence.1126232. phys. Res., 105,26,871-26,890. Kaufman, Y., C. Ichoku, L. Giglio, S. Korontzi, D. A. Chu, W. M. Lin, J. C., T. Matsui, R. A Pielke Sr., and C. Kummerow (2006), Hao, R.-R. Li, and C. O. Justice (2003) Fires and smoke obEffects of biomass burning-derived aerosols on precipitation and served from the Earth Observing System MODIS instrument: clouds in the Amazon basin: a satellite-based empirical study, J. Products, validation, and operational use, Int. J. Remote Sens., Geophys. Res., Ill, D19204, doi:10.1029/2005JD006884. 24,1765-1781. Longo, K. M., A. M. Thompson, V. W. J. H. Kirchhoff, L. A. ReKaufman, Y. J., and T. Nakajima (1993), Effect of Amazon smoke mer, S. R. de Freitas, M. A F. S. Dias, P. Ariaxo, W. Hart, J. D. on cloud microphysics and albedo--Analysis from satellite imSpinhime, and M. A Yamasoe (1999), Correlation between agery, J. Appl. Meteorol., 32, 729-744. smoke and tropospheric ozone concentration in Cuiaba during Kaufman, Y. J., et al. (1998), Smoke, Clouds, and Radiation-Brazil Smoke, Clouds, and Radiation-Brazil (SCAR-B), J. Geophys. (SCAR-B) experiment, J. Geophys. Res., 103(D24), 31,783Res., 104,12,113-12,129. 31,808. Longo, K. M., S. R. Freitas, M. Silva Dias, and P. Silva Dias (2006), Kesselmeier, J., et al. (2002), Concentrations and species composiNumerical modeling ofthe biomass-buming aerosol direct radition of atmospheric volatile organic compounds (VOCs) as obative effects on the 'thermodynamics structure of the atmosphere served during the wet and dry season in Rondonia (Amazonia), and convective pre~ipitation [CD-ROM], in Proceedings of 8th J. Geophys. Res., 107(D20), 8053, doi:l0.1029/2000JD000267. Intemational COl1r~rence on Southem Hemisphere Meteorology Kesselmeier, J., A. Guenther, T. Hoffmann, M. T. Piedade, and and Oceanography (ICSHMO), Foz do Iguar;u, Brasil, pp. 283J. Warnke (2009), Natural volatile organic compound emissions 289, INPE, Sao Jose dos Campos, ISBN 85-17-00023-4. from plants and their roles in oxidant balance and particle formaLongo, K. M., S. R. Freitas, A Setzer, E. Prins, P. Ariaxo, and M. O. tion, Geophys. Monogr. Ser., doi:1O.1029/2008GM000717, this Andreae (2007), The Coupled Aerosol and Tracer Transport volume. model to the Brazilian developments on the Regional AtmosKirchhoff, V. W. J. H., A W. Setzer, and M. C. Pereira (1989), pheric Modeling System. Part 2: Model sensitivity to the bioBiomass burning in Amazonia: Seasonal effects on atmospheric mass burning inventories, Atl/1os. Chem. Phys. Disc., 7, 85710 3 and CO, Geophys. Res. Lett., 16,469-472. 8595. Kirchhoff, V. W. J. H., 1. M. O. da Silva, and E. V. Browell (1990), Lyons, W. A, T. E. Nelson, E. R. Williams, J. A Cramer, and Ozone measurements in Amazonia: Dry season versus wet seaT. R. Turner (1998), Enhanced positive cloud-to-ground lightson, J. Geophys. Res., 95,16,913-16,926. ning in thunderstorms ingesting smoke from fires, Science, 282, Kirkman, G. A, A Gut, C. Ammann, L. V. Gatti, A M. Cordova, 77-80. M. A. L. Moura, M. O. Andreae, and F. X. Meixner (2002), Sur- Marengo, J. A, A Cornejo, P. SatyamUliy, C. A Nobre, and W. Sea face exchange of nitric oxide, nitrogen dioxide, and ozone at a (1997a), Cold waves in the South American continent: The strong cattle pasture in Rondonia, Brazil, J. Geophys. Res., 107(D20), event of June 1994, Mon. Weather Rev., 125, 2759-2786. 8083, doi:10. 1029/2001JD000523. Marengo, J. A, C. A Nobre, and A D. Culf(1997b), Climatic imKley, D. (1997), Tropospheric chemistry and transport, Science, pacts of the "friagens" in the Amazon region, J. Appl. NJeteorol., 276, 1043-1045. 36, 1553-1566. Koren, 1., Y. J. Kaufman, L. A Remer, and J. V. Martins (2004), Marengo J. A, C. A Nobre, J. Tomasella, M. Oyama, G. Sampaio, Measurement of the effect of Amazon smoke on inhibition of H. Camargo, L. M. Alves, R. de Oliveira (2008), The drought of cloud formation, Science, 303, 1342-1345. Amazonia in 2005, J. Clim., 21, 495-516. Koren, 1., J. V. Martins, L. A Remer, and H. Afargan (2008), Marengo, J., C. A Nobre, R. A Betts, P. M. Cox, G. Sampaio, and Smoke invigoration versus inhibition of clouds over the AmaL. Salazar (2009), Global warming and climate change in Amazon, Science, 321,946-949. zonia: Climate-vegetation feedback and impacts on water reKuhn, U., S. Rottenberger, T. Biesenthal, C. Ammann, A Wolf, sources, Geophys. Monogr. Ser., doi:10.1029/2008GM000743, G. Schebeske, S. T. Oliva, T. M. Tavares, and J. Kesselmeier this volume.
230
LONGO ET AL.
BIOMASS BURNING IN AMAZONIA
Martins, J. A., M. A. F. Silva Dias, and F. L. T. Goncalves (2009), Impact of biomass burning aerosols on precipitation in the Amazon: A modeling case study, J Geophys. Res., 114, D02207, doi: 10.1U£9/2007JD009587. Mason, S. A., R. 1. Field, R. 1. Yokelson, M. A. Kochivar, M. R. Tinsley, D. E. Ward, and W. M. Hao (2001), Complex effects arising in smoke plume simulations due to inclusion of direct emissions of oxygenated organic species from biomass combustion,.!. Geophys. Res., 106(DI2), 12,527-12,539. Mason, S. A., 1. Trentmann, T. Winterrath, R. 1. Yokelson, T. 1. Christian, L. 1. Carlson, T. R.'Warner, L. C. Wolfe, and M. O. Andreae (2006), Intercomparison of two box models of the chemical evolution in biomass-blU'ning smoke plumes, J Atmos. Chem., 55, 273-297, doi:10.1007/sI0874-006-9039-5. Mauzerall, D. L., 1. A. Logan, D. 1. Jacob, B. E. Anderson, D. R. Blake, 1. D. Bradshaw, B. Heikes, G. W. Sachse, H. Singh, and B. Talbot (1998), Photochemistty in biomass burning plumes and implications for tropospheric ozone over the tropical South Atlantic, J Geophys. Res., 103, 8401-8423. McMillan, W. W., C. Barnet, L. Strow, M. T. Chahine, M. L. McCourt, J. X. Warner, P. C. Novelli, S. Korontzi, E. S. Maddy, and S. Datta (2005), Daily global maps of carbon monoxide from NASA's Atmospheric Infrared Sounder, Geophys. Res. Lett., 32, U1801, doi: 10. 1029/2004GL02 I 821. Morton, D. C., R. S. Defries, Y. E. Shimabukuro, O. L. Anderson, F. del B. Espirito Santo, M. Hansen, and M. Carroll (2005), Rapid assessment ofmmual deforestation in the Brazilian Amazon using MODIS data, Earth Interact., 9(8), EIl39, doi: 10.1175/EIl39.1. Morton, D. C., R. S. DeFries, Y. E. Shimabukuro, L. O. Anderson, E. Arai, F. Espirito-Santo, R. Freitas, and J. Morisette (2006), Cropland expansion changes deforestation dynamics in the southern Brazilian Amazon, Proc. Natl. Acad. Sci. U S. A., 103(39), 14,637-14,641. Nobel', F. J., H.-F. Graf, and D. Rosenfeld (2003), Sensitivity ofthe global circulation to the suppression of precipitation by anthropogenic aerosols, Global Planet. Change, 37, 57-80. Nobre, C. A., P. 1. Sellers, and 1. Shukla (1991), Amazonian deforestation and regional climate change, J Clim., 4, 957-988. Nobre, C. A., G. O. Obregon, 1. A. Marengo, R. Fu, and G. Poveda (2009), Characteristics ofAmazonian climate: Main features, Geophys. Monogr. Ser., doi:IO.102912008GM000720, this volume. Oliveira, P. H. F., P. Artaxo, C. Pires, S. De Lucca, A. Procopio, B. Holben, 1. Schafer, L. F. Cardoso, S. C. Wofsy, and H. R. Rocha (2007), The effects of biomass burning aerosols and clouds on the CO 2 flux in Amazonia, Tellus, Ser. B, 59, 338-349, doi:IO.1IIl/j.1600-0889.2007.00270.x. Olivier, 1., A. Bouwman, J. Berdowski, 1. Bloos, A. Visschedijk, C. van der Mass, and P. Zandveld, (1999), Sectoral emission inventories of greenhouse gases for 1990 on a per country basis as well as on 1 x 1 degrees, Environ. Sci. Policy, 2, 241263. Pereira, G. (2008), 0 uso de satelites ambientais para a estimativa dos fluxos de gases tra<;os e de aerossois liberados na queima de biomassa e sua assimila<;ao emmodelos numericos de qualidade do aI', Master thesis in Remote Sensing-Instituto Nacional de Pesquisas Espaciais (INPE), Sao Jose dos Campos, Brazil.
(Available at http://urlib.netlsid.inpe.br/mtc-mI7@80/2008/02. 13.16.15). Pickering, K. E., R. R. Dickerson, G. J. Huffman, J. F. Boatman, and A. Schanot (1988), Trace gas transport in the vicinity of frontal convective clouds, J Geophys. Res., 93(0 I), 759773. Prins, E. M., J. M. Feltz, W. P. Menzel, and D. E. Ward (1998), An overview of GOES-8 diurnal fire and smoke results for SCARBand 1995 fire season in South America, J Geophys. Res., 103(024),31,821-31,835. Procopio, A. S., L. A. Remer, P. Artaxo, Y. 1. Kaufman, and B. N. Holben (2003), Modeled spectral optical properties for smoke aerosols in Amazonia, Geophys. Res. Lett., 30(24), 2265, doi: 10.1 029/2003GLO 18063. Procopio, A. S., P. Artaxo, Y. 1. Kaufman, L. A. Remer, 1. S. Schafer, and B. N. Holben (2004), Multiyear analysis of amazonian biomass buming smoke radiative forcing of climate, Geophys. Res. Lett., 31, L03108, doi:l0.1029/2003GLOI8646. Radke, L. F., D. A. Hegg, P. V. Hobbs, 1. D. Nance, J. H. Lyons, K. K. Laursen, R. E. Weiss, P. 1. Riggan, andD. E. Ward (1991), Particulate and trace gas emissions from large biomass fires in North America, in Global Biomass Burning, edited by 1. Levine, MIT Press, Cambridge, Mass. Richardson, 1. L., 1. Fishman, and G. L. GregOly (1991), Ozone budget over the Amazon: Regional effects from biomass-burning emissions, J Geophys. Res., 96, 13,073-13,087. Riggan, P. 1., R. G. Tissell, R. N. Lockwood, 1. A. Brass, J. A. R. Pereira, H. S. Miranda, A. C. Miranda, T. Campos, and R. Higgins (2004), Remote measurement of energy and carbon flux from wildfires in Brazil, Ecol. Appl., 14, 855-872. Rissler, J., A. Vestin, E. Swietlicki, G. Fisch, 1. Zhou, P. Artaxo, and M. O. Andreae (2006), Size distribution and hygroscopic properties of aerosol particles from dry-season biomass burning in Amazonia, Atmos. Chem. Phys., 6, 471--491. Roberts, G. C., A. Nenes, 1. H. Seinfeld, and M. O. Andreae (2003), Impact of biomass bmning on cloud properties in the Amazon Basin, J Geophys. Res., 108(D2), 4062, doi:10.10291 2001JD000985. Robock, A. (1991), Surface cooling due to forest fire smoke, J Geophys. Res., 96, 20,869-20,878. Rosenfeld, D., Y. 1. Kaufman, and 1. Koren (2006), Switching cloud cover and dynamical regimes from open to closed Benard cells iil response to the suppression of precipitation by aerosols, Atmos. Chem. Phys., 6,2503-2511. Rosenfeld, D., U. Lohmann, G. B. Raga, C. D. O'Dowd, M. Kulmala, S. Fuzzi, A. Reissell, and M. O. Andreae (2008), Flood or drought: How do aerosols affect precipitation?, Science, 321, 1309-1313. Ross, 1. L., P. V. Hobbs, and B. Holben (1998), Radiative characteristics of regional hazes dominated by smoke from biomass buming in Brazil: Closure tests and direct radiative forcing, J Geophys. Res., 103, 31,925-31,941. Rummel, U., C. Ammann, A. Gut, F. X. Meixner, and M. O. Andreae (2002), Eddy covariance measurements of nitric oxide flux within an Amazonian rain forest, J Geophys. Res., 107(020), 8050, doi: 10.1029/200IJD000520.
Rmrullel, u., C. Arrunann, G: A. Kirkman, M. A. L. Moura, T. Foken, M. O. Andreae, agd F. X. Meixner (2007), Seasonal variation of ozone deposit.ion to a tt'opical rainforest in southwest Amazonia, Atmos. CIJ~m. Phys., 7,5415-5435. Satyammty, P., C. No~re, and P. Silva Dias (1998), South America, in Meteorology ojlhe Southern Hemisphere, edited by D. Karoly and D. Vincent, Meteorol. Monogr., 27(49),119-139. Schafer, J. S., B. N. Holben, T. F. Eck, M. A. Yamasoe, and P. Artaxo (2002a), Atmospheric effects on insolation in the Brazilian Amazon: Observed modification of solar radiation by clouds and smoke and derived single scattering albedo of fire aerosols, J Geophys. Res., 107(D20), 8074, doi:l0.1029/2001JD000428. Schafer, J. S., T. F. Eck, B. N. Holben, P. Artaxo, M. A. Yamasoe, and A. S. Procopio (2002b), Observed reductions of total solar irradiance by biomass-burning aerosols in the Brazilian Amazon and Zambian Savanna, Geophys. Res. Lett., 29(17), 1823, doi: 10.1029/200 IGL014309. Schafer, J. S., T. F. Eck, B. N. Holben, P. Artaxo, and A. F. Duarte (2008), Characterization of the optical properties of atmospheric aerosols in Amazonia from long-term AERONET monitoring (1993-1995 and 1999-2006), J Geophys. Res., 113, 004204, doi: 10.1 029/2007JD009319. Schroeder, W., A. Alencar, E. Arima, and A. Setzer (2009), The spatial distribution and interannual variability of fire in Amazonia, Geophys. Monogr. Ser., doi:IO.1029/2008GM000724, this volume. Seinfeld, 1. H., and S. N. Pandis (1998), Atmospheric Chemisliy and Physics: From Air Pollution to Climate Change, John Wiley, New York. Setzer, A. W., and M. C. Pereira (1991), The Operational detection of fires in Brazil with NOAA-AVHRR, 24th Intemational Symposium on Remote Sensing of the Environment, Rio de Janeiro, Brazil, pp. 76-77. Setzer, A. W., and J. P. Malingreau (1996), AVHRR monitoring of vegetation fires in the tropics: Towards a global product, in Biomass Burning and Global Change, edited by 1. Levine, pp. 25-39, MIT Press, Cambridge, Mass. Silva Dias, M. A. F., et al. (2002), Cloud and rain processes in a biosphere-atmosphere interaction context in the Amazon Region, J. Geophys. Res., 107(D20), 8072, doi:IO.1029/2001JD000335. Smith, A. M. S., and M. 1. Wooster (2005), Remote classification of head and backfire types from MODIS fire radiative power and smoke plume observations, Int. J Wildland Fire, 14, 249-254. Stull, R. B. (1988), An/ntroduction to Boundary Layer Meteorology, Springer, Dordrecht. Talbot, R. W., M. O. Andreae, T. W. Andreae, and R. C. Han'iss (1988), Regional aerosol chemistly ofthe Amazon Basin during the dry season, J Geophys. Res., 93, 1499-1508. Thompson, A. M., et al. (1996), Ozone over southern Afi'ica during SAFARI-92/TRACE A, J. Geophys. Res., 101, 23,79323,807. Thompson, A. M., 1. C. Witte, R. D. Hudson, H. Guo, 1. R. Herman, and M. Fujiwara (2001), Tropical tt'opospheric ozone and biomass burning, Science, 291, 2128-2132. Trentmann, 1., R. 1. Yokelson, P. V. Hobbs, T. Winterrath, T. J. Christian, M. O. Andreae, and S. A. Mason (2005), An analysis
231
ofthe chemical processes in the smoke plume fi'om a savanna fire, J Geophys. Res., 110,012301, doi:IO.1029/2004JD005628. Turquety, S., et al. (2007), Inventory of boreal fire emissions for North America in 2004: Importance ofpeat burning and pyroconvective injection, J Geophys. Res., 112, D12S03, doi:10.10291 2006JD007281. Van del' Werf, G. R., 1. T. Randerson, L. Giglio, G. 1. Collatz, and P. S. Kasibhatla (2006), Intermllmal variability in global biomass burning emission fi'om 1997 to 2004, Annos. Chem. Phys., 6, 3423-3441. Vera, C., et al. (2006), The South American Low-LevelJet Experiment, Bull. Am. Meteorol. Soc., 87(1), doi:l0.1175/BAMS-87-1-63. Vestin, A., 1. Rissler, E. Swietlicki, G. Frank, and M. O. Andreae (2007), Cloud-nucleating properties of the Amazonian biomass burning aerosol: Cloud condensation nuclei measurements and modeling, J Geophys. Res., 112, D14201, doi:IO.10291 2006JD008104. Wang, J., S. A. Christopher, U. S. Nair, 1. S. Reid, E. M. Prins, 1. Szykman, and 1. L. Hand (2006), Mesoscale modeling of Central American smoke transport to the United States: I. "Top-down" assessment of emission strength and diumal variation impacts, J Geophys. Res., 111, D05S17, doi:10. 1029/2005JD006416. Ward, D. E., R. A. Susott, 1. B. Kauffman, R. E. Babbitt, D. L. Cummings, B. Dias, B. N. Holden, Y. 1. Kaufman, R. A. Rasmussen, and A. W. Setzer (1992), Smoke and fire characteristics for Cenado and deforestation burns in Brazil: BASE-B experiment, J Geophys. Res., 97,14,601-14,619. Williams, E., et al. (2002), Contrasting convective regimes over the Amazon: Implications for cloud electt'ification, J Geophys. Res., 107(020), 8082, doi:1 0.1029/200 IJD000380. Yamasoe, M. A., C. vop Randow, A. O. Manzi, 1. S. Schafer, T. F. Eck, and B. N. Holben (2006), Effect of smoke and clouds on the transmissivity of photosynthetically active radiation inside the canopy, Atmos. Chem. Phys., 6, 1645-1656. Yanenko, N. N. (1971), The Method ofFractional Steps: The Solution ofProblems ofMathematical Physics in Several Variables, Springer, New York. Yokelson, R. 1., D. W. T. Griffith, and D. E. Ward (1996), Openpath Fourier transform infi'ared studies of large-scale laboratOlY biomass fires, J Geophys. Res., 101(DI5), 21,067-21,080. Yokelson, R. 1., R. Susott, D. E. Ward, J. Reardon, and D. W. T Griffith (1997), Emissions from smoldering combustion of biomass measured by open-path Fourier transform infi'ared spectroscopY,J Geophys. Res., 102, 18,865-18,877. YOkelSOll, R. J., 1. T. Bertschi, T. 1. Christian, P. V. Hobbs, D. E. Ward, and W. M. Hao (2003), Trace gas measurements in nascent, aged, and cloud-processed smoke from African savanna fires by airborne Fourier transform infrared spectroscopy (AFTIR),J. Geophys. Res., 108(013), 8478, doi: 10.10291 2002JD002322. Yokelson, R. 1., T. G. Karl, P. Artaxo, D. R. Blake, T. 1. Christian, D. W. T. Griffith, A. Guenther, and W. M. Hao (2007), The tropical forest and fire emissions experiment: Overview and airborne fire emission factor measurements, Atmos. Chem. Phys., 7,5175-5196. Yokelson, R. J., T. 1. Christian, T. G. Karl, and A. Guenther (2008), The tropical forest and fire emissions experiment: LaboratOlY fire
232
BIOMASS BURNING IN AMAZONIA
measurements and synthesis of campaign data, Almos. Chem. Phys., 8, 3509-3527. Yokelson, R., et al. (2009), Emissions from biomass burning in the Yucatan, in press, Almos. Chelll. Phys. Disc., 9, 767-835. Zhang, Y., R. Fu, H. Yu, R. E. Dickinson, R. N. Juarez, M. Chin, and H. Wang (2008), A regional climate model study of how biomass burning aerosol impacts land-atmosphere interacti9ns over the Amazon, J. Geophys. Res., 113, D14S15, doi:10.1029/2007JD009449.
M. O. Andreae, Max Planck Institute, P.O. Box 3060, D-55020 Mainz, Germany. ([email protected])
P. N1axo, Institute of Physics, University of Sao Paulo, Rua do Matao, Travessa R. 187, Sao Paulo, SP 05508-900, Brazil. ([email protected]) S. R. Freitas, Center for Weather Forecast and Climate Studies, National Institute for Space Research, Rodovia Presidente Dutra, km 40, Cachoeira Paulista, SP 12630-000, Brazil. (saulo.freitas@ cptec.inpe.br) K. M. Longo, Center for Space and Atmospheric Sciences, National Institute for Space Research, Av. Dos Astronautas, 1758 Jardim da GranjaSao, Jose dos Campos, SP 12227-010, Brazil. ([email protected]) R. Yokelson, Department of Chemistry, University of Montana, Missoula, MT 59812, USA. ([email protected])
Aerosol. Particles in Amazonia: Their Composition, Role in the Radiation Balance, Cloud FOll11ation, and Nutrient Cycles Paulo Artaxo,] Luciana V. Rizzo,] Melina Paixao,] Silvia de Lucca,] Paulo H. Oliveira,] Luciene L. Lara,] Kenia T. Wiedemann,! Meinrat O. Andreae? Brent Holben,3 Joel Schafel} Alexandre L. Correia,3 and Theotonio M. Pauliquevis4 The atmosphere above tropical forests plays a velY active part in the biogeochemical cycles that are critically important for the processes that maintain the ecosystem, including processes involving the vegetation, soil, hydrology, and atmospheric composition. Aerosol particles control key ingredients of the climatic and ecological environment in Amazonia. The radiative balance is strongly influenced by the direct and indirect radiative forcing of aerosol particles. Nutrient cycling is partially controlled by dly and wet deposition of key plant nutrients. It was observed that the aerosol particles that act as cloud condensation nuclei influence cloud fonnation and dynamics, having the potential to change precipitation regimes over Amazonia. The IO-year-long record of aer9s01 optical thickness measurements in Amazonia shows a strongly negative radiative forcing of -37 W m-2 average d over 7 years of dly season measurements in Alta Floresta. There is a strong influence of biomass-buming aerosols on the cloud thicrophysical properties during the dly season. The connections between the amount of aerosol particles and carbon uptake trough photosynthesis highlighted the close connection between forest natural processes and the aerosol loading in the atmosphere. Climate change combined with socioeconomic drivers could alter significantly the emission of trace gases, aerosols, and water vapor fluxes from the forest to the atmosphere. It is a vital task to quickly reduce Amazonian deforestation rates, and to implement solid and long-tenn conservation policies in Amazonia. 1. INTRODUCTION
The composition of the tropical atmosphere is controlled by a variety of processes ranging from emissions to lInstitute of Physics, University of Sao Paulo, Sao Paulo, Brazil. 2Max Planck Institute for Chemistry, Mainz, Getmany. 3NASA Goddard Space Flight Center, Greenbelt, Matyland, USA. 4LBA Central Office, Instituto Nacional de Pesquisas da Amazonia, Manaus, Brazil. Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000778
the processing, transport, and deposition of trace gases and aerosol particles [Andreae el al., 2002]. The understanding of the chemical composition of the tropical atmosphere requires a knowledge of the natural processes related to the production and emission of naturally released chemical species, as well as the anthropogenic emissions associated with land use changes [Andreae and Crulzen, 1997]. Once in the atmosphere, chemical compounds are subject to transport, . especially associated with convection that can transport compounds over large heights and distances. Processes associated with deposition of trace gases and aerosol particles occur in both dry and wet regimes, and the removal processes are quite efficient in tropical areas. Photochemical processing is also important, since the interaction with solar 233
234
ARTAXO ET AL.
AEROSOL PARTICLES IN AMAZONIA
radiation occurs in both the trace gas and particle phases. Cloud processing of trace gases and aerosols is another important process that is not completely understood. During each ofthcme steps, trace gases and aerosol particles strongly influence the Amazonian climate in many different ways [Davidson and Artaxo, 2004; S. T. Martin et aI., Sources and properties of Amazonian aerosol particles, submitted to Reviews ofGeophysics, 2008, hereinafter referred to as Martin et aI., submitted manuscript, 2005]. Aerosol particles influence -the global and regional climate via changes in the radiative balance of the atmosphere as well as through their influence on the hydrological cycle [Andreae et al., 2004, 2005]. Due to its short residence time in the atmosphere and the fact that its concentration varies in space and time, it is not easy to make an accurate quantitative estimation of the aerosol radiative forcing. According to the latest Intergovernmental Panel on Climate Change report [IPCC, 2007], the largest uncertainty in climate forcing arises from a lack of understanding of the effect of aerosols on climate (in particular, the indirect effect, i.e., changes in cloud radiative properties induced by aerosol particles) [Forster et al., 2007]. A better understanding of the role that natural aerosol particles play in climate is also critical to quantitatively assess the environmental changes that regions such as Amazonia are suffering [Andreae, 2007]. Natural biogenic aerosol particles emitted by plants play an important role in nutrient cycling in tropical ecosystems [Artaxo et at., 2005]. Tropical ecosystems maintain a delicate nutrient balance characterized by intense internal recycling and depend on atmospheric input of cettain nutrients to fulfill certain requirements [Davidson and Artaxo, 2004]. Aerosol particles influence the climate in two primary manners: first, the aerosol direct radiative effect, which involves the scattering and absorption of solar radiation by aerosol particles changing the net radiative fluxes in the atmosphere and at the ground; second, the aerosol indirect forcing, which is related with aerosol-induced changes in cloud properties. Clouds are a critical ingredient ofthe radiation balance, and the increase in aerosol particle population changes the number of cloud condensation nuclei (CCN) and clouds properties such as albedo and lifetime. Aerosols also change the atmospheric thermodynamic properties such as the temperature profile and relative humidity variability over large areas; this is generally called the semidirect effect [Rosenfeld et al., 2008]. Significant changes related to human activities are occurring in Amazonia [Nobre et al., 2004] that could have global effects on the carbon balance, the concentrations of greenhouse gases and aerosol particles, and on the oxidizing power of the atmosphere. Amazonia is one of the major direct sources of organic aerosols to the global atmosphere.
The size and elemental composition of aerosol particles are important variables that influence their role as CCN. Although aerosols are efficiently scavenged by precipitation, the long-range transport of only a small fraction of Amazonian aerosols could make a major contribution to the global budget of the free troposphere [Andreae and Crutzen, 1997]. In pristine areas of Amazonia, background primary and secondary biogenic aerosol concentrations are usually vety low [Artaxo et ,a I. , 2002; Zhou et al., 2002; Roberts et al., 2001], comparable to the observed levels in other remote locations ofthe planet. The two main sources ofnatural aerosol particles are the direct emission of primary particles (mostly biogenic), and second, the oxidation ofvolatile organic compounds (VOC) emitted by the vegetation [Guenther et al., 1995; Claeys et al., 2004]. Once in the atmosphere, VOCs are subject to chemical and photochemical h'ansfOlmations that can convert some of them to aerosol particles. Due to this predominantly biogenic character of particle formation, aerosol particles in Amazonia are mostly organic [Graham et al., 2003a, 2003b]. Amazonia has been subject to an intensive process ofland use change in the last 40 years [see Alves et al., this volume; Soares-Filho et al., 2006]. These land use changes are mostly in the region known as the "arc of deforestation," in the southern and eastern parts of the Amazon basin. Large areas are convetted from natural forests to pasture or largescale arable agriculture [see Walker et al., this volume]. The main tool used by farmers to remove biomass is fire [Bowman et al., 2009]. At the beginning of the dty season, the number of large fires detected by remote sensing tools rises to a few thousand per day. Most of biomass burning takes place in the states of Rondonia, Mato Grosso, and Pani, following road building during the 1970s and 1980s. During the months of August to October, a large part of Amazonia and South America is covered with smoke. This heavy smoke covering millions of square kilometers has profound effects on the radiation balance, cloud formation, and the health of the Amazonian population. Aerosols from biomass burning can travel over large distances [Andreae et al., 2001] and influence areas far from the source regions [see Longo et al., this volume]. Biomass-burning aerosols have very complex physical and chemical propetties [Andreae, 1991]. Several experiments as part of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) were dedicated to the study of aerosol particle properties and their effects on the Amazonian ecosystem. A few of these includes the LBA/Cooperative LBA Airborne Regional Experiment (CLAIRE) 1998 and 2001 and LBAJEuropean Studies on Trace gases and Atmospheric Chemistry (EUSTACH) [Andreae et al., 2002]. During the
dry-to-wet season transition in 2002, the LBA/smoke aerosols, clouds, rainfall, gnd climate (SMOCC) campaign was conducted in RonQ0nia during September to November [Andreae et al., 20,04]. The main objective of the experiment was to characteryte the optical, physical, and chemical properties of biomass-burning aerosol particles. A review article by Fuzzi et al. [2007] synthesized LBA/SMOCC results and the complexities 'of aerosol particles emitted through biomass burning in tropical areas [see also Longo et al., this volume]. Recently, the Amazonian Aerosol Characterization Experiment-08 experiment has focused on understanding the natural biogenic aerosol properties (Martin et aI., submitted manuscript, 2008), with a field measurement campaign in Janumy-March 2008 in Manaus. Long-term measurements are essential to provide scientific knowledge on the temporal variability of key atmospheric properties. An important example is the operation of NASA's Aerosol Robotic Network (AERONET) that maintains a network of ground-based Sun photometers in Amazonia [Holben et at., 1998]. The strong change in the radiation fluxes at the surface has important consequences on several aspects of the Amazonian ecosystem functioning. A net surface cooling of around 2-3 0 can be modeled through the result of the aerosol layer, as well as a heating at the levels of2-3lan in the atmosphere. This effect stabilizes the vettical temperature profile, reducing convection and the transport of water vapor to upper levels. The radiation field is also strongly affected, with a reduction of direct solar radiation on the surface, and increase in the diffuse solar radiation reaching the forest. The aim of this chapter is to review the main issues regarding atmospheric chemistly in Amazonia, with emphasis on the role of aerosol patticles in the ecosystem functioning including deposition of nutrients and radiation balance.
235
nisms ofparticle emission are still not well understood. They probably include mechanical abrasion by wind, biological activity of microorganisms on plant surfaces and forest litter, and plant physiological processes such as transpiration and guttation. Fungi, through the release of fungal spores, are a major source of primary biogenic aerosol particles and components, and especially in the humid Amazonian environment, they are expected to be a significant source of aerosol particles [Elbert et al., 2007]. These processes may generate particles containing biogenic-related elements such as Na, Mg, P, S, K, Ca, Zn, and Rb. The transpiration of plants can lead to migration of Ca 2+, soi-, Cl-, K\ Mg 2+, and Na+ to the atmosphere. The biogenic-related elements (Mg, K, P, S, Zn, Rb, and others) are essential to plants; they are present in the fluids circulating in the plant and are released from the leaves to the atmosphere. Table 1 shows the average elemental concentration for aerosol particles in the wet season in Rondonia [Artaxo et al., 2002]. The vety low aerosol concentration under natural conditions is evident from the 2.21 flg m- 3 of fine mode mass in the wet season. In particular, an average of88ng m-3 for sulfur is vety low for any continental region. Soil dust also appears with vety low concentrations, with Fe concentrations of only 34 ng m-3 in the coarse-mode aerosols. These low elymental concentrations contrast with the high aerosol loading during the dty season in Rondonia. The optical properties of aerosols in the wet season show a surprisingly strohg absorption of radiation in the visible wavelengths by the biogenic particles [Schmid et al., 2006; Guyon et al., 2004]. The biogenic aerosols absorb light very efficiently due in part to their morphology and elemental composition, with the presence of humic substances. The absorption efficiency is higher than for biomass-burning particles. This strong absorption effect has important implications in the atmospheric radiative budget and ground2. PROPERTIES OF NATURAL PARTICLES based temperature over large areas of Amazonia [Hoffer et IN AMAZONIA al., 2006; Schmid et al., 2006]. Biogenic volatile organic compounds (BVOC), which folVegetation has long been recognized as an impOltant source lowing reaction with Oll or 0 3, are important precursors to of both primary and secondmy aerosol particles [Artaxo and secondary organic aerosol production in the Amazon basin, Hansson, 1995; Martin et aI., submitted manuscript, 2008]. are emitted from plants during growth, maintenance, decay, Aerosol particles are responsible for the airborne transpOlt and consumption. Tropical forests are the dominant global of phosphorus, calcium, sulfur, nitrogen compounds, and source of atmospheric BVOCs, and the Amazon basin is a other essential nutrients. Only a few studies of natural bio- major contributor [Rasmussen and Khalil, 1988]. The basin genic aerosols from vegetation in tl'opical rain forests have contains on the order of 10 5 plant species, each with a unique been undertaken [Artaxo et al., 1990; Artaxo and Hansson, BVOC emission signature. This high species diversity is cou1995; Echalar et al., 1998; Artaxo, 2001]. Biogenic aerosols' pled with a dramatic ecological complexity and a seasonality consist of many different types ofparticles, including pollen, that is vety different from temperate regions, where BVOC spores, bacteria, algae, protozoa, fungi, fragments of leaves, emissions have been studied more extensively. These factors excrement, and fragments ofinsects. This aerosol component combine to make estimates of BVOC emissions from the is mainly in the coarse size fraction (dp > 2 flm). The mecha- whole of Amazonia an important but challenging task. Prior
236
AEROSOL PARTICLES IN AMAZONIA
ARTAXO ET AL.
Table 1. Average Elemental Concentration for Fine- and Coarse-Mode Aerosol Particles in the Wet Season in Rondonia at the Rebio Jam Large-Scale BiosphereAtmosphere Experiment in Amazonia (LBA) Towera
3. PROPERTIES AND EFFECTS OF BIOMASSBURNING A~ROSOL PARTICLES IN AMAZONIA
Wet Season Aerosol Composition in Rondonia Fine Fraction
Mass TC EC n BCe Mg Al Si P S Cl K Ca Ti Cr Mn Fe Ni Cu Zn Br Sr Zr Pb
Coarse Fraction
Mean
SD
11
2.21 896.7 76.6 276.9
1.39 504.6 59.3 137.0
(28) (11 ) (11) (28)
34.5 42.5 5.6 87.7 6.1 26.8 8.4 3.8
30.9 37.9 1.9 87.2 7.9 26.5 6.1 4.1
(24) (28) (28) (28) (11) (28) (28) (14)
0.56 21.6 0.33 0.42 0.68 1.05 0.16 0.18 0.84
0.36 29.7
(28) (26) (1 ) (27) (23) (1) (10) (6) (18)
0.44 0.72 0.07 0.08 0.92
Mean
SD
removed over the continent [Andreae and Andreae, 1988; Worobiec et al., 2007].
11
3.77
1.32
(28)
29.0 52.5 52.0 23.4 23.5 8.9 48.2 12.2 6.5 2.5 0.87 34.2
23.2 50.8 59.8 9.3 14.8 5.4 14.9 7.5 7.3 1.5 0.57 46.8
(23) (24) (25) (28) (28) (28) (28) (28) (14) (8) (28) (26)
0.32 0.72
0.20 0.39
(28) (28)
0.26 0.77 0.46
0.13 0.78 0.17
(9) (4) (3)
"Mean and standard deviations (SD) are shown; 11 is the number of samples where the detected elements were observed above detection limit. Mass concentrations are expressed in I-Ig m-3 ; equivalent black carbon (BC e) and trace elements concentrations are in ng m-3. Table 1 is adapted from Artaxo et al. [2002].
The very clean atmospheric conditions that predominate in the wet season in Amazonia change strongly in the dly season with large emissions of biomass-burning aerosols [Andreae et al., 2002; Echalar et al., 1995]. Typical aerosol number concentrations for the wet season is around 200-300 particles cm-3. In the dry season, particle number concentrations jump to 10,000-20,000 cm-3 [Artaxo et aI., 2002]. Aerosol mass concentrations (PM lO) for the wet season are typically 10-12 /1g m-3 , while in the dly season, they can reach extremely high values of600 /1g m-3. This large aerosol concenh'ation can lead to important effects on human health, clouds, and radiation balance [Kmifman et al., 1998]. Figure 2 shows the deforestation rate in square kilometers per year from the late 1970s to 2008. Large year-to-year variability is observed, due to climatic and to socioeconomic drivers. Figure 1 shows that deforestation has declined significantly from 2004 to 2007, but an increase in 2008 has changed the recent trend. A deforestation of 10,000 to 20,000 km 2 per year from burning injects huge amounts of particles in the atmosphere [Yokelson et al., 2007, 2008]. Emission factors from primary deforestation fires and pasture maintenance fires in tropical rainforests range from 6 to 25 g kg- I for total PM and 7.5 to 15 g kg- I for PM 2.5 , expressed as mass of emitted primary particles per mass unit of dly fuel. For Amazonia, the esti-
Deforestation
to new studies conducted in the past decade, Amazonian BVOC emission estimates were based on a few measurements conducted by Zimmerman et al. [1988] and Kesselmeier et al. [2000]. However, due to high biodiversity and spatially different radiation and temperature fields, much more extensive work on VOCs and the production of secondary aerosols are needed in Amazonia. A prominent contribution of particles from outside the Amazon basin is that made by Saharan dust. The importance of transatlantic transport of dust was recognized by Prospero et al. [1981] and has been observed in several subsequent measurement campaigns [Swap et al., 1996; Artaxo et al., 1988, 1998; Formenti et al., 2001]. ImpOlied dust occurs at its highest concentrations in those parts of the basin that are north ofthe Inteliropical Convergence Zone (ITCZ). The maximum dust concentrations at the surface are typically reached around March and April, coinciding with the wet season in the central basin. The dust at ground stations
is observed in pulses of high concentrations that last from one to several days. Given the large transport distance from Africa, a significant fraction of the mineral dust depositing in the Amazon basin is submicron. Marine aerosol particles consist largely of primary sea spray particles, which are composed mainly of coarse-mode inorganic salts mixed with lesser amounts of the primary biological material that was partitioned to the ocean's surface [Andreae and Rosenfeld, 2008]. Marine emissions dominate the particle population that enters the Amazon basin with the trade wind flow, being progressively removed by wet and dry deposition as an air mass travels deeper into the basin. Even so, the relative contribution by marine particles to the total Amazonian particle mass concentration remains significant even over the central parts of the basin during the wet season. This can be explained by the large concentrations of marine particles present in the air, as it crosses the coast and the relatively slow rates at which aerosol particles are
35000 30000
mates for the emission rates ofPM 2.5 and PM IO are 8 and 10 Tg a-I, respectively [Yokelson et al., 2008]. One of the consequences of these large areas being burned is the emission oflarge number ofpmiicles and the high aerosol concenh'ations [Artaxo et al., 2002; Andreae and MerIel, 2001; Hoffer etal., 2006]. Table 2 shows the average elemental concenh'ations for aerosol pariicles in the illy season in Rondonia collected at the Rebio Jam LBA tower. Fine-mode sulfur now appears at a high concentration of 533 ng m-3, which is six times higher than the values measured in the wet season, a similar increase as black carbon. Fine-mode potassium concentration increases by 18 times. This large change in trace element concentrations affects the biogeochemical cycles of several key nutrients such as phosphoms. The Amazonian atmosphere is not isolated from the global atmosphere and interacts strongly with the nearby continents and oceans [Andreae et al., 2001]. Swap et al. [1996] showed that particles originating in the Sahara desert reach Amazonia and can be impOliant in terms of nutrient cycling over large time spans. Formenti el al. [2001] also observed Saharan dust particles over the northern part of Amazonia and in Suriname, indicating that this process is more important than previously thought. The biomass-burning emissions of soluble iron also have important implications in the South Atlantic primary productipn, since Fe is critically important to the ocean biogeochemistry and is emitted in large amounts by biomass burning [L~to et al., 2008]. MercUlY emissions from biomass burning can be significant, as was observed in the study of Artaxo et al. [2000], with airborne measurements of Hg and black carbon showing a strong relationship. It is
in Amazonia 1977-2008 in km 2 per year
I"'-------------------~_~
~
_
----~-------~---------------~---~-~---~---
"i:'
~ ~
25000
1
20000
'"a.
~
'"c:
o
15000
<8
10000
I '"
o
5000 0
10 ~ "-
'"
~
'" '"
0
~
'"
:3 '"
'"~
;;;
~
'"
'"
~
;J;
'" ~ '"'"
"-
;);
'"
237
'" '"
~
'" '"
~
0
~
'"
~
0
'"
~
::J
8
"
~
::J
~
'" ~
0
"-
~
0
'"
~
0
Figure 1. Deforestation rate in the Brazilian Amazonia in km 2 a-I from late 1970s to 2008. It is possible to observe a decrease in deforestation rate from 2003 to 2007, followed by an increase in 2008.
238
AEROSOL PARTICLES IN AMAZONIA Table 2. Average Elemental Concentration for Aerosol Particles in the Dry Season in Rondonia at the Rebio JalU LBA Tower" D1Y Season Aerosol Composition in Rondonia
zr
Coarse Fraction
Fine Fraction Mean Mass TC EC" BC e Al Si P S CI K Ca Ti Cr Mn Fe Cu Zn Br
33.5 17.1 0.74 1.82 91.0 118.8 26.7 532.7
2004 506.2 25.1 6.0 10.3 1.87 28.6 1.66 4.3 10.2
SD 21.8 8.8 0.34 1.30 63.7 97.1 11.7 288.8 1l.5 355.3 26.2
404 6.2
lAO 28.0 1.57 3.5 6.4
/I
(79) (28) (28) (79) (72) (80) (81) (81) (19) (81) (75) (50) (15) (79) (81) (67) (81) (43)
Mean
SD
II
6.6
3.0
(79)
80.0 95.0 47.7 59.8 10.1 93.1 52.3 8.0 1504 4.5 48.0 3.9 4.3 25.1
50.0 133.8 23.0 26.6 5.1 56.2 74.8 5.3 17.1 5.5 47.5 6.1 4.8 15.6
(26) (62) (81) (81) (20) (79) (76) (42) (31 ) (79) (77) (24) (7) (5)
"Mean and standard deviations (SD) are shown; /I is the number of samples where the detected elements were observed above detection limit. Mass concentrations are expressed 3 in Ilg m-3 ; equivalent black carbon (BC e) and trace elements concentrations are in ng m- . Table 2 is adapted from Guyon et al. [2003a, 2004b].
always important to emphasize that the burning of the forest not only releases patiicles, but also large amounts of important trace gases, which include, VOCs, methane, CO, CO 2, and several other key species [Karl et al., 2007a, 2007b; Yokelson et al., 2007, 2008; Andreae and Merlet, 2001]. 4. AEROSOL OPTICAL DEPTH MEASUREMENTS THROUGH THE AERONET NETWORK IN AMAZONIA One ofthe strong points in the atmospheric chemistry component of LBA is the long-term aerosol monitoring stations operated in Amazonia for the last 10 years. One of the set of measurements was obtained using Sun photometers within AERONET. AERONET is a globally distributed network of well-calibrated and standardized Sun photometers, maintained by NASA and expanded by national and international collaborations [Holben et al., 1998]. A CIMEL Sun-sky radiometer (manufactured by CIMEL Electronique, France) and total radiation sensors and photosynthetic active radiation sensors are deployed at several sites in Brazil for measuring the aerosol optical thickness (AOT) and solar flux in the total solar spectrum. Currently, in Brazil, there are seven Sun
photometers in operation: Ii Parana (IP) (Rondonia (RO)), Alta Floresta (AF) (Mato Grosso (MT)), Cuiaba (CB) (MT), Rio Branco (Acre), Campo Grande (Mato Grosso do SuI (MS)), Sao Paulo (SP), and Petrolina (Pernambucco), but several other sites such as Balbina (BA) (Amazonas), Belterra (Para (PA)), Santarem (PA), Brasilia (Distrito Federal), and Reserva Biol6gica Jaru (RO) have been operated [Schafer et al., 2008]. The CIMEL Sun-sky radiometer measures radiances in several wavelengths, in near real-time and provides, from direct Sun observations, aerosol properties such as AOT, column water vapor and, from sky observations, aerosol size distribution, absorption propetties, and other key aerosol propetiies [Holben et al., 1998]. This network is the only long-term project (with a record including observations from more than 11 years at some locations) ever to have provided ground-based remotely sensed column aerosol properties for this critical region. The monitoring sites generally include measurements from 1999 through the present day, but some sites have measurement records that date back to the initial days of the AERONET program in 1993 [Schafer et al., 2008]. All LBA-AERONET sites exhibit similar seasonal trend in atmospheric properties with vety low aerosol loading dur-
Figure.2' ~ime series of dai~y a:erage o~ aerosol optical thickness (AOT) in Ji Parana (JP), Alta Floresta (AF), Balbina, and CUiaba. Note the ve1Y h1gh 1I1crease 111 AOT every year in the d1y season.
ing the wet season (January-June). As the dty season starts, aerosols from biomass burning increase very significantly the amount of aerosols in the atmosphere. Figure 2 shows the time series of daily average AOT (at 500 nm) in JP, AF, BA, and CB. In the wet season, very low AOT (around 0.10.2) is observed for all sites. BA shows some increase during the dry season, but this increase is modest. During the dty season, values as high as 3.5 are frequently observed in AF, CB, and JP. These are the highest AOT values observed for all worldwide AERONET sites. The AERONET Sun photometers also measure continuously the total column water vapor (CWV) [Yamasoe et al., 1998]. Figure 3 shows the time series of the water vapor column in AF (Mato Grosso), JP (Rondonia), and Belterra (near Santarem, Pani). In the northern part ofthe basin, water vapor changes slightly on an annual basis, as can be seen for the BA CWV data, since the dry season is not vety strong in this part of the basin. In the southern part of Amazonia, the annual cycle of water vapor is much more pronounced, with water vapor sometimes as low as 1.5 em [Schafer et al., 2008].
3:
Figure 4 shows an analysis ofthe average seasonality with the weekly average values of AOT sites in the northern part of Amazonia, as well as sites in the cenado and in forest sites in the southern part of the basin. It is clear that in the south the most intense biomass burning occurs in August and Sep~ tember, while in the northern pati, the most impacted period is November-December [Schafer et al., 2008]. Aerosol forcing is the most important uncertainty in global and regional climate change [IPCC, 2007], making it important to reduce unce~tainties in the parameters relevant to the radiative forcing calculations to get aerosol optical depth with high spatial resolution. Currently, using the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor from AQUA and TERRA satellites, effOlis are being undertaken for getting AOT with a high spatial resolution of 1 Ian x 1 km in Amazonia. For this purpose, models of aerosol optical properties divided in single-scattering albedo ranges were obtained using AERONET data. This approach involves obtaining AOT not just for a few AERONET sites, but for the whole area of Amazonia. The main problem is the high cloud
Figure Time series of daily average of columnar water vapor (in centimeters of water vapor) in AF (Mato Grosso), JP (Rondoma), and Belterra (near Santarem, Panl). Data from the AERONET Sun photometer network.
240 AEROSOL PARTICLES IN AMAZONIA
• ill
L
0
ARTAXO ET AL.
• Forest (South) ill Cerrado (South)
:::. 2.0 E c a 't
::!. 1 •5 .<: .w
0-
l])
CI
('j 1.0 :,:; 0-
0
"0
e
0.5
l])
<{
til
0,
.i(
0.0 0
50
100
Figure 4. Weekly averages of aerosol optical depth (at 440 nm) for regionally grouped sites in Amazonia [modified after Schafer et aI., 2008]. The increase in aerosols due to biomass burning is more pronounced in the southern part of the Amazon Basin.
cover, but improved algorithms are being developed to increase the accuracy of AOT measurements using MODIS. 5. DIRECT RADIATIVE FORCING OF AEROSOL PARTICLES AND EFFECTS ON THE AMAZONIAN ECOSYSTEM The deforestation processes have been disturbing the Amazonian forest ecosystems in several different ways [Ometto
Dec-99
Oct-OO
Jul-01
May-02
Mar-03
Jan-04
et al., 2005]. As an example, the conversion of natural forest to pasture changes the energy and water balances and, consequently, can alter the atmospheric water content and precipitation patterns [Silva Dias et al., 2002]. These processes release a large amount of aerosol particles to the atmosphere, leading to strong changes in the surface radiation balance [Chand et al., 2006; Schafer et al., 2008; Procopio et al., 2004]. The interaction between the downward solar radiation with these aerosol particles and clouds affect directly the atmospheric radiative budget reducing the direct incoming solar radiation and increasing its diffuse fraction [Schafer et al., 2002a, 2002b; Niyogi et al., 2004; Gu et al., 1999,2003]. LBA made long-term measurements of detailed aerosol optical properties using several AERONET Sun photometers and radiometer in several sites [Eck et al., 1998]. The reduction in the surface radiation fluxes was monitored for AF, and Figure 5 shows a time series of 9 years of aerosol direct radiative forcing in AF (located in the northern portion of the Mato Grosso state), a representative area of biomass-burning influence. Instantaneous direct radiative forcing of up to -300 w m-2 (the radiative forcing is negative because this effect subtracts solar radiation at the ground) is observed most of the years during the dry season. The observations over many years of dly season measurements resulted in an average surface radiative forcing of -37 W m-2 [Procopio et al., 2004], which is a velY significant cooling effect in the surface level. As shown by Oliveira et al. [2007], a small increase in atmospheric aerosol loading increases the fraction of diffuse versus direct radiation. Therefore, the vegetation increases the efficiency of the use of solar radiation and consequently
Nov-04
Sep-05
Jul-06
Feb-OB
Apr-07
Dec-OB
0
-50 N
!E -100 .W u
....5 -150 (I)
.~
:ilm -200 '" -250
- - - ~ -
\ -300
.
....
.. .. ... ..
.~ ... ..
. ~
.~
•
...
Figure 5. Instantaneous direct aerosol forcing (in W m-2) measured over 9 years in AF, Mato Grosso state, a region heavily impacted by biomass-burning emissions. Note the very large instantaneous forcing of up to -300 W m-2 measured almost every year during the dry season,
Forest site, year: 2000-2001
o ~(/)
~ry
Season - NEE Increase: 46 %
dNat Season - NEE increase: 24 %
-10
E
"0
-=E
-20
z
-30
-·-H
UJ UJ
0.0
0.2
0.4
0.6
0.8
241
emissions [Longo et aI" 1999]. The ozone levels observed, which are sometimes higher than 100 ppb, can play an important role in the primaty productivity of Amazonian vegetation because some species and cultivars are sensitive to high levels of ozone [Bulbovas et al., 2007]. Ozone is also a greenhouse gas and changes photochemistlY significantly, potentially altering the production of secondaty organic aerosols in ways still to be studied (Martin et aL, submitted manuscript, 2008).
1.0
Relative Irradiance
Figure 6. Net ecosystem exchange (carbon flux in J-lmol m-2 S-I) as a function of aerosol loading in the atmosphere expressed as the relative irradiance index. Relative in'adiance 1 means no aerosols, with increase in aerosols as the radiation decreases. Note that negative values of NEE correspond to net ecosystem uptake and that relative irradiance index decreases with increasing AOT [Oliveira et ai" 2007].
increases the net primary productivity (NPP, defined as the net flux of carbon from the atmosphere into the vegetation per unit oftime), but up to a certain point. From clean conditions, characterized as with AOT, around 0.1 at 500 nm to AOT of 1.2, the NPP increases by 30-50% in the dly season and 24% in the wet season in Rondonia [Oliveira et al., 2007]. As soon as the AOT becomes larger than 1.2 at 500 nm, the reduction in the total flux statis to shut off carbon assimilation, and for AOT of about 3--4, the vegetation stops assimilating carbon due to the large reduction in the solar radiation flux, as can be seen in Figure 6. As the area of the biomass-burning plumes in South America, Africa, and Southeast Asia is very large, the effect of biomass-burning aerosols on the carbon exchange is an important factor in the Southern Hemisphere [Artaxo and Andreae, 2007]. Particles emitted through vegetation fires absorb solar radiation very efficiently because of the large amount ofblack carbon in these patiicles [Artaxo et al., 1988, 1998]. Taking into account that due to biomass-burning emissions there are large areas in Amazonia where there is a significant aerosol loading for about 4 months, the effects of aerosols on the carbon uptake by the Amazonian forest is possibly velY significant. The effects observed in Amazonia are certainly also present in tropical forests of Africa and Southeast Asia because of the similar biomass-burning .conditions and type of forest. Aerosol patiicles from biomass burning affect the global carbon budget, and quantification at a global scale would bring important extra knowledge of the global carbon cycle. Additionally, it was observed that large amounts of ozone are formed from biomass-burning
6. AEROSOLS AND CLOUDS: NUCLEATION PROPERTIES OF BIOGENIC AND BIOMASSBURNING PARTICLES Clouds are a critically important component of the Amazonian ecosystem because of their effect on the hydrological cycle and the radiation balance. The fonnation of clouds in the Amazon basin takes place under conditions of high water vapor availability, low CCN concentrations, high temperature, and solar radiation, which are quite different from other continental areas of the world [Williams et al., 1997; Kaufman and Koren, 2006]. Precipitating clouds are generally divided in two classes: low-level stratus type clouds (up to 2-5 km in altitude) and high-level convective systems (morie than 6 km altitude). In Amazonia during the wet season, most cloud fields are comprised of the so-called low-levyl watm clouds. High convective systems are responsible foi' most of the precipitation, and they form and develop under special thermodynamic conditions [Silva Dias et aI" 2002]. However, even during the dly season, the warm clouds are present in fair weather cumuli as well as in precipitating clouds. This seasonal difference is mainly driven by large-scale phenomena, which control the wet/dly season patterns. In such an environment where warm clouds play an important role in the hydrological cycle, the concentration of atmospheric CCN and ice nuclei and updraft velocities are the critical characteristic of the atmosphere in the formation and propeliies of convective systems [Prenni et aI" 2009; McFiggans et al., 2006], A surprising result obtained during the LBNCLAIRE intensive campaign (March-April 1998, in central Amazonia) is that, when free of anthropogenic emissions, the typical concentration of CCN in Amazonia is velY low at about 200 cm-3 at 1% supersaturation (SS) [Roberts et al., 2000, 2001] (Figure 7). During the dly season, CCN concentrations at 1% SS goes to velY high values of around 3000 cm-3 . It means that, under natural conditions, the typical concentration of natural biogenic CCN in Amazonia resembles more those ones found in oceanic than in continental areas. Typical oceanic CCN concentrations are about 100-200 per cm3, whereas typical background
242 AEROSOL PARTICLES IN AMAZONIA Dry season
-{}- Pasture site
1000
z
100
Wet Season
...sv- Pasture site
~JaruTower -...- CLAIRE '98
lOL-->-...l.----'---l._L.-.L..-<---L.->._L-->-...l.--L.---l.---''----.J 1.4 1.6 1.2 1.0 0.2 0.4 0.6 0.0
Figure 7. Number of cloud condensation nuclei particles versus supersaturation for wet season, transition, and dly season in several regions in Amazonia. Data from Roberts et al. [2001].
continental values are within 600-1000 cm-3 [Pruppacher and Klett, 1998]. It was found also that a significant fraction of the aerosol patiicles in Amazonia (40-60%) actually could act as CCN [Gunthe et al., 2009]. This is due to their relatively large size and the predominantly water-soluble chemical composition [Roberts et al., 2001]. Because of these two features, the low CCN concentration and the predominance of warm clouds, the Amazon basin has been called "the green ocean" [Williams et al., 2002]. The comparison of cloud structure between oceanic and Amazonian conditions is valid, as pointed out in the pioneer study of Squires [1956]. As over oceans, clouds in the Amazon basin typically present low veliical development and are velY efficient at producing rain quickly. The role of the low CCN concentration is impOliant because it yields cloud droplets that are initially already large, with a fast and efficient growth mechanism through water vapor diffusion. The large radii of these droplets improve the effectiveness ofthe collision-coalescence phase of rain formation, making the production of precipitation velY efficient. Most of these CCN particles are biogenic in origin. Most of them originate fi'om gas-to-patiicle conversion of VOCs emitted naturally by the vegetation [Guenther et al., 1995; Claeys et al., 2004], while some of the larger particles are primaty biological particles. The composition of such particles is mainly organic [Artaxo et al., 1990, 1994, 1998], with ve1Y low concentrations of sulfur and heavy metals. Further, they are rich in water-soluble organic compounds (WSOC), and the presence of soluble matter helps in the effectiveness
ARTAXO ET AL. of an aerosol particle in acting efficiently as a CCN [Rissler et al., 2006]. The scenario is completely different in regions dominated by biomass-burning activity. In these affected areas, the particle population rises from a few hundred per cubic centimeter [Zhou et al., 2002] to concentration levels as high as 10,000 cm-3 during most of the dly season [Rissler et al., 2006]. This extra loading of patiicles is released cally in the fine-mode fraction [Guyon et al., 2005; Artaxo et al., 2002], which makes the biomass-burning aerosols ve1Y susceptible to large-scale transport by winds, and influencing large areas that are free of local biomass-burning emis" sions. The chemical composition of these biomass-burning particles, as demonstrated in many studies, is predominantly organic matter distributed in a myriad of components, which makes them ve1Y efficient as CCN [Mircea et al., 2005; Graham et al., 2003a, 2003b; Mayol-Bracero et al., 2002; Falkovich et al., 2005; Decesari et al., 2006; Fuzzi et al., 2007]. Actually, WSOC affect the CCN properties by contributing to the solute material, altering the surface tension of the growing droplet and affecting the mechanisms responsible for the growth of activated droplets [Vestin et al., 2007; Sun and Ariya, 2006]. The complete understanding of these propeliies are largely unknown even for the relevant WSOC species found in the atmosphere and are currently an area of intense research [Svenningsson et al., 2006; McFiggans et al., 2006]. This large increase in CCN and droplet concentrations has profound effects in cloud microphysics properties [Freud et al., 2008]. Roberts et al. [2003] used a one-dimensional cloud parcel model to assess the impact of biomass-burning aerosols on cloud properties. They found that properties such as cloud droplet effective radius and maximum SS are more sensitive at low CCN concentrations, which would lead to larger interannual variation of cloud properties during the wet season than the d1y season. However, the authors also conducted measurements of CCN spectra, and they observed few differences between forested and deforested regions during the wet season and that the resulting modifications of cloud properties are small compared to those between wet and dly seasons. They postulate that, for the case of the wet season, differences in surface albedo between forested and deforested regions may dominate the impact of deforestation on the hydrological cycle and convective activity during the wet season. The large population of droplets and the inhibited growth mechanisms can make it difficult for droplets to reach the 25-llm threshold for precipitation, leading to a larger cloud fraction that evaporates instead of precipitates [Andreae et al., 2004]. The most extreme example of the influence of biomass-burning aerosol particles is the formation of pyroclouds (i.e., clouds that form in the smoke
plume over an active fire). The pyroclouds feed directly on ~he smoke and heat ~~'()m the fires. They receive conflicting Impacts: on extreme concentrations of CCN suppress the onset of precipita~ibn and the fire-generated heat invigorates the updrafts a~f further suppresses warm rain processes [Andreae et al., 2004, 2008]. Microphysical propeliies of those pyroclouds are velY differe~t to natural Amazonian clouds. First, a pyrocloud will be difficult to precipitate due to the strong inhibition of the c?llision-coalescence process that makes droplets grow to a SIze where they could precipitate. As the cloud droplet concentration is velY high, the effective cloud droplet diameter (D eff) is significantly reduced, and the smaller the D eff the less efficient is the evolution of the droplet size distribl~tion
at higher altitudes into the cloud. On the other hand clouds in oceanic are~s ("Blue Ocean") and in the green oce~n presents an effective droplet-growing process along the vertical coordinate. This is not the case of pyroclouds as is shown in Figure 8 [Andreae et al., 2004]. It is possible to observe that for blue and green ocean clouds, there is a pronounced broadening in the droplet size distribution as the cloud becomes higher, ~ ~Iear consequence of the growth of droplets due to the colhslOn and coalescence process. This growth cannot be observed for the pyrocloud measurements. Ins~ea~, the droplet distribution ofthe pyrocloud does not grow slg?lficantly after 2800 m height, a consequence of the saturation effe.ct of too. many patiicles and droplets. Figure 8c shows an mtermedlate case: A cloud formed under smoky
Blue Ocean, 18 Oct, 11 UT ·········400m --550m ··· ..····650m
Green Ocean, 5 Oct, 20 UT ··········1597m - - 1890m ......... 2338m ··•··..··2781m
--1120m --1550m
--2980m - - 3640m - - 4403m
........ \
\
\\, ... '.
~L~\j~L
50 0
10
20
30
40
50
Drop diameter (flm) 1 Smoky, 4 Oct 02,15 UT
c 0.1
~ 0.001 ..l
0.0001
10.
5
243
Pyro, 1 Oct < 4000 m; 4 Oct> 4000 m
......·.. 1731m --1899m ·..······2150m --3069m ......... 3635m --4265m
--2300m ......... 2167m 2341m --2648m ·..······2842m --3162m ··.....··3419m - - - - • 4165m
o ... Cloud \
\.
.I,,'f.\ \ . \."
=:~ci1~
.~\
o~'~··'~;'~··.Lc1~!-,-~':J=~lcJl'JliLc..1!.,~.J50 0 '~LL\H l~lJ~ !i:uu_ 10 20 30 40 50 Drop diameter (flm)
Figure 8. The of. cloud drop diameter distribution (DSD) with hel'gl1t 111 . glOW111g . ' convective . clouds, 111 . the f, . . I' evolution . OUI aelOSO legunes of (a) blue ocean, off the northeast Brazilian coast (4°S 38°W)' (b) gl'een 0 . th I . th t·· f l o ' cean 111 e c ean atr at e w~s ern tip 0 t le Amazon (6 .~ 73°W); (c) smoky clouds in Rondonia (lOoS 62°W); and (d) pyroclouds. The lowest D~~;n e;clh plot re~resents condItions at cloud base, except in Figure 8d, where a size distribution for large ash particles ou Sl .eo. t 1.e cloud IS also show~. Note. the narrowing of CDSD and the slowing of its rate of broadening with hei ht for the plOglesslvely more aerosol-nch regimes from Figures 8a to 8d [Andreae et al., 2004]. g
244
AEROSOL PARTICLES IN AMAZONIA
conditions. Despite the fact that it is possible to observe an evolution and broadening of droplet spectra, it is slower than in the blue and green ocean conditions. The consequence is that warm lain, if it really occurs, will be at higher altitudes than in the natural case. Another important effect of biomass smoke in cloud properties is the change in atmospheric thermodynamic conditions [Koreli' et al., 2008]. The increase of black carbon loaded particles in the mid-atmosphere increases the stability of the lower troposphere. The preset'lce ofthese absorbing particles results in a strong cooling at the surface, with a heating effect at altitudes around 3-4 km. This has been observed to suppress cloud formation during the biomass-buming season in the Amazon basin by remote sensing observations [Koren et al., 2004]. Modeling calculations show that the suppression is possible under certain conditions [Jiang et al., 2006]. This effect of the inhibition of low cloud formation was well documented by Koren et al. [2004,2008]. Figure 9 shows that cloud cover for low clouds is strongly dependent on the atmospheric aerosol loading, expressed as AOT at 500 nm. For "natural" aerosol conditions (AOT <0.1), the average cloud cover in this study was about 40%. As the amount of aerosol increases, the cloud cover is strongly reduced. For AOT larger than 1.2, virtually all low clouds are inhibited [Koren et al., 2004]. This value of AOT is frequently reached in the dry season, as can be seen in Figure 6. The mechanisms for this cloud suppression could be the high BC content of biomass-burning aerosol coupled with high solar radiation, in addition to the smaller droplet size. The small cloud droplets tend to evaporate if the core has high absorption properties [McFiggans et al., 2006].
ARTAXO ET AL.
7. DRY AND WET DEPOSITION OF TRACE ELEMENTS AND IMPACTS ON NUTRIENT CYCLING
The atmosphere plays an important role in nutrient cycling through wet and dry deposition of trace elements that are critically important to the forest ecosystem. Rain is an efficient scavenging process for atmospheric pollutants, and nutrients present in the atmosphere can be removed both in the gas and aerosol phases. Dty and wet deposition provides a key pathway for essential nutrients to reach terrestrial and aquatic ecosystems [Galloway et al., 1984]. Wet deposition could have several harmful effects on various ecosystems with possible acidification due to high nitrate and sulfate values and therefore can have important impacts in the biogeochemical cycles. Systematic measurements of deposition provide a simple way to evaluate the influence of human activities on the atmospheric composition and to improve the knowledge of physicochemical processes related to the atmospheric transpOli and deposition of nutrients and pollutants and their impacts on the ecosystems [Galloway et al., 1984; Lara et al., 2001]. Various factors can affect the chemical composition of precipitation, including local emissions, regional-scale pollutant and nutrient transport processes, sea level and meteorological conditions. In addition, the rainwater chemical composition is also directly controlled by in-cloud and below-cloud scavenging of the atmospheric aerosols and trace gases derived fi'om natural or anthropogenic sources [Seinfeld and Pandis, 2006]. The aqueous cloud droplet environment is also adequate for the absorption of soluble trace gases, worldng as a catalytic factor to many chemical reactions possible only in aqueous media [Hegg et al., 1984], and bioh'ansformation by microorganisms. On the other hand, during the below-cloud processes, the falling raindrops scavenge the airborne aerosols Nea peroptical depth present in the atmosphere [Pruppacher and Klett, 1998]. This East c10ud fraction mechanism of aerosol removal is one of the major processes West c10ud fraction by which the atmosphere is cleansed [Wallace and Hobbs, 2006]. Dry deposition is the h'anspOli of gaseous and particulate' species fi'om the atmosphere onto surfaces in the absence of precipitation. The dty deposition flux is commonly described as a function of the "deposition velocity." Obviously, the larger and/or heavier the particle, the greater is its deposition velocity [Seinfeld and Pandis, 2006], and consequently, the coarse-mode patiicles (d> 2.5 Ilm) will be more subject to gravitational settling than smaller ones. In Amazonia, very different pattems of deposition occur in the dty 1.2 0.4 0.6 0.8 1 0.2 and wet seasons. For both dry and wet seasons, the atmosSmoke optical depth pheric cycling of phosphorus is critically important for the Figure 9. Cloud cover in Amazonia as a function of atmospheric maintenance of the Amazonian ecosystem [Mahowald et al., 2005]. aerosol loading, expressed as AOT at 500 nm.
245
Previous studies on wet deposition conducted in pristine cant differences between the two seasons possibly in this areas in the Amazou.basin [Pauliquevis et al., 2007; Wil- site, there are no anthropogenic interferences, and it could liams et al., 1997;;4ndreae et al., 1990; Stallard and Ed- be assumed to be a pristine area. The dominant compound mond, 1981] hav9/reported low anthropogenic influence in in the rainwater composition is H+ during both dty and wet the rainwater co/¢position. A significant fraction of aerosol seasons. Despite this during the whole year, both dly and emitted by biomass burning is soluble as showed by Yama- wet seasons, the acidity in this area is always associated soe et al. [2000]. As should be expected in pristine areas, with the organic compounds. The organic compounds are rainwater composition is characterized by lower concentra- dominated by acetate that is up to 10 times higher than fortions of the major compounds than those values found in mate and oxalate, which was expected since biogenic emisareas subjected to biomass-burning emissions and anthropo- sions are one of the main sources of acetate in pristine forest genic activities [Lara et al., 2001; Trebs et al., 2006]. The areas. natural acidity (pH ranging fi'om 4.9 to 5.2) is commonly On the other hand, the rainwater composition changes comlinked to the organic acids, mainly acetic and fOlmic acids. pletely when there are influences of a heavy polluted atmoIn contrast, in areas with anthropogenic activities and land sphere due to the biomass burning. Most of the compounds use changes, especially during the dry season, this scenario emitted by biomass burning and soil emissions such as is completely different, with important changes in concen- S042-, NH4+, K+, Cl-, Ca 2+, and others become significantly trations of sulfate and nitrate, in addition to the organic acids emiched in Rondonia during the dty season. The observed [Artaxo et al., 2003]. volume-weighted mean concentrations are significantly There are significant differences in the concentrations of higher than the ones observed in clean areas such as BA. most rainwater compounds between dty and wet seasons Nitrogen is a fundamental nutrient for ecosystems. How(Figure 10), including H+, K+, Mg 2+, Ca+ Cl- N0 3- and ever, because of human activities, the input of nitrogen on 2 ' , , S04 -. The significant increase of the average concentra- terrestrial ecosystems has dramatically increased in the last tions of N0 3- could be attributed only to an increase in 50 years. In the tropics, the increase in N deposition is renatural biogenic emissions or reduced deposition due to lated with the widespread use of biomass burning as a tool reduced precipitation. Since the compounds originating for land use change. When wood is burned, the biomassfrom biomass-burning emissions have shown no signifi- associated N is volatilized, and a large fraction is emitted
1000.00 DWETSEASON
-
II DRY
SEASON
100.00
::::::
0-
(!)
E
10.00
l:
0
:0::;
e
1.00
l:
(!) (J
l:
0
0.10
U
0.01
Figure 10,. Rainwater composition in Rondonia, a region impacted by biomass burning, for the wet and dry seasons. ConcentratIOns expressed in Ileq L-1, except dissolved organic carbon and dissolved inorganic carbon both expressed inIlML- I . '
246
AEROSOL PARTICLES IN AMAZONIA
in the form of gaseous NH 3 , which may result in considerable N losses from tropical ecosystems [Kauffman et al., 1998]. As biomass burning is also a rich source of NO.,(= [NO] -l [NOzD, there is a concern on the role of these emissions on the N cycle [Trebs et al., 2006]. An important implication is the alteration of the dominant form of nitrogen deposition from nitrate to ammonium [Lara et al., 200l].'It is particularly critical because, once deposited, ammonium releases acidity since the nitrogen is either accumulated in organic nitrogen form or nitrified and leached as nitrate. 8. CONCLUSIONS
ARTAXO ET AL.
tion and quantification of local, regional, and long-range sources of all particle types; (3) improved characterization and understanding of Amazonian aerosol particles by application of a combination of advanced measurement techniques such as bulk and single-particle mass spectrometly, X-ray microanalysis, fluorescence spectroscopy, electron microscopy, and DNA analysis; (4) development of process models describing the emission of primmy biological particles from the Amazonian ecosystem and implementation of these process models in regional and global models of atmospheric chemistly, transport, and climate (Martin et al., submitted manuscript, 2008). We need studies on the hygroscopic behavior of Amazonian aerosol pmticles, since this is critically important for CCN activation as well as for radiative forcing. We also need to understand new particle formation mechanisms in Amazonia and to study the radiation balance and the relationship between aerosol particles and clouds at the large scale using advanced remote sensing techniques. The main message of this chapter for public policies is that deforestation is not only changing carbon pools, but has profound impacts on the functioning of the Amazonian ecosystem itself. The changes in the hydrological cycle tlu'ough changes in CCN population have implications for the whole South American continent as well as teleconnections with regions far from Amazonia. The changes in cloud structure have important implications on water vapor transport over large areas. The Amazonian region plays a critical role in global climate change, as well as the possible effects that the changing climate could have over tropical forest ecosystems [IPCC, 2007]. These points make a strong argument for reducing deforestation rates as quickly as possible and to look at Amazonia as a key region in regulating planetary climate from the atmospheric chemistlY point of view.
This review ofLBA science in terms of atmospheric chemistry shows a lot of progress in understanding the role of atmospheric chemistry as a main driver of the ecosystem functioning in several key areas of Amazonia. The land use changes are altering the atmospheric emissions and deposition as well as the radiative balance and the hydrological cycle over large areas. The changes in land use also alter the surface albedo that is a critical parameter in the atmospheric radiative balance. The large aerosol concentrations reported in this chapter have important health effects on the Amazonian population, and this effect must be taken into account in designing policies to reduce deforestation and biomassburning emissions. During the wet season, Amazonia is one of the very few regions on our planet where we can still observe "natural" conditions in terms of aerosol particles. This helps the understanding of how a pristine atmosphere works before humans started to change atmospheric composition in a significant way. The study of natural biogenic particles and their role in cloud droplet nucleation and precipitation formation is a key area that just started to emerge. The coupling between Acknowledgments, The authors would like to thank the finanthe atmosphere, vegetation, and climate is very strong in cial support from CNPq through the Millelmium Institute Program, Amazonia, and we have major gaps in,understanding the with the project: "Integrayao de abordagens do ambiente, uso da complex relationship of biosphere-atmosphere interactions. terra e dinamica socialna Amazonia: as relayoes homem-ambiellte We need more long-term environmental measurements at e 0 de'safio da sustentabilidade-MilenioLBA2", CNPq Processo a spatial scale compatible with the area of Amazonian for- 42019912005-5, We also acknowledge the support from FAPESP est. Very few measurements of atmospheric propelties have for several scholarships and projects that supported a significant been taken in the western part of Amazonia (between Tefe part of the results from this book chapter. We thank the support in and Sao Gabriel da Cachoeira), and the high precipitation field work from Alcides C. Ribeiro and Fernando Morais, as well rate and carbon uptake in this area indicated processes that as the laboratory support of Ana Lucia Loureiro, are different from eastern Amazonia. Opportunities for progress in identifying the sources of REFERENCES primary particles in the Amazon basin and quantifying their emissions include (l) characterization and quantification of Alves, 0, S., 0, C. Morton, M. Batistella, D, A. Roberts, and C. Souza Jr. (2009), The changing rates and patterns of deforestadifferent types of primary biological, biomass burning, mintion and land use in Brazilian Amazonia, Geophys, Monogr, eral dust, and marine aerosol pmticles, including long-term Ser., doi: 10.1029/2008GM000722, this volume, trends, seasonal cycles, diurnal variability; (2) discrimina-
All~,rea~, M,
0,.0
~91), Biomass burning its hist01y, use and distnbutl~n and Its lln~a9t on environmental quality and global clim~te, III ~lobal B~f!4nass Burning: Atmospheric, Climatic and B~osphel'lc II1!pli~{Jtiol1s, edited by I. S. Levine, pp. 3-21, MIT Pless, Cambndg~, Mass, Andreae, M. 0. {1007), Aerosols before pollution Science 315 50-51. ' "
247
Art~xo, P" F, Gerab, M, A. Yamasoe, and J. V. Martins (1994) Fme, m~de a.ero~ol composition at three long-term atmospheri~ momtonng sites m the Amazon Basin, J. Geophys. Res 99(011) 22,857-22,868. " ,
Artaxo, P" E, T. Fernandas, I. V, Martins, M, A. Yamasoe P V Hobb W M i l l ' . . .~' . a.el aut, K. M. Lon,go, and A. Castanho (1998a), Large scale aelOsol source apportlOnment in Amazonia, J. GeoAndreae, M. 0" and T. W. Andreae (1988), The cycle of biogenic phys. Res" 103(024), 31,837-31,847, . sulfur-compounds over the Amazon Basin I, DiY season J. GeoAlta~o, P., ~, C. ~e Campos, E. T. Fernandes, I. V. Martins, Z, phys. Res" 93(D2), 1487-1497. ' ' XlaO, 0. Lmdqvlst, M. T. Fermlndez-Jimenez, and W. Maenhaut And~'eae, M, 0:, and P, I. Cmtzen (1997), Atmospheric aerosols: (2000), Large sc~le mercury and trace element measurements in BIOgeochel11lCal sources and role in atmospheric chemistlY Sci- A the Amazon Basm, 1tmos, Environ., 34, 4085--4096, ence, 276, 1052-1058. ' rtaxo: P., J. V. Martms, M. A. Yamasoe, A. S, Procopio T M Andr~ae, M. 0., a?d P. Merlet (2001), Emission of trace gases and Pauhquevis, M, 0. Andreae, P. Guyon, L. V. Gatti and A C' aelOsols from bIOmass burning, Global Biogeochem C)lcles 15 Leal (2002), Physical and chemical properties of ~erosol~ in 'th~ 955-966, ' " wet and dly season in Rondonia, Amazonia J. Geophys Res Andr~ae,. M . and D. Rosenfeld (2008), Aerosol-cloud-precipi107(020), 8081, doi:1O.1029/2001JD000666~ ", tatIOn mteIactlOns, Part I. The nature and sources of cloud-active Artaxo, P., L: ~, ~. S, Lara, and T. M. Pauliquevis (2003), DIy and aerosols, Earth Sci, Rev., 89, 13--41. ~et depOSItIO~ In, Amazonia: From natural biogenic aerosols to h' Andreae, M. 0., R. W, Talbot, H. Berresheim and K M B bIOmass burl1lng Impacts, IGAC Newsletter 12-16 ' . . eec el 1990) P .. , ( , reclpltatlon chemistly in central Amazonl'a ' J.. G eopIIYS, Artaxo , P ., L . V . G attl,. A . M, C, Leal, K. M. Longo, , .S. R, de FreiR es" 95(0 10) 16,987-16,999. ta~, L. L. Lara, ~. ~, Pauliquevis, A. S, Procopio, and L. V. Andreae, M. 0., et al. (2001), Transport of biomass burning smoke RIZZO, (2~05), Qun11lca atmosferica na Amazonia: A Floresta e to ~he upper troposphere by deep convection in the equatorial as ~mIssoe~ ~e queimadas controlando a composiyao da atmo. regIOn, Geophys, Res. Lett., 28, 951-954. B sfeia amaz0111ca, Acta Amaz6nica, 35(2), 185-198, Andreae, M, 0" et al. (2002), Biogeochemical cycling of carowman, 0, M., et, al. (2009), Fire in the Earth system, Science, bon, water, energy, trace gases, and aerosols in Amazonia: The 324,481--484, dOl) 0.1 126/science.1 163886, LBA-EU.STACH experiments, J. Geophys, Res" 107(D20) BUlb.ovas, P" S. R. & Souza, R, M. de Moraes, F, Luizao, and P, 8066, d01:10.l029/2001JD000524, ' AI~a~o (2007), Re~postas de Glycine max 'TracaJ'a' exposta ao Andreae, M, 0., D, Rosenfeld, P, Artaxo, A. A. Costa, G. P. Frank, OZ01110 N . , s 0 b con d'rypes controladas, Pesqllisa Agropecllaria BraK. M. Longo, and M, A. F. Silva-Dias (2004), Smoking rain st/elra, 42(5), 641-646. clouds over the Amazon, Science 303 1342-1345 Cha?d, D" P, Guyon, P, Artaxo, 0. Schmid, G. P. Frank L V A ill " , n 'eae, M. 0., C. D. Jones, and P. M. Cox (2005) Stl'on RIZZO, 0, ~. Mayol-Bracero, L. V, Gatti, and M. 0. And~'ea~ present-day aerosol cooling implies a hot future Natt:re 43; (2006), OptIcal and physical properties of aerosols in the bound1187-1190, ' " my la,Yer and free. troposphere over the Amazon Basin during Andreae, ~, .et al. (2008), The WMOIIUGG International Aerothe bIOmass burnmg season, Atmos, Chem Ph)ls 6 2911 ." 2925. ' sol PI:eclpttatlOn Science Assessment Group (IAPSAG) Aerosol Pollutt~n Impact on Precipitation: A Scientific Review, edited by CI:eys, M" et al. (~OO~), Fo~nation of secondmy organic aerosols Z. Levm and W. Cotton, Springer New York ISBN 978-1-4020 ~ough photooxidatlOn of Isoprene, Science, 303, 1173-1176 8689-2. ' Da.vIds.on, E: A., and P, Artaxo (2004), Globally significant chang'es Altaxo, P: (2001), The atmosphedc component of biogeochemical m bIOlogIcal processes of the Amazon Basin: Results of th cycles I~ the ~mazon Basin, in The BiogeochemistJy ofthe Amalarge-scale biosphere-atmosphere experiment Glob I CI e zon Baslll, edited by~. E. McClain, R. Victoria, and I.'E, Richey, Bio!., 10, 519-529. ' a lange pp, 42-52, Oxford Umv. Press, New York ISBN 0-19-51143 Deces~~'i, S" et al. (2006), Characterization of the organic comAl'ta~o, P., and~, 0, Anill'~~e (2007), Biomass burning as a driver pOSItIOn of aerosols. from Rondonia, Brazil, during the LBAf01 atmosphe1'1c compOSitIOn and ecosystem changes iLEAPS SMOCC 2002 expe1'1ment and its representation through model Netvsletter, (4), 12-14. ' compounds, Atmos, Chelll, Ph)'S, 6 375--402 i ' , . Artax~, P., and H.-C. Hansson (1995), Size distribution of bio- Ec h aI ar, F" H. Cachier . . .A. Gaudichet, . ' and P. Artaxo (1995) , A elOgemc aerosol particles fi'om the Amazon basin Atmos Environ so emISSions by tropICal forest and savanna biomass burning' 29(3),393--402. ' ' ., Characteristic trace elements and fluxes Geoph)ls Res Lett 22' Artaxo, P., H. Storms, F. Bmynsee1s, R. Van Grieken and W 3039-3042. " , " , Maenhaut (1 ~88), Composition and sources of aerosol; from th~ Echalar, F., P, Artaxo, I. V. Martins, M. Yamasoe F. Gerab W Amazon Basm, J. Geophys, Res" 93(02), 1605-1615. Maenhaut ' , adnB. H 0Ib en (1998), Long-term "monitoring of. , Artax~, P., W. Mae~h~ut, H, Sto1'1ns, and R, Van Grieken (1990), ~tmosphe1'1c aerosols in the Amazon Basin: Source identl'fi _ AelOsol characte1'1stlcs and sources for the Amazon basin during tlOn d . ca an apportIOnment, J. Geophys, Res 103(024) 31 849 the wet season, J. Geophys. Res., 95(D1 0), 16,971-16,985, '" , 31,864,
M
.a".
?"
248
AEROSOL PARTICLES IN AMAZONIA
Eck, T., B. N. Holben, 1. Slutsker, and A.. Set~er (1998), Mea~ure ments of irradiance attenuation and estllTIatlOn of aerosol smgle scattering albedo for biomass burning aerosol in Amazonia, J. GeaphySitRes. 103(D24),31,865-31,878. Elbert, W., P. E: Taylor, M. O. Andreae, and U. Po~chl (2007), Contribution of fungi to primatY biogenic aerosols m the a~mo sphere: Wet and dry discharged spores, carbohydrates, and 1110rganic ions, Atmas. Chem. Phys., 7, 456~-4588. . '11, A . HER Maen., . Graber . , G. Schkolmk, Y. RudIch, . W. , F alkOVIC haut, and P. Artaxo (2005), Low molecular ,;elght.orgamc ~cids in aerosol particles from \Rondonia, BrazIl, dunng the blOmass-burning, transition and wet periods, Atmas. Chem. Phys., 5,781-797. . 0 . Andreae 1.. Lange, G. Roberts, J. CafmeyeI, Forment1,· P ., M. , l' Id 1. Rajta, W. Maenhaut, B. N. Holben, P'. Artaxo, a~d J. I.e Ie~e (2001), Saharan dust in Brazil and ~unna~e dunng ~he LargeScale Biosphere-Atmosphere Expenment m Amazom~ (LBA)Cooperative LBA Regional Experiment (CLAIRE) ll1 March 1998,J. Geaphys. Res., 106(Dl4), 14,919-14,~34. . . ter, P ., ea. t 1 (2007) , Changes in atmosphenc constrtuents F DIS I and . I radiative forcing, Chapter 2, Climate Change 2007: The P Iyslca Science Basis, Contribution of Wor/ring Group I to the F~urth Assessment Report of the Intergovernmental ~anel Ol~ CI/lnate Change, edited by S. Solomon et aI., Cambndge Umv. Press, Cambridge U. K., ISBN 978-0-521-88009-1. Freud, E., D. Rosenfeld, M. Andreae, A. Costa, a~d P. Art~xo (2008), Robust relations between CCN and the. vertIcal evolutlOn of cloud drop size distribution in deep convectrve clouds, Atmos. . Chem. Phys., 8,1661-1675. t al (2007) FUZZI. S ., e . , Overview ofthe inorganic. and orgamc '1 f comh position of size-segregated aerosol in Rondoma, BrazI, rom t e biomass-buming period to the onset of the wet season, J. Geophys. Res., 112, D01201, doi:l0.l029/2005JD006741. . Galloway, J. N., G. E. Likens, and M. E. Hawley (1984), .Acld deposition: Natural versus anthropogenic components, SCience, 226,829-831. Graham, B., P. Guyon, P. E. Taylor, P. Artaxo, W. Maenhaut,~. M. Glovsky, R. C. Flagan, and M. O. Andreae (2003a), Orgamc compounds present in the natural Amazonian aerosol: Characterization by gas chromatography-mass spectrometry, J. Geo108(D24) " 4766 doi: 10.1 02912003JD003990. P IIYS. R es. . , b'l' · h B etI al, (2003b) Ity Gra a t T . , . , Composition and dnunal vana ( 1 4) of the natural Amazonian aerosol, J. Geophys. Res., 108 D2 , 4765, doi:l0.1029/2003JD004049. Gu, 1.., J. D. Fuentes, H. H. Shugart, R. M. Staebler, and T. A. Black (1999), Responses of net ecosystemexchanges of carbo~ dioxide to changes in cloudiness: Results from two North American deciduous forests, J. Geophys. Res., 104(D24), 31,42131,434. M' h 1 Gu, 1.., D. D. Baldocchi, S. C. Wofsy, J. W. Munger, J. J. IC asky, S. P. Urbanski, and T. A. Boden (2003): Response of deciduous forest to the Mount Pinatubo eruptIOn: Enhanced photosynthesis, Science, 299, 2035-2038. . t a.l (1995) , A global model of natural volattle orGuen tller, A ., e 3 ganic compound emissions, J. Geophys. Res., lOO(D5), 887 8892.
ARTAXO ET AI.. Gunthe, S. S., et al. (2009), Cloud condensation nuclei in pristine tt'opical rainforest air of Amazonia: Size-resol:?d measurements and modeling of atmospheric aerosol compOSItIOn and CCN activity Atmos. Chem. Phys. Disc., 9, 3811-3870. O. Boucher, B. Graham, J. Beck, O. 1.. Mayol- Bracero, Guyon, G. C. Roberts, W. Maenhaut, P. Artaxo, and M. O. Andreae (2003a), Refractive index of aerosol particles over the Amazo.n tropical forest during LBA-EUSTACH 1999, J. Aerosol Sc/;, 34(7),883-907. Guyon, P., B. Graham, J. Beck, O. Boucher, E. Gerasopoulos, O. 1.. Mayol-Bracero, G. C. Roberts, P. Artaxo, ~ndM. Andrea~ (2003b), Physical properties and concentr~tlOn of aelOsol patticles over the Amazon tropical forest dunng background and biomass buming conditions, Atmos. Chem. Phys., 3, 951-967. Guyon, P., B. Graham, G. C. Roberts, O. 1.. Mayol-Bracero, Maenhaut, P. Artaxo, and M. O. Andreae (2004), Sources of optrcally active aerosol patiicles over the Amazon forest, Atmos. Environ., 38(7),1039-1051, doi: 10.1016/j.atmosenv.2003.10.051. (2005) P et al. Guyon,., , Airborne measurements . .of ,trace gas and . aerosol particle emissions from blOmass bummg m Amazoma, Atmos. Chem. Phys., 5, 2989-3002. . Rutledge A numencal , and P. V. Hobbs (1984), . H egg, D . A ., S . A. model for sulfur chemistry in warm-frontal rall1bands, J. Geophys. Res., 89(D5), 7133-7147. Hoffer, A., A. Gelencser, M. Blazso, P. GU~o~, P. ~rtaxo, and.M. O. Andreae (2006), Diel and seasonal vanatlOns ll1 the chemIcal composition of biomass burning aerosol, Atmos. Chem. Phys., 6 3505-3515. Holben, B. N., et al. (1998), AERONET-A feder.ate~ instrument network and data archive for aerosol charactenzatlOn, Remote Sens. Environ., 66,1-16. IPCC (2007), Climate Change 2007: The 4th Assessment R.eport o{ the Intergovernmental Panel 0/1 Climate Change. (Avatlable a http://www.ipcc.ch) . H. H . J lang, , .Xue " A. Teller G. Feingold, and Z. Levll1 (2006), I Aerosol effects on the lifetime of shallow cumulus, Geop IYs. Res. Lett., 33, Ll4806, doi: 10.1 029/2006GL026024. . Karl, T. G., T. J. Christian, R. J. Yokelson, P. Artaxo, W. MI~ f.Iao, and A. Guenther (2007a), The tropical forest and ~re emISSIons experiment: Method evaluation of volatile orgamc compo~nd emissions measured by PTR-MS, FTIR, and GC from tropIcal biomass burning, Atmos. Chem. Phys., 7, 5883-5897. Karl, T., A. Guenther, R. J. Yokelson, J. Greenberg, M. Potosnak, D. R. Blake, and P. Aliaxo (2007b), The tropical forest and fire emissions experiment: Emission, chemistry, and transport of biogenic volatile organic compounds in the lower a~mosphere over Amazonia, J. Geophys. Res., 112, D18302, dor: 10.1029/ 2007JD008539. I Koren (2006) , Smoke and pollution aerosol and. K au fiman, Y .J., .. . effect on cloud cover, Science, 313, 655-658. Kaufman, Y. J., et al. (1998), Smoke, Clouds, and RadIatIOn-Brazil (SCAR-B) experiment, J. Geophys. Res., 103(D24), 31,783-
P.,
0:
w,.
31,808. . . Kesselmeier, 1., et al. (2000), Atmospheric volatrle orgamc compounds (VOC) at a remote tropical forest site in central Amazonia, Atmos. Environ., 34, 4063-4072.
249
Koren, I., Y. J. Kaufman,L. A. Remer, and 1. V. Martins (2004), clouds on the CO 2 flux in Amazonia, Tellus B, 59B, 338-349, Measurement of the effect of Amazon smoke on inhibition of doi: 10.1 11 l/j.1600-0889.2007.00270.x. cloud formation, Sqience, 303, 1342-1345. Ometto, J. P., A. D. Nobre, H. R. Rocha, P. Artaxo, and 1.. A. MarKoren, 1., 1. Vanqr'rlei Matiins, 1.. A. Remer, and H. Afargan tinelli (2005), Amazonia and the modern carbon cycle: Lessons (2008), Smoke ~i1Vigoration versus inhibition of clouds over the learned, Oecologia, 143(4),483-500. Amazon, Scielf~e, 321, 946-949. Pauliquevis, T., L. I.. Lara, M. I.. Antunes, and P. Artaxo (2007), Lara, 1.. B. 1.. S., P. Artaxo, 1.. A. Martinelli, R. 1.. Victoria, P. Aerosol and precipitation chemistty in a remote site in Central B. Camargo, A. Kmsche, G. Ayers, E. S. B. Ferraz, and M. V. Amazonia: The role of biogenic contribution, Atmos. Chem. Ballester (2001), Chemical composition of rainwater and land Phys. Disc., 7, 11,465-11,509. use changes in Piracicaba river basin: Southeast Brazil, Atmos. Prenni, A. J., M. D. Petters, S. M. Kreidenweis, C. 1.. Heald, Environ., 35,4937-4945. S. Martin, P. Artaxo, U. Poeschl, A. Wollny, and R. Garland Longo, K. M., A. M. Thompson, V. W. 1. H. Kirchhoff, 1.. Re(2009), Wet season ice nuclei budget in the Amazon Bamer, S. R. de Freitas, M. A. F. S. Dias, P. Artaxo, W. Hart, 1. sin: Biogenic emissions and Saharan dust, Nat. Geosci., in D. Spinhirne, and M. A. Yamasoe (1999), Correlation between press. smoke and tropospheric ozone concentration in Cuiaba during Procopio, A. S., P. Artaxo, Y. J. Kaufman, 1.. A. Remer, J. S. Smoke, Clouds, and Radiation-Brazil (SCAR-B), J. Geophys. Schafer, and B. N. Holben (2004), Multiyear analysis ofAmazoRes., 104(DIO), 12,113-12,130. nian biomass burning smoke radiative forcing of climate, GeoLongo, K. M., S. R. Freitas, M. O. Andreae, R. Yokelson, and P. phys. Res. Lett., 31, 1.03108, doi: 10.1 029/2003GLO 18646. Artaxo (2009), Biomass burning in Amazonia: Emissions, long- Prospero, J. M., R. A. Glaccum, and R. T. Nees (1981), Atmorange transport of smoke and its regional and remote impacts, spheric transpoli of soil dust from Africa to South America, NaGeophys. Monogr. Ser., doi:IO.1029/2008GM000847, this volture, 289, 570-572. ume. Pmppacher, H. R., and J. Klett (1998), Mycrophysics ofClouds and Luo, C., N. Mahowald, T. Bond, P. Y. Chuang, P. Aliaxo, R. Precipitation, Springer. Siefert, Y. Chen, and 1. Schauer (2008), Combustion iron distri- Rasmussen, R. A., and M. A. K. Khalil (1988), Isoprene over the bution and deposition, Global Biogeochem. Cycles, 22, GBI012, Amazon Basin, J. Geophys. Res., 93(D2), 1417-1421. doi: 10.1 029/2007GB002964. Rissler, J., A. Vestin, E. Swietlicki, G. Fisch, J. Zhou, P. Artaxo, Mahowald, N. M., P. Artaxo, A. R. Baker, T. D. Jickells, and M. O. Andre,ae (2006), Size distribution and hygroscopic G. S. Okin, 1. T. Randerson, and A. R. Townsend (2005), Impacts properties of aerosol particles from dry-season biomass burning of biomass burning emissions and land use change on Amazoin Amazonia, Atmos. Chem. Phys., 6, 471-491. nian atmospheric phosphoms cycling and deposition, Global Bi- Roberts, G. c., P. Attaxo, and M. O. Andreae (2000), The chemisogeochem. Cycles, 19, GB4030, doi: 1O.1029/2005GB002541. try and role of cloud condensation nuclei in the Amazon Basin, Mayol-Bracero, O. 1.., P. Guyon, B. Graham, G. Robelis, M. O. J. Aerosal Sci., 31, S62-S63. Andreae, S. Decesari, M. C. Facchini, S. Fuzzi, and P. Artaxo Roberts, G. C., M. O. Andr·eae,J. Zhou, andP. Artaxo (2001), Cloud (2002), Water-soluble organic compounds in biomass burning condensation nuclei in the Amazon Basin: "marine" conditions aerosols over Amazonia 2. Apportionment ofthe chemical comover a continent?, Geophys. Res. Lett., 28(14), 2807-2810. position and importance of the polyacidic fraction, J. Geophys. Roberts, G. C., A. Nenes, J. H. Seinfeld, and M. O. Andreae Res., 107(D20), 8091, doi:IO.1029/2001JD000522. (2003), Impact of biomass burning on cloud properties in the McFiggans, G., et al. (2006), The effect of physical and chemiAmazon Basin, J. Geophys. Res., 108(D2), 4062, doi:l0.l029/ cal aerosol properties on warm cloud droplet activation, Atmas. 2001JD000985. Chem. Phys., 6, 2593-2649. Rosenfeld, D., U. Lohmann, G. B. Raga, C. D. O'Dowd, M. KulMircea, M., et al. (2005), Impoliance of the organic aerosol fi'acmala, S. Fuzzi, A. Reissell, and M. O. Andr'eae (2008), Flood tion for modeling aerosol hygroscopic growth and activation: A or drought: How do aerosols affect precipitation?, Science, case study in the Amazon Basin, Atmos. Chem. Phys., 5, 3111321(5894), 1309-1313. 3126. Schafer, J. S., T. F. Eck, B. N. Holben, P. Artaxo, M. A. YamaNiyogi, D., et al. (2004), Direct observations of the effects of soe, and A. S. Procopio (2002a), Observed reductions of total aerosol loading on net ecosystem CO 2 exchanges over different irradiance by biomass-burning aerosols in the Brazilian Amalandscapes, Geophys. Res. Lett., 31, 1.20506, doi:IO.l029/2004 zon and Zambian Savanna, Geophys. Res. Lett., 29(17), 1823, GL020915. doi: 10.1029/2001 GL014309. Nobre, C. A., M. A. Silva Dias, A. D. Culf, 1. A. Polcher, 1. H. C. Schafer, J. S., B. N. Holben, T. F. Eck, M. A. Yamasoe, and P. ArGash, 1. A. Marengo, and R. Avissar (2004), The Amazonian taxo (2002b), Atmospheric effects on insolation in the Brazilian climate, in Vegetation, Water, Humans and the Climate: A New' Amazon: Observed modification of solar radiation by clouds and Perspective on an Interactive System (IGBP Series), edited by P. smoke and derived single scattering albedo of fire aerosols, J. Kabat et aI., Springer, Berlin. Geophys. Res., 107(D20), 8074, doi:lO.l029/2001JD000428. Oliveira, P. H. F., P. Artaxo, C. Pires Jr, S. de Lucca, A. Proco- Schafer, J. S., T. F. Eck, B. N. Holben, P. Aliaxo, and A. F. Duarte pio, B. Holben, 1. Schafer, 1.. F. Cardoso, S. C. Wofsy, and H. (2008), Characterization of the optical propeliies of atmospheric R. Rocha (2007), The effects of biomass burning aerosols and aerosols in Amazonia from long term AERONET monitoring
250
AEROSOL PARTICLES IN AMAZONIA
(1993-1995 and 1999-2006), J. Geophys. Res., 113, 004204, Williams, E., et al. (2002), Contrasting convective regimes over the Amazon: Implications for cloud electrification, J. Geophys. Res., doi: 10.1 029/2007JD0093 19. 107(020), 8082, doi: 10.1029/200 IJD000380. Schmid, 0., P. Artaxo, W. P. Arnott, D. Chand, L. V. Gatti, G. P. Fral1k, A. Hoffer, M. Sclmaiter, and M. O. Andreae (2006), Williams, M. R., T. Fisher, and 1. M. Melack (1997), Chemical composition and deposition of rain in the central Amazon, BraSpectral light absorption by ambient aerosols influenced by biozil, Atmos. Environ., 31(2), 207-217. mass burning in the Amazon Basin-I. Comparison and field Worobiec, A., I. Szaloki, 1. Osan, W. Maenhaut, E. A. Stefaniak, calibration of absorption measurement techniques, Atmos. Chem. and R. Van Grieken (2007), Characterization of Amazon Basin Phys., 6, 3443-3462. aerosols at the individual particle level by X-ray microanalytical Seinfelct, J. H., and S. N. Pandis (2006), Atmospheric ChemistlJI techniques, Atmos. Environ., 41,9217-9230. and Physics: From Air Pollution to Climate Change, John WiYamasoe, M. A., Y. 1. Kaufman, O. Dubovik, L. A. Remer, B. N. ley, New York. ' Holben, and P. Artaxo (1998), Retrieval of the real pat1 of the Silva Dias, M. A. F., et al. (2002), Cloud and rain processes in a refractive index of smoke particles from Sun/sky measurements biosphere-atmosphere interaction context in the Amazon Region, during SCAR-B, J. Geophys. Res., 103(D24), 31,893-31,902. J. Geophys. Res., 107(D20), 8072, doi:IO.1029/200IJD000335. Soares-Filho, B. S., D. C. Nepstad, L. M. Curran, G. C. Cerqueira, Yamasoe, M. A., P. Artaxo, A. H. Miguel, and A. G. Allen (2000), Chemical composition of aerosol pat1icles from direct emissions R. A. Garcia, C. A. Ramos, E. Voll, A. McDonald, P. Lefebvre, of biomass burning in the Amazon Basin: Water-soluble species and P. Schlesinger (2006), Modeling conservation in the Amaand trace elements, Atmos. Environ., 34, 1641-1653. zon Basin, Nature, 440, 520-523. Squires, P. (1956), The micro-structure ofcumuli in maritime and Yokelson, R. 1., T. Karl, P. Artaxo, D. R. Blake, T. 1. Christian, D. W. T. Griffith, A. Guenther, and W. M. Hao (2007), The Tropicontinental air, Tellus, 8(4), 443-444. cal Forest and Fire Emissions Experiment: Overview and airSun, 1. M., and P. A. Ariya (2006), Atmospheric organic and biobome fire emission factor measurements, Atmos. Chem. Phys., aerosols as cloud condensation nuclei (CCN): A review, Atmos. 7,5175-5196. Environ., 40, 795-820. StallardR. F., and J. M. Edmond (1981) GeochemistlY of the Ama- Yokelson, R. J., T. 1. Christian, T. G. Karl, and A. Guenther (2008), The tropical forest and fire emissions experiment: LaboratOlY fire zon: Precipitation chemistry and the marine contribution to the measurements and synthesis of campaign data, Atmos. Chem. dissolved load at the time of peak discharge, J. Geophys. Res., Phys., 8, 3509-3527. 86, 9844-9858. Svenningssoll, B., 1. Rissler, E. Swietlicki, M. Mircea, M. Bilde, Zhou, 1., E. Swietlicki, H. C. Hansson, and P. Artaxo (2002), Submicrometer aerosol particle size distribution and hygroM. C. Facchini, 1. Zhou, J. Monster, and T. Rosenorn (2006). scopic growth measured in the Amazon rain forest during the Hygroscopic growth and critical supersaturations for mixed aerowet season, J. Geophys. Res., 107(D20), 8055, doi:IO.10291 sol pa11icles of inorganic compounds of atmospheric relevance, 2000JD000203. Atmos. Chem. Phys., 6,1937-1952. Swap, R., M. Garstang, S. A. Macko, P. D. Tyson, W. Maenhaut, Zimmerman, P. R., 1. P. Greenberg, and C. E. Westberg (1988), Measurements of atmospheric hydrocarbons and biogenic emisP. Artaxo, P. KiHlberg, and R. Talbot (1996), The long-range sion fluxes in the Amazon boundaty layer, J. Geophys. Res., transport of southern African aerosols to the tropical South At93(02),1407-1416. lantic, J. Geophys. Res., 101(Dl9), 23,777-23,791. Trebs, I., L. L. Lara, L. M. M. Zeri, L. V. Gatti, P. Artaxo, R. Dlugi, 1. Slanina, M. O. Andreae, and F. X. Meiner (2006), Dry and wet deposition of inorganic nitrogen compounds to a tropical pasture site (Rondonia, Brazil), Atmos. Chem. Phys., 6,447-469. M. O. Andreae, Max Planck Institute for Chemistty, P.O. Box Vestin, A., J. Rissler, E. Swietlicki, G. P. Frank, 'and M. O. Andreae (2007), Cloud-nucleating properties of the Amazonian bio- 3060, D-55020 Mainz, Germany. ([email protected]) P. A11axo, S. de Lucca, L. L. Lara, P. H. Oliveira, M. Paixao, mass burning aerosol: Cloud condensation nuclei measurements and modeling, J. Geophys. Res., 112, D14201, doi:10.10291 L. V. Rizzo, and K. T. Wiedemann, Institute of Physics, University of Sao Paulo, Sao Paulo, SP 05508-900 Brazil. ([email protected]) 2006JD008104. A. L. Correia, B. Holben, and 1. Schafer, NASA Goddard Space Walker, R., R. DeFries, M. del C. Vera-Diaz, Y. Shimabukuro, and A. Venturieri (2009), The expansion of intensive agriculture Flight Center, Greenbelt, MD 20771, USA. (bholben@pop900. and ranching in Brazilian Amazonia, Geophys. Monogr. Ser., gsfc.nasa.gov) T. M. Pauliquevis, LBA Central Office, Instituto Nacional de doi: 10.1 02912008GM000735, this volume. Wallace, J. M., and P. V. Hobbs (2006), Atmospheric Science: An Pesquisas da Amazonia, Manaus, AM CEP 69060-000, Brazil. ([email protected]) Introductory Survey, Academic, Burlington, Mass.
Modeling the Regional and Remote Climatic Impact of Deforestation M. A. Silva Dias,l R. Avissm} and P. Silva Dias1,3 The o?servations a~d models agree in that CUlTent levels and pattems of Amazoman deforestatlOn actually enhance mass and energy transfers between th~ land and the ~tmosphere through the creation of thermally driven circulations WIt? effects. on ramfall that are significant but vary considerably with seasons and ~·eglOns. ThIS has also indicated the necessity to identify the threshold where the mcreased deforestation .actually implies a decrease in rainfall, as poil;ted out by most o~ the low-resolutIOn general circulation models. Most of the studies of the ~'e~ote Im~ac.t are still exp~oratOlY, but they indicate that pattems of global climate m Iemote legIOns of EuraSia and North America may be affected.
1. INTRODUCTION . Becau~e .the. Ai~azonia~ rainforest is located in the tropiCS, preCipitatIOn is dommated by moist convection The rainy season is governed by the advection of moismre' from the Atlantic, which supplies, as an average, about half of the moist air used for precipitation over the region [Salali et al., 1979J. The remainder is supplied through evapotranspiration (ET), primarily driven by the deep-rooted Amazonian rainfore~t. He~ce,.t~e Amazonian rainforest plays an impOliant role m mamtammg the hydrologic balance of the region. As noted by Alves et al. [this volumeJ, deforestation of Amazonia continues at an alarming pace, so that it is impOltant to quantify the effects of such a reduction in rainforest area on the regional and global climate. As part of the
[Departamento de Ciencias Atmosfericas, IAGIUniversidade de Sao Paulo, Sao Paulo, Brazil. 2Rosenstiel School of Marine and Atmospheric Science Univer' sity of Miami, Miami, Florida, USA. 3Laborat6rio Nacional de Computa~ao Cientifica Petr6polis Brazil. ' , Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1 029/2008GM00081 7
Large-Scale Biosphere-Atmosphere (LBA) Experiment in Amazonia project, ,several modeling experiments have been conducted to study the impact of Amazonian deforestation. These provide additional insights to the earlier shldies performed by Henderson-Sellers and Gornitz [1984J, Dickinson and Henderson-Sellers [1988J, Lean and Warrilow [1989J, and Shukla et al. [1990J, among others. While the results of these earlier studies VaIy, most agreed that deforestation cau~es a. local reduction in precipitation and evaporation, a Sl.lght mcrease in surface temperature, and a reduction in ~Ols.ture converg~nce over the region. The latter change Impltes a ~onnectIOn between deforestation and large-scale atmosphenc flow, and indeed, models have indicated a reduction in large-scale vertical motion over Amazonia in response to deforestation [Nobre et al., 1991; HendersonSellers et al., 1993; Hahmann and Dickinson, 1997; Costa and F~/ey, 2.000J. Eltahir [1996J developed a conceptual ~nodel m WhiCh a reduction in net radiation at the surface m response to deforestation (due to the increase in surface albedo) ~'educes convective precipitation. This change is accompamed by a reduction in convectively induced vertical motion and in the convection-low-level moismre convergence feedback. In this chapter, we review the numerical experiments that have ~een conducted as part of LBA to improve our understandmg of the various mechanisms involved in landatmosphere interactions at the various scales (spatial and tempor~l) relevant to the Amazon basin. It is important to emphaSize that considerable progress in the capability of
251
252
SILVA DIAS ET AL.
MODELING REGIONAL AND REMOTE CLIMATIC IMPACT OF DEFORESTATION
modeling these interactions is credited to the many observational field campaigns that were carried out during LBA and provided valuable data sets for the evaluation and refinement of these models. While not the focus of this review, we do mention such evaluation efforts for cross-references to the other relevant chapters of this book. 2. FUNDAMENTAL MECHANISM AND SUPPORTING OBSERVATIONS Local interactions between vegetation cover and the atmosphere can generate "vegetation breezes" that result from the physical differences between contrasting land cover types (Figure 1). The various land cover types, such as crop fields, pastures and!or urban areas, replacing the tropical forest, redistribute the radiative energy absorbed at the Earth's surface into sensible and latent heat differently to the forest, as extensively discussed by Betts et al. [this volume]. A land cover capable of "pumping" soil moisture from deep in the ground (such as deep-rooted trees), in places where upper soil moisture is limited, will typically transpire much more than one with shallow roots (such as pasture), which can only use the water stored in the upper soil. When ET is limited, the radiative energy absorbed by the land surface is mostly used to heat the atmosphere immediately above it. For this reason, the atmosphere above a well-transpiring forest is relatively cool and moist, whereas that above a dty pasture is relatively warm and dty [cf. Betts et al., this volume]. The atmospheric pressure gradient resulting from the
virtual temperature difference (or density differences generated by temperature and moisture gradients), between contrasting land cover types triggers atmospheric circulations in the lower troposphere, with a near-land-surface component directed from the relatively cool area (which has higher surface pressure) toward the relatively warm area (which has lower surface pressure). Indeed, such a circulation has been detected in the wind data in the SW Amazon Basin Souza et aI" 2000]. Following the conceptual diagram in Figure 1, it is expected that converging vegetation breezes would lead to enhanced cloudiness and a potential effect on rainfall. Clltrin et al. [1995] first noted that shallow convective clouds during the dry season are correlated with deforested areas in the Amazon Basin. Negri et aI, [2004] analyzed geosynchronous visible and infrared satellite data over Brazil to estimate the percentage of cloudiness. They also estimated rainfall using the Tropical Rainfall Measuring Mission (TRMM) and the Special Sensor Microwave Imager. They concluded that, during the dty season, when the effects of the surface are not overwhelmed by synoptic-scale weather events, shallow cumulus cloudiness, deep convective cloudiness, and rainfall occurrence are all larger over the deforested and nonforested (savanna) regions than over areas of dense forest. Analysis of the diurnal cycle of cloudiness reveals a shift in the onset of convection toward afternoon hours in the deforested and toward the morning hours in the savanna regions when compared to the neighboring forested regions.
Durieux et al. [2003] examined whether cloudiness has already changed 1091\lly in the heavily deforested Brazilian "arc of deforestatidll," where over 15% of the primatY forest has been conve!¥{ed to pasture and agriculture. Three pairs of "deforestesVforested" areas at a scale comparable with that of a grid element of a climate model were selected to evaluate changes in land cover with changes in cloudiness observed in satellite data during the lO-ye|
3. MODELING THE REGIONAL EFFECTS OF DEFORESTATION
High Pressure
Low Prell:Ulre Hlgll
~l
(low lE, hlgll H)
Figure 1. Schematic representation of the "vegetation breeze." The pressure gradient created by the contrast of Bowen ratio (P) between the forest and the clearing is at the origin of the atmospheric circulations that develop between the two
land cover types.
The abovementioned fundamental mechanism was extensively discussed during the preparation phase ofLBA. Using a mesoscale model with a highly simplified (yet well identified) land-surface forcing, Avissar and Schmidt [1998] had found that land cover patchiness at a scale as small as 3-5 km was large enough to trigger vegetation breezes. In addition, Avissar and Liu [1996] concluded that, in the moist environment typical of tropical regions, vegetation breezes resulting from deforestation could trigger thunderstorms. Yet, it was thought that the vegetation breezes were mostly relevant for the dly season and preliminaty modeling studies focused on simulating that season. Pre-LBA studies such as Silva Dias and Regnier [1996] and Dolman et al. [1999] pointed out deficiencies in our understanding of the coupling between biosphere and atmosphere that could lead to significant deficiencies in the description of boundary layer structure and thus affect the simulation of local circulations. A
253
simple theory for convective circulations induced by surface heterogeneities was proposed by Souza et al. [2000]. The theory is based on the thennodynamics of heat engines and provides a simple physical explanation for the general characteristics of circulations forced by surface inhomogeneities. It predicts that the intensity of the mesoscale convective circulation forced by deforestation depends on the difference of the near-surface temperature and humidity between the forest and cleared regions and on the depth of the convective boundaty layer. Baidya Roy and Avissar [2002] simulated the impact of deforestation on Rondonia's regional hydrometeorology. For that purpose, their mesoscale model was set up with a horizontal resolution of 1 km and was forced with realistic atmospheric background conditions and land-surface characteristics. They found that vegetation breezes formed and converged above deforested areas, resulting in strong updrafts canying the moisture that had transpired from the forest and generating shallow-convection clouds. These results were supported by an analysis of images derived fi'om the Geostationaty Earth Observing Satellite 7 (GOES 7). Wang et aI, [2000] used a numerical mesoscale model to investigate the impact of mesoscale circulations on the distribution of precipitation and cloudiness over a deforested area in Rondonia. Observed patterns of deforestation with scales on the order of 10 km were used to describe land surface conditions. Their results suggest that the synoptic forcing, in terms! of atmospheric stability and background horizontal wind,· dominates during the rainy season when synoptic conditions were so favorable to moist convection that the added effect of surface heterogeneity was negligible. During the dry season, a noticeable impact of mesoscale circulations resulting in .enhancement of shallow clouds was simulated; the mesoscale circulations also triggered scattered deep convection that altered the spatial distribution of precipitation, During the break period, the transition from the rainy season to the dty se
SILVA DIAS ET AL. 254
255
MODELING REGIONAL AND REMOTE CLIMATIC IMPACT OF DEFORESTATION
high-resolution topography, coastlines, and large rivers. In a control simulation, the biophysical parameters of current vegetation ~ere used. In the other simulation, the forestedarea biophysical parameters were replaced by those corresponding to the pasture areas of the region. Near coastal zones and along large rivers, deforestation resulted in reduced cloud cover and precipitation. However, increased cloud co'ver and precipitation were simulated in upland areas, especially on slopes facillg river valleys. Wind speed near the surface was the meteorological variable that presented the most significant change due to deforestation. The reduction in roughness coefficient resulting from the shift from forest to pasture produced increased wind speeds neal' the Atlantic coast. The greater wind speeds diminished local humidity convergence and, consequently, reduced rainfall totals in nearby regions. Ramos da Silva and Avissar [2006] simulated the evolution of convection in Rondonia during the rainy season to elucidate some of the complex land-atmosphere interactions taking place in that region. Simulations were either initialized 01' evaluated with several of the TRMM/Wet Season Atmospheric Mesoscale Campaign/LBA data sets. They explained that to simulate properly the domain-average accumulated rainfall in Rondonia, reliable initial profiles of relative humidity and soil moisture are necessary, as they affect the timing and spatial rainfall accumulation. In general, more water in the soil and/or the atmosphere produces more rainfall. But these conditions affect the onset of rainfall in opposite ways; while higher relative humidity leads to early rainfall, higher soil moisture delays its formation. As illustrated in Figure 2, which summarizes these findings, an increase (decrease) of the initial relative humidity by only 10% generates significantly more (less) and earlier (delayed) rainfall. The impact of soil moisture content on the timing and rainfall location creates a negative feedback that works to homogenize the spatial distribution of rainfall and land water content. Indeed, a wet soil delays convection and produces more precipitation downstream. Early morning atmospheric humidity appears to be quite important for the simulation of rainfall in this region. A nocturnal rainfall event raises the soil moisture and delays the onset of the next-day convective rain. Otherwise, the moisture remains in the atmosphere and leads to earlier rainfall the next morning. Thus, Ramos da Silva and Avissar [2006] concluded that models that are unable to represent nocturnal rainfall in Amazonia will likely fail to simulate rainfall due to misrepresentation of early morning atmospheric and soil moisture. Other models, such as the operational European Centre for Medium-Range Weather Forecasts Integrated Forecast System, simulated rainfall too early in the wet season in Rondonia [Betts and Jakob, 2002]. In another study,
"'-RHt - S-Pol RHControl FOG 8
Wet Soil
. Dry Soil
b
a 6
4 2
-
Cii 8 c: 6 '(ij
a:
4 2
15
20
15
Local Time (hr) Figure 2, Domain-average accumulated precipitation for different lUns simulated with RAMS and estimated from S-Pol on (a) 4 February, (b) 6 FeblUary, (c) 14 February, and (d) 23 February [from
Ral1lOs da Silva and Avissar, 2006]. Silva Dias et at. [2002] simulated a squall line that fonned in February 1999 for the same region and noticed a delayed precipitation compared to observations. Correia et at. [2007] investigated the impact of land covel' changes on the regional climate in Amazonia. They performed four, 13-month integrations for the following scenarios: (1) no deforestation, (2) current conditions, (3) deforestation predicted for 2033, and (4) large-scale deforestation. All initial and prescribed boundary conditions were kept identical for all integrations, except the land cover changes. The results show that during the dry season, the postdeforestation decrease in root depth plays an important role in the energy budget, since there is less soil moisture available for ET. In all scenarios, there was a significant increase in the surface temperature, from 2.0°C in the first scenario up to 2.8°C in the last one. In scenarios 1, 2, and 3, a negative feedback mechanism was observed in the hydrological cycle, with greater amounts of moisture being carried to the deforested areas. The increase in moisture convergence was greater than the reduction in ET for both scenarios 2 and 3. These and the mesoscale thermodynamic processes caused an increase in precipitation. A different situation was observed in the large-scale deforestation scenario 4: a local increase of moisture convergence was observed, but not sufficiently intense to generate an increase in precipitation; the local ET decrease was dominant in that scenario. These results indi-
cate that partial deforestation in Amazonia may cause a local increase in precipitation. However, if the deforestation increases, this condjtion becomes unsustainable, leading to drier conditions a9tl, consequently, to reduced precipitation in the region. ,/l Avissar et at [2002] have summarized possible changes in rainfall accumulation due to a progressive expansion of deforestation. Accordingly, three scenarios could be considered: (1) rainfall will decrease linearly with deforested area; (2) rainfall will first decrease rapidly, then more slowly as the deforested area expands; (3) given the abovementioned finding, it could be anticipated that rainfall will first remain unaffected (or possibly even increase) as a result of initial deforestation, but will then fall rapidly as the deforested area crosses some threshold estimated at 30-50% ofdeforestation. These three scenarios are illustrated in Figure 3. Sampaio et at. [2008] used a global climate model (GCM) to address the threshold of noticeable effects of deforestation. Their results showed that changes in vegetation cover in Amazonia modified the calculated fields of radiation, energy, water balance, and the dynamical structure of the atmosphere and, consequently, the moisture and mass convergence in low levels of the atmosphere, mainly in dry season. The main impacts on Amazonian climate, because of deforestation, occurred over eastern and central Amazonia and were more evident when total deforested area was larger than 40%. However, the Sampaio et at. results are dependent on the relatively low resolution of the GCM that hinders the development of truly local circulations generated by the patchy nature of deforestation as shown by Ramos da Silva et at. [2008].
1.2
c
1.1
0
'.p
ro .....
1
'u w .... 0-
0.9
'a.. w
> 0.8
'.p
ro
Qj 0::
0.7 0.6
0
20
40
60
80
100
Deforestation (%) ~igure
3. Conceptual impact of deforestation on relative precipitation. The three curves indicate different models, among many other possible ones [from Avissar et al., 2002].
While the numerical experiments described above were aimed at understanding the basic mechanisms affecting the hydrometeorology of the basin fi'Om a larger-scale point of view, other experiments were designed to simulate the combined effects of such processes at the seasonal scale at smaller spatial scales. Scenarios of land covel' and climate change on the hydroclimatology of the basin were also conducted. Among others, Ramos da Silva et at. [2008] used a regional climate model and the socioeconomic scenarios of future land covel' changes designed by Soares-Filho et at. [2004] to simulate the impacts of projected deforestation in. the Amazon Basin on its hydroclimatology. Basically, ramfall gradually decreases as the deforestation increases but the magnitude of the impacts depends on the prevail~ ing climate regime, and there is a significant spatial variability in the response. Impacts are stronger under drier, El Nino conditions and are also more intense in the western and southern regions, downstream of the easterly trade winds (which intensify in the basin after deforestation). This leads to large areas of low precipitation, higher temperatures, and hi~her sensible heat fluxes in the basin. These results agree WIth those of Votdoireand Royer [2004], who used the ARPEGE-Climat GCM and found that during El Nino periods, the effects of deforestation are enhanced. The climate regi?1e also seems to control the relationship between deforestation extent and the hydroclimatological respo~se: Wetter, La Nina conditions lead to a linearly decreasmg trend of ptecipitation as deforestation expands, but under dly conditions, a steep decrease may occur even for minor levels of deforestation. FmihelIDore, the geographical locations within the basin experience deforestation in a different way. Indeed, some regions (e.g., the high elevation of the eastern part of the basin) even see an increase rainfall as a result of the basin deforestation. Experiments performed with GCMs suggest that deforestation may establish a permanent savanna in Amazonia mostly over the east [Oyama and Nobre, 2003]. But Ramo; da Silva et at. [2008] show that progressive deforestation can affect the propagation of squall lines toward the west and with stronger impacts during EI Nino years. Thus, the combination of the large-scale effects on the dlying of the eastern part of the basin, and the impacts on the local mesoscale systems (i.e., the squall lines), may cause a much stronger effect if they occur during an El Nino year. Similar to GCMs, regional models typically show that temperature and wind speed increase, and latent heat flux decreases as a result of expanding deforestation in the Amazon basin. However, two important differences are noticed when comparing the two types of models. First, the response to deforestation obtained with a regional model is of lesser magnitude than that obtained with the GCMs. Second, while
256
MODELING REGIONAL AND REMOTE CLIMATIC IMPACT OF DEFORESTATION SILVA DIAS ET AL.
the GCMs display an increase in the domain-averaged sensible heat flux, the regional models typically show a decrease. This is an interesting example of possible nonlinear feedbacks causecr"by scale interactions in higher-resolution models. Since the regional models show spatial patterns ofchange in the sensible heat flux that couelate with changes in downward solar radiation and fractional cloud cover, this discrepancy betWeen the models is interpreted as being caused by the nonlinear land-atmosphere-cloud interactions resolved in regional models that are neither resolved nor parameterized in the GCMs. As mentioned by Castro et al. [2005], a GCM will show a better representation of the large-scale variability, but the utility of a higher-resolution regional model, or its value added, is to resolve the smaller-scale features, which have a greater dependence on the surface boundary. The interactions that take place in regional models induce positive precipitation anomalies and explain why the effects of deforestation are less evident during wet years compared to dly years, when local circulations and cumulus clouds are better correlated with landscape heterogeneities. One last comment over the deforestation effect can be drawn fi'om the work of Costa et al. [2007], who examined the effect of land cover change from forest to cropland, specifically soybeans, instead of grassland as in most of the other studies. They showed that the decrease in precipitation after a soybean extension is significantly higher when compared to the change after a pastureland extension, a consequence of the veIY high albedo of the soybean. Berbet and Costa [2003] had shown that most ofthe spatial and seasonal variability in the simulated climate after a tropical deforestation can be explained by the difference in the radiation reflected by the surface covered by different vegetation. Also, D 'Almeida et al. [2007] point out that although most of the model simulations indicate that complete deforestation leads to a restrained water cycle, while the simulated effect of small, disturbed areas shows a contrasting tendency, some differences between coarsely spatially averaged observations and finely sampled data sets have also been encountered. According to their work, the contrasts between the two scenarios are only partially explained by the different spatial resolutions among models and observations, since they seem to be further associated with the weakening of precipitation recycling under scenarios of extensive deforestation and with the potential intensification of convection over areas of land-surface heterogeneity. 4. MODELING THE REMOTE IMPACTS OF AMAZON DEFORESTATION Most pre-LBA studies of deforestation impacts using GCMs focused on the local effects of heating and dlying.
More recently, the remote impacts of Amazonian deforestation have been studied. Gedney and Valdes [2000] used a GCM to show that complete Amazonian deforestation could result in changes in the climate far afield from the region of deforestation. In particular, the model predicted statistically significant changes to winter rainfall over the NE Atlantic, extending toward western Europe. These were associated with large-scale circulation changes in middle and high latitudes. Simulation of these circulation changes with a simple model confirmed that the physical mechanism responsible was planetary wave propagation. Tropical heat sources associated with deep convection have been shown to be relevant for the organization of middle- and higher-latitude atmospheric teleconnection patterns through planetary wave propagation. The PacificlNorth American and the Eurasian patterns are examples of important anomalous atmospheric circulation patterns that have been connected to the Amazonian heat source [Grimm and Silva Dias, 1995]. The larger interhemispheric response associated with the Amazonian heat source has been attributed to the Kelvin waves generated by the diurnal forcing of convection in Amazonia/Central Brazil and their interaction with the slow Rossby modes [Raupp and Silva Dias, 2004]. The interaction between the high-frequency waves generated by convection in the tropical sector of South America (such as the waves generated by the diurnal convective forcing) and the low-frequency modes (such as the synoptic or planetary scale waves with period of the order of a few days) may also have some implications for atmospheric predictability at longer, intraseasonal timescales. In fact, atmospheric models generally exhibit rather poor simulation and low predictability of the Madden-Julian Oscillation [Jones et al., 2000; Hendon et al., 2000]. Raupp and Silva Dias [2005, 2008] propose that such a low skill may be due, at least in part, to the difficulty the models have in representing the interaction between moist convection and large-scale dynamics. In this process, the role of high-frequency modes is crucial, suggesting the importance of Kelvin and inertiogravity waves (and therefore of the latent heat release) for the generation of the low-frequency variability of the tropical atmospheric circulation. As the inertiogravity waves are directly associated with precipitation and moist convection, there is a clear link between the predictability and realistic representation of intraseasonal oscillations in atmospheric models and well founded moist convective parameterization. These results are particularly relevant for the interaction between the Amazonia/Central Brazil heat source and the larger-scale circulation in view of the large amplitude of the gravity modes generated by convection, as shown by Raupp and Silva Dias [2008].
Theoretical evidence ofthe coupling between anomalous Amazonian convection/and global-scale atmospheric telec~nnections is clearlw'supported by simulations performed With GCMs. For iI;istance, based on mesoscale modeling
studies that indicated that deforestation would enhance the thunderstorm activity in the basin, Avissar et al. [2002] speculat~d that land-atmosphere teleconnections should develop outSIde of the Amazon basin. Using the NASA Goddard
/f.-I',J
4 3 2 1
60N
•
0 8
30N
6
4
EO
2
30S
0 10 8 6
60S
4 2
60W
o
257
60E
"0
0 8
..........
6
E ......
E
4
c
2
:.p
0
2
60N
......... ~
1.5
30N
1
EO
0 5
30S
4 3 2
0
0
+'
'0.
'0 ~
a..
0.5
60S
1 0 9
3 (-2,117.5)
J
DO
J
12
Figu.~e 4. Worldwide locations where precipitation has either significantly (top) decreased or (bottom) increased durin a pellod of at least 3 months of the year, as a result of tropical deforestation as simulated by a superensemble of thre: GCMs: GISS, AM, and CCSM. The superensemble-mean annual cycle of precipitation (nnn d- I ) for the control (black cu~es) and deforeste~ (shaded ~ur:es) cases at continental locations most severely affected by the deforestation is also reples~nte~. ~he shadmg scale ~ndICates the number of months registering a statistically significant change (Student's t test 95 Yo slgmficance level) dunngthe annual cycle.
258
SILVA DIAS ET AL.
MODELING REGIONAL AND REMOTE CLIMATIC IMPACT OF DEFORESTATION
Institute for Space Studies (GISS) GCM, Werth and Avissar [2002] indeed found several remote areas where there was a noticeable response. The remote effect tended to be strongest in the ii'i'eas closest to Amazonia and decreased "downstream" as one passes over the Pacific. More recently, Hasler et al. [2009] revisited this issue wi~h three different GCMs: the GISS GCM, the GISS Atmosphenc Model and the National Center for Atmospheric Research COlll1~unity Climate Model Version 3. Using the simulations of all three models in a multimodel ensemble, they confirmed the existence of these land-atmosphere teleconnections with the Amazon basin showing monthly precipitation changes reaching up to 34% outside the deforested areas, .mai~ly in the tropics and at a few locations in nOli?ern mIdl~t1tudes (Figure 4). But because the three models sll~u~ate.d different patterns of midlatitude geopotential and precIpItatiOn change due to tropical deforestation, the multimodel ensemble clearly reduced the intensity of these teleconnections. Studies using other GCMs to predict the large-scale effects of tropical deforestation have produced a range of responses from weak [pindell et al. 2006] to strong [Gedney and Valdes, 2000], with the study of Hasler et al. [2009] being somewhere in between. 5. CONCLUSIONS
With the first phase of LBA completed and the second phase on its way, it is important to build upon what we have learned during the first phase, as described in this chapter and recommend the future research directions: 1. High-resolution, intermmual regional simulations with improved hydrology schemes and ecological dynamics (including fire and aerosol) are likely to provide additional insights on eco-hydro-climate (EHC) interactions and feedback at all spatial and temporal scales. While some work in that direction has been initiated, it still requires a sustained effort by the LBA community to elucidate the complex mechanisms and processes involved in these interactions. 2. On the basis of the various GCM experiments described above and theoretical results based on simple shallow water models, it is reasonable to assume that a midlatitude response to Amazonian deforestation exists, but it is unclear what the precise pattern or its magnitude is. Fmihermore, the use of prescribed climatological sea surface temp~ra~~'es (SSTs) can act to dampen the model interannual vanabIhty. Reproducing these results with transient observed SST, or with coupled ocean-atmosphere models, is an essential next st~p to these studies to provide additional insights on the spatlal and temporal variability of these teleconnections. 3. For long-term climatological simulations, it is also essential to simulate the ecosystem dynamics as well as the various parameters believed to change the climate (e.g., incr~ase of emission of carbon dioxide, land use change under agncultural expansion especially with the development of biofuels, as a patiial alternative to fossil fuel). Such simulations are likely to highlight new feedback not yet considered. 4. Interactions between regional and global-scale processes are still simulated in a one-way nesting strategy. In other words, using the current modeling capability, macroscale dynamics are forced into the regional-scale ~elds, but the opposite (i.e., regional-scale to macroscale) IS not simulated. The new generation of Earth system models, such as the Ocean-Land-Atmosphere Model [Walko and Avissar, 2008a, 2008b], are capable of simulating these two-way interactions. They should be used to elucidate the importance of these interactions.
One of the more impOliant contributions of modeling the regional climate impacts of deforestation in the Amazon Basin with high-resolution numerical models has been to provide an understanding and a qua.ntitative analys~s of the processes whereby rainfall may be ll1creased assocIated with local deforestation or may compensate the large-scale reduction of moisture convergence and ET. The observations and models agree in that CUlTent levels and patterns of Amazonian deforestation actually enhance mass and energy transfers between the land and the atmosp.here through the creation of thernmlly driven circulation's wIth effects on rainfall that are significant but vary considerably along the seasons and regions. This has also indicated the neces~ity to identify the threshold where the increased deforestatiOn actually implies a decrease in rainfall, as pointed out by most Acknowledgments. M. A. Silva Dias and P. Silva Dias acknowledge the support ofFAPESP, CNPq, and Instituto do Milenio. The of the low-resolution GCMs. From the point of view of remote impacts of Amazonian Moore Foundation funded part of the research. deforestation on global climate, considerable progress has come from the logical induction that land use change implies effects on rainfall, which alter the featur~s associated REFERENCES with a tropical heat source, which interacts WIt~ planetary scale waves that provide the global teleconnectiOns. Most Alves, D. S., D. C. Morton, M. Batistella, D. A. Roberts, and C. Souza Jr. (2009), The changing rates and patterns of deforestaof the studies of the remote impact are still exploratory but tion and land use in Brazilian Amazonia, Geophys. Monogr. indicated that patterns of global climate in remote regions of Ser., doi:10.l02912008GM000722, this volume. Eurasia and North America may be affected.
Avissar, R., and Y. Liu (1996), Three-dimensional numerical Shldy of shallow convective clouds and precipitation induced by landsurface forcing. J. Gep"phys. Res., 101(D3), 7499-7518. Avissar, R., and T. SI;,limidt (1998), An evaluation of the scale at which ground-sur~ce heat flux patchiness affects the convective boundary layer us'ing a Large-Eddy Simulation model, J. Atmos. Sci., 55, 2666-2689. Avissar, R., P. L. Silva Dias, M. A. F. Silva Dias, and C. Nobre (2002), The Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA): Insights and future research needs, J. Geophys. Res., 107(020),8086, doi:l0.l029/2002JD002704. Baydia Roy, S., and R. Avissar (2002), Impact of land use/land cover change on regional hydrometeorology in Amazonia, J. Geophys. Res. 107(D20), 8037, doi:lO.l029/2000JD000266. Berbet, M. L. C., and M. H. Costa (2003), Climate change after tropical deforestation: Seasonal variability of surface albedo and its effects on precipitation change, J. Clim., 16(12), 2099-2104. Betts, A. K., and C. Jakob (2002), Evaluation ofthe diurnal cycle of precipitation, surface thelIDodynamics and surface fluxes in the ECMWF model using LBA data, J. Geophys. Res., 107(D20), 8045, doi:lO.l029/200lJD000427. Betts, A. K., G. Fisch, C. von Randow, M. A. F. Silva Dias, 1. C. P. Cohen, R. da Silva, and D. R. Fitzjarrald (2009), The Amazonian boundary layer and mesoscale circulations, GeophyS: Monogr. Ser., doi: 1O.1029/2008GM000725, this volume. Castro, C. L., R. A. Pielke Sr., and G. Leoncini (2005), Dynamical downscaling: Assessment of value retained and added using the Regional Atmospheric Modeling System (RAMS), J. Geophys. Res., 110, D05108, doi: 1O.1029/2004JD004721. Chagnon, F. J. F., and R. L. Bras (2005), Contemporary climate change in the Amazon, Geophys. Res. Lett., 32, Ll3703, doi: 10.1 029/2005GL022722. Correia, F. W. S., R. C. S. Alvahl, and A. O. Manzi (2007), Modeling the impacts ofland cover change in Amazonia: A regional climate model (RCM) simulation study, Them'. Appl. Climatol., 93(3--4), 225-244, Published Online First: October 2007, doi:lO.1007/s00704-007-0335-z. Costa, M. H., and 1. A. Foley (2000), Combined effects of deforestation and doubled atmospheric CO 2 concentrations on the climate of Amazonia, J. Clim., 13, 18-34. Costa, M. H., S. N. M. Yanagi, P. J. O. P. Souza, A. Ribeiro, and E. 1. P. Rocha (2007), Climate change in Amazonia caused by soybean cropland expansion, as compared to caused by pastureland expansion, Geophys. Res. Lett., 34, L07706, doi:1O.1029/ 2007GL029271. Cutrin, E., D. W. Matiin, and R. Rabin (1995), Enhancement of cumulus clouds over deforested lands in Amazonia, Bull. Am. Meteorol. Soc., 76, 1801-1805. D'Almeida, C., C. 1. Vorosmarty, G. Hurtt, 1. A. Marengo, S. L. Dingman, and B. Keim (2007), The effects of deforestation on the hydrological cycle in Amazonia: A review on scale and resolution, Jnt. J. Climato!., 27, 633-647. Dickinson, R. E., and A. Henderson-Sellers (1988), Modelling tropical deforestation: A study of GCM land surface parametrizations, Q. J. R. Meteorol. Soc., 1(14),439--462.
259
Dolman, A. 1., M. A. F. Silva Dias, J.-C. Calvet, M. Ashby, A. S. Tahara, C. Delire, P. Kabat, G. F. Fisch, and C. A. Nobre (1999), Meso-scale effects of tropical deforestation in Amazonia: PreparatOly LBA modelling studies, Ann. Geophys., 17, 1095-1110. Durieux, L., L. A. T. Machado, and H. Laurent (2003), The impact of deforestation on cloud cover over the Amazon arc of deforestation, Remote Sens. Environ., 86,132-140. Eltahir, E. B. (1996), Role of vegetation in sustaining large-scale atmospheric circulations in the tropics, J. Geophys.Res., 101(D2), 4255--4268. Findell, K. L., T. R. Knutson, and P. C. D. Milly (2006), Weak simulated extratropical responses to complete tropical deforestation, J. Clim., 19(12),2835-2850. Gandu, A. W., 1. C. P. Cohen, and 1. R. S. Souza (2004), Simulation ofdeforestation in eastern Amazonia using a high-resolution model, Theor. Appl. Climatol., 78, 123-135. Gedney, N., and P. 1. Valdes (2000), The effect of Amazonian deforestation on the Northern Hemisphere circulation and climate, Geophys. Res. Lett., 27(19), 3053-3056. Grimm, A. M., and P. L. Silva Dias (1995), Analysis of tropicalextratropical interactions with influence functions of a barotropic model, J. Atmos. Sd, 52, 20, 3538-3555. Hahmann, A., and R. E. Dickinson (1997), RCCM2-BATS model over tropical South America: Applications to tropical deforestation, J. Clim., 10, 1944--1964. Hasler, N., D. Werth, and R. Avissar (2009), Tropical deforestation impact on global hydroclimate: A multimodel ensemble analysis, J. Clim., 22, 1124--1141. Henderson-Sellers, A.:, and V. Gornitz (1984), Possible climatic impacts of land cover 'transformations, with particular emphasis on tropical deforestation, Clim. Change, 6,231-257, doi:l0.1007/ BF00142475. Henderson-Sellers, A., R. E. Dickinson, T. B. Durbidge, P. 1. Kennedy, K. McGuffie, and A. 1. Pitman (1993), Tropical deforestation~modeling local-scale to regional-scale climate change, J. Geophys. Res., 98,7289-7315. Hendon, H. H., B. Liebmann, M. Newman, 1. D. Glick, and J. Schemm (2000), Medium range forecast enors associated with active episodes ofthe Madden-Julian Oscillation, Mon. Weather Rev., 128, 69-86. Jones, C., D. E. Waliser, J. K. Schemm, and W. K. Lau (2000), Prediction skill of the Madden-Julian oscillation in dynamical extended range forecasts, Clim. Dyn., 16, 273-289. Lean, 1., and D. A. Warrilow (1989), Simulation of the regional climatic impact of Amazon deforestation, Nature, 342(41), 1-13. Machado, L. A. T., H. Laurent, and A. A. Lima (2002), The diurnal march of the convection observed during TRMMWETAMCILBA, J. Geophys. Res., 1 07(D20), 8064, doi: 10.1 029/ 2001JD000338. Negri, A. J., R. F. Adler, L. Xu, and 1. Surratt (2004), 1. The impact of Amazonian deforestation on dly season rainfall, J. Clim., 17, 1306--1319. Nobre, C., P. Sellers, and J. Shukla (1991), Amazonian deforestation and regional climate change, J. Clim., 4, 957-988.
260
MODELING REGIONAL AND REMOTE CLIMATIC IMPACT OF DEFORESTATION
Oyama, M. D., and C. A Nobre (2003), A new climate-vegetation equilibrium state for Tropical South America, Geophys. Res. Lett., 30(23), 2199, doi:l0.l029/2003GLOI8600. Ramos da Silva, R., and R. Avissar (2006), The hydrometeorology of a deforested region ofthe Amazon, 1. Hydromeleorol., 7, 1028-1042. Ramos da Silva, R., D. Werth, and R. Avissar (2008), Regional impacts of futnre land-cover changes on the Amazon basin during the wet-season climate impacts, 1. Clim., 21,1153-1170. Raupp, C. F. M., and P. L. Silva Dias (2004), Effects of nonlinear processes on the interhemispheric energy propagation forced by tropical heat sources, Brazilian 1. Meteorol., 19(2), 177-188. Raupp, C. F. M., and P. L. Silva Dias (2005), Excitation mechanism of mixed Rossby-Gravity waves in the equatorial atmosphere: Role of the nonlinear interaction among equatorial waves,1. Almos. Sci., 62(5), 1446-1462. Raupp, C. F. M., and P. L. Silva Dias (2008), Resonant wave interactions and their potential role on tropics-extratropics connection,1. Almos. Sci., 65, 3398-3418. Salati, E., A Dall'Olio, E. Matsui, and J. R. Gat (1979), Recycling of water in the Amazon, Brazil: An isotopic study, WalerReSOliI'. Res., 15(5), 1250-1258. Sampaio, G., C. Nobre, and P. Satyamurty (2008), Climatic consequences of gradual conversion of Amazonian tropical forests into degraded pasture or soybean cropland: A new vegetationclimate equilibrium state in Amazonia, Intel'llational Scientific Conference Amazon in Perspective, Integrated Science for a Sustainable Future, Manaus, November 17-20, 2008. http:// www.1baconferencia.org/lbaconC2008/eng/index.htrn. Shukla, J., C. A. Nobre, and P. Sellers (1990), Amazon deforestation and climate change, Science, 247,1322-1325. Silva Dias, M. A F., and P. Regnier (1996), Simulation of mesoscale circulations in a deforested area of Rondonia in the dry season, in Amazonian Deforestation and Climate, edited by J. Gash et a1., pp. 531-547, John Wiley. Silva Dias, M. A F., et al. (2002), A case stndy of convective organization into precipitating lines in the Southwest Amazon during
the WETAMC and TRMM-LBA, 1. Geophys. Res., 107(D20), 8078, doi: 1O.102912001JD000375. Soares Filho, B., A. Alencar, D. Nepstad, G. C. Cerqueira, M. Vera Diaz, S. Rivero, L. Solorzano, andE. Voll (2004), Simulating the response of land-cover changes to road paving and governance along a major Amazon highway: The Santarem-Cuiaba corridor, Global Change Bioi., 10(7), 745-764. Souza, E. P., N. O. Renno, andM. A F. Silva Dias (2000), Convective circulations induced by surface heterogeneities, 1. Almos. Sci., 57, 2915-2922. Voldoire, A, and E. J. F. Royer (2004), Tropical deforestation and climate variability, Clim. Dyn., 22, 857-874. Walko, R. L., andR. Avissar (2008a), The Ocean-land-Atmosphere Model (OLAM). part I: Shallow-water tests, Mon. Weather Rev., 136,4033-4044. Walko, R. L., and R. Avissar (2008b), The Ocean-Land-Atmosphere Model (OLAM). part II: Formulation and tests of the nonhydrostatic dynamic core, Mon. Weather Rev., 136, 4045-4062. Wang, J., R. L. Bras, and E. A B. Eltahir (2000), The impact of observed deforestation on the mesoscale distribution of rainfall and clouds in Amazonia, 1. Hydrometeorol., 1, 267-286. Werth, D., and R. Avissar (2002), The local and global effects of Amazon deforestation, J. Geophys. Res., 107(D20), 8087, doi:l0.1029/200IJD000717. .
R. Avissar, Rosenstiel School of Marine and Atmospheric Science, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149-1031, USA M. A Silva Dias and P. Silva Dias, Depaltamento de Ciencias Atmosfericas, IAGlUniversidade de Sao Paulo, Rua do Matao 1226, Sao Paulo, SP 05508-900, Brazi1. ([email protected]. usp.br) P. Silva Dias, Laboratorio Nacional de Computa<;ao Cientifica (LNCC), Petropolis, RJ 25651-075, Brazi1.
Evapotranspiration Humberto R. da Rocha Departamenlo de Ciencias Almosfericas, Universidade de Silo Palllo, Silo Palllo, Brazil
Antonio O. Manzi Inslilulo Nacional de Pesquisas da Amazonia, Manaus, Brazil
Jim Shuttleworth Deparlment ofHydrology and WalerResources, University ofArizona, Tucson, Arizona, USA
We review the measurements of latent and sensible heat flux made at seven fl.ux. tower s~tes dUri~~ the Large-S.cale Biosphere-Atmosphere Experiment for tIoplcal humId, transItIonal and semldeciduous forests, floodplain (with celTado), and cerrado ecosystems. Measurements over falTlllands in Amazonia VaIy from 1.2 (for bar~ so~l) to 3 mm d- l , with a reduction in the dlyiseason. Estimates of e:apotranspuatIOn for Amazonia based on atInospheric re,\nalysis are generally hIgh~r than the measurements. Remarkably, for all the seven sites, the mean annual s~nslbl~ heat flux ranged from 20 to 38 W m-2, lower during the wet season and hl~her ~n the late dly season, consistent with the variation of net radiation and SOlI mOls~re. Based on the seasonal evapotranspiration, the sites are divided into two ~~ctIOnal g:o.up~: tropical forest and savanna. At the northern sites (Manaus, S~ntaIem), preCIpItatIOn IS above 1900 mm a-I, monthly evapotranspiration is ~a1rly constant during the wet season, ranges from 2.8 to 3.6 mm d- I, progressively Illcreases along. the dry.season up to 4 mm d- I, and is dominated by net radiation a~d.va~~r de~slty defiCIt. The western semideciduous forest in Rondonia presents sImIlantIes WIth the forest group, with monthly evapotranspiration that varies little but concurrent with. ~et radiation year round, and peaks more exactly in the dlYto-wet seas~n tranSItIon. At the southern and eastern sites, precipitation is below 1700 mm a ~'I ~easonal evapotranspiration is limited by soil moisture, ranges from 3 to 4~. d III the wet season, and decreases in the dry season to 2.5 mm d- l in ~he tran~ltlOnal forest (Mato Grosso) and floodplain (Tocantins), and to 1 mm d- l III the Sao Paulo cerrado.
1. INTRODUCTION Amazonia and GIobal Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000744
Evapotranspiration influences rainfall through the process recycling and, particularly via this mechalllsm III the tropics, also regional patterns of air temperature o~ atr~lOspheric
261
DA ROCHA ET AL.
262 EVAPOTRANSPIRATION and humidity, and soil moisture. Its importance properly lies within the context of the surface energy partition because it is the combination of surface sensible heat flux and latent heat fluxes Which controls the humidity and thermal stability of the planetary boundaty layer and thus the triggering of convective rainfall. As well, evapotranspiration controls ecosystem functionality because transpiration is linked to the photosynthetic activity of the canopy. Several attempts to define the annual area-average surface latent heat flux (or evapotranspiration) of tropical forests in Amazonia based on model reanalysis data (i.e., retrospective analysis of general circulation model simulations, with assimilation of available observational data) or climate simulations carried out in the context of investigating the climatic effects of the large-scale deforestation in Amazonia, present I results that vary widely from 2.7 to 5.2 mm d- , with an b
10
8
~
1U
E :;:; 6
~ 0
1ti .....
4
0
I-
2
o 2
345
6
Mean annual evaporation in Amazonia calculated by models (mm d- 1)
Figure 1. Overview of estimated values ofthe mean annual evaporation for the whole Amazon basin calculated by mass and energy balance models and global model reanalysis data (in mm d- 1). The histogram columns include the values given by the following sources: (a) Polcher [1995], Villa Nova et al. [1976]; (b) Polcher et al. [1994a], Willmott et al. [1985], Lean and Rowntree [1993], Matsuyama [1992], Molion [1975], Baumgartner [1975], Zhang and Henderson-Sellers [1996], Vorosmarty et al. [1989], Marques et al. [1980]; (c) Diclanson and Kennedy [1992]; Manzi [1996]; Costa and Foley [1999], Sud et al. [1982], Nizhizawa and Koike [1992]; (d) da Rocha et al. [2004]; Leopoldo [2000], Marengo [2005], Marengo et al. [1994], Russell and Miller [1990], (e) Franken and Leopoldo [1984], Nobre et al. [1991], Leopoldo et a1. [1982]; Zeng [1999]; (f) Jordan and Heuveldop [1981], Polcher et al. [1994b].
average value of 3.9 mm d- I and with a standard deviation among 29 available estimates of 0.7 mm d- I (see Figure 1). Marengo [2005] reports using different estimates of precipitation, evaporation, and runoff can result in an imbalance in the regional water budget of up to 50% and estimates that the mean hydrological cycle of the Amazon basin has regional precipitation of 2117 mm a-I, runoff of 511 mm a-I and evapotranspiration of 1570 mm a-I. Marengo [2005] and Zeng [1999] used global reanalysis data to estimate that the mean evaporation in the basin is 4.3 mm d- I. This value is substantially larger than suggested by the existing field obI servations made using flux towers which suggest 3.5 mm dI for Manaus at Ducke [Shuttleworth, 1988], 3.5 1l11ll d- for I Santarem at K83 [da Rocha et al., 2004], 3.7 mm d- for I Rondonia at Janl [von Randow et al., 2004], and 3.l1l11ll dfor Santarem at K67 [Hutyra et al., 2007]. Over continental surfaces, a proportion of the regional evapotranspiration (E) is recycled in the atmosphere to fall as precipitation (P) elsewhere over the continent. Early studies, including Oltman [1967], Molion [1975], and Marques et al. [1980], estimated the EIP ratio to be around 50%, but only part of the regional evapotranspiration will contribute to the basin-wide precipitation by supplementing the incoming horizontal humidity flux into the basin, so the proportion of evaporated water recycled is less than the EIP ratio. Nonetheless, qualitative discussion based on isotopic measurements [Salati et al., 1979] and quantitative analysis [Eltahir and Bras, 1994] both suggest there is a substantial contribution of evaporation to local atmospheric water vapor which increases inland. Estimates of the mean areal recycling over Amazonia based on interpreting reanalysis data indicate values in the range 20-27%, (see, e.g., Costa and Foley, [1999], Eltahir et al. [1996], Bosilovich and Chern [2006], 2005, and Brubaker et al. [1993], who report 20%, 25%, 27%, and 24%, respectively). Lettau et al. [1979], Eltahir et al. [1994], and Trenberth et al. [2003] reported that recycled precipitation increases downwind in Amazonia from about 10% in the east to as much as 50% in the southwestem sectors Of the basin. The fact that precipitation recycling is significant in Amazonia suggests that complex biosphereatmosphere interactions may exist, with surface processes, especially evapotranspiration, playing a significant role in the regional hydrological cycle and affecting spatial pattems of soil moisture and productivity and the occurrence of extreme events such as floods and droughts. The potential impact of forest to pasture conversion on temperature and precipitation pattems in Amazonia as a result of changes in the surface energy and radiation budgets [Gash et al., 1996] remains an important and poorly understood question. Field observations have shown evapotranspiration from fannlands is generally less than for rainforest
and that it varies with season. Estimates for pastureland ValY, from 2.2 to 1.9 1lll11d- 1 in the wet and dly seasons near Santarem [Sakai et al.,'2004], 2.9 to 2.2 mm d- 1 in the wet and dry seasons in/Rondonia [von Randow et al., 2004], and 2.5 to 2.2 mrp.i'd- I during the dly season near Manaus [Wright et aI., 1992; da Rocha et al., 1996], while Sakai et al. [2004] report evapotranspiration of 2.7 mm d- l during the wet season for a Santarem rice crop. In this chapter, we review measurements of evapotranspiration made at several flux tower sites for a range of ecosystems along a large-scale biome and water balance gradient, which included tropical humid and semideciduous forest, transitional forest, floodplain, and cerrado sensu stricto (s.s.). We compare pattems between sites in the long-term average of data collected during the Large-Scale BiosphereAtmosphere Project, with focus on the observed seasonality and environmental and climate controls. 2. SITE DESCRIPTION AND DATA The field measurements reviewed in this study were made at seven flux towers in Brazil. A concise biogeographical characterization of these sites is given in Table 1, imd their location, photograph, and seasonal climates are illustrated
263
in Figure 2. Three sites are tropical terra finne humid forests near 3°S, respectively, Manaus-K34, Santarem-K67, and Santarem-K83, hereinafter referred to as K34, K67, and K83, respectively. Three other sites are nearly 10 0 S, respectively the Rondonia-Jaru site, surrounded by tropical semi·· deciduous forest, hereinafter referred to as JRU; the Mato Grosso-Sinop site, located in transitional tropical forest, hereinafter referred to as SIN; and the Tocantins-Javaes site, hereinafter referred as JAV, classified as floodplain which included a mixture of cerradao (tall woodland savanna), cerrado and campo (natural grassland). The Sao Paulo-Pe deGigante site, hereinafter referred to as PEG, is surrounded by cerrado s.s. and is in the subtropical moist climate zone with a 4-month dly season at 2l oS. Although the JRU and SIN sites are much further south of 3oS, they belong to the same climate zone as the sites near Santarem, i.e., have a tropical moist climate with 3- to 4-month dly season (see Figure 2). On the other hand, the K34 site is in a tropical moist climate zone with a 2-month dly season. The JAV site lies in a floodplain region about 1 km east of the Javaes river and has seasonal flooding between JanualY and May lasting typically 3 to 5 months. This site is in the tropical moist climate zone, but is close to the transition to subzones that have dly seasons of about 5 months.
Table 1. Description of the Flux Tower Sites in Brazil Site lD
K34 K67 K83 JRU
SIN
JAY
Site (State, City, Local) Amazonas, Manaus K34 Para, Santarem_K67 Para, Santarem K83 Rondonia, JiParana Jan] MatoGrosso, Sinop
Tocantins, Pium, Javaes river at Bananal Island PEG Sao Paulo, SantaRita, Pe deGigante
Reference"
(Lat oS, Lon OW)
Mean Tower DIy Season N0l111al Tower Elevation, Height, Length, Precipitationb, Temp,oC month m m mm
(02.60, 60.20)
130
50
2
2286
25.9
2
(02.85, 54.95)
130
63
4
1911
25.3
3
(03.01,54.97)
130
64
4
1911
25.8
4
(10.08,61.93)
191
60
3-4
2173
25.3
5
(11.24,55.19)
423
42
5
1694
26.4
6
(09.82,50.15)
120
42
5
1755
24.5
7
(21.61,47.64)
690
21
6
1478
22.5
Ecosystem Type Tropical humid forest Tropical humid forest Tropical humid forest Tropical semideciduous forest Transitional forest (Cerradao) Floodplain (CerradaoCerradoGrassland) Cerrado s.s. (woodland savanna)
aSee references for detailed descriptions of sites: I, Armijo et al. [2002]; 2, Saleska et al. [2003] and Hutyra et al. [2007]; 3, Goulden et al. [2004]; 4, von Randow et al. [2004]; 5, Vourlitis et a1. [2002]; 6, Borma et al. [2009]; and 7, da Rocha et al. [2002]. Source: adapted from da Rocha et al. [2009]. bNOlmal annual precipitation at nearest climatological station [INMET, 1994].
264
DA ROCHA ET AL.
EVAPOTRANSPIRATION
~ ~
~
~
~
<'l
~ E
~
~
{:l
.~
z
ro
..,i; :§:
:§::
E
~
~ ~ I:<
a"
f/)
c:
<>
g
~
i l
~
~
oS
<>
~
1i
~
~ ::2.
l!!
oS
~
'til
,til
E
l!l
~
I:<
~
~
~ ~
l;;
"
.><
a"
E
l!!
]j
~
~ '"N ....c:
~
'"
"~
(f)
e <>
g
:e
<>
N
"
a"
'"
l:ll~
,- (!) (!)w
]j
(l)o. 'tJ~
~
~
'
"a
0.
'if
o
oS
""
'.
-.'I:<
'"
N
l!l
f/)
a"
:::J
]l
'c:" ::iii :§:
oS
""'"
"lil
" ::l
"
"a "~
~
~
t:l I'l
Z
~
~
'"
'if
(l
~
~
"~
§
.
.~
~ ~
(l)
~
<>
The measurements of climate and turbulent fluxes were made using sensors ll}ounted above the canopy near the top of a micrometeorololfical tower at each study site (the tower heights are given Table 1). This analysis used the energy fluxes measurectilsing the eddy covariance method and calculated from the 30-min covariance of vertical wind velocity and with temperature for the sensible heat flux (H), and water vapor for the latent flux (LE). Additional environment measurements included the 30-min averages of air temperature, net radiation (R n), soil heat flux (G), and accumulated precipitation. Monthly mean values of all these variables were calculated for data collected for site-specific periods whenever available between 1999 and 2006 [da Rocha et al., 2009]. Climate records of surface air temperature and precipitation were available from the nearby climate stations for the 30-year period 1961-1990 [INMET, 1994]. Additional information on the instruments used for data acquisition, local patterns of climate, and turbulent fluxes of energy and CO 2 have been reported elsewhere (see Table 1).
~
"<> N
~
3. TEMPERATURE AND PRECIPITATION Overall, the seasonal variation of air temperature is well correlated with incoming solar radiation at all sites and regulated by cloud cover and the large-scale subsidence. The measurements of monthly mean air temperature measured at the top of the tower (Figure 2) peaked in October at the sites K34, K67, and K83, consistent with the long-term temperature cycle measured at the nearest climate stations. The month ofOctober is the late dry season in the Manaus region, and the middle of the dry season in the Santarem region. The tower-top and climate station temperature records at JRU, SIN, and JAV both show a bimodal annual variation, with a minimum in June/July around the winter solstice and a second minimum during December/January when precipitation is greatest (Figures 2d-2f). The measured tower-top temperatures are in most cases similar or a little warmer (by less 2°C) than the long-term average climate station values, the difference generally being somewhat less (~1 0c) in the wet season. The K34 and SIN sites are the exceptions; there, the
265
tower-top temperature is relatively cooler (by less than 2°C) (Figures 2a and 2f). The normal annual precipitation (Table 1) is above 2100 mm at the K34 and JRU sites, is less at the K67 and K83 sites (1911 mm), reduces to ~ 1700 mm at the JAV and SIN sites, and is least and equal to 1478 mm at the PEG site. Overall, the annual precipitation decreased and the length of the dty season increased, from west to east and north to south. From 3° to 1O o S, the dty season at the K67 and K83 sites is comparable in length to that at the JRU, although later in the year, whereas at the SIN and JAV sites, the dry season is longer than those (see shaded bars in Figure 2). As might be expected, given its subtropical climate, the dty season at the PEG site is earlier, longer, and cooler than at the other sites, with 6 months of below 60 mm per month rainfall from April to September (Figure 2f). 4. SEASONAL VARIABILITY The seasonal variability is interpreted with the mean monthly net radiation, sensible heat flux, and latent heat flux (for simplicity, hereinafter referred to as evaporation). One consistent pattern appears common to a first set of sites K34, K67, K83, namely, the evaporation is reduced in the wet season and higher in the dty season, coincident with the pattern of net radiation (Figures 3a-3c). This suggests that net radiation is a s~rong control on the evaporation over the tropical forests at! 3°S, a result which corroborates previous studies [e.g., Shuttleworth et al., 1988; da Rocha et al., 2004]. The peak evaporation in the dty season is roughly between 110 and 120 W m-2 and generally occurs in October at these sites, but it is earlier at the K83, in August (Figure 3c). In general, the minimum evaporation in the wet season remains fairly constant between 80 and 110 W m-2 (Figures 3a-3c) then rises in the dry season. The increase in evaporation begins in June/July at the K34, K67, and K83 sites (Figures 3a-3c). The sites in Manaus and Santarem are well-established forests, presumably deep-rooted, where soil moisture depletion apparently does not limit transpiration [Bruno et al., 2006]. Transpiration and photosynthesis
]l
Cl.
0
c:
iii
~11l l:l
~
'0
N
~
E:
t:l I'l ~
z
! ]l
f
a
g
g
'"
N
~
"~
~
~
~
Figure 2. (opposite) Measured precipitation, in mm month-1 (white bar), top tower temperature, in °C (dashed line), sUlTounding vegetation and used tower (pictures), at the flux tower sites in Brazil, specifically (a) Manaus, K34 with scaffolding tower, (b) Santarem, K67 with triangle tower (left) and vegetation around (middle), (c) Santarem, K83 with triangle tower (left) and vegetation (right), (d) Rondonia, Jaru (aerial photograph), (e) Tocantins, Javaes at Bananal Island with scaffolding tower (left), Cerradao vegetation around tower (middle top) and Cen'ado vegetation with natural grassland 1 km apart oftower (middle bottom), (f) Mato Grosso, Sinop with scaffolding tower (left) and vegetation (bottom), (g) Sao Paulo, Pe de Gigante with ,scaffolding tower (left), and surrounding vegetation in summer (middle top) and winter (middle bottom). The shaded bar is the 30-year average precipitation measured at the nearby climate station, The solid line is the 30-year average temperature measured at the nearby climate station; (h) central map of Brazilian Amazonia with the dry season length (in months) for the Thomthwaite Tropical moist climate type. Source: adapted from da Rocha et al. [2009].
266
DA ROCHA ET AL.
EVAPOTRANSPlRAnON
(a) Manaus (K34)
(Rn, LE, H)
(Rn, LE, H) (b) Santarem (K83)
150
150
120
120
(Rn, LE, H)
(e) Santarem (K67)
15 L
.
~,p·
,90 r - -_ _6
3 Jon Mar May Jul
Jan
Sap Noy
(f) levees (JAV)
(h)
KG7 & Ksa trop fore~t
pA
lOS
:JAV
~Ioodplain
Jan Mar May Jul
Sep NoY
Climatological dry season 205
_W,_'~_PP0D Rn Net Radiation
(W m-2) (W m-2)
- - - LE Latent Heat Flux _ .. H Sensible Heat Flux (W m- 2)
(d) Jaru (JRU)
(Rn, LE, H) l::Ju~~
.....•
P
120
(Rn, LE, H)
(g) Pe deGlgente(PEG)
(e) Sinop (SIN)
150
12
120
1..---..
90-
90
60
60
30 Jan Mar May Jul
Sep Noy
Jan Mar May
.H
Jul
Sep Nov
Jan Mar May
may be limited in dry years [Malhi et aI., 2002], more likely eddy covariance technique at different sites and little sysin long-term droughts.exceeding 6 months [see Meil' et al., tematic seasonal difference in the error with which measurethis volume]. ments are made at the same site. This is not necessarily the At the JRU sitefalso, the evaporation peaks in October, case, and it is important to introduce a note of caution when which, however)1s more specifically the transition to the drawing conclusions from these data because such systemearly wet seas011 (Figure 3d), and the increase in evaporation atic errors may impact interpretation of the results. To take happens in September. The lower evaporation at JRU rela- an example, the measured mean annual evaporation was tive to the other tropical forest sites during most of the dly largest (3.7 mm d- 1) at the K83 site and lowest (2.5 mm season may be because the colder temperatures and drier air d- 1) at the PEG site, while the mean millual evaporation at in the winter (Figure 2d) reduce the surface conductance and JRU (2.8 mm d- 1) was the lowest among the values reported the canopy photosynthesis [see Saleska et al., this volume] for the tropical forest sites. To address how systematic siteand thus inhibit transpiration, prior to the onset of the wet to-site differences in energy balance closure influence these season. As well, net radiation is least in June: the effect of values, an index of the energy budget closure is presented as the winter solstice at lOGS significantly reduces the top ofthe the ratio [(H + LE) / (R n ~ G)], with the use of the monthly atmosphere, incoming solar radiation, notwithstanding the averages for each individual term. This index neglects other effect of cloud cover. The observed increase in net radia- possible contributions to the energy budget, including ention during the dry season is not well reflected in increased ergy stored in biomass and canopy air below the measureevaporation at JRU site. ment height, the chemical energy associated with net CO 2 On the other hand, for a second set of sites, SIN, JAV, and exchange, and any energy advected horizontally by the wind. PEG, the mean monthly evaporation increases with the onset Figure 4 shows the alillual variation at different sites of the of and along the wet season and decreases progressively dur- index of closure and demonstrates that, on average, the sum ing the dly season (Figures 2e~2g). The peak evaporation at of the turbulent fluxes of sensible and latent heat fluxes corthose sites are between 110 to 120 W 111-2 , and the minima respond to 70% to 100% of the available energy depending are of ~70 W m-2 for the SIN and JAV, and particularly less on the site. The PEG and JRU sites have the minimum cloand of ~ 30 W m-2 for the PEG site. This seasonal pattern sure index (~0.7), tfie K34 and K67 sites have an intermediis opposite to that for tropical forest but consistent with the ate closure index (":'0.85), while the JAV and K83 sites have seasonality in the net radiation as was that case. However, a the largest closure index (between ~0.85 and ~ 1.1). The cloparticular exception to such a seasonal pattern is at the JAV sure index at the JAV site has the largest seasonal variation, site, where net radiation progressively increases during the increasing to a maximum of ~ 1.1 in the high wet season, a dly season (Figure If), that is, similar to what is shown at period that coincides with the peak flooding within the fetch the sites near 3°S. A feature worthy of note is that, at all sites, the mean monthly sensible heat flux is low during the wet season but 1.1 increases during the middle to the late dly season to reach G' a maximum (Figure 3). This pattern is consistent with that in net radiation and the increasing soil moisture depletion #,_ ..-. which happens almost 1 month later than minimum rainfall ~ 0.7 (Figure 2). Although the mean annual sensible heat flux is in + :r: the range 20 to 40 W m-2 at all sites, the dly season st:nsible '-" 0.5 I I I I I I I I I I I I heat flux is consistently and substantially larger than that in Jan Mar May Jul Sep Nov the wet season, with seasonal differences between 19% and 44% [see cia Rocha et al., 2009]. At all the sites, the mean monthly sensible heat flux is lower than the latent heat fluxes Figure 4. Annual variation in the index of closure defined as [(H + (i.e., the Bowen ratio is less than 1) except for the PEG site LE) / (R n ~ G)] and calculated from mean monthly latent heat flux (LE), net radiation (R n ), sensible heat flux (H), and soil heat flux during the late dry season.
~ 0.9~~"
(Rn, LE, H)
150
267
Jul
Sap Noy
Figure 3. Mean monthly latent heat flux (LE) (black s?lid ne), ~et radiation (R) (gray solid line), and ~,ensible heat flux (H) (dashed line), alI in W m-2 for the flux tower sites In Braztl, namely, (a) Manaus, K~4, (b) San~arem, K83, (c) Santarem, K67, (d) Rondonia, Janl, (e) Mato Grosso, Sinop, (f) Tocantins, Javaes, (g) .Santa Rita, Pe deGlgante, (h) map of Brazil with the experimental sites (black triangles). The climatological dry season IS shaded. Source: da Rocha et al. [2009].
(G) measured at the six flux tower sites in Brazil, which are labeled
5. COMPARISON BETWEEN SITES All of the previous discussion is made assuming that there are no significant systematic differences in the error with which measurements of energy fluxes are made using the
as folIows: forest Santarem K83 (black solid line), forest Santarem K67 (gray solid line), forest Manaus K34 (black dashed line), forest Rondonia Janl JRU (gray dashed line), transitional forest Sinop SIN (gray dashed line with circle), floodplain-Cel1'ado Tocantins Javaes JAV (black solid line with circle) and Cerrado Sao Paulo, Pe deGigante (black dashed line with circle).
268
EVAPOTRANSPIRAnON
DA ROCHA ET AL.
of tower when there may be an enhanced contribution to the locally measured fluxes from horizontal advection. Figure 5a shows the annual cycle of monthly mean net radiation at an the study sites. The mean monthly net radiation at the PEG site is anomalous in that it changes through the year with the largest solar zenith angle likely being the strongest control. At all of the other sites between 3° and 10 S, the mean monthly net radiation increased between May and October, a period that approximately coincides with the onset and end of the dry season. It 'is interesting that, despite the fact that there are variations in incoming solar radiation at the top of the atmosphere within this latitude belt, the regional distribution of cloudiness and local surface conditions (e.g., albedo and emissivity) appear to work together to provide a comparable across-region evolution in net radiation through the dry season. However, during the wet season, the mean net radiation varies considerably between the sites, with the greatest seasonal amplitude (~11 0 to 150 W m-2) near Santarem (at both the K83 and K67 sites) and lowest at the JRU site (~130 to 150 W m-2 ). To minimize the impact of different site-to-site energy closure on comparisons of the seasonality of energy fluxes, 0
180
the yearly variation in the ratio between the mean monthly flux and the maximum mean monthly flux for LE and H, respectively, is shown in Figures 5b-5e. The ratio [LE / LEmaxl shows that evaporation at the rainforest sites (K34, K67, K83, and JRU) in the dry season reflects the control of net radiation (Figure 5b) in that it progressively increases after the late wet season (roughly in May) through the dry season and eventually peaks between September and October. During the wet season, the ratio [LE / LE max ] at these sites is minimum and varies between 0.75 and 0.9. At the rainforest sites, the annual variation in [H / H max ] is similar to that for [LE / LEmax] with a maximum around September and is minimum in the range 0.6 to 0.75 (Figure 5d). In comparison with [LE / LEmax ], the ratio [H / H max ] appears more consistent between sites, and its seasonal variation is better defined. The decline in [H / H max ] after September is more abmpt than its increase into the dly season and more abmpt than the decline in [LE /LE max ] at this time. Soil moisture depletion likely plays a role in this variation, as it generally progresses fairly smoothly in long dly spells and reduces water loss fi'om the soil evaporation with little impact on the rainfall interception loss. On the other hand,
(a)
_
k83 kG? k34 JRU JAV SIN PEG
Tropioal humid for.st 03'S Tropical humid forest 03'S Tropical humid forest 03'S Semi deciduous forest 10'S Floodplain (Cerrado) 10'S Transitional foros! 10'S Cerrado sensu slr/clo 20'S
I I I I I I I I I I I I Jan Mar May Jul Sep Nov
the readily availability ofsoil moisture in shallow upper layers after early rainfall..events [Bruno et al., 2006], together with a flush of green leaves before the onset of the wet sea2004] will tend to increase the surface son [Goulden et conductance, ar~itablychanging the balance between LE and H in such i way that there is a more rapid decrease in sensible heat flux. At the other group of sites (SIN, JAV, and PEG), the pattern of [LE /LEmax ] is quite different to that at the rainforest sites, with maxima between the middle and late wet season (December to March) and minima near the onset of the wet season around September (Figure 5c). On the other hand, the seasonality in the monthly ratio [H / H max ] is broadly similar to that for the rainforest sites (Figure 5d), having, however, a substantial difference in terms of the absolute values, that is, at the PEG site, the ratio [H / H max ] has a minimum value of just ~0.3, the least for all the sites, whereas at the JAV and SIN sites, the annual variation is comparable to that at the rainforest sites.
aV,
6. CONCLUSIONS This chapter reviews the estimates of the evapotranspiration in Amazonia using general circulation models, mass and energy balance models, and local measurements using the eddy covariance technique at several flux tower sites in Amazonia and bordering regions, for a range of ecosystems, including especially tropical humid forest (Manaus
K34, Santarem K83 and K67), semideciduous tropical forest (Rondonia, Jam), transitional tropical forest (Mato Grosso, Sinop), floodplain over cerrado (Tocantins, Javaes at Bananaiisland), and cerrado s.s. (Sao Paulo, Pe de Gigante), and over farmlands in Amazonia. Measurements of mean monthly tower-top air temperature and precipitation at the flux tower sites showed a seasonal pattern that agrees reasonably well with the climatological average values measured at nearby climate stations. This suggests that the experimental conditions, instmmental approach, environmental characteristics, and (albeit limited) period of measurement are such that the results arguably reflect regional characteristics and are not solely reflective ofthe microclimatic conditions present in the fetch of the towers. Estimates of area-average evapotranspiration in Amazonia made using global climate models, and mass and energy balance models, vmy substantially fi'om 2.7 to 5.2 mm d- 1, while estimates of atmospheric reanalysis data suggest a value around 4.3 mm d- 1• These estimates are quite large relative to earlier field observations made using flux towers and are also large with respect to data for the seven study sites reviewed here. Table 2 summarizes the measured mean evapotranspiration with eddy covariance, during the wet and dry seasons, which include existing reports for Amazonian rainforest and farmlands, and savanna. Studies of evaporation over pasture and croplands [Sakai et al., 2004] in Am~zonia have reported flux tower measurements that varied fi'om ~ 1.2 mm d- 1, for bare soil prior to
Table 2. Mean Evaporation Measured Using the Eddy Covariance Method for Ecosystems in Amazonia and Cerrado in the Wet and Dry Season, Respectively, Using the Data Reviewed in This Study Together With Values Taken From Previous Reports" Measured Mean Evaporation, mm d- 1
1
0.7
~ 0.9 E 0.8 ::. 0.7 :I: 0.6
0.6 1
0.5 1
~ 0.9
.5 ..... 0.8
...... ~
~ 0.8
~
E !:!:l 0.6
--..
0.4 0.2
Tropical forest
(e)
:I:
I I I I I I I I I I I I Jan Mar May Jul Sep Nov
Tropical semideciduous forest
Site
Wet Season
Manaus K34 Santarem K67 Santarem K83 Rondonia JRU Mato Grosso SIN(5)
2.8 2.9 3.6 2.8 3.4
Tocantins JAV Sao Paulo PEG Brasilia(6) Santarem(l) Rondonia(2) Manaus(3) Manaus(4) Santarem(l) Santarem(l)
3.8 (Oct-May) 3.0 (Sep-Apr) 3.6 (Jan) 2.2 (Apr-Jun) 2.9 (Jan-Mar)
(Nov-Jul) (Dec-luI) (Dec-Jul) (Oct-May) (Jan-May)
0.8
E :I: 0.6
...... ~
Ecosystem
a
0.4 0.2
@'~e .... ~ ....
"'~""'..5 I I I I I I I I I I I I Jan Mar May Jul Sep Nov
Figure 5. (a) Mean monthly net radiation (R n), and annual variation in ratio of the mean monthly heat flux relative to the maximum mean monthly heat flux in the year for (b and c) the latent heat flux (LE), and (d and e) the sensible heat flux (H) at the flux tower study sites in Brazil, which are labeled as follows: forest Santarem K83 (black solid line), forest Santarem K67 (gray solid line), forest Manaus K34 (black dashed line), forest Rondonia Janl JRU (gray dashed line), transitional forest Sinop SIN (gray dashed line with circle), floodplain-Cerrado Tocantins Javaes JAV (black solid line with circle) and Cel1'ado Sao Paulo, Pe deGigante (black dashed line with circle). Source: da Rocha et al. [2009].
Floodplain (Cerrado) Cerrado s.s. Pastureland
Rice crop Bare soil
DIy Season 3.4 (Aug-oct) 3.3 (Aug-Nov) 3.9 (Aug-Nov) 2.6 (Jun-Sep) 3.0 (transition Sep-oct) 2.4 (Jul-Oct) 3.3 (Jun-Sep) 1.3 (May-Aug) 1.7 (Jul-Sep) 1.9 (Aug-Oct) 2.2 (Jul-Sep) 2.5 (Oct) 3.2 (Jul-Sep)
2.7 (Mar-Jun) 1.2 (Nov-Dec)
"Reports referenced are noted in parentheses: 1, Sak;i et al. [2004]; 2, von Randow et al. [2004]; 3, Wright et al. [1992]; 4, da Rocha et al. [1996]; 5, Vourlitis et al. [2002]; and 6, Miranda et al. [1996].
269
270
EVAPOTRANSPIRAnON
planting, to subsequent rates that are typically less than 3 mm d- 1 which tend to reduce strongly during the dly season (Table Such values and seasonal variation are.consistel:t with the villues calculated for agricultural lands 1ll numencal model experiments investigating large-scale Amazonian
2).
deforestation. The pattern of the evapotranspiration discussed for the seven flux tower sites clearly divides them into two nmctional groups, namely, the tropical forest and sava~na. This simple classification only relies on the fact that ~all1fall. an~ seasonality are the most impOliant controls of blOme distnbution in tropical South America, although other factors are involved, for example, the soil parameters of fertility and nutrients, drainage, and water holding capacity [Lloyd et al., this volume]. For the group of the rainforest sites at nearly 3°S (Manaus and Santarem), the mean monthly evaporation is fairly constant during the wet season and in the range 2.8 to 3.6 mm d- 1. The evaporation progressively increases along the dly season and is about 10% higher than in the wet season (Table 2). At these sites, the dly season length does not exceed 4 or 5 months, annual precipitation is above 1900 mm, and atmospheric conditions of net radiation and vapor density deficit are well correlated and exert the dominant control on the evapotranspiration. The western forest site, in Rondonia at lOGS, is also identified with the former group. The evapotranspiration varies little but is also concunent with net radiation year round and peaks at ~3.0 mm d- 1 more exactly in the dly-to-wet season transition (Table 2). For a remaining group of sites which include transitional forest in Mato Grosso, floodplain cerrado in Tocantins, and cerrado in Sao Paulo, where the dry season length exceeds 4 months and annual precipitation is below 1700 mm, the evapotranspiration has an opposite seasonal pattern to the 1 first group: it varies in the range ~3 to 4 mm d- in the wet season and decreases in the dly season to ,minimum values of 2.5 mm d- 1 in the transitional forest and floodplain, and ~1 mm d- 1 in the cerrado s.s. (Table 2). Such a decrease was likely due to soil moisture limitation, and the marked seasonality is not unexpected because it is known that the typical Brazilian cenado undergoes strong seas~nal va~'i ability in leaf area index and green biomass, assoclate.d wIth the decline in photosynthesis [see Saleska et al., thIS volume; Lloyd et al., this volume], tree leaf senescence, and dOlmancy of grasses.
Acknowledgments. We thank the LBA scientific committee, the Brazilian Ministry of Science and Technology, the Brazilian funding agencies Fapesp, CNPq, and Finep, the NASAlLBA-Eco coordination and staff, and all who contributed with constmctive
DA ROCHA ET AL. work during the last decade, namely, M. Goulden, S. Miller, S. Saleska, S. Wofsy, L. Hutyra, O. Cabral, D. Fitzjarrald, O. Moraes, R. Sakai, O. Acevedo, P. Artaxo, M. Keller, M. Assunyao S. Dias, P. Dias, C. Nobre, A Nobre, 1. Loyd, A. Miranda, H. Miranda, L. BOlma, C. von Randow, F. Cardoso, J. Gash, B. Kruijt, J. Tota, Y. Malhi, J. Grace, G. Vourlitis, H. Freitas, M. Aidar, L. Sa, M. H. Costa, F. Zanchi, R. Avissar, A Araujo, 1. Maia, R. Aguiar, E. Collichio, F. Lobo, 1. Tota, R. Bmno, R. Tallius, L. Oliveira, R. Juarez, D. Kurzatkowski, D. Rezende, and many others. author acknowledges Fapesp (02/09289-9) and CNPq (Instituto do MiIenio-LBA, Ed. Universal-Ol, Ed. CT-Hidro 03) and the Brazilian institutions Universidade Federal do Tocantins, Instituto Florestal de Sao Paulo, Embrapa, Universidade Federal do Para, IBAMA Jim Shuttleworth was funded under NASA contract NNX06AG91G#1.
REFERENCES Araujo, A. C., et al. (2002), Comparative measurements of carbon dioxide fluxes from two nearby towers in a central Amazonian rainforest: The Manaus LBA site, J, Geophys. Res, 107(D20), 8090, doi:lO.l029/2001JD000676. Baumgartner, A. (1975), The World Water Balance, p. 464, Elsevier, New York. Borma, L. S., et al. (2009), Atmosphere and hydrological controls of the evapotranspiration over a floodplain forest in the Bananal Island region, Amazonia, J, Geophys. Res., 114, GOI003, doi: I 0.1 02912007JG000641. Bosilovich, M., and 1. Chern (2006), Simulation of water sources and precipitation recycling for the MacKenzie, Mississippi, and Amazon river basins, J, Hydrometeorol., 7(3), 312. Brubaker, K., D. Entekhabi, and P. Eagleson (1993), Estimation of continental precipitation recycling, J, Clim., 6, 1077-1089. Bruno, R. D., H. da Rocha, H. Freitas, M. Goulden, and S. Miller (2006), Soil moisture dynamics in an eastern Amazonian tropical forest, Hydrol. Processes, 20, 2477-2489. Costa, M. H., and J. A Foley (1999), Trends in the hydrologic cycle of the Amazon basin, J. Geophys. Res., 104, 14,189-14,198. da Rocha, H. R., C. A Nobre, J. P. Bonatti, 1. R. Wright, and P. 1. Sellers (1996), A vegetation-atmosphere interaction study for Amazonian deforestation using field data and a single column model, Q. J, R. Meteorol. Soc, 122, 567-598. da Rocha, H. R., H. C. Freitas, R. Rosolem, R. 1. N. Juarez, R. N. Tamms, M. V. Ligo, O. M. R. Cabral, and M. A F. Silva Dias (2002), Measurements of CO 2 exchange over a woodland saval1lla (Cerrado Sensu stricto) in southeast Brasil, Biota Neotropica, 2(1). (Available at http://www.scielo.br/scielo.php/ script_sci_serial/lng-ptlpid_1676-0603/nrm_iso) da Rocha, H. R., M. Goulden, S. Miller, M. Menton, L. Pinto, H. Freitas, and A. S. Figueira (2004), Seasonality of water and heat fluxes over a tropical forest in eastern Amazonia, Ecol. Appl., 14(4), S22-S32. da Rocha, H. R., et al. (2009), Patterns of water and heat flux across a biome gradient from tropical forest to savanna in Brazil, J, Geophys. Res., 114, GOOBI2, doi:lO.102912007JG000640.
Dickinson, R. E., and P. Kennedy (1992), Impacts on regional climate of Amazon deforestation, Geophys. Res. Lett., 19(19), 1947-1950. Eltahir, A, and R. BIfls (1994), Precipitation recycling in the Amazon basin, Q. J,l} Meteoro!' Soc., 120, 861-880. Franken; W., andP. Leopoldo (1984), Hydrology of catchment areas in central-Amazonian forest streams, in The Amazon, Limnology and Landscape Ecology ofa Mighty Tropical River and its Basin, edited by H. Sioli, pp. 501-519. Gash, 1. C. H., C. A. Nobre, J. M. Roberts, and R Victoria (1996), Amazonian Deforestation and Climate, pp. 1-14, John Wiley, Chichester, U. K. Goulden, M., S. Miller, H. Rocha, M. Menton, and H. Freitas (2004), Physiological controls on tropical Forest CO 2 exchange, Ecol. Appl., 14(4), S42-S55. Hutyra, L. R, 1. W. Munger, S. R. Saleska, E. Gottlieb, B. C. Daube, A. L. Dunn, D. F. Amaral, P. B. de Camargo, and S. C. Wofsy (2007), Seasonal controls on the exchange of carbon and water in an Amazonian rain forest, J, Geophys. Res., 112, G03008, doi: 10.102912006JG000365. INMET (1994), Normais Climatol6gicas de Supeljicie 1931-1990, Instituto Nacional de Meteorologia, Brasilia, DF, Brasil. Jordan, C., and 1. Heuveldop (1981), The water balance of an Amazonian rain forest, Acta Amazonica, 11,87-92. Lean, 1., and P. Rowntree (1993), GCM simulation of the impact of Amazon deforestation on climate using an improved canopy representation, Q. J, R. Meteorol. Soc., 119, 509-530. Leopoldo, P. (2000), 0 ciclo hidrologico em bacias experimentais da Amazonia central, in Amazonia: um Ecossistema em Transformar;{io, edited by E. Salati, M. Absy, and R. Victoria, pp. 87-177, INPA, Manaus. Leopoldo, P., W. Franken, E. Matsui, and M. Ribeiro (1982), Estimativa da evapotranspirayao da floresta amazonica de terra firma, Acta Amazonica, 12,23-28. Lettau, H., K. Lettau, and L. Molion (1979), Amazonia's hydrologic cycle and the role of atmospheric recycling in assessing deforestation effects, Mon. Weather Rev., 107, 227-238. Lloyd, 1., M. L. Goulden, J. P. Ometto, S. Patino, N. M. Fyllas, and C. A. Quesada (2009), Ecophysiology offorest and saval1lla vegetation, Geophys. Monogr. Ser., doi: IO.l029/2008GM000740, this volume. Malhi, Y., E. Pegoraro, A D. Norbe, M. G. P. Pereira, 1. Grace, A D. Culf, and R. Clement (2002), Energy and water dynamics of a central Amazonian rain forest, J, Geophys. Res., 107(D20), 8061, doi: lO.l029/200IJD000623. Manzi, A (1996), A simulation of Amazonian deforestation using a GCM calibrated with Abracos and Anne data, in Amazonian Deforestation and Climate, edited by J. H. C. Gash et aI., pp. 505--':530, John Wiley, Chichester. Marengo, J. (2005), Characteristics and spatio-temporal variability of the Amazon river basin water budget, Clim. Dyn., 24, . 11-22. Marengo, 1., et al. (1994), Calculations of river-runoff in the GISS GCM: Impact of a new land surface parameterization and runoff routing on the hydrology of the Amazon river, Clim. Dyn., 10, 349-361.
271
Marques, 1., E. Salati, and 1. Santos (1980), Calculo da evapotranspirayao real nabacia amazonica au'aves do metodo aerologico, Acta Amazonica, 10, 357-361. Matsuyama, H. (1992), The water budget in the Amazon river basin during the FGGE period, J, Meteol'o!. Soc. Jpn., 70, 10711083. Meir, P., et al. (2009), The effects of drought on Amazonian rain forests, Geophys. Monogr. Sel'., doi: I 0.1 02912008GM000882, this volume. Miranda, A., et al. (1996), Carbon dioxide fluxes over a cerrado sensu stricto in central Brazil, in Amazonian Deforestation and Climate, edited by J. H. C. Gash et aI., pp. 353-363, John Wiley, Chichester. Molion, L. (1975), A climatonomic study of the energy and moisture fluxes of the Amazon basin with consideration of deforestation effects, PhD thesis, Univ. of Wisconsin, Madison. Nizhizawa, T., and Y. Koike (1992), Amazon Ecology and Development, ]wanami, Tokyo, p. 221. Nobre, C. A, P. 1. Sellers, and 1. Shukla (1991), Amazonian deforestation and regional climate change, J, Clim., 4,957-987. Oltman, R. E. (1967), Reconnaissance investigation of the discharge and water quality of the Amazon basin, Biota Amazonica, 3,163-185. Polcher,1. (1995), Sensitivity ofu'opical convection to land surface processes, J, Atmos, Sci., 52, 3143-3161. Polcher, 1., and K. Laval (1994a), The impact ofAfrican and Amazonian deforestatiqn on tropical climate, J, Hydrol., 155, 389405. Polcher, 1., and K. La,val (1994b), A statistical study ofthe regional impact of deforestation on climate in the LMD GCM, Clim. Dyn., 10,205-219. Russell, G., and 1. Miller (1990), Global river runoff calculated fi'om a global atmosphere general circulation model, J, Hydrol., 155,241-254. Sakai, R., D. Fitzjarrald, O. Moraes, R Staebler, O. Acevedo, M. J. Czikowsky, R da Silva, E. Brait, and V. Miranda (2004), Landuse change effects on local energy, water, and carbon balances in an Amazonian agricultural field, Global Change BioI., 10(5), 895-907. Salati, E., A Dall'olio, E. Matsui, and 1. R. Gat (1979) Recyling of water in Amazon basin: An isotopic study, Water Resoll/·. Res., 15, 1250-1258. Saleska, S., H. da Rocha, B. Krnijt, and A Nobre (2009), Ecosystem carbon fluxes and Amazon forest metabolism, Geophys. Monogr. Ser., doi:lO.l029/2008GM000728, this volume. Saleska, S. R., et al. (2003), Carbon in Amazon forests: Unexpected seasonal fluxes and disturbance-induced losses, Science, 302, 1554-1557. Shuttleworth, W. 1. (1988), Evaporation from Amazonian rain forest, Proc. R. Soc. London, Ser. B, 233, 321-346. Sud, Y., et al. (1982), Biogeophysical consequences of a u'opical deforestation scenario: A GCM simulation study, J. Clim., 9, 3,226-3247. Trenberth, K., A Dai, R. Rasmussen, and D. Parsons (2003), The changing character of precipitation, Bull. Am. Meteorol. Soc., 84, 1205, doi:l0.l175/BAMS-84-9-1205.
272
EVAPOTRANSPIRATION
Villa Nova, N., E. Salati, and E. Matsui (1976), Estimativa da evapotranspirayao na bacia amazonica, Acta Amazonica, 6, 215228. von Randow, C., et al. (2004), Comparative measurements and seasonal variations in energy and carbon exchange over forest and pasture in southwest Amazonia, TheaI'. Appl. Climatol., 78, 5~26, doi: 10.1 007/s00704-004-0041-z. Vorosmarty, C. 1., B. Moore III, A. L. Grace, M. P. Gildea, J. M. Melillo, B. 1. Peterson, E. B. Rastetter, and P. A. Steudler (1989), Continental scale models of water balance and fluvial transport: An application to South Ame~'ica, Global Biogeochem. Cycles, 3,241-265. Vourlitis, G. L., N. P. Filho, M. M. S. Hayashi, J. de S. Nogueira, F. T. Caseiro, and 1. H. Campelo Jr. (2002), Seasonal variations in the evapotranspiration of a transitional tropical forest of Mato Grosso, Brasil, Water Resour. Res., 38(6), 1094, doi:10.10291 2000WROOO 122. Willmott, C., et al. (1985), Climatology of the terrestrial seasonal water cycle, J. CUmatol., 5, 589-606.
Wright, 1. R., 1. H. C. Gash, H. R. da Rocha, W. 1. Shuttleworth, C. A. Nobre, G. T. Maitelli, C. A. G. P. Zamparoni, and P. R. A. Carvalho (1992), Dry season micrometeorology of central Amazonian ranchland, Q. J. R. Meteol'ol. Soc., 118,1083-1099. Zeng, N. (1999), Seasonal cycle and interannual variability in the Amazon hydrologic cycle, J. Geophys. Res., 104, 9097-9106. Zhang, H., and A. Henderson-Sellers (1996), Impacts of tropical deforestation. Pali I: Process analysis oflocal climate change, J. CUm.,9,1497-1517.
H. R. da Rocha, Departamento de Ciencias Atmosfericas, lAG, Universidade de Sao Paulo, Rua do Matao, 1226 Cidade Universitaria, Sao Paulo, SP CEP 05508-090, Brazil. (humberto@model. iag.usp.br) A. O. Manzi, Instituto Nacional de Pesquisas da Amazonia, Manaus, AM CEP 69060-001, Brazil. 1. ShuttlewOlih, Department ofHydrology and Water Resources, University of Arizona, Tucson, AZ 85721, USA.
Global Warming and Climate Change in Amazonia: Clilnate-Vegetation Feedback and Irnpacts on Water Resources Jose Marengo, 1 Carlos A. Nobre,1 Richard A. Betts,2 Peter M. COX,2,3 Gilvan Sampaio, 1 and Luis Salazar l This chapter constitutes an updated review of long-term climate variability and change in the Amazon region, based on observational data spanning more than 50 years of records and on climate-change modeling studies. We stati with the early experiments on Amazon deforestation in the late 1970s, and the evolution of these experiments to the latest studies on greenhouse gases emission scenarios and land use changes until the end of the twenty-first century. The "Amazon dieback" simulated by the HadCM3 model occurs after a "tipping point" ofCO 2concentration and warming. Experiments on Amazon deforestation and change ofclimate suggest that once a critical deforestation threshold (or tipping point) of 40-50% forest loss is reached in eastemAmazonia, climate would change in a way which is dangerous for the remaining forest. This may favor a collapse of the (tropical forest, with a substitution of the forest by savanna-type vegetation. The concept of "dangerous climate change," as a climate change, which induces positive feedback, which accelerate the change, is strongly linked to the occurrence' of tipping points, and it can be explained as the presence of feedback between climate change and the carbon cycle, patiicularly involving a weakening of the CUlTent terrestrial carbon sink and a possible reversal from a sink (as in present climate) to a source by the year 2050. We must, therefore, cUlTently consider the dlying simulated by the Hadley Centre model(s) as having a finite probability under global warming, with a potentially enormous impact, but with some degree of uncertainty; 1. INTRODUCTION
thropogenic climate change across the world. This change, which seems to be more important than natural climate variAccording to the Fourth Assessment Report of the Inter- ability, has been affecting climate and the hydrologic cycle governmental Panel on Climate Change (IPCC AR4), over and extremes with impacts on the availability of global and the last 50 years, there has been an especially intensive an- regional water resources. Amazonian rainforest plays a crucial role in the climate system, helping to drive atmospheric lCenh'o de Ciencias do Sistema Terrestre/Instituto Nacional de circulations in the tropics by absorbing energy and recycling Pesquisas Espaciais, Cachoeira Paulista, Brazil. about half ofthe rainfall that falls on it. Previous studies have 2Met Office Hadley Centre, Exeter, UK. characterized the variability of water resources over Amazo3School of Engineering, Computer Science and Mathematics,. nia and their dynamics with time and distribution over the University of Exeter, Exeter, UK. region, but only due to natural climate variations and on interannual and decadal timescales. Furthermore, human ecoAmazonia and Global Change Geophysical Monograph Series 186 nomic activities such as urbanization, cattle growing, and Copyright 2009 by the American Geophysical Union. ranching, as well as agricultural development have affected 10.102912008GM000743 vegetation coverage, and the changes in land use and land 273
274
CLIMATE-VEGETAnON FEEDBACK AND IMPACTS ON WATER RESOURCES
cover due to intensive large scale deforestation could show impacts on the regional and global climate. As the agricultural front expands, soybean monoculture may lead td'the alteration of Amazonian ecosystems. Deforestation and subsequent biomass burning result in the injection of large volumes of greenhouse gases (GHGs) and aerosols, and could exacerbate the changes already produced by natural climate variability. In addition to the foreseeable increased deforestation, the following are also a threat: extinction and/or reduced diversity offish specie's in an area considered a fisheries hotspot; the accumulation in reservoirs of sediments and toxic levels of mercUlY; impacts on riverbank dwellers and indigenous peoples, as well as urban communities. Amazonia can be categorized as a region at great risk due to climate variability and change. The risk is not only due to projected climate change but also through synergistic interactions with existing threats not related to climate change, such as land clearance, forest fragmentation, and fire. Some model projections have shown that over the next several decades, there is a risk of an abrupt and ineversible change over a part or perhaps the entirety of Amazonia, with forest being replaced by savanna-type vegetation with large-scale loss of biodiversity and loss of livelihoods for people in the region, and with impacts ofclimate in adjacent regions. However, the uncertainties of these possible future scenarios are still high. On the basis of what is now known on climate variability and climate change in Amazonia, a question arises: What would be the possible impacts ofregional-scale deforestation or of the increase of GHG concentrations in the atmosphere on the climate of Amazonia and neighboring regions? In the austral summer 01'2005, drought hit western Amazonia, with devastating results: river levels fell, fish rotted, and routes to schools and hospitals were cut off, whole lagoons were evaporated, forest fires were widespread, fish, crops, and protected species died, boats and the villagers who travel in them were stranded. Four years later, in 2009, the extensive floods in the region determined that highest levels of the Rio Negro at Manaus from the 106 years of river records. Therefore, the main purpose of this chapter is to investigate and assess the risk associated with global warming and consequent climate change on the hydroclimatology of Amazonia and its impact on water resources, natural ecosystems, and society. There is also some discussion of climate feedback, the drought 01'2005, and prospects of similar events in the future. 2. WARMING IN AMAZONIA AND OBSERVED CLIMATE TRENDS Global mean surface temperatures have risen by 0.74°C ± 0.18°C when estimated by a linear trend over the last 100
years (1906-2005). The rate of warming over the last 50 years is almost double that over the last 100 years (O.13°C ± O.03°C versus 0.07°C ± 0.02°C per decade), according to IPCC AR4 [Trenberth et al., 2007]. The observed warming in Brazil is about 0.7°C during the last 50 years at annual timescales, while for winter, the trend is 1°C. For the Amazon region, Victoria et al. [1998] detected an observed warming of +0.56°C/l 00 years until 1997, while Marengo [2003] updated this wanning to +0.85°C/ 100 years until 2002. Similarly, positive trends have been identified in the nighttime and daytime air temperatures in some stations in Amazonia. Observational studies have shown no clear signs of unidirectional negative trends in rainfall in Amazonia. The magnitude and size of the trends depend on the rainfall data sets, length of records, etc., and the uncertainty is high since studies have found trends that V31Y in direction when different length periods are used. While rainfall in northern/southern Amazonia show slight negative/positive rainfall trends [Marengo, 2004], what is important are not the trends themselves, but the presence of decadal scale rainfall variability, with relatively wetter periods during 1945-1976 and relatively drier periods during 1977-2000 in northern/ southern Amazonia. The presence of dty years and drought in Amazonia on interannual timescales is related to occurrences of strong El Nino events in the tropical Pacific, or to an anomalously warm sea surface temperature (SST) in the tropical North Atlantic. The observed increasing trends in discharge and precipitation in all but the eastern parts of the Amazon Basin between the late 1950s and the early 1980s, detected by Gently and Lopez-Parodi [1980], were attributed by them to upstream deforestation. A possible explanation of this pattern was linked to the wet 1945-1976 period, part of the natural climate variability in the Pacific Ocean on decadal timescale [Zhang et al., 1997]. Analysis of extreme rainfall events in Amazonia [Haylock et al., 2006] have been hampered by the degree of spatial coherence and uncertainty due to the reduced number of stations used for the region. Using the records at Manaus and Belem, since 1961, an increase in the rainfall exceeding the 99th and 95th percentiles has been detected, suggesting a tendency for more intense and extreme events during the last 40 years, even though the annual totals may have not exhibited a significant positive trend. These findings thus support the idea that the atmospheric fluctuations induced by remote forcings [Fu et al., 2001] can potentially affect rainfall variability in the region, particularly the decadal variability of the Pacific Ocean, but it is still unclear if these teleconnections can be strong enough to offset or overshadow the effects of deforestation [Chen et al., 2001]. However, these tendencies or decadal variations do not provide information on whether significant changes in precipitation would occur in the future.
MARENGO ET AL.
The existence of trends on additional terms of the hydrological cycle in Amaz9nia have also been documented in previous studies [see H;yiews by Marengo et al., 2008a, 2008b], as have the swing~> and tendencies of significant changes in spatial averages f6r the input and output fluxes of water vapor (decreasing); these V31Y according to the type and length of time series used. The use of spatially aggregated point data may not be appropriate for the detection of trends, due to the inevitable "dilution" ofthe signal during the up-scaling process, while the use of gridded data sets may also create artificial trends. What has also been observed is decadal timescale variability rather than any trend toward systematic dtying or moistening of Amazonia in the long term. 3. BRIEF EVOLUTION OF MODELING OF CLIMATE CHANGE IN AMAZONIA: DEFORESTATION AND EXPERIMENTS ON CLIMATE CHANGE AND AMAZONIAN DEFORESTATION
275
of Amazon rainforest have been destroyed. About 60~70% of deforestation in Amazonia results in the creation of cattle ranches, while the rest mostly results in small-scale subsistence agriculture. Despite the widespread press attention, large-scale fanning (i.e., soybean) currently contributes relatively little to total deforestation in Amazonia. Costa et al. [2003] have identified increases in the annual mean and the high flow season discharge of the Rio Tocantins in southeastern Amazonia since the late 1970s, even though rainfall has not increased. They suggest that changes in the land cover in the basin for agricultural purposes and urban development have altered the hydrological cycle of the basin. Callede et al. [2004] suggest that increases in the mean annual discharge of the reconstructed series of the Amazon River at 6bidos during 1945-1998 could be the consequence of Amazon deforestation. In an attempt to investigate the possible impact of Amazonian deforestation on the regional climate and hydrology, general circulation models (GCM) have been used to simulate the effects of land use changes, where forest is replaced by grassland in the whole basin. These suggest a possible change in the regional and global climate as the results of h'opical deforestation [see reviews by Salati and Nobre, 1991; Marengo and Nobre, 2001; Marengo, 2006; Sampaio et al., 2007]: Under a hypothesized entire Amazon basin deforestation scenario, almost all models show a significant reduction in precipitation and evapotranspiration (Table 1), and most found a .decrease in streamflow, precipitation and increases in air temperature. Total deforestation results in an increased surface temperature, largely because ofdecreases in evapotranspiration. Sampaio et al. [2007] studied the effects of Amazonian deforestation on the regional climate, using simulated land cover maps fi'om a business-as-usual scenario of future deforestation in which the rainforest was gradually replaced by degraded pasture or by soybean cropland. The results for eastern Amazonia, where changes in land cover are expected to be larger, showed that the reduction in precipitation in eastern Amazonia is more evident when deforestation exceeds 40% of the original forest cover, and this reduction in precipitation occurs mainly during the dty season. In the same study, the combined effect of deforestation and a doubling of CO 2, including the interactions among the processes was simulated to increase temperature by some + 1.4 dc.
A variety of human activities can act to modify various aspects of climate and the surface hydrologic systems. Historically, land surface changes in Amazonia intensified in the mid- and early 1970s, when strategic governmental plans, such as Brazil's Programa de Integrayao Nacional, first attempted to promote the economic development of the region. Those plans included the constlUction of extensive roads throughout the basin and the implementation of fiscal incentives for new settlers, triggering a massive migration of landless people into the region. Changes in land cover can significantly affect the smface water and energy balance through changes in net radiation, evapotranspiration, and runoff. However, because of the intricate relationships between the atmosphere, tenestrial ecosystems, and surface hydrological systems, it is still difficult to gauge the importance of human activities in the Amazonian hydrological cycle. The aerosols and smoke from the biomass burning during the dry season in Amazonia seems to have an impact on the onset of the rainy season in southern Amazonia, and ultimately, increase in the concentration of GHGs and aerosols could affect the energy balance and thus climate of the region. Recent data from remote sensing show that large areas of Amazonia (mostly Brazilian Amazonia) have been changed from forest to pasture and agricultural land, and that observed deforestation rates in Brazilian Amazonia increased in 2004 relative to 2003. Deforestation rates had stabilized somewhat 4. MODEL PROJECTIONS OF CLIMATE CHANGE in the early 1990s, mainly in the Brazilian Amazonia, but the IN THE AMAZON BASIN FROM IPCC underlying pressures to continued land use change are still' AR4 GLOBAL MODELS present: a growing population in the developing nations of Amazonia and plans for a road network crisscrossing the reSection 2 summarized how the issue of deforestation has gion. Between May 2000 and August 2005, Brazil lost more been explored in various numerical experiments since the than 132,000 !Gn2 offorest, and since 1970, over 600,000 km2 1980s, all of which show that Amazonia would become drier
Z';@ji
276
CLIMATE-VEGETATION FEEDBACK AND IMPACTS ON WATER RESOURCES
MARENGO ET AL.
Table 1. Comparison of Climate Simulation Experiments of Amazonian Deforestation From Global Climate Models a Experiment It
Dicldnson and Henderson-Sellers [1988] Dicldnson and Kennedy [1992] Henderson-Sellers et al. [1993] Hahman and Dicldnson [1995] Zeng et al. [1996] Hahmann and Dickinson [1997] Costa alld Foley [2000] Lean and Warrilow [1989] Lean and Wardlow [1989] Lean and Rowntree [1993] Lean and Rowntree [1997] Lean et al. [1996] Manzi and Planton [1996] Nobreetal. [1991] Shukla et al. [1990], Nobre et al. [1991] Dil'lneyer and Shukla [1994] Sud et al. [1996a] Sud et al. [1996b] Walker et al. [1995] Polcher and Laval [1994a] Polcher and Laval [1994b] Zhang et al. [2001] Voldoire and Royer [2004] Sampaio et al. [2007]b
till
I1T
M
-0.5 -0.7 -0.6 -0.4 -2.0 -0.4 -0.6 -0.9 -0.6 -0.6 -0.8 -0.8 -0.3 -1.4 -1.4 -0.4 -1.2 -1.0 -,1.2 -2.7 -0.4 -0.4 -0.6 -1.0
+3.0 +0.6 +0.5 +0.8
0.0 -1.4 -1.6 -0.8 -3.1 -1.0 -0.7 -1.4 -1.3 -0.8 -0.3 -0.4 -0.4 -1.8 -1.8 -0.7 -1.5 -0.7 -1.5 +1.0 -0.5 -1.1 -0.4 -0.9
+1.0 +1.4 +2.4 +2.0 +1.9 +2.3 +2.3 -0.5 +2.5 +2.5 +2.0 +3.0 +3.8 +0.1 +0.3 -0.1 +3.1
M
+0.5 -0.7 -1.0 -0.4 -1.1 -0.6 -0.1 -0.5 -0.7 -0.3 +0.5 +0.4 -0.1 -0.4 -0.4 -0.3 -0.3 +0.3 -0.3 +3.7 -0.1 -0,01
This is also confirmed by Boulanger et al. [2006,2007], Cavalcanti et at. [2006], Grimm and Natori [2006] and Li et at. [2006]. On air t~.lnperahlre changes, Meehl et at. [2007] show that all mod~ls feature warming in South America with the strongest waylning being in tropical South America, especially Amazonia and Northeast Brazil, reaching, in some models, increases of up to 6-8°C warmer than the present by 2100 and with the degree of warming vaIying among models. Li et at. [2006] show that the GCMs from IPCC AR4 predict velY different changes of rainfall over the Amazon under the Al B scenario for global climate change. Five of the 11 models studied by Li et al. [2006] predict an increase of annual rainfall, 3 models predict a decrease of rainfall, and the other 3 models predict no significant changes in Amazonian rainfall. Figure I show trends ofprecipitation and temperature anomalies fi'om 15 AOGCM for two emission scenarios (A2 and B1) during the twenty-first centulY. The results reveal larger differences among models than among emission scenarios for
0.5
(if
32. 0.0 E E
7
timate for the low emission scenario (B1) is 2.2°C (likely range is 1.8°C to 2.6°C), and the best estimate for the high scenario (A2) is 4.5°C (likely range is 3.9°C to 5.1 0C). The new assessment of the IPCC AR4 likely ranges now relies on a larger number of climate models of increasing complexity and realism, as well as new information regarding the nature offeedback from the carbon cycle and constraints on climate response from observations. Annual rainfall projections from the multimodel IPCC AR4 average for Amazonia for the AlB scenario for 2080-2099 in relation to 1980-1999 show positive anomalies around 0.3 mm d- 1 in western Amazonia, while eastern Amazonia exhibits a reduction of about 0.2 mm d- 1. The analysis of the climate change projections for the AlB scenario made by Vera et al. [2006] show a substantial agreement among IPCC-AR4 models in precipitation changes over some parts of South America for the period 2070-2099 relative to 1970-1999, particularly an increase in summer precipitation over the northern Andes and southeastern South America. Over Amazonia, results are mixed.
Precipitation Anomaly
1.0
aResults show the differences between deforested minus control run. till is change in evapotranspiration (lUlU d- 1), I1Tis the change in surface air temperature (OK), M is the change in precipitation (mm d- l ), M is runoff, calculated as the difference of M and till (M = M - M). Modified from Marengo [2006]. bOnl y for eastern Amazonia.
and warmer. Even though at present there are no clear signs of trends for reduction of rainfall in the basin due to deforestation [Marengo, 2004, 2009], one study [Costa et al., 2003] has detected changes in the Rio Tocantins discharge as a result of land use changes in its upper basin following the constmction of the city of Brasilia in the 1960s. On the basis of what is now known on the climate variability in Amazonia and the moisture transport in and out of the basin based on observational studies and model simulations, the question now arises: What are the possible impacts on the Amazonian ecosystem of regional-scale deforestation or the increase of GHG concenh'ations in the atmosphere and subsequent global warming? Analysis of models from the Coupled Model Intercomparison Project, which contributed to the assessment of future climate change scenarios in South America in the IPCC AR4 [Vera et al., 2006; Li et al., 2006; Christensen et al., 2007; Meehl et al., 2007] suggest that waIming in Amazonia at the end of the twenty-first centulY (2090-2099), relative to 1980-1999 can VaIy from between models. The best es-
the same model. The pmjected temperature warming for the Amazon basin ranges fi'om 1°C to 4°C for emissions scenarios B I and from 2°C to 7°C for A2. The analysis is much more complicated for rainfall changes. Different climate models show distinct pattern, even with almost opposite projections (the mean ofthe 15 models does not show significant precipitation anomaly in the twenty-first cenhuy). The large disagreements found between the AR4 simulations raises numerous questions regarding the ability to properly reproduce atmospheric convection and the climate and carbon cycle in Amazonia, which is also linked to a realistic representation of the dynamics of vegetation. Since the early 2000s, new developments in atmosphere-oceanbiosphere coupled models by the Met Office Hadley Centre in the UK, the Instihlte Pierre et Simon Laplace-University of Paris in France, and others have allowed for more sophisticated simulation of fuhlre climate change scenarios [Cox et al., 2000, 2004; Betts et al., 2004; Friedlingstein et al., 2006]. Projections for future climate change fi'om the
1.5
+0.1
r-_-..__-:li:::em::J;p:,::e::::ral:.::;ureram.:::::::ma::::;lyt..-_--r_----, ,
6
,
::: : : :
-0,5
.
:ES
, •:: : .:
:::::.
~~~ ~~'.1.- ~~~{; ;'Hii:':'~~t::::.'::,....~.~:..~:;~.~; "",'
•••
[;
'"
......... :::.: ::::::.,
" SRES 61
"" ~~;;
. ,
61
.:,~"'" <"
•• ""
<"'«,
-1.0
M••nSRES A2
1«.,
"'.
5
....
-1.5 2020-29
4
2050-59 Year
2090-99
3
2
0<----'--
2020-29
..1-
2050-59 Year
277
.1.-_-'1
2090-99
Figure 1. Changes in precipitation and temperature for the periods 2020-2029, 2050-2059, and 2090--2099 from those of 1961-1990 simulated from 15 AOGCMs for the SRES A2 and B1 scenarios (data from IPPC/AR4) averaged for Amazonia. The solid lines show the mean of the 15 models for each scenario, and the dotted lines show individual model projections. The Amazonian domain is indicated by the box in the map (l2°S-4'soN; 74.5°W-46.5°W). Models are BCCR-BCM2.0, CCSM3, CGCM3.1 (T47), CNRM-CM3, CSIRO-MK3, ECHAM5, GFDL-CM2, GFDL-CM2.l, GISS-ER, INM-CM3, IPSL-CM4, MIROC3.2 (MEDRES), MRI-CGCM2.3.2, UKMO-HADCM3, and ECHO-G.
278
CLIMATE-VEGETATION FEEDBACK AND IMPACTS ON WATER RESOURCES
Hadley Centre model HadCM3LC using a dynamic vegetation scheme have shown that an increase in the concentration of GHGs in;fhe atmosphere will produce a wmming above 6°C in the high emission scenario, with rainfall reductions and associated decreases in soil water and thus determining changes in vegetation. Some ,model projections [Betts et al., 2004; Cox et al., 2004; Oyama and Nobre, 2004; Salazar et al., 2007; Sitch et al., 2008] exhibit over the nex,t several decades a risk of an abrupt and irreversible replacement of forests by savannah with large-scale loss of biodiversity and loss of livelihoods for people in the region. This process is referred as the "dieback" of Amazonia, and it represents a process simulated by a few climate models, where after reaching a "tipping point" in climate (C0 2 concentration, air temperature), the forest stops behaving as a carbon sink and becomes a carbon source, and after that, the forest enters into a state of collapse, being replaced by a savanna-type vegetation, in a process that has been referred to as "savamlization" starting by 2050-2060. A recent study by Sampaio et al. [2007] identified another "tipping point" when the deforested area in eastern Amazonia reached 40-50% level, also leading to savannization. Therefore, after 2050, the Amazon Basin would behave as a "source of moisture and carbon" rather than a sink as in today's climate [see Houghton et al., this volume; Cox et al., 2000]. Cox et al. [2004] suggest that more frequent droughts would wipe out around 65% of Amazonian forest cover by 2090. As a consequence, there is a risk of significant biodiversity loss through species extinction in Amazonia and, in many areas, of tropical Latin America. However, the probability of this dangerous event is not clear. These projections for drought in Amazonia after 2040 also show systematic warming in the tropical Pacific indicative of an El Ninolike mode of valiability becoming persistent after 2040. However, the likelihood of this extended El Nino or more frequent/intense El Nino-mode scenarios in a global warming world is still an open issue. The projected decrease in precipitation also appears to be linked to anomalous warming in the tropical North Atlantic [Cox et al., 2008; Good et al,,2008]. 5. PROJECTIONS OF CHANGE IN HYDROLOGY AND MOISTURE TRANSPORT IN AND OUT OF AMAZONIA Macroscale hydrological models, which include the land surface hydrological dynamics of continental-scale river basins, have rapidly developed during the last decade [Russell and Miller, 1990; Miller et al., 1994; Marengo et ai" 1994; Nijssen et ai" 1997, 2001]. These models can act as links
between global climate models and water resources systems on large spatial scales and long-term timescales. Predictions of river discharges in the Amazon Basin for present climate and double CO 2 future scenarios have been calculated by Russell and Miller [1990] and Nijseen et al. [2001] using global models. Some problems with the parameters of the models or perhaps the lack of suitable runoff data for validations indicate that, in most of the models, the rainfall and runoff in Amazonia are underestimated. This also generates an unceliainty in the projected values of runoff in the future, forced either by increase in GHG or in changes in land use and land cover. Simulations by Coe et al. [2003] using a terrestrial ecosystem model have been successful in simulated intermillual and seasonal runoff variability in Amazonia, and even though the discharge is consistently underestimated, the model captures climate variability and the impacts ofEl Nino since the early 1950s. It is hard to attribute changes in the hydrology of Amazonian rivers to deforestation and human-induced land use changes or to an increase in the concentration of GHGs. The study by Milly et al. [2005] that is refened in the IPCC AR4 assesses changes in streamflow in various rivers worldwide. This study uses 12 IPCC AR4 models (CCSM3, CGCM3.1(T63), ECHAM5/MPI-OM, ECHO-G, FGOALSgl.O, GFDL-CM2.0, GFDL-CM2.1, GISS-AOM,MIROC3.2 (hires), MRI-CGCM2.3.2, UKMO-HadCM3, and UKMOHadGEM1), for the period 2041-260 relative to 1900-1970 for the AlB scenario. The average projection is for a 10-15% reduction in Amazonia. A possible reduction in streamflow in the Amazon basin can have negative impacts in transportation, biodiversity, and water resources. Even though Amazonia can be considered as a closed system, the region constitutes a source of atmospheric moisture for other regions in the continent. Moisture transpOli and out of the Amazon basin has also been studied since the 1990s using a variety of data sets. The regional circulation features responsible for this transport and its variability in time and space have been detected and studied using observations collected during short-term field experiments. This feature is the South American Low Level Jet (LLJ) east of the Andes (Figure 2) and represents a regional circulation pattern in South America that could be described as a moisture corridor that brings moisture from the Amazon basin to the southern Brazil-northern Argentina region of the Parana-La Plata Basin. This occurs especially during the warm rainy season, where the trade winds over Amazonia and the tropical North Atlantic are stronger, while during the cold season, the LLJ is fed with moisture coming from the South Atlantic associated with a more intense Subtropical Atlantic anticyclone, which is closer to the continent during austral winter [Marengo et al., 2004; Vera et al., 2006].
MARENGO ET AL.
Fi~Ul:e 2. Conceptuall~odel of the South America Low Level Jet LL ' tUle, flOm the Amazon mto the Parana-La Plata Basin [M . (J) east of the AIJdes transportmg atmospheric ll1oisSociety, mengo et al., 2004]. Copynght 2004 American Meteorological
Choncerns are ,great regarding the possible role of climate c ange (warmlllg and drying) in Amazonia on the climate of the Parana-La Plata Basin.
. Figure 3 shows that, in the future for the A2 sc "LLJ IS f enano, me~ore r~qu~nt almost all year long, especially in sumand w111teI, compared to the present. The fact ofh ' more. freque.nt and p,ossibly intense LLJ in the future ir:;ti~~ a. wan:ler chmate With stronger and more fi'equent LLJ _ SI?ly Irregular in terms of distribution. This is consi~~~t WIth I~ore ~requent and intense extreme rainfall events at the eXIt regIOn of the jet in southern Brazil as suggested b Marengo et al. [2009b] simulated by HadRM3P regional climate model by the end of the twenty-first centuly,
H:rojections. of change in moisture transpOli using the ~3 regIO.nal model [Soares and Marengo, 2008] have ' shown of , ,l11terestlllg results (Figures 3 and 4) . The Impacts l11Creaslllg the concentration of GHG on the functioning of t1:e LLJ, and the moisture transport into and out of Am _ ma ar h 'F' azo , e s ownlll 19ure 3, In warmer climates (A2 scenario) fOl both the mean and LLJ composites, the meridional wind ~omponent ~nd the flux outside Amazonia suggest more ' llltense LLJ 111 summer. Moisture transpoIi l'nto A . t' ' mazoma 6. A COMBINATION OF THE INCREASE IS s longer S111ce the northeast trades are accelerated in th IN GREENHOUSE GAS CONCENTRATIONS future. I~ ~resent climates, the moisture transport outsid: AND LAND USE CHANGES IN AMAZONIA Amazoma 111to the La Plata basin and tIle LLJ are stronger d AS FORCINGS FOR CLIMATE CHANGE an more frequent during sunllneliime season [J.r t I 2 marengo ~. a ., OO~]. Therefore, the integrated moisture transport . It can be presumed that changes to the cycles of wat ' IS stronger III the future but mainly due t t ,'d" ' 0 a s ronger meenergy, ~arbon, and nutrients that result from replacement ~f n ~onal wl11d, even though moisture content may be lower Amazoman. vegetation will have consequences for climate ~~~'e~lso leads to increased moisture convergence in th~ and th~ ~nvIronn:ent at local, regional, and global scales. The conveI SIOn of pnmaly tropical forest to agricultural areas or
279
280
MARENGO ET AL.
CLIMATE-VEGETATION FEEDBACK AND IMPACTS ON WATER RESOURCES 15N
A
100
15N
90
iON
iON
5N
5N
EO
EO
58
58
::l
108
108
15...
158
158
~ 30
208
208
258
258
308
308
358
358
1980-1989
408
85W 80W75W 70W 6 W 60W55W 5 W 4 W 4 W 35W 3 W 25W
408
t/)
EO
58 108 158
~~
i
208
208
258
258
308
308
358
358-
408
85W 80W 75W 70W 65W 60W 55W 50W 45W 40"" 35W 30W 25W • • •
8
0
85W 8 W 75W7 W 65W 6 W5 W 5 W 4W 4 W 3 W 3 W2 W
. 'de b oxes ).m 10 8 kg S-l for (a and (1000-300 hPa) and net moistnre convergence (mSl Figure 3. Integrated mOlstnre ux b d( d d) LLJ composites for the present (1980-1989) and the futnre S d ,,,. 2008] b) mean summer December-January-Fe mary an can . Resu1ts are Slmu . 1at'ons from HadRM3P [ oares all lYlmellgo, . (2080-2099) for the A2 scenano. 1 .
fl
secondary vegetation represents one of the most profound changes to the natural environment of the present a~e. Two recent studies further investigated the relatIv.e roles of future changes in GHGs versus future changes m .Iand cover. Voldoire [2006] compared the climate change Sl1I~U lated under a 2050 Special Report on Emissions Scenanos (SRES) B2 GHGs scenario to the one under a 2050 S~S B2 land cover change scenario. He shows that .the .relatIve impact of vegetation change to GHG concentration mcr~ase . f the order of 10% and can reach 30% over locahzed IS 0 ., f£ t t tropical regions. Similarly, he found no slgmficant e ~c a the global scale, but a potentially large effect at the reg.IOnal scale, such as a warming of 2°C by 2100 over Amazoma for
20
o A2 2080·2089
158
"
50 40
10
108
\,
70 60
58
\1,2
Q.
..,
J..
~~"f~~~('
iON 5N
80
.~
Integrated moisture flux (a(rOroJwF)S) antdh mpOriesst~~ C(?~85~~~)n~~d'~~~~~~~6~~_~~)1 ~ A2 Kg.s-1 for summer mean for e 15N--=V-~~~~'---~~~~~-:;C;:--j
-8
the A2 land cover change scenario, coming with a red~c tion of the diurnal temperature range. The general findmg of these studies is that the climate change due to land ~over changes may be important relative to GHGs at the regIOnal level where intense land cover change occurs. GI~bally, the impact ofGHG concentration increase seems to dominate over the impact ofland cover change. In the case of Amazonian forest, if warming due to increase in concentrations of GHG (either natural or anthropogenic) is ~bove 3.5°C to 4°C, there is a risk of a "tipping point" leadmg to savannization. A recent study by Sampaio et al. [2007] and Sampaio [2008] identified another "tipping point" when the deforested area reached 40-50% level, also leading also to
DJF
MAM
JJA
SON
Figure 4. Number ofLLJ episodes during all seasons for the present (1980-1989, open) and the futnre (2080-2099, solid), for the A2 scenario. Results are simulations from HadRM3P [Soares and Marengo, 2008].
savannization. They found two biosphere-atmosphere equilibrium states for South America as did Oyama and Nobre [2003]: (1) present-daypotential biomes; (2) a new vegetationclimate equilibrium where the eastern part ofthe Amazonian tropical forest is replaced by tropical savanna and semidesert; desert areas also appear northeast ofBrazil. In this study, they found that 50% of total deforested area in Amazonia is the thTeshold for transition from present-day potential biomes to a new vegetation-climate equilibrium state in Amazonia. The impacts of the GHG increase and the land use changes combined with climate change in the region is still under study. 7. EXPERIENCES ON CLIMATE CHANGE PROJECTIONS FROM REGIONAL MODELS IN AMAZONIA The issue of the spatial resolution in scenarios must be put in the context of other uncertainties of climate change. Studies and analyses of climate change impact and adaptation assessments recognize that there are a number of sources of uncertainty in such studies, which contribute to uncertainty in the final assessment. The importance of high resolution climate scenarios for impacts and adaptation studies remains to be thoroughly explored in South America. Regional Climate Models can represent the local land surface variables affecting regional climate as well as internal climate variability, and they represent an effective method of downscaling to add fine-scale detail to simulated patterns of climate variability and change. Downscaling in South America has been developed for better understanding of the physical processes in the atmo-
281
sphere, as well as for weather and climate forecasts, and is now being developed for climate change studies. One initiative has been the implementation of Regional Climate Change Scenarios for South America (CREAS) [Marengo and Ambrizzi, 2006; Marengo et al., 2009b]. This project aims to provide high-resolution climate-change scenarios for South America for raising awareness among government and policy makers in assessing climate change impact, vulnerability, and in designing adaptation measures. The CREAS project lUns three regional models: Eta CCS (I. A. Pisnitchenko and T. A. Tarasova, The new version of the Eta regional model developed for climate-change simulations, submitted to Meteorological Applications, 2007), RegCM3 [Ambrizzi et al., 2007], and HadRM3P [Jones et al., 2004; Marengo and Ambrizzi, 2006]. The models are nested within the HadAM3P global atmospheric GCM, which is driven by GHG concentrations from the SRES emissions scenarios and SSTs from the HadCM3 ocean-atlnosphere GCM. We examine two 30year simulations: the present climate that examines the time period 1961-1990 and the future climate that covers the time slice of 2071-2100 under the IPCC SRES A2-high emission and B2-10w emission scenarios using the three regional models above indicated, which were lUn at a resolution of 50 km latitude-longitude. Figure 5 shows projections of annual rainfall and temperature anomaliesfi'om the A2 and B2 scenarios for the future, as represented by the ensemble of the three regional models used for CREAS. Figure 5 shows that changes in A2 are more radical and regionally comprehensive compared to those of B2. While the drier region (between 1 and 2 mm 1 d- ) in B2 covers mostly northern Amazonia, the dly region in A2 extends into eastern Amazonia and the entire state of Para, with the largest reductions nearby the mouth of the Amazon River. In relation to annual temperature changes, in the scenario A2, the entire tropical South American region may become 4-6°C warmer, and up to 8°C warmer than normal in central equatorial Amazonia. In the B2 scenario, the warming all across Amazonia varies between 2°C and 4°C with the warmest in central equatorial Amazonia by 4-5°C. The possible impacts on these changes are shown in Figure 5, with the more intense impacts on the A2 scenario. It is expected that as a consequence of global warming, there will be an intensification of the hydrological cycle that will increase the frequency and intensity of extreme rainfall events and also dly spells [[PCC, 2007a, 2007b]. Projections of climate extremes derived from the multimodel - ensemble fi'om the IPCC AR4 [Tebaldi et al., 2007; Meehl et al., 2007] and from the downscaling of a singe realization of the HadCM3 using Providing Regional Climates for Impacts Studies (PRECIS) [Marengo et al., 2009a] suggest that, in general, the temperature-based indices of extremes exhibit
282
MARENGO ET AL.
CLIMATE-VEGETATION FEEDBACK AND IMPACTS ON WATER RESOURCES
Western Amazonia is projected to experience an increase in extreme rainfall events/ Heavy precipitation events, which are velY likely to increase in frequency in this region, would augment flood risk lh lower Amazonia.
iON 5N It
EQ
/
58
8. UNCERTAINTY ASSESSMENTS
108 158 208 258 308 358 408 10N
iON
5N
5N EQ
EQ
ti,
58 108 158 208
ti,
58 108 158 208 258
258
308
308
358 W
Figure 5. Projected changes in a1Umal rainfall and air temperature anomalies for the scen7i: (Ie~) ~2 ~nd~~:I~t~~~i Units are mm d- 1 (rainfall) and (0C) for temperatures. Results are th~. ensemble mean 0 t ee reglOna ~ d I Re CM3 and HadRM3P) driven by lateral boundary condltlons from the Had~3.P global cllm~te .mo e CPTEC, . g , . I ' t 1961-1990 The box represents Amazonia as defined 111 tins study [Ambl/zzi et al., for the penod 2071-2100 Ie atlve 0 . 2007].
changes consistent with the expected warmer climate in the future relative to the present under the IPCC S~S A2 a~d B2 scenarios, with the greatest warming trends bemg seen m A2 especially in tropical South America. . From the PRECIS projections for 2071-2100 relatJ.ve to 1961-1990, positive trends in the TN90 index (warm mghts) are detected over most of Amazonia, and the presence of RlOmm (number of days with rainfall abo~e ~O mm) events show a positive trend in western Amazoma m the B2 scenario and in the A2 scenario, the changes are more accentuated. Thus, this region would experience more extreme precipitation events in the future than at present, under ~he HadCM3-driven scenarios. On the other hand, the CDD mdex shows an increasing trend in the region extending. from eastern Amazonia to Northeast Brazil in the B2 scenano un-
til the end of the twenty-first century. These trends, derived from the downscaling of the HadCM3 using PRECIS, are consistent with those trends in extremes derived from the IPCC AR4 models up to 2100 from Tebatdi et at. [2007]. The IPCC AR4 model projections suggests an increas.e ~f about 10% in accumulated rainfall in western Amazoma m both A2 and B2 scenarios at annual and seasonal levels, and the extJ'eme rainfall projections indicates that this inc~'ease of rainfall would be in the form of irregular intense ramfa~l events. In eastern Amazonia, the frequency of dry spells IS projected to increase together with a reduction in the frequency of intense rainfall events. . The resilience of many ecosystems could be exceeded th~s century by this unprecedented combinat.ion of. changes m mean and extremes of climate and assocIated disturbances.
283
els in the Amazon region [Li et at., 2006; Cox et at., 2008] and also with regard to ENSO variability [Cox et at., 2004]. We must therefore currently consider the drying simulated by the Hadley Centre model(s) as having a finite probability under global warming, with a potentially enormous impact. All these projected changes in Amazonia may have climatic, ecological, and environmental implications for the region, the continent, and for the globe. A sound knowledge of how the natural system functions is thus a prerequisite to defining optimal development strategy.
Some of the studies reviewed suggest that increases in the concentrations of GHG and aerosols in the atmosphere, as well as land changes in land cover for agriculture, have already affected the hydrology of the Amazon basin. Some uncertainties can be attached to these results due to model 9. DANGEROUS CLIMATE CHANGE IN AMAZONIA limitations, and to the lack of continuous and long-term AS PROJECTED BY THE UK MET OFFICE observational series of climate and hydrological variables. HADLEY CENTRE MODEL Large uncertainties were also identified in anthropogenic aerosol forcing and response, and changes in land use leadOne definition of "dangerous climate change" is a climate ing to biomass burning and their impacts on rainfall in the change, which induces positive feedback, which accelerates basin [see Longo et at., this volume; Artaxo et at., this vol- the change. Feedback between climate change and the carume]. Analyses of the few available long-term rainfall and bon cycle, particularly involving a weakening of the current river series suggest an absence of significant unidirectional terrestrial carbon sink and a possible reversal to a source have been identified as one such potential positive feedback trends toward drier or wetter conditions in the basin. However, more evidence of variability at interanpual and [Cox et at., 2000; Friedlingstein et at., 2006]. While a major decadal timescales is apparent, and these observed trends process in this is increased soil respiration worldwide due to may be due to natural climate variability and the observed rising temperatures, a further feedback would result from the climate shifts, while no signal of land use changes have yet loss of forest cover ip Amazonia if the climate of the region appeared in the long-term climate and hydrologic variability were to become drier. of the region. The reduction of rainfall will also have imBetts et at. [2004] quantified the contribution of Amazopacts on the river and water levels, as well as on the rainfall nian forest dieback to the overall carbon cycle feedback on distribution in them with longer dry periods and intense rain- climate change in the UK Met Office Hadley Centre coupled fall events concentrated on few days. climate-carbon cycle model HadCM3LC [Cox et at., 2000], The uncertainty level is still high since, among other rea- which is a version of the HadCM3 GCM with an interacsons, models still have problems in representing the carbon tive carbon cycle included. They performed two simulations cycle, aerosols, and convection in tropical regions. Despite driven by historical CO 2 emissions since 1860 and proproviding greater spatial detail due to the higher resolution, jected twenty-first century emissions according to the IPCC the use of regional climate models may still suffer limita- IS92a "business as usual" scenario. One simulation (named tions because many ofthe downscaling experiments in South "CARBCLIM" here) included CO 2 acting as both a GHG and America derived from the use of one or more regional mod- as a fertilizer of photosynthesis, while in the other (named els nested on one global model, particularly the HadCM3 "CARB"), CO 2 only exerted its fertilization effect and did model. So far, we do not have a multimodel ensemble of not act as a GHG. The CARE simulation therefore did not include radiatively forced climate change; the carbon cycle regional models as in the global models fi'om IPCC AR4. The Amazon dieback is a product of the HadCM3 model. only responded to the increase in C02. In CARBCLIM, An argument for using the Hadley model relies on the however, the carbon cycle responded to both the increasing grounds that it is actually among the more realistic of the C02 and the resulting climate change. The CARB simulation projected an uptake of 60 gigatons IPCC AR4 models. Several AR4 models do not give such an extreme climate change simply because they are umealisti- of carbon (Gt C) in South American vegetation relative to cally dly in the first place. Of course, we acknowledge that ,1860, and a total uptake of220 Gt C in global vegetation and there are significant unceliainties, and maybe the Hadley 400 Gt C in global soils (Figure 6). This was due to an inmodel is too extreme in terms of warming and dlying (Fig- crease in net primary productivity, the net uptake of carbon ure 1), but several of the more realistic models do agree on a by vegetation, as a result offertilization from enhanced phodlying of the region. Although HadCM3 produces the most tosynthesis by the rising C02. As well as increasing the storextreme dlying of current GCMs, it is among the better mod- age of carbon in the vegetation itself, the more productive
MARENGO ET AL. 284
285
CLIMATE-VEGETATION FEEDBACK AND IMPACTS ON WATER RESOURCES a 400
Chan e in Ve el:at',on Carbon HodCM 3LC with lnteroctrve CO. HodCM3LC with lnteroctive CO.. but no climate change
_
300
""
c'
200 u
<':'5
100 ,
... "" """
0
.
Giobol South America
-100
L~~~--'----L-~-"-~--'-----'--::~~~"'=:=-:---'---'--';;2;;:0~5:-;:0-;-~~~2--;1 00
-200 1 850
1900
1950
2000
Year
b 400
Chan e in Soil Carbon
H
300
200
~
100
o Globol South America
-100
L~~~--'----'-~-'-~~----'~~~~"'=:=-:---'---'-~2;;:0~5:-;:0-;-~~~2~·1 00
- 200 1 850
1 900
1950
2000
Year
" ' b 1and South American terrestrial carbon stores with both C02 rise and clim~te Figure 6. Time senes of changes m Gl~ a , h (bl k l' ) for the globe (solid lines) and South Amenca '., , ) d 'th CO rise without climate c ange ac mes, change (gray lines an WI 2 '1 b F M t Office Hadley Centre presentation, Bntlsh Crown (dotted lines): (a) vegetation carbon and (b) SOl car on. rom e copyright.
vegetation also released more. carbon ~n litter and hence in, creased the storage of carbon m the soIl. In CARBCLIM, the mean precipitation across Amazoma 1 decreased by approximately 3.0 rom d- by 2100, Tree cover reduced throughout the twenty-first century, and shrub.s .and grasses spread to replace the forest. Eventually, cond1tions in some locations became unfavorable even ~or shrubs or grasses, and in the driest grid boxes, the bare soIl was largely left exposed. By the end ofthe twenty-first ce~tury, the mean broadleaf tree coverage of the Amazon bas1~ was reduced from over 80% to less than 10%. In approximately h~lf of this area, the trees had been replaced by C4 grass leadmg to a savanna-like landscape. Elsewhere, even grass~s could not be supported, and the conditions became essentially desertlike.
was more than offset by an increase in soil respiration under higher temperature~( This contrasts with the carbon store changes projected bytCARB with no enhanced greenhouse effect. Ii Therefore, relatite to the increased carbon stores in CARB, the total terrestrial carbon deficit was 710 Gt C. Of this, 130 Gt C of the total deficit was due to the Amazon forest dying back instead or becoming more productive. Of the total deficit, 550 Gt C was due to a 150-01 C release of global soil carbon rather than a 400-Gt C uptake. While most of the release of soil carbon globally is due to enhanced soil respiration under greenhouse warming, some of the loss of soil carbon in Amazonia was due to forest dieback itself. Part of this was due to reduced input of carbon from the forest and part due to additional warming arising by biogeophysical feedback on local surface temperatures. Precise quantification of this would require an additional simulation neglecting changes in NPP and forest cover, but it was estimated that approximately 30 Gt C was lost from Amazonian soils through these mechanisms. The total loss of carbon fi'om aboveground and belowground biomass attributable to forest dieback was therefore 160 Gt C. T~erefore, the Amazon dieback can be considered to conh'ibute approximately 20% to the total land feedback on atmospheric CO 2 rise in this model. It is therefore clear that regional climate change simulated in Amazonia could exert a feedback on global climate change through the effects of forest dieback on atmospheric CO 2 rise. Since regional climate change predictions are subject to considerable uncertainty, the importance of the feedback to global climate implies that the uncertainties in global climate are also subject to uncertainties that depend on the regional-scale uncertainties. Assessments of the risk of "dangerous climate change" from carbon cycle feedback therefore need to quantify the risk of a significantly dlying climate in Amazonia.
The Amazon forest dieback in CARBCLIM caused the aboveground carbon store in South America to reduce by 70 Gt C. Although much of the rest of the global vegetation cover took up carbon in CARBCLIM as in CARB, the change of the South American vegetation carbon sto~e from a 60-Gt C uptake in CARB to a 70-01 C release m CARBCLIM means that the total global vegetation carb?n store increase was much reduced in CARBCLI~. With some other losses in other regions, the increase m glob~l vegetation carbon was limited to 60 Gt c., The global .solIl carbon store was reduced by 150 Gt C (F1gure 6), mam Y due to enhanced soil respiration under rising tem~eratul:es. As well as leading to Amazon forest dieback, the mcluSlOn of climate change in CARBCLIM means that the large .fl~x of carbon into soils via increased net primary produchVlty
10. EFFECTS OF AEROSOLS IN THE PROJECTIONS OF FUTURE DROUGHT EVENTS IN AMAZONIA Amazonia is estimated to contain about 10% of the carbon stored in land ecosystems, and to account for 10% of global net primary productivity. Despite large-scale human deforestation, it seems likely that the region is currently acting as a net sink for anthropogenic CO 2 emissions [see Phillips et al., 2008, this volume; Saleska et al., this volume; Houghton et al., this volume]. The resilience of the forest to the combined pressures of deforestation and climate change is therefore of great concern, especially since at least one major climate model predicts a severe drying of Amazonia in the twenty-first centulY.
Biomass burning emits aerosols into the atmosphere [see Schroeder et al., this volume; Longo et al., this volume]. The composition and amount of these aerosols depends on meteorological and biospheric conditions as well as on human activities. In the dly season of the Amazon basin, biomass burning increases the aerosol loading to number concentrations at least several times larger than those in the wet season [Andreae et al., 2004]. It seems that the main effect of the aerosols in the hydrological cycle is on large-scale circulation transition from dly to wet season and, thus, on the onset of the rainy season. Recent work of Zhang et al. [2008] suggests that the biomass burning aerosols may tend to re-enforce the dry season circulation pattern and, thus, weaken 01' slowdown the circulation transition to the wet season. Yu et al. [2007] suggest that biomass-burning aerosols could amplify the original climate anomalies during the transition season (AugustOctober), i.e., the aerosols tend to reduce cloudiness during the anomalous dry transition season and increase cloudiness when the transition season is relatively humid. 11. EXTREME EVENTS IN THE CONTEXT OF CLIMATE CHANGE: IS THE DROUGHT OF AMAZONIA IN 2005 AN INDICATOR OF HOW EXTREMES WILL BE IN A WARMER CLIMATE? In 2005, large sedtions of the western Amazon basin experienced the most severe drought in the past 40 years and also one ofthe most intense of the last 100 years. During this event, the worsening drought forced the Brazilian government to extend emergency warnings across Amazonas State. The military was called in to distribute supplies and medicine to tens of thousands of people. This drought provoked extensive forest fires in the State of Acre, although the area has yet to be estimated [Brown et al., 2006]. Navigation along sections of the Rio Madeira and upper and central Amazon River (known in Brazil as the Rio Solimoes) had to be suspended because the water levels fell to extremely low levels. Various Amazonian countries declared a state of public emergency in September 2005. The drought left thousands of people short of food, caused problems in river transportation, agriculture, and generation of hydroelectricity. It also affected directly and indirectly the populations living along the Amazon River tributaries. An observational study of the causes of this drought by Marengo et ai, [2007] suggests that the drought was not caused by El Nino but by (l) the anomalously warm tropical North Atlantic, (2) the reduced intensity in northeast trade wind moisture transport into southern Amazonia during the peak summertime season, and (3) the weakened upward
MARENGO ET AL. 286
287
CLIMATE-VEGETAnON FEEDBACK AND IMPACTS ON WATER RESOURCES
motion over this section of Amazonia, res~lting in reduced convective development and rainfall. The dIfference between the spatial felitures of the El Nino drought years and that of 2005 and 1963-1964 was that the later two stluck ha~dest in western and southern Amazonia, a feature not ~ssocIated with a typical EI Nino but probably with the tropIcal North Atlantic being warmer and more active than normal. While several studies have analyzed the droughts of 19821983 [Aceituno, 1988] and 1997-1998 [Nepstad eta~., 1999] and their impacts on climate, hydrology, and fires m Amazonia, there are only casual references to the drou~ht eve~t of 1963-1964. Figure 7 shows the levels of the ~1O N~gro at Manaus and the levels of the Amazon River m Iqmtos. The levels at Manaus are shown for 2005 and other drought
years while the Iquitos levels are shown just for 2005, both com~ared to the long-term means. In Manaus, the l~we~t levels were detected during the drought of 192~, ~hlle III 2005 reductions in the levels were detected startmg m June 2005' and reaching the lowest values in August [Marengo et at., 2008a, 2008b; Zeng et at., 2008]. , The drought of 1964 was detected all year long, wh,Ile the drought in 2005 appears only after August 2005, wIth, the levels before May 2005 being above the normal. In Iqmtos, the levels were below normal from JanuaI!' ~005" rea?hing their lowest in September 2005. A sImIlar sItuation was observed in the Rio Solimoes levels at F.onte. Boa ~~d Tabatinga, and in the levels of the Amazon RIver m LetiCI~ (Colombia). In 2006, the levels went back to normal until
30 25
_ _ MLT
-II- 2004·05 , 1925·26 -1963·64 --1982·83 " • "1997·98
5 Rio Negro at Manaus·Dry years (anomalies)
0
o
s
A
J
J
M
A
M
F
J
D
N
118 _ _ LMT
116
-
-2004·05
114
-:
'"!"
112 110
E 108 106 104
Amazon River at Iquitos-2005 (anomalies)
102
s
o
N
D
J
F
M
A
M
J
J
Months
. naus durin some drought years in Amazonia (2004-2005, 1925-1926, 1963gR'" I 't s during 2005 [Marengo et ai" 2008a, 2008b], comFigure 7. Levels of the RIO Negro at Ma 1964 1982-1983 1997-1998) and of the Amazon lYer 111 qUi 0 pared with the MLT (or long-term mean in English).
A
May 2006, and after that, the levels were about three to four below nOlmal until the/end of October 2006. However, the situation in 2006 was/hot that of the drought in 2005. This indicates tliat the drought of 2005 affected western Amazonia, unlik.lf El Nino-induced droughts which affect central and east~rn Amazonia. After October 2006, rainfall began again and alleviated the situation, and by Februmy 2006, rainfall was' above the normal, producing floods. This event is illustrative of the consequences of a future warmer climate, in which the probability of events like this may increase by the end of the twenty-first century, as simulated by the HadCM3 GCM. In addition, land use changes, and the increased fires and the subsequent injection of aerosols to the atmosphere have the potential to affect the onset and the amount of rainfall in the region [Andreae et at., 2004]. However, this does not imply that changes in the large-scale circulation leading to this drought in 2005 were a consequence of regional deforestation or of global warming and climate change; no evidence either for or against this possibility has yet been established. As extreme climatic events, such droughts induced either by natural climate variability or human activities could fragment the forest of Amazonia and transform an area of approximately 600,000 km2 into a savanna These results, reported by Hutyra et at. [2005], were obtained by mapping areas of the forest most sensitive to drought using rainfall records during the last 100 years. The region corresponding approximately to 11 % of the forest area extending fi'om Tocantins to Guiana would be the most affected. We have learned about impacts on the population, biodiversity, and local and regional economies by studying the 2005 drought in Amazonia. The impacts on the local economy were mainly reflected in the closing of the airpOlis due to the large amounts of smoke and the closing of pOlis due to the extremely low levels in river which prevented navigation. To this, we should add the problems of the population due to the collapse oflocal agriculture, the number ofpeople treated in hospitals due to thermal stl'ess, and respiratOly and intestinal diseases from smoke fi'om forest fires and polluted water. This showed that people in Amazonia are vulnerable to drought, and this vulnerability could be aggravated in warmer-drier climates in the future as suggested by some climate change projections. In a recent paper, Cox et at. [2008] examine the possible links of the drought of Amazonia in 2005 to climate change. This was done by comparing the relationship between Atlantic SST patterns and rainfall in Amazonia inferred from observations, to those predicted using a GCM. They analyze results from the HadCM3LC coupled climate-carbon cycle model. HadCM3 has performed well in GCM intercomparison exercises and was recently selected as one of the two
GCMs, which simulate the Amazonian climate with reasonable accuracy [Cox et at., 2008]. HadCMLC, in addition, includes dynamic vegetation and an interactive carbon cycle, so that atmospheric C02 concentrations can be updated based on anthropogenic emissions, taking account of the effects of climate change on ocean and land C02 uptake. They consider two separate simulations with HadCM3LC for the period from 1860 to 2100, In both cases, the model is driven with C02 emissions consistent with the IS92a scenario, which is approximately in the center ofthe spread of future emissions represented by the more recent SRES scenarios. Both model experiments also include prescribed time-varying concentrations of trace GHGs (CH4, N20) based on IS92a. The second run additionally includes changes in tropospheric and stratospheric ozone, solar variability, and most notably, forcing from sulfate and volcanic aerosols. This model experiment was able to reproduce the observed warming and C02 increase over the twentieth centuly to good accuracy, especially when a revised estimate of the net C02 flux from land use change was used. It is shown that the reduction of dly season (July-October) rainfall in western Amazonia is well-conelated with an index ofnOlih-south SST gradient across the equatorial Atlantic related to the Atlantic Multidecadal Oscillation (AMO). Their climate model can reproduce this relationship and also the observed twen~ieth centmy decadal variability in the AMO when the effects of aerosols are included. These results suggest that anthropogenic aerosols have acted to suppress waIming of the subtropical nOlih Atlantic relative to the south and thereby delayed drying in western Amazonia. Simulations with the same model for the twenty-first centuly show a strong tendency for the SST conditions associated with the 2005 drought to become much more common under increased C02 and reduced nOlihern hemisphere aerosol loadings. This drought was the most severe the past 40 years and also one of the most intense of the last 100 years. Model experiments with changes in C02 concentration and aerosols suggest that the probability of a "2005-like" drought will be 1-in-2 year event by 2025 and a 9-in-10 year event by 2060. 12. IMPACTS OF CHANGES IN CLIMATE AND HYDROLOGY IN WATER RESOURCES IN AMAZONIA All these projected changes in Amazonian climate and hydrology may have climatic, ecological, and environmental implications for the region, the continent, and for the globe. A sound knowledge of how the natural system functions is thus a prerequisite to defining optimal development strategies. The complex interactions between the soil, vegetation, and climate
288
MARENGO ET AL.
CLIMATE-VEGETAnON FEEDBACK AND IMPACTS ON WATER RESOURCES
must be measured and analyzed so that the limiting factors to vegetation growth and soil conservation can be established. New knowledge and improved understanding of the functioning of the AJazonian system as an integrated entity and of its interaction with the Ealih system will support development of national and regional policies to prevent the exploitation trends from bringing about irreversible changes in the Amazonian ecosystem. Such knowledge, in combination with enhancement of the research capacities and networks between the Amazonian countries, will ~timulate land managers and decision makers to devise sustainable alternative land use strategies along with forest preservation strategies. These changes due to an increase of GHG concentrations may result in a Walmer and drier Amazonia. Conditions would then allow the propagation of forest fires, and the tropical forest collapsing, as it changes from a sink of carbon to a source of carbon, and becomes savanna-type vegetation by the years 2040-2060. The drier climate in the region would also impact levels of rivers in the region and also moisture export for other regions outside Amazonia, thus also affecting the hydrological cycle outside the region. A variety of human activities can act to modify various aspects of climate and the surface hydrologic systems. Historically, land surface changes in Amazonia intensified in the mid- and early 1970s, when strategic governmental plans first attempted to promote the economic development of the
region. Those plans included the constmction of extensive roads throughout the basin and the implementation of fiscal incentives for new settlers, triggering a massive migration of landless people into the region. The constmction of reservoirs for hydroelectric generation in Amazonia has some impacts on the hydrological regime as well as on the biodiversity and on the water quality [Tundisi et al., 2002], depending on the size and inundated area of the tropical rain forests. Brazil has five reservoirs operating for hydroelectricity generation (Coaracy Nunes, Cuma-Una, Tucumi, Balbina, and Samuel) and six other planned to be built (Manso, Cachoeira, Ji-Parana, Karanao, Barra do Peixe, and Couto Magalh1ies). Land use changes have also been repOlied near the site of the reservoir due to human settlements in the region. Table 2 lists natural and anthropogenic forcings that affect water resources in Amazonia, the expected impacts on water resources of the region, and the possible consequences on the components of the climate system. The major forcings can be natural or human induced, and their impact could be on various timescales: intraseasonal (the onset of the rainy season in the region), interanriual (drought of flood associated with El Nino) and on long-term (changes in vegetation, soil hydrology, increase of GHG concentration). The table can serve as a starting point in the assessment of possible impacts of future climate change in Amazonia.
Table 2. Natural and Anthropogenic Forcings That May Affect Climatic and Hydrological Tendencies in Amazonian Countries
Forcing (Natural or Anthropogenic)
Impacts on Water Resources
El Nino and tropical Atlantic sea surface temperature anomalies
Changes in the rainfall distribution in Amazonia
Climate change due to increase in concentration of greenhouse gas (GHG)
Possible changes in the hydrological cycle; changes in the energy balance and warm· ing; changes in biodiversity and natural ecosystems
Deforestation and land use change
Possible changes in the hydrological and energy cycles. Changes in water quality and chemistry due to deforestation in the east flank of the Andes (Upper Amazon countries) Changes in the water and energy cycles; changes in the air quality
Biomass burning (natural and man-made)
Consequences Drought in northern Amazon region; problems in transportation due to low river water levels; high risk of forest fires at seasonal level; impacts on natural river ecosystems; impacts on agriculture; impacts water storage for hydroelectric generation Dynamics of vegetation affected; Amazon forest die and become savanna; drying of the Amazon region; floods or extremely low water levels likely to occur; more fi'equent forest fires; impacts on water storage for hydroelectric generation Regional rainfall reduction; regional warming; erosion; sedimentation along the main channel and accumulation of sediments in reservoirs; possible effect on water quality and biodiversity Impacts on the onset of the rainy season and physics of rainfall; impacts on the air quality and sensitivity to warming due to release of large amounts of GHG and aerosols
Decreased precipitation during dIy months will affect many Amazonian streams and fi'eshwater systems. Small, shallow habitats (ponds, headwater streams, marshes, and small lakes) willliktly experience the first effects of reduced precipitation [Cmjenter et al., 1992]. While prospects for successful relocation of spawning activities for fishes exist, some ~ay be thwarted by the strong imprinting and homing behaVIOr present in'many species. Acknowledgments. We would like to thank the Brazilian National Climate Change Program from the Ministry of Science and Technology ~CT; the Brazilian National Research Council CNPq; the UK ForeIgn and Commonwealth Office Strategic Programme FU~d (SPF) via .the projects "Using Regional Climate Change Scenanos for Studies on Vulnerability and Adaptation in Brazil and South America" and "Dangerous Climate Change in Brazil" (PGL GCC 0207); the Joint DECC, Defra, and MoD Integrated Climate Programme-DECClDefi'a (GA0110l), MoD (CBC/2B/0417 Annex. C5); the GEOMA; and the LBA2 Millennium Institute~ and the INCT-Mudanr;as Climaticas from the MCT-CNPq.
REFERENCES Aceituno, P. (1988), On the functioning ofthe Southem Oscillation in the South-American Sector .1. Surface climate, Mon. Weather Rev" 116, 505-524. Ambrizzi, T., R. Rocha, I. A, Marengo, I. Pisnitchenko, and L. Alves (2007), Cenarios Regionalizados de Clima no Brasil para o Seculo XXI: Proje<;i5es de Clima Usando TI'I!s Modelos Regionais, Relatorio 3, Ministerio do Meio Ambiente-MMA Secretaria de Biodiversidade e Florestas-SBF, Diretoria d~ Conservar;ao da Biodiversidade-DCBio Mudanr;as Climaticas Globais e Efeitos sobre a Biodiversidade-Sub projeto: Caracterizar;ao do clima atual e definir;ao das alterar;oes climaticas para 0 territ6rio brasileiro ao longo do Seculo XXI. Brasilia, Fevereiro. Andreae, M., D. Rosenfeld, P. Artaxo, A A Costa, G. P. Frank, K. M. Longo, and M. A F. Silva Dias (2004), Smoking rain clouds over the Amazon, Science, 303,1337-1342. Artaxo, P., et al. (2009), Aerosol particles in Amazonia: Their compositi.on, role in the radiation balance, cloud formation, and nutnent cycles, Geophys. Monogr. Ser., doi:l0.1029/ 2008GM000778, this volume. Betts, R., P. Cox, M. Collins, P. Harris, C. Huntingford, and C. Jones (2004), The role of ecosystem-atmosphere interactions in simulated Amazonian precipitation decrease and forest dieback under global change wanning, Theor. Appl. Climatol., 78, 157-175. Boulanger, J.-~., F. Martinez, and E. C. Segura (2006), Projection of future clImate change conditions using IPCC simulations, neural networks and Bayesian statistics. Part 1: Temperature mean state and seasonal cycle in South America, Clim. Dyn., 27, 233-259, doi: 10.1 007/s00382-006-0134-8. Boulanger,. I.-P., F. Martinez, and E. C. Segura (2007) Projection of future cIrmate change conditions using IPCC simulations, neural
289
networks and Bayesian statistics. Part 2: Precipitation mean state and seasonal cycle in South America, Clim. Dyn., 28, 255-271, doi: 10.1007/s00382-006-0 182-0. Brown, I. F., W. Schroeder, A Setzer, M. Maldonado, N. Pantoja, A Duarte, and J. Marengo (2006) Fires in southwestem Amazonian rain forests, Eos Trans. AGU, 87(26), 253. Callede, I., I. L. Guyot, J. Ronchail, Y. L'Hote, H. Niel, and E. De ~liveira (2004), Evolution of the River Amazon's discharge at Obidos from 1903 to 1999, Hydrol. Sci. J, 49, 85-98. Carpenter, S. R, S. G. Fisher, N. B. Grimm, and J..F. Kitchell (1992), Global change and freshwater ecosystems, Annll. Rev. Bcol. Syst., 23, 119-139. Cav~lca~ti, I. F., I. Camilloni, and T. Ambrizzi (2006), Escenarios chmatlcos regionales, in EI Cambio Climatico en la Cuenca del Plata, edited by V. BaiTOs, R Clarke, and P. Silva Dias, chap. 13, CONICET, Buenos Aires. Chen, T.-C..' I.~H. Yoon, K. I. St. Croix, and E. S. Takle (2001), Suppressmg Impacts of the Amazonian deforestation by the global circulation change, Bul!. Am. Meteorol. Soc. 82 2209-2215 Christensen, I. H., T. R Carter, M. Rummukaine~, a~d G. Amana~ ti~is (2007), Evaluating the performance and utility of regional cIrmate models: The PRUDENCE project, Clim. Change, 81, 1-6. Co~, M., ~. H. Costa, A Botta, and C. Birkett (2002), Long term SImulations of discharge and floods in the Amazon Basin, J Geophys Res., 107(D20), 8044, doi:IO.1029/2001JD000740. Costa, M. H., and J. AI Foley (2000), Combined effects of deforestation and doubled' atmospheric CO 2 concentrations on the climate of Amazonia, J Clim., 13,18-34. Costa, M. H., A ~otta~ and I. A Cardille (2003), Effects of largesc.ale changes m land cover on the discharge of the Tocantins River, Southeastern Anlazonia, J Hydrol., 283, 206-217. Cox, P., ~. Betts, C. Jones, S. Spall, and T. Totterdell (2000), AcceleratIOn of global warming due to carbon-cycle feedbacks in a coupled climate model, Nature, 408, 184-187. Cox, P., R. Betts, M. Collins, P. Harris, C. Huntingford, and C. Jones (2004), Amazonian forest dieback under climate-carbon cycle projections for the 21stcentuly, Theor. Appl. Clilllatol. 78, 137-156. ' Cox, P., P. Harris, C. Huntingford, R Betts, M. Collins, C. Jones, T. Jupp, J. Marengo, and C. Nobre (2008), Increase risk ofAmazonian Drought due to decreasing aerosol pollution, Nature 453 212-216. ' , Dickinson, ~., and A Henderson-Sellers (1988), Modeling tropical deforestatIOn: A study of GCM land-surface parameterization Q. J. R. Meteorol. Soc., 114,439-462. ' Dickinson, R., andP. Kennedy (1992), Impacts on regional climate .of Amazon deforestation, Geophys. Res. Lett., 19, 1947-1950. Dumeyer, P., and I. Shukla (1994), Albedo as a modulator of climate response to tropical deforestation, J. Geophys. Res., 99 20,863-20,877. ' Friedling.stein, P., et al. (2006), Climate-carbon cycle feedback an~lysls, results from the C4MIP model intercomparison, J Chm., 19, 3337-3353, doi: 10. 11 75/JCLI3800. 1. Fu, R., R..E. Dickinson, M. X. Chen, and H. Wang (2001), How do tropical sea surface temperatures influence the seasonal
290
MARENGO ET AL.
CLIMATE-VEGETATION FEEDBACK AND IMPACTS ON WATER RESOURCES
distribution of precipitation in the equatorial Amazon?, 1. Clim., 14, 4003-4026. Gentry, A., and H. J. Lopez-Parodi (1980), Deforestation and increased 1il:ooding of the upper Amazon, Science, 210, 1354-
measured ABRACOS vegetation characteristics, in Amazonian Deforestation and Climate, edited by J. Gash et aI., pp. 549576. Li, W., R. Fu, and R. E. Dickinson (2006), Rainfall and its seasonality over the Amazon in the 21 st century as assessed by the coupled models for the IPCC AR4, 1. Geophys. Res., 111, D02111, doi: 10.1029/200510006355. Longo, K. M., S. R. Freitas, M. O. Andreae, R. Yokelson, and P. Al1axo (2009), Biomass burning in Amazonia: Ernissions, long-range transport of smoke, and its regional and remote impacts, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000847,
1356. Good, P., J. A. Lowe, M. Collins, and W. Moufouma-Okia (2008), An objective tropical Atlantic sea surface temperature gradient index for studies of south Amazon dry-season climate variability and cl;ange, Philos. Trans. R. Soc. Ser. B., 363(1498), 17611766, doi: 10.1 098/rstb.2007 .0024. Grimm, A M., and A A. Natori (2006), Climate change and interthis volume. annual variability of precipitation in South America, Geophys. Manzi, 0., and S. Planton (1996), Calibration of a GCM using Res. Lett., 33, Ll9706, doi:l0.102912006GL026821. ABRACOS and ARME data and simulation of Amazonian defoHahmann, A, and R. Dickinson (1995), Performance and sensitivrestation, in Amazonian Deforestation and Climate, edited by ity of the RCCM2/BATS model to tropical deforestation over J. H. C. Gash et aI., John Wiley, New York. the Amazon Basin, paper presented at General Assembly XXI, Marengo, J. A (2003), Condiyoes climiiticas e recursos hidricos no In!. Union of Geod. and Geophys., Boulder, Colo. Norte Brasileiro, in Clima e Recursos Hidricos no Brasil, vol. Henderson-Sellers, A., R. E. Dickinson, T. B. Durbidge, P. J. 9, edited by C. E. Tucci, and B. Braga, pp. 117-161, AssociaKennedy, K. Mcguffie, and A J. Pitman (1993), Tropical deyao Brasileira de Recursos Hidricos FBMCI ANA, Porto Alegre, forestation: Modeling local- to regional-scale climate change, J. Brasil. Geophys. Res., 98(D4), 7289-7315. Haylock, M. R., et al. (2006), Trends in total and extreme South Marengo, J. A (2004), Interdecadal variability and trends of rainfall across the Amazon basin, TheaI'. Appl. Climatol., 78, 79-96. American rainfall 1960-2000 and links with sea surface tempeMarengo, J. A (2006), On the hydrological cycle of the Amazon rature,1. Clim., 19,1490-1512. Basin: A historical review and current, Rev. Bras. Meteorol., 21, Houghton, R. A, M. Gloor, J. Lloyd, and C. Potter (2009), The 1-19. regional carbon budget, Geophys. Monogr. Ser., doi:l0.10291 Marengo, J. A, (2009), Long-term trends and cycles in the hy2008GM000718, this volume. drometeorology of the Amazon basin since the late 1920s, HyHutyra, L. R., J. W. Munger, C. A. Nobre, S. R. Saleska, S. A drol. Processes, doi:l0.1002/hyp.7396. Vieira, and S. Wofsy (2005), Climatic variability and vegetation vulnerability in Amazonia, Geophys. Res. Lett., 32, L24712, Marengo, J. A, and T. Ambrizzi (2006), Use of regional climate models in impacts assessments and adaptations studies from doi: 10.1029/2005GL024981. continental to regional and local scales, The CREAS (Regional Intergovernmental Panel on Climate Change (IPCC) (2007a), CliClimate Change Scenarios for South America) initiative in South mate Change 2007: The Physical Science Basis. Contribution 0/ America, Proceedings 0/8 ICSHMO, Foz do Iguar,:u, Brazil, Worlring Group I to the Fourth Assessment Report of the InterApril 24-28, pp. 291-296, INPE. govel'llmental Panel on Climate Change, edited by S. Solomon Marengo, J. A, and C. A Nobre (2001), The Hydroclimatological et al., 996 pp., Cambridge Univ. Press, New York. framework in Amazonia, in BiogeochemistlJl of Amazonia, edIntergovernmental Panel on Climate Change (IPCC) (2007b), Cliited by J. Richey, M. McClaine, and R. Victoria, pp. 17-42. mate Change 2007: Impacts, Adaptation and Vulnerability. ConMarengo, J., J. Miller, G. Russell, C. Rosenzweig, and F. tribution of Worlring Group II to the Fourth Assessment Report Abramopoulos (1994), Calculations of river-runoff in the GISS of the Intergovel'llmental Panel on Climate Change, edited by GCM: Impact of a new land surface and runoff routing model in M. L. Parry et aI., 976 pp., Cambridge Univ. Press, Cambridge, the hydrology of the Amazon River, Clim. Dyn., 10, 349-361 . U.K. Marengo', J. A., et al. (2003), Ensemble simulation of regional Jones, R. G., M. Noguer, D. Hassell, D. Hudson, S. Wilson, G. rainfall features in the CPTECICOLA atmospheric GCM. Skill Jenkins, and J. Mitchell (2004), Generating high resolution cliand Predictability assessment and applications to climate predicmate change scenarios using PRECIS, report, Met Office Hadley tions, Clim. Dyn., 21,459-475. Centre, Exeter, U. K Lean, J., and P. R. Rowntree (1993), GCM simulation of the impact Marengo, J. A, W. Soares, C. Saulo, and M. Nicolini (2004), Climatology ofthe LLJ east of the Andes as derived from the NCEP of Amazon deforestation on climate using an improved canopy reanalyses,1. Clim., 17,2261-2280. representation, Q. J. R. Meteorol. Soc., 119, 509-530. Lean, J., and P. R. Rowntree (1997), Understanding the sensitivity Marengo, J. A, L. Alves, M. Valverde, R. Rocha, and R. Laborbe (2007), Eventos Extremos em Cenarios Regionalizados de Clima of a GCM simulation of Amazonian deforestation to specification of vegetation and soil characteristics, J. Clim., 6, 1216-1235. Lean, J., and D. Warrilow (1989), Climatic impact of Amazon deforestation, Nature, 342, 311-313. Lean, J., C. Bunton, C. Nobre, and P. Rowntree (1996), The simulated impact of Amazonian deforestation on climate using
no Brasil e America do Sui para a Seculo XXI: Projer,:i5es de Clima Futuro Usando Tres Modelos Regionais, Relatorio 5, Ministerio do Meio Ambiente~MMA, Secretaria de Biodiversidade e Florestas~SBF, Diretoria de Conservayao da Biodiversidade~DCBio Mudanyas Climiiticas Globais e Efeitos sobre a
Biodiversidade~Sub projeto: Caracterizayao do clima atual e
definiyao das alterayoesclimiiticas para 0 territ6rio brasileiro ao longo do Seculo XXI, Brasilia, Fevereiro. Marengo, J. A., C. N~b're, J. Tomasella, M. Oyama, G. Sampaio, H. Camargo, L. A}:ves, and R. Oliveira (2008a), The drought of Amazonia in 2005, 1. Clim., 21, 495-516. Marengo, J. A., C. Nobre, J. Tomasella, M. Cardoso, and M. Oyama (2008b), Hydro-climatic and ecological behaviour ofthe drought of Amazonia in 2005, Philos. Trans, R, Soc, Ser, B, 21, 1-6. Marengo, J. A., R. Jones, L. Alves, and M. Valverde (2009a), Future change of temperature and precipitation extremes in South America as derived from the PRECIS regional climate modeling system, Int. 1. Climatol., doi: 1O.1002/joc. 1863. Marengo, J. A, T. Ambrizzi, R. P. Rocha, L. M. Alves, S. V. Cuadra, M. C. Valverde, S. E. T. Ferraz, R. R. Torres, and D. C. Santos (2009b) Future change of climate in South America in the late XXI centulY: Intercomparison of scenarios from three regional climate models, Clim. Dyll., in press. Meehl, G. A, et al. (2007), Global climate projections, in Climate Change 2007: The Physical Science Basis: Contribution of
War/ring Group I to the Fourth Assessment Report a/the Intergovel'llmental Panel on Climate Change, edited by S. Solomon et aI., pp. 747-846, Cambridge Univ. Press, Cambridge, U. K Miller, J., G. Russell, and G. Caliri (1994), Continental Scale River Flow in climate models, 1. Clim., 7, 914-928. . Milly, P. C. D., K A. Dunne, and A V. Vecchia (2005), Global pattern ofh'ends in streamflow and water availability in a changing climate, Nature, 438, doi: 1O.1038/nature043 12. Nepstad, D., et al. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505-508. Nijssen, B., D. Lettenmaier, X. Liang, S. Wetzel, and E. F. Wood (1997), Streamflow simulation for continental-scale river basins, Water Resour. Res., 33(4), 711-724. Nijssen, B., G. O'Donnell, and D. Lettenmaier (2001), Predicting the discharge of Global Rivers, 1. Clim., 14, 3307-3323. Nobre, C. A, P. J. Sellers, and J. Shukla (1991), Amazonian deforestation and regional climate change, J. Clim., 4, 957-988. Oyama, M. D., and C. A Nobre (2003), A new climate-vegetation equilibrium state for tropical South America, Geophys. Res. Lett., 30(23), 2199, doi:l0.1029/2003GLOI8600. Oyama, M. D., and C. A Nobre (2004), A simple potential vegetation model for coupling with the simple biosphere model (SIB), Rev. Bras. Meteorol., 1(2),203-216. Phillips, O. L., S. L. Lewis, T. R. Baker, K.-J. Chao, and N. Higuchi (2008), The changing Amazon forest, Philos. Trans. R. Soc. Ser. B., 363(1498), doi: 10.1 098/rstb.2007 .0033. Phillips, O. L., N. Higuchi, S. Vieira, T. R. Baker, K-J. Chao, and S. L. Lewis (2009), Changes in Amazonian forest biomass, dynamics, and composition, 1980-2002, Geophys. Monogr. Ser., doi:1O.1029/2008GM000739, this volume. Polcher, J., and K. Laval (1994a), The impact of African and Amazonian deforestation on tropical climate, J. Hydrol., 155, 389-405. Polcher, J., andK Laval (1994b), A statistical study of the regional impact of deforestation on climate in the LMD GCM, Clilll. Dyn., 10,205-219.
291
Russell, G. J., and J. Miller (1990), Global river mnoff calculated from a global atmosphere general circulation model, 1. Hydrol., 155,241-254. Salati, E., and C. A. Nobre (1991), Possible climatic impacts of tropical deforestation, Clim. Change, 19,177-196. Salazar, L. F., C. A. Nobre, and M. D. Oyama (2007), Climate change consequences on the biome distribution in tropical South America, Geophys. Res. Lett., 34, L09708, doi:1O.10291 2007GL029695. Saleska, S., H. da Rocha, B. Kmijt, and A. Nobre (2009), Ecosystem carbon fluxes and Amazonian forest metabolism, Geophys. Monogr. Ser., doi: 1O.1029/2008GM000728, this volume. Sampaio, G. (2008), Climatic consequences of gradual conversion of Amazonian Tropical Forests into degraded pasture or soybean cropland: A GCM simulation study, PhD thesis in Meteorology-(INPE-15263-TDII1346), 417 pp., Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos. (Available at http:// urlib.netlsid.inpe.br/mtc-mI7@80/2008/02.28.17.17) Sampaio, G., C. Nobre, M. H. Costa, P. Satyamurty, B. S. SoaresFilho, and M. Cardoso (2007), Regional climate change over eastern Amazonia caused by pasture and soybean cropland expansion, Geophys. Res. Lett., 34, LI7709, doi: 10.10291 2007GL030612. Schroeder, W., A Alencar, E. At'ima, and A Setzer (2009), The spatial distribution and interannual variability of fire in Atnazonia, Geophys. Monogr. Ser., doi:1O.1029/2008GM000724, this volume. Shukla, J., C. Nobre, and P. Sellers (1990), Amazonia deforestation and climate change, Science, 247,1322-1325. Sitch, S., et al. (2008), Evaluation of the telTestrial carbon cycle, future plant geography and climate-carbon cycle feedbacks using five Dynamic Global Vegetation Models (DGVMs), Global Change BioI., 14, 2015-2039. Soares, W. R., and J. A Marengo (2008), Assessments of moisture fluxes east of the Andes in South America in a global warming scenario,!nt.1. Climatol., 29, 1395-1414, doi: 10.1002/joc.1800. Sud, Y. c., R. Yang, and G. K Walker (1996a), Impact of in situ deforestation in Amazonia on the regional climate: General circulation model simulation stlldy, J. Geophys. Res., 101(D3), 7095-7109. Sud, Y., G. Walker, J.-H. Kim, G. Liston, P. Sellers, and W. Lau (1996b), Biogeophysical consequences of a tropical deforestation scenario: A GCM simulation study, 1. Clim., 9, 3226-3247. Tebaldi, C., K Haohow, J. Arblaster, and G. Meehl (2007) Going to extremes: An intercomparison of model-simulated historical and future changes in extreme events, Clim. Change, 79, 185211, doi:l0.1007/s10584-006-9051-4. Trenberth, K. E., et al. (2007), Observations: Surface and atmospheric climate change, in Climate Change 2007: The Physical
Science Basis: Contribution 0/ War/ring Group I to the Fourth Assessment Report 0/ the Intergovel'llmental Panel on Climate Change, edited by S. Solomon et aI., pp. 235-336, Cambridge Univ. Press, Cambridge, U. K. Tundisi, J. G., T. M. Tundisi, and O. Rocha (2002), Ecossistemas de Aguas interiors, in Aguas Doces do Brasil: Capital Ecologico, Usa e Conservar,:iio, 2nd ed., edited by A Rebouyas, B. Braga, and J. G. Tundizi, pp. 153-192.
292
CLIMATE-VEGETATION FEEDBACK AND IMPACTS ON WATER RESOURCES
Vera, C., et al. (2006), Towards a unified view of the American Zhang Y., J. M. Wallace, and D. S. Battisti (1997) ENSO-like interdecadal variability: 1900-93, Journal of Climate, 10, 1004monsoon systems, J. Clim., 19,4977-5000. 1020. Victoria, R., L. Martinelli, J. Moraes, M. V. Ballester, A., Krushche, G. PeYlegrino, R. Almeida, and J. Richey (1998), Surface Zhang, Y., Y. Xu, W. Dong, L. Cao, and M. Sparrow (2006), A future climate scenario of regional changes in extreme climate air temperature variations in the Amazon region and its border events over China using the PRECIS climate model, Geophys. during this century, J. Clim., 11, 1105-1110. Res. Lett., 33, L24702, doi:1O.1029/2006GL027229. Voldoire, A. (2006), Quantifying the impact of future land-use changes against increases in GHG concentrations, Geophys. Res. Zhang, Y., R. Fu, H. Yu, R. E. Dickinson, R. N. Juarez, M. Chin, and H. Wang (2008), A regional climate model study of Lett., 33, L04701, doi:10.102912005GL024354. biomass burning aerosol impacts land-atmosphere interactions Voldoire, A., and J. F. Royer (2004), Tropical deforestation and over the Amazon, J. Geophys. Res., 113, D14S15, doi: 10.1029/ climate variability, Clim. Dyn., 22,857-874. 2007JD009449. Walker, G., Y. Sud, and R. Atlas (1995), Impact of the ongoing Amazonian deforestation on local precipitation: A GCM simulation study, Bull. Am. Meteorol. Soc., 76,346-361. Wilby, R. L., and T. M. L. Wigley (1997), Downscaling general circulation model output: A review of methods and limitations, Prog. Phys. Geogr., 21,530-548. R. A. Betts, Met Office Hadley Centre, Fitz Roy Road, Exeter, Yu, H. B., R. Fu, R. E. Dickinson, Y. Zhang, M. Chen, and H. Devon EX1 3PB, UK. ([email protected]) Wang (2007) Dynamical and thermodynamic controls on smokeP. M. Cox, School of Engineering, Computer Science and cloud interactions over the Amazon, Remote Sens. Environ., 111, Mathematics, University of Exeter, Exeter, Devon EX4 4QF, UK. 435-449. Zeng, N., J. Yoon, J. A. Marengo, A. Subramaniam, C. A. Nobre, ([email protected]) J. Marengo, C. A. Nobre, L. Salazar, and G. Sampaio, Centro and A. Mariotti (2008), Causes and impacts ofthe 2005 Amazon de Ciencias do Sistema Terrestre'CCSTlInstituto Nacional de Pesdrought, Environ. Res. Lett., 3, 1-9. Zhang, H., A. Henderson-Sellers, and K. McGuffie (2001), The quisas Espaciais, 12630-000 Cachoeira Paulista, SP, Brazil. (jose. compounding effects of tropical deforestation and greenhouse [email protected]; [email protected]; [email protected]. br; [email protected]) warming on climate, Clim. Change, 49, 309-338.
Biogeochemistry and Ecology of Terrestrial Ecosystems of Amazonia Yadvinder Malhi Environmental Change Institute, School of Geography and the Environment, University of Oxford, Oxford, UK
Eric A. Davidson Woods Hole Research Center, Falmouth, Massachusetts, USA
The lastdecade of research associatedwith the Large-ScaleBiosphere-Atmosphere Experiment inAmazonia (LBA) has led to substantial advances in our understanding ofthe biogeochemistry and ecology ofAmazonian forests and savannas, in particular in relation to the carbon cycle ofAmazonia. In this chapter, we present a synthesis of results and ideas that are presented in more detail in subsequent chapters, drawing together evidence from studies' of forest ecology, ecophysiology, trace gas fluxes and atmospheric flux towers, large-scale rainfall manipulation experiments and soil surveys, satellite remote sensing, and quantification of carbqn and nutrient stocks and flows. The studies have demonstrated the variability of the functioning and biogeochemistty of Amazonian forests at a range of spatia~ and temporal scales, and they provide clues as to how Amazonia will respond to ohgoing direct pressure and global atmospheric change. We conclude by highlighting key questions for the next decade of research to address. 1. INTRODUCTION The Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) has resulted in an unprecedented international effort to understand the functioning of the vegetation of Amazonia and its interaction with the atmosphere and hydrosphere. As a result, Amazonia is now almost certainly the best studied major tropical forest region of the world, although our journey to a comprehensive understanding of the functioning of this system is only just beginning. Some of the key questions that LBA asked in relation to the vegetation and soils of Amazonia included the following: (1) How does the structure and functioning of Amazonian forests vary across the region; what factors drive this variation? (2) How much carbon is stored in Amazonian vegetation .
and soils; is this carbon store increasing or decreasing in response to contemporary environmental change? (3) How does the supply of nitrogen, phosphorus, and other nutrients affect Amazonian forests and the viability and sustainability of management practices after conversion of forest to other land uses? (4) How does Amazonian vegetation respond to seasonal, interannual, and long-term drought? (5) What role does Amazonia play in global climatic teleconnections and in the budgets ofthe major atmospheric trace gases? (6) How are Amazonian forests changing, and how will they change in response to climate change? In this introduction, we provide a briefoverview ofthe major themes discussed in this section. We do not introduce the chapters in sequence but rather when appropriate to our narrative. 2. NUTRIENT SUPPLY AND LIMITATIONS
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2009GM000905
The supply of nutrients affects forest composition and function, and also the viability ofagriculture or cattle production on converted forest lands, and the rate of recuperation 293
294
MALHI AND DAVIDSON
BIOGEOCHEMISTRY AND ECOLOGY OF AMAZONIAN TERRESTRIAL ECOSYSTEMS
of forests on degraded lands. Davidson and Martinelli [this volume], Luiziio et al. [this volume], Malhi et al. [this volume], and qoyd et al. [this volume] review ~ range ofLBA research on nutrient limitation in Amazoma. It has been suggested that the main limiting nutrient in many lowland tropical forests growing on highly weathered soils may be phospho!,us rather than nitrogen. LBA research over recent years has both confirmed and refined this picture. On the regional scale, the growth rate,and net primary productivity of trees in old-growth forests does seem most strongly correlated with leaf phosphorus concentration, which in turn is strongly correlated to soil phosphorus status [Malhi et al., this volume; Lloyd et al., this volume]. Old-growth Amazonian forests appear to have plentiful nitrogen supply, which leads to a leaky nitrogen cycle and significant emissions of gaseous NzO from the soil [Bustamante et al., this volume]. However, nitrogen limitation does appear on sandy soils, in montane regions, and perhaps also in the dry season in seasonally dry forests, when dry leaf litter and soil surface brings decomposition and mineralization to a halt [Lloyd et al., this volume]. Perhaps most impoliantly, secondary forests appear to be nitrogen-limited because of losses of actively cycling N during the agricultural phase [Luiziio et al., this volume], which takes several decades of secondary forest succession to replenish [Davidson and Martinelli, this volume]. Agricultural systems can therefore be limited in both phosphorus and nitrogen as well as other nutrients [Luiziio et al., this volume]. Hence, the broad picture ofphosphorus limitation in lowland Amazonian forests has been confirmed, but research has also revealed an ephemeral but impoliant role for nitrogen limitation in the dynamics of forest and nonforest landscapes. Nutrient management is a key aspect of research on sustainable agriculture and forest management in the Amazon Basin [Luiziio et al., this volume]. 3. TRACE GAS EXCHANGES Amazonia plays a major global role not only in the cycling of water, energy, and carbon, but also in important trace gases such as nitric oxide (NO), nitrous oxide (NzO), and methane, which play important roles in atmospheric chemistry and can be major greenhouse gases. Bustamante et al. [this volume] review recent research findings on trace gas sources and sinks in Amazonia and the cerrado. The availability of nitrogen and the supply of water are major detenninants of natural emissions of NO and NzO. Land use change can cause a temporary increase in NzO and NO emissions, but cattle pastures older than a few years have lower emissions than native forests, due, in part, to declines in nitrogen availability [Luiziio et al., this volume]. Rates of uptake of atmospheric methane also decline as pastures age and as
soils become compacted. Large uncertainties in regional estimates remain regarding regional balances and the effects of land use change, and further change may occur as cattle pastures are replaced by fertilized row crop agriculture. 4. BIOGEOGRAPHY OF AMAZONIAN FORESTS AND SAVANNAS One of the achievements ofthe LBA era has been a significant improvement in understanding of the regional variation in vegetation structure and dynamics, and tree community composition across the Amazon region, and how these variations are related to soil and climatic conditions [Malhi et al., this volume; Phil/lips et al., this volume; Lloyd et al., this volume]. In particular, the Rede Amazonica de Inventarios Florestais (RAINFOR) long-term forest plots have described and explored a broad east-west gradient in forest function, and have shown how many traits in forest function can covary in response to shifts in environmental conditions. Lowland forests in western Amazonia tend to have higher wood productivity than those in eastern Amazonia, and thereby higher tree recruitment and mOliality rates, shorterlived trees, lower mean wood density, and thinner, more nutrient-rich leaves. Most surprisingly, perhaps, the higher productivity in western Amazonia is associated with slightly lower biomass, as forest turnover times are higher, and wood densities are slightly lower. It appears that soil fertility, and in particular phosphorus, drives this east-west gradient: soils in western Amazonia tend to be of Pleistocene or Holocene age, deposited on the floodplains of meandering rivers carlying recently eroded material off the Andes. Soils in the center and east tend to be ancient, heavily weathered, infertile Ferralsols. The second key environmental gradient in Amazonia is the broad rainfall gradient from the high rainfall climate of the northwest to the strong seasonality of the southeast. Another LBA focus has been on the impact of this rainfall seasonality on vegetation community and on the seasonality of vegetation function. The most important transition along this gradient is from forest to savanna, and this forestsavanna boundmy is broadly related to interannual frequency of intense drought. However, soil fertility also appears to have an influence, with fertile sites favoring seasonal tropical forest and infertile sites favoring savannas [Lloyd et al., this volume]. 5. METABOLISM AND ITS SEASONAL VARIATION A number of aspects of LBA research have focused on gaining detailed understanding at a few focal sites. Here a broad spectrum of measurements, some of them applied for
295
the first time in a tropical context, have painted a detailed interesting feature is that the highest forest net primmy propicture of the physiologY, metabolism, and cycling of car- ductivity measured, to date, in Amazonia has been for forbon, nutrients, and water in these tropical forest stands. A est growing on the highly fertile terra preta soils, created by particular feature h1}'g been the establishment of eddy covari- pre-Columbian indigenous communities [Malhi et al., this ance flux towers 5}la number of sites across Brazilian Ama- volume; Luiziio et aI., this volume]. zonia (results from these are reviewed by Saleska et al. [this 6. RESPONSE TO DROUGHT volume]; see also Houghton et al. [this volume] and Lloyd et al. [this volume]). These flux tower studies have provided new insights into the seasonality of carbon dynamics. Sites Perhaps one of the most remarkable components of LBA in the central Amazon appear to have little decline in forest was the implantation oflarge-scale rainfall manipulation exphotosynthesis in the dly season or even some enhancement periments: two droughting experiments in mature forests at of photosynthesis under cloud-free dly season conditions. Tapaj6s and Caxiuana, and a dly season watering experiment Many of these forests tend to have sufficient dly season at a secondmy forest at Castanhal. Meir et al. [this volume] water supply in most years because of the high water hold- synthesize the outcomes ofthese experiments and their impliing capacity of the well-drained but clay-rich soils, which cations for our understanding of the drought response of the can be accessed through deep root systems [Meir et al., Amazon forest. The two drought experiments showed that in this volume; Lloyd et al., this volume]. Forests in southern the first few years, the forest ecophysiology showed surprisAmazonia, however, appear to have greater decline to dly ing resilience to drought, partially through access to adequate season photosynthesis. This pattern of seasonal variation has water within the top few meters of the soil surface, through been broadly corroborated by observations of the Enhanced some access to deeper soil water stores through deep roots Vegetation Index derived from satellite imagely acquired and through the closure of leaf stomata to limit water loss. from the Moderate Resolution Imaging Spectroradiometer Soil respiration shows either no change or a slight decline (MODIS) sensor on the Terra and Aqua satellites [Saleska with imposition of drought, and overall, short-to-mediumterm drought did not seem to cause a substantial carbon efet al., this volume]. Eddy covariance measurements have enabled quantifica- flux from the forest ~o the atmosphere. This resilience seems tion of bulk ecosystem photosynthesis and respiration, and to have a limit, though, and after a few years of drought, there their sensitivity to seasonal and interannual environmental was a pulse of mort~lity at both sites. However, the immedivariation. At the same time, numerous studies of the vari- ate physiological mechanism of this mOliality is not known. ous component processes of production and respiration have The most likely explanations are that ongoing depletion of painted a detailed picture of the stand-level cycle of car- soil water reserves leads to hydraulic failure within plant vesbon production, allocation, respiration, and decomposition sels, or that the trees experience "carbon starvation," as res[Malhi et al., this volume]. When compared, the "top-down" piration costs outpace photosynthetic intake, or other factors flux tower or soil gas-exchange estimates and "bottom- such as pathogens. The imposition of drought also greatly up" component measurements tend to be surprisingly con- increased the flammability of the near-ground region, and sistent. These studies highlight that wood production, the it is likely that the spread of fire would have a greater role main subject of carbon-focused studies, accounts for only th~m ecophysiological drought resilience in determining the a small fraction (about 10%) of the ecosystem, carbon cy- response of Amazonian forests to drought. cle. More generally, the carbon-use efficiency, the fraction of gross photosynthesis that ends up producing new organic 7. CARBON STORES tissue, is as low as 30% in old-growth forests on infeliile soils in eastern Amazonia, but may rise to about 50% after Tropical forests are one of world's largest reservoirs of disturbance, and also may rise with increasing soil fertility carbon, but the exact magnitude of this store remains a fo(phosphorus concentration). Both belowground and above- cus of research. The question of carbon storage has moved ground productivity appear to rise as soil fertility increases, from being one of purely academic interest to one of policy with little shift in relative allocation, but disturbance ap- relevance, as global attention focuses on the impacts of carpears to cause a disproportionate allocation aboveground. . bon emissions from deforestation on global climate change The emerging picture is that shifts in carbon allocation may and potential mechanisms to pay for this carbon to remain be as important as shifts in photosynthesis and bulk respi- in living, healthy forests. Unceliainty in forest biomass is ration in determining the wood productivity, biomass, and the largest uncertainty term in estimating emissions from destructure of tropical forests, but our understanding of the forestation [Houghton et al., this volume]. Major sources of detelminants of allocation is much more limited. Another unceliainty include lack of sufficient data (from forest hare
296 BIOGEOCHEMISTRY AND ECOLOGY OF AMAZONIAN TERRESTRIAL ECOSYSTEMS MALHI AND DAVIDSON 297 vests) on the allometric relationship between tree biomass and tree dimensions, differences in methodology that make cross-site cOlnparison of biomass estimates difficult, and the sheer challenge of extrapolating from even a few hundred sample sites to a region as vast as Amazonia. Over the past decade, significant progress has been made on all fronts (as described by Malhi et al. [this volume]). Allometric relationships have been developed and tested in central and eastern Amazonia based on rigorous and careful harvesting of over 300 trees. Use of the Amazon-wide RAINFOR network has allowed for a consistent methodology to be applied to examine regional variations in biomass. Application of remote sensing and climatic data layers has allowed for more information-rich spatial eXh'apolation. These recent efforts suggest that about 120 Pg C is stored in live biomass in Amazonia. There are further steps to be taken, such as development of allomehies for other forest types or regions of Amazonia, the expansion and merging ofthe forest inventory and remote sensing approaches, and the incorporation of new remote sensing products such as latest-generation radar and lidar. Amazonian forests also contain substantial stores of carbon in and on the soil. Trumbore and de Camargo [this volume] review what is known about these stores and their stability or otherwise (with further insights given by Malhi et al. [this volume] and Lloyd et al. [this volume]). Carbon in the dead wood and litter layer can be a substantial and fairly labile store of carbon, as much as the labile stores in the soil. The deeper soil (below 10 em) holds large, fairly stable reserves of carbon in forms that have mean ages of many centuries, and in deep soils, this soil carbon store can be greater than the live biomass carbon store. These stores are, however, unlikely to show a major response to climate change on a centuly timescale. In understanding the response of tropical forest soils to climate or land use change, it is essential to take the very different residence times of different soil carbon pools into account; failure to do so can lead to substantial overestimation of the sensitivity of soil carbon reserves. 8. THE CARBON BALANCE OF AMAZONIA Quantifying the carbon balance of Amazonia has been an aspiration of LBA because it enables understanding of the possible carbon sequestration service provided by forests and the carbon emission damage being done by deforestation and degradation. It also opens a door into more mechanistic understanding of how Amazonia will respond to twenty-first century climate change. Houghton et al. [this volume] review the evidence informing our understanding of the contemporary carbon balance of Amazonia. A few key features stand out. First, deforestation is clearly a substantial source
of CO 2 to the atmosphere. Second, long-term observations of intact old-growth Amazonian forests suggest that they are increasing in biomass and absorbing carbon (the latter result still generates some conh'oversy, which is discussed by Phillips et al. [this volume] and Trumbore and de Camargo [this volume]). An acceleration of biomass production, perhaps stimulated by increasing atmospheric CO 2 or increased diffuse radiation, could lead to a modest soil carbon sink, but the heavily weathered forest soils of eastern Amazonia are unlikely to be significant sinks of C [Trumbore and de Camargo, this volume]. The deforestation carbon emission and intact forest carbon absorption are of similar magnitude and may approximately cancel each other out, resulting in a net carbon balance close to zero [Houghton et al., this volume]. Atmospheric approaches, ranging from global analyses of carbon dioxide concentration to short-time scale local aircraft studies through to eddy covariance flux towers, they each have their methodological issues and have thus far failed to conclusively quantify the net carbon balance of the region [Houghton et al., this volume]. A clear next step forward would be a long-term, multisite atmosphe~ic sampling campaign using tall towers and/or aircraft combined with surface observations of biomass change and forest carbon cycling, and detailed remote sensing estimations of monthly carbon emissions from fires. Such a plan is underway through the LBA programs Balanyo Regional de Carbono na Amazonia (BARCA) and Amazon Integrated Carbon Analysis (AMAZONICA), and may finally provide a conclusive answer to the carbon balance question.
10. CONCLUSION
Houg~ton, R. A, M. Gloor, 1. Lloyd, and C. Potter (2009), The regIOnal carbon budget, Geophys. MOl/ogl'. Sel'., doi: 10.10291
In summary, a decade ofLBA-associated research has led 2008GM0007l8, this volume. to substantial advafbes in understanding of the ecology and Lloyd, 1., M. L. Goulden, 1. P. Ometto, S. Patino, N. M. Fyllas, and biogeochemishy~of Amazonian forests, savannas, and conC. Po" Quesada (2009), Ecophysiology offorest and savanna vegetatIOn, Geophys. MOl/ogl'. Sel'., doi: lO.l02912008GM000740 velted landscapes. Only now are researchers in a position to ~w~~ , step back and synthesize these findings, and this book repLuizao, F. 1., P. M. Fearnside, C. E. P. Cerri, and 1. Lehmann resents a major milestone in that process. This process will (2009), The maintenance of soil fertility in Amazonian managed continue, as will new research spawned by these findings systems, Geophys. MOl/ogl'. Sel'., doi: lO.l029/2008GM000742 and new research programs implemented by the cadre of scithis volume. ' entists that have been trained through LBA. Inevitably, this Malhi, Y., S. Saatchi, C. Girardin, and L. E. O. C. Aragao (2009), The research throws up as many questions as it answers. Here production, storage, and flow of carbon in Amazonian forests, Geoare a few: Ifphosphorns does limit forest productivity, what phys. MOl/ogl'. Ser., doi:1O.102912008GM000779, this volume. physiological process is it limiting? How will response to Meir, P., et a1. (2009), The effects of drought on Amazonian rain twenty-first centuly environmental change ValY across speforests, Geophys. MOl/ogl'. Sel'., doi: 10.1 029/2008GM000882 this volume. ' cies, and what implications does this have for forest composition and ecology? How would a potential twenty-first Phillips, O. L., N. Higuchi, S. Vieira, T. R. Baker, K.-1. Chao, and S. L: Lewis (2009), Changes in Amazonian forest biomass, dycentuly temperature rise in excess of 4°C affect the forests na~l11cs, and composition, 1980-2002, Geophys. MOl/ogl'. Ser., and their fauna? What is the potential for agricultural intendOI:1O.l029/2008GM000739, this volume. sification on better managed soils, perhaps bon'owing lesSaleska, S., H. da Rocha, B. Knlijt, and A. Nobre (2009), Ecosons from the terra preta soils? How will the combination system carbon fluxes and Amazon forest metabolism, Geophys. of changing climate and changing land use and interactions MOl/ogl'. Ser., doi: 10.1 029/2008GM000728, this volume. of these two trends affect fire susceptibility of forests and Tr~mbore, S., and P. B. de Camargo (2009), Soil carbon dynamfuture forest-savanna transitions? How will climate change ICS, Geophys. MOl/ogl'. Ser., doi:lO.l02912008GM00074l this and land use change affect the quantity and quality of water volume. ' draining from Amazonian lands? What is the carbon balance ofAmazonia, and how does it change in drought years? Will newer approaches enable us to map forest biomass from satellites? All of these questions need answers, and some need urgent answers. The challenge is set for this coming decade of scientific research to deliver.
9. CHANGES IN INTACT VEGETATION OVER TIME If old-growth Amazonian forests are increasing in biomass, this begs the question why they are doing so, and what implications this has for the ecology and functioning of these forests. Phillips et al. [this volume] discuss these issues. A particularly remarkable feature is that forests appear to have accelerated in both tree growth and death rates, resulting in a reduction in overall turnover times. This in turn is likely to affect forest composition and ecology, favoring faster-growing disturbance species over slow growing species. The authors argue that the overall pattern and broad extent of the change suggests that a global atmospheric driver (probably either CO 2 or changing light quality) may be the most likely cause. Whatever the cause, Phillips et al. [this volume] underline that even the most remote Amazonian forests are changing and will continue to change as greenhouse gas concentrations and temperatures continue to rise. Understanding the nature, causes, and consequences of this change is one of the great challenges facing Amazon forest ecologists this centmy.
REFERENCES Bustamante, M. M. C., M. Keller, and D. A da Silva (2009), Sources and sinks of trace gases in Amazonia and the cerrado Geophys. MOl/ogr. Sel'., doi: 10.1029/2008GM000733, thi~ volume. Davidson, E. A, and L. A. Martinelli (2009), Nutrient limitations to secondmy forest regrowth, Geophys. MOl/ogl'. Sel'., doi:1O.l0291 2008GM000732, this volume.
E. A Davidson, Woods Hole Research Center, 149 Woods Hole Road, Falmouth, MA 02540-1644, USA Y. Malhi, Environmental Change Institute, School of Geography and the Environment, University of Oxford, South Parks Road Oxford OXI 3QY, UK. ([email protected]) ,
Nutrient Limitations to Secondary Forest Regrowth Eric A. Davidson Woods Hole Research Center, Falmouth, Massachusetts, USA
Luiz A. Mmiinelli CENA, University ofSilo Paulo, Piracicaba, Brazil
The old, highly weathered soils of the lowland forest within the Amazon Basin generally exhibit conservative P cycles and leaky N cycles. This generalization applies to mature forests, but accelerating land use change is altering Amazonian landscapes. About 16% of the original forest area has been cleared, and about 160,000 km2 is in secondary forest cover. Secondmy forests are common in agricultural regions, but few persist in one place for much more than 5 years. The nutrients within ephemeral forests are important for smallholder traditional slashand-burn agriculture and in alternatives developed to con1\erve nutrients. Forest clearing causes an initial loss of nutrients through timber harvesting, fire, erosion, soil gaseous emissions, and hydrologic leaching, with N losses exceeding P losses. In contrast, the Ca, Mg, and K present in woody biomass: are largely conserved as ash following fire, redistributing these nutrients to the soil. After the initial postclearing pulse of nutrient availability, rates of N cycling and loss consistently decline as cattle pastures age. Feliilization experiments have demonstrated that growth of young forests in abandoned agricultural land is nutrient limited. Several N cycling indicators in a secondmy forest chronosequence study also demonstrated a conservative N cycle in young forests. Variable N limitation in young forests helps explain a negative relationship observed between the burn frequency during previous agricultural phases and the rate of forest regrowth. Recuperation of the N cycle gradually occurs during decades of secondary forest succession, such that mature lowland forests eventually recover abundant N relative to a conservative P cycle. 1. INTRODUCTION Most of the Brazilian Amazonian forests fall within the categOly of lowland tropical forests growing on very old, highly weathered soils [Sombroek, 2000]. Soil age is imp 01'Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.102912008GM000732
.
tant because it affects nutrient availability to plants. Nitrogen (N) is derived primarily from the atmosphere and, therefore, is in short supply in young soils and gradually accumulates as soils age [Walker and Syers, 1976]. Hence, soils forming within a few thousands of years since glaciations, volcanic activity, or erosion events tend to have limited supplies of actively cycling N, whereas very old soils and the ecosystems that they support have had a long time to accumulate N. In contrast, rock-derived phosphorus (P) is often abundant in young soils but becomes bound in unavailable forms to
299
300
DAVIDSON AND MARTINELLI
NUTRIENT LIMITAnONS TO SECONDARY FOREST REGROWTH
the iron and aluminum oxides that form as soil weathering proceeds over thousands and millions of years [Uehara and Gilman, 19~1]. As a result of these age-depend~nt characteristics of soil properties, mature lowland tropIcal forests on highly weathered soils generally exhibit conservative P cycling processes, whereas conservative N cycling properties are more common on younger soils, including most temperate f~rests and tropical montane forests [Vitousek, 1984]. This effect of soil age on N anq P cycling characteristics was elegantly demonstrated along a soil age chronosequence in the Hawaiian Islands [Chadwick et al., 1999; Crews et al., 1995; Hedin et al., 2003]. Excluding montane systems (such as the western Amazon on the slopes of the Andes), soil age generally increases with decreasing latitude. In global-scale analyses, the N:P ratios of green foliage [Reich and Okeksyn, 2004] and litterfall [McGroddy et al., 2004] of mature forests have been shown to increase with decreasing latitude, further supporting the generalization that P becomes less available to plants (more conservative P cycle) and N becomes more available to plants (less conservative N cycle) as soils age. While these stoichiometric generalizations appear robust for mature forests, accelerating land use change is altering Amazonian landscapes, and the consequences for nutrient cycling in secondary forests growing after agricultural abandonment are unclear [McGrath et al., 2001]. In the Brazilian Amazon Basin, about 16% of the original forest area has been cleared (Instituto Nacional de Pesquisas Espaciais, Monitoramento da Cobertura Florestal da Amazonia pOl' Satelites Sistemas Prodes, Deter, Degrad e Queimadas 2007-2008, report, Sao Jose dos Campos, Brazil, 2008, available at http://www. 0 bt.inpe.br/prodes/Relatorio]rodes2008.pdf), but basin-scale estimates of secondary forest cover are more difficult to obtain because of frequent confusion among young secondmy forests and other cover types in satellite imagery. As much as 30-50% of cleared land has been estimated to be in some stage of secondary forest succession following agricultural abandonment [Hirsch et al., 2004]. NeejJ et al. [2006] estimated that from 1978 to 2002 there was an increase in the area of secondmy forests in the Amazon fi'om approximately 30,000 km2 to 160,000 km2, the latter figure representing about 20% of deforested land and about 4% of the natural forest area. Secondary tropical forests are playing an increasingly important role for maintaining genetic diversity [Vieira et al., 1996] and hydrological functioning of altered landscapes [Holscher et al., 1997b; Nepstad et al., 2001], but biogeochemical processes remain poorly studied in Amazonian secondary forest succession. Where fertilization is not economically viable, agricultural lands often become unproductive and are abandoned. A native secondary forest then begins to grow. Aggrading forests create a strong demand
for essential plant nutrients [Vitousek and Reiners, 1975]. It is often assumed that P limitation also applies to young tropical forests as well as mature forests, but N losses during land use change may alter the stoichiometric balance of N and P cycling processes. Nutrient cycling processes during secondary succession may have implications for the development of forest structure, such as stand density, height, leaf area index, and species composition. Our objective here is to review the evidence that nutrient cycling processes change significantly during secondmy forest succession in lowland Amazonian forests and to discuss some of the implications of those changes for the future Amazonian landscape. 2. NUTRIENT LOSSES DURING AND FOLLOWING DEFORESTATION Forest clearing causes an initial loss of nutrients from Amazonian tenestrial ecosystems through timber harvesting, fire, erosion, soil gaseous emissions, and hydrologic leaching of nutrients [Davidson et al., 2004a; McGrath et al., 2001]. By far, the largest of these losses of nutrient capital occurs during the initial phase of biourass removal through a combination of logging and/or fire. Fire is used both for site preparation and for subsequent weed control, resulting in significant loss of Nand P and sometimes K through emissions of aerosols and wind-blown ash [Holscher et al., 1997a; Kauffinan et al., 1995, 1998]. Significant N loss also occurs through volatilization as nitrogen oxides. Mass balance studies have shown that losses ofN from the telTestrial ecosystem caused by fire are 51-62% and 36-86% of the aboveground biomass N in Amazonian forests and pastures, respectively, and 7-32% and 1-36% of the aboveground biomass P in Amazonian forests and pastures, respectively [KaujJman et al., 1995, 1998]. Losses ofN exceed losses of P during fire not only because N can be volatilized but also because N is a relatively abundant nutrient in the Amazonian forest biomass that is being burned [Martinelli et al., 1999]. In contrast to Nand P, the relatively large quantities of Ca, Mg, and K present in woody biomass are largely conserved as ash following fire, resulting in significant redistribution of these important plant nutrients from aboveground biomass to the soil [Johnson et al., 2001; McGrath et al., 2001]. Markewitz et al. [2004] demonstrated a nearly stoichiometric recovelY of aboveground forest stocks of Ca and Mg in the exchangeable cation pool ofpasture and secondary forest soils. The more mobile K was partially lost from the terrestrial ecosystem. Even in Oxisols, which have inherently low cation exchange capacity because of nearly nonexistent permanently charged minerals, a significant increase in cation exchange capacity and exchangeable cation concentrations can occur following fire, because the organic matter and
some of the edges of kaolinite have modest pH-dependent cation exchange capasity [Uehara and Gilman, 1981]. The ash inputs following,clearing and burning both increase this cation exchange clpacity and provide the Ca, Mg, and K that become ele9t1'ostatically bound to the exchange sites. This effect of deforestation on soil cations appears to persist for several decades [Holmes et al., 2005; Johnson et al., 2001; Markewitz et al., 2001; Moraes et al., 1996]. Hence, the growth of pasture or secondary forest vegetation when fire is involved in land clearing is probably not limited by availability of these cations for several decades following deforestation. Although deforestation causes significant net loss of N and P, it also provokes a transient pulse of nutrients in readily available fmills for plant uptake, because the increase in soil pH from ash inputs renders the soil nutrients that remain more available to plants [Uehara and Gilman, 1981]. The availability of P is particularly pH-dependent. Indeed, this release of a fraction of the nutrients bound in aboveground biomass to readily available forms in the soil is the basis for the feltilization effect of slash-and-burn agriculture, and it also helps with pasture establishment. Within a fyw years, however, this pulse of available Nand P and, possibly, K becomes spent. In a pasture chronosequence study in Rondonia, Garcia-Montiel et al. [2000] found an initial increase in available-P pools during the first 3-5 years following pasture establishment and a gradual decline thereafter. Townsend et al. [2002] also observed declines in plant-available P forms with increasing pasture age in pasture chronosequences on Oxisols and Entisols near Santarem, Para. The process of gradual reacidification of the soil, as exchangeable cations from ash inputs are lost or as they accumulate in biomass, requires several decades [Holmes et al., 2005; Markewitz et al., 2001; Moraes et al., 1996], although P availability appears to drop more rapidly. Some of the P also appears to become bound into slowly cycling soil organic matter fractions with increasing pasture age [Garcia-Montiel et al., 2000; Townsend et al., 2002]. Additional losses of ecosystem N and P occur with repeated fire used for weed control or with harvest of cattle [Dias-Filho et al., 2001]. Although the largest nutrient losses occur with initial and repeated fire, additional modest losses of nutrients following disturbance can occur through inputs to groundwater and stream runoff and through gaseous emissions from soils. In Amazonia, increased hydrologic export of Nand P was measured after clearing of forest cover in a small watershed near Manaus [Williams and Melack, 1997]. However, the period of hydrologic and gaseous nutrient loss is often quite short in Amazonia. In all but the most severely degraded sites, growth of crops, pasture grasses, and secondary vegetation is usually rapid and vigorous after deforestation [Feld-
301
pausch et al., 2004; Gehring et at., 2005; Johnson et al., 2001; Salim on and Brown, 2000; Steininger, 2000; Vieira et al., 1996], thus limiting the period oflow nutrient uptake by plants and significant hydrologic nutrient loss after forest clearing. Soil solution concentrations of nitrate and cations tend to decline within 1 or 2 years after cutting and burning [Uhl and Jordan, 1984; Williams et al., 1997; Schroth et al., 1999]. Concentrations of nitrate in soil solution were lower in pastures compared to forests in both eastern Para [Marlcewitz et at., 2004], and Rondonia [Neill et at., 2001]. The initial pulse of nutrient availability after clearing and burning also supports temporarily increased soil emissions of nitrous oxide (N20), but net N mineralization, net nitrification, nitrate leaching, and soil efflux of nitric oxide (NO) and N20 consistently decline as tropical cattle pastures age [Bustamante et al., this volume]. Melillo et al. [2001] measured increased soil ammonium and nitrate concentrations for 6 months after forest clearing and higher fluxes of N20 from young pastures for 2 years after clearing on an Ultisol in Rondonia. After 3 years, N20 emissions from pastures declined to below rates measured in the forest. Verchot et al. [1999] detected no elevated N20 emissions from Oxisols in eastern Para, even fi'om pastures as young as 6 months. In addition to lower gaseous emissions and lower nitrate concentrations in s.oil solutions, another common indicator of N impoverishment in pasture soils is the ratio of extractable soil nitrate to ammonium [Davidson et al., 2000]. Whereas extractablb nitrate:ammonium ratios >1 in forests indicate excess N availability that suppmts nitrification, ratios <1 have been commonly found in pastures [Neill et al., 1995; Verchot et al., 1999], indicating less nitrification and a more conservative N cycle in pastures. Assays of net N urineralization and net nitrification are also lower in pasture soils compared to mature forest soils in both regions [Neill et al., 1997; Verchot et al., 1999]. Pasture feitilization with P is common on Oxisols in eastern Amazonia but is rare in pasture management of most other palts of the region [Dias-Filho et al., 2001]. Most of the P added to the Oxisols at Fazenda Vitoria, near Pm'agominas, Para, is retained in the soil [Davidson et al., 2004b], but enough must be taken up by pasture plants to make it economically rewarding for the ranchers. Smallholder fmillers commonly use N-P-K fertilizer when they can afford it. Soybean farmers routinely apply lime and P to Amazonian soils. Hence, nutrient management beyond ash inputs from slash-and-burn agriculture is common in Amazonian agriculture. However, the high demand for nutrients by crops and pasture grasses and the capacity of highly weathered soils to immobilize added nutrients means that agricultural abandomnent due, at least in part, to nutrient impoverishment is also common. Hence, secondaly forest species that
304
DAVIDSON AND MARTINELLI
NUTRIENT LIMITATIONS TO SECONDARY FOREST REGROWTH
sampled and homogenized before isotopic and chemical analyses. Hence, changes in species composition with forest age could have affected the N concentrations and isotopic compositiobs in these chronosequence studies, but the continuous trends with forest age shown in Figures la and Ib demonstrate that no single species that may have dominated during one stage of succession exerted an anomalous effect. The N:P ratios of fine litterfall also increased with forest age (Figure lc). The mean litterfall N:P ratios in the mature forests bound the globallnean of 62 for tropical forests, whereas the mean litterfall N:P values in the youngest forests are closer to the global mean of 29 for broadleaftemperate forests [McGroddy et al., 2004]. The litterfall mass:N ratio also declined with forest age (Figure Id). The values in the young forests are similar to N-limited temperate forests, while the advanced and mature forests values are commonly found in mature lowland tropical forests [Vitousek, 1984]. Although both foliar Nand P may be reabsorbed prior to leaf fall, only in the advanced successional and mature forests did the trees reabsorb more P relative to N before leaf abscission. The balance shifts from N economy to P economy in litterfall as forests age during secondary forest succession. Extractable soil nitrate also increased with forest age (Figure Ie), indicating increasing availability of soil N as the forests mature. Ammonium (which tended to decrease with forest age but not significantly so) is often the dominant form of inorganic soil N in N-limited systems, whereas nitrate accumulates where available N is more abundant [Davidson et al., 2000]. In a related sixth index, the ratio of extractable nitrate to extractable ammonium increased from values ::;1 for forests ::;20 years to values> 1 for forests ;:::40 years (Figure If). Ratios of <1 in these young secondaly forests are similar to the ammonium:nitrate ratios reported in old cattle pastures [Neill et al., 1995; Verchot et al., 1999]. Finally, N20 efflux from the soil increased with forest age (Figure Ig). These results help explain previously observed relative rates of soil N20 efflux among tropical land covers: mature forest greater than secondaty forest, which is greater than old pasture [Davidson et al., 2000; Keller and Reiners, 1994; Verchot et al., 1999]. The availability of N that supports N20 production via nitrification and denitrification is highest in the mature forest, falls to lowest levels in old pastures, and gradually recuperates to intermediate levels under secondaty forest vegetation. These analyses of green foliage, litterfall, soil inorganic N, and soil N 20 emissions demonstrate that the young successional forests exhibit a conservative N cycle, the mature forest exhibits a leaky N cycle, and advanced successional forests have intermediate N cycling properties. The log linear relationships of these indicators ofN and P cycling show that the largest nutrient cycling changes occur relatively early
during succession and that the rate of change declines as the secondary forests mature. The indicators of conservative N cycling in young secondaty forests are consistent with positive responses of tree growth to N addition to young forests discussed above [Davidson et al., 2004b] and to slower rates of secondary forest growth in sites that had been repeatedly burned during an agricultural phase [Zarin et al., 2005]. Taken together, these results indicate that young secondary forests often start out N limited at the time of agricultural abandonment, that recuperation of the N cycle gradually occurs during decades of secondary forest succession, and that mature lowland forests eventually recover abundant N relative to a conservative P cycle. Secondary forest chronosequence studies also demonstrate considerable variability in rates of biomass accumulation [Feldpausch et al., 2007; Zarin et al., 2001] and vegetation structure and species composition [Chazdon, 2008]. Some of this variation may be attributable to the duration and intensity of previous land use [Zarin et al., 2005] and to stochastic processes, such as proximity to seed sources [Chazdon, 2008]. Similarly, rates of recuperation of N cycling processes during secondary succession may also be influenced by intensity and duration ofprevious land uses, clearing size, and proximity to potential colonizing species. The patterns of Nand P cycling shown here for Amazonian secondaty forest succession are similar to the Walker and Syers [1976] paradigm regarding soil weathering over millennia1 timescales, except that the shift between Nand P limitation occurs in only decades to centuries in the case of secondaty succession (Figure 2). Actively cycling N in terrestrial ecosystems can be lost either by land use change, such as forest clearing, burning, and agricultural practices, or by natural processes such as fires, landslides, glaciers, and volcanic activity. Mature forests of the western Amazon, on the slopes of the Andes, may exhibit symptoms ofN limitation due to losses of N from landslides and erosion, which expose N-poor bedrock and primary minerals and which rejuvenate the soil profile. In the lowland eastern Amazon, accumulation of total ecosystem N alleviates N limitation as soils age over timescales of 104 to 107 years. Similarly, actively cycling N accumulates in the forest ecosystem over 101 to 10 3 years during secondaty succession, resulting in a similar, but more rapid, successional trajectOly from a conservative N cycle following agricultural abandonment to the leaky N and conservative P cycles expected in mature lowland tropical forests on old soils. The main reason that available N accumulates so much more rapidly during secondaty forest succession compared to primary succession on newly exposed rock is probably that the soil still contains large stocks of organic N at the end of the agricultural phase, albeit organic N that is present in
l<novlIltrends of C:N:P stoichiometry in mature forest ecosystems
r - - - - - - - - - - - - - - - -I ,-
_
: Okl, highly-weathered soils I (e.g., lowland tropical Minerai weathering I mature forests)
5011 age
Young soils (e.g., temperate and montane mature forests)
Conservative N
Leaky N cycle Conservative P cycle Forest
age
Secondary forest succession
Young forests on highlyweathered tropical soils
Conservative N cycle Conservative P cycle A new dimension of tropical land-use change (recuperation of the N cycle during secondary succession) addressed here
Figure 2. Simple schematic integrating the patterns of Nand P cycling for secondary succession, shown here in the vertically oriented dashed box on the right, and those previously demonstrated for primmy succession, shown here in the horizontally oriented dotted box at the top. Actively cycling N in tel1'estrial ecosystems can be lost either by land use change, such as forest clearing; by burning and agricultural practices; or by natural processes such as fires, landslides, glaciers, and volcanic activity. Just as accumulation of total ecosystem N alleviates a N limitation as soils age over thousands and millions of years, actively cycling N accumulates over decades and centuries during secondmy forest succession, resulting in a similar successional trajectory ft'om a conservative N cycle following agricultural abandonment to the leaky N and conservative P cycles expected in mature lowland tropical forests on old soils. From online supplementary material linked to Davidson et al. [2007].
fonus that are relatively recalcitrant to decomposition. The rate of recuperation ofN cycling processes during secondary forest succession may reflect, in pati, the kinetics of mobilization of recalcitrant forms of soil N to an actively cycling N pool, as well as the legacy of the degree of degradation during agricultural phases. Another possibility is the importance of biological N fixation during secondary succession, although this process is still poorly understood and poorly quantified, as reviewed in section 4. Nutrient amendment experiments do not perfectly replicate variation in nutrient availability, and chronosequence studies require the assumption that all factors other than age are constant among study sites. Despite these limitations of each methodological approach, their strengths and weaknesses are different. Therefore, evidence regarding the important role of N during early forest successional stages derived from these two independent research approaches provides important corroboration for each.
305
4. BIOLOGICAL N FIXATION IN SECONDARY FORESTS OF AMAZONIA The legume family is well represented in the canopy of most terra firme tropical forests in Amazonia [Moreira et al., 1992; Ometto et al., 2006; tel' Steege et al., 2006]. However, data on the natural abundance of foliar 15N indicate that most Amazonian legumes do not fix significant amounts of N fi'om the atmosphere, despite generally having higher foliar N concentration than nonlegume trees [Gehring et al., 2005; Ometto et al., 2006; Vitousek et al., 2002; Yoneyama et al., 1993]. A notable exception is the work conducted in a rain forest on an Oxisol in French Guiana, where Roggy et al. [1999a, 1999b] reported a substantial presence ofNfixing species and also estimated that N derived from the atmosphere contributed an average of 54% ofthe N nutrition of those N-fixing species. On the basis of the above results, it has been suggested that symbiotic N fixation occurs in tropical terra finne forests only during temporary N shortages [McKey, 1994; Vitousek et al., 2002]. Because secondaty forests are often N-deficient, as already noted in this chapter, it would be reasonable to expect that symbiotic N fixation would confer a competitive advantage to secondaty forest species [Gorham et al., 1979; Vitousek an~ Field, 1999]. Results from the only two studies addressing N fixation in Amazonian secondary forests provide equi",ocal results. Gehring et al. [2005] found isotopic evidence for significant symbiotic N fixation during the first 25 years of secondaly vegetation regrowth situated in central Amazon near the city of Manaus. In contrast, a recent study conducted in secondaty forests in the eastern Amazon in the state of Para did not find a significant difference in foliar ()15N between legume and nonlegume species [Davidson et al., 2007]. It is possible that strategies to compete for other limiting' resources, such as rapid height growth to compete for light [Vitousek and Field, 1999] and rapid root growth to obtain P [Vitousek and Field, 1999] or to obtain water in the dly season [Nepstad et al., 2001], are more important characteristics of early secondary forest successional species than is investment in energy costly N fixation. A newly proposed framework for explaining latitudinal trends of N fixation in mature forests suggests that N fixation may confer an advantage in P-limited soils of lowland tropical forests by meeting the high demand for N required by extracellular phosphatase enzyme production [Houlton et al., 2008]. This fi'amework has yet to be applied to tropical secondary forests, where both Nand P are conservatively cycled [Davidson et al., 2007]. Obviously, we do not have enough studies to evaluate the importance ofsymbiotic N fixation by plants in the secondaty forests of the Amazon. Part of the difficulty in quantifying
306 NUTRIENT LIMITATIONS TO SECONDARY FOREST REGROWTH or even confirming the existence of symbiotic N fixation arises from overlapping 015N values of legumes and nonlegume plants, especially in situations where N is in short supply [Bultamante et al., 2004]. Where N is a relatively abundant nutrient, as in most mature lowland Amazonian forests, ecosystem N stocks tend to become enriched in 15N (Figure la), and so the isotopic difference between soil N and atmospheric N is usually large enough to distinguish sources [Martinelli et al., 1999]. In contrast, fractionation by nitrification and denitrification is less important in Nimpoverished systems, such as young secondaIy forests, and so the natural abundance of actively cycling ecosystem N may be near zero per mil, which is also the isotopic signature of atmospherically derived N inputs, including N fixation. Hence, detection and quantification of N fixation rates may be inherently more difficult in secondaIy forests. Stochastic events that influence the composition of colonizing species [Chazdon, 2008] may also influence the potential for N fixation. This is clearly an area that deserves further investigation together with studies devoted to addressing other symbiotic N-fixing systems, such as endophytic bacteria and heterotrophic N fixation, which could be also contribute to N inputs to secondaIY forests [Vitousek et al., 2002]. 5. IMPLICATION FOR THE FUTURE OF AMAZONIAN SECONDARY FORESTS The mean age of Amazonian secondaIy forests has been estimated to vary only between 4.4 and 4.8 years during the last 3 decades [NeefJ et al., 2006]. This result implies that secondary forests continue to be recleared frequently for new phases of agriculture. Only in areas where the initial deforestation occurred many decades ago, such as eastern Para and Maranhao, do more advanced successional forests occur to any appreciable extent, but even there, they make up only about 5-20% of the area [Neeff et al., 2006; Vieira et al., 2003]. The authors are aware of anecdotal accounts of secondary forests being set aside as community forest reserves for hunting and other nontimber products and for emergency use as future agricultural lands in hard times, but this type of management is rare. The Amazonian landscape does not yet include anything similar to the advanced successional secondaIy forests of Europe and North America, nor is it likely to happen in the near future. Much of the secondaIy forests of Amazonia are the fallow fields that are allowed to grow for a few years between cropping cycles in slash-and-burn agriculture. In ranching areas, so-called "dirty pastures" that contain early successional forest growth are also common. Repeated fire can lead to degradation of nutrient stocks, especially if pressure for agricultural land results in decreases in the length of the fal-
low period. Nutrient cycling shldies have demonstrated that alternatives to slash-and-burn agriculture, such as chop-andmulch cropping systems, can retain nutrients on site [Denich et al., 2005; Sommer et al., 2004] and reduce net greenhouse gas emissions [Davidson et al., 2008], leading to a potentially more sustainable cropping system that continues to depend on secondaIy forests in a fallow phase; Not surprisingly, this research has been conducted in the Zona Bragantina in eastern Para, where several generations of slash-and-burn smallholder farmers have applied nine or more cropping cycles to the land [Vieira et al., 2003]. This is among the first areas where one would expect concern about nutrient impoveriShment and a search for alternatives to slash-and-burn cropping systems to emerge. Expansion of soybean and other industrial-scale farming systems in Amazonia is sometimes justified as making productive use of already degraded lands. The term "degraded" is not well defined in this context, but it presumably means abandoned agricultural land that is now in some stage of secondaIy forest succession. The secondaIy forests of many of these "degraded lands" would likely evenhlally become tall-statured advanced successional forests if left alone. The key considerations for selection of sites for industrial-scale cropping systems have little to do with degradation in terms of nutrients or species composition but rather are based on soil physical properties that allow good drainage and on topographic features of flat or only gently rolling terrain. The nutrient legacy of previous agricultural practices are usually of little importance to soybean farmers because they have the necessary capital to use fertilizers and lime as needed. Where agricultural land is left abandoned, nutrient impoverishment could delay rates of regrowth and canopy closure, thus leaving the young forest susceptible for a longer period to drought and to fire escaping from nearby agricultural land [Cochrane et al., 1999; Nepstad et al., 1999]. Increasing occurrences of drought with anticipated regional climate change [Li et al., 2006; Malhi et al., 2008] may also increase the probability of fire and slow rates of secondary forest succession.Repeated agricultural phases also result in legacies of altered species composition in affected secondaIy forests, with more lianas persisting longer and fewer palms in secondary forests regrowing after multiple cycles of clearing [Gehring et al., 2005]. However, the current high rate ofreclearing of secondary forests is probably the most important contemporaIy reason for lack of large-stahlred, advanced secondaIy forest succession in the region. Only when agricultural expansion slows are advanced secondaIy forests and their legacies of prior land uses likely to become important components of the landscape. In the interim, secondary forests are likely to remain part of a dynamically changing landscape, with young second-
DAVIDSON AND MARTINELLI aIY forests making up a significant fraction of agricultural regions, but with fevy; secondaIy forests persisting in one place for much morS;ifhan 5 years. The nutrients within those ephemeral forests.l,will continue to play an important role in the success or fa.Vlure of smallholder agriculhlre, both in traditional slash-a~d-burn agriculture and in alternatives that may be developed to manage nutrients more effectively. Hence, our improving understanding of losses and inputs of the major macronutrients and their ratios in secondaIy forests may help the development of viable management schemes that include secondary forests. Acknowledgment. This work was supported by grant NNG06GE88A of NASA's Terrestrial Ecology Program as part of the LBAproject.
REFERENCES Amundson, R., A T. Austin, E. A. G. Schuur, K. Yoo, V. Matzek, C. Kendall, A Uebersax, D. Brenner, and W. T. Baisden (2003), Global patterns of the isotopic composition of soil and plant nitrogen, Global Biogeochem. Cycles, 17(1), 1031, doi:lO.10291 2002GBOO 1903. . Buschbacher, R., C. Uhl, and E. A S. SelTao (1988), Abandoned pastures in eastern Amazonia. II. Nutrient stocks in the soil and vegetation, J Ecol., 76,682-699. Bustamante, M. M. C., L. A. MaIiinelli, D. A Silva, P. B. Camargo, C. A Klink, T. F. Domingues, and R. V. Santos (2004), 15N natural abundance in woody plants and soils of central Brazilian savannas (celTado), Ecol. Appl., 14, suppl., 200-213. Bustamante, M. M. C., M. Keller, and D. A da Silva (2009), Sources and sinks of trace gases in Amazonia and the cerrado, Geophys. Monogr. Ser., doi: 10.1029/2008GM000733, this volume. Chadwick, O. A, L. A Derry, P. M. Vitousek, B. 1. Huebert, and L. O. Hedin (1999), Changing sources of nutrients during four million years of ecosystem development, Nature, 397, 491--497. Chazdon, R. L. (2008), Chance and detelminism in tropical forest succession, in Tropical Forest Community Ecology, edited by W. Carson and S. Sclmitzer, pp. 384--408, Blackwell, Hoboken, N. 1. Cochrane, M. A, A Alencar, M. D. Schulze, C. M. Souza, D. C. Nepstad, P. Lefebvre, and E. A Davidson (1999), Positive feedbacks in the fire dynamic of closed canopy tropical forests, Science,284,1832-l835. Crews, T. E., K. Kitayama, J. H. Fownes, R. H. Riley, D. A. Herbert, D. Mueller-Dombois, and P. M. Vitousek (1995), Changes in soil phosphOlus fractions and ecosystem dynamics across a long chronosequence in Hawaii, Ecology, 76, 1407-1424. da Silva, 1. M. C., C. Uhl, and G. MUlTay (1996), Plant succession,' landscape management, and the ecology of frugivorous birds in abandoned Amazonian pastures, Conserv. Bioi., 10,491-503. Davidson, E. A, M. Keller, H. W. Erickson, L. V. Verchot, and E. Veldkamp (2000), Testing a conceptual model of soil emissions of nitrous and nitric oxides, BioScience, 50, 667-680.
307
Davidson, E. A., C. 1. R. de Carvalho, 1. C. G. Vieira, R. de O. Figueiredo, P. Moutinho, F. Y. Ishida, M. T. P. dos Santos, 1. B. GuelTero, K. Kalif, and R. T. Saba (2004a), Nutrient limitation of biomass growth in a tropical secondary forest, Ecol. Appl., 14, suppl., 150-163. Davidson, E. A., C. Neill, A V. Krusch, V. V. R. Ballester, D. Markewitz, and R. de O. Figueiredo (2004b), Loss of nutrients from terrestrial ecosystems to streams and the atmosphere following land use change in Amazonia, in Ecosystems and Land Use Change, Geophys. Monogr. Ser., vol. 153, edited by R. S. DeFries, G. P. Asner, and R. A Houghton, pp. 147- 158, AGU, Washington, D. C. Davidson, E. A, et al. (2007), Recuperation of nitrogen cycling in Amazonian forests following agricultural abandonment, Nature, 447, 995-998. Davidson, E. A, T. D. de A Sa, C. J. R. Carvalho, R. de O. Figueiredo, M. do S. A Kato, O. R. Kato, and F. Y. Ishida (2008), An integrated greenhouse gas assessment of an alternative to slashand-bill'n agriculture in eastern Amazonia, Global Change Bioi., 14, 998-1007. Denich, M., P. L. G. Vlek, T. D. de A Sa, K. Vielhauer, and W. Lucke (2005), A concept for the development of fire-fi'ee fallow management in the eastern Amazon, Brazil, Agric. Ecosyst. Environ., 110,43-58. Dias-Filho, M., E. A Davidson, and C. 1. R. de Carvalho (2001), Linking biogeochemical cycles to cattle pasture management and sustainability in the Amazon Basin, in The Biogeochemistry of the Amazon, edited by M. E. McClain, R. L. Victoria, and 1. E. Richey, pp. 84-105, Oxford Univ. Press, New York. Feldpausch, T. R., 1'1. A Rondon, E. C. M. Fernandes, S. 1. Riha, and E. Wandelli (2004), Carbon and nutrient accumulation in secondaIy forests regenerating on pastures in central Amazonia, Ecol. Appl., 14, suppl., 164-176. Feldpausch, T. R, C. de C. Prates-Clark, E. C. M. Femandes, and S. 1. Riha (2007), Secondary forest growth deviation fi'om chronosequence predictions in central Amazonia, Global Change Bioi., 13,967-979. Garcia-Montiel, D. C., C. Neill, 1. M. Melillo, S. Thomas, P. A Steudler, and C. C. CelTi (2000), Soil phosphOlus transformations following forest clearing for pastill'e in the Brazilian Amazon, Soil Sci. Soc. Am. J, 64, 1792-1804. Gehring, C., M. Denich, M. Kanashiro, and P. L. G. Vlek (1999), Response of secondary vegetation in eastem Amazonia to relaxed nutrient availability constraints, Biogeochemistry, 45, 223-241. Gehring, C., P. L. G. Vlek, L. A. G. de Souza, and M. Denich (2005), Biological nitrogen fixation in secondary regrowth and mature rainforest of central Amazonia, Agric. Ecosyst. Environ., 111,237-252. Gorham, E., P. M. Vitousek, and W. A. Reiners (1979), The regulation of chemical budgets over the course of terrestrial ecosystem succession, Annu. Rev. Ecol. Syst., 10,53-84. Hedin, L. 0., P. M. Vitousek, and P. A. Matson (2003), Nutrient losses over foill' million years of tropical forest development, Ecology, 84, 2231-2255. Hirsch, A I., W. S. Little, R. A Houghton, N. A. Scott, and 1. D. White (2004), The net carbon flux due to deforestation and forest
308
DAVIDSON AND MARTINELLI
NUTRIENT LIMITATIONS TO SECONDARY FOREST REGROWTH
re-growth in the Brazilian Amazon: Analysis using a processbased model, Global. Change Bioi., 10, 908-924. Holmes, K. W., P. C. Kyriakidis, O. A Chadwick, J. O. V. Soares, and D. 1\ Roberts (2005), Multi-scale variability in tropical soil nutrients' following land-cover change, Biogeochemistly, 74, 173-203. Holscher, D., B. Ludwig, R. F. Moller, and H. Foister (1997a), Dynamic of soil chemical parameters in shifting agriculture in the eastem Amazon, Agric. Ecosyst. Environ., 66,153-163. Holscher, D., T. D. da A. Sa, T. X. Bastos, M. Denich, and H. FOIster (1997b), Evaporation from young secondaly vegetation in eastern Amazonia, J. Hydrol., 193,293-305. Houlton, B. Z., Y.-P. Wang, P. M. Vitousek, and C. B. Field (2008), A unifYing framework for dinitrogen fixation in the terrestrial biosphere, Nature, 454, 327-330. Johnson, C. M., I. C. G. Vieira, D. J. Zarin, 1 Frizano, and A H. Johnson (2001), Carbon and nutrient storage in primaly and secondmy forests in eastern Amazonia, For. Eea!. Manage., 147, 245-252. Kauffman, 1 B., D. L. Cummings, D. E. Ward, and R. Babbitt (1995), Fire in the Brazilian Amazon: Biomass, nutrient pools, and losses in slashed primary forests, Oecologia, 104, 397-409. Kauffman, J. B., D. L. Cummings, and D. E. Ward (1998), Fire in the Brazilian Amazon 2. Biomass, nutrient pools, and losses in cattle pastures, Oecologia, 113,415-427. Keller, M., and W. A Reiners (1994), Soil-atmosphere exchange of nitrous oxide, nih'ic oxide, and methane under secondmy succession of pasture to forest in the Atlantic lowlands of Costa Rica, Global Biogeochem. Cycles, 8, 399-409. Li, W., R. Fu, and R. E. Dickinson (2006), Rainfall and its seasonality over the Amazon in the 21 st centmy as assessed by the coupled models for the IPCC AR4, J. Geophys. Res., 111, D02111, doi:1 0.1 02912005JD006355. Malhi, Y., 1 T. Roberts, R. A Betts, T. J. Killeen, W. Li, and C. A Nobre (2008), Climate change, deforestation and the fate of the Amazon, Science, 319, 169-172. Markewitz, D., E. A Davidson, R. Figueiredo, R. L. Victoria, and A V. Krnsche (200 I), Control of cation concentrations in sh'eam waters by surface soil processes in an Amazonian watershed, Nature, 410, 802-805. Markewitz, D., E. A Davidson, P. Moutinho, and D. C. Nepstad (2004), Nutrient loss and redistribution after forest clearing on a highly weathered soil in Amazonia, Eco!. Appl., 14, suppl., 177-199. Martinelli, L. A., M. C. Piccolo, A R. Townsend, P. M. Vitousek, E. Cuevas, W. McDowell, G. P. Robertson, O. C. Santos, and K. Treseder (1999), Nitrogen stable isotopic composition of leaves and soil: Tropical versus temperate forests, Biogeochemist/y, 46, 45-65. McGrath, D. A, C. K. Smith, H. L. Gholz, and F. de A. Oliveira (2001), Effects of land-use change on soil nutrient dynamics in Amazonia, Ecosystems, 4, 625-645. McGroddy, M. E., T. Daufresne, and L. Hedin (2004), Scaling of C:N:P stoichiomehy in forests worldwide: Implications of terrestrial Redfield-type ratios, Ecology, 85, 2390-2401.
McKey, D. (1994), Legumes and nih'ogen: The evolutionmy ecology of a nitrogen-demanding lifestyle, in The Nitrogen Factor, edited by 1 L. Sprent and D. McKey,Adv. Legume Syst., 5, 211-228. Melillo, 1 M., P. A Steudler, B. 1 Feigl, C. Neill, D. Garcia, M. C. Piccolo, C. C. Cerri, and H. Tian (2001), Nitrous oxide emissions from forests and pastures of various ages in the Brazilian Amazon,J. Geophys. Res., 106, 34,179-34,188. Moraes, 1 L., B. Volkoff, C. C. Cerri, and M. Bernoux (1996), Soil properties under Amazon forest and changes due to pasture installation in Rondonia, Brazil, Geoderma, 70,63-81. Moran, E. F., E. S. Brondizio, J. M. Tucker, M. C. da Silva-Forsberg, S. McCracken, and I. Falesi (2000), Effects of soil fertility and land-use on forest succession in Amazonia, For. Ecol. Manage., 139,93-108. Neeff, T., R. M. Lucas, 1 R. dos Santos, E. S. Brondizio, and C. C. Freitas (2006), Area and age of secondary forests in Brazilian Amazonia 1978-2002: An empirical estimate, Ecosystems, 9, 609-623. Neill, C., M. C. Piccolo, P. A. Steudler, 1 M. Melillo, B. 1 Feigl, and C. C. CelTi (1995), Nitrogen dynamics in soils of forests and active pastures in the western Brazilian Amazon Basin, Soil Bioi. Biochem., 27, 1167-1175. Neill, C., M. C. Piccolo, C. C. Cerri, P. A Steudler, J. M. Melillo, and M. Brito (1997), Net nitrogen mineralization and net nitrification rates in soils following deforestation for pasture across the southwestern Brazilian Amazon Basin landscape, Oecologia, 110,243-252. Neill, C., L. A. Deegan, S. M. Thomas, and C. C. Cerri (2001), Deforestation for pasture alters nitrogen and phosphorns in small Amazonian streams, Ecol. Appl., 11,1817-1828. Nepstad, D. C., et al. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505-508. Nepstad, D. C., P. R. S. Moutinho, and D. Markewitz (2001), The recovery of biomass, nutrient stocks, and deep-soil functions in secondmy forests, in Biogeochemist/)I of the Amazon Basin, edited by M. E. McClain, R. L. Victoria, and 1 E. Richey, pp. 139-155, Oxford Univ. Press, New York. Ometto,l P., et al. (2006), The stable carbon and nitrogen isotopic composition of vegetation in tropical forests of the Amazon Basin, Brazil, Biogeochemist/y, 79,251-274, doi:IO.1007/sI0533006-9008-8. Reich, P. B., and 1 Okeksyn (2004), Global patterns of plant leaf Nand P in relation to temperatm'e and latitude, Proc. Nat!. Acad. Sci. U. S. A., 101,11,001-11,006. Roggy, J. C., M. E. Prevost, 1 Garbaye, and A. M. Domenach (1999a), Nitrogen cycling in the h'opical rain forest of French Guiana: Comparison of two sites with contrasting soil types using 8 1SN, J. Trap. Eea!., 15, 1-22. Roggy, J. C., and M. F. Prevost (1999b), Nitrogen-fixing legumes and silvigenesis in a rain forest in French Guiana: A taxonomic and ecological approach, New Phytol., 144, 283-294. Salimon, C. I., and I. F. Brown (2000), Secondmy forests in western Amazonia: Significant sinks for carbon released from deforestation?, lnterciencia, 25, 198-202. Schroth, G., L. F. da Silva, R. Seixas, W. G. Teixeira, 1 L. V. Macedo, and W. Zech (1999), Subsoil accumulation of mineral
nitrogen under polyculture and monoculture plantations, fallow and primmy forest ina ferralitic Amazonian upland soil, Agric. Ecosyst. Environ., ~1, 109-120. Sombroek, W. (200!}); Amazon landforms and soils in relation to biological diver~Jk Acta Amazonica, 30, 81-100. Sommer, R., P. I;/t. Vlek, T. D. de A. Sa, K. Vielhauer, R de F. R. Coelho, and H. Foister (2004), Nutrient balance of shifting cultivation by burning or mulching in the eastern Amazon-Evidence for subsoil nutrient accumulation, NutI'. Cycling Agroecosyst., 68, 257-271. Steininger, M. K. (2000), Secondmy forest structure and biomass following short and extended land-use in central and southern Amazonia, J. Trap. Eeal., 16, 689-708. tel' Steege, H., et al. (2006), Continental-scale patterns of canopy tree composition and function across Amazonia, Nature, 443, 444-447. Townsend, A R., G. P. Asner, C. C. Cleveland, M. E. Lefer, and M. M. C. Bustamante (2002), Unexpected changes in soil phosphorns dynamics along pasture chronosequences in the humid tropics, J. Geophys. Res., 107(D20), 8067, doi:IO.1029/ 2001JD000650. Uehara, G., and G. Gillman (1981), The Mineralogy, Chemist/y, and Physics of Tropical Soils With Variable Charge Clays, Westview Press, Boulder, Colo. Uhl, C. (1987), Factors controlling succession following slash-andburn agriculture in Amazonia, J. Ecol., 75,377-407. Uhl, C., and C. F. Jordan (1984), Succession and nutrient dynamics following forest cutting and burning in Amazonia, Ecology, 65, 1476-1490. Verchot, L. V., E. A. Davidson, H. CaWlnio, I. L. Ackerman, H. E. Erickson, and M. Keller (1999), Land use change and biogeochemical controls of nitrogen oxide emissions from soils in eastern Amazonia, Global Biogeochem. Cycles, 13(1), 31-46. Vieira, I. C. G., R de P. Salomao, N. Rosa, D. C. Nepstad, and 1 Roma (1996), 0 renascimento da f10resta no rastro da agricultura, Cienc. Hoje, 20, 38-45.
309
Vieira, I. C. G., A. S. de Almeida, E. A. Davidson, T. A Stone, C. 1 R. de Carvalho, and 1 B. Guerrero (2003), ClassifYing successional forests using Landsat spectral properties and ecological characteristics in eastern Amazonia, Remote Sens. Environ., 87,470-481. Vitousek, P. M. (1984), Litterfall, nutrient cycling, and nutrient limitation in tropical forests, Ecology, 65, 285-298. Vitousek, P. M., and C. B. Field (1999), Ecosystem consh'aints to symbiotic nih'ogen fixers: A simple model and its implications, Biogeochemist/)I, 46, 179-202. Vitousek, P. M., and W. A Reiners (1975), Ecosystem succession and nutrient retention: A hypothesis, Bioscience, 25, 376-381. Vitousek, P. M., et al. (2002), Towards an ecological understanding of biological nitrogen fixation, Biogeochemist/y, 57, 1-45. Walker, T. W., and 1 K. Syers (1976), The fate ofphospholUs during pedogenesis, Geoderma, 15, 1-19. Williams, M. R, and 1 M. Melack (1997), Solute expOli from forested and partially deforested catchments in the cenh'al Amazon, Biogeochemistly, 38, 67-102. Williams, M. R., T. R. Fisher, and J. M. Melack (1997), Solute dynamics in soil water and groundwater in a central Amazon catchment undergoing deforestation, Biogeochemistry, 38, 303-335. Yoneyama, T., T. Moraoka, T. Murakami, and N. Boonkerd (1993), Natural abundance of lsN in tropical plants with emphasis on tree legumes, Plant Soil, 153,295-304. Zarin, D. 1, M. 1 Ducey, J. M. Tucker, and W. A Salas (2001), Potential biomass accumulation in Amazonian regrowth forests, Ecosystems, 4, 658-668. Zarin, D. A, et al. (2005), Legacy of fire slows carbon accumulation in AmaZ01}ian forest regrowth, Front. Ecol. Environ., 3, 365-369. . E. A Davidson, Woods Hole Research Center, 149 Woods Hole Road, Falmouth, MA 02540-1644, USA ([email protected]) L. A Martinelli, CENA, University of Sao Paulo, Avenida Centenario 303, Piracicaba, SP 13416-000, Brazil.
The Maintenance of Soil Fertility in Amazonian Managed Systems Fhivio J. Luizao and Philip M. Fearnside Department a/Ecology, INPA, Manaus, Brazil
Carlos E. P. Cerri ESALQ, University a/Sao Paulo, Piracicaba, Brazil
Johannes Lehmann Department a/Crop and Soil Sciences, Camel! University, Ithaca, New York, USA
Most of Brazilian Amazonia faces important limitations for conventional agriculture and pastures due to a generally poor chemical fertility as well as the region's environmental conditions, especially high temperature and moisture. Without proper management, degradation of the soil and resulting unsustainability of agricultural and ranching production occur within a few years, leading to land abandonment. Use ofperennial crops, especially those based ~n native tree species, would be instrumental in order to achieve best management such as that which assure recycling processes similar to those in the primary forest. Recommended alternative land uses are those producing high soil organic matter, recycling of nutrients, substantial agricultural production, and economic viability. These include agroforestry systems, enrichment of second growth with valuable native timber or fruit species, accelerated fallow regrowth via enrichment plantings, sequential agroforestly with slash-and-mulch, and diversified forest plantations. Improvement of agricultural soils can be based on lessons learned from the study of processes involved in the formation and maintenance of the rich "dark earths" (terra preta), which owe their high carbon content and fertility in part to high content of charcoal. Addingpowdered charcoal combined with selected nutrients can increase soil carbon in modern agriculture. Considering that limitations to expansion of intensified land uses in Amazonia are serious, regional development should emphasize the natural forest, which can maintain itself without external inputs of nutrients. Instead of creating conditions to further expand deforestation, these forests may be used as they stand to provide a variety of valuable environmental services that could offer a sustainable basis for development of Amazonia.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 1O.1029/2008GM000742 311
LUIZA.O ET AL.
312 MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS 1. INTRODUCTION 1.1. Rr:1e ofSoil-Fertility Maintenance in Amazonian Developmellt The Amazon tropical forest is one of the w?rld's last remaining forests that is sufficiently large a~ld mtact to provide globally important environmental serv~ces [ASB, 200?], occupying 7.86 million km2 in nine countnes, and covenng approximately 45% of the Sotlth America~ continent. ~or: than 60% of the Amazon forest is located m northern Brazil [IBGE/SIDRA, 1997], an area larger than the whole of western Europe [INPE, 2000]. In recent decades, Brazil's Amazon forest has been rapidly destroyed and replaced by cattle pasture. Much smaller areas are maintained under agricultural uses such as S?y- Figure 1. Brazil's Legal Amazon r~gi?n [IBGE/SID~, 19~7]: beans or other armual crops, and a very small proportIOn Savannas (mostly cerrado) are shown m light gray, deforested ~I~as becomes perennial crops, as in agroforestry sys.tems (AFS!. in the Amazonian biome are depicted in dark gray, and remammg Amazonian forest is shown in white. The forested and forn~er~y According to the last official survey ofland use m ~r~azoma forested areas, together with the small patches of savarmas wIthm [IBGE, 2006], 29% of Brazilian l~nds ~o: ?49 mr.llIOn ha) this area, constitute "biological" Amazonia. were occupied by agrosilvopastonl actIvrtIes, mamly pastures (70% of the total). While in the country, as a whole, pasture cover decreased by 3.8% rel~tive t? the ye~r 1985, external inputs such as inorganic fertilize~·s. ?f c?urse, this in the 5 million km2 Legal Amazoma regIOn, there was a implies the conservation of functional bIOdrversrty and of 44.2% expansion in pastureland between 1~85 and 200.6. large tracts of natural habitats. . Pastures represent 82.3% of the land occupred by ag.rosrlThe lack of proper land management, together wrth natuvopastoril activities in the Brazilian Legal Ama~oma, or ral constraints such as excessive moisture and high tempera61.6 million ha [IBGE, 2006]. In national terms, thrs mea~s ture, high acidity, and low supply of soil nutrien~s.in most of that 36% of the country's cattle ranching is concentrated ~n the region, have been causes of limited productIvrty of food Legal Amazonia, as well as 39% of the so~~ean and :7 Yo and fiber in Brazilian Amazonia. However, development. of of the cotton cultivation in the country. AddrtIOnally, 6 Yo of new technologies and the understanding of principles beh~nd the sugarcane ethanol produced by Brazil came fr~m ~eg~~ the practices of traditional populations living in the reg~on Amazonia [Smera1di and May, 2008]. Althoug~ cenado may be used to change this situati?n. Also, a better .valuat~on occupies 16% of the Legal Amazonia and has lugh ~rodu~ of the many environmental servrces of the standmg forest tivity of pastures and crops, large areas of Amazoman .fo~ may be used as a sound basis for sustainable development ests were also converted into pastures or croplands. T~us rs in Amazonia. Particularly true for cattle ranching: 73 % of the ca~le m the ' " mea mng areas. . the " Amazon b region are located m lOme, 1.2. Soil-Fertility Dynamics Under Amazonian Land Uses of Amazon forest [Smeraldi and May, 2008]. On the oth~r hand, only 2% of soybean production in Legal Ama.zoma Starting in the early 1960s, the Brazilian Government was supposedly produced in former forest. lands, whrle no tried to use the Amazon's abundant natural resources (forcotton production took place (all produced m celTado lands, ests, agricultural lands, and minerals). t? fuel regional and although some of these were located within the limits of the national economic growth. However, mrtral attempt~ to de"biological" Amazonia; Figure 1). . . velop the region through government-oriente~ establ.rshment These land use changes have important rmpacts on sorl of agricultural settlements ran into seriou~ ~rfficultI~s . Folfertility, often leading to degradation of the .soil and res~lt lowing the establishment offederally sUbs~drzed~redrt r~ the ing in unsustainable agricultural and .ran~~m~ .productr~n. late 1960s, hundreds of agricultural and mdustnal projects Here, our working definition of "sustamabrhty rs the m~r~ were approved and implemented in the Amazon, but m?st of tenance of the basic functions and mechanisms of the ?ngrthe agricultural projects failed and were abandoned. Lrl~ely, nal ecosystem, making the man-made system economrcally the main reason for failure was that the process of assrsted productive for long periods without the constant need for
migration and colonization was rapid and intense, and millions of hectares of forested land were handed over to newcomers with little k.r10wledge of the potential of these areas to support agricultpt:e, and little consideration was given to soil, water, or wajershed conditions when sites were chosen .f [Walker and He/fnma, 1996]. In the Amazon, two land use systems (neither of which sustainable if applied in a continuous way) are widely practiced by farmers [ASB, 2002]: (1) traditional pasture and (2) traditional (short) annual crop/fallow rotation. In traditional pasture, generally after growing annual crops for 2 or 3 years, farmers plant pasture grass, usually Brachiaria sp., and the pastures are burned to control weeds and insects, sometimes annually, with little or no other management. In the practice oftraditional (short) annual crop/fallow rotation, the annual food crops, usually grown for only 2 years, are followed by 3 or more years of natural bush fallow. Usually, this is again followed by a rotation of annual crops, after which the plot is usually dedicated to pasture. The use of slash-and-burn as the main land preparation method for agriculture in the Brazilian Amazon occurs during the dry season found in most of the region (with one to four dry months between June and October), which stands in contrast to the slash-and-mulch system found in more humid Amazonian areas but seldom practiced in Brazil [ASB, 2002]. In the slash-and-burn process, the cleared vegetation is allowed to dry and is then burned before the onset of the rains, when the area is planted to annuals, perennials, or pasture grasses, using the nutrients released from biomass burning. However, despite this initial fertilization through ashes, which are rich in cations, crop and pasture production are not durable or economically sustained in the region due to several natural and human-induced constraints. One reason is that soil quality varies widely and is patchy, but the predominant soil types, Oxisols and Ultisols, usually have relatively low natural fertility, with high levels of acidity, low phosphorus contents, low levels of cation exchange, and high levels of aluminum toxicity [Cochrane and Sanchez, 1982]. Although generalizations on soil fertility levels for a large region like Amazonia must be avoided, only 7% of the land area in Brazilian Amazonia is considered to be free from major limitations on plant growth, while soil phosphorus (P) deficiencies are believed to constrain productivity in 90% of the area, and aluminum (AI) toxicity occurs in 73% of the soils [Cochrane and Sanchez, 1982]. Such limitations appear to be particularly evident in the central and eastern Amazon, far from the Andes, and where most soils are very old and weathered. These soils sometimes have nutrient concentrations that are far too low for plant growth (e.g., Table 1). Such infertile soils are fragile and need special care if they are used for any kind of cultivation after forest removal.
313
Table 1. Volumetric Content of Exchangeable Bases in the Top 1m of an Alic, Dystrophic Oxisol, Near Manaus (Amazonas) and in a Eutrophic Oxisol From Rio Grande do Sul a K Manaus, Amazonas (Hapluslox) Rio Grande do Sui (Hapludoll)
Ca
Mg
103 3,330
242 120 42,300 9920 "Exchangeable base values are given in kg ha- 1 m- I (E. C. Fernandes, personal communication, 2006).
On old, heavily weathered soils, regular inputs of new nutrients from soil parent material are very small or even negligible [Schubart et a1., 1984; Brinlonann, 1989]. Thus, aboveground biomass and the litter layer are two important reservoirs of plant nutrients for Amazonian forest [Brinkmann, 1989; Anderson and Spencer, 1991]. Under such conditions, atmospheric deposition (both dry and wet) also becomes an important source of nutrients, especially for base cations [Brinlonann, 1989], compensating for the small losses fi'om the ecosystem caused by leaching. However, most ofthe nutrient demand of the plants is met through biologically mediated remineralization of nutrients from the organic matter. Generally, nutrients are rapidly and efficiently cycled in the lowlapd evergreen rain forest ecosystem, and most nutrients have a very short residence time in the soillitter system, being quickly remineralized and made available to the plants [lrIerrera et al., 1978; Luiziio et al., 2004]. The concentration pf mineral nutrients in the primary forest litter in the Amazon is generally high for nitrogen; however, for phosphorus and base cations, these considerations are lower than in other tropical lowland evergreen rain forests [Proctor, 1984]. The lower concentrations can be considered as an indication of nutrient limitation for the plants in this region [Vitousek, 1984], and these nutrients are strongly translocated and retained in the plants prior to foliar abscission. However, other patterns exist in the region as well: for instance, in Maraca, northern Amazonia, Scott et al. [1992] found nutI'ient concentrations in litter (especially for phosphorus and calcium) that are much higher than the average values for Amazonia despite high translocation for some nutrients. In general, rapid cycling of nutrients in tropical forests is achieved through high decomposition rates, which are made possible by high temperatures and high annual rainfall, which boosts biological activity in soil and litter. Despite the intrinsic low chemical fertility in the majority of Amazonian soils, and the relatively low content of soil carbon (C) in most of the region [Monies et a1., 1996; Cerri et a./., 2003], soils generally have good physical structure. This, together with the complex, specialized and very active soil biota, can maintain the natural fertility of soils if
314 MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
they are kept well covered and protected against direct sun and raindrop impact, as occurs under the natural forest cover [Ross et al., 1990]. Some oftthe limiting factors for crop production can be surpassed through technological advances, but others cannot be overcome at a reasonable cost in the region, where there are restrictions for both the intensification of agriculture and ranching uses and the scale to which these land uses can be expanded [Fearnside, 1997a]. These restrictions include agronomic limits on per-hectare yields, physical resource limits such as phosphate deposits available for soil fertilization, and environmental risks. Thus, other strategies should be adopted to benefit Amazonia's human population, and a number of projects have been searching for sound alternatives to slash-and-burn agriculture [Palm et aI" 2001, 2005; Almeida et al., 2006]. In recent years, special emphasis has been given to keeping the forest standing through the creation of compensation mechanisms for the environmental services of intact forest [Fearnside, 1997b, 2008a]; scientists and decision makers are approaching a consensus that such mechanisms must be pursued and put into action. Therefore, agricultural production would be directed to the already deforested areas in Amazonia, most of which are abandoned or degraded and need alternative practices (e.g" fallow enrichment for no-bum agriculture, agroforestly, etc.) to become productive again, 1.3. Conversion ofForest to Well-Managed Pasture: Effect on Carbon Accumulation in Soils
Cattle pastures represent the largest single use of cleared forest land in most of the Brazilian Amazon. Estimates show that 70% of the deforested land has been converted to pastures at one stage or another [Serrao and Toledo, 1990; Dias-Filho et aI" 2001]. About 45% of the deforested land in Brazilian Amazonia is occupied by actively grazed cattle pasture or 24.7 Mha [Fearnside and Barbosa, 1998]. Similar statistics were reported by Homma [1994] and Kitamura [1994]. Farmers were motivated to convert lands cleared from forest into pasture because of the real or perceived increases in land value. Farmers not only maintained cattle as standing "bank accounts" and obtained cash from sales of animals and milk, but also increased the value of these "savings" by investing time and resources in pasture, fencing, corrals, and ponds [Flljisaka et aI" 1996]. Despite the enOlWOUS scale of pasture expansion in the Amazon, there is still no clear understanding of the direction ofthe resulting changes in soil C stocks. Fearnside and Barbosa [1998] reported that conversion of Amazon forest to pasture can produce a net soil C sink (well-managed pasture) or a net C source (overgrazed pasture), depending on man-
agement. Neill and Davidson [1999] observed that conversion of forest to pasture in the Amazon occurs on a variety of soils and in regions that differ in the amount and timing of precipitation, The sequence leading to pasture development also differs. Some pashlres are created by planting grasses directly into forest slash, while others are created after one or two years of annual cropping or after a cropping and fallow sequence. The choices of grass species and the practices of interplanting with legumes also differ. These factors can influence whether a pashlre soil will become a source or sink of C. Once established, pasture management by stocking rate, burning fi'equency, effectiveness of weed control, fertilizing, or disking may also affect soil C balance [Neill and Davidson, 1999]. Therefore, in some locations, C stocks in pastures are lower compared with the original forest [R, C. Luizao et al., 1992; Desjardins et al., 1994]. In other locations, pasture grass productivity declines in older pastures, but soil C concentrations remain relatively constant [Falesi, 1976; serrao et al., 1979; Buschbacher et al., 1988]. Yet in other locations, inputs of C from roots of pasture grasses cause increases in soil C stocks [Cerri et aI" 1991; Bonde et al., 1992; Trumbore et al., 1995; Moraes et al., 1996; Neill et al., 1997; Bernoux et aI" 1998; Cerri et aI" 2003]. Neill and Davidson [1999] synthesized the available literature on soil C stocks in pasture following deforestation in the Amazon. They repOlied that 19 of 29 pastures examined accumulated C in surface soils and 10 showed C losses. They also observed a strong relationship between pasture grass species and the change in surface soil C stocks. Pasture planted to Braclliaria humidicola lost C and those planted to Panicum maximum and Brachiaria brizantha gained C. Moraes et al. [1996] found that total soil C contents to 30 cm depth in 20-year-old well-managed pastures were 17% to 20% higher than in the original forest sites in the western Amazon. A comparison of C budgets for forest and pastures in the eastern Amazon was made by Trumbore et al. [1995]. In a rehabilitated and feliilized pasture of B. brizantha, they estimated gains, relative to forest soil C stocks, of over 20 Mg C ha- I in the top 1 m of soil and a loss of about 0.5 Mg C ha- I in the 1-8 m soil depth during the first 5 years following pasture rehabilitation. More than 50% of the forest-derived C in sUlface soils of pastures in converted Amazon forest turns over in 10 to 30 years [Chone et al., 1991; Trumbore et al., 1995]. Cerri et at. [1999] reported carbon sequestration of 0.27 Mg C ha- I a-I for the 0-30 cm depth range (C-sequestering rates for a 20-year time range), while Neill et al. [1997] reported annual soil C accumulation rate, in the top 50 cm, in the range of 0.2 to 0.3 Mg ha- i for secondgrowth ages from 3 to 23 years. These results are in the range (0.2-3.9 Mg C ha- I a-i) of those reported by Sampson et al.
LurzAo ET AL.
315
[2000, p. 199] for pastures in wet tropical areas ofthe world. For the present Shldy,.
There are numerous factors and processes that must be considered in estimating the direction and rate of change in soil organic carbon (SOC) contents from changes in soil management. Post and Kwon [2000] reported that impOliant factors for increasing carbon sequestration include: (l) increasing the input rates of organic matter; (2) changing the decomposability of organic input increasing the light organic carbon fraction; (3) placing organic matter deeper in the soil either directly by increasing belowground inputs or indirectly by enhancing surface mixing by soil organisms; and (4) enhancing physical protection through formation of aggregates or organomineral complexes.
from aboveground forest biomass to the atmosphere. There are also additional benefits of intact forests in ameliorating floods, conserving soils, maintaining stable regional climates, preserving biodiversity, and supporting indigenous communities and ecotourism industries [Lauran~e et al., 2001]. Therefore, any policy changes that reduce the rate of deforestation would have the greatest potential for reducing the net emission of greenhouse gases. Moreover, it is also greatly desired that these policies enhance the rate of carbon sequestration in soil.
Amazonian soils seem particularly prone to rapid declines in natural fertility. One reason for this is their medium to low level of soil organic matter (SOM), which is generally associated with rapid C and nutrient mineralization due to favorable moisture conditions during most of the year. Low contents of SOM imply low retention of cations and, consequently, soils susceptible to nutrient leaching. Any human intervention in the forest implies some degree of change in struch1re and functioning of the ecosystem, although some interventions are not destructive and may be sustainable over a fairly long time. Forest-management practices, such as selective logging, which supposedly causes little impact on soil, may still represent a considerable impact on soil properties, including soil compaction, erosion, and leaching. Also, selectively logged forests become susceptible to fire due to the addition of large quantities of dly material on the soil surface and due to opening of pathways for wind to enter the forest; the wind dries out the soil surface and creates additional combustible material [Nepstad et al., 1999]. Thus, changes in carbon and nutrient cycling are introduced in managed forests after the first intervention. The impacts of selective logging may be considerably reduced if carefully designed interventions are conducted, such as the reduced-impact logging (RIL) procedures (Table 2). In Para state, a comparison between plots submitted to conventional selective logging (CL) and plots harvested by RIL procedures, during 4 years, showed that the total area of soil , affected (tractor tracks plus skid roads plus log decks) under CL was between 8.9% and 15.3%, while under RIL, it was between 4.6% and 8.6% of the total area [Asner et aI" 2004a]. The largest proportion of the plot damages (4-12%) was caused by the skid tracks, and by reducing this specific damage, RIL reduced canopy openings and the susceptibility
1,4, Conversion From Degraded to Well-Managed Pasture
More than half of the cattle pasture areas in the Brazilian Amazon are degraded [Serrao and Toledo, 1990; DiasFilho, 2003], which represents approximately l.3 Mha. Productivity of Amazonian pastures is often good during the first 3 to 5 years after establishment. A rapid decline in productivity of the planted grasses occurs after this period due to encroachment by herbaceous and woody invaders [Uhl et al., 1988; Sen·ao and Toledo, 1990]. Ifuncontrolled, invader species gradually dominate and severely degrade pastures, a condition characterized by a complete dominance of the weedy community [Dias-Filho et al., 2001]. If the entire area now under degraded pasture could be well managed, and assuming the rate of soil C sequestration for well-managed pastures of 0.2 to 3.9 Mg C ha- I a-I in the 0- to 30-cm soil layer [Sampson et al., 2000], the potential soil C sequestration from converting degraded to well-managed pasture in the Brazilian Amazon varies from 2.6 to 51 Tg C a-i. Note, however, that the high value of3,9 Mg C ha- I is much higher than is likely to occur over a wide area.
2, WHAT ARE THE CAUSES OF DECLINE IN SOIL FERTILITY UNDER AGRICULTURAL AND FOREST MANAGEMENT AND WHAT IS THEIR RELATIVE IMPORTANCE?
316
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
LUIZAo ET AL.
Table 2. Percent of the Forest Soil Affected by Different Actions During Forest Management in Some Selective Logging Studies in Brazilian Amazonia"
Table 3. Comparative Studies on Soil Erosion, Runoff, and the Runoff/Total Rainfall Ratio in Several Brazilian Amazon Locations" Vegetation Coverb ,/'
Paragominas c
f
Manaus CL (%)
Moju CL (%)
RIL (%)
CL (%)
RIL (%)
CL(%)
Acre RIL(%)
>12.0
0.4 1.2 7.7 9.3
0.6 0.6 3.9 5.1
1.0 1.3 7.7 10.0
0.6 2.0 5.1 7.7
1.5 3.4 10.1 15.0
0.9 1.1 1.8 3.8
b
Log decks ,Skid tracks Log tracks Total area affected
Fundayao Floresta Tropical d
C
"Abbreviations are RIL, reduced impact logging; CL, conventional logging. bBIONTE [1997] (timber harvesting: 34 m 3). cSilva et al. [2001] (harvesting: 35 m3). dHolmes et al. [2002] (harvesting: CL= 25 m 3, RIL= 26 m3). eJohns et al. [1996] (harvesting: CL= 37 m 3 , RIL= 30 m3). fOliveira and Braz [1995] (harvesting: 20 m\
of the residual forest to fire, producing less coarse woody debris (CWD) [Asner et al., 2004a; Keller et al., 2004]. The amount of coarse litter (dead wood) found in plots, I year after selective logging, was 2.7 times higher under CL than under RIL, and only 8-18% of the mass of the debris was small-diameter wood «10 cm), which undergoes faster decomposition. At Floresta Nacional do Tapaj6s, dead wood increased from 50.7 ± 1.1 Mg ha- I in intact forest to 76.2 ± 10.2 in RIL, while in Cauaxi, it increased from 55.2 ± 4.7 to 74.7 in RIL and to 108 ± 10.5 Mg ha- I in CL [Keller et al., 2004]. However, despite the many advantages ofRIL, most logging in Amazonia is not done in this way, and serious damage continues to be caused to the logged plots. The impacts of selective logging on soil physical, chemical, and biological features are obviously stronger in the log decks and parts of the logged plots such as the tractor tracks, which are submitted directly and repeated times to mechanical impacts, certainly affecting soil physical properties, thus changing the soil water availability to plants. Near Manaus, the Biomass and Nutrients in Tropical Moist Forest (BIONTE, National Institute for Amazonian Research! Department for Intemational Development (INPAJDFID)) Project has evaluated the effect of extracting 34.3 m3 of timber per ha [6-10 h'ees, with diameter at breast height (DBH) >55 cm] on the soil properties, nutrient cycling, and forest regeneration [BIONTE, 1997]. Soon after the logging, tensiometer measurements showed an increase of 246% in soil water tension in the tractor tracks (to 430 kPA) and a 37% increase in the center of clearings (to 240 kPA), compared to the forest control (175 kPA) [Mello-Iva and Ross, 2006]. The tractor tracks created by a former selective logging procedure 7 years before were still perfectly visible, treeless, and compacted, showing water accumulation on the surface, as a clear indication that these microsites could not be rehabilitated in the short or medium tenn.
317
Contrary to fOlmer assumptions that grasses produce a fair cover of the soil surface after slash-and-bum [Falesi, 1976], some studies show strong losses through erosion/leaching in Amazonia [Barbosa and Fearnside, 2000] (Table 3). For instance, in their 3.5-year study on soil erosion under two land uses, primary forest and adjacent pasture derived from forest in Apiau, Roraima, Barbosa and Fearnside [2000] found that, for a slope of20%, soil erosion under pasture (1128 kg ha- I a-I) was 7.5 times higher than under forest (150 kg ha- 1 a-I). The mnoff was almost three times higher in pasture (31.8 cm a-I) than in primmy forest (11.3 cm a-I). Additional studies have also shown strong losses of soil through laminar erosion after forest conversion, such as the studies conducted on the Transamazon Highway area by Smith [1976] and Fearnside [1980], both using the stake methodology. Both studies stressed that conversion of primmy forest into pastures increases soil laminar erosion, which will certainly be reflected in the regional and global socioeconomy. The major reason is that removal of forest cover exposes the soil surface to direct impacts of sun and of raindrops (e.g., an increase of 571 mm or 37% in the Apiau study), in addition to the direct compaction caused by cattle trampling. The above results confhm studies carried out in forested sites such as the study by Ross et al. [1990], on Maraca Island, Roraima (Table 3). These authors found that, just after the treatments, clear cut of forest increased fourfold the soil and nutrient losses in a topographic gradient, compared to primmy forest with intact litter and canopy cover, and losses were almost 2.5 times higher than in forest plots where canopy was removed but litter layer was kept in place. In the clayey Oxisols of central Amazonia, the soil micropores (between 0.01 and 0.03 ,"un) result from the compact assemblage of clayey particles of kaolinite, while soil macropores (between 0.1 and 100 /lm) result mainly from soil fissures and biological activity in various forms: galleries,
V
Pasture Bh t Primary forest // Clean pasture Pili Pasture with weeds Pili Primary forest Clear cut Canopy removal Primary forest Pasture Bh Primary forest
Soil Loss (kg ha- I a-I)
Runoff, 106 L (ha- l a- 1)
1703 158 3556 664 330 1140 475 270 1128 150
2.32 0.27 9.87 5.09 0.37
3.18 1.13
Runoff: Rainfall (%)
49.8 25.7 2.2
15.1 7.4
Location
Author
Manaus, Amazonas
Feal'llside [1986]
Ouro Preto do Oeste, Rondonia
Fearnside [1989]
Maraca, Roraima
Ross et al. [1990]
Apiau, Roraima
Barbosa and Feal'llside [2000]
aAdapted ft'om Barbosa and Feal'llside [2000]. bBh, Brachiaria humidicola; Pm, PanicuIII maximum.
channels, chambers, etc. [Grimaldi et al., 1993]. Under forest, Oxisols have a bimodal pore spectmm (small and large pores together). A study carried out near Manaus showed that mechanized deforestation before buming and planting caused a 70-80% decrease in pores >0.1 /lm, which are the pores containing water available to plants, and the pore spectmm became virtually unimodal, with sh'ong soil compaction between 20-40 cm depth. Under these conditions, hydraulic conductivity decreases, rainwater accumulates on the soil surface, and infiltration can be 10 times slower than under forest [Grimaldi et al., 1993]. In the same study, it was found that a young and well-managed pasture at FUCADA (Center for Technical Support of the Manaus Agriculture and Ranching District), 3-5 years old following manual deforestation, showed a strong decrease in pores >0.1 /lm, implying limitations to grass root growth, water infiltration, and oxygen diffusion. Besides the impact caused by cattle h'ampling, part ofthe soil compaction found in managed pastures may be a consequence of the strong reduction in the diversity of soil biota, thus influencing soil physical stmcture and fertility. An experimental soil manipulation canied out in cenh'al Amazonia showed that a clayey Oxisol under an old B. hllmidicola pasture was compacted in the upper layer by the action ofthe pantropical earthworm Pontoscolex corethrllrlls [Barros et al., 2001]. This earthworm was very abundant and widely dominant in the area; its feces contained a very high propOltion of fine clay that, once deposited on the soil surface, produced a tiny compact layer. When a I_m 3 block of the compacted pas- . ture soil was moved and inserted into the neighboring forest in less than 1 year, the forest organisms colonized the formerly compacted soil and produced aggregates and porosity similar to the forest soils; in the opposite way, the wellstmctured forest soil, once moved to the pasture became
compact within a year [Barros et aI., 2001]. The same pattem was observed in Maraba (PA), showing that soil biodiversity decreased sharply by 70% under grazed pastures compared to forest: soil compaction, which begins with rice cropping following slash-and-bum, was accelerated with pasture age and was stronger in the 2- to 5- and 5- to 10-cm soil layers (T. Desjardins, personal communication, 2007). These results emphasize th~ need for keeping a diversified soil biota (and a diversified mixture ofplant species capable offeeding it) in order to mai~tain the soil shucture and fertility. 3. EVIDENCE OF FERTILITY DECLINE IN MANAGED AMAZONIAN FORESTS Changes in soil fertility in Amazonia are connected to the different levels of human intervention on the forest ecosystem. They may be slight and reversible, for example, as in careful selective logging or highly damaging as in a poorly managed old pasture that degrades the soil system and makes recovery difficult. Even worse, changes can be highly deshl1ctive, as in mining operations, which, although normally resh'icted to small localized areas, generally represent a very intense destructive impact on native ecosystems. For instance, bauxite mining involves the removal of all vegetation and the entire topsoil layer, causing soil impoverishment, erosion, and toxicity, thus affecting the flora, fauna, water quality, and people in the region in many ways. 3.1. Soil Fertility Under Selective Logging
The selective logging conducted in central Amazonia by the BIONTE Project [BIONTE, 1997] resulted in an export of65.3 kg N, 0.86 kg P, and 18.8 kg Ca per hectare of forest. Nutrient expOlt was relatively modest and at the soil (rather
318
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
than the ecosystem) level was largely compensated by the addition of a new pool of nutrients from plant residues resulting from the logging. However, pati of the clearings had microsites v,Yith newly exposed soil surface where fine litter decomposition rates were lower in the first months, and available soil nutrients were lost (at least temporarily) from the rooting zone through soil percolation (in the first weeks after logging) due to the addition of large amounts of new organic material and the absence of absorbing roots [MelloIvo et al., 1996]. Additional losses of N may occur, since logging can increase emissions of N20 and NO from 30% to 350% depending upon soil conditions [Bustamante et al., this volume], affected by changes in nutrient and water circulation, along with soil compaction by heavy machinery in skid trails and log storage decks. In the mONTE Project, the microsites of the clearings, where an accumulation of plant debris (branches, fallen canopies, etc) took place, which were usually located near the remaining forest edges, showed an increase in nutrient availability in the upper soil layers. This occurred because greater amounts of organic substrate, together with higher soil moisture, induced higher decomposition rates in the two first years, resulting in higher concentrations of available nutrients in the soil (especially Ca and Mg) after 1.5 years [BIONTE, 1997]. A parallel study confirmed that wood residues from selective logging caused increases in soil-nutrient availability (via decomposition), especially for the exchangeable bases K, Ca, and Mg in the rainy season [Ferreira et al., 2001]. Decomposition of CWD (with diameter> 10 cm) released half of its C content in the first 5 years (19.9 Mg C ha- 1); the remaining content is released in another 20 years. However, CWD with diameter between 2 and 10 cm is decomposed in less than 5 years [Summers, 1998]. A high proportion of the nutrients is released in the first 4 years of wood decomposition, especially P and K: fluxes were four times higher for P and K, and three times higher for Mg in the logged plots (though still lower than fluxes via fine litter in intact forest) [Summers, 1998]. A recent study in an open tropical forest in luruena, Mato Grosso (10028'S; 58°30'W), at the southem fringe of Amazonia, showed that selective logging induced a strong increase in the mOliality of palms (120-340%) and a decrease in annual production of CWD in the plots logged more recently: respectively, only 1.1 Mg ha- 1 a-I and 2.8 Mg ha- 1 a-I in the plots logged 2 and 6-7 years previously, versus 5.3 Mg ha- 1 a-I in a plot logged 11-12 years previously and 5.7 Mg ha- 1 a-I in the undisturbed forest [Pauletto, 2006]. However, the main change observed was the sharp increase of 105% in the stocks and 37% in the volume of the CWD fraction with smaller diameters (2-10 cm). This fraction represented only 15-16% of total stock of CWD, but it stored 29% of total N, 35-40% ofP, 18-20% ofK, 37-42% ofCa,
and 30-35% of Mg. There was an overall increase of 54109% in nutrient contents of this fraction in the logged plots compared to the undisturbed forest. The reason for such in.. creases in relation to the total stocks of nutrients in CWD is that nutrient concentrations in this fraction are much higher (especially for Ca and Mg) than in dead wood with larger diameters. In the clearings produced during selective logging operations, the microsites under decomposing CWD had significantly higher concentrations of soil C and nutrients. This was particularly strikingly for Ca, which reached concentrations up to 590% higher in soil microsites under CWD. However, it must be remembered that nutrient release from CWD is restricted to spatially limited and specific microsites within the clearings, and there is a risk of not being absorbed by roots and, subsequently, being removed from the rooting zone of the forest ecosystem [Mello-Ivo et al., 1996]. Similar removals were found in a parallel study of nutrients released by the decomposition of fine litter in the clearings: early in the wet season, 5 months after logging, concentrations of the exchangeable bases K+, Ca2+, Mg2+, and Na+ were fi'om two- to fourfold higher than those in the forest control, but 2 months later, near the end of the wet season, strong decrease in the concentrations, probably due to leaching, especially for Mg, were observed in the clearings [Ferreira et al., 2006]. The general result of selective logging is, then, a strong redistribution of carbon and nutrients from the standing biomass, associated with the creation of microsites, with heavy additions to the soil surface of new organic materials such as fine or coarse litter [Luizi'io et al., 1998], which can affect the natural forest regeneration, favoring either pioneer or climax species at different times after logging. Also, a short-term removal of some nutrients by soil percolation is observed in the cleared areas due to high release through decomposition of new plant residues and lack of absorbing roots in these microsites. 3.2. Forest Fragmentation and Soil Fertility
One of the consequences of selective logging (and the majority of other human interventions on dense forests) is forest fragmentation, which also affects the cycles of mineral elements. Long-term studies carried out by the Biological Dynamics of Forest Fragments Project (BDFFP, INPA! Smithsonian Institution), near Manaus, found an increase in the stocks of both fine and coarse litter on the soil surface as a consequence of forest fragmentation, mostly as a result of edge effects [Nascimento and Laurance, 2004]. Annual production of fine litterfall, measured during 3 years, was 0.68 Mg ha- 1 higher in forest areas suffering edge effects
LUIZAO ET AL.
than in the forest interior, >300 m fi'om the edge (9.50 ± 0.23 versus 8.82 ± 0.14 Mg lla-1nVasconcelos and Luizi'io, 2004]. The concentrations of'Ca in leaf litterfall were higher near the edges, probabllbecause of strong soil Ca mobilization by pioneer speci~/growing by the edges of forest fragments [Lucas et al., 1993]. Thus, forest fragmentation may also affect litter quality by favoring the recruitment of successional tree species at the expense of old-growth species [Laurance et al., 1998, 2004], as these two groups can differ strongly in nutrient contents and in decomposition rates [Mesquita et al., 1998]. These changes are closely related to variations in the abundance, species richness, and composition ofmany groups of soil invertebrates in response to edge effects or changes in vegetation cover [Didham, 1998]. There are indications that land cover changes in Amazonia affect decomposition mainly through changes in plant species composition, which in tum affects litter quality in fragment edges, particularly in heavily disturbed edges where successional tr'ees become dominant [Vasconcelos and Laurance, 2005]. 3.3. Soil Fertility Under Slash-and-Burn Practices
Despite the fact that impacts of selective logging, and especially forest fi'agmentation, may be severe in some aspects or for some organisms and for the forests functioning over the long run, by far, the greatest impacts on soil feliility and dynamics are caused by the generalized use ofthe slash-andbum practice for clearing and preparing land for agriculture or pastures in Amazonia. Biomass buming is used for releasing nutr'ients stored in biomass to fertilize chemically poor soils in most of the Amazon. The increase in nutrients incorporated into the soil by the ashes has been confirmed in many studies in the Amazon [ASB, 2002; Palm et al., 2005]. Seubert et al. [1977] calculated that the buming of the primary vegetation on an Ultisol in Peruvian lowland forest incorporated 67 kg ha- 1 ofN, 6 kg ha- 1 ofP, 38 kg ha- 1 ofK, 75 kg ha- 1 ofCa, and 16 kg ha- 1 of Mg to the soil. The newly improved soil feliility situation may endure for several years, although crop production generally decreases sharply years before the fertility declines [Seubert et al., 1977; Sanchez et al., 1983; Desjardins et al., 2000]. This implies that other factors are also involved, among which the partial gaseous losses of nutrients and the decrease of biological activities involved in organic matter decomposition and mineralization. The buming used for releasing mineral nutrients from the biomass also represents a direct and considerable loss of nutr'ient pools by volatilization just at the beginning of the cultivation process [Kauffinan et al., 1998; Kato et al., 2004; Palm et al., 2005]. A field experiment on biomass buming of a 7-year-old second growth in Para state showed strong nutrient losses through fire: 98% of the C; 96% of the N;
319
47% of the P; 48% of the K; 35% of the Ca; 40% of the Mg; and 76% of the S are stored in bumed material [Mackensen et al., 1996] (SHIFT Project). Further losses can be expected in the succeeding periods because forest conversion to agricultural use through the slash-and-bum system changes both the quantity and the quality of organic matter deposited on the soil surface, altering the soil moisture and temperahlre regimes and, consequently, the biological processes that control litter decomposition and the dynamics of the soil organic matter. Biomass buming has the immediate effect of decreasing the biomass stock and nutrient pools of the system, and preventing the continuous and high input of carbon and nutrients to forest soils through litter fall, either fine or coarse. In central Amazonia, the annual input of carbon and nutrients to the soil surface of an upland forest on a plateau with clayey Oxisol was 3.9 Mg ofC, 151 kg ofN, 3 kg ofP, 15 kg ofK, 37 kg ofCa, and 14 kg ofMg through fine litterfall alone [Luizi'io, 1989]. This material is rapidly decomposed, generally within a year after litter fall, releasing carbon and nutrients to plant roots. Without a litter-layer cover, the soil becomes exposed to heating from the direct sun and to the impact of raindrops, which increase compaction, erosion, and consequent nutrient losses. In Rondonia, a 16-month study of the soil solution in an intact forest compared to a slash-and-bum area showed different results at distinct times [Piccolo et al., 1994]. In the first wet season'ifluxes of the most abundant ions (Si 4+, NH4+, N0 3-, Mg 2+, soi-, K+, Ca2+, and Mn 2+) were higher under bumed forest than under intact forest; however, in the following dry and wet seasons, the nutrient fluxes were higher under intact forest. This suggests a significant decline in the 25-cm soil solution nutrient concentrations in bumed areas, after a shOli-lived nutrient enrichment. After the installation of pa,stures, periodic bumings are common as a cultural practice for cleaning and refertilizing the land, representing fmiher losses of carbon and nutrients from the crop system. A study carried out in a 7-year-old pasture in Apiau, Roraima state, Brazil, during its third bum, showed that 210 days after bum, the C stocks (20.2 Mg ha- 1) were significantly lower than in the prebum (26.0) in the top 20 cm of soil [Barbosa and Fearnside, 2003]. The authors point out that losses of C by buming and mineralization are higher than gains from humification of roots and plant remains, but the cumulative imbalance only becomes apparent after a certain time (in this case, after 210 days). Changes in soil physical conditions, such as those described above, have strong influences in nutrient recycling and availability to plants, likely because they affect biological transformations of nutrients such as N, P, and S in soil. This was confirmed by a reanalysis of former data [e.g., Falesi, 1976], which stated that pashlres improve soil
320
LUIZAO ET AL.
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
fertility [Fearnside, 1980]. In fact, available P (PzOs by North Carolina extraction: 0.05 N HCl and 0.025 N HzS04), considered a critwally limiting nutrient with initial PzOs concentrations of6.9 f!g g-I in the forest, increased to 41.8 f!g g-I after biomass burning in the Paragominas region, but after 5 years, it decreased to a plateau of 4.6 f!g g-I (one order of magnitude lower), remaining there until 10 years of age in the pastUl:e [Fearnside, 1980]. Even considering the unlikely possibility of P now stored in t~e grass biomass, a considerable net decrease in soil available concentrations was observed instead of enrichment. Probably, some of the P in the soil was fixed in oxides, while part was lost from rooting zone by slow percolation. Forest clearing changes soil physical properties, which in turn affect chemical processes: it exposes soil and residual litter to rainfall events that can accelerate leaching, leading to removal of nutrients at rates higher than those for nutrient mineralization of dead roots and soil organic matter via microbial decomposition. In the long tern1, these changes can decrease soil nutrient stahlS and soil organic matter concentrations [Woomer et al., 1999]. These severe impacts on soil functioning after forest conversion, generally involving the removal of the forest canopy, should be expected because they are a result of the replacement of a dense tropical forest, with high biological diversity and biomass, by a velY simplified crop production system (generally a monoculture) or by pastures composed of a single grass species, usually exotic and implanted in areas with precarious infrastructure. Under these conditions, the basic mechanisms for the functioning of the native ecosystem are disrupted, including its efficient nutrient recycling process that is based on the stock and biological transformations of organic matter. Soil biota is seriously reduced and becomes dominated by a few species that are resistant to the impacts. New litter production is
generally very low in the first years, and its quality also may be low, failing to provide adequate diversity of substrates and proper cover of the soil surface, which are essential for recovery of soil biodiversity needed to promote the nutrient recycling in the system. In addition, some nutrients, such as Nand S, can be lost in high propOliions in the initial and additional burnings and can become limiting to the new system [Fernandes et al., 1997]. Most of the pastures in Brazilian Amazonia are planted with B. humidicala, which can generally grow well in the first years, covering the soil surface and can recover, within a few years (5-7 years), the soil C content [Cerri et al., 1991]. However, the quality of the new organic matter that has originated from the grass is poor (with higher C:N ratio and lower mineralizable Nand P contents) and does not allow the soil biota to act effectively in nutrient recycling, leading to nuh'itional deficiencies in the soil [R. C. Luiziio et al., 1992; Feigl et al., 1995]. Because pasture management is generally poor and inadequate in the region, the degradation factors of soil and/or pasture production evolve very quickly and can cause pa,sture abandomnent in a few years. The study of a pasture chronosequence from 2 to 13 years of age on clayey Oxisols (>70% clay), all located within a 10-km radius and all on a flat plateau near Manaus, showed significant changes in soil C and N dynamics [Luiziio et al., 1999]: (1) soil microbial biomass, C and soil N mineralization increased up to 5 years in pastures, followed by a gradual decrease, which was very accentuated after 8 years (Figure 2); (2) pastures showed a higher propOliion of soil N as N-N03 (while the forest control, as a result of nutrient conservation, had more N as N-NH4), facilitating losses by leaching, denitrification, or complexation in the soil; (3) the low N mineralization rates corresponded to a de-
2500
25
2000
20 .g>
2: 1500 () 1000
15 2: r."
500
5
~
z
OJ
OJ
0
iii
0
2
5
6
7
8
12
13
0
Pasture age (years)
Figure 2. Soil microbial biomass-C (Ilg C g-l) and concentrations ofN-NH4 (Ilg N g-l) in the 0- to 10-cm soil layer in pastures of Brachial'ia hlllllidicoia aged fi'om 2 to 13 years, in Manaus, Amazonas. Redrawn from LlIizGO et al. [1999].
crease of organic N, leading to N deficiency in the soil in the oldest pashlres; (4) .~he oldest pashlres (12-13 years old) also showed an accentUated decrease of soil organic C. This . 'I pattern agreed wrth 9ie one observed for fine root mass in the upper 0- to 20-cny§oillayer [LlIiziio et al. 1999]: for roots with diameter 0.1-1 mm, biomass decreased from 1110 g m-z (5-year-old pashlre) to 361 g m-z (7 years) and then to 243 g m-z (12 years). This is a dramatic change if compared to figures obtained for a nearby young and well-managed 2-year-old pasture on similar soils where potentially renewable fine root (0.1-1mmin diameter) mass was 8.9 Mg ha- I a-I versus 5.1 Mg ha-Ia- 1 in the forest, for the upper 20 cm of soil [F. J. LlIiziio et al., 1992]. In the chronosequence, the C:N ratio in roots increased with pasture age, which also indicates a decrease in the nutritional quality of organic matter derived from roots. The Shldy of two pasture chronosequences in Santarem, Para, one on clayey and the other on sandy soil [Asner et aI., 2004b], showed that the C stocks in aboveground biomass and soil decrease with pasture age. Plant biomass decreases are probably related to lower C, available P and exchangeable Ca concentration in the soil; in addition, ecosystem P decreases with pasture age. Another Shldy of pasture chronosequence, also in Santarem [Townsend et aI., 2002], showed significant losses of organic matter and soil total P with pashlre age in soils that were already deficient in P; however, losses occurred for the inorganic P fractions, while organic P forms remained constant or even increased, despite losses of organic matter. The observed losses were attributed to changes in soil microorganism communities. Losses of soil N to the atmosphere in young pastures can be substantial, as demonstrated by a comparative study of NzO fluxes in forest, burned areas, and young pasture, all adjacent and on clayey Oxisol (>70% of clay), in Manaus [Luiziio et al., 1989]. The NzO annual flux increased threefold in pasture compared to a forest control (which, in the h'opics, is considered to be naturally high) [Davidson et al., 2004b]: 1.9 kg ha- I a-I in the forest and burned forest against 5.7 kg ha- I a-I in the young pasture. There was a strong seasonal effect: in the dry season, NzO fluxes were similar in forest and pasture, but in the wet season, fluxes were from threefold to fivefold higher in pasture (>10 ng cm-z h- 1 in March and April, the rainiest months). Later measurements iri neighboring pastures showed that older pastures decrease the emissions of nitric and nitrous gases as well as N mineralization rates and soil available P; thus, aging pashlres decrease the concentrations of soil nitrate (due to denitrification, nitrate leaching, or soil acidification) and increase the relative concentrations of ammonium [Luizelo et al., 1999]. Lower mineralization and nitrification rates are
321
related to decreases in demineralization potential and associate lower rates of NzO and NOx emissions from soils in aging pastures [Bustamante et al., this volume]. 4. NUTRIENT MANAGEMENT REGIMES IN USE IN AGRICULTURE AND FORESTRY AFS are often mentioned as a type of sustainable agriculture that is appropriate for the edaphoclimatic conditions of Amazonia because the species selection and arrangement can produce a nutrient management regime suitable for keeping or even improving soil fertility [Fernandes et al., 1997]. However, long-term studies on the sustainability of this land use do not exist in Amazonia. Smallholders in Nova California, Rondonia, Brazil, are conducting the first large and organized experiment on the sustainability of such systems (now spanning 19 years), implemented after the usual land clearing through slashand-burn of native dense forest. The systems are based only on three regionally imporiant fi'Uits: "cupuayu" (Theobroma grandiflorllm, Sterculiaceae), peach palm (Bactris gasipaes, Palmae), and Brazil nut (Bertholletia excelsa, Lecythidaceae). All farmers also have pashlre plots on their fanns. Most of the 200 farmers involved (in the Mixed and Dense Economic Re$toration Project (RECA) association, Mixed and Dense Economic Reforestation Project) had no leguminous cover plants included in their AFS in the first 10-12 years, when ad evaluation was made, comparing AFS to pastures of similai' ages and control forest plots [Alfaia et al., 2004]. The AFS soils maintained their improved chemical conditions derived from biomass burning, especially the increased levels of exchangeable Ca and Mg and the reduction of exchangeable AI, while maintaining stable levels of organic C, even when compared to adjacent primary forest soils. In contrast, the improved soil conditions in the pastures were transitOly and short lived: after the first years, low soil pH, high level of AI, and low levels of exchangeable bases returned. However, K and P fell to extremely low levels in the AFS. This reduction was reasonably attributed to nutrient exports by consecutive harvests of T. grandiflarum and B. gasipaes fruits, since K is one ofthe most impOliant nutrients for the yields of these fruit species. Considering the high cost of mineral fertilizers in the region and the soil characteristics that favor mineral nutrient leaching, a simple solution adopted by farmers, was to stop burning T. grandiflorllm fiuit rinds, rich in K (and also in Nand P), grinding them and adding the product on the AFS soils. Additionally, some of the farmers decided to introduce leguminous cover plants (mostly PlIeraria phaseolaides) in their plots, and apparently, the nutrient balance problem is now under control for the RECA farmers [So Alfaia, personal communication, 2007].
LUIZAO ET AL.
322 MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS In central Amazonia, the efficiency of a multistrata AFS for organic matter recycling was studied at full (FF) and low (LF) fertilization levels, compared to natural fallow [Uguen, 2001; Schroth et al, 2002]. (Full initial feliilization level stands for 1 applications of 38-215 g planC I a-I of N, 18-90 g planC a-I of P, 39-331 g planCI a-I of K, and 0.2-1.5 kg of dolomitic lime to each h'ee; LF level corresponded to application of only 30% of the full levels of fertilizer and lime.) The organic and nutrient inputs through litterfall and plUnings were assessed during 1 year in a 5-year-old AFS composed of four h'ee species: Brazil nut (B. excelsa), "cupuayu" (T. grandijlorum), peach palm (B. gasipaes), annatto (Bixa orellana). The soil was heterogeneously covered by a legume cover crop of P. phaseoloides. Large differences in leaf litter nUh'ient concenh'ations were found between the four tree species [Uguen, 2001]. Annatto leaf litter had the highest nutrient concenh'ations for all measured macronuh'ients, while Brazil nut and "cupuayu" had the lowest N, P, and K concentrations. Total organic inputs were lower in the AFS (4.56 and 3.59 Mg ha- I in FF and LF, respectively) than in the fallow (5.1 Mg ha- 1). Fine litterfall was relatively low (1.6 and 1.5 Mg ha- I in FF andLF, respectively) in the AFS, and plUning biomass accounted for more than 50% ofthe total organic inputs (2.95 and 2.12 Mg ha- I in FF and LF, respectively). Chemical fertilization had no significant effect on litterfall but significantly increased plUning biomass, especially for annatto, from which the main contribution to total nutrient inputs came. Ca and Mg inputs were enhanced for all species, and P inputs were increased only for peach palm and annatto. In a neighboring similar AFS, Pueraria (used as a legume cover crop) had a noticeable high litter production (1300 g m- 2 ), showing its importance to the nutrient balance and recycling in the AFS [Uguen, 2001]. Chemical fertilization, which increased plUning biomass, also enhanced Pueraria growth and, thus, the cover crop litterfall. There were large differences in nutrient inputs between species. For all nutrients except calcium, "cupuayu" litterfall represented the lowest nutrient input, while annatto had the highest contribution to total nutrient inputs. Major nutrient inputs in the agroforestry system came from plUnings: about two thirds of the Nand P, and 80% of the K, whereas 60% of Mg and 50% of the Ca came fi'om litter [Uguen, 2001; Schroth et al., 2002]. Thus, it appears that the spatial alternation of species with high and low nutrient cycling could favor a good soil cover and reduce nutrient leaching from prunings. Additionally, the presence of a cover crop such as Pueraria, producing abundant and high-quality litterfall, seems to have high importance for soil fertility rehabilitation under AFS. At the Experimental Station of the Centre for Agroforestry Research/Brazilian Agency for Agricultural Research
(CPAA/Embrapa), near Manaus, different formulations of AFS were tested on abandoned pasturelands, after an initial slash-and-burn of young (4-6 years old) fallow and a light P fertilization (20 kg ha -1), together with 1.5 Mg ha-I of lime, applied only at the onset of the experiment [Femandes et al., 1999]. In the first years, the system, which was planned to reproduce an improved pasture (B. brizantha mixed with the legume Desmodiwn ovalifolium) planted together with commercial timber species (Schilozobium amazonicum and Swetenia macrophylla), showed the best soil cover and conditions for nutrient recycling and soil biological activities [LuizQO et a I., 2006]. However, a few years later, the fast tree growth in the two AFS with flUit, palms, and timber trees showed better conditions for soil nutrient recycling through higher litter production together with higher litter quality and green manures coming from prunings of Inga edulis (planted in the tree lines in the improved pasture and in the multistrata AFS) and Gliricidia sepium (used as green hedge in all AFS). When the palm-based and multish'ata AFS were 6-7 years old, their fine-litter production corresponded to only 25-30% of the amount produced by the adjacent second growth (used as a control). However, most of the AFS species had higher nutrient contents in litterfall, and the AFS received regular additions of green manures, originated from periodic plUnings of the leguminous plants G. sepium (fi'om the green hedges) and 1. edulis (planted in the tree rows within the AFS); both ofthese were scattered as mulch on the smface of the AFS soil. Because of the pruning addition, a similar or even better nutrient balance could be observed compared to the second growth as early as 6-7 years of age (Table 4). These results suggest that a young agroforestly system, especially if it is not dense enough, requires additional sources of organic matter, e.g., from cover crops or green manures, to replenish soil organic matter and to balance nutrient cycling, thus allowing early and proper nutrient cycling, which is essential for optimal crop development. Traditional smallholders cultivating rice, beans, maize, or most frequently, manioc (cassava), after slash-and-burn cleaning.ofthe land, usually do not apply any inorganic fertilizer to the soil, and velY seldom do they use organic manures in their fields in the Brazilian Amazon. The most conunon practice is to burn any organic residues generated on the property, using the resulting ashes to fertilize the soil. On the other hand, the rapidly expanding area of largescale industrial cultivation of soybeans relies on heavy doses of fertilizers and pesticides to guarantee good yields in Amazonia [Fearnside, 2001]. Significant applications of fertilizers and pesticides are also required by sugarcane to overcome climatic and edaphic limitations Sugarcane plantations are currently present only in a small portion of the Amazonian biome but have the potential to spread to many
323
Table 4. Annual Nutrient Input to the Soil in Two 6-Year-Old Agroforestry Systems (AFS) and in a 10-Year-Old Second Growth Via Fine Litter and Green Mal}ltre (From PlUnings)" P K Ca Mg Treatmentb CO'l1Jponents N ASI
t:
./Litter Prunings Total
36.8 (54) 16.8(46) 53.6
2.35 (72) 0.94 (28) 3.29
5.76 (48) 6.23 (52) 12.0
32.7 (92) 2.87 (8) 35.6
8.64 (80) 2.11 (20) 10.8
Litter PlUnings Total
36.3 (60) 24.5 (40) 60.8
1.90 (59) 1.33 (41) 3.23
5.01 (37) 8.57 (63) 13.6
28.7 (84) 5.31 (16) 34.0
8.58 (45) 10.8 (55) 19.4
Litter = Total
64.1
3.82
12.6
45.2
13.6
AS2
CAP
"Nutrient input values are given in kg ha-1. Values in parentheses represent the relative contribution (%) of each source (fine litter and green manure) to the total nutrient input in the AFS. Source is Gallardo-Ordillola [1999]. bSix-year-old agroforestIy: SI (palm-based) andAS2 (multistrata) and CAP: lO-year-old second growth.
parts ofthe region in the next few years [Smeraldi and May, 2008]. 5. CARBON SEQUESTRATION POTENTIAL IN AGROFORESTRY SYSTEMS CONVERTED FROM DEGRADED PASTURE Agroforestly is a possible option not only for carbon sequestration but also to increase the value ofpreviously cleared forest land in the humid tropics [Fujisaka and White, 1998]. Wherever agroforestly succeeds in maintaining soil fertility at a satisfactory level and increases farmers' incomes, additional clearing of primmy forest and accompanying carbon emissions are drastically reduced. When established on degraded soils, timber and tree crops in these systems sequester carbon in the biomass and soil and also provide firewood and charcoal as offsets for fossil fuel. On the other hand, when AFS or tree crop plantations are established on previously cleared fallow or secondmy forest land, carbon is released from the fallow vegetation that would have accumulated carbon in biomass and litter. Instead, establishment of pioneer trees, timber trees, annual crops, and sometimes cover crops affect the ecosystem C budget through soil management, feliilizer application, and suppression of spontaneous vegetation through weeding [Schroth et al., 2002]. Few data are available on the rate of carbon sequesh'ation throughagroforestly in the Amazon. A mean rate of2.7 Mg C ha- 1 a-lover 25-30 years is supported by the literature elsewhere, in which the values vmy fi'om 0.5 to 3.8 Mg C ha- I a-I. Woomer et al. [1999], measuring total system carbon in chronosequences in Brazilian Amazonia (Rondonia and Acre states), Cameroon, Indonesia, and PelU, reported that AFS sequestered about 3.3 Mg C ha- I a-I in soils and
vegetation. Sampson et al. [2000, p. 199] mentioned a range of 0.5 to 1.8 Mg C ha- 1 a-I of carbon accumulation for agroforest management in the tropics. McCajJeTJi et al. [2002] found that AFS based on native flUits and palms, planted on a severely degraded pasture in the Central Amazon accumulated up to 33 Mg C ha- I in the aboveground biomass after 12 years of manageqlent. Biomass in degraded pasture was 9 MgCha- 1, indicating anetC uptakeof2MgC ha- 1 a-I. For the same systems, Rr,JI1don et al. [2000] reported soil carbon stocks (to 1 m depth) under AFS of120 Mg C ha- I compared with the degraded pasture soil which stored 110 Mg C ha- I. This resulted in soil carbon accumulation rates of 0.83 Mg C ha- I a-I. Schroth et al. [2002] reported that trees planted as monocultures accumulated carbon at lower rates, i.e., 1.0 Mg ha- I a-I for "CitlUs", 1.3 Mg ha- 1 a-I for "cupuayu" (T. grandijlorum), and2.5 Mgha- I a-I forlUbberh'ees (Hevea brasiliensis). However, multistrata agroforestly favors carbon sequestration more than monocultures. A fast-growing system in central Amazonia accumulated 3.8 Mg ha- I a-I of carbon in the FF treatment (as described in section 3.4) and 3.0 Mg ha- 1 a-I of carbon under LF (only 30% of the full levels of fertilizer and lime). These high rates compared to most monocultures were due to a relatively high tree density and to the association of the smaller tl'ee crops, such as "cupuayu" and peach palm (B. gasipaes) for heart-of-palm production, with larger and faster-growing trees, i.e., lUbber and Brazil nut trees (B. excelsa). Schroth and colleagues also observed that in all of the investigated plantation systems, there was more than twice as much carbon in the soil organic matter than in the biomass and litter combined. Changes in the soil organic-matter stocks could, therefore, be of clUcial importance to evaluate the effect of land use transformations on the carbon budget.
LUIZAO ET AL. 324
325
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
neous second growth ("capoeira"). After a fa~low peri?d, However, Schroth and colleagues observed n? effects ofveg- a new crop or pasture can be established, agam followmg etation types and plant species on the orgamc-matter stocks slash-and-burn for clearing the land. In the abandoned a~'eas, of the soil ptofile to 2-m depth, although carbon conte~lt of the development of second-growth biomass and plant dlv.erthe topsoil was affected, being stored more s.uperficrall~. sity will depend on the previous use of the area (clean.ng These authors proposed two possible explanatI~ns for thIS methods, intensity of use, and management) and on the SIze trend. First the conversion of primary forest to dIfferent tree of the area cultivated [Uhl et al., 1988; Moran et al., ~OOO; crop plant:tions may have affected the distribution of carbon Mesquita et al., 2001]. However, even an area subJ.e~ted in the soil, but not its total quantity. Such ch~nges m~y occur to only moderate use may only recover partly ~he ong~nal through an altered distribution 'Of root mass m th~ s.oll profil~ plant diversity and biomass, even after long penods of tI~e or through differences in the abundance and actiVIty of ~UI [Moran et al., 2000). A recent s~~y fro~ the Muse.u EmllIO rowing soil fauna among vegetation types and plant speCIes. Goeldi (Belem, Pal'll) in assocIatIOn WIth the L~Ige-Scale Second, the total carbon stock in the soil to 2-m dept~ may Biosphere-Atmosphere Experiment in Amazoma ~LB.A) be less sensitive than the carbon content of the t?psor! as a Project, in the Bragantine Zone (the oldest colomzatIOn measure for soil organic matter loss over a relatively short zone in Amazonia), showed that 70-year-old second gro",:,th time period [Schroth et al., 2002). The a~thors als~, observe~ had recovered only 20-35% of the number of tree. s~ecles that the tree crops with low litter quahty (e.~., cupua 9u present in the forest control and 50-60~o o~ the ongmal C and Brazil nut) restored and maintained orgalllc ~atter lev- stocks (I.C.G. Vieira, personal commumcatIOn, 200~). Anels in the topsoil comparable to those i~ ~he p~nnary for- other study done in the same region showed that the llltrogen est, even when they were grown in association w~th tree and (N) cycle takes 70 years to recover to a condition similar cover crops that produced easily decomposable htter. Whe~ to that of primary forest [Da,:idson et al., 2007). Th~s, o~e integrated into multistrata AFS, such tree crops reduc~ soll can suppose that cycles of organic matter and ~utnents.m organic matter loss. Thus, these systems must be estabhsh~d these second growths are also distinct from those m the ongon sites with low standing biomass (e.g., degraded pastu~es inal forest and that the pace and dynamics of tree growth or other degraded soils) while preserving vigorously growmg are affected by shortages of key nutrients [Markewitz et al., secondary forests. However, Amazo~ian pasture~ are often 2004). A fertilizer experiment in 6-year-old second growth prone to topsoil compaction and erOSIOn [Feamslde, 1985], in Paragominas showed that additions of Nand of.N plus. P with adverse effects on growth and yields of tree crops. produced an increase in the biomass of,:oody specIes, whlle Considering the complexities involved, .it is ~pparent that the addition of P increased only the bIOmass of herbs and the estimate of6.5 to 49.4 Tg C a-I potential gam from congrasses. Thus, regenerating forest b~omass is limite~ by N ~n version of degraded pasture to agroforestry ",:ithin a. period old pasturelands that have been subjected to successl:,e bUI~ of 25-30 years is rather crude. This estimate IS obtar~ed by ings [Davidson et al., 2004a). In a feltilizing expenm~nt 111 multiplying 13 Mha of degraded pasture area by the soll pl~~ the Bragantina region, it was shown that second~l~ regrowth biomass carbon accumulation range of 0.5 to 3.8 Mg C h~ was mainly P-limited and, to a lesser extent, N-hmlted [Geha-I. This estimate is based on the assumption that the entire ring et al., 1999]. In a chronosequence of second growth ~p area under degraded pasture is conveited to agroforestly. Fur- to 14 years old in cenh'al Amazonia, it was fo~nd that solls thermore, soil organic carbon sequestration in. agroforestr: up to 45 cm depth accumulated N but lost avarlable P, sugis affected by species. Though the total area m Amazoma gesting possible limitations by P (and Ca) for second growth dedicated to agroforestly is relatively small co~p~re~ to [Feldpausch et al., 2004]. . other management systems, and there are severe hmltatIOns In order to prevent new cycles of deforestatIOn, shortthat restrict the potential of converting degr~ded pas~res lived cropping and/or pasture systems an,d. co~sequent land into AFS [Fearnside, 1995], agroforests remam as a Viable abandomnent, alternative land use, and utIhzatIOn of alr~a~y alternative land use for the region, and the area under these deforested areas have been proposed in recent years [VIeIra systems has been steadily increasing since the mid-1980s. et al., 1993; Feamside, 1997a). These large areas of ab~n 6. PROMISING ALTERNATIVES TO SLASH-AND-BURN AGRICULTURE Lack of proper knowledge of soils and of crop or p~sture management generally leads to short-lived crops (typIcally 2-3 years) or pasture (mostly from 5 to 10 'years) systems, after which they are abandoned and cololllzed by sponta-
doned land could be re-utilized through different altern~tlve techniques: (1) emichment of second growth ~espec~allY with introduction of valuable timber and/or fruIt s~ecIes), (2) creation of new cropping systems using plant bIOmass without burning, and (3) use of AFS. . Use of alternatives to slash-and-burn agriculture m the Brazilian Amazonia is relatively infrequent, and many
times, these practices are not known outside a specific site 2008). Even though strong increases in CH4 emissions were where they are appliegrHowever, they do exist and could found in the first 2 years after mulching, the C02-equivalent be adopted more widely if properly advertised and tested. emissions calculated for the entire crop system were at least For instance, the s9ihers ("caboclos") in the Arroz Cru set- five times lower than in the traditional slash-and-burn process tlement (XillgU l:~ion), .coming from Xingu river islands, [Davidson et al., 2008]. Although a moderate NPK fertilizareceived plots of 100 ha for cultivation: they used 4-year tion was required in the chop-and-mulch system for the first fallows for mixed plantations, obtained good and contin- year, available P and mineral N (~ and N03) increased ued production of manioc, corn, beans, and bananas [Silva- significantly by the end of the second year, contrasting with Forsberg and Feamside, 1995]. This could be a potentially a sharp decline in the slash-and-burn plots. These results suitable system for cultivating second-growth areas, with no indicate that, if proper incentives and technical supervision are given in Amazonia, it is possible to prevent, or at least need for new forest slash-and-burn. In Central America, the "fi'ijol tapado" is a slash/mulch reduce, the use offire by farmers during land preparation for bean (Phaseolus vulgaris L.) production system that has cropping. This chop-and-mulch experiment has now been been practiced since pre-Columbian times in the region now replicated in the states of Acre, Amapa, Amazonas, Maraknown as Costa Rica [Melendez et al., 1999). Under slash/ nhao, Rondonia, and Roraima, using annual, semiperennial mulch systems, crops are sown in direct association with (black pepper, passion fiuit) species, and pastures. in Sitli slashed vegetative mulches. They are still common Also in Brazilian Amazonia, as part of the "Crop producthroughout humid tropical Central America and nOlthern tion without burning" Project (Viver, Produzir e Preservar South America, where high rainfall prevents burning and Foundation), at the end of the 1990s, small fanners used induces rapid decomposition and nutrient mineralization. diversified cropping systems mixing annual and perennial By providing continuous ground cover, they are particularly species, adopting principles of the agroecological transition well adapted to the steep slopes that account for most of the now included in the Pro-Ambiente Program in Brazil [Sa land area of Central America. Traditionally, the "frijol ta- et al., 2007]. Well-established farmers with sufficient ecopado" system allowed a fallow period of 3 years or longer, nomic resources may also be able to adopt the "Bragantine but presently, its fallow period can be as shOlt as 9 months, System" of continuous cropping of varied and mixed speor even shOlter, raising serious concerns about its sustain- cies under rotation llnd consOltium systems, using mulchability [Melendez et al., 1999]. ing and keeping the: soil cultivated and covered throughout In northeast Para, Brazil, the Tipitamba Project (Embrapa- the year [Cravo et al., 2005). This management system reAmazonia OrientaVSHIFTILBA Projects) used early fallow quires liming and fertilization (P and micronutrients), at least emichment and a chop-and-mulch procedure, eliminating temporarily. Direct and fast recovelY of soil properties in abandoned the use of fire for land preparation, then cultivating annual crops successfully for several years [Sa et al., 2007]. The ex- or degraded lands, used for agriculture and/or pastures in periment consisted of the initial emichment of smallholder Amazonia, has been tried in some experiments in Brazil. For fallows, on the occasion of land abandonment, with fast- instance, in central Amazonia, the use of two different cover growing legumes (Acacia sp. or other tree species) in order crops was tested for rehabilitating soil structure and functo increase and maintain biomass production and to diversify tion. The leguminous species P. phaseoloides and D. ovalithe chemical quality of the organic material produced. After folium were planted as cover crops together with an Elaeis a short period of fallow (reduced to approximately 3 years guineensis (oil palm) plantation, in order to reverse the cominstead of the minimum 7 years that is usual in the region), paction effects of both mechanized and manual deforestaland preparation for new cultivation without use of fire was tion for land clearing [Grimaldi et al., 1993]. After 2 years, tested (meaning, fire was not used in the process). Plant bio- the soils cultivated with oil palm and a cover of Desmodium mass of the improved fallow was cut and ground using a recovered a bimodal pattern of soil-pore distribution similar large grinder coupled to a tractor, and the resulting organic to that under the intact forest, with pores >0.1 /lm in the material was deposited on the soil surface as mulch, followed plots deforested manually, but not in the plots deforested by planting beans, maize, manioc, passion fruit, etc. Despite with machinery. When the soil cover was Pueraria, in both some problems (and lack of specific incentives to farmers), .situations, soil macroporosity was recovered after 2 years the system was found to be economically more efficient than due to strong root development plus the biological activity the traditional slash-and-burn procedure after 5 years. This favored by Pueraria. coincided with the period when the soil fauna (especially the In 'Amazonia, some traditional AFS have been shown to macroinvertebrates) is recomposed to levels similar to those be sustainable as parts of indigenous [Hecht et al., 1989] found in nearby forests (T. Sa, personal communication, and smallholder [de long, 1996] agriculture and are now
326
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
considered as a promising and sustainable land use in Amazonia that is especially suitable fordegraded areas [Fernandes et al., 1997]. Multistrata AFS are expected to be more sustainable lanll uses than are annual crops because they include various long-lived woody species and have a larger canopy and a more complete ground cover, thus limiting nutrient losses through runoff and leaching, being as efficient as or even better than natural fallow for soil rehabilitation [Young, 1997]. They can maintain soil organic matter and biological activity at satisfactory levels for soil fertility and to restore degraded soil [Ewel, 1986; Young, 1997]. Depending on the type and diversity of the new ecosystem, the soil microbial biomass, which shows surprising capacity for recovelY, may return to previous forest levels within a few years [Woomer et al., 1999; Barros et al., 2003]. In Manaus, soil microbial biomass was measured in four different 5-year-old AFS [Tapia-Coral et al., 1999]. All the AFS had from amounts of microbial biomass 1.3 to 3.1 higher than the 10-year-old second-growth system used as a control, and the highest values were found in the most diverse and dense multistrata system. In spite of some limits to expanding AFS to large productive systems, mainly due to market constraints [Fearnside, 1995], they seem to be good alternatives for smallholders, as illustrated by some successful experiments carried out in the region. In Brazil, the RECA system in Rondonia is an example of success, after the fmmers overcame the market constraints on selling their crops. The Manaus experiment carried out by Embrapa/CPAA, using four different fonnulations of AFS, has also yielded good results and helped to understand processes involved as well as to establish useful principles to be applied in other AFS [Fernandes et al., 1997; LuizflO et al., 2006]. The evolution of the agroforestry system in Manaus showed the following: (l) In the first 2 years, the systems with improved pashlres, with a fast and efficient establishment of soil cover (by a mixhlre of B. humidicola grass and the leguminous cover crop D. ovalifolium) allowed a soil organic cover and a better recovelY of soil fauna diversity [Barros et al., 2001]. (2) After 4 years, there was a decrease in concentrations of all soil cations due to nutrient expOlt through crops, and tree growth in the agroforestry system. (3) After 5 years, the most diversified agroforestly system had higher diversity of taxonomic groups and higher biomass of soil macroinvertebrates, which were also related to the quantity and quality (C/N, C/P, NIP, P) of litter produced and deposited on the soil surface [Tapia-Coral et al., 1999]. (4) After, 6-7 years, the agroforestry system composed of timber and fruit trees, plus a leguminous green hedge, produced fine litterfall equivalent to only 25-30% of the total litterfall in the lO-year-old second growth (taken
as a control), but with higher nutrient contents. Together with the green manures from the G. sepium hedge and the 1. edulis planted in the AFS, this input produced a balance of nutrients in the agroforestly system (see Table 4). (5) After 10 years, the agroforestly system presented similar or higher C and nutrient stocks (highly variable, depending upon the composition of tree species in the agroforestly system) than the second growth, indicating the high potential of AFS as C sink and for recovering the ecosystem nutrient recycling mechanisms. (6) After 10 years, the agroforestly system also had better soil structure (especially the macroporosity, produced by soil macroinveltebrates and roots and essential for water movement and availability in the soil) than the second growth, indicating the recovelY of another ecosystem service in the agroforestly system: the water recycling in the system [Cortes-Tan'a, 2003]. Among the tree species in the agroforestly system, 1. edulis had the highest biomass of soil macroinvertebrates per hectare and the best soil macroporosity in the soil directly influenced by the trees, probably because of the high-quality litter, which is associated with the production of highly energetic root and branch exudates. (7) After 12 years, the agrofbrestly system had soil P and cation concentrations that were higher than second-growth soils, which can at least partly be attributed to better litter quality and pruning inputs to the soil (Table 5). Forest plantations with native tree species are still surprisingly rare in Brazilian Amazonia, since they represent a natural vocation for areas that have been deforested in the region. Such plantations could be used for (1) emichment of mahlre or secondmy forest with valuable timber species [Yared, 1996], (2) timber or charcoal production, (3) rehabilitating degraded lands [Higuchi et al., 1998] (Jacaranda Project). Only a few examples are known for all three cases, and generally, they lack evaluations of soil nutrients. Forest plantation in Amazonia should have a mixed species approach in order to prevent problems with diseases, which easily spread through a monoculture, or nutrient shortages in the soil (all trees demanding the same nutrient at the same time and position in soil), or problems with the timing for the nutrient-release rates. One example is a Brazil nut plantation in Itacoatiara, Amazonas, which produces slowly decomposing litter (Table 6). On the other hand, its litter covers the soil surface, thereby improving moisture and other soil conditions. A few experiments have been done on the rehabilitation of mining sites in Amazonia, such as the recent plantation of climax and pioneer native species at the oil extraction site in Urucu, Amazonas (CT-Petro experiments). Another example is from Trombetas, Para, where diversified forest plantations were installed to rehabilitate soils mined for bauxite. To rehabilitate these badly damaged sites, reforestation pro-
LUIZAO ET AL.
327
Table 5. Soil Surface (G-IO cm Depth) Characteristics in a 12-Year-OldAFS Based on Palms and Fruit Trees and iii Improved (Mixed) Pasture and Second Growth" AFS
pH'? C (%) N (%) P (mg kg-I) K (mg kg-I) 1 Mn(mgkg- )
Palms
Fmit trees
Mixed Pasture
Secondmy Forest
4.5a±0.14 3.1 ± 0.9 0.2 ± 0.03 58.2 a ± 26.5 147±126 6.5±2.1
4.4ab±0.08 2.8 ± 0.5 0.2 ± 0.Q3 68.8 a ± 41.4 152±140 7.1±2.4
4.5a±0.11 3.1 ± 0.7 0.21 ± 0.Q3 54.9 a ± 27.4 151±122 5.9±1.8
4.3b±0.1 3.0 ± 0.7 0.19 ± 0.02 23.5 b ± 10.6 119±122 4.5±1.0
"Means followed by different letters indicate significant differences between treatments (analysis of variance,p < 0.05); n = 3. Source is Silva [2005].
grams using native tree species have been implemented by the "Systems of Production and Studies of Degraded Land, Reconstitution and Rehabilitation of Forest Ecosystems" project [UNESCOlEconomic European Community (EEC)I Deutsche Gesellschaft fur Technische Zusammenarbeit (GTZ)/Minerac;ao Rio do Norte (MRN)]. At the Saraca Mine in the eastern part of Para, reforestation treatments (topsoil replacement versus nO topsoil replacement; nahlral regeneration versus reforestation) were compared in treatlnent plots including a l-year-old reforested plot where no topsoil replacement occurred, two 11-year-old plots where the topsoil had been replaced (one reforested and the other with vegetation allowed to regenerate naturally), and a primmy forest that was used as a control. The l-year-old plantations without topsoil replacement always had velY low concentrations of nutrients in the soil, palticularly for P and K (Table 7). Soil exchangeable Ca and Mg were higher in the two 11year-old treatments [Costa et al., 2002]. Thus, reforestation with native tree species (and to a lesser extent, the nahlrally regenerated forest) allows the recovelY of the soil litter cover and microclimate within a few years. This, in turn, leads to the recovery of soil microbial activity and the decomposition process. Topsoil nutrient availability is improved in one to two decades due to the mobilization of
nutrients from the soil, particularly Ca and Mg, promoted by planted or pioneer species. The organic carbon content of the soil is the slowest to recover and may therefore act as the ultimate indicator ofthe rehabilitation process. In mining sites, topsoil replacement prior to planting native tree species can speed up the soil rehabilitation process [Costa et al., 2002]. Studies on natural regeneration in the central Amazonia have shown that pioneer species, such as Cecropia sp. and some planted agroforestry species, produce leaf litter that is rich in base cations, particularly Ca [Lucas et al., 1993; Gallardo-Ordinola, 11999; Tapia-Coral et al., 2005]. These species appear to have a high capacity to extract Ca and Mg from soil compounqs, incorporating these nutrients as biomass and releasing them to the soil surface in a more available fonn as plant litter. Secondmy forest associated with agriculture in Amazonia follows a clear pattern of development. During pastlJre use, burning and weeding delay succession, but the forest begins to regenerate once the field is abandoned. Secondmy vegetation is established through four main processes: regeneration of remnant individuals, gennination fi'om the soil seed bank, sprouting from cut or crushed roots and stems, and dispersal and migration of seeds from other areas [Tucker et al., 1998]. Variations in the speed of forest regrowth are evident
Table 6. Litter Production and Decomposition Rates in Managed Ecosystems in the Brazilian Amazonia Location Manaus
Capitao Poc;:o
Ecosystem 10-Year-old Bertholletia excelsa plantations 5-Year-old B. excelsa plantations Forest (control) AgroforestIy I AgroforestIy 2 3-Year-old fallow field
Litter Production (Mg ha- 1)
Litter Decomposition (half-life) (days)
Authors
1.3
518
Kato [1995]
0.67 7.G-1O.0 2.0 2.3 5.0
518 83 25-116 25-116
Gallardo-Ordinola [1999] Dantas and Phillipson [1989]
328
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
LUIZAO ET AL.
Table 7. Soil Nutrients From the Upper Layer (0-10 cm) in 1-Year-01d Reforested Plots With and Without Topsoil Replacement, in 11-Year-Old Natural Regeneration, in 11-Year-Old Reforested Plots, and in Primary Forest at the Saraca Mine, POlio Trombetas, Para in the Wet Seasona Soil Nutrients
Treatmentb RWTS1 RWTS1 NR11 REF 11 PF
Total N (g kg-I) 0.80 b ± 0.20 0.23 b ± 0.11 1.02 b ± 0.31 2,33 a ± 0.20 2.11 a ± 0.12
P (mg dm-3 Mehlich-1) 1.38 a ± 0.88 b ± 1.50 a ± 1.68 a ± 1.46 a ±
0.08 0.001 0.03 0.07 0,09
K+ (cmolc dm-3)
Ca++ (cmo1 c dm-3)
0,09 abc ± 0,02 0.01 ± 0.01 0.11 ab ± 0.02 0.16 a ± 0.01 0.06 bc ± 0.03
0.46 c ± 0.07 0.03 c ± 0.01 1.53 ab ± 0,35 2,34 a ± 0.24 0.07 c ± 0.23
Mg++ (cmo1 c dm-3) 0.16 cd ± 0,01 d ± 0.38 ab ± 0.49 a ± 0.11 cd ±
0,03 0,01 0,08 0.04 0.00
aValues are means ± SE (n = 3), Source is Costa et ai, [2002]. Means followed by similar letters are not significantly different (p < 0.05) according to the Tukey test. bAbbreviations are RWTS 1, plots with topsoil replacement; (RWTH1, plots without topsoil replacement; NR11, an 11-year-old natural regeneration; REF11, 11-year-old reforested plots; and PF, primary forest.
across regions and along a soil-fertility gradient in Brazilian Amazonia. The rate of forest succession is determined by several factors. Original floristic composition, neighboring vegetation, and soil fertility and texture may affect regrowth. In addition, farmers' land use decisions (such as clearing size, clearing procedures, crops planted, frequency and duration of use) influence tree establishment and path of secondmy succession [Moran et al., 2000; Mesquita et al., 2001]. At the regional scale, soil fertility and land use history are the critical factors influencing forest regrowth [Tucker et al., 1998]. In Amazonia, secondary forests have high rates ofregeneration following slash-and-bum agriculture but slower regeneration after abandonment of degraded pasture [Fearnside and GUimariies, 1996; Mesquita et al., 2001]. Brown and Lugo [1990] reported that abandoned agricultural lands reverting to forests accumulated carbon at rates proportional to the initial forest biomass. Rates ranged from about 1.5 Mg C ha- I a-I in forests with initial b~omass of <100 Mg C ha- 1 to about 5.5 Mg C ha- I a-I for forests with biomass of> 190 Mg C ha- I. Woomer et al. [1999] observed a rate of 6.2 ± 1.3 Mg C ha- I a-I of carbon sequestration in secondalY forest regrowth in agricultural fallows in Brazilian Amazonia (Rondonia and Acre). Sampson et al. [2000, p. 200] suggested a range ofcarbon accumulation from 3.1 to 4.6 Mg C ha- I a-I for tropical regions over 40 years. Schroth et al. [2002] repOlied that secondary forest on an infertile upland soil in Manaus, central Amazonia, accumulated carbon in above- and belowground biomass and litter at a rate of about 4 Mg ha- I a-I. The rate of accumulation in aboveground biomass reported by Nepstad et al. [2001] ranged from 2.5 to 5 Mg C ha- I a-I for a 20-year-old secondmy forest in Para, eastem Amazonia. In Paragominas, Para, a
19-year-old secondary forest accumulated 20% of the forest biomass at a rate of9 kg C ha- I a-I [Markewitz et ai" 2004]. In Manaus, Feldpausch et al. [2004] reported C accumulation of 128 Mg C ha- I for a'12-year-old secondary forest dominated by Vismia spp. regenerated on an abandoned and severely degraded pasture. However, soil degradation under typically managed pasture can severely slow the subsequent regeneration of secondaly forest [Uhl et al., 1988; Fearnside and Guimariies, 1996]. Potential carbon sequestration when degraded pastures in the Brazilian Amazon are abandoned for secondary forest regrowth, were calculated by multiplying 13 Mha (degraded pasture area) by the accumulation range of 1.5 to 5.5 Mg C ha- I a-lor 19.5 to 71.5 Tg C a-I. This potential for carbon sequestration accounts for carbon in soil plus aboveground biomass. As of 1986, secondmy forests covered 30% of the area of the Brazilian Amazon that had been cleared by the 1980s [Houghton et al., 2000]. It should be noted that less secondary forest is present in the region now than was the case in 1986, the year of Landsat satellite imagery studied by D. S~ole that is believed to provide the basis of the assumption of Houghton et al. [2000] that 30% of the region was under secondary forest. As of 2002, Landsat satellite imagery indicated that 16.1 million ha of secondmy forest were present in Brazilian Amazonia or 19% of the deforested area [Neeff et ai" 2006]. If this 16.1 million ha area were to remain abandoned, it could sequester carbon in soil and plant biomass. Thus, multiplying the same accumulation range values (1.5 to 5.5 Mg C ha- I a-I), there would be an additional sequestration of 24 to 86 Tg C a-I. Therefore, potential soil + biomass carbon sequestration in the Brazilian Amazon due to secondaly forest regrowth is between 43.5 and 157.5 Tg C a-I.
By fixing carbon in biomass and gradually restoring soil physical and chemical/propeliies, forests that develop on abandoned land als@/ counteract many of the deleterious impacts of forest ~6nversion to agriculture and cattle pasture. These fore;l.i'§ play an important role in the regional carbon budget, as they re-assimilate part of the carbon that was released upon cutting and buming of the original forest vegetation. Secondary forests allow the expansion of native plant and animal populations from mature forest remnants back into agricultural landscapes [Nepstad et al., 2001]; they restore hydrological functions performed by mature forests and reduce the flammability of agricultural landscapes; they also transfer nutrients from the soil to living biomass, thereby reducing the potential losses of nutrients from the land through leaching and erosion. In Paragominas, Para, degraded and managed pastures, as well as 19-year-old secondary forest showed similar or higher contents of nutrients, especially basic cations (Ca, Mg) in relation to the forest [Markewitz et al., 2004]. In part, this can be attributed to the long-term effect of biomass buming and the initial release of nutrients such as Ca from the ashes. The basic cations K, Ca, and Mg were mostly retained in soil after 19 years of second growth, although soil was slowly retuming to an acidic condition [Markewitz et al., 2004]. Reaccumulation of macronutrients in vegetation in the 19-year-old second growth was equivalent to 20% of the N, 21% of the P, 42% of the K, 50% of the Ca, and 27% of the Mg of the original forest. In the degraded pastures, the reaccumulation was much smaller, as expected: 2% of the N, 4% of the P, 15% of the K, 11% of the Ca, and 6% of the Mg of the original forest. Both managed and degraded pastures only possessed about 2% of original N of the forest biomass, implying that N was not transferred to the soil by land use conversion; rather, the N was probably lost from the ecosystem through fire or by leaching to the deep soil solution or to streams [Markewitz et al., 2004]. Thus, plant demands are probably supplied by N mineralization from soil organic N, but in second growth, the rates of actively cycling N, including soil organic N, are low. Lower N in soil solutions in secondmy forests indicates low cycling of available forms of N and possible N limitation [Davidson et al., 2004a]. Phosphorus (Mehlich-III extraction) in both pastures and the 19-year-old second growth was very low «1 Ilg g-I), and only 8.8 kg ha- I of P was found in the second~growth biomass, representing an accumulation of only 0.25 kg ha- I a-I from bioavailable P. This also indicates that losses of P from the ecosystem though intemal cycling of P occurs predominantly through litterfall and grass tumover, which recycle nutrients annually in a way that conservatives ecosystem stocks [Markewitz et al., 2004].
329
7. TERRA PRETA SOILS AND CONTEMPORARY NUTRIENT MANAGEMENT IN AMAZONIA Patches ofsoil with varying spatial extent show high fertility in an otherwise comparatively infertile soilscape. These so-called "terras pretas do indio" or Amazonian Dark Earths are man-made soils that occur throughout the Amazon Basin, but with greater concentrations in the middle Amazon and along the larger tributaries [Sombroek et aI., 2003]. The total contribution to Amazonian soils and particularly agricultural soils in the Amazon Basin is unclear and could be as high as several percent but more likely less than 0.1 % or 0.3% [Sombroek et al., 2003]. In any event, the proportion of agricultural soils that are terra preta is small and is rather local in importance. However, such patches of terra preta are highly valued by local fanners due to their superior productivity [Lehmann et ai" 2003]. The relevance of terra preta currently lies less in its quantitative importance for Amazonian agriculture than in the lessons it may provide for recreation of terra preta. A deeper understanding of the mechanisms by which terra preta has maintained its high feliility over millennia is required to develop meaningful recommendations for modem agriculture. Also, having a high concentration of stable C, it could become an impOliant sink and stock of a~mospheric C if its area is augmented to 5-10% of Amazonia [Sombroek et al., 2003]. Our understanding, to date, suggests that one of the most important aspects Of the fertility of terra preta is its high black carbon (biochar or charcoal) content [Glaser et al., 2001]. The black C increases cation exchange capacity through greater surface area and charge density [Liang et al., 2006]. However, this greater ability of tena preta to retain cations does not, by itself, explain the fact that total nutrient contents are often greater by orders of magnitude than in other soil types, principally for P and Ca [Lehmann et al., 2003]. The most likely source of large amounts ofP and Ca is fish residues from the preparation of meals [Lima, 2001; Lehmann et al., 2004]. Other refuse materials may also play a role, and their importance will most likely change significantly as a function of site characteristics and the history of human habitation. While soil management with fish residues may be feasible in certain situations where fish wastes are available in sufficient quantities, the application of biochar has a much broader applicability. Indeed, research on biochar application to soil has intensified over the past years and provided clear evidence for the potential to improve crop productivity in highly weathered tropical soil [Lehmann and Rondon, 2006]. Biochar can be applied to field crops on an annual basis with as little as a few tons per hectare as well as only once in larger amounts of several tens of tons to recapitalize
330
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
soil functions, Such applications have been done using the broadcast method or by application directly to the root zone. In addition, tree crops in plantations or in AFS can receive biochar pl~ed in the planting hole at the time of site establishment or placed around the stem after planting. Whether or not biochar soil management will be able to have a major impact on agricultural production in the Amazon depends not only on the refinement of a biochar product but also on the biochar production itself. Lehmann and Rondon [2006] demonstrated that even relatively young secondaly forests in the Amazon provide sufficient quantities of biomass to produce biochar in amounts that significantly improve crop productivity. Therefore, charring the biomass instead of buming it during a shifting cultivation cycle would provide the quantities of biochar that are necessalY to transform a soil with low cation exchange capacity into a soil with a significantly improved ability to retain cations. Over long periods of time as shown by terra preta soils, approximately a doubling of cation exchange capacity may be achieved [Liang et aI" 2006]. However, other factors constraining adoption of biochar application at a larger scale have not been sufficiently taken into account up to now, The work load associated with charring instead ofbuming the slashed biomass may constitute a significant obstacle. Even though most farmers are familiar with the techniques of making charcoal or have access to the technology, they may find the time or financial expense too large to engage in biochar production. The altemative of marketing as a fuel any charcoal produced provides immediate retums. The full financial benefit ofbiochar has not been sufficiently demonstrated to farmers to build confidence in the long-term retums, as would be needed to justify the required short-term investments. 8. LIMITS TO THE INTENSIFICATION OF AGRICULTURE AND RANCHING Severe limits restrain both the intensification of agriculture and ranching uses and the scale to which these land uses can be expanded [Fearnside, 1997a]. In addition to agronomic limits on per-hectare yields, physical resource limits, such as phosphate deposits, restrain land uses that depend on these inputs. Amazonian soils are poor in phosphoms, and the vast extent of Amazonia means that convelting these areas to land uses that require feltilization with phosphates would quickly exceed existing deposits, both within Brazil and globally [Fearnside, 1998]. Market limits restrain the potential expansion of some of the less-destructive production systems, such as agroforestly [Fearnside, 1995], but provide little restraint on the most destmctive land uses, such as cattle pasture [Fearnside, 2005]. In addition, convelting
tropical forest to these uses carries environmental risks that make policies leading to forest loss unwise as a strategy for developing the region. In summalY, degradation of soil and forest is not inevitable in Amazonia, Cattle pastures, which dominate deforested landscapes in Brazilian Amazonia, can have increased soil organic matter if the best management techniques are used, including certain soil amendments. Soil organic matter sustains the levels of a series of nutrients, leading to greater plant productivity. Agricultural soils can be improved based on lessons leamed from the region's histOly prior to European contact: indigenous populations left many patches of rich "dark earths" (tena preta), which owe their high carbon content and fertility, in part, to high content of charcoal. Soil carbon and nutrient retention can be increased in modem agriculture by adding powdered charcoal to the soil, especially if combined with supply of nutrients through fertilization. AFS also represent a means of maintaining and increasing soil fertility because of they have greater capacity to cycle nutrients than do other land uses and because the soil surface is protected by a litter layer. The various means of maintaining soil fertility are subject to severe limitations that would prevent their being expanded to the vast areas that have already been deforested, let alone to the much larger areas that are still covered by standing forest but that could meet the same fate if deforestation trends continue unchecked. Some limiting factors can be faced through technological advances or by applying what is already known about Amazonian productive systems. Any successful production system in Amazonia needs to respect the nature of the region by always keeping the soil covered by forest and by maintaining its high biodiversity in order to remain productive for a long time. It is essential to recognize that some limiting factors cannot be overcome. Thus, other strategies should be adopted to benefit the Amazonian population. The limits to expansion of intensified land uses mean that fulther deforestation must be prevented and that development should emphasize the natural forest, which can maintain itself without outside inputs of nutrients. 9. ENVIRONMENTAL SERVICES AS A BASIS FOR DEVELOPMENT Maintenance ofsoil feltility in Amazonian managed systems is not a goal in itself, but rather a means of achieving ends such as supporting the region's human population in a sustainable way and maintaining the region's environmental services. The choices of systems to be implanted and managed will be key factors in determining the extent to which these larger goals are met. Cattle pasture, which is by far the most common land use in deforested areas in Brazilian Amazonia, is not only the
LUIZAo ET AL.
least sustainable, as managed in the region; it is also one ofthe worst in telms of maintaining the human population, Vety few employees are needeQ'to maintain fences and herd cattle on the large and medium )!anches that account for most Amazonian deforestation [Felrnside, 1983], Soybeans are grown in large properties with fuechanized agriculture that provides little employment [Fearnside, 2001]. The environmental services provided by Amazonian forests offer a far more valuable and sustainable basis for development than does the expansion of deforestation [Fearnside, 1996, 1997b, 2008a]. Tapping the enviromnental services of intact forest, including maintaining ecosystem carbon stocks in order to avoid global warming [Fearnside, 2008b], maintaining the water cycle [Fearnside, 2004] that supplies rainfall both to Amazonia and the remainder of Brazil, and maintaining biodiversity [Fearnside, 1999], must be recognized as sound strategies for the region. 10. CONCLUSIONS 1. Amazonian managed systems have velY different effects on soil feltility depending on the type of managed system, the way it is managed, and the initial stage of soil degradation. 2. Pasture management affects the organic matter content of the soil and thereby the soil's capacity to retain nutrients. Well-managed pastures lead to higher fertility than pastures that are degraded through the extensive management practices that predominate in Brazilian Amazonia today. Note that conversion of Amazon forest to pasture always implies a large loss of ecosystem carbon stocks due to the much higher biomass of the forest. 3. AFS can have an important role in maintaining and improving soil fertility and offer potentially sustainable livelihoods to small farmers in Amazonia. 4. The possibility of large-scale expansion of intensive pasture, agroforestly, or other uses with fertilizer inputs is subject to severe limits from available nutrient sources, especially phosphate deposits. AgroforestlY expansion is limited by markets for the products. These limits add to the evidence indicating the wisdom of halting further clearing of Amazonian forest, and of administering the already cleared area in ways that sustain production (a challenge that includes maintaining and improving soil fettility). The soils in deforested areas need to be used in a way that maintains the region's human population, a role that can be served by smallholder agroforestly but not by large cattle ranches or' agribusiness operations such as growing soybeans. 5. The limits to expansion ofintensified land uses mean that further deforestation must be prevented and that development should be based on maintaining the natural forest. Amazonian
331
forests provide environmental services that are more valuable and more sustainable as a foundation for the region's development than is the expansion of deforestation. REFERENCES Alfaia, S, S" G. A. Ribeiro, A. D, Nobre, R. C. Luizao, and F, J. Luizao (2004), Evaluation of soil fertility in smallholder agroforestry systems and pastures in western Amazonia, Agric, Ecosyst, Environ., 102, 409-414, Almeida, E., C. Sabogal and S. Brienza Jr, (2006), Recuperw;iio de Areas Alteradas na Amazonia Brasileira: Experiencias Locais, Lh;oes Aprendidas e Implica<;:oes para Polfticas P/lblicas, CIFOR, Belem, Brazil. Anderson, J. M" and T. Spencer (1991), Carbon, Nutrient and Water Balances of Tropical Rainforest Subject to Disturbance, 54 pp., MAB Digest No.7, UNESCO, Paris, France, ASB (Alternatives to Slash-and-Burn) (2002), Alternatives to Slash-and-Burn in Brazil-SumIlIaIJ! Report alld Synthesis of Phase II, edited by J. Lewis et aI., 100 pp" Nairobi, Kenya, Asner, G. P., M. Keller, R. Pereira Jr., J. C. Zweede, and J. N. M. Silva (2004a), Canopy damage and recovery after selective logging in Amazonia: Field and satellite studies, Eco!. Appl., 14, S280-S298. Asner, G, P" A. R. Townsend, M. M. C. Bustamante, G. B, Nardotto, and L. P, Olander (2004b), Pasture degradation in the central Amazon: Linking changes in carbon and nutrient cycling with remote sensing, Global Change BioI" 10, 844-862, Barbosa, R. I., and :P, M. Fearnside (2000), Erosao do solo na Amazonia: Estudd de caso na regiao do Apiall, Roraima, Brasil, Acta Amazonica, 30, 601-613. Barbosa, R, I., and P, M, Fearnside (2003), Burning of pasture in Amazonia: Short-term changes in soil carbon stocks, Braz, J. Ecol., 1-2, 11-16, Barros, E., P. Cunni, Y, Hallaire, A. Chauvel, and P. Lavelle (2001), The role of macrofauna in the transformation and reversibility of soil structure of an Oxisol in the process of forest to pasture conversion, Geodel'lna, 100, 193-213, Barros, E., A. Neves, E. Blanchart, E. Fernandes, E. Wandelli, and P. Lavelle (2003), Soil macrofauna community of Amazonian agroforesny systems, Pedobiologia, 47, 267-274, Bernoux, M" D, Arrouays, C. C. Cerri, and H. Bourennane (1998), Modeling vertical distribution of carbon in Oxisols of the western Brazilian Amazon, Soil Sci., 163, 941-951. BIONTE (1997), Relatorio Final Projeto BIONTE-Biomassa e Nutrientes na Floresta Tropical Umida, 365 pp., INPA/DFID, Manaus, Brazil. Bonde, T. A., B. T. Christensen, and C. C. Cerri (1992), Dynamics of soil organic matter as reflected by natural 13C abundance in particle size fractions of forested and cultivated oxisols, Soil BioI. Biochefll., 24, 275-277, Brinkmann, W. L. F, (1989), System propulsion of an Amazonian lowland forest: An outline, GeoJournal, 19, 369-380, Brown, S., and A. E, Lugo (1990), Tropical secondmy forests, J. Trop, Eco!., 6, 1-32,
332
LUIZAO ET AL.
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
integrated greenhouse gas assessment of an alternative to slashBuschbacher, R., C. Uhl, and A. S. Serrao (1988), Abandoned pasand-burn agriculture in eastern Amazonia, Global Change BioI., tures in eastern Amazonia. II. Nutrient stocks in the soil and veg14,998-1007. etation, J. Ecol., 76,682--699. de Jong, W. (1996), Swidden-fallow agroforestly in Amazonia: Di" Bustamante,~. M. C., M. Keller, and D. A da Silva (2009), Sources versity at close distance, Agrofor. Syst., 34, 277-290. and sinks of trace gases in Amazonia and the cerrado, Geophys. Desjardins, T., F. Andreux, B. Volkoff, and C. C. Cerri (1994), Monogr. Ser., doi:10.1029/2008GM000733, this volume. Organic carbon and 13C contents in soils and soil size-fractions, Cerri, C. C., B. Volkoff, and F. Andreux (1991), Nature and beand their changes due to deforestation and pasture installation in havior of organic matter in soils under natural forest, and after eastern Amazonia, Geoderma, 61, 103-118. deforestation, burning and cultivation, near Manaus, For. Ecol. Desjardins, T., et al. (2000), Degradadion des paturages amazoManage., 38, 247-257. niens, Etude et Gestion des Sols, 7,353-378. Cerri, C. c., M. Bernoux, B. J. Feigl, and M. C. Piccolo (1999), Carbon dynamics in forest and pasture soils of the Brazilian Dias-Filho, M. B. (2003), Degradar;iio de Pastagens: Processos, Causas e Estrategias de Recuperar;iio, Embrapa Amazonia OriAmazon, Workshop on Tropical Soils, Academia Brasileira de ental, Belem. Ciencias, Rio de Janeiro, Brazil, pp. 65-72. Dias-Filho, M. B., E. A Davidson, and C. J. R. Carvalho (2001), Cerri, C. E. P., K. Coleman, D. S. Jenkinson, M. Bernoux, R. L. Linking biogeochemical cycles to cattle pasture management Victoria, and C. C. Cerri (2003), Soil carbon dynamics at Nova and sustainability in the Amazon basin, The Biogeochemistl)! of Vida Ranch, Amazon, Brazil, Soil Sci. Soc. Am. J., 67, 1879the Amazon Basin, edited by M. E. McClain et aI., pp. 84-105, 1887. Oxford Univ. Press, New York. Chone, T., F. Andreux, J. C. Correa, B. Volkoff, and C. C. CelTi Didham, R. K. (1998), Altered leaf-litter decomposition rates in (1991), Changes in organic matter in an Oxiso1 from the central tropical forest fragments, Oecologia, 116, 397--406. Amazonian forest during eight years as pasture, determined by l3C composition, Diversity of Environmental Biogeochemistry, Ewel, J. J. (1986), Designing agricultural ecosystems for the humid tropics, Annu. Rev. Ecol. Syst., 17,245-271. edited by J. Berthelin, pp. 307--405, Elsevier, New York. Falesi,1. C. (1976), Ecossistema de Pastagem Cultivada na AmazoCochrane, T. T., and P. A. Sanchez (Eds.) (1982), Land Resources, nia Brasileira, Boletim Tecnico 1, 193 pp., Centro de Pesquisa Soils and Their Management in the Amazon Region: A State of Agropecuaria do Tr6pico Umido, Empresa Brasileira de Pesquisa Knowledge Report, Centro Internaciona1 de Agricultura Tropical Agropecuaria Belem, Brazil. (CIAT), Cali, Colombia. Fearnside, P. M. (1980), The effect of cattle pasture on soil fertilCortes-TalTa,1. L. (2003), Re1a<;oes entre os grupos funcionais da ity in the Brazilian Amazon: Consequences for beef production macrofauna e 0 volume dos macroporos do solo em sistemas sustainability, Trop. Ecol., 21,125-137. agroflorestais na Amazonia central, Disserta<;ao de Mestrado, Feamside, P. M. (1983), Development alternatives in the Brazilian 80 pp., INPA, Manaus, Brazil. Amazon: An ecological evaluation, Interciencia, 8, 65-78. Costa, E. S., F. J. Luizao, R. C. Luizao, and A Macmillan (2002), Litter layer and soil microbial biomass in reforested areas after bauxite Fearnside, P. M. (1985), Agriculture in Amazonia, Key Environments: Amazonia, edited by G. T. Prance and T. E. Lovejoy, mining in eastern Amazon, Tropical Plantation-Forests and Their pp. 393--418, Elsevier, New York. Soil-Litter System: Littel~ Biota and Soil-Nutrient Dynamics, edFearnside, P. M. (1986), Human Canying Capacity of the Brazilited by M. V. Reddy, pp. 127-145, Science, Enfield, N. H. ian Rainforest, Columbia Univ. Press, New York. Cravo, M. S., J. Corteletti, O. L. Nogueira, T. J. Smyth, andB. D. L. Souza (2005), Sistema Bragantino: Agricultura SustentGvelpara Fearnside, P. M. (1989), A ocupa<;ao humana de Rondonia: Impactos, limites e planejamento, in Relatbrio de Pesquisa 5, 76 pp., a Amazonia, Embrapa-Amazonia Oriental, Belem, Brazil. CNPq, Brasilia. Dantas, M., and J. Phillipson (1989), Litterfall and litter nutrient content in primary and secondary Amazonian' 'terra firme' rain Fearnside, P. M. (1995), Agroforestry in Brazil's Amazonian development policy: The role and limits of a potential use for deforest,J. Tl'Op. Ecol., 5, 27-36. graded lands, Brazilian Perspectives on Sustainable Development Davidson, E. A., C. J. R. Carvalho, 1. C. G. Vieira, R. O. Figueiofthe Amazon Region, edited by M. CHisener-Godt, and 1. Sachs, redo, P. Moutinho, F. Y. Ishida, M. T. P. Santos, J. B. Guerpp. 125-148, UNESCO, Paris, and Parthenon, Carnforth, U. K. rero, K. Kalif, and R. T. Saba (2004a), Nitrogen and phosphorus Fearnside, P. M. (1996), Amazonian deforestation and global limitation of biomass growth in a tropical secondary forest, Ecol. warming: Carbon stocks in vegetation replacing Brazil's AmaAppl., 14, S150-S163. zon forest, For. Ecol. Manage., 80, 21-34. Davidson, E. A, F. Y. Ishida, and D. C. Nepstad (2004b), Effects of an experimental drought on soil emissions of carbon dioxide, Fearnside, P. M. (1997a), Limiting factors for development of agriculture and ranching in Brazilian Amazonia, Rev. Bras. BioI., methane, nitrous oxide, and nitric oxide in a moist tropical for57,531-549. est, Global Change BioI., 10, 718-730. Davidson, E. A, et al. (2007), Recuperation of nitrogen cycling in Fearnside, P. M. (1997b), Environmental services as a strategy for sustainable development in lUral Amazonia, Ecol. Econ., 20, Amazonian forests following agricultural abandonment, Nature, 53-70. 447,995-998. Davidson, E. A, T. D. A Sa, C. J. R. Carvalho, R. O. Figueir- Fearnside, P. M. (1998), PhospholUS and Human Canying Capacity in Brazilian Amazonia, Phosphorus in Plant Biology: Reguedo, M. S. A. Kato, O. R. Kato, and F. Y. Ishida (2008), An
latOly Roles in Molecular, Cellular, Organismic and Ecosystem Processes, Edited by J.iP. Lynch and J. Deickman, pp. 94-108, American Society o(Plant Physiologists, Rockville, MD. Fearnside, P. M. (19Q9), Biodiversity as an environmental service in Brazil's Amaz,~nian forests: Risks, value and conservation, Environ. Conse,,(,., 26, 305-321. Fearnside, P. M. (2000), Global warming and tropical land-use change: Greenhouse gas emissions from biomass burning, decompositions and soils in forest conversion, shifting cultivation and secondary vegetation, Clim. Change, 46,115-158. Fearnside, P. M. (2001), Soybean cultivation as a threat to the environment in Brazil, Envil'On. Conserv., 28, 23-38. Fearnside, P. M. (2004), A agua de Sao Paulo e a floresta amazonica, Cienc. Hoje, 34(203), 63--65. Feamside, P. M. (2005), Deforestation in Brazilian Amazonia: Histmy, rates and consequences, Conserv. BioI., 19, 680-688. Fearnside, P. M. (2008a), Amazon forest maintenance as a source of environmental services, An. Acad. Bras. Cienc., 80,101-114. Feamside, P. M. (2008b), Deforestation in Brazilian Amazonia and global warming, Ann. Arid Zone, 47(3--4), 1-20. Feamside, P. M., and R. 1. Barbosa (1998), Soil carbon changes from conversion of forest to pasture in Brazilian Amazon, For. Ecol. Manage., 108, 147-166. Fearnside, P. M., and W. M. Guimaraes (1996), Carbon uptake by secondary forests in Brazilian Amazonia, For. Ecol.' Manage., 80,35--46. Feigl, B. J., J. Melillo, and C. C. CelTi (1995), Changes in the origin and quality of soil organic matter after pasture introduction in Rondonia (Brazil), Plant Soil, 175,21-29. Feldpausch, T. R., M. A. Rondon, E. C. M. Fernandes, S. J. Riha, and E. Wandelli (2004), Carbon and nutrient accumulation in secondalY forests regenerating on pastures in central Amazonia, Ecol. Appl., 14, S164-S176. Fernandes, E. C. M., Y. Biot, C. Castilla, A Canto, J. C. S. Matos, S. Garcia, R. Perin, and E. Wandelli (1997), The impact of selective logging and forest conversion for subsistence agriculture and pastures on telTestrial nutrientdynamics in the Amazon, Cienc. Cult., 49, 34--47. Fernandes, E. C. M., R. Perin, E. Wandelli, S. G. Souza, J. C. Matos, M. Arco-Verde, T. Ludewigs, and A Neves (1999), Agroforestly systems to rehabilitate abandoned pastureland in the Brazilian Amazon. International Symposium on multi stl·ata agroforestry systems with perennial crops, Turrialba, 28, 24-26. Ferreira, S. J. F., S. Crestana, F. J. Luizao, and S. A F. Miranda (2001), Nutrientes no solo em floresta de terra firme cortada seletivamente na Amazonia Central, Acta Amazonica, 31, 381396. Ferreira, S. J. F., F. J. Luizao, S. A F. Miranda, M. S. R. Silva, and A R. T. Vital (2006), Nutrientes na solu<;ao do solo em floresta de terra firme submetida 11 extl'a<;ao seletiva de madeira na Amazonia central, Acta Amazonica, 36, 59-68. . Fujisaka, S., and D. White (1998), Pasture or permanent crops after slash-and-bum cultivation? Land-use choice in three Amazon colonies, Agrofor. Syst., 42, 45-59. Fujisaka, S., W. Bell, N. Thomas, L. Hurtado, and E. Crawford (1996), Slash-and-bum agriculture, conversion to pasture, and
333
deforestation in two Brazilian Amazon colonies, Agric. Ecosyst. Environ., 59, 115-130. Gallardo-Ordinola, J. L. E. (1999), Produ<;ao e qualidade de liteira em sistemas agroflorestais e seus efeitos sobre as propriedades quimicas do solo, Disserta<;ao de Mestrado, 97 pp., INPAlFUA, Manaus, Brazil. Gehring, C., M. Denich, M. Kanashiro, and P. L. G. Vlek (1999), Response of secondary vegetation in Eastem Amazonia to relaxed nutl'ient availability constraints, Biogeochemisf7y, 45, 223-241. Glaser, B., L. Haumaier, G. Guggenberger, and W. Zech (2001), The Terra Preta phenomenon-A model for sustainable agriculture in the humid tropics, Naturwissenschaften, 88, 37--41. Grimaldi, M., M. SalTazin, A. Chauvel, F. Luizao, N. Nunes, M. R. Lobato-Rodrigues, P. AmbiaI'd, and D. Tessier (1993), Effects de la deforestation et des cultures sur la structure des sols argileux d' Amazonie bresilienne, Cahiers Agricultures, 2, 36--47. Hecht, S. B., D. A Posey, and W. Balee (1989), Preliminary results on soil management techniques ofthe Kayapo Indians. Resource management in Amazonia: Indigenous and folk strategies, Adv. Econ. Bot., 7, 174-188. Henera, R., C. F., Jordan, H. Klinge, and E. Medina (1978), Amazon ecosystems. Their structure and functioning with pal1icu1ar emphasis on nutrients, Interciencia, 3, 223-232. Higuchi, N., M. A. Campos, P. T. B. Sampaio, andJ. Santos (1998), Pesquisas Florestais para a Conservar;iio da Floresta e Reabilitayiio de Areas Degradadas da Amazonia, 264 pp., INPA, Manaus, Brazil. . Holmes, T. P., G. M. Blate, and J. C. Zweede (2002), Financial and ecological indicators of reduced impact logging performance in the eastem Amazolj., For. Ecol. Manage., 163, 93-110. Homma, A. K. O. (1994), Amazonia: Desenvo1vimento economico e questao ambiental, Agricultura e Meio Ambiente, edited by E. F. Vilhena and L. C. Santos, pp. 25-37, UFV-NEPEMA, Vi<;osa, Brazil. Houghton, R. A, D. L. Sko1e, C. A Nobre, J. L. Hackler, K. T. Lawrence, and W. H. Chomentowski (2000), Annual fluxes of carbon from deforestation and regrowth in the Brazilian Amazon, Nature, 403, 301-304. IBGE (2006), Censo Agropecuario do Brasil, Instituto Brasileiro de Geografia e Estatistica-IBGE, Rio de Janeiro, Brazil. IBGE/SIDRA (1997), Sistema IBGE de Recupera<;ao Automatica de Dados Agregados (online), Instituto Brasileiro de Geografia e Estatistica. (Available at http://www.sidra.ibge.gov.brl) INPE (2000), Monitoramento da floresta amazonica brasileira por satellite-Projeto PRODES (online), Instituto Naciona1 de Pesquisas Espaciais. (Available at http://www.obt.inpe.br/prodesl) Johns, J. S., P. Barreto, and C. Uhl (1996), Logging damage during planned and unplanned logging operations in the eastern Amazon, For. Ecol. Manage., 89, 59-77. Kato, A K. (1995), Dinamica da entrada de nutrientes via liteira em plantios de castanheira-do-Brasil (Bertholletia excelsa H.B.K.) em ecossistemas de pastagens degradadas e de floresta primaria, Ph.D. thesis, INPAIUniversidade Federal do Amazonas, Manaus, Brazil. Kato, O. R., M. S. Kato, T. A Sa, andR. Figueiredo (2004), Plantio direto na capoeira, Cienc. Ambiente, 29, 99-111.
334
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
Kauffman, J. B., D. L. Cummings, and D. E. Ward (1998), Fire in the Brazilian Amazon, 2, Biomass, nutrient pools and losses in cattle pastures, Oecologia, 113,415-427. Keller, M., ~, Palace, G. P, Asner, R. Pereira, and J. N, M, Silva (2004), C'oarse wood debris in undisturbed and logged forests in the eastern Brazilian Amazon, Global Change Bioi., 10, 784795, Kitamura, P. C, (1994), A Amazonia e 0 Desenvolvimento Susten{(j{Jel, 182 pp" Embrapa, Brazil. Laurance, W. F" L. V, Ferreira, J. M. Rankin-de Merona, and S, G. Laurance (1998), Rainforest fr\agmentation and the dynamics of Amazonian tree communities, Ecology, 79, 2032~2040, Laurance, W, F" M. A Cochrane, S. Bergen, P. M. Fearnside, P, Delamonica, C, Barber, S. D'Angelo, and T. Fernandes (2001), The future of the Brazilian Amazon, Science, 291, 438-439, Laurance, W, F., et al. (2004), Pervasive alteration of tree communities in undisturbed Amazonian forests, Nature, 428, 171-175. Lehmann, J., and M. Rondon (2006), Bio-char soil management on highly weathered soils in the humid tropics, in Biological Approaches to Sustainable Soil Systems, edited by N. Uphoff, pp, 517~530, CRC Press, Boca Raton, Fla, Lehmann, J., D. C. Kern, L. A. Gelman, J. McCann, G, C. Mariins, and A. Moreira (2003), Soil feliility and production potential, in Amazonian Dark Earths: Origin, Properties, Management, edited by J. Lehmann et aI., pp, 105-124, Springer, Dordrecht, The Netherlands, Lehmann, J., C. V, Campos, J. L. V, Macedo, and L German (2004), Sequential fractionation and sources ofP in Amazonian Dark Earths, in Amazonian Dark Earths: Explorations in Time and Space, edited by B. Glaser and W. 1. Woods, pp. 113-123, Springer, Berlin, Germany. Liang, B" et al. (2006), Black carbon increases cation exchange capacity in soils, Soil Sci. Soc. Am, J, 70, 1719~1730. Lima, H. N. (2001), Genese, quimica, mineralogia e micromorfologia de solos da Amazonia Ocidental, PhD thesis, Universidade de Vil(osa, Vil(osa, Brazil. Lucas, Y., F. J. Luizao, A Chauvel, J. Rouiller, and D. Nahon (1993), The relation between biological activity of the rain forest and mineral composition of soils, Science, 260, 521-523, Luizao, F, J. (1989), Litter production and mineral element input to the forest floor in a central Amazonian fbrest, GeoJournal, 19,407-417. Luizao, F" P, Matson, G. Livingston, R. Luizao, and P. Vitousek (1989), Nitrous Oxide Flux Following Tropical Land Clearing, Global Biogeochem, Cycles, 3,281-285, Luizao, F. J., R, C. Luizao, and A Chauvel (1992), Premiers resultats sur la dynamique des biomasses racinaires et microbiennes dans un "latossol" d'Amazonie Centrale (Bresil) sous foret et sous paturage, Calliers ORSTOM, ser, Pedol" 27, 69-79. Luizao, F. J., J. Proctor, J. Thompson, R. C, C. Luizao, R. H. Marrs, D. A Scott, and V. Viana (1998), Rain forest on Maraca Island, Roraima, Brazil: Soil and litter process response to artificial gaps, For. Eco!. Manage" 102, 291~301. Luizao, F. J., S. Tapia-Coral, J. Gallardo-Ordinola, G, C, Silva, R. C. C. Luizao, L. Trujillo-Cabrera, E. Wandelli, and E. C. M, Fernandes (2006), Ciclos biogeoquimicos em agroflorestas na
Amazonia, in Sistemas Agroflorestais: Bases Cient(ficas para 0 Desenvolvimento Sustentavel, edited by A, C. Gama-Rodrigues et aI., pp, 87-100, UENF, Campos dos Goytacazes, Brazil. Luizao, R, C., T, A, Bonde, and T. Rosswall (1992), Seasonal variation of soil microbial biomass: The effect of clearfelling a tropical rainforest and establishment of pasture in the central Amazon, Soil Bio!. Biochem" 24,805-813. Luizao, R. C. C" E. S. Costa, F, J. Luizao (1999), Mudanl(as na biomassa microbiana e nas transfonnal(oes de nitrogenio do solo em uma seqiiencia de idades de pastagens ap6s den'uba e queima da floresta na Amazonia central, Acta Amazonica, 29, 43-56. Luizao, R, C. c., F. J. Luizao, R. Q, Paiva, T. F. Monteiro, L. S. Souza, and B, Kruijt (2004), Variation of carbon and nitrogen cycling processes along a topographic gradient in a central Amazonian forest, Global Change Bioi" 10, 592-600, Mackensen, J., D, Holscher, R, Klinge, and H. Foister (1996), Nutrient transfer to the atmosphere by burning of debris in eastern Amazonia, For, Ecol. Manage., 86, 121-128, Markewitz, D" E, Davidson, P. Moutinho and D, Nepstad (2004), Nutrient loss and redistribution after forest clearing on a highly weathered soil in Amazonia, Ecol. App!., 14, SI77~SI99. McCaffely, K., M. Rondon, J. Gallardo, S. Welsch, T. Feldpausch, E. Fernandes, S, Riha, E. Wandelli (2002), Carbon and nutrient stocks in agroforestly systems and secondary forest in the Central Amazon, Proceedings ofthe Second Scientific Conference of the LBA Project, Manaus, Brazil. Melendez, G., R. Ocampo, F. Herrera, and J. Briceno (1999), La biodiversidad vegetal y el funcionamiento ecol6gico del frijol tapado, inEI Frijol Tapado en Costa Rica: Fortalezas, Opciones y Desafios, pp. 79-102, Univ. of Costa Rica, San Jose, Costa Rica. Mello-Ivo, W" S. FelTeira, Y. Biot, and S. Ross (1996), Nutrients in soil solution following selective logging of a humid tropical "terra-finne" forest north of Manaus, Brazil, Environ, Geochem. Health, 18, 69-75. Mello-Ivo, W. M. P., and S, Ross (2006), Efeito da colheita seletiva de madeira sobre algumas caracteristicas fisicas de um latossolo amarelo sob floresta na Amazonia Central, Rev. Bras. Genc. Solo, 30, 769~776, Mesquita, R, C. G., S. W. Workman, and C, L. Neely (1998), Slow litter decomposition in a Cecropia-dominated secondary forest of central Amazonia, Soil Bioi, Biochem, 30, 167-175. Mesquita, R. C. G., K. Ickes, G, Ganade, and G, B, Williamson (2001), Alternative successional pathways in the Amazon Basin, J £Col" 89, 528-537. Moraes, J. F. L., B, Volkoff, C, C. CelTi, and M, Bernoux (1996), Soil properties under Amazon forest change due to pasture installation in Rondonia, Brazil, Geoderma, 70, 63~81. Moran, E. F., E. S. Brondizio, J. M. Tucker, M, C. Silva-Forsberg, S. McCracken, and 1. Falesi (2000), Effects of soil feliility and land-use on forest succession in Amazonia, For, Ecol. Manage., 139,93-108, Nascimento, H, E. M" and W. F. Laurance (2004), Biomass dynamics in Amazonian forest fragments, Ecol, Appl., 14, SI27-S138. Neeff, T" R. M" Lucas, J. R. Santos, E. S. Brondizio, and C. C. Freitas (2006), Area and age of secondary forests in Brazilian
LUIZAo ET AL. Amazonia 1978-2002: An empirical estimate, Ecosystems, 9, 609~623.
Neill, C" and E. A. D,\1vidson (1999), Soil carbon accumulation or loss following deJ'orestation for pasture in the Brazilian Amazon, in Global Climate Change and Tropical Ecosystems, edited by R, Lal, J. M:Kimble, and B, A, Stewart, pp, 197~211, CRC Press, Boca Raton, Fla, Neill, C., C. C, Cerri, J. Melillo, B, J. Feigl, P, A Steudler, J. F. L. Moraes, and M, C. Piccolo (1997), Stocks and dynamics of soil carbon following deforestation for pasture in Rondonia, in Soil Processes and the Carbon Cycle, edited by R, Lal et aI., pp, 9~28, CRC Press, Boca Raton, Fla. Nepstad, D" P. R. S, Moutinho, and D. Markewitz (2001), The recovery of biomass, nutrient stocks, and deep-soil functions in secondary forests, in The BiogeochemistlY of the Amazon Basin, edited by M, E. McClain, R. L. Victoria, and J. E. Richey, pp, 139-155, Oxford Univ. Press, New York. Nepstad, D. C" et al. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505-508, Oliveira, M, V., and E. M. Braz (1995), Damage reduction through planned harvesting in Brazilian moist tropical forest, Commonw. For. Rev., 74,208-210. Palm, C. A, K. E. Giller, P. L. Mafongoya, and M. J. Swift (2001), Management of organic matter in the tropics: Translilting theOly into practice, Nutrient Cycling in Agroecosystems, 61, 63-75. Palm, C, A, S, A. Vosti, P. Sanchez, and P. 1. Ericksen (2005), Slash-and-Burn Agriculture-the Search for Alternatives, Columbia Univ, Press, New York, Pauletto, D. (2006), Estoque e produl(ao de liteira grossa em floresta submetida ao manejo florestal no noroeste de Mato Grosso, Dissertal(ao de Mestrado, 85 pp., INPAlUFAM, Manaus, Brazil. Piccolo, M. c., F. Andreux, and C. C. Cerri (1994), Hydrochemistry of soil solution collected with tension-fr'ee lysimeters in native and cut-and-burned tropical rain forest in Central Amazonia, Geochim. Brasil" 81, 51-63, Post, W. M" and K. C. Kwon (2000), Soil carbon sequestration and land-use change: Processes and potential, Global Change Bioi" 6, 317~327. Proctor, J. (1984), Tropical forest litterfall. II. The data set, Tropical Rain Forest: The Leeds Symposium, edited by A C, Chadwick and S. L. Sutton, pp. 83-113, The Leeds Philosophical and Literary Society, Leeds, U. K. Rondon, M., E. Fernandes, R. Lima, and E. Wandelli (2000), Carbon storage in soils from degraded pastures and agroforestry systems in Central Amazonia: The role of charcoal, Proceedings of the Second Scientific Meeting of the LBA-Eco Project, Atlanta, Ga, Ross, S. M" J. B. Thomes, and S. Nortcliff(1990), Soil hydrology, nutrient and erosional response to the clearance of terra firme forest, Maraca Island, Roraima, northern Brazil, Geogr, J, 156, 267-282, Sa, T. D, A, O. R. Kato, C. J. R. Carvalho, and R. O. Figueiredo (2007), Queimar ou nao queimar? De como produzir na Amazonia sem queimar, Revista USP, 72,90-97, Sampson, R. N., et al. (2000), Additional human-induced activities-Article 3.4, Land Use, Land Use Change, and Forestry,
335
A special report of the fPCC, edited by R, T, Watson et aI., pp. 181-281, Cambridge Univ. Press, Cambridge, U. K, Sanchez, P. A, J. H. Villachica, and D. E. Band (1983), Soil fertility dynamics after clearing a tropical rainforest in Peru, Soil Sci, Soc, Am, J, 47, 1171-1178. Schroth, G., S. A. D'Angelo, W. G. Teixeira, D. Haag, and R. Lieberei (2002), Conversion of secondary forest into agroforestly and monoculture plantations in Amazonia: Consequences for biomass, litter and soil carbon stocks after 7 years, For. Ecol. Manage" 163,131-150. Schubari, H. 0, R., W, Franken, and F, J. Luizao (1984), Uma floresta sobre solos pobres, Genc. Hoje, 2, 26-32. Scott, D. A, J. Proctor, and J. Thompson (1992), Ecological studies on a lowland evergreen rain forest on Maraca Island, Roraima, Brazil. II, Litterfall and nutrient cycling, J Ecol" 80, 705-717, Serrao, E. A. S., and J. M, Toledo (1990), The search for sustainability in Amazonian pastures, Alternatives to Deforestation: Steps Towards Sustainable Utilization of Amazon Forests, edited by A B. Anderson, pp, 195~214, Columbia Univ. Press, New York. Serrao, E. AS" 1. C. Falesi, J. B. Veiga, and J. F, T. Neto (1979), Productivity of cultivated pastmes on low fertility soils of the Amazon Basin, Pasture Production in Acid Soils ofthe Tropics, edited by P. A Sanchez and L. E. Tergas, pp, 195-225, Centro Internacional de Agricultura Tropical, Cali, Colombia. Seubert, C, E., P. A. Sanchez, and C. Valverde (1977), Effects of land clearing metljods on soil propeliies and crop performance in an Ultisol of Arnazon jungle of Peru, Trop, Agric, (Trin.), 54, 307-321. Silva, G. C. (2005), F;luxos e estoques de nutrientes, colonizal(ao pOl' micorrizas arbusculares e influencia das raizes na decomposil(ao da liteira em sistemas agroflorestais e vegetal(iio secundaria na Amazonia central, Tese de Doutorado, 154 pp" INPA/UFAM, Manaus, Brazil. Silva, S. M. A, J. N. M. Silva, A M. V. Baima, N, M. Lobato, 1. S. Thompson, and P. P, Costa-Filho (2001), Impacto da exploral(ao madeireira em floresta de tena finne no municipio de Moju, estado do Para, in A Silvicultura na Amazonia Oriental: Contribui<;oes do Projeto EmbrapalDFID, edited by J. N. M. Silva, J. O. P. Carvalho, and J. A G. Yared, pp. 309-323, Embrapa/DFID, Belem, Brazil. Silva-Forsberg, M, C" and P. M. Fearnside (1995), Agricultural management of caboclos of the Xingu river: A starting point for sustaining populations in degraded areas in the Brazilian Amazon, Management and Rehabilitation of Degraded Lands and SecondOl)! Forests in Amazonia, edited by J. A Parrota and M, Kanashiro, pp, 90-95, International Institute of Tropical Forestry, USDA-Forest Service, Rio Piedras, Puelio Rico. Smeraldi, R" and P. May (2008), 0 Reino do Gado: Uma Nova Fase na Pecuariza<;(fo da Amazonia, Amigos da Terra-Amazonia Brasileira, Sao Paulo. Smith, N. J. H. (1976), Transamazon Highway: A cultural-ecological analysis of colonization in the humid tropics, Ph,D. dissertation in Geography, Univ. of California, Berkeley, Calif, Sombroek, W., M. L. Ruivo, P. M, Fearnside, B. Glaser, and J. Lehmann (2003), Anthropogenic dark earths as carbon stores
336
MAINTENANCE OF SOIL FERTILITY IN AMAZONIAN MANAGED SYSTEMS
and sinks, Amazonian Dark Earths: Origin, Properties, Management, edited by J. Lehmannet aI., pp, 125-139, Springer, Dordrecht, The Netherlands. Summers, :ij-. M. (1998), Estoque, decomposiyao e nutrientes da liteira grossa em fioresta de terra firme na Amazonia central, Dissertayao de Mestrado, 103 pp., INPA/UFAM, Manaus, Brazil. Tapia-Coral, S. C" F. J. Luizao, and E. V. Wandelli (1999), Macrofauna'da liteira em sistemas agrofiorestais sobre pastagens abandonadas na Amazonia central, Acta Amazonica, 29, 447--495. Tapia-Coral, S, C" F, J. Luizao,:E. V. Wandelli, and E. C. M. Fernandes (2005), Carbon and nutrient stocks in the litter layer of agroforestry systems in Central Amazonia, Agrofor, Syst., 65, 33--42. Townsend, A. R., G, P. Asner, C, C, Cleveland, M. E. Lefer, and M. M. C, Bustamante (2002), Unexpected changes in soil phosphorus dynamics along pasture chronosequences in the humid tropics, J. Geophys, Res., 107(D20), 8067, doi:10.l029/ 2001JD000650. Trumbore, S. E., E. A. Davidson, P. B. Camargo, D. C. Nepstad, and L. A. Matiinelli (1995), Below ground cycling of carbon in forests and pastures of eastern Amazonia, Global Biogeochem, Cycles, 9, 515-528. Tucker, J. M" E. S, Brondizio, and E, F. Moran (1998), Rates of forest regrowth in Eastern Amazonia: A comparison of Altamira and Bragantina regions, Para State, Brazil, Interciencia, 23, 64-73. Uguen, K. (2001), Effect de la qua lite de litiere sur la mineralization d' Azot du Sol en systemes agroforestiers, Ph.D, thesis, 108 pp., Universite Paris VI, Paris, France. Uhl, C., R. Buschbacher, and E. A. S. Serrao (1988), Abandoned pastures in Eastern Amazonia, 1. Patterns of plant succession, J. Ecol., 76,663--681.
Vasconcelos, H. L., and W. F. Laurance (2005), Influence ofhabitat, litter type, and soil invertebrates on leaf-litter decomposition in a fragmented Amazonian landscape, Oecologia, 144,456-462. Vasconcelos, H. L., and F. J. Luizao (2004), Litter production and litter-nutrient concentrations in a fragmented Amazonian landscape: Edge and soil effects, Ecol. Appl., 14, 884-892. Vieira, 1. C. G" D. C, Nepstad, S. Brienza Jr., and C. Pereira (1993), A importancia de areas degradadas no contexto agricola e ecol6gico da Amazonia, Bases Cientijicas para Estrategias de Preservat;iio e Desenvolvimento da Amazonia, vol. 2, edited by E. G. Ferreira et al. pp. 43-53, INPA, Manaus, Brazil. Vitousek, P. M. (1984), Litterfall, nutrient cycling, and nutrient limitation in tropical forests, Ecology, 65, 285-298. Walker, R., and A. K. O. Homma (1996), Land use and land cover dynamics in the Brazilian Amazon: An overview, Ecol, Econ" 18,67-80. Woomer, P, L., et al. (1999), Slash-and-burn effects on carbon stocks in the humid tropics, Global Climate Change and Tropical Ecosystems, edited by R. Lal, J. M. Kimble, and B. A. Stewart, pp. 99-115, CRC Press, Boca Raton, Fla. Yared, J. A. G. (1996), Efeitos de sistemas silviculturais na fioristica e na estrutura de fiorestas secundaria e primaria na Amazonia Oriental, Ph.D. thesis, 179 pp., Viyosa, Brazil. Young, A. (1997), Agroforestry for Soil Management, 306 pp., CAB International and ICRAF, Wallingford, U. K.
Sources and Sinks of Trace Gases in Amazonia and the Cerrado M. M. C. Bustamante Department ofEcology, University ofBrasilia, Brasilia, Brazil
M. Keller! Intel'llational Institute of Tropical Forest/y, USDA Forest Service, Rio Piedras, Puerto Rico
D. A. Silva Department ofEcology, University ofBrasilia, Brasilia, Brazil
Data for trace gas fluxes (NOx, NzO, and ClL() from the Amazon and cerrado region are presented with focus on the processes of production and consumption of these trace gases in soils and how they may be changed because ofland use changes in both regions. Fluxes are controlled by seasonality, soil moisture, soil texture, topography, and fine-root dynamics. Compared to Amazohian forests where the rapid cycling of nitrogen supports large emissions of NzO, nitrification rates and soil emissions ofN oxide gases in the cerrado region are very low. Several studies report CH4 consumption during both wet and dry seasons in forest soils, but there is occasionally net production of CH 4 during the wet season. A few studies suggest an unknown source ofCH4 from upland forests. As with N oxide emissions, there are few data on CH4 emissions from cerrado soils, but CH4 consumption occurs during both wet and dry seasons. Clearing natural vegetation, burning, fertilization of agricultural lands, intensive cattle ranching, and increasing dominance by legume species in areas under secondary succession after land conversion have all been identified as causes of increasing NzO and NO emissions from tropical regions. Large uncertainties remain for regional estimates of trace gas fluxes. Improvement of models for the N oxides and CH4 fluxes for Amazonia and the cerrado still depends upon gathering more data from sites more widely distributed across two vast biomes and more importantly on basic theOly about the controls of emissions from the ecosystem to the atmosphere.
C. E. P. Cerri, ESALQ, University of Sao Paulo, Piracicaba, SP 13400-970, Brazil. P. M. FearnsideandF. J. Luizao, Department of Ecology, INPA, Manaus, AM 69060-001, Brazil. ([email protected]) J. Lehmann, Department of Crop and Soil Sciences, Cornell University, Ithaca, NY 14853, USA.
1Also
at NEON Inc., Boulder, Colorado, USA.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000733
1. INTRODUCTION Tropical forest and savanna regions are substantial natural source regions for the trace gases nitrous oxide (NzO) and methane (C}4). The Amazon region contains the largest extent of tropical forest on Earth and a large area of savanna, 337
338
BUSTAMANTE ET AL.
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
known in Brazil as cerrado. The cerrado of Brazil, covering tion. Anaerobic conditions are found in soils when the rate of 2 millionlanz, is the largest savanna biome in South Amer- Oz diffusion is slower than Oz consumption. Within the Soil ica and an area of extremely rapid agricultural development environment, the aerobic process of nitrification (predomi[Klink and~\1achado, 2005]. The tropical forests of Amazo- nant at <60% water-filled pore space (WFPS)) is maintained nia are also suffering rapid rates of change. Deforestation primarily by autotrophic bacteria and archea [Leininger et in Brazil has averaged nearly 20,000 km z a-lover the past al., 2006], resulting in the conversion of ammonium (NH/) 2 decades [see Alves et al., this volume] and selective log- to nitrate (N03-) via nitrite (NOz-). There are two functional ging covers a similar area [see Asner et al., this volume]. groups of nitrifiers, namely, the ammonium oxidizing nitriThe conversion of native vegetation formations to pasture fiers, which convert NH/ via hydroxylamine to NOz-, and and cropland causes substantial changes in biogeochemical the nitrite oxidizing nitrifiers, which oxidize NOz- to N03-. processes, including the ecosystem-atmosphere exchange of Denitrification, on the other hand, is an anaerobic process (predominant at >60% WFPS) in which denitrifiers reduce NzO, nitric oxide (NO), and CH4. N0 Nitrous oxide is a stable greenhouse gas in the troposphere 3 (via NOz-), NO, and nitrous oxide (NzO) to molecular nitrogen (Nz). The complete denitrification pathways result that absorbs strongly in the infrared and has a long atmoin the reduction of N0 3- to N z, but significant amounts of spheric lifetime of about 120 years. As a result, its 100-year NO and NzO can be emitted before complete reduction to global warming potential is 298 times that ofCO z [Forster et Nz. Soil pH, metallic ion composition, and soil organic matal., 2007]. In the stratosphere, NzO is destroyed by photolyter (SOM) all control the abiotic process of chemodenitrificasis and reaction with excited oxygen, which is a source ofnition, whereby microbially produced NO z- is decomposed to tric oxide that contributes to depletion of the ozone [Crutzen, NO and NO [Davidson, 1992]. The conceptual hole-in-the1970]. Therefore the cunent increase from the preindustrial z pipe model has been formalized in simulation models such atmospheric mixing ratio of 270 ppbv to 319 ppbv (in 2005) as the NASA-Carnegie-Ames-Stanford approach (CASA) has the potential to impact global climate over the next cen[Potter et al., 1998] and the Tenestrial Ecosystems Model tury [Lashof and Ahuja, 1990; Prather and Ehhalt, 2001; [Melillo et al., 2001]. Forster et al., 2007]. The global budget ofNzO is seriously Methane (CH4) is an important greenhouse gas, and its atperturbed with anthropogenic sources accounting for nearly mospheric concentration has more than doubled since preina 50% increase over natural sources from terrestrial ecosysdustrial times from about 0.7 to about 1.8 ppm today [Spahni tems and the oceans [Hirsch et aI., 2006]. Atmospheric NzO et al., 2005; Bosquet et al., 2006; Forster et al., 2007]. Most is produced primarily as a result of microbial processes in atmospheric CH4 is produced by anaerobic degradation of soils especially through denitrification and through nitrifiorganic matter under conditions where anaerobic respiration cation when oxygen tensions are low. Soils of the tropical by microbes is limited by the paucity of alternate electron acforest biome are the most important natural source for NzO ceptors [Fenchel et al., 1998]. Natural wetlands are the most [Matson and Vitousek, 1990]. important global source of C~, producing up to about 200 Nitric oxide (NO) is a shOli-lived reactive gas that influTg CH4 a-I (up to 40% of the estimated total C~ producences the oxidant balance of the troposphere and the produc[Reebwgh, 2003]. The CH4 source from the wetlands tion) tion of ozone, a component of photochemical smog and a of the Amazon Basin has been estimated to be 29 Tg C~ greenhouse gas. Although most atmospheric NO is produced a-I [Melack et al., 2004]. This important source is discussed either by fossil fuel combustion or by ligh,tning, biological by Melack: et al. [this volume] together with other aspects of nitrification and denitrification, as well as chemodenitrificathe wetland carbon cycle. Microbial production of C~ in tion in soils, are also impOliant [Davidson and Kingerlee, termite guts is another natural source of CH4 that may be im1997]. A global estimate of NO emissions from soils is less certain than for NzO but is about 21 Tg N a-I [Davidson and portant in tropical ecosystems. Although this source is generally considered to be a minor term in the global budget, it Kingerlee, 1997]. In soils, the production and emission ofNzO and NO are is extremely difficult to quantify and highly unceliain [Marmainly regulated by the same processes. These have been tius et al., 1993]. Reaction with OH radicals and transpOli summarized in a conceptual model known as "the hole- to the stratosphere are the primary sinks for tropospheric in-the-pipe" [Firestone and Davidson, 1989; Davidson et al., CH4. In soils under aerobic conditions, CH4 tends to be con2000]. According to this model, total emissions ofNzO and sumed from the atmosphere in a process mediated primarily NO are proportional to the inorganic nitrogen (ammonium by bacteria that specialize in C 1 compounds, collectively and nitrate ions) flowing through the nitrification and denitri- known as methylotrophs [Hanson and Hanson, 1996]. Soil fication pipes. Gases leak out through holes that are regulated CH4 consumption accounts for less than 10% of the annual by soil conditions such as moisture and oxygen concentra- destruction of atmospheric C~ [Reeburgh, 2003].
2. FLUXES OF NzO AND NO FROM NATURAL ECOSYSTEMS TO THE ATMOSPHERE 2.1. N 2 0 and NO Fluxes FrolJl Amazon Forest Soils »
The rapid cy{{ing of nitrogen in Amazonian forests supports large emissions ofNzO. A summalY of reported average annual fluxes of NO and NzO for forests in the Amazon regions is presented in Table 1. The data indicate annual emissions of NzO from soils of mature Amazonian forests ranging from 1.1 [Davidson and Kingerlee, 1997] to 6.9 kg N ha- I a-I [Keller et al., 2005]. Estimates of annual emissions of NO from Amazonian forests ranged from 0.9 [Davidson et al., 2004] to 2.4 kg N ha- I a-I [Garcia-Montiel et al., 2003]. The controls of NO and NzO fluxes including seasonality, soil moisture, soil texture, topography, and fineroot dynamics are discussed in this section.
339
Variability of rainfall is a more important control than the small variation in soil temperature for the emissions of N oxides from soils in forests in the Amazon region [Verchot et al., 2000; Keller et al., 2005] (Figure 1). Nitrous oxide emissions are controlled by carbon and nitrogen availability and soil aeration in relation to water saturation. Both nitrifying and denitrifying bacteria produce NzO, but the largest emissions result from denitrification under anaerobic conditions. For example, fluxes ofNzO were generally higher during the wet season (January-June) than the dry season [Davidson et al., 2004] in Para. The significant increase ofNzO emission in the wet season could have been due to a combined effect of increased soil water content and nitrogen mineralization. The ratios ofNzO:NO fluxes were positively correlated with volumetric water content in soils in Para [Davidson et al., 2004]. Working on sandy soils in Rondonia (Brazil), Garcia-Montiel et al. [2003] also found over 84% ofNzO fluxes
Table 1. Summary of Reported Average Annual Fluxes of NO, NzO and CH4 From Natural and Managed Ecosystems in the Amazon . Region Location! Ecosystem Amazonas State Forest Amazon region Forest Secondary forest Active pasture Old pasture Para State Primary forest Secondary forest Active pasture Degraded pasture Forest Primmy forest clay Primmy forest sandy loam Logged forest clay Logged forest sandy loam Rondonia State Forest Pasture (4 years) Pasture (22 years) Forest Forest Pasture (1-3 years) Pasture (6 years)
Reference Kaplan et al. [1988] Davidson Davidson Davidson Davidson
and Kingerlee and Kingerlee and Kingerlee and Kingerlee
NO Flux (kg N-NO ha-I a-I)
NzOFlux (kg N-NzO ha- I a-I)
1.4
1.4 and 1.9
[1997] [1997] [1997] [1997]
CH4 Flux (kg CH4 ha- 1 a-I)
1.1 0.3 0.5 0.4
Verchot et al. [1999,2000] Verchot et al. [1999,2000] Verchot et al. [1999,2000] Verchot et al. [1999,2000] Davidson et al. [2004] Nepstad et al. [2002] Keller et al. [2005] Keller et al. [2005]
1.5 0.7 0.5 0.3 0.9 1.7 7.9 7.7
2.4 0.9 0.3 0.1 2.6 2.3 6.5 1.4
-0.7 -3.5
Keller et al. [2005] Keller et al. [2005]
28.6
7.0 6.5
16.8 7.5
Feigl et al. [2001] Feigl et al. [2001] Kirkman et al. [2002] Garcia-Montiel et al. [2003] Neill et al. [2005] Neill et al. [2005] Neill et al. [2005]
~30%
0.17 2.4 1.4 0.2 0.2
1.9-2.8 < than forest 3.2 4.3 3.1-5.1 0.1-0.4
-2.1 -1.0 -1.3 -3.1 -1.1
-6.7 3.6
340
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
BUSTAMANTE ET AL.
25
•
20 ~
'"
......I
1:
Clay Clay, r 2
•
=0.85
Sandy loam
. ......... Sand, r 2
=0.46
•
15
<;l
E () ZI
10
•
OJ
c:
6
5
N
Z 0
a -5 20
0
60
40
%WFPS 5
•
4
'"
•
Clay Sandy loam
- - All data, r
2
=0.12
'"
3
......I "'0
NI
2
'"
E OJ
E
~
:c
0
•
"
()
-1
•
-2
-3 0
"'!
'" '"
•
'"
b 40
20
'"
• •
60
%WFPS
Figure 1. (a) NzO and (b) C~ fluxes versus soil WFPS from clay (circles) and sandy loam Ultisol (triangles) soils from undisturbed sites in the Tapaj6s National Forest. Reprinted from Keller et al. [2005].
were emitted during the wet season. Similarly, during the wet season, soil N20 fluxes in forest and pasture sites near Santarem, Para, Brazil, were also positively correlated to CIN ratio in leaflitter, NH/-N, and the ratio N03--N/(N03--N + NH/-N) but were negatively correlated to soil CH4 consumption and bulk density, indicators of soil aeration [Wick et al., 2005]. In contrast to NzO, emissions ofNO were more evenly distributed over the wet and dly seasons in Rondonia [Garcia-Montiel et al., 2003; Vasconcelos et al., 2004]. In laboratory experiments on forest soil samples collected across Brazilian Amazonia, van Dijk and Meixner
[2001] measured NO production and the rate constant of NO consumption under variable temperature and moisture conditions. NO production increased exponentially with soil temperature. Under velY dly and velY wet soil conditions, the response of the NO production to a change in temperature was less pronounced than under conditions of intermediate soil moisture. NO production peaked, independently of soil temperature, at soil moisture of 0.10 g g-1 (0.27 and 0.38 WFPS for forest and pasture soils, respectively). On the other hand, NO consumption was most efficient at high soil temperatures (>25°C) and under dry soil conditions «0.20 g
g-I or 0.53 WFPS for fOl'est soils). The rate constants of NO consumption were within 5% of NO release with comparable values for forest..a'ild pasture soils. Tropical humi~Jrforests have generally high rates of net mineralization alJd net· nitrification that can lead to the accumulation ofNH/ and N0 3 - during dry periods [Neill et al., 1995]. The accumulation ofNH4+ and N03- in dry soils, followed by wetting, provides more favorable conditions for microbial activity and the development of soil anoxia stimulating the production of both N20 and NO [Davidson, 1991; Garcia-Mendez et al., 1991]. Soil accumulation of NH/, NO z-, and N0 3- may occur in thin water films of microsites near oxidizing sites. Upon soil wetting, soil microbes can quickly use these pools, and produce pulses ofN oxide gases. The effects of moisture and substrate availability on soil fluxes of NO and NzO in an Amazonian regrowth forest were quantified by Vasconcelos et al. [2004] through irrigation during the dry season and removal of aboveground litter. Fluxes of N20 and NO increased during dly season inigation, while litter removal had no significant impact on N oxide emissions. Net soil nitrification did not respond to dry season irrigation but was somewhat reduced by litter removal. In contrast, pulse releases following wet~up events contributed relatively little to total annual emissions ofNzO and NO in the forest and pastures in studies in Rondonia [Neill et al., 2005]. Curiously, an experiment investigating the effects of drought simulating the predicted climate change in the Amazon region showed no effect on NO and NzO emissions [Davidson et al., 2004]. Luiziio et al. [2004] showed a significant differentiation of nitrogen mineralization and inorganic N concentration along a topographic gradient near Manaus, Amazonas, Brazil, with a positive rate of net N mineralization on the plateau and slope location and negative net N immobilization in the valley on sandy soils [Luiziio et al., 2004]. The decrease in total N concentrations followed the clay content in the topographic gradient, with higher concentrations in the clayey Oxisol and much lower concentrations for the sandy Podzol. This result explains early observations that noted considerably greater NO emissions from clay soils on plateaus as compared to sandy soils in valleys in the Manaus region [Bakwin et al., 1990]. Consistent differences in soilatmosphere fluxes of NzO and NO with soil texture were determined by Keller et al. [2005] in a study over 2 years in undisturbed forest, near Santarem, Para. Annual soilatmosphere fluxes ofNzO (mean plus/minus standard error) were 7.9 ± 0.7 and 7.0 ± 0.6 ng N cm-z h- 1 for a clay-texture' Oxisol compared to 1.7 ± 0.1 and 1.6 ± 0.3 ng N cm-2 h- 1 for a sandy loam-texture Ultisol for 2000 and 2001, respectively (Figure 1). The annual fluxes of NO from undisturbed forest soil in 2001 were 9.0 ± 2.8 ng N cm-2 h- 1 for the Oxisol and
341
8.8 ± 5.0 ng N cm- z h- 1 for the Ultisol [Keller et al., 2005]. These results are consistent with earlier studies from the same region that showed significantly greater net nitrogen mineralization, net nitrification, and denitrification enzyme activity in the clay soils compared to the sandy loams [Silver et al., 2000]. Along with soil texture, fine-root dynamics can affect the production and emission of trace gases, especially in tropical rain forests that are characterized by high root biomass density and rapid turnover of fine roots. Working in the Santarem region, Varner et al. [2003] and Silver et al. [2005] examined patterns in fine-root dynamics on clay and sandy loam soils in a lowland moist forest and its effect on rates of C and N trace gas fluxes. Root production did not differ significantly with soil texture. However, root decay was faster in clay than in sandy loam soil, leading to greater standing stocks of dead roots in the sandy loam. Rates of NzO emissions were significantly greater in the clay soil (13 ± 1 ng N cm-z h- I) than in the sandy loam (1.4 ± 0.2 ng N cm-z h-1). Root mortality and decay following trenching doubled rates of NzO emissions in the clay and tripled them in the sandy loam over a I-year period. Trenching also increased NO fluxes, with greater fluxes in the sandy loam than in the clay soil. The authors concluded that fine-root mortality and dec9mposition associated with disturbance and land use change can contribute significantly to increased rates ofN oxide eJV.issions. 2.2. Interactions oiNO and N0 2 With Amazonian Forest Canopy In the uppermost soil layers, NO is simultaneously produced and consumed by microbiological processes, generally resulting in a net emission [Conrad, 1996]. However, when the soil is covered by tall vegetation, not all NO emitted from the soil leaves the canopy. A substantial portion of biogenically emitted NO can react with ozone (0 3), mixed into the canopy from aloft, to form nitrogen dioxide (N02) in the subcanopy atmosphere. The NO z can subsequently be deposited on vegetation elements [Jacob and Wofty, 1990; Jacob and Balrwin, 1991; Meixner, 1994; Ammann et al., 1999]. In a forest in Rondonia, Rummel et al. [2002] compared the quantity of NO emitted from the soil and fluxes above the forest canopy measured by eddy covariance and found that 92% of soil-emitted NO was consumed in the canopy during the daytime. At night, when rates of vertical mixing of 0 3 into the forest canopy were much slower, about 8% of the. NO was consumed, resulting in a daily reduction flux of NO from soil to atmosphere of 52%. Other studies suggested canopy consumption in the range of6Q-.81 % [Balrwin
342
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
et al. 1990' Jacob and WofSy, 1990]. On the basis of these " . -1 -1 results and on an annual soil emission of 1.4 kg N ha a from a forest in Rondonia, Neill et al. [2005] estimated that the net NQ flux to the atmosphere from forest would be 0.7 kg N ha- 1 a-I.
Table 2. Summary of Reported Average Daily Fluxes for Dry and Rainy Seasons and Annual Fluxes Estimation of NO and NzO From Natural and Managed Ecosystems in the Cerrado Region"
Ecosystem
2.3. N20 and NO Fluxes From Cerrado Soils Under Native Vegetation
Most of the cerrado vegetation occurs on dystrophic and clayey but well-drained soils. Drainage of water is very rapid even during the wet season because of the formation of soil microaggregates. Organic matter content is low and concentrated in a thin surface layer. Because cerrado soils are mainly well drained and aerated, nitrification is a more important pathway for N oxide production, and conditions favorable for denitrification are rare [Pinto et al., 2002; Varella et al., 2004; Pinto et al., 2006]. Frequent fires in the cerrado region provoke losses of N fi'om the ecosystem, and this is reflected in low soil N availability [Pivello and Coutinho, 1992; Bustamante et al., 2006]. Nitrification rates in celTado soils are very low [Nardoto and Bustamante, 2003], and only rarely does N03- production exceed the demand by microorganisms and plant roots. The combination of low nitrification rates and the dominance of NH4+ in the inorganic N pools contribute to low soil emissions of N oxide gases in the cerrado region. Fires have an impact in the trace gases emission not only as a consequence of biomass combustion but as a result of emissions by soils after fires. Pinto et al. [2002] measured soil fluxes of NO and NzO in two cerrado vegetation types in plots near Brasilia characterized by differing amounts of woody plant canopy cover that were either burned every 2 years or protected from fire. The two vegetation types were cerrado in the strict sense (20-50% woody plant canopy cover) and campo sujo (open and grass dominated). NO fluxes varied according to the propOltion of woody cover and fire regime. Annual fluxes ofNO from the fire-protected areas were 0.1 kg N ha- I for the campo sujo and 0.4 kg N ha- I for the cerrado in the strict sense. An increase in the annual soil-atmosphere flux of NO with fire was observed only for the campo sujo (0.5 kg N ha- I) [Pinto, 2003] (Table 2). NO emissions increased after burning(1.0 ng N cm-2 h- I), but the flux diminished quickly to levels even lower than prefire levels. The timing of burning (early dry season, middle dry season, or late dry season burning) had little influence on soil NO emissions. In the same experiment, soil moisture was the critical control on soil-atmosphere NO fluxes. Large NO fluxes were observed immediately following precipitation events that broke long droughts. NO emissions increased to 1.0 ng N
BUSTAMANTE ET AL.
Cerrado Cerrado burned plus 45 days Cerrado burned plus 17 days Cenado
Reference
Annual N-NO Flux
Anderson and Path [1998] Anderson and Path [1998] Anderson and Path [1998] Salllinez
NzO Flux
DS
RS
Annual
UO UD UD 0.3--0.5
[1999] Pasture
Salllinez
Soybean between lines Soybean line
Salllinez
0.5
[1999]
1.1
[1999]
1.6
Sa III inez [1999]
Pinus
0.4
Salllinez [1999]
Eucalyptus
0.3
Salllinez [1999]
Campo sujo
0.1
UD
UD
UD
Pinto [2003]
0.5
UD
UD
UD
Pinto [2003] Pinto [2003] Pinto [2003]
0.4 0.4 O.Olb
UO UO UD
UO UO UD
UD UD UO
Pinto [2003]
O.Ol b
UD
UD
UD
Pinto [2003] Pinto [2003]
0.04 b O.OOb
UD UD
UO UD
UD UO
Fel'1landes
0.3
Pinto et al. [2002]
Campo sujo burned Cerrado Cerrado burned Fertilized pasture Legume-grass pasture Young pasture Traditional pasture Corn
[2008]C Bean
Fel'1landes
0.3
[2008]" Soybean,
Fel'1landes
0.2
[2008]C Cotton
Fel'1landes
0.7
[2008]C aAverage daily fluxes are given in mg N m-z d- 1. Annual fluxes estimations are given in kg N ha-1 a-I. Abbreviations are DS, dry season; RS, rainy season; and UD, under detection limit. bFlux is integrated for the wet season. cFlux is integrated for one crop cycle. Some crops are cultivated twice a year.
2 I cm- h- with the first rains in unburned cerrado in the strict sense and to 1.9 ng N cm- 2 h- I in burned cerrado in the strict sense. Wetting of c\~ry' soil in cerrado causes an increase in NO emissions off factor of 10 or more but that pulse was short-lived. Flu~\is fell to background values within 3 days of a wetting pulse [Pinto et al., 2002; Varella et al., 2004]. In the cerrado, N20 production is limited both by low N availability and by dry conditions, and fluxes of N20 are generally very low «0.6 ng N cm-z h- 1) [Pinto et al., 2002; Varella et al., 2004] regardless of the vegetation type or fire regime. Even the increase of soil water content in the wet season is not sufficient to stimulate high NzO production. 3. EFFECTS OF LAND USE CHANGES ON N20 AND NO EMISSIONS FROM SOIL TO THE ATMOSPHERE
343
One Large-Scale Biosphere-Atmosphere Experiment in Amazonia study approached this topic through measurement of soil-atmosphere fluxes of N20 and NO on two soil types (clay Oxisol and sandy loam Ultisol) over 2 years (2000-2001) in both undisturbed forest (as discussed in section 2.1) and recently logged forest. Studies were conducted in the Tapaj6s National Forest, near Santarem, Para, Brazil, where a demonstration logging project used reduced-impact forest management techniques [Keller et al., 2005]. Keller et al. measured fluxes in logged areas and compared excess fluxes by subtraction of a background flux from undisturbed forest. Logging increased emissions of N 20 and NO from 30 to 350% depending upon soil conditions. Along with the hypothesized effects of changes in nutrient and water cycles discussed above, the authors found that compaction of soils by heavy machinery in skid trails and log storage decks tended to augment emissions N20 and NO. While significant effects were measurable locally, Keller et al. cautioned that logging-induced fluxes were unlikely to increase regional emissions ofN 20 by more than about 6%.
Deforestation causes environmental changes resulting in changes of trace gas emissions. Il11ll1ediately following clearing, the elimination ofthe plant sink for nutrients results in a pulse of nutrient availability, including N, in soils and streams. Emissions depend upon the prior site fertility and 3.2. Emissions ofN Oxides With SecondGlJI the rate of vegetation regrowth after disturbance.' Clearing Forest Succession natural vegetation [Luizfio et al., 1989; Keller et al., 1993; Davidson and Kingerlee, 1997], burning [Levine et al., In the Amazon ~asin, about 30-50% of cleared land is 1996; Neff et al., 1995; Serc;a et al., 1998], fertilization of in some stage of secondary forest succession following agagricultural lands [Matson et al., 1996; Mosier and Delgado, ricultural abandonrpent [Hirsch et al., 2004]. Davidson et 1997; Veldkamp and Keller, 1997], intensive cattle ranching al. [2007] demonstrated through the comparison of forest [Lima et al., 2001; Primavesi et al., 2004], and increasing chronosequences (stands ranging in age fi'om 3 to 70 years dominance by legume species in areas under secondary suc- and remnant mature forests in eastern Amazonia-Para) that cession after land conversion [Erickson et al., 2002; David- young successional forests growing after agricultural abanson et al., 2007] have been identified as causes of increasing donment on highly weathered lowland tropical soils exhibN2 0 and NO emissions from tropical regions. ited conservative N cycling properties much like N-limited forests on younger soils in temperate latitudes. As second3.1. Effects ofLogging on N 20 and NO Emissions ary succession progressed, N cycling properties recovered From Soils with increasing availability of soil nitrate and a concomitant decrease of ammonium concentration. The dominance of a In the Brazilian Amazon region, selective logging removes conservative P cycle typical of mature lowland tropical fortimber volume in the range of only 20-60 m3 ha-I (about 3-9 ests reemerged. Because of the successional shifts in N:P trees ha-I); however, current practices result in high levels of cycling ratios with forest age, soil emissions of N 0 were 2 collateral damage to the forest canopy and soils [Verissimo et initially low and then increased gradually. N2 emissions 0 al., 1992; Pereira et al., 2002; see Asner et al., this volume]. increased with forest age at clayey and sandy soils, although Felling a tree transfers fresh leaf and root nutrients to the the more clay-rich site exhibited higher emissions. These soil, causing pulses of decomposition [Lodge et al., 1991; results showed that increasing emissions ofN 0 as the suc2 Scatena et al., 1996]. Loss of nutrient and water uptake by cessional forests age can be attributed to the gradual recuroots added to the forest floor leads to wetter sites and greater operation ofN cycling processes. nutrient leaching [Parker, 1985; Brourver, 1996]. In Guyana, nitrate losses from selective harvest varied in proportion to 3.3. Conversion ofForest to Pasture and Crops the area of soil disturbance [Brouwer, 1996]. The changes in nutrient and water cycles provoked by Following a site disturbance such as deforestation N logging affect the soil-atmosphere exchange of trace gases. availability in the soil often temporarily increases, cau~ing
344
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
significant increases in emissions of NO a~d NzO. Fire is the main tool for clearing forest.in Amazoma [Setzer et al' 2 1998; Kauffman et al., 1992]. An average o~ 19,000 ~m of forests
BUSTAMANTE ET AL.
tern to that described above for NzO emissions, except that NO is relatively more impOltant where there is a more pronounced dry season along the eastern and southern flanks .of the Amazon Basin [Davidson et al., 2001]. Nitric oxide emISsions did not vary with pasture age in one study in Rondonia, and mean annual NO emission from pastures was 0.2 kg N ha- I a-I compared to 1.4 kg N ha- I a-I in nearby forest [Neill et al., 2005]. This study did not measure NO fluxes ~om pastures less than 9 years old, and Neill et a1. recogmzed that younger Pilstures might have greater NO emissions assuming behavior similar to NzO. Kirkman et al. [2002] comp.a~·ed forest and old pasture soils in Rondonia during two trans.ItJon season periods (wet-dry and dry-wet) and obser:,ed sOlI-atmosphere NO fluxes fi'om pastures that were 9 tJ~es lower than old-growth forest fluxes under similar soil mOIsture and temperature conditions. They attributed this ~attern to the combination of a diminished velocity for the soIl N cycle and lower effective soil NO diffusion rates in the pasture soils.. Over the past 40 years, large areas of cattle pasture 111 Amazonia have suffered declining productivity as a result of nutrient losses. Practices adopted by ranchers to restore productivity to degraded pastures have the potenti~l to alter soil N availability and gaseous N losses from soIls. These practices involve varying amounts of tillage com?ined with fertilizer and herbicide applications, and the plant111g of cash crops prior to the planting of forage grasses. Passianoto et al. [2003] reported on the emissions of NzO and NO from the first 6 months of three restoration treatments for pastures in central Rondonia. The treatments were (1) control pasture; (2) conventional tillage followed.~y planti~~ offorage grass (Brachiaria brizantha) and fertIhzer addItJon (42 kg
40
cut~i ';">,
30
(1)
';"(1)
20
0
z 0 0 000
0 0
o
o
0
JUN 1992
JAN
: III
a~~.~ :
10
OJ
:. :
I
~
-
:, , b~i
"0
.c
i~Burn
JUN 1993
JAN
III:
:0
..:,, 0 q.,,, , ,
JUN 1994
JAN
•
0
tt 0
JUN 1995
0
+
0 III
0
JAN
JUN
JAN
1996
F' 2 NO emissions (plus standard error) from the reference forest (open circles) and a new. 3 .ha pasture (solid !gure ,2 circles) that was establ'IShed"'101. tlIe study in 1994 . The error bars for the forest measurements are wlthm symbols. . t th Meadata . t 1abeled "a" were taken surements to generate the data pom . 2 days before. burn, and the measurements to geneta e e point labeled "b" were taken 3 days after the burn. Repnnted from Melillo et al. [2001].
I N ha- ); (3) no tillage with herbicide treatment followed by two plantings, the fir~t being rice, feltilized with 33 kg N I ha- , followed by fl,torage grass. The cumulative NzOand NO emissions ov~t the first 2 months from the tillage regime 1 (0.94 kg N-NzOjhaand 0.98 kg N-NO ha- I) were much .! higher than the'releases from either the no-tillage/herbicide regime (0.64 kg N-NzO ha- I and 0.72 kg N-NO ha- I) or the control pasture treatment (0.04 kg N-NzO ha- I and 0.12 kg N-NO ha-1). Tillage increased soil NH/ and N0 3- pools, while these pools remained relatively constant in the control and no-tillage treatments. Cumulative rates of net N mineralization and net nitrification during the first 6 months after treatment varied widely but were highest in the tilled treatment [do Carmo et al., 2005]. In this experiment, while emissions of NO and NzO increased with tillage and N fertilization, there were no clear relationships among rates of N fertilizer application, net mineralization, net nitrification, NO, NzO, and total N oxide emissions. The magnitude of the increased N oxide fluxes differed based on the timing of fertilizer application relative to the presence of plants and the magnitude of plant N demand. Emissions of N oxides decreased with the use of restoration sequences that minimized reductions In pasture grass cover [do Carmo et al., 2005]. A few authors have attempted to estimate the regional effects of land use changes on the fluxes of N oxides. Potter et al. [1998] using the NASA-CASA ecosystem model estimated a total flux of 0.5 Tg NzO-N a-I from Brazilian Amazonia. In a subsequent study [Potter et al., 2001], a version of this model was applied to two forest sites, located in the Brazilian states of Rondonia and Para, differing in tenus of seasonality of rainfall, length of the annual dly period, and soil properties. The measured fluxes of soil NzO for forests closely matched the proposed models for the forest in Para but not for that in Rondonia. This result suggested that algorithms controlling nitrogen trace gas fluxes, particularly in relatively sandy tropical soils require further development. Melillo et al. [2001] used a constant value of 1.4% of modeled N mineralization to estimate NzO emission. With this approximation, Amazon Basin-wide (area of 6.9 x 10 6 kmz) emissions fi'om 1978 to 1995 averaged 1.3 Tg NzO-N a-I with 0.8 Tg NzO-N a-I for the Brazilian pOltion of the Basin. 3.4. Conversion ofCerrado to Pasture and Crops
Planted pastures (mainly Brachiaria spp.) are the most extensive land use in the cerrado, and with an area of approximately 50 million ha they occupy nearly one fourth of the biome [Sano et al., 2000]. In well-managed pastures on clayey soils in the cerrado region, productivity and longterm soil C stocks can surpass levels for native vegetation
345
[Corazza et af., 1999; Bustamante et al., 2006]. On the other hand, poor management practices, especially overgrazing, lead to pasture degradation after a few years. According to Oliveira et al. [2004], degraded pastures in the tropical region of Brazil occupy 25 million ha. Although few studies have been conducted on N oxide emissions from pasture soils in the cerrado region (Table 2), they suggest a decrease in emissions with pasture age as presented in section 3.3 for the Amazon region. Varella et al. [2004] measured fluxes of NO in a 20-year-old B. brizantha pasture in central Brazil and found aJIDual NO emissions of only 0.1 kg N ha- I a-I. Emissions of soil NzO were below the detection limit «0.6 ng N-NzO cm-z h- I = 0.5 kg N ha- I a-I). Saminez [1999] measured annual NzO soil emission of 0.5 kg N ha- I a-I in both native cerrado and in a 5-year-old Andropogon gayannus pasture. Because planted pastures in the cerrado tend to lose productivity with time, ranchers use a variety of agronomic approaches including tillage and fertilization to refonu their pastures. Pinto et al. [2006] studied the effects of pasture reformation on N dynamics (net N mineralization, net nitrification, available inorganic N and NO, and NzO gas fluxes). The study focused on three areas of cerradao (closed savanna woodland) converted to B. brizantha in 1991 that exhibited characteristics of d~gradation after 9 years. In 1999, different restoration tJ'eatments were tested: (1) fertilization (60 kg 1 N ha- and 12 kg P ha- 1); (2) association of grasses and legumes (B. brizantha and Stylosanthes guianensis) with addition of 12 kg P ha-1, and (3) a traditional plot without management. These treatments were also compared with a fourth area of cerradao converted to B. brizantha pasture in 1999 (young pasture). Measurements were carried out during the wet season of 2001-2002. Ammonium was the predominant inorganic N form in the soils. All plots showed high variability of soil N gases emissions. Peak emissions of NO (3.6 ng N-NO z cm- h- I ) and NzO (6.7 ng N-NzO cm-z h- I ) were probably caused by cattle urination and defecation that led to an irregular distribution of organic residues. Despite these peaks ofN gas emissions, overall nitrogen oxide emissions were low and only amounted to 0.03 kg N ha-I during the JanuaJy to April growing season. A water addition experiment during the dly season (September 2002) indicated that the transition fi'om dly to wet season might be an important period for the production of NO in the young pasture with fluxes ranging fi'om 6.8 ng N-NO cm-z h- I within an hour after watering to 3.0 ng N-NO cm-z h- I 2 days after the treatment. In 2006, approximately 14 million ha of cerrado were cropped with soybean, maize, cotton, common bean, and rice (see. the Companhia Nacional de Abastecimento Web site at http://www.conab.gov.br). Soybean production catalyzed the agricultural expansion in the cerrado during the last 2
346
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
decades. This crop occupies more than 6 million ha in the plateau regions of the cenado. However, continuous monoculture of soybeans coupled with inadequate tillage practices has caused severe erosion and soil degradation [Resck et at., 1991]. Because of the problems associated with conventional systems, low-tillage agriculture was introduced. In 1989-1990, low-tillage agriculture in Brazil occupied an area of '1 x 106 ha, and by 2004 it had expanded to 23.6 x 106 ha (see the Federac,;ao Brasileira de Plantio Direto na Palha Web site at http://www.febrapdp.org.br). Of this total low-tillage area, about 8 x 10 6 ha were in the cerrado region [Duarte et al., 2007].
Few studies track the effects of the conversion of cerrado to row crops for nitrogen oxide emissions. Work done to date has shown that NzO emissions are not great. Results from a soybean-corn crop rotation showed that NzO flux increased modestly from 0.8 ng NzO-N cm-2 h- I to 2.5 ng NzO-N cm- z h- I under soybean cultivation [Sam inez, 1999] compared to other published accounts of elevated emissions from tropical agricultural soils. Low emissions may be explained by the relatively dry cenado climate that does not favor large NzO emissions. Metay et al. [2007] studied two 5-year-old systems during a cropping cycle in Goiania, Goias: tillage with disking in the first 15 em and no tillage with a directsowing mulch-based crop system and an additional cover crop. Emissions of NzO were vely low, and no significant difference between treatments was observed. Monthly mean NzO emissions increased exponentially with monthly mean (WFPS) in the 0- to 10-cm layer. NO emissions may be quantitatively more important for N cycling than NzO emissions in croplands in the cenado region. Carvalho et at. [2006] compared NO fluxes in a cornfield under no-tillage and tillage systems fertilized with urea. Inigation was perfolTlled immediately following the fertilization. No significant differences in nitrogen oxide fluxes were found between plots under tillage and no-tillage systems. The response to fertilization was rapid. Following broadcast fertilization, high NO fluxes were measured 15 min (SA ng N cm-z h- I ) and 3 days after N feIiilization (4.8 ng N cm-z h- I ), but fluxes decreased to 1.2 ng N cm- z h- I after 5 days. Unfortunately, the database related to N oxide emissions in the cerrado is even more restricted than in Amazonia. Most of the data are from the core region of the cerrado (state ofGoias and the federal district) and these studies have focused on the most common soil types in the cenado region (Latossolos VelTllelho-Escuro and Vermelho-Amarelo in the Brazilian system and Acrustox in the U.S. Department of Agriculture taxonomy system). The diversity of crop systems and management practices in the region limit extrapolation.
4. METHANE FLUX BETWEEN UPLAND ECOSYSTEMS OF AMAZONIAN FOREST AND CERRADO AND THE ATMOSPHERE
4,1. Exchange of CH4 Between Uptand Soils and the Atmosphere in Amazon Forests and the Cerrado In general, well-drained soils consume atmospheric CH4, and by convention these fluxes are shown as negative. Early studies from well-drained upland forest soils in Amazonia confolTll to global trends [Keller et at., 1983, 1986; Steudler et al., 1996]. On the basis of the results of 22 studies fi'om humid tropical forests, Potter et at. [1996] repOlied an average flux of CH4 from the atmosphere to soil of -3,8 ± 0.6 kg CH4 ha- I a-I. Several studies report CH4 consumption during both wet and dly seasons in forest soils in the municipalities of Santarem and Beltena (Para) although all observe occasional net production of CH4 during the wet season [Davidson et at" 2004; Wick et al., 2005; Keller et at" 2005]. Keller et al. [2005] compared soil texture effects CH4 fluxes and found annual average (plus/minus standard error) fluxes of-0.3 ± 0.2 and'-O.l ± 0.9 mg CH4 m-z d- I on a clay-textured Oxisol and -1.0 ± 0.2 and -0.9 ± 0.3 mg CH4 m-z d- I on a sandy loam Ultiso1 for two subsequent years (2000 and 2001) with greater variability ofCH4 fluxes in the Oxisol than in the Ultisol, especially during the wet season. The size of the regional CH4 sink was calculated by Davidson and Artaxo [2004], who estimated the net soil uptake of CH4 as 1 to 3 Tg CH4 a-I. In general, studies of soil CH4 in Amazon forests demonstrate that soil CH4 consumption is limited by gas diffusivity [Verchot et at., 2000]. Net C~ production has also been associated with soil respiration possibly because high rates of soil respiration create anaerobic microsites resulting in CH4 production [Verchot et al., 2001]. Manipulations of soil moisture confirm that moisture controls CH4 fluxes. In an experiment investigating the effects of drought, in Amazonian forest, soil consumption of atmospheric CH4 was increased on soils affected by drought by a factor of >4 [Davidson et at., 2004]. Vasconcelos et al. [2004] reported significantly z lower CH4 fluxes in the dly season (-0.3 ± 0.1 mg CH4 mI z d- I ) than in the wet season (0.1 ± 0.1 mg CH4 m- d- ) in a secondmy forest and that irrigation during the dly season increased CH4 efflux in relation to control plots (0.2 ± 0.4 and -0.5 ± 0.2 mg CH4 m-z d- I , respectively).
BUSTAMANTE ET AL.
ers, 199.4], soil ~ompaction resulting from forest to pasture con~er~I~n
restncts spil-atmosphere gas exchange and leads to dimIlllshed CH4"consumption. Along with compaction, other factors that,/avor CH4 emission from pasture soils in th~ ~mazon re~6n include higher pH values, lower availabIlIty of alternate electron acceptors such as N-NO - and 'bl 3 , pOSSI Y greater availability of organic matter [Feigt et at. 2001]. . , Methan~ fluxes between soils and the atmosphere were ~easured m two tropical forest-to-pasture chronosequences
m the.state of Rond6nia, Brazil [Steudter et at., 1996]. Forest sOll.s alwa~s showed a net consumption of atmospheric CJ:I4 wIth maXImum uptake rates in the dry season. Pasture soIls consumed atmospheric C~ during the dly season but at lower rates than those in the forests. When soil moisture increased in the pasture soils, they became a source of CH4 to the atmosphere. In~egrated over the year, forest soils were a net sink of approxImately -1.7 mg CH4 m-z d- I , while pastures were a net source ~fabout +1.0 .mg C~ m-z d- I . Thus forest-to-pasture converSIOn resulted m a net change in the soil flux of CH 4 from the soil of almost 2.7 mg CH4 m-z d- 1. Logging in Brazilian Amazonia annually affects an area almost as large as the area clear-cut for conversion to agriculture and pasture [Asner et at., 2005]. Logging operations ?epend on heavy machinely that compacts forest soils, leadmg to a change from CH4 consumption to CH4 production [Keller et at., 2005]. Ground damage is limited to only a portion of the logged area that depends upon the intensity of harvest and the quality of the harvest management [Pereira
CH4 Flux
Path et aI, [1995]
Anderson and Path [1998]
Saminez [1999]
4,2. Effects ofLand Use Changes on the Soil-Atmosphere Exchange of CH4
Soil aeration is the primary conh'ol on CH4 flux. As observed in Costa Rica [Keller et at., 1993; Keller and Rein-
et al., 2002]. Critically, CH4 production is dominated by small. areas know~ as log decks that are used to stage logs for truc~mg to sawmIlls. These log decks occupy approximately 1-2 Yo of the logged area dependent upon management, and they become waterlogged in the wet season because the co~pacted soils do not readily drain. Waterlogged decks emI~ CH4 at rates similar to tropical swamps while they are lllundated, accounting for >80% of the CH 4 emitted in logged areas. As pointed out for N oxide emissions, there are few data on CH4 emissions from cerrado soils (Table 3). In the first measurements of CH4 flux made on cenado soils [Anderson and Poth, 1998], newly burned cerrado had a CH4 flux of -4.~ mg CH4 m-z d- I that increased to -2004 mg CH4 m-z d- I at sItes burned 30 days earlier. However, in an area burned 1 year before, CH4 uptake by soils had disappeared. CH 4 fluxes were not correlated with any soil chemical parameters measured [Poth et ai" 1995]. Poth et al. hypothesized that the absence of uptake by unburned soils would indicate the p:esence of a ~alance ~etween the source of soil CH4, possIb.ly t~e foragmg termIte community, and the sink for CH4, OXIdatIOn by the soil microbiota [Seiter et al., 1984]. San~lueza [2007] has presented limited measurements indicatmg that savanna grasses in the Venezuelan llanos produced small amounts of GH4. Saminez [1999] compared fluxes of CH4 from Oxiso1s under cerrado native vegetation, 5-yearold pas.ture, soybe~n-corn rotation, and eucalyptus and pine plantatIOns. CH4 consumption occurred under all land uses during both wet and dry seasons although values during the wet season were lower (Table 3). Low but positive C~
Table 3. SummalY of Reported Average Daily Fluxes for DIy and Rainy Seasons and Annual Flux Estimation of CH F' Natural and Managed Ecosystems in the Cerrado Region 4 10m
Reference
Metay et al. [2007]
347
Vegetation Cover! Land Use
DIy Season (mg Cl-Lt m-2 d- I )
cerrado cerrado burned plus 2 days cerrado bumed plus 30 days celTado celTado burnt plus 45 days cerrado burnt plus 17 days cerrado pasture soybean between lines soybean line Pinus Eucalyptus rice, tillage (15 em) rice, direct seeding mulch
1.12 -3.93 -20.22 -1.68 -1.01 -1.03
Rainy Season (mg Cl-Lt m-2 d- 1)
Annual (kg CH4 ha- 1 a-I)
-1.24 -0.88 -0.69 -1.09 -1.17 -1.39
-5.5 and -6.0 -4.0 -3.3 -2,3 -4.8 -5.6 0,54 0.33
-----------,.-------------~---·~I
348
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
fluxes were measured by Metay et al. [2007] in a cultivated cerrado area under low till (0.33 kg CH4 ha- I a-i) and tillage (0.53 kg CH4 ha- i a-i).
4.3. Methane Production by Ruminant Livestock
Brazil has the largest commercial cattle herd in the world, and it is second only to India in total cattle population [Lerner et al., 1988]. The cattle population of Brazil is increasing rapidly, reaching 207 million head in 2005, with a concentration of animals in the center west region (mainly cerrado, 72 x 106 head) and north region (Amazon, 42 x 106 head) (see the Banco de Dados Agregados Web site at http://www. sidra.ibge.gov.br). One result of the cattle population increase is that emissions from livestock have become a significant source of atmospheric CH4. Ruminants generate CH4 as a by-product of the anaerobic digestion offood in the rumen. CH4 is released mainly by exhalation and eructation. The CH4 production rate is affected by factors such as quantity and quality of feed, body weight, age, and exercise and varies among animal species as well as among individuals ofthe same species [Mosier et al., 1998]. CH4 emission from dairy cattle fed with tropical grasses (between 121 and 147 kg CH4 a-I per animal) is greater than emissions from dairy cattle in temperate climate (between 100 and 118 kg CH4 a-I per animal) probably because of differences in forage quality [Primavesi et al., 2004]. The estimate of CH4 emissions from all cattle raised in Brazil (including beef and dairy cattle) for 1994 was 9.77 Tg CH4 a-I (96% from enteric fermentation and 4% from animal wastes) with beef cattle contributing 81 % of the total. Of this total, the center west region (most of the cerrado) makes the greatest contribution (3.09 Tg CH4 a-I), representing 30% of the total [Lima et al., 2001]. 4.4. Atmospheric CH4 Measurements and CH4 Sources
Evidence from satellite sensors as well as ground- and aircraft-based in situ sampling suggests that the CH4 emissions of the Amazon region have been underestimated. Frankenberg et al. [2005] compared total column CH4 concentrations retrieved using the Scanning Imaging Absorption Spectrometer for Atmospheric Cartography insttument on the European ENVISAT satellite with a global atmospheric chemistry and transport model. On the basis of severalmonths of observations from 2003, they concluded that tropical forest-derived CH4 emissions have been underestimated by at least 30 Tg CH4 a-I representing more than 5% of global annual emissions. In a subsequent compilation of data for 2003 and 2004, Frankenberg et al. [2006] confirmed anomalous column abundances of CH4 over
Amazonia compared to accepted emISSIOn inventories. These anomalies can be accommodated in part with an increased wetland source in Amazonia of 41 Tg CH4 a-I as suggested by an inverse model analysis [Bergamaschi et al., 2007]. However, this wetland source is substantially greater than the most complete bottom-up estimate of Amazon Basin wetland CH4 emissions (29 Tg CH4 a-I [Melack et al., 2004]). Measurements of CH4 mixing ratios in the canopy layer of three forest sites across Brazilian Amazonia (Caxiuana National Forest in Para, Cuieiras Reserve in Amazonas, and Sinop in Mato Grosso) in both wet and dry seasons showed net CH4 emissions ranging from 2 to 21 mg CH4 m-z d- I [do Carmo et al., 2006]. While the measurements are sparse, they suggest an unknown source of CH4 from upland forests (as opposed to wetlands) of between 4 and 38 Tg CH4 a-I if extrapolated over the Amazon forest area. Estimates of CH4 emissions based on episodic aircraft sampling fi'om 2001 through 2006 by Miller et al. [2007] averaged 35 ± 23 mg CH4 m-z d- I and 20 ± 17 mg CH4 m-z d- I between the Brazilian coast and the interior Amazonian cities of Santarem and Manaus, respectively. These are also greater than the estimates of average CH4 emissions for the wetlands of the region (16 mg CH4 m-z d- I [Miller et al., 2007]). What sources could account for greater C~ emissions in Amazonia compared to bottom-up inventories? While recently, laboratory experiments have indicated that aerobic production of C~ by plants may playa significant role in the atmospheric budget of C~ [Keppler et al., 2006], the early estimates may have exaggerated the magnitude of the source. Limited evidence from field smdies in Venezuelan savannas supports the plant source although with a smaller magnimde [Sanhueza, 2007]. The extt'apolation of the laboratOly smdy has been controversial, and several analyses indicate that even if the laboratOly results are reliable, the global impOliance of plants as a CH4 source has been exaggerated [Kirschbaum et al., 2006; Hourveling et al., 2006; Ferretti et al., 2007]. A recent laboratory smdy using l3C-labeled plants found no evidence for plant CH4 emission above trace levels that would be globally insignificant [Dueck et al., 2007]. It remains difficult to justify a plant source of C~ that could account for excess C~ observed over Amazonia. Other known sources that may be underestimated in current inventories include biomass burning (especially the manufacmre of charcoal) and tennite-derived C~ [Christian et al., 2007]. 5. CONCLUSIONS Emissions of NzO and NO from soils of mature Amazonian forests range from about 1 to 7 kg N ha- I a-I and 1 to 3 N ha-I a-I, respectively. Environmental controls are mainly
BUSTAMANTE ET AL.
the variability of rainfall and soil texmre as influences on soilmoismre content with higher fluxes on clayey soils. In addition, fine-root d:Y'llamics affect the production and emission of tt'ace gase~( especially in tropical rain forests that are characterized by.ill large biomass and rapid turnover of fine roots. The proportion of N oxides emitted from soil to atmosphere is also dependent on the interaction with the vegetation canopy. Smdies indicated a daily reduction flux of NO from soil to atmosphere of about 50%. Land use change does not necessarily lead to a greater emission of greenhouse gases. Logging increased emissions ofNzO and NO from 30 to 350% depending upon soil conditions, but logging-induced fluxes were unlikely to increase regional emissions of NzO by more than about 6%. Conversions of tropical forests to cattle pasmres do not cause 10ng-teIID increases in the contribution of soil emissions to atmospheric NzO or NO. Lower N oxide emissions measured from pasmres compared to old-growth forests are related to a progressive decline in N availability with pasttlre age combined with strongly anaerobic conditions in some pasmres during the wet season. Compared to the primaIy forest, predicted changes in soil nitrogen cycling lead to a doubling in annual emissions of NzO gas during the first year following deforestation with lower emissions thereafter (Figure 3). After pasmre abandomnent and with forest regrowth soil emissions ofNzO were initially lower and then increased gradually. Comparisons of N oxide emissions from agriculmral areas indicated that soil preparation had an effect, with higher fluxes under tillage. Smdies of intensively managed agriculmre in Amazonia are still rare, but the evidence so far accumulated indicates that namral ecosystems often emit more nitrogen oxides than agroecosystems. In the case of the cerrado, the combination of low nitrification rates and the dominance of NH4+ in the inorganic N pools contribute to low soil emissions of N oxide gases. Large NO fluxes have been observed immediately following precipitation events that broke long drought periods, but these pulses are short-lived and contribute only slightly to annual emissions. NO emissions increased after burning, but again the flux remrned quickly to prefire or even lower levels. Soilmoismre and vegetation type were more important in controlling NO fluxes than fire regime. In the cerrado, NzO production is limited both by low N availability and by dry conditions; fluxes ofNzO are generally very low regardless of the vegetation type or fire regime. Although few smdies have been conducted on N oxide emissions from pasmre soils in the cerrado region, the existing data suggest a decrease in emissions with pasmre age as observed for Amazonia. The conversion of cerrado to row crops does not contribute significantly to NzO emissions, but NO emissions appear to be more important. However, the association ofN
(a)
I
FOREST
Nitrification NO
N,O
349
Denitrification NO
YOUNG PASTURE
OLD PASTURE
(b)
I
CERRADO
? YOUNG PASTURE
? OLD PASTURE
? AGRICULTURE/IRRIGATION
? Figure 3. Fluxes of inorganic nitrogen for (a) moist forest and (b) cerrado environments schematically represented by a modified hole-in-the-pipe model. The relative widths of the horizontal arrows to the left of the boxes represent nitrification and denitrification processes and mineralized N or feliilizer N entering into the soil system. Horizontal arrows on the far right represent the presumed losses of N z through denitrification. These losses have not been well quantified and may be negligible in the cerrado as noted by the question mark. The relative widths of the vertical arrows represent the gaseous losses of NO and NzO from the nitrification and denitrification processes. Where no arrows are shown, losses are below the measurable threshold.
350
BUSTAMANTEET AL.
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
fertilization with irrigation could lead to higher N20 emissions (Figure 3). In general studies of soil CH4 in Amazonian forests demonstrate ~hat soil CH4 consumption is limited by gas diffusivity. On an annual basis, undisturbed forest soils always showed a net consumption of atmospheric CH4 with maximum uptake rates in the dry season. Pasture soils consumed atinospheric CH4 during the dry season but at lower rates than those in the forests. When soil moisture increased in the pasture soils, they beca~e a source of CH4 to the atmosphere, and thus forest-to-pasture conversion resulted in a net change in the direction of the flux of C~ from the soil. Measurements of CH4 mixing ratios in the canopy layer of three forest sites across Brazilian Amazonia in both wet and dry seasons showed evidence for net CH4 emission in upland forests. This indicates the existence of an unknown source of CH4 from upland forests. As pointed out for N oxide emissions, there are few data on CH4 emissions from cerrado soils, but they indicated a predominance of CH4 consumption over emission under different land uses during both wet and dry seasons. Davidson and Artaxo [2004] have summed the 100-year global warming potentials (GWP) of the annual sources and sinks of CH4, N20, and CO 2, indicating that the Amazonian forest-river system currently may be nearly balanced in terms of net GWP for these biogenic atmospheric gases. Unfortunately, large uncertainties remain for these estimates. There is still a lack good predictive models for regionalization of site-specific data. The current models lack geographic data for parameterization and testing and, more importantly, basic theory on the controls of emissions from the ecosystem to the atmosphere. From a regional perspective, the unknown source ofCH4 fi'om upland forests is the greatest unceliainty. Improvement of models for the N oxides and CH4 fluxes for Amazonia and the cerrado still depends upon gathering more data from sites with different edaphic and vegetation characteristics, more widely distributed across two'vast biomes. Aclmowledgment. Mercedes Bustamante and Dulce Alves Silva acknowledge support from the LBA-ND07 project.
REFERENCES Alves, D. S., D. C. Morton, M. Batistella, D. A Robelis, and C. Souza Jr. (2009), The changing rates and patterns of deforestation and land use in Brazilian Amazonia, Geophys. Monogr. Ser., doi: 10.1029/2008GM000722, this volume. Ammann, M., R. Siegwolf, F. Pichlmayer, M. Suter, M. Saurer, and C. Brunold (1999), Estimating the uptake of h'af'fic-derived N 0 2 from 15N abundance in Norway spmce needles, Decologia, 118, 124-131.
Anderson, J. C., and M. A. Poth (1998), Controls on fluxes oftrace gases from Brazilian cerrado soils, J Environ. Qual., 27, 11171124. Asner, G. P., D. E. Knapp, E. N. Broadbent, P. 1. C. Oliveira, M. Keller, and 1. N. Silva (2005), Selective logging in the Brazilian Amazon, Science, 310, 480-482. Asner, G. P., M. Keller, M. Lentini, F. Meny, and C. Souza Jr. (2009), Selective logging and its relation to deforestation, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000723, this volume. Bakwin, P. S., S. C. Wofsy, S.-M. Fan, M. Keller, S. E. Trumbore, and J. M. Da Costa (1990), Emission of nitric oxide (NO) from h'opical forest soils and exchange of NO between the forest canopy and atmospheric boundary layers, J Geophys. Res., 95, 16,755-16,764. Bergamaschi, P., et al. (2007), Satellite chartography of atmospheric methane from SCIAMACHY on board ENVISAT: 2. Evaluation based on inverse model simulations, J Geophys. Res., 112, D02304, doi:10.102912006JD007268. Bosquet, P., P. Ciais, and J. B. Miller (2006), Contribution of anthropogenic and natural sources to atmospheric methane variability, Nature, 443, 439-443. Brouwer, L. C. (1996), Nutrient Cycling in Pristine and Logged Tropical Rain Forest: A Study in Guyana, Tropenbos Guyana Ser., vol. 1, Elinkwijk Press, Uti'echt, Netherlands. Bustamante, M. M. C., M. Corbells, E. Scope1, and R. Roscoe (2006), Soil carbon storage and sequestration potential in the celTado region of Brazil, in Carbon Sequestration in Soils ofLatin America, edited by R. Lal et al., pp. 285-304, Hmworth, Binghamton, N. Y. Carvalho, A. M., M. M. C. Bustamante, A R. Kozovits, L. N. Miranda, L. J. Vivaldi, and D. M. Sousa (2006), Emiss5es de oxidos de nitrogenio associadas a aplicayao de ureia sob plantio convencional e direto, Pesqui. Agropecu. Bras., 41, 679-685. Christian, T. J., R. 1. Yokelson, 1. A Carvalho Jr., D. W. T. Griffith, E. C. Alvarado, 1. C. Santos, T. G. S. Neto, C. A G. Veras, and W. M. Hao (2007), The tropical forest and fire emissions experiment: Trace gases emitted by smoldering logs and dung from deforestation and pasture fires in Brazil, J Geophys. Res., 112, Dl8308, doi:lO.l029/2006JD008147. Conrad, R. (1996), Soil microorganisms as controllers of atmospheric trace gases (H2, CO, CH4, OCS, N 20, and NO), Microbioi. Rev., 60,609-640. Corazza, E. 1., J. E. Silva, D. V. S. Resck, and A C. Gomes (1999), Compoi'tamento de diferentes sistemas de manejo como fonte ou deposito de carbono em relayao a vegetayao de Cerrado, Rev. Bras. Cien. Solo, 23, 425-432. Crutzen, P. 1. (1970), The influence of nitrogen oxides on the atmospheric ozone content, Q. J R. Meteorol. Soc., 96, 320-325. Davidson, E. A (1991), Fluxes of nitrous and nitric oxide from terrestrial ecosystems, in Microbial Production and Consumption ofGreenhouse Gases: Methane, Nitrogen Oxides and Halomethanes, edited by 1. E. Rogers and W. B. Whitman, pp. 219-235, Am. Soc. for Microbiol., Washington, D. C. Davidson, E. A (1992), Sources of nih'ic oxide and nitrous oxide following wetting of dry soil, Soil Sci. Soc. Am. J, 56, 95-102. Davidson, E. A, and P. Artaxo (2004), Globally significant changes in biological processes of the Amazon Basin: Results of large-
scale biosphere-atmosphel'e experiment, Global Change Bioi., 10,519-529. Davidson, E. A, and W.iKingerlee (1997), A global inventOly of nitric oxide emissions .~om soils, NutI'. Cycling Agroecosyst., 48, 3750. , . D aVl'dson, E. A., M./'Keller, H. W. Enckson, L. V. Verchot, and E. Veldkamp (2000), Testing a conceptual model of soil emissions of nitrous and nitric oxides, BioScience, 50, 667-680. Davidson, E. A., M. M. C. Bustamante, and A S. Pinto (2001), Emissions of nitrous oxide and nitric oxide fi'om soils of native and exotic ecosystems of the Amazon and cerrado regions of Brazil, Sci. World, 1,312-319. Davidson, E. A, F. Y. Ishida, and D. C. Nepstad (2004), Effects of an experimental drought on soil emissions of carbon dioxide, methane, nitrous oxide, and nitric oxide in a moist tropical forest, Global Change Bioi., 10, 718-730. Davidson, E. A, et al. (2007), Recuperation of nih'ogen cycling in Amazonian forests following agricultural abandonment, Nature, 447, 995-999. do Cmmo, J. B., C. Neill, D. C. Garcia-Montiel, M. de C. Piccolo, C. C. Cerri, P. A Steudler, C. A. de Andrade, C. C. Passianoto, B. 1. Feigl, and 1. M. Melillo (2005), Nitrogen dynamics during till and no-till pasture restoration sequences in Rondonia Brazil NutI'. Cycling Agroecosyst., 71,213-225. " do Carmo, 1. B., M. Keller, 1. D. Dias, P. B. de Camargo, and P. Crill (2006), A source ofmethane from upland forests in the Brazilian Amazon, Geophys. Res. Lett., 33, L04809, doi:10.l029/ 2005GL025436. Duarte, 1. 0., 1. C. Garcia, and M. J. Matoso (2007), Area de plantio direto e area plantada com sorgo no Cerrado: Existe alguma cOl'l'elayao entre elas?, Com un. Tec. 151,8 pp., Embrapa Milho e Sorgo, Sete Lagoas, Brazil. Dueck, T. A, et al. (2007), No evidence for substantial aerobic methane emission by terrestrial plants: A l3C-labelling approach, New Phytol., 175,29-35. Erickson, H., E. A Davidson, and M. Keller (2002), FOlmer landuse and tree species affect nitrogen oxide emissions from a tropical dry forest, Decologia, 130, 297-308. Feigl, B., M. Bernoux, C. C. Cerri, and M. C. Piccolo (2001), 0 efeito da sucessao floresta/pastagem sobre 0 estoque de carbono e 0 fluxo de gases em solos da Amazonia, in Mudmu;:as Climaticas Globais e a Agropecuaria Brasileira, edited by M. A Lima, O. M. R. Cabral, and 1. D. G. Miguez, pp. 257-271, Embrapa Meio Ambiente, Jaguariuna, Brazil. Fenchel T., G. M. King, and T. H. Blackbum (1998), Bacterial Biogeochemisfl)J: The Ecophysiology ofMineral Cycling, Academic, San Diego, Calif. Fernandes, E. B. (2008), Emiss5es de CO 2, NO x e N 20 de solos sob diferentes sistemas de cultivo no cerrado, Ph.D. thesis, 138 pp., Univ. of Brasilia, Brasilia, Brazil. Fel'l'etti, D. F., 1. B. Miller, J. W. C. White, K. R. Lassey, D. C. Lowe, and D. M. Etheridge (2007), Stable isotopes provide revised global limits of aerobic methane emissions from plants, Atmos. Chem. Phys., 7,237-241. Firestone, M., and E. A Davidson (1989), Microbial basis of NO and N20 production and consumption, in Exchange of Trace
351
Gases Between Ecosystems and the Atmosphere, edited by M. O. Andreae and D. S. Schimel, pp. 7-21, John Wiley, New York. Forster, P., et al. (2007), Changes in atmospheric constituents and in radiactive forcing, in Climate Change 2007: The Physical Science Basis: Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, edited by S. Solomon et al., pp. 129-234, Cambridge Univ. Press, Cambridge, U. K. Frankenberg, C., 1. F. Meirink, M. van Weele, U. Plalt, and T. Wagener (2005), Assessing methane emissions from global space-borne observations, Science, 308, 1010-1014. Frankenberg, c., 1. F. Meirink, P. Bergamaschi, A P. H. Goede, M. Heimann, S. Korner, U. Platt, M. van Weele, and T. Wagner (2006), Satellite chmiography of atmospheric methane from SCIAMACHY on board ENVISAT: Analysis of the years 2003 and 2004, J Geophys. Res., 111, D07303, doi:lO.1029/2005JD006235. Garcia-Mendez, G., J. M. Mass, P. A Matson, and P. M. Vitousek (1991), Nih'ogen transformations and nitrous oxide flux in a tropical deciduous forest in Mexico, Decologia, 88, 362-366. Garcia-Montiel, D. C., P. A Steudler, M. C. Piccolo, 1. Melillo, C. Neill, and C. C. CelTi (2001), Controls on soil nitrogen oxide emissions from forest and pastures in the Brazilian Amazon, Global Biogeochem. Cycles, 15, 1021-1030. Garcia-Montiel, D. C., P. A. Steudler, M. C. Piccolo, 1. Melillo, C. Neill, and C. C. Cerri (2003), Nitrogen oxide emissions following wetting of dly soils in forest and pastures in Rondonia, Biogeochemisfly, 64" 319-336. Hanson, R. S., and T. E. Hanson (1996), Methanotrophic bacteria, Microbiol. Rev., 60,439-471. Hirsch, A, W. S. Litt,le, R. A Houghton, N. A Scott, and J. D. White (2004), The net carbon flux due to deforestation and forest re-growth in the Bt'azilian Amazon: Analysis using a processbased model, Global Change Bioi., 10, 908-924. Hirsch, A 1., A M. Michalak, L. M. Bmhwiler, W. Peters, E. 1. Dlugokencky, and P. P. Tans (2006), Inverse modeling estimates of the global nitrous oxide surface flux from 1998-200 I, Global Biogeochem. Cycles, 20, GB1008, doi:1O.1029/2004GB002443. Houweling, S., T. Rocknlann, 1. Aben, F. Keppler, M. Krol, 1. F. Meirink, E. 1. Dlugokencky, and C. Frankenberg (2006), Atmospheric constraints on global emissions of methane from plants, Geophys. Res. Lett., 33, Ll5821, doi:10.102912006GL026162. Jacob, D. J., and P. S. Bakwin (1991), Cycling of NOx in tropical forest canopies, in Microbial Production and Consumption of Greenhouse Gases: Methane, Nitrogen Oxides and Halomethanes, edited by 1. E. Roger and W. B. Whitman, pp. 237-253, Am. Soc. for Microbiol., Washington, D. C. Jacob, D. 1., and S. C. Wofsy (1990), Budgets of reactive nih'ogen, hydrocarbons, and ozone over the Amazon forest during the wet season, J Geophys. Res., 95,16,737-16,754. Kaplan, W. A, S. C. Wofsy, M. Keller, and 1. M. Da Costa (1988), . Emission of NO and deposition of 0 3 in a tropical forest system, J Geophys. Res., 93, 1389-1395. Kauffman, 1. B., K. M. Till, and R. W. Shea (1992), Biogeochemistty of deforestation and biomass burning, in The Impact ofHuman Activities on the Environment, edited by D. A Dunnette and R.1. O'Brien, ACS Symp. Ser., 483, 426-456.
352
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
Keller, M., and W. A. Reiners (1994), Soil-atmosphere exchange of nitrous oxide, nitric oxide, arid methane under secondaty succession of pasture to forest in the Atlantic lowlands of Costa Rica, GloMI Biogeochem. Cycles, 8, 399-409. Keller, M., T, 1. Goreau, S, C, Wofsy, W, A. Kaplan, and M. B, McElroy (1983), Production of nitrous oxide and consumption of methane by forest soils, Geophys, Res. Lett., 10, 11561159. , Keller, M" W, A. Kaplan, and S. C, Wofsy (1986), Emissions of N20, CH4 and CO 2 from tropici'll forest soils, J. Geophys, Res., 91,11,791-11,802, Keller, M" E. Veldkamp, A. M. Weitz, and W. A. Reiners (1993), Pasture age affects on soil-atmosphere trace gas exchange in a deforested area of Costa Rica, Nature, 365, 244-246, Keller, M., R. Varner, J. D. Dias, H, Silva, P. Crill, R. C. de Oliveira Jr., and G, P, Asner (2005), Soil-ahnosphere exchange of nitrous oxide, nitric oxide, methane, and carbon dioxide in logged and undishlrbed forest in the Tapajos National Forest, Brazil, Earth Interact., 9(23), EIl25, doi:l0.1l75IEIl25.1. Keppler, F., J. T, G. Hamilton, M. BraE, and T. Rockmann (2006), Methane emissions from terrestrial plants under aerobic conditions, Nature, 439, 187-191. Kirkman, G. A., A. Gut, C. Ammann, L. V. Gatti, A. M. Cordova, M, A. L. Moura, M. O. Andreae, and F. X. Meixner (2002), Surface exchange of nitric oxide, nih'ogen dioxide, and ozone at a cattle pasture in Rondonia, Brazil, J. Geophys. Res., 107(D20), 8083, doi:10.1029/2001JD000523. Kirschbaum, M, U. F., D, Brulm, D. M, Etheridge, 1. R. Evans, G. D. Farquhar, R. M, Gifford, K. 1. Paul, and A. J. Winters (2006), A comment on the quantitative significance of aerobic methane release by plants, Funct. Plant BioI., 33, 521-530, Klink, C. A., and R. B, Machado (2005), Conservation of the Brazilian cerrado, Conserv, BioI., 19,707-713, Lashof, D. A., and D. R, Ahuja (1990), Relative global warming potentials of greenhouse gas emissions, Nature, 344, 529-531. Leininger, S" T. Urich, M. Schloter, L. Schwark, J. Qi, G. W. Nicol, J. 1. Prosser, S, C. Schuster, and C. Schleper (2006), Archaea predominate among ammonia-oxidizing prokaryotes in soils, Nature, 442, 806-809. Lerner, 1., E. Matthews, and 1. Fung (1988), Methane emission fi'om animals: A global high-resolution data base, Global Biogeochem, Cycles, 2,139-156, Levine, 1. S., E. L. Winstead, D. A. B. Parsons, M, C. Scholes, R. J. Scholes, W. R. Cofer III, D. R. Cahoon Jr., and D. 1. Sebacher (1996), Biogenic soil emissions of nih'ic oxide (NO) and nih'ous oxide (N20) fi'om savannas in South Africa: The impact of wetting and burning, J. Geophys, Res., 101,23,689-23,697, Lima, M. A., R. C. Boeira, V. L. S, Castro, M. A. V, Ligo, 0. M, R. Cabral, R. F. Vieira, and A. 1. B, Luiz (2001), Estimativa das emissoes de gases de feito estufa provenientes de atividades agricolas no Brasil, in Mudal/{;as climaticas Globais e a Agropecuaria Brasileira, edited by M, A. Lima, O. M. R, Cabral, and J. D, G. Miguez, pp. 169-189, Embrapa Meio Ambiente, Jaguariuna, Brazil. Lodge, D. J" F. N, Scatena, C, E, Asbuty, and M, J. Sanchez (1991), Fine litterfall and related nutrient inputs resulting from
hurricane Hugo in subtropical wet and lower montane rain forests of Puerto Rico, Biotropica, 23, 336-342. Luizao, F" P, Matson, G. Livingston, R. Luizao, and P. Vitousek (1989), Nitrous oxide flux following tropical land clearing, Global Biogeochem, Cycles, 3, 281-285, Luizao, R. C. C" F, J. Luizao, R. Q. Paiva, T. F. Monteiro, L. S. Sousa, and B. Kruijts (2004), Variation of carbon and nitrogen cycling process along a topographic gradient in a central Amazonian forest, Global Change BioI" 10, 592-600. Martius, C., R, Wassmann, U. Thein, A. Bandeira, H, Rennenberg, W. Junk, and W. Seiler (1993), Methane emission from wood-feeding termites in Amazonia, Chemosphere, 26, 623632. Matson, P. A., C. Billow, S. Hall, and J. Zachariassen (1996), Fertilization practices and soil variations control nitrogen oxide emissions fi'om tropical sugar cane, J. Geophys, Res., 101, 18,533-18,545. Matson, P. A., and P. M. Vitousek (1990), Ecosystem approach for the development of a global nitrous oxide budget, Bioscience, 40, 667-672. Meixner, F. X. (1994), Surface exchange of odd nitrogen oxides, Nova Acta Leopold., 288, 299-348. Melack, 1. M., L. L. Hess, M. Gastil, B. R. Forsberg, S. K. Hamilton, 1. B, T. Lima, andM. L. M.'Novo (2004), Regionalization of methane emissions in the Amazon Basin with microwave remote sensing, Global Change BioI., 10, 530-544. Melack, 1. M., E. M, L. M, Novo, B. R, Forsberg, M. T, F. Piedade, and L. Maurice (2009), Floodplain ecosystem processes, Geophys. Monogr, Ser., doi:l0.l02912008GM000721, this volume. Melillo, 1. M., P. A. Steudler, B. J, Feigl, C. Neill, D, Garcia, M, C. Piccolo, C. C. Cerri, and H. Tian (2001), Nitrous oxide emissions from forests and pastures of various ages in the Brazilian Amazon,J. Geophys, Res" 106, 34,179-34,188, Metay, A., R. Oliver, E. Scopel, 1. M. Douzet, J: A. A. Moreira, F, Maraux, B. 1. Feigl, and C. Feller (2007), N 20 and CH4 emissions from soils under conventional and no-till management practices in Goiania (cerrados, Brazil), Geoderma, 114, 7888. Miller, J. B., L. V. Gatti, M. T. S. d' Amelio, A. M. Crotwell, E. J. Diugokencky, P. Bakwin, P. Artaxo, and P. P. Tans (2007), Airborne measurements indicate large methane emissions fi'om the eastern Amazon basin, Geophys. Res. Lett" 34, Ll0809, doi: 10.1 029/2006GL029213. Mosier, A.R., and 1. A. Delgado (1997), Methane and nitrous oxide fluxes in grass lands in western Puerto Rico, Chemosphere, 35, 2059-2082, Mosier, A. R., 1. M, Duxbuty, 1. R, Freney, 0. Heinemeyer, K. Minami, and D. W. Johnson (1998), Mitigation agricultural emissions of methane, Clim. Change, 40, 39-80. Nardoto, G. B., and M. M, C. Bustamante (2003), Effects of fire on soil nitrogen dynamics and microbial biomass in savannas of central Brazil, Pesqui, Agropecu, Bras" 38, 955-962, Neff, 1. C., M. Keller, E. A. Holland, A. W, Weitz, and E. Veldkamp (1995), Fluxes of nitric oxide fi'om soils following the clearing and burning of a secondary h'opical rain forest, J. Geophys. Res" 100, 25,913-25,922,
BUSTAMANTE ET AL. Neill, C" M, C, Piccolo, P, Steudler,1. M. Melillo, B, 1. Feigl, and C. C, Cerri (1995), Nitrogen dynamics in soils offorests and active pastures in the ,y,testem Brazilian Amazon Basin, Soil BioI, Biochem., 27, 1167fl175. Neill, C., P. A. Steudl9', D. C. Garcia-Montiel, J. M. Melillo, B, J, Feigl, M. C. Piccolo, and' c. C. Cerri (2005), Rates and conh'Ols of nih'ous oxide and nitric oxide emissions following conversion of forest to pasture in Rondoni!:!, Nutl'. Cycling Agroecosyst" 71, 1-15. Nepstad, D, C" et al. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505-508. Nepstad, D. C" et al. (2002), The effects of partial throughfall exclusion on canopy processes, aboveground production, and biogeochemishy of an Amazon forest, J. Geophys, Res., 107(D20), 8085, doi: 10.1 029/200 IJD000360, Oliveira, 0. C., 1. P. Oliveira, S. Urquiaga, B, J. R. Alves, and R. M. Boddey (2004), Chemical and biological indicator of decline/ degradation of Brachiaria pastures in the Brazilian cerrado, Agric. Ecosyst, Environ" 103, 289-300, Parker, G. G, (1985), The effect of disturbance on water and solute budgets of hillslope tropical rainforest in northeastern Costa Rica, Ph,D. thesis, Univ. of Ga., Athens. Passianoto, C. C., T, Ahrens, B, 1. Feighl, P. A. Steudler, 1. B. do Carmo, and 1. M. Melillo (2003), Emissions of CO2, N 20, and NO in conventional and no-till management practices in Rondonia, Brazil, BioI, Fertil, Soils, 38, 200-208. . Pereira, R., 1. C, Zweede, G. P. Asner, and M, Keller (2002), Forest canopy damage and recovery in reduced impact and conventional logging in eastern Para, Brazil, For. Ecol. Manage" 168, 77-89, Pinto, A. de S, (2003), Contribuiyao dos solos de cerrado do Brasil Central para as emissoes de gases trayo (C0 2, N 20 e NO): Sazonalidade, queimadas prescritas e manejo de pastagens degradadas, Ph,D. thesis, 114 pp., Univ. de Brasilia, Brasilia. Pinto, A. de S" M. M, C. Bustamante, K. Kisselle, R, Burke, R. Zepp, L, T. Viana, R. F, Varella, and M. Molina (2002), Soil emissions of N20, NO, and CO 2 in Brazilian savannas: Effects of vegetation type, seasonality, and prescribed fires, J. Geophys. Res" 107(020), 8089, doi: 10, 1029/2001JD000342, Pinto, A. de S" M, M, C. Bustamante, M. R, S. S, da Silva, K. W, Kisselle, M. Brossard, R. Kluger, R. G. Zepp, and R. A. Burke (2006), Effects of different treatments of pasture restoration on soil trace gas emissions in the celTados of central Brazil, Earth Interact" 10(1), EI146, doi:10.1l75/EI146.1. Pivello, V. R., and L. M. Coutinho (1992), Transfer ofmacro-nutrients to the atmosphere during experimental burnings in an open cerrado (Brazilian savanna), J. Trop, Ecol., 8, 487-497. Poth, M., 1. C. Anderson, H, S. Miranda, A. C. Miranda, and P. J. Riggan (1995), The magnitude and persistence of soil NO, N 20, CH4 and CO2fluxes from burned tropical savanna in Brazil, Global Biogeochem. Cycles, 9,503-513, Potter, C" E. Davidson, D. Nepstad, and C, R. de Carvalho (2001), Ecosystem modeling and dynamic effects of deforestation on trace gas fluxes in Amazon tropical forests, For, Ecol, Manage., 152,97-117, Potter, C. S" E. A. Davidson, and L. V, Verchot (1996), Estimation of global biogeochemical controls and seasonality in soil methane consumption, Chemosphere, 32, 2219-2246,
353
Potter, C, S" E. A. Davidson, S, A. Klooster, D. C. Nepstad, G, H. De Negreiros, and V. Brooks (1998), Regional application of an ecosystem production model for studies of biogeochemistly in Brazilian Amazonia, Global Change BioI" 4, 315334, Prather, M., and D, Ehhalt (2001), Atmospheric chemistly and greenhouse gases, in Climate Change 2001: The Scientific Basis: Contribution o/Working Group J to the Third Assessm~nt Report o/the Intergovel'llmenlal Panel on Climale Change, edited by 1. T, Houghton et al., pp, 239-288, Cambridge Univ. Press, Cambridge, U, K. Primavesi, 0., R. T. S. Frighetto, M. S. Pedreira, M, A, Lima, T. T. Berchielli, and P, F. Barbosa (2004), Metano enterico de bovinos leiteiros em condiyoes tropicais brasileiros, Pesqui, Agropecu, Bras., 39, 277-283, Reeburgh, W. S. (2003), Global methane biogeochemistly, in Treatise on Geochemistry, vol. 4, The Atmosphere, edited by R. F, Keeling, pp. 65-89, Elsevier, New York. Resck, D. V, S., J. Pereia, and 1. E. da Silva (1991), Dinamica da materia organica na Regiao dos Cerrados, Doc. 36, Empresa Bras, de Pesqui, Agropecu., Planaltina, Brazil. Rummel, U., C. Ammann, A. Gut, F, X. Meixner, and M. 0. Andreae (2002), Eddy covariance measurements ofnitric oxide flux within an Amazonian rain forest, J. Geophys, Res" 107(D20), 8050, doi:IO.1029/2001JD000520, Saminez, T. C. 0. (1999), Efeito do sistema de cultivo, tensao da agua biomassa mic1'l;:>biana e temperatura do solo nos fluxos de CH4 e N20 em solos de cerrados, MSc. thesis, 99 pp" Univ. de Brasilia, Brasilia. Sanhueza, E. (2007), 11ethane soil-vegetation-atmosphere fluxes in tropical ecosystems, Interciencia, 32, 30-34. Sano, E, E., A. 0, Barcellos, and H, S. Bezerra (2000), Assessing the spatial distribution of cultivated pastures in the Brazilian savanna, Pasturas Trop" 22, 2-15. Scatena, F. N., S, Moya, C. Estrada, and J. D. Chinea (1996), The first five years in the reorganization of aboveground biomass and nutrient use following hurricane Hugo in Bisley Experimental watersheds, Luquillo Experimental Forest, Puerto Rico, Biotropica, 28, 424-440. Seiler, W., R, Comad, and D, Scharffe (1984), Field studies of methane emission from termite nests into the atmosphere and measurements of methane uptake by tropical soils, J. Atmos, Chem., 1,171-186. Serya, D., R. Delmas, X, L. Roux, A. B. Parsons, M, C, Scholes, L. Abbadie, R. Lensi, 0, Ronce, and L. Labroue (1998), Comparison of nih'ogen monoxide emissions fi'om several African tropical ecosystems and influence of season and fire, Global Biogeochem. Cycles, 12, 637-651. Setzer, A. W., M, C, Pereira, A. C, P. Pereira, and S, 0, Almeida (1998), Relat6rio de atividades do projeto IBDF-INPE "SEQE"ano 1987, Publ. INPE_4534-RPE/565, INPE, Inst. Nac. de Pesqui. Espaciais, sao Jose dos Campos, Brazil. Silvel:, W, L., 1. Neff, M, McGroddy, E, Veldkamp, M, Keller, and R. Cosme (2000), Effects of soil texture on belowground carbon and nutrient storage in a lowland Amazonian forest ecosystem, Ecosystems, 3, 193-209.
354
SOURCES AND SINKS OF TRACE GASES IN AMAZONIA AND THE CERRADO
Silver, W. L., A W. Thompson, M. E. McGroddy, R. K. Varner, J. D. Dias, H. Silva, P. M. Crill, and M. Keller (2005), Fine root dynamics and trace gas fluxes in two lowland tropical forest soils, Glob~ Change BioI., 11, 290-306. Spahni, R., et al. (2005), Atmospheric methane and nitrous oxide of the late Pleistocene from Antarctic ice cores, Science, 310, 1317-1321. Steudler P. A., R. D. Jones, M. S. Castro, J. M. Melillo, and D. L. Lewis (1996), Microbial controls of CH4 oxidation in temperate forest and agricultural soils, in Microbiology ofAtmospheric Trace Gases, edited by J. C. Muhell and D. P. Kelly, NATO AS! Ser., Ser. 1, vol. 39, pp. 69-85, Springer, Berlin. van Dijk, S. M., and F. X. Meixner (2001), Production and consumption of NO in forest and pasture soils from the Amazon Basin, Water Ail' Soil Pol/ut. Focl/s, 1, 119-130. Van Gestel, M., R. Merckx, and K. Vlassak (1993), Microbial biomass responses to soil drying and rewetting: The fate of fast- and slow-growing microorganisms in soils from different climates, Soil BioI. Biochem., 25, 109-123. Varella, R. F., M. M. C. Bustamente, A S. Pinto, K. W. Kisselle, R. V. Santos, R. A Burke, R. G. Zeep, and L. T. Viana (2004), Soil fluxes of COl, CO, NO and NlO from an old pasture and from native savanna in Brazil, Ecol. Appl., 14(sp4), 221-231. Varner, R. K., M. Keller, J. R. Robertson, J. D. Dias, H. Silva, P. M. Crill, M. McGroddy, and W. L. Silver (2003), Experimentally induced root mortality increased nitrous oxide emission from tropical forest soils, Geophys. Res. Lett., 30(3), 1144, doi: 10.1 029/2002GLO 16164. Vasconcelos, S. S., et al. (2004), Moisture and substrate availability constrain soil trace gas fluxes in an eastern Amazonian regrowth forest, Global Biogeochem. Cycles, 18, GB2009, 10. 1029/2003GB0022 10
Veldkamp, E., and M. Keller (1997), Nitrogen oxide emissions from a banana plantation in the humid tropics, J. Geophys. Res., 102,15,889-15,898. Verchot, L. V., E. A Davidson, H. Cattanio, 1. L. Ackerman, H. E. Erickson, and M. Keller (1999), Land use change and biogeochemical controls of nitrogen oxide emissions from soils in eastern Amazonia, Global Biogeochelll. Cycles, 13, 3146. Verchot, L. V., E. A. Davidson, J. H. Cattanio and 1. L. Ackerman (2000), Land use change and biogeochemical controls of methane fluxes in soils of eastern Amazonia, Ecosystems, 3, 41-56. Verchot, L. V., Z. Holmes, L. Mulon, P. M. Groffman, and G. M. Lovett (2001), Gross vs net rates of N mineralization and nitrification as indicators of functional differences between forest types, Soil BioI. Biochem., 33, 1889-1901. Verissimo, A, P. Barreto, M. Mattos, R. Tarifa, and C. Uhl (1992), Logging impacts and prospects for sustainable forest management in an old Amazonian fi'ontier: The case of Paragominas, For. Ecol. Manage., 55, 169-199. Wick, B., E. Veldkamp, W. Z. de Mello, M. Keller, and P. Crill (2005), Nitrous oxide fluxes and nih'ogen cycling along a pasture chronosequence in central Amazonia, Brazil, Biogeosci. Discuss., 2, 499-535.
M. M. C. Bustamante and D. A Silva, Department of Ecology, University of Brasilia, Asa Norte ICC SuI, Brasilia, DF CEP 70910-900, Brazil. ([email protected]) M. Keller, NEON Inc., 5340 Airport Boulevard, Boulder, CO 80301, USA
The Production, Storage, and Flow of Carbon in Amazonian Forests Yadvinder Malhi, l Sassan Saatchi,2 Cecile Girardin,l and Luiz E. O. C. Aragao 1 The carbon stores and dynamics of tropical forests are the subject of major international scientific and policy attention. Research associated with the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) has generated substantial advances in our understanding of the cycling of carbon at selected forest sites in Brazilian Amazonia and generated new insights into how these processes may vary across the wider Amazonian region. Here we report on aspects of this new understanding. We present, in particular, a comprehensive synthesis of carbon cycling in three focal LBA sites (Manaus, Tapaj6s, and Caxiuana), drawing on studies of productivity, litterfall, respiration, physiology, and ecosystem fluxes. These studies are placed in the context of the wider Amazonian region by utilizing the results of the Amazon Forest Inventory Network (RAINFOR) and other forest plots. We discuss the basin-wide distribution of forest biomass derived by combining these plots and a suite of satellite data, and examine the dynamics of carbon cycling in the context of regional carbon stores in the forest. Particular: attention is drawn to the strong relationship between forest productivity and turnover, which suggests that higher levels of forest productivity increase forest dynatnism rather than forest biomass. We conclude by discussing what the scientifi<; priorities should be for a synthetic region-wide understanding of the carbon dynamics and stores of Amazonian forests.
1. INTRODUCTION There is cunently unprecedented interest in the carbon cycling and carbon storage of tropical forests, stimulated by a renewed global effort to limit rates of carbon dioxide emissions from deforestation as a strategy contributing to the mitigation of global atmospheric change, and also by a concern that climate change may result in net carbon emislEnvironmental Change Institute, School of Geography and the Environment University of Oxford, Oxford, UK. lJet Propulsion LaboratOly, California Institute of Technology, Pasadena, California, USA Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10. I029/2008GM000779
sions from tropical forests, and thus a positive feedback on climate change. Much ofthis interest is focused on assessing the magnitude of carbon stores in these forests, understanding what determines the magnitude of these stores and exploring how these stores will respond to either mitigate or accelerate climate change. Amazonia is home to half of the world's tropical forests, yet until recently, there has been little detailed exploration of the carbon dynamics of its forests, nor the spatial variation of their carbon cycle. The Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) program and associated research provided a unique opportunity to explore these is. sues, both through intensive studies at a number of sites, and from a Pan-Amazonian perspective of the role of Amazonia in the earth system. In' this chapter, we review and synthesize some recent advances that LBA-associated research has made in our understanding of the carbon cycling of Amazonian forests, 355
356
PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS
focusing on three questions: (1) What synthetic picture do studies at key LBA study sites provide about stocks and flows of carbon at these sites? (2) How do these focal LBA sites fit into''tthe wider context of Amazonian forests? (3) What does the relationship between carbon allocation, biomass production, and biomass stock tell us about spatial and temporal variations in carbon cycling and carbon storage in old-growth Amazonian forests? To pursue these questions, we will draw on three main lines of research. First, we foclls on a detailed synthesis of carbon cycling research at three LBA study sites (Manaus, Tapaj6s, and Caxiuana; hereafter termed the "focal sites"), where there has been an overlap of forest mensuration and net primalY productivity (NPP) studies, ecophysiological and respiration measurements, and eddy covariance studies of above-canopy fluxes. This synthesis is largely drawn fi'om Malhi et al. [2009]; detailed analysis and discussion of the caveats of this synthesis are presented in that paper. Here we summarize the results of the synthesis without dwelling on methodological issues. Second, to put these sites into the wider context, we draw on forest plots and NPP studies across Amazonia from the Amazon Forest InventOly Network (RAINFOR) [Malhi et al., 2001], in particular, studies of the spatial variation of productivity [Malhi et al., 2004; Aragiio et al., 2009] and biomass [Baker et al., 2004; Malhi
MALHI ET AL.
et al., 2006]. The temporal shifts in forest carbon dynamics observed in this network are discussed by Phillips et ill. [this volume], and spatial relations to plant ecophysiology are discussed by Lloyd et al. [this volume]. Our focus here is on mean annual or longer-term budgets in the carbon cycle: we will not discuss seasonal or interannual variations (somewhat discussed by Saleska et al. [this volumeD. Third, we also draw on and discuss the remote sensing-based apPl'(lach to spatial extrapolation of biomeuy data, as employed by Saatchi et al. [2007]. The three focal study sites are situated on deep, highly weathered Oxisols in terra firme forests on upland areas of lowland eastem Amazonia. At Manaus and Caxiuana, the Oxisol landscape is occasionally dissected by valleys with seasonally waterlogged Spodosols and a lower biomass forest, whereas the site at Tapaj6s sits on a broad plateau (~90 m above sea level) with little sU'eam development. The majority of detailed process studies have focused on the Oxisolltena filme landscape, whereas the flux tower footprint extends across a broader landscape of plateaus and river valleys. For further discussion of the sites, see Malhi et al. [2009]. We also report on new assessments 'ofNPP at other sites, particularly in westem Amazonia (Colombia and Peru), as reported by Aragiio et al. [2009]. A summary of these sites is given in Table 1, and the sites are plotted in Figure 1.
Manaus Tapajos
VI L",
Table 1. Site Codes, Locations, and Climatic Characteristics of the 10 Net Primary Productivity Amazonian Sites Reported in This Study Including the Three Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) Focal Sites Caxiuana, Manaus K34, and Tapaj6s km 67 a Study Sites Amazon Forest InventOly Network Sites Code AGPOI AGP02 CAX03 CAX06 CAX08 MAN05,MAN12, BNT04 TAM05 TAM06 TAP04 ZAROI
357
....~ __,_,_,_.1
'''''''''''''_,~~,
~
Figure 1. A map of the Large-Scale Biosphere-Atmosphere Experiment in Atnazoriia (LBA) and Amazon Forest InventOlY Network (RAINFOR) net primary productivity (NPP) sites mentioned in this chapter. The focal LBA sites (Manaus Tapaj6s, and Caxiuana) are underlined. Site codes are listed in Table I. From Aragiio et al. [2009]. ,
Climate Location
2. STOCKS OF CARBON
Name
Country
Latitude
Longitude
Rainfall, mma- I
Agua Pudre plot E Agua Pudre plot U . Caxiuana drought experiment control plot Caxiuana flux tower site LBA Caxiuana terra preta site Manaus K34 flux tower site LBA Tambopata RAINFOR plot 3 Tambopata RAINFOR plot 4 Tapaj6s krn 67 flux tower site LBA Zafire, Varillal
Colombia Colombia
-3.72 -3.73
-70.3 -70.4
2723 2723
0.0 0.0
25.5 25.5
Brazil
-1.72
-51.5
2314
4.0
26.9
Brazil Brazil
-1.72 -1.72
-51.5 -51.5
2314 2314
4.0 4.0
26.9 26.9
Brazil
-2.5
-60.0
2272
3.0
27.1
-12.8 -12.9
-69.7 -69.8
2417 2417
3.5 3.5
25.2 25.2
Brazil
-2.5
-55.0
1968
4.5
26.1
Colombia
-4.0
-69.9
2723
0.0
25.5
Peru Peru
DSL, months
MAT,OC
"The climate data presented in this table are mean values from 1960 to 1998 derived fi'om the University of East Anglia Observational Climatology [New et al., 1999] and published in the work of Malhi et al. [2004]. Cumulative annual rainfall is given in mm a-I, dry season length (DSL) in months, corresponds to the sum of consecutive months with rainfall < 100 mm month-I, and temperature is the mean annual temperature (MAT) in degrees Celsius. Modified from Aragiio et al. [2009].
logical ones. Palace et al. [2008] estimated tree density and size for seven sites in Amazonia, including the three focal A synthesis of repOlied values of carbon stocks at each fo- sites in this chapter, with the same u'end in u'ee density and cal site is summarized in Figure 1. Detailed discussion ofthe tree size based on crown width (more trees per hectare, but data sets and the procedure employed to average across stud- smaller trees, in Manaus, in comparison with larger but less ies are presented by Malhi et al. [2009]. All carbon stocks trees at Caxiuana). are in Mg C ha- 1; 1 Mg C ha- I is equal to 100 g C m-2 • In the wider context of Amazonia, these focal sites tend to have higher-than-average biomass [Baker et al., 2004; 2.1. Aboveground Live Biomass Malhi et al., 2006]. Typical AG live biomass in these eastem Amazonian deep Oxisol forests is between 300 and 350 Mg Aboveground (AG) live biomass is highest at Caxiuana, dly weight ha- I (Plate 2), equivalent to 150-175 Mg C ha- I . slightly lower on the Manaus plateau, and lowest at Tapa- Similar high biomass values are found in the Guyanas, but j6s km 67 and in the Manaus valley bottoms. Some of these biomass tends to decline into the drier margins of Amazonia differences are reflected in forest structure: Caxiuana has , as wood volume declines (Figure 2). This decline probably a greater propOliion of large trees than Manaus. All these occurs because seasonal drought intensifies the relative imstudies applied the allometric equation of Chambers et al. portance of root competition for water over crown compe[2001], modified by species-specific densities as applied by tition for light, the median intertree spacing consequently Baker et al. [2004]. Hence, the differences between sites increases, and the number of trunks per unit area declines. largely reflect structural differences rather than methodo- In contrast, biomass also tends to decline as one heads west,
358
PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS
l:g I
!11l!
iii i,~!
..... '.
::::II
inespective of whether this is to dry southwestern Amazonia or wet northwestern Amazonia (Plate 2). This decline appears associated with an increase in wood productivity ~nd a decrease in mean wood density (Plate 1). Hence, the hrgh AG biomass in the eastern Amazonian focal sites appears driven largely by the influence of soils on forest structure and tree life-history traits. Infertile soils tend to favor slow growing, long-lived tree species, which invest more in herbivore and pathogen defense strategies such as high wood density. Mean wood density is about 15% higher in the e~st ern Amazonian RAINFOR sites than in western Amazonran sites [Baker et al., 2004]. 2.2. Aboveground Dead Biomass
AG coarse woody debris (CWO) has been measured at Manaus [Chambers et al., 2004] and Tapaj6s, and is much greater at multiple sites across Tapaj6s. The CWO stocks at Tapaj6s appear to be in disequilibrium across a wider area [Pyle et al., 2008] and in combination with the low AG biomass, this suggests that the Tapaj6s local region underwent a recent and widespread major disturbance most likely in the 1990s [Keller et al., 2004; Pyle et al.: 2008]. CWD has not been measured at Caxiuana, but was estimated by taking estimates of mortality biomass inputs and dividing by a wood decay constant, kwood' of 0.16 ± 0.04 (see below) [Malhi et al., 2009], producing values very close to those observed at Manaus (15 ± 5 Mg C ha- 1 a- r versus 14 ± 2 Mg C ha- 1 a-I, respectively). This assumes near-equilibrium conditions, but at the Caxiuana tower plot, there is little evidence of the numerous large fallen trees seen at Tapaj6s km 67 (Y. Malhi, personal observation). A recent unpublished CWD census for Caxiuana (D. Metcalfe, unpublished data, 2008) confirms this estimate. Values of CWO at Tapaj6s are the highest reported for any site in lowland Amazonia [Baker et al., 2007].
MALHI ET AL.
naus by Klinge [1973J and more recently by N. Higuchi et al. (unpublished data, 2098). Estimates based on soil cores or pits tend to underest!i~ate biomass by being forced to exclude the core rooy/areas immediately underneath trees. In the absence of ejilect excavation, the best estimate of BG biomass may come from generalized empirical ratios for tropical rainforests: Malhi et al. [2009] estimated root biomass by multiplying the AG biomass values for the plateaus by a root: shoot ratio of 0.21 ± 0.03, encompassing the values reported for the tropics in the global surveys of Jackson et al. [1996J and Cail'11s et al. [1997]. A similar value of 0.21 has been confirmed by extensive and comprehensive BG biomass sampling of 131 trees in the vicinity of the Biomassa e Nutrientes na Floresta Tropical (BIONTE) plots near Manaus (N. Higuchi et a1., unpublished data, 2008). There is very little information on likely regional-scale variation in root biomass. Root biomass would be expected to be low in shallow soils (as may occur in crystalline shield regions or on montane slopes), soils with impermeable and shallow hardpans, or where they are limited by anoxia associated with seasonally high water tables, as are extensive in the broad poorly drained landscape between Manaus and the Andes foothills. Given suitable soils, the proportion of biomass in roots may be greater in seasonally drier forests and is known to be much greater in cerrado and cerradao regions, where more than 71 % of total live biomass can be BG [Castro and Kauiinann, 1998]. 2.4. Soil Carbon
Soil carbon (C) stocks are usually reported only for the top 30 cm or top 1 m of soil, and range between 74 and 127 Mg C ha -I for studies in our focal sites. Quesada reporis (reported by Malhi et. al. [2009]) C stocks for all three sites to 2 m depth, showing substantial C stocks at these depths and evidence for further carbon storage at greater depths, particularly at Tapaj6s.
2.3. Belowground Biomass
Belo~ground (BG) biomass stocks are predominantly in the coarse roots of live trees, with fine roots being a very minor component of the stock (though a large component of the turnover; see below). Root biomass has rarely been measured by direct harvesting, except in the vicinity ofMa-
Figure 2. (opposite) Carbon stocks in the aboveground (AG) and belowground (BG) compartments of the three focal LBA. Amaz~ nian forests. (a) Manaus, (b) Tapaj6s, (c) Caxiuana. Umts are III MgCha- 1•
2.5. Total Carbon Storage
Total AG C stocks are similar at all three sites (Manaus 199, Tapaj6s 202, Caxiuana, 231 Mg C ha- I ), with the smaller amount of biomass in living vegetation at Tapaj6s compensated by the higher CWO. Total carbon stocks to 2 m depth al'e presented in Figure 2 from Malhi et al. [2009J; data were derived from Quesada et al. [2009]. BG C stocks to 2 m depth are very similar in magnitude to AG stocks, with Tapaj6s showing the highest soil carbon stocks. Total C stocks (to 2 m depth) are 406 Mg C ha- 1 at Manaus, 422 Mg C ha- 1 at Tapaj6s, and 427 Mg C ha- 1 at Caxiuana. These values would certainly increase if greater soil depths
359
are considered and would be much lower for Manaus if the full plateau-valley landscape was considered. Overall, as much C is stored BG as AG, overwhelmingly in the soil carbon pool. How important these stocks are when considering the carbon value of rainforests depends on how vulnerable they are following land use change. Conversion to cattle ranching has little impact on soil C stocks, whereas intensive ploughing-associated agriculture may substantially oxidize carbon stocks in the upper soil layers. 3. NET PRIMARY PRODUCTIVITY ANOITSCOMPONENTS The NPP, the net amount of carbon fixed per unit time into organic matter, is a fundamental property predicted by many ecosystem models and a metric ofresource use by ecosystems. Comprehensive measurements of NPP have been rare in tropical forests [Clark et al., 2001a; Chambers et al., 2004J, with most studies repOliing only wood productivity or total AG productivity (woody production plus fine litterfall). The intensity of effort at the focal LBA sites provides an opportunity for a more comprehensive assessment of NPP, putting component measurements into context. ]Vialhi et al. [2009J examined and synthesized the studies of NPP components at the focal sites, including an analysis ofuncertainty and self-consistency, and discussion of caveats. Synthesized values ofNPP (and respiration) for the LBA focal sites are illustrated Al Figure 3, and for ten RAINFOR sites, the components of AG and BG NPP are plotted in Figure 4 [from Aragao et al., 2009J. We discuss each of the major terms in turn. 3.1. Woody Biomass Productivity
AG wood productivity, NPP stem , is the most visible aspect of forest productivity and can be measured by recensus of tree diameters and new recruits. General allometric relationships [e.g., Chambers et al., 2001; Chave et al., 2005] are then employed to converi these estimates to changes in woody biomass, and the changes ofthese terms per unit time are SLImmed over individual trees and then used to estimate total AG productivity, with some correction for the fraction of trees missed between censuses [Malhi et al., 2004J. All values reported here employ the Chambers et al. [2001] allometry, as modified by Bakel' et al. [2004] to incorporate . wood density where wood density values were given. The alternative widely employed tropical allometric equation from Chave et al. [2005] seems to overestimate the biomass oflai'ge Amazonian trees [Pyle et al., 2008]. As defined here, NPP stem includes the net woody production of the tree crown associated with changes in tree size and form, but excludes
360
PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS
any turnover and replacement of crown branches (this term, NPPbranch is discussed below). NPP stem at the Manaus and Caxiuana plots ranges between 2.0 and 2.9 Mg C a-I, values that are typical of eastern Amazonian forests [Malhi et aI., 2004]. Tapaj6s tends to show higher wood productivity over a wide area, particularly in the largerscale censuses [Pyle et aI., 2008]. As indicated above, this appears to be correlated with high CWD, indicating a large-scale disturbance event in the Tapaj6s region in recent years. The values at Manaus and Caxiuana are among the lower values of NPP stem reported for Amazonian forests [Malhi et al., 2004] (Figure 4). Wood productivity shows a distinctive regional trend across Amazonia, with highest values found in western Amazonia, both in the wet northwest and in the seasonally dry southwest. The values at the eastern site of Tapaj6s, however, compare with the high values found in western Amazonia. The generally higher values across western Amazonia irrespective of rainfall suggest that climate is not the major factor in detennining wood productivity. Instead, soils seem to exert more influence, most likely through the supply of phosphoms [Davidson et al., 2007], and there is a general trend of increasing wood productivity with increasing soil feliility and specifically soil phosphorus [Malhi et al., 2004; Quesada et al., 2009; Aragiio et al., 2009]. Lowland forest soils in western Amazonia tend to be less infertile, being typically of Pleistocene or Holocene age and initiated by meandering rivers depositing sediment eroded from the Andes. The lowest values of NPPstem are found on white sand soils in the upper Rio Negro region (northwest Amazonia). Soils in lowland eastern Amazonia have generally been weathered, eroded, and redeposited over a much longer timescale than those in western Amazonia.
3.2. Canopy Productivity The NPP of the canopy, NPPcanopy, is the annual rate of net fixation of carbon into the stmctures of leaves, flowers, fmit, and (for methodological convenience) small twigs (typically <1 cm diameter). For a near-equilibrium forest, its annual value can be measured through an array of litter traps collected at frequent (e.g., biweekly) intervals to minimize decomposition [Clark et al., 2001b]. Productivity estimates based on litterfall rely on the assumption that the litterfall is approximately equal to productivity ofthat component. This assumption is weaker if interannual variability is significant, especially for larger components such as branches, and is also complicated by the trapping and in situ decomposition of dead material in the canopy. Measurements are also challenged if the spatial pattern of litterfall is aggregated rather than a uniform "rain" of litter.
For Caxiuana and Manaus, fine litterfall values average 3.6-3.8 Mg C ha- I a-I (Figure 3). Tapaj6s is again an exception, with higher mean values approximately in proportion to the higher wood productivity. At Caxiuana, 73% of the litter was from leaves, 12% from flowers and fruit, 8% twigs, and 7% unidentifiable (Almeida and L. E. O. C. Aragao, unpublished data, 2008). In the context of the wider Neotropics, Malhi et al. [2004] reported a fairly strong linear relationship, between NPPlitter and NPP stem (this analysis included older data from the two focal sites Caxiuana and Manaus), with the highest value reported being fi'om Barro Colorado Island in Panama. The newer data from the 10 Amazonian NPP sites are consistent with this relationship (Figure 5). The lower wood productivity at Caxiuana and Manaus is reflected in lower canopy productivity, whereas both wood and canopy productivity at Tapaj6s are among the highest reported in Neotropical forests, even when compared with the productive sites of western Amazonia or the feliile terra preta site.
MALHI ET AL. GPP nux lower
Manaus (a)
361
=30.4; Predicted GPP =29.9±4.8
= 29.3±4.7
NPPtolal;:: 10.1±1.4 NPPAG = 7.3±1.3 NPP sG = 2.8±O.7
Predicted R soli = 12.6±2.3 Measured R soli = 12.1 ±1.7
3.3. Coarse Woody Littel1all Coarse woody litterfall can be an important component of forest carbon cycling, but is notoriously difficult to measure. It can be divided into three components: hunk mOliality, nonlethal large branch shedding, and twig and small branch fall (pieces between 1 cm and 10 cm diameter). The division from fine litterfall is often set at 1 cm diameter for twigs. This division is largely for methodological convenience: small branches are too heterogeneous in placement and too large to be adequately captured by fine litter traps. Trunk mortality, D stem , should be close to wood productivity for quasi-equilibrium forests, although for many plots, it has been reported to be slightly lower, resulting in a slight net AG biomass increment over time. Branch loss rates have been reported for Manaus and Tapaj6s [Chambers et al., 2001; Nepstad et al., 2002; Palace et al., 2008], and have a typical value of 1 ± 1 Mg C ha- I a-I [Malhi et al., 2009]. This is an approximate measure of branch turnover, NPPbranch (but see discussion by Malhi et al. [2004]) and is added to the total estimate of AG NPP. 3.4. Volatile Organic Compound Emissions The emission of volatile organic compounds (VOCs) from vegetation is a source of carbon from the vegetation, and can play an impOliant role in local and regional atmospheric chemistry [see Kesselmeier et al., this volume]. For the K34 tower near Manaus, Kuhn et al. [2007] report a total 24-h VOC flux (isoprene plus monoterpenes) of 24 Mg C m-2 day-I for the period 17-25 July 2001, which if fairly invari-
: Fdoc
=O.19±O.07
" Figure 3. Components of forest productivity of the three focal LBA Amazonian forests. (a) Manaus, (b) Tapaj6s, (c) Caxiuana. Units are in Mg C ha-1 a-I. The diagram shows the partitioning ofthe gross primmy productivity into components ofNPPtotal and autotrophic respiration.
ant over the year is equivalent to an annual total of 0.088 Mg C ha-I a-I. Similar values were reported by Greenberg et al. [2004] at Tapaj6s and by previous studies north of Manaus (summarized by Kuhn et al. [2007]). In carbon terms alone, the VOC flux is clearly a small component of the internal carbon cycle, and for our synthesis, we allocate a value of 0.1 ± 0.05 Mg C ha- I a-I for all three focal sites. Methane emissions from terra firme tropical forests are a new subject of interest, but still controversial. do Carmo et al. [2006] applied a canopy budget model to measured soil-atmosphere fluxes of methane to estimate net meth~ ane emissions from upland forests of 2-21 Mg CH4 m-2 d- I, equivalent to 0.005-0.06 Mg C ha- I a-I. Taking the midrange, Malhi et al. [2009] allocate 0.03 ± 0.03 Mg C ha- I a-I to methane emissions. Combining these with the estimate of isoprene and monoterpene emissions, they arrive
at an estimate of total volatile emissions of 0.13 ± 0.06 Mg C ha- 1 a-I.
3.5. Fine Root Productivity Two major components in BG productivity, DrOOl> are coarse root productivity and fine root productivity. A third component is the export of organic material in the form of exudates, or to symbionts such as mycorrhizal fungae and nitrogen-fixing bacteria. This third term is hard to quantify and is often treated as part of rhizosphere respiration (e.g., exudate that is rapidly metabolized is, for many practical purposes, indistinguishable from root respiration). Fine root productivity is defined as the production of root material less than a threshold diameter, usually 2 mm. The value calculated for fine root production can depend on
362
PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS GPPnux tower'" 31.4±0.4; Predicted GPP '" 29.3±4.4
Tapaj6s (b)
MALHI ET AL.
,.--------, R total::: 29.8±4.4
NPPtotal "'14.4±1.3
R aut = 14.9±4.2 R het = 14.9±1.4
NPPAG ", 11.4±1.2
~
~
c:::::> ~
n IN?? leaves,flowers,frUiI
L.:>
R leat "7.4'±4.0 ~
V
J;J0
065':7~'c.?:"'........ 01 -
:
30.8±4.2
NPPtotal = 10.0±1.2
=21.4±4.1
NPPAG '" 7.2±1.1 NPP SG = 2.8±O.6
INPp voc "o.13±0061
••.•• W"
,o'
j
::J
R lear= 8.9±4.0
o
+---v
6.5±0 7
Predicled R soil "14.1±1.2 Ir--'----'--------,
R cwo" 4.5±1.1
(C)
NPPSG '" 3.0±O.4
I0,,,_., Measured R soil = 12.0±O.6
GPP nux tower = 38.2±2.0; GPPecoPhysiology"'31.2; Predicted GPP '" 31.4±4.4
Caxiuana
= =
363
I 0,,"" "eO,,,
R slem = 5.1 ±O.5 - - - - I
0
38±O.1
Predicted R soil" 14.6±O.9
Irl-D---'cw-o-=~'2-.-9±-1-.0--1
Dcwo " 3.7±1.0
i~?j
~:~~~L~~' R soli hel,"10.4±O,9 : Fdoc
O,0004±O,0004
~
:
F doc
:::
O.19±O.07
~
Figure 3b. (continued) Figure 3c. (continued) sampling methodology, in particular, the technique employed and the depth to which the soil is sampled. Fine root production values have been reported for Caxiuana and Tapaj6s. For Caxiuana, rhizotron and ingrowth core methods were used to 30 cm [Metcalfe et af., 2007b]; for Tapaj6s, a combination of sequential coring and root trenching were employed [Silver et af., 2000]. An important discrepancy between these two studies is the depth to which root production is measured, 30 cm at Caxiuana versus 10 cm at Tapaj6s. Mafhi et af. [2009] attempt to COlTect for this discrepancy by using root profiles to standardize to 1 m depth. Once this is done, there is no significant difference in estimates offine root productivity between the two sites (Caxiuana, 2.2 ± 0.6 Mg C ha- 1 a-I; Tapaj6s, 2.0 ± 0.3 Mg C ha- 1 a-I). For Manaus, no data were available, and we take the mean of the Caxiuana and Tapaj6s values, with conservative error bars of ±1.0 Mg C ha- 1 a-I. At the RAINFOR NPP sites, root productivity was estimated from ingrowth cores as reported by Aragfio et af. [2009].
The resulting values are shown in Figures 3 and 4. 3.7. Dissofved Organic Carbon Leakage
Coarse root productivity is the productivity of larger, more lignified roots. These can be divided into roots <10 cm diameter, 'which can be expected to be reasonably homogeneous in distribution and amenable to conventional random or grid-based sampling, and massive structural roots, which are velY difficult to assess for biomass without excavation, and even more difficult to monitor over time. For massive structural roots, the best approach may be to assume that wood production per unit biomass is the same as for AG woody biomass and use the estimate of BG biomass being 21 ± 3% of AG biomass (see carbon stocks section above). When added to estimates of fine litterfall above, this yields values of total BG NPP of 2.9 ± 0.6 Mg C ha- I a-I (Caxiuana), 2.8 ± 0.7 Mg C ha- I a-I (Manaus), and 3.0 ± 0.5 Mg C ha- 1 a-I (Tapaj6s).
the same study, DOC concentrations in rainfall were 1.2 mg
1-1, resulting in millual DOC deposition rates in dissolved
3.6. Coarse Root Productivity
The possibility of waterborne carbon disappearing laterally through either surface runoff or groundwater slow into streams has been invoked as a possible reason for apparent carbon sinks in eddy covariance studies and as a partial source of the high CO 2 respired by rivers [Richey et af., 2002; see also Richey et af., this volume]. Waterloo et af. [2006] measured dissolved organic carbon (DOC) runoff in the Igarape Ayu catchment, which covers an area of 6.8 km2 , including the Manaus K34 micrometeorological tower. Net carbon exports over 2 years (2002 and 2003) amounted' to 0.19 ± 0.07 Mg C ha-I a-I. Almost all ofthis export seemed to originate in the riparian zone of the valley bottoms, which cover 35% of the watershed area in the local landscape. For plateau regions, the total DOC exports through groundwater seemed to be much less, around 0.0005 Mg C ha- I a-I. In
rainwater of 0.03 Mg C ha-I a-I. A significant proportion of DOC inputs are probably derived from scavenging of aerosols during rainfall formation and dry deposition. Dissolved inorganic carbon was not reported but is unlikely to be much greater in magnitude. For their synthesis, Mafhi et af. [2009] applied the values of DOC transfer reported from the Igarape Ayu catchment for Caxiuana and Manaus. The Tapaj6s site is a plateau cut by few rivers, so they employed the values for only plateaus reported by Waterloo. They report net DOC export (DOC runoff - DOC deposition). In all cases, DOC transfer is clearly a very small component of the forest carbon cycle, The figures for the Ayu catchment are comparable with the annual average carbon export for the Rio Negro basin over 1982-1984 of 0.126 Mg C ha- I a-I [Richey et af., 1990], suggesting that broadly similar processes operate across the
364
PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS 12 ..,------------
- ------ ---
MALHI ET AL.
-- ---.- --- ----- -- --- ------------------ ------- -------,
Malhi eta!.
6.0
8 +-------------
m-2 at Caxiuana. When these are scaled by the Stem Area Index for trees> 10 cm diameter at breast height (calculated using the formula shown by Chambers et al. [2004]), the per unit ground area fluxes are 4.2 ± 1.0 Mg C ha- l a-I, 3.8 ± 1.0 Mg C ha-] a-I, 5.1 ± 0.5 Mg C ha-I a-I, respectively. Similar values were reported by Meir and Grace [2002]: a mean value of ~0.6 ~mol m-2 stem area S-I for 23 species at Jarl], Rondonia, Brazil. foda [1983] reported that stem respiration rates increase with height along the tree; hence, the branch respiration estimate of Chambers et al. [2004] is probably low. More recently, Cavaleri et al. [2006] reported that, at La Selva, Costa Rica, per unit area respiration rates are much higher in branches than on the main tmnk. This suggests that our estimates may be biased to being too low, although the total stem respiration reported at La Selva (5.08 ± 1.35 Mg C ha- I a-I) is not very different from that reported at these Amazonian sites.
7.0
10-1----
•
TAP-O'!
2009
6
5.0
~
'7(0
4
'7(0
2
.c
o
0>
~
-2
~
-4
ll..
-6 -8 -1-------------------------------------------10
!il NPPfinelitter
B NPPstem 0 NPPbrancl1 0 NPPfineroot
III NPPcoarseroot
365
1.0
0 NPPvoc
0.0 MAN
CAXfower CAXSand
ZAR01
TAP04
AGPOl
AGP02
TAM05
TAM06
CAXfP
Sites
0.0
1.0
2.0
3.0
4.0
5.0 1
Aboveground wood production (Mg C ha· Figure 4. The components ofNPP at the RAINFOR NPP sites, plotted in sequence of increasing soil phosphorus from left to right. Units are in Mg C ha-] a-I. BG NPP values are plotted as negative. From Aragiio et af. [2009].
Amazon Basin, although there is likely to be local variation according to soil type and precipitation regime. 4. RESPIRATION FLUXES Next, we turn our attention to the efflux of gaseous CO 2 from the system. This is termed "respiration" and can be divided into two terms, "autotrophic respiration" (C0 2 directly respired by plants as a breakdown product from their own metabolic activity), and "heterotrophic respiration" (C0 2 respired by herbivores, detritivores, and higher trophic levels as they consume and break down organic matter). In terms of the carbon cycle, the respiration is often conceived as a loss term (net carbon gain = photosynthetic input - respiratmy loss), but this can be misleading. Respiration rates reflect the (usually efficient) allocation of metabolic activity of the plant or heterotrophic communities, whether to stem growth, leaf, or root tissue construction, protein maintenance, or reproduction. Plants or ecosystems that respire a smaller fraction of their fixed carbon are not necessarily any more "efficient," they are simply prioritizing other activities than biomass constmction. The ratio between NPP and autotrophic respiration reflects partitioning of the energy captured in plant photosynthesis [the gross primmy productivity (GPP)] between constmction of new organic material and work done in metabolic activity:
GPP = NPP + Rautotrophic. Similarly, the ratio between autotrophic and heterotrophic respiration reflects the relative amount of metabolic activity occurring at the lowest trophic level (the plants and photosynthesizing bacteria) and in the sum of all other trophic levels (animals, fungi, bacteria). 4.1. Root Respiration
The partitioning of soil respiration into autotrophic (root) and heterotrophic respiration is helpful in terms of interpreting processes, but presents numerous methodological challenges [Baggs, 2006]. Malhi et al. [2009] report on measurenients from Caxiuana and Manaus. At Caxiuana, Metcalfe et al. [2007a] applied the direct extraction approach. Silver et al. [2005] quantified root respiration at Tapaj6s using two approaches: (1) a trenching experiment around a 2.5 m x 2.5 m block of land to 1 m depth, (2) a steady state mass balance approach based on quantifying AG and BG litter input, assuming that heterotrophic respiration rates are equal to litter input rates, and allocating the remaining soil respiration to root respiration. The trenching approach had varied success: here only results from the mass balance approach are reported. There is a factor of two difference between the Silver et al. [2005] and Metcalfe et al. [2007a] estimates (Figure 2). This
4.3. LeafRespiration
a·1)
Figure 5. The relationship between the AG wood carbon production and the totallitterfall production, both in Mg C ha -1 a-I. Gray dots are values fro111 across the Neotropics reported by Malhi et al. [2004], black dots are the 10 sites presented here. The linear fit (black line) incorporated all points and was forced through the origin, yielding a relationship, NPPfinclitter= 1.61(±0.07) x NPP stcm . From Aragiio et al. [2009].
could reflect (l) genuine between-site differences, (2) an overestimation of root respiration by the extraction method at Caxiuana, (3) an overestimation of heterotrophic respiration by the trenching and mass balance approaches. Details are discussed by Malhi et al. [2009]. For Manaus, we have no direct data and take the mean of the Caxiuana and Tapaj6s values, with conservative error bars of ±2 Mg C ha- I a-I to encompass the mean values of both other sites. For further review of soil and root carbon dynamics, see Trumbore and de Camargo [this volume]. 4.2. Stem Respiration
The respiration of carbon dioxide from stems reflects the metabolic activity of stem maintenance and growth (and pos-' sibly some efflux of CO 2 carried in the stem water stream). Stem respiration has been measured at all three sites [Chambers et al., 2004; Nepstad et al., 2002; Teixeira et al., unpublished data, 2008]. Per unit stem area, the respiration rates were 0.6 ~mol m-2 at Manaus and Tapaj6s and 0.78 ~mol
The respiration of leaves is a major plant metabolic activity, but is complicated by a number of definition and measurement issues. The first issue is to distinguish between photorespiration, the release of some CO 2 mediated by Rubisco and an intrinsic part of plant photosynthetic processes, and mitochondrial ("dark") respiration, which reflects the metabolic activity iin the plant liberating energy utilized for plant maintenance and growth. In a diurnal cycle, mitochondrial respiration would be expected to increase with leaf temperature (and the amplitude of the diurnal cycle varies considerably within the canopy according to leaf position and sun and wind exposure), but can also decrease strongly with increasing solar radiation [Atkin et al., 2000]. This decrease occurs because photosynthesis becomes a direct provider of adenosine triphosphate (ATP) for plant metabolic processes, reducing the demand for this from mitochondria. The approach we adopt here is to attempt to estimate daytime photoinhibition and, hence, arrive at a total leaf dark respiration term that incorporates all leaf mitochondrial activity. An alternative approach that is sometimes adopted [e.g., Litton et al., 2007] is to focus only on nighttime dark respiration [e.g., Meir et al., 2008; Lloyd et al., 2002] and ignore daytime dark respiration. We apply a 67% reduction of measured daytime dark respiration rates to allow for daytime photoinhibition, based on the Atkin et al. [2000] photoinhibition equations. Details of measurements and corrections are discussed by Malhi et al. [2009]. At Manaus, we estimate a leaf respiration rate of 10.0 Mg C ha- I a-I modified from data reported by Chambers et al. [2004]; at Tapaj6s, 7.4 Mg C ha- I a-I modified from data reported by Domingues et al. [2005] and at Caxiuana, 8.9 ±
366
PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS
1.4 Mg C ha- I a-I, derived from L. E. O. C. Araguo et al. (unpublished data, 2008). In summary, leaf respiration is probably the largest single term in the futernal carbon budget, but determination of its exact magnitude remains complex. The estimated sampling uncertainty for Caxiuanu (±I.4 Mg C ha- I a-I) does not account for potential systematic uncertainties in process and scaling, imd for their synthesis table, Malhi et al. [2009] applied a more conservative uncertainty estimate of ±4.0 Mg C ha- I a-I to all three focal LBA ~ites. 5. ECOSYSTEM PHOTOSYNTHESIS AND RESPIRATION Above-canopy eddy covariance studies attempt to measure the net carbon flux or net ecosystem exchange (NEE) in and out of the forest canopy [see Saleska et al., this volume] and, hence, enable the estimation of ecosystem GPP and ecosystem respiration (R e), by consideration of amplitude of the diurnal cycle of net carbon flux as outlined by Reichstein et al. [2005]. The method relies on assuming that nighttime flux data (or an appropriately filtered subset of these data) are reliable and extrapolating into the daytime (usually on the basis of temperature) to estimate daytime respiration. The difference between measured net carbon uptake and the estimated ecosystem respiration (which is in the opposite direction) is then the estimated gross primary production. The major uncertainties in this approach are (l) the estimation of nighttime ecosystem respiration fluxes, which are highly problematic in calm tropical conditions, particularly within tall canopies, which decouple subcanopy air from abovecanopy turbulence; (2) the assumptions behind extrapolation into daytime, in pmiicular, which measurement of temperature to use and how to account for photoinhibition of leaf respiration. Once ecosystem respiration has be estimated, GPP can be calculated as:
respectively) and significantly higher at Tapaj6s (14.4 ± 1.3 Mg C ha- I a-I). The largest components ofNPP are leaf/ flower/fruit/twig production, followed by stem production. There is little evidence of any significant variation in BG NPP between the three LBA sites (though estimated fine root turnover at Manaus is simply an average of the other two sites). Hence, the high NPP at Tapaj6s is entirely explained by a disproportionate allocation to AG wood and foliar production. Once other sites across Amazonia are considered (Figure 4), it seems that BG NPP tends to increases almost in step with AG NPP, as soil fertility increases. Hence, disturbance appears to shift allocation AG, as trees compete for light in newly created gaps, whereas fertility does not appear to cause as large a shift in allocation. The components of autotrophic respiration are much more challenging to quantify and the largest source of uncertainty in our calculations. Total estimates of autotrophic respiration are 19.8 ± 4.6 Mg C ha- I a-I (Manaus), 14.9 ± 4.2 Mg C ha- I a-I (Tapaj6s), and 21.4 ± 4.1 Mg C ha- I a-I (Caxiuanu). Leaf respiration is the largest component and the most uncertain. There are considerable methodological differences between sites in measurements of leaf and root respiration [outlined in Malhi et al., 2009] that may explain some ofthe difference between sites. Such comprehensive measurements of carbon cycling at these sites enable two independent checks of self-consistency. First, we can compare against measurements of soil respiratOlY CO 2 efflux [Malhi et al., 2009]. The expected soil respiration can be calculated from rates of carbon inflow into the soil.
R soil , expected = Rroot + R SOM . If we assume quasi-equilibrium conditions on an mmual time scale and negligible intermmual variability, the heterotrophic respiration is:
GPP = R e - NEE, where a negative NEE indicates a net carbon flux into the forest canopy. Reported estimates of GPP are summarized in Figure 3. 6. A COMPREHENSIVE VIEW OF THE FOREST CARBON CYCLE AT THE LBA FOCAL SITES The values ofNPP and respiration distilled from the three focal LBA study sites are displayed in Figures 2 and 3. Considering the components of net primmy production first (Figure 3), the NPP at Manaus and Caxiuanu is similar (l0.1 ± 1.4 Mg C ha- I a-I and 10.0 ± 1.2 Mg C ha- I a-I,
R SOM = NPPfineroot + NPPfinelitter + F(cw-soil) . (NPPtrunk + NPPbranch) + NPPBG - i1C - F doc ,
where F(cw-soil) is the fi'action of CWD that is transferred to the soil estimated as 0.24 ± 0.15 [Malhi et al. 2009], F(cw-soil) is the BG root biomass fraction (estimated as 0.21 ± 0.03; see above), and i1C is the change in soil carbon stocks. We assume there is negligible change in soil carbon stocks, (i.e., i1C < < Rsoil), an assumption supported at Tapaj6s by radiocarbon studies [Telles et al., 2003], and the resulting calculations of expected R soil are shown in Figure 3. The largest contributors to soil respiration are fine litter, which is fairly
MALHI ET AL.
well-quantified, and root respiration, which carries greater methodological uncSiftainty. The material derived from CWD component 1/jS relatively minor, and hence, assumptions about the e1'3ct value of F(cw-soil) are not particularly important. Whed expected soil respiration is compared to measured soil respiration (Figure 3); there is a high consistency between approaches, increasing confidence that our understanding of the bulk flows of the BG carbon cycle at these sites is fairly complete (although there are many details in the processes and their sensitivities to environmental factors that are still to be understood). Agreement is particularly good at Manaus, but at the other two sites, there is some suggestion that less respiration is being measured than expected. As a second cross-check, we can calculate GPP from bottom-up measurements with those estimated from flux towers. The GPP of the forest is by definition the sum of NPP and autotrophic respiration: GPP = NPP + Rantotrophic. These predicted values of GPP for the focal sites .are 29.9 ± 4.8 Mg C ha- I a-I (Manaus), 29.3 ± 4.4 Mg C ha- I a-I (Tapaj6s) and 31.4 ± 4.4 Mg C ha- I a-I (Caxiuanu). The error bars in these estimates are dominated by the large elTors we ascribe to leaf respiration. When these estimates are compared to those from flux towers (Figure 3), the agreement is close at Manaus and Tapaj6s. This greatly increases confidence in both these approaches. At Caxiuanu, the flux tower estimate is substantially higher, but our estimate is almost identical to the Fishel' et al. [2007] estimate (31.2 Mg C ha-I a-I) derived from measured photosynthetic parameters and canopy hydrology. This hints the problem may be with the Caxiuanu flux tower, rather than with the "bottom-up" measurements (the Caxiuanu tower is situated 6 km downwind from a velY large water body, which generates large-scale circulations, which may complicate flux measurements). The dominant term in ecosystem respiration appears to be leaf respiration, followed by root respiration, stem respiration, and fine litter decomposition, all ofroughly equal magnitude. With some caveats, the fairly close agreement between the two approaches (flux towers or ecophysiology, and bottomup measurements) indicates there are no velY large terms missing, such as enhanced respiration from branches, litter decomposition in situ in the canopy, understory respiration, etc. [Malhi et al., 2009]. It is now possible to calculate the ecosystem carbon use efficiencY,the fraction ofGPP that is allocated to NPP. NPP CUE cco = - GPP
NPP = 1- R auto (NPP + R auto ) GPP
.
367
The values ofCUEeco at Caxiuanu (0.32 ± 0.07) are similar to those at Manaus (0.34 ± 0.10), confirming the picture of low carbon use efficiency in old-growth tropical forests suggested by Chambers et al. [2004] for Manaus. At Tapaj6s, however, CUE eco is higher with a mean value of 0.49 ± 0.16, closer to the values reported in many temperate broadleaf forests. Given the large error bars around the CUE estimates, however, the difference is not significant (z test, p = 0.14). Fundamentally, this difference reflects the fact that the observations of higher wood and litter production at Tapaj6s are not matched by a higher GPP as seen by both the flux tower and the sum of "bottom-up" measurements. Hence, at Tapaj6s, there is some suggestion of disproportionate allocation to above ground productivity (canopy and woody growth), with a commensurate reduction in metabolic activity such that overall GPP is velY similar between the three focal LBA sites. The difference in carbon cycling between Tapaj6s and the other two sites may therefore primarily reflect differences in allocation rather than differences in photosynthesis. The most plausible hypothesis to explain this difference is that there is likely to have been a significant mortality event in Tapaj6s in the 1990s [Pyle et al., 2008] and that in the aftermath of the mortality, there is a surge in growth with surviving and newly recruiting individuals competing for increased light availability by allocating disproportionately to wood and canopy production, thus causing an increasq in CUE. 7. SPATIAL INTERPOLATION TO THE WIDER AMAZON REGION The three LBA focal study sites are located in the region associated with some of the least dynamic, slowest-growing forests in Amazonia [Malhi et al., 2009]. The new RAINFOR NPP sites (Figure 4) present part of the first comprehensive multisite assessments ofthe carbon cycle published for other parts of Amazonia. Cavaleri et al. [2008] present a fairly comprehensive assessment for the La Selva forest in Costa Rica. Plate 1 [from Malhi et al., 2006] presents regional extrapolations of basal area, AG wood productivity, and wood residence time (defined as AG live biomass/stem productivity). A simple extrapolation based on luiging is employed here to indicate general trends; more sophisticated studies using soil maps and/or remote sensing metrics would indicate differences in detail but the same broad trends. The increase in productivity from east to west is mirrored by a corresponding decrease in biomass residence time (defined as AG woody biomass divided by AG woody productivity). This is the average time that carbon stays fixed in live biomass in an Amazonian forest. The mean residence
MALHI ET AL.
368 PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS 200-250
a
100N
b
100 N
250-270 270-275 275-280 280-285 285-290 290-300 300-305 305-310 310-315 315-320 320-325 325-330 330-335 335-340 340-345 345-350 >350
~5-20
20-25 25-26 26-27 27-28 28-29 29-30 30-31 > 31
10'N ,-.,
.-<
~
.-<
.2 U
~
6 ~ ~
~
2.3-2.5 2.5-2.6 2.6-2.7 2.7-2.8 2.8-2.9 2.9-3.0 3.0-3.1 3.1-3.3 3.3-3.5 3.5-3.7 >3.7
10'5
20-30
0-25 25-50 50-75 75-100 100-150 150-200 200-250 250-300 300-350 350-400 >400
10°5
Plate 2. Spatial variation of AG biomass across Amazonia determined by (a) Malhi et al. [2006] using geo-statistics and (b) Saatchi et al. [2007] using remote sensing. Note that Plate 2b covers a slightly larger area. Units are in Mg illy weight ha- 1 (1 Mg dty weight ~ 0.5 Mg C).
30-35 35-40 40-45 45-50 50-55 55-60 60-65 65-70
> 70
80'W
60'W
Plate 1. Spatial variation of (a) basal area (m 2 ha- 1), (b) AG coarse wood productivity (NPP stem - Mg ha- 1 a-i) and (c) wood residence time (years) across Amazonia. White areas are outside the boundaries of intact evergreen forest. From Malhi et al. [2006], copyright Wiley-Blackwell.
time in eastern Amazonia is 65-70 years, but declines to 30--40 years in western Amazonia. Our basic extrapolation suggests that the region-wide AG wood productivity of Amazonia is around 1.7 Pg C a- 1 (per unit area, 2.93 Mg C ha- 1 a-I). This number increases by about 21% (to 2.06 Pg C a-I) if BG biomass is taken into account. If we take our estimate of NPPcanopy = 1.61 x NPP stem to be general (Figure 4), the total AG NPP of Amazonian forests is 4.4 Pg C a-I. Malhi et al. [2006] use a similar but larger data set to estimate an AG live biomass of 93 ± 23 Pg C a-I (see below).
Dividing the AG biomass by the wood productivity suggests a mean residence timyfor live woody biomass of 55 years. In conclusion, Amazo~lian tropical forests incorporate about 2 thousand millio~!tons of carbon in woody biomass each year, which resi~s in the live biomass for about 55 years. Two recent regional extrapolations of biomass are compared in Plate 2. Plate 2a plots an inverse-distance-weighting extrapolation based on forest plot data corrected for wood density variation [Malhi et al., 2006]; Plate 2b plots a remote sensing-based extrapolation built from a different, partially overlapping biomass data set [Saatchi et al., 2007]. The interpolation of tree biomass presented by Malhi et al. [2006] incorporates in-depth understanding of forest structure and wood density, but relies on relatively crude extrapolation from a few sample points to the wider region. It may identify key regional trends, but is unreliable as a predictor of biomass for any particular region. In particular, Malhi et al. [2006] identify some decrease in biomass in the more productive regions, which reflects a decrease in mean wood density and, in turn, reflects a life-history trade-off as faster growing species with low wood densities increase in abundance relative to high wood density, slow-growing species. As an alternative approach, Saatchi et al. [2007] apply multiple remote sensing layers to a (different) biomass data set and utilize a tree-based regression approach and direct estimation techniques to map the AG live biomass of forests at a moderate spatial resolution (l km) over the entire Amazon basin and sUlTounding areas. The methodology relies on the sensitivity of the remote sensing measurements to various attributes of forest cover such as canopy roughness and moisture, tree density, leaf area index, crown and stem volume to extrapolate the ground biomass data over the basin, but incorporates less direct ecological understanding of forest composition. These attributes are known to be strongly correlated with the biomass density [Saatchi et al., 2007; Chambers et al., 2007; Liddell et al., 2007; Alves and Santos, 2002]. The biomass distribution captured both the large-scale variations of the carbon stores across the Amazonian and finer-scale heterogeneities at the landscape level associated with variations in soil, geomorphology, topography, and moisture gradients. Both the remote sensing and ecologically derived maps show similar trends with the highest biomass in the northeast and central Amazonia and lower biomass in the west and south. At the large scale, these high-biomass regions COlTe- . spond to areas with high rainfall and short dty season [Malhi et al., 2006; Saatchi et al., 2007]. At the finer scale, areas of northwest and southwest Amazonia with lower biomass have larger heterogeneity that may be associated with species composition, higher turnover, wood density-basal area
369
ratio and, to some extent, the soil fertility and topographical variations [Bakel' et al., 2004; Saatchi et al., 2007]. In total, Malhi et al. [2006] estimate an AG live biomass of 93 ± 23 Pg C over a forest area of 5.76 x 106 km2 (including a 10% correction for the biomass of small trees and lianas, which are not usually included). Saatchi et al. [2007] arrive at a smaller estimate of 66 ± 15 Pg C over an area of 5.46 x 10 6 km 2 . The difference can partially be explained by the inclusion of savmma in the Saatchi et al. estimate and by the omission of a 10% cOlTection for small trees and lianas. A logical next step is to combine both approaches, utilizing the ecological insight from in-depth plot level studies with multiple remote sensing measurements of forest structure, moisture and phenology. One such approach would be to first generate and interpret maps of relevant parameters such as wood density, forest structure (e.g., basal area, height, fraction oflarge trees) and then build up to a regionwide map of biomass. 8. WHAT CONTROLS THE BIOMASS AND DYNAMISM OF AN AMAZONIAN FOREST? Figures 3 and 4 highlight that woody production accounts for only a small proportion of the NPP in the three Amazonian forests stud,ed here and an even smaller proportion of the GPP. Hence, it is quite likely that small shifts in carbon allocation can generate large shifts in wood productivity (Figure 4). The~e shifts may matter more than shifts in photosynthesis in detennining spatial and temporal patterns in wood production. An initial assumption may be that areas of high productivity correspond to areas of high biomass. This is manifestly not the case (Plate 1). An emerging insight is that biomass of old-growth systems seems to be determined less by productivity and more by turnover or residence times. Put another way, for near-equilibrium old-growth forests, the rate of mortality generally increases as wood productivity increases. This suggests that for given environmental conditions, there is some fOlm of cap on the "canying capacity" of biomass of an old-growth tropical forest. A likely cause of this cap is competition between tree crowns for light resources. A closed tropical forest canopy already captures or reflects almost all of the incident photosynthetically active radiation, and in the absence of changes in incoming solar radiation, increased supply of other limits to productivity is likely to intensify competition for light resources and thereby enhance mortality among those individuals that lose out in light competition. Hence, a stand-level constraint on biomass emerges that cannot easily be identified from understanding response of individual trees to the environment. A boost in productivity thereby induces a boost in mortality in the long-term,
370
PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS
although there may be transient increases in biomass as the system tries to re-equilibrate. This insight emerges from analysis of spatial patterns of productivity llnd biomass. When considering changes in the biomass of old-growth forests over time, for instance, in response to rising carbon dioxide considerations, similar constraints may come into play. There may be a short-term increase ih biomass, but the long term competition for light may present a stand-level constraint on total biomass. It may be the response of this stand-level constraint to environmental change that determines future changes in the biomass of old-growth forests, rather than individual-level responses in productivity. For example, increased atmospheric CO 2 concentrations may increase water use efficiency, affecting the maximum height of canopy trees. Any stimulation of productivity (whether stimulated by soil fertility over space, or by CO 2 over time) could result in a more dynamic forest with increased abundance of pioneers, lianas, and other disturbance-favoring taxa. The interaction between increasing dynamism, forest structure, and plant traits also have the potential to act as a positive feedback, or ecological "amplifier." As forest mortality rates increase, there may be greater gap fi'equency in the canopy and greater light penetration to the understory. This will favor fast-growing, short-lifetime species, which fmiher increases mortality and canopy openness. Such discussion is speculative, but highlights our lack of knowledge of the fundamental controls on the biomass of old-growth tropical forests. In conclusion, intensive LBA research at the three focal sites presented here has painted a comprehensive picture of the allocation of productivity at at least these sites in eastern Amazonia and demonstrated how sensitive the woody biomass growth is to small shifts in carbon allocation priorities within the forest. The RAINFOR project has demonstrated the generally higher productivity of Western Amazonia, but it is still a subj ect of active research as to whether that higher productivity is driven by increased photosynthesis or by a shift of allocation to woody production. We emerge fi'om a decade ofLBA-related research with a fairly comprehensive picture of fine-scale local dynamics at particular sites and a gradually emerging (but far from complete) sense oflargescale regional variations in NPP and carbon cycling. Studies of the contemporaty carbon cycle of the Amazon region can now be based more on rich data validation at multiple sites rather than model assumption. These insights have answered some questions and inevitably raised more questions and challenges that will be the focus of another decade of research. We would venture to suggest that many answers to these questions lie in new study sites away from the LBA sites that have been the focus ofthis chapter. In particular, the barely studied forests of western Amazonia, the floodplains,
the crystalline shield, and the Andes beckon. Our journey toward comprehending the greatest "carbon machine" on the land surface is only just beginning.
REFERENCES Alves, L. F., and F. A. M. Santos (2002), Tree allometry and crown shape oHour tree species in Atlantic rain forest, southeast Brazil, J. Trop. Ecol., 18,245-260. Aragao, L. E. O. C., et al. (2009), Above- and below-ground net primary productivity across ten Amazonian forests on contrasting soils, Biogeosci. Discuss., 6, 2441-2488. Atkin, O. K., l R. Evans, M. C. Ball, H. Lambers, and T.L. Pons (2000), Leaf respiration of snow gum in the light and dark interactions between temperature and irradiance, Plant Physiol., 122,915-923. Baggs, E. M. (2006), Partitioning the components of soil respiration: A research challenge, Plant Soil, 284,1-5. Baker, T. R., et al. (2004), Variation in wood density determines spatial patterns in Amazonian forest biomass, Global Change Bio!., 10,545-562. Baker, T. R., E. N. Honorio Coronado, O. L. Phillips, l Martin, G. M. F. van der Heijden, M. Garcia, and J. S. Espejo (2007), Low stocks of coarse woody debris in a southwest Amazonian forest, Oecologia, 152,495-504. Cairns, M. A., S. Brown, E. H. Helme, and G. A Baumgardner (1997), Root biomass allocation in the world's upland forests, Oecologia, 111, I-II. Castro, E. A., and l B. Kauffman (1998), Ecosystem structure in the Brazilian Cerrado: A vegetation gradient of aboveground biomass, root mass and consumption by fire, J. Trop. Ecol., 14, 263-283. Cavaleri, M. A, S. F. Oberbauer, and M. G. Ryan (2006), Wood CO 2 efflux in a primary tropical rain forest, Global Change BioI., 12,2442-2458. Cavaleri, M. A., S. F. Oberbauer, and M. G. Ryan (2008), Foliar and ecosystem respiration in an old-growth tropical rain forest, Plant, Cell Environ., 31(4), 473--483, doi:IO.1111/j.1365-3040. 2008.01775.x. Chambers, l Q., l dos Santos, R. l Ribeiro, andN. Higuchi (2001), Tree damage, allometric relationships, and above-ground net primary production in central Amazon forest, For. Ecol. Manage., 152, 73-84. Chambers l Q., et al. (2004), Respiration from a tropical forest ecosystem: Partitioning of sources and low carbon use efficiency, Ecol. Appl., 14, S72-S88. Chambers, J. Q., G. P. Asner, D. C. Morton, L. O. Anderson, S. S. Saatchi, F. D. B. Espirito-Santo, M. Palace, and C. Souza Jr. (2007), Regional ecosystem structure and function: Ecological insights from remote sensing of tropical forests, Trends Ecol. Evol., 22(8), 414--423. Chave, l, et al. (2005), Tree allometry and improved estimation of carbon stocks and balance in tropical forests, Oecologia, 145, 87-99.
MALHI ET AL. Clark, D. A., S. Brown, D. W. Kicklighter, l Q. Chambers, l R. Thomlinson, l Ni, andE. A. Holland (200Ia), Net primary production in tropical forests: An evaluation and synthesis of existing field data, Eco!YAppl., 11, 371-384. Clark, D.. A, S. B0vn: D. W. Kickligh:er, l Q.. Chambers, l.R. Thomlmson, and l Nl (200 Ib), Measunng net pl'1ll1aly productIOn in forests: Concepts and field methods, Ecol. Appl., 11, 356-370. Davidson, E. A, et al. (2007), Recuperation of nitrogen cycling in Amazonian forests following agricultural abandonment, Nature, 447, 995-998. do Carmo, lB., M. Keller, l D. Dias, P. B. de Camargo, and P. Crill (2006), A source of methane from upland forests in the Brazilian Amazon, Geophys. Res. Lett., 33, L04809, doi:10.1029/ 2005GL025436. Domingues, T. F., l A BelTY, L. A. Martinelli, l Ometto, and l R. Ehleringer (2005), Parameterization of canopy strncture and leaf-level gas exchange for an eastern Amazonian tropical rain forest (Tapaj6s National Forest, Para, Brazil), Earth Interact., 9(17), EII49, doi:10.1175/EII49.1. Fisher, R. A, M. Williams, A L. da Costa, Y. Malhi, R. F. da Costa, S. Almeida, and P. Meir (2007), The response of an Eastern Amazonian rain forest to drought stress: Results and modeling from a throughfall exclusion experiment, Global Change BioI., 13, 2361-2378. Greenberg, l P., A. B. Guenther, G. Petron, C. Wiedinmyer, O. Vega, L. V. Gatti, l Tota, and G. Fisch (2004), Biogenic VOC emissions from forested Amazonian landscapes, Global Change Bio!., 10, 651-662. Jackson, R. B., J. Canadell, J. R. Ehleringer, H. A Mooney, O. E. Sala, and E. D. Schulze (1996), A global analysis of root distributions for terrestrial biomes, Oecologia, 108, 389--411. Keller, M., M. Palace, G. P. Asner, R. Pereira, and l N. M. Silva (2004), Coarse woody debris in lU1disturbed and logged forests in the eastem Brazilian Amazon, Global Change BioI., 10, 784-795. Kesselmeier, l, A Guenther, T. Hoffmann, M. Piedade, and l Warnke (2009), Natural volatile organic compound emissions from plants and their roles in oxidant balance and particle formation, Geophys. Monogl'. Sel'., doi: 10. 1029/2008GM000717, this volume. Klinge, H. (1973), Root mass estimation in lowland tropical rain forests of central Amazonia, Brazil. 2. Coarse-root-mass of trees and palms in different height classes, An. Acad. Bras. Cienc., 45, 595-609. Kuhn, U., et al. (2007), Isoprene and monoterpene fluxes from Central Amazonian rainforest inferred from tower-based and airborne measurements, and implications on the atmospheric chemistlY and the local carbon budget, Atmos. Chem. Phys., 7, 2855-2879. Liddell, M. J., N. Nieullet, O. C. Campoe, and M. Freiberg (2007), Assessing the above-ground biomass of a complex tropical rainforest using a canopy crane, Austral Ecol., 32(1), 43~58, doi: I0.III1/j.1442-9993.2007.0 1736.x. Litton, C. M., l W. Raich, and M. G. Ryan (2007), Carbon allocation in forest ecosystems, Global Change BioI., 13,2089-2109. Lloyd, l, et al. (2002), Seasonal and annual variations in the photosynthetic productivity and carbon balance of a central Siberian pine forest, Tellus, Sel'. B, 54, 590-610.
371
Lloyd, l, M. L. Goulden, l P. Ometto, S. Patino, N. M. Fyllas, and C. A. Quesada (2009), Ecophysiology of forest and savmma vegetation, Geophys. MOl/ogr. SCI'. , doi: 10.1 02912008GM000740, this volume. Malhi, Y., et al. (2001), An international network to understand the biomass and dynamics of Amazonian forests (RAINFOR), J. Veg. Sci., 13,439--450. Malhi, Y, et al. (2004), The above-ground coarse wood productivity of 104 Neotropical forest plots, Global Change BioI., JU, 563-591. Malhi, Y., et al. (2006), The regional variation of aboveground live biomass in old-growth Amazonian forests, Global Change BioI., 12(7), 1107-1138, Malhi, Y, et al. (2009), Comprehensive assessment of carbon productivity, allocation and storage in three Amazonian forests, Global Change BioI., 15,1255-1274. Meir, P., and J. Grace (2002), Scaling relationships for woody tissue respiration in two tropical rain forests, Plant Cell Environ., 25,963-973. Meir, P., D. B. Metcalfe, A C. L. Costa, and R. A Fisher (2008), The fate of assimilated carbon during drought: Impacts on respiration in Amazon rain forests, Philos. Trans. R. Soc. Sel'. B, 363,1849-1855. Metcalfe, D. B., et al. (2007a), Factors controlling spatio-temporal variation in carbon dioxide efflux from surface litter, roots, and soil organic matter at four rain forest sites in the eastern Amazon, J. Geophys. Res., I! 2, G04001, doi: 10.1029/2007JG000443. Metcalfe, D. B., P. Meir, and M. Williams (2007b), A comparison of methods for cOlwerting rhizotron root length measurements into estimates of l'OOt mass production per unit ground area, Plant Soil, 301, 279-288, doi:10.1007/s11104-007-9447-6. Nepstad, D. c., et al.(2002), The effects of partial throughfall exclusion on canopy processes, aboveground production, and biogeochemistry of an Amazon forest, J. Geophys. Res., 1U7(D20), 8085, doi: 10.1 0291200 IJD000360. Palace, M., M. Keller, and H. Silva (2008), Necromass production: Studies in undisturbed and logged Amazon forests, Eco!. Appl., 18, 873-884. Phillips, O. L., N. Higuchi, S. Vieira, T. R. Baker, K.-l Chao, and S. L. Lewis (2009), Changes in Amazonian forest biomass, dynamics, and composition, 1980-2002, Geophys. Monogr. Sel'., doi: 10.1029/2008GM000739, this volume. Pyle, E. H., et al. (2008), Dynamics of carbon, biomass, and structure in two Amazonian forests, J. Geophys. Res., 113, GOOB08, doi: 10.1029/2007JG000592. Quesada, C. A, l Lloyd, L. O. Anderson, N. M. Fyllas, M. Schwarz, and C. 1. Czimczik (2009), The soils of Amazonia with special reference to the RAINFOR sites, Biogeosci. Discuss., 6, 3851-3921. Reichstein, M., et al. (2005), On the separation of net ecosystem exchange into assimilation and ecosystem respiration: Review and improved algorithm, Global Change BioI., 11, 1424-1439. Richey, l R., l 1. Hedges, A. H. Devol, P. D. Quay, R. Victoria, L. A Martinelli, and B. R. Forsberg (1990), Biogeochemistry of carbon in the Amazon river, Lillll/ol. Oceanogl'., 35, 352-371.
372
PRODUCTION, STORAGE, AND FLOW OF CARBON IN AMAZONIAN FORESTS
Richey, J. E., J. M., Melack, A. K. Aufdenkampe, V. M. Ballester, and L. Hess (2002), Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric CO 2, Nature, 416, 617-620. "" Richey, J. E., A. V. Krusche, M. S. Johnson, H. B. da Cunha, and M. V. Ballester (2009), The role of rivers in the regional carbon balance, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000734, this voh,nne. Saatchi, S. S., R. A. Houghton, R. C. Dos Santos Alvala, J. V. Soares, and Y. Yu (2007), Distri~ution of aboveground live biomass in the Amazon basin, Global Change BioI., 13(4), 816-837, doi: 1O.llllIj.l365-2486.2007.01323.x. Saleska, S., H. da Rocha, B. Kruijt, and A. Nobre (2009), Ecosystem carbon fluxes and Amazon forest metabolism, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000728, this volume. Silver, W. L., J. Neff, M. McGroddy, E. Veldkamp, M. Keller, and R. Cosme (2000), Effects of soil texture on belowground carbon and nutrient storage in a lowland Amazonian forest ecosystem, Ecosystems, 3, 193-209. Silver, W. L., A. W. Thompson, M. E. McGroddy, R. K. Varner, J. D. Dias, H. Silva, P. M. Crill, and M. Keller (2005), Fine root dynamics and trace gas fluxes in two lowland tropical forest soils, Global Change BioI., 11, 290-306.
Telles, E. D. C., P. B. de Camargo, L. A. Martinelli, S. E. Trumbore, E. S. da Costa, J. Santos, N. Higuchi, and R. C. Oliveira Jr. (2003), Influence of soil texture on carbon dynamics and storage potential in tropical forest soils of Amazonia, Global Biogeochem. Cycles, 17(2),1040, doi:IO.102912002GBOOI953. Trumbore, S., and P. B. de Camargo (2009), Soil carbon dynamics, Geophys. Monogr. Ser., doi:IO.1029/2008GM000741, this volume. Waterloo, M. J., et al. (2006), Export of organic carbon in lUn-off from an Amazonian rainforest blackwater catchment, Hydrol. Processes, 20, 2581-2597. Yoda, K. (1983), Community respiration in a lowland rain forest in Pasoh, peninsular Malaysia, Jpn. J. Ecol., 33,183-197.
L. E. O. C. Aragao, C. Girardin, and Y. Malhi, Environmental Change Institute, School of Geography and the Environment, University of Oxford, South Parks Road, Oxford OXI 3QY, UK. ([email protected]) S. Saatchi, Jet Propulsion LaboratOly, California Institute of Technology, Pasadena, CA 91109,USA.
Changes in Amazonian Forest Biomass, Dynamics, and Composition, 1980-2002 Oliver L. Phillips, I Niro Higuchi,2 Simone Vieira,3 Timothy R. Baker, I Kuo-Jung Chao, I and Simon L. Lewis I Long-tem1, on-the-ground monitoring offorest plots distributed across Amazonia provides a powerful means to quantifY stocks and fluxes ofbiomass and biodiversity. Here we examine the evidence for concerted changes in the structure, dynamics, and functional composition of old-growth Amazonian forests over recent decades. Mature forests have, as a whole, gained biomass and undergone accelerated growth and dynamics, but questions remain as to the long-term persistence of these changes. Because forest growth on average exceeds m0l1ality, intact Amazonian forests have been functioning as a carbon sink. We estimate a net biomass increase in trees 2:10 cm diameter of 0:62 ± 0.23 t C ha- I a-I through the late twentieth centmy. If representative of the wider forest landscape, this translates into a sink in South American old-growth forest of at least 0.49 ± 0.l8 Pg C a-I. If other biomass and necromass components also increased proportionally, the estimated South American old-growth forest sink is 0.79 ± 0.29 Pg C a: l , before allowing for possible gains in soil carbon. If tropical forests elsewhere ai'e behaving similarly, the old-growth biomass forest sink would be 1.60 ± 0.58 Pg C a-I. This bottom-up estimate of the carbon balance of tropical forests is preliminaly, pending syntheses of detailed biometric studies across the other tropical continents. There is also some evidence for recent changes in the functional composition (biodiversity) of Amazonian forest, but the evidence is less comprehensive than that for changes in structure and dynamics. The most likely driver(s) of changes are recent increases in the supply of resources such as atmospheric carbon dioxide, which would increase net primaly productivity, increasing tree growth and recruitment, and, in tum, mortality. In the future the growth response of remaining undisturbed Amazonian forests is likely to saturate, and there is a risk of these ecosystems transitioning from sink to source driven by higher respiration (temperature), higher mOliality (drought), or compositional change (functional shifts toward lighterwooded plants). Even a modest switch from carbon sink to source for Amazonian forests would impact global climate, biodiversity, and human welfare, while the documented acceleration of tree growth and mortality may already be affecting the interactions of thousands of plant and millions of animal species.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000739
IEarth and Biosphere Institute, School of Geography, University of Leeds, Leeds, UK. 2Laborat6rio de Silvicultura Tropical, Instituto Nacional de Pesquisas da Amazonia, Manaus, Brazil. 3Laborat6rio de Ecologia Isot6pica, USP, CENA, Piracicaba, Brazil. 373
374
PHILLIPS ET AL.
CHANGES IN AMAZONIA
1. INTRODUCTION Given the scale of the anthropogenic experiment with the atmosphere~iosphere system, it is self-evident that all eco-
systems on Earth are now affected by human activities. Processes such as deforestation are physically obvious, but others, such as hunting and surface fires, although subtler, still affect biodiver'sity in insidious ways [cf. Lewis et al., 2004a; Malhi and Phillips, 2004]. Anthropogenic atmospheric change will become more significant during this centmy, as carbon dioxide concentrations reach levels unprecedented for at least 20 million years [e.g., Retallack, 2001] and climates move beyond Quaternary envelopes [Meehl et al., 2007]. Moreover, the rate of change in these basic ecological drivers is without precedent in the evolutionary span of most species on Earth today. This is the Anthropocene [Crutzen, 2002]: we live in epoch-making times. Changes in tropical forests matter for three reasons. First, tropical forests play an important role in the global carbon cycle and, hence, affect the rate of climate change, as ~40% of terrestrial vegetation carbon stocks lie within tropical forests [Malhi and Grace, 2000]. Second, as tropical forests are home to at least half of all Earth's species, changes here have large impacts on global biodiversity and the cultures, societies, and economies that are bound to that biodiversity [Gl'Oombridge and Jenkins, 2003]. Finally, as different plant species vary in their ability to store and process carbon, climate and biodiversity changes are linked by feedback mechanisms [e.g., Cox et al., 2000; Lewis, 2006]. 2. A NETWORKED APPROACH Biodiversity change as a consequence of recent climate change is now widely documented in better-studied temperate areas [e.g., Parmesan and Yohe, 2003]. However, documentation in the tropics is much sparser and often focused on a few well-known locations; while this brings benefits, it is also risky. Inevitably, site-centric science is skewed, since peculiar features of that site, such as anthropogenic isolation, unusual soil conditions, cyclones, or fires, can all color interpretations. In most fields, such as climate change, it would be obvious folly to infer the presence or absence of global effects from records at one or two sites, but in ecological science, attempts are often made to scale from one or two local case studies to the regional and global. To avoid the pitfalls of the single-site approach, since 2000, we and others have tried to develop a standardized, international, long-term network of permanent plots in mature forests across Amazonia, by drawing together the existing efforts of local botanists and foresters, often working hitherto largely in isolation, and extending the site network when
375
possible to fill spatial and environmental gaps. This network ofAmazonian-forest researchers, known as "Red Amaz6nica de Inventarios Forestales" (RAINFOR or Amazon ForestInventOly Network, http://www.geog.leeds.ac.uk/projects/ rainfor/), now represents the long-term ecological monitoring effOlis of 35 institutions worldwide including from all Amazonian countries except Suriname. Here we synthesize recent results from the network to assess how Amazonian forests are changing on average. Where appropriate, we also discuss results from additional, individual sites where these may shed further light on the processes involved. 3. METHODOLOGY For these analyses, we define a monitoring plot as an area of old-growth forest where all trees 2:1 0 cm diameter at breast height (dbh, measured at 1.3 m height or above any buttress or other defonuity) are tracked individually over time. All trees are marked with a unique number, measured, mapped, and identified. Periodically (generally evelY 5 years), the plot is revisited, and all surviving trees are re-measured, dead trees are noted, and tree's recmited to 10 cm dbh are uniquely numbered, measured, mapped, and identified. This allows calculation of (l) the cross-sectional area that tree trunks occupy (basal area), which can be used with allometric equations to estimate tree biomass [Higuchi et al., 1998; Baker et al., 2004a; Chave et al., 2005]; (2) tree growth (the sum of all basal-area increments for surviving and newly recmited stems over a census interval); (3) the total number of stems present; (4) stem recruitment (number of stems added to a plot over time); and (5) mortality (either the number or basal area of stems lost from a plot over time). We present results from 50 to 91 plots, depending upon selection criteria for different analyses (most critically, the number of census intervals from a plot and whether only stem-count data or the full tree-by-tree data set is available). More plots are used to assess stem-density change than biomass change because full tree-by-tree data are required to calculate biomass (using the methods of Baker et al. [2004a]), whereas stem-change data can often be obtained from published studies. The plots span the Amazonian forests of northern South America (Figure I), including Bolivia, Brazil, Ecuador, French Guiana, Pem, and Venezuela, from the driest to the wettest and the least to the most fertile Amazonian forests. Most are 1 ha in size and comprise ~600 trees of 2:10 cm dbh, but the smallest is 0.25 ha and the largest 9 ha. Many plots have been monitored for more than a decade, although they range in age from 2 to 25 years. The earliest plot inventory was in 1971, the latest in 2007. Here we analyze in full results of censuses completed up to 2002. Details of the exact plot locations, inventOly and monitoring methods, and
Figure 1. Plot locations used in this study. Symbols represent approximate locations of each plot; gray circle for plots monitored for 5-10 years, black for those with ~10 years of monitoring. The approximate extent of seasonal and highly seasonal areas within South America north of the tropic of Capricom and excluding local rain shadow climates are indicated.
issues relating to collating and analyzing plot data are omitted from this chapter for reasons of space but are discussed in detail elsewhere [Phillips et al., 2002a, 2002b; Baker et al., 2004a, 2004b; Malhi et al., 2002, 2004; Lewis et al., 2004b; Phillips et al., 2004]. Scaling from individual tree to biomass is based on the diameter-based allometric equations detailed by Baker et al. [2004a, 2004b]. In brief, we used an equation developed for the Manaus area [Chambers et al., 2001a], modified by taking account of the taxon-specific wood density of each tree relative to the mean wood density of trees in the Manaus region. Alternatively, biomass can be estimated by universal, tropical forest equations such as those of Chave et al. [2005]. The Manaus equation is based
on a smaller sample size but has the advantage ofbeing local. For simplicity, we do not show results using universal equations here, but note that while different methods celiainly result in systematic differences in "biomass" estimates [e.g., Chave et al., 2003; Peacock et al., 2007], the rates of biomass "change" calculated across Amazonia appear largely insensitive to the equation used [Baker et al., 2004a]. We summarize findings fi'om old-growth forests in terms of (a) stmctural change, (b) dynamic-process change, and (c) functional change, over the two to three decades up to ~2002. Results assembled after this manuscript was prepared [Phillips et al., 2004] update some ofthe pattern documented here for the early twenty-first centulY.
378
PHILLIPS ET AL.
CHANGES IN AMAZONIA
Table 1. (continued)
Land Cover Forest Area Class (ha 106) Global GLC2000
FRACS
FRARS
humid tropical forest dly tropical forest flooded tropical forests
Aboveground Coarse Aboveground Biomass Coarse Increase Trees Biomass (Trees ~10 cm :0;10 cm and Necromass dbh) Increase Lianas ~1 cm Increase
Mean
Mean
834.2 302.8
276.6
1110.8 403.2
55.8
538.9 195.6
178.7
717.6
260.5
3.2
4.1
39.6
14.4
13.1
52.7
19.1
1881.1 1652.5 422.6 2075.1 1595.1
682.9 599.9 153.4 753.3 579.1
Mean
CI
Mean
Mean
1093.8
680.5
247.0
67.4
86.4
706.6
439.6
159.6
43.5
51.9
32.3
11.7
\
Total Aboveground Biomass and Belowground Total Biomass and Necromass Necromass Biomass Increase Increase Increase Mean
CI
total
1852.3
1152.3
418.3
114.1
146.3
1412.7 512.8
468.4
closed forest open forest
1627.2 416.1
1012.3 367.5 258.9 94.0
100.2 25.6
128.6 32.9
1241.1 450.5 317.4 115.2
411.5 105.2
total foresttotal
2043.3 1570.7
1271.1 977.1
125.8 96.7
161.4 124.1
1558.4 565.7 1198.0 434.9
516.7 397.2
461.4 354.7
CI
aWe take the net gain in aboveground coarse biomass (trees 2:10 em dbh) recorded in Amazonia (0.62 ± 0.22 t C ha- t a-I), and scale by the estimated ratio oftrees <10 em dbh and lianas 2:1 em dbh to trees 2:10 em dbh in Amazonia (=0.099, [Phillips et al., 1998]), by the most comprehensive estimate of~oarse necromass: aboveground coarse biomass ratio available for Amazonia (=0.127) [Chao et al., 2009], and by the latest estimate of belowground: aboveground biomass ratio (=0.370) (N. Higuchi et al., unpublished central Amazonian estimate, 2008). Values for each region are estimated by assuming the same allometty and behavior as Amazonian forests. Forest area estimates are taken from Mayaux et al. [2005]. Abbreviations are GLC, global land cover; FRA CS, Food and Agriculture Organization (FAO) [2000] countty statistics; FRARS, FAO [2000] remotely sensed values. Scaled-up estimates based on FRARS highlighted in bold are mentioned in the text. Units for biomass stock increases are 106 t C a-I. Totals for each continent are given in italics.
The finding of increased biomass has proved controversial [cf., for example, Clark, 2002; Phillips et al., 2002a, 2002b; Wright, 2005; Lewis et al., 2006a]. While there is no space here to review the debate fully, one important aspect concerns the role of recent disturbance and the role of coarse woody debris (CWD) in total aboveground carbon balance. Results from a single LBA site in eastern Amazonia (Tapajos) show that over intervals of a few years, atmospheric carbon :fluxes from CWD may exceed biomass gains [Rice et al., 2004]. Clearly, recent disturbances can drive patterns of local biomass change, which is a key reason why long-term monitoring over decades is so valuable. However, the LBATapajos site has uniquely abnormally high values of CWD [Palace et al., 2007; Chao et al., 2009] and at least twice as much as at most sites in western Amazonia where CWD :flux rates (mortality and decomposition) are also faster [Baker et al., 2007]. Even ifit were possible to accurately track CWD inventories through time at our sites, their impact on our Amazonia-wide long-term aboveground carbon balance estimate would be small, unless there have been recent, large,
region-wide secular changes in the rate of CWD production or decomposition. Have such changes occurred? We find that mortality has indeed increased on average over recent years (see next section); this would imply an additional carbon sink, not a source, in necromass (see Table 1), but because CWD averages only ~12% of biomass and has short residence times [Baker et al., 2007], the additional carbon sink it represents must be small. Still, it could be argued that our turn-of-thecenhlly Amazonian mature forest plots could be recovering from unobserved, earlier megadisturbances. By definition, such a suggestion is impossible to falsify completely, but it is inconsistent with the evidence of other, simultaneous structural and dynamic changes including increasing growth rates (see below) and the fact that episodic climate anomalies have occurred during the monitoring period itself, including a drought associated with the strong 1997-1998 El Nino. (The 2005 drought, particularly severe in southwestern Amazonia, struck after the monitoring period analyzed here). Long-term forest rebound from much earlier human distur-
bances also merits consideration [e.g., Phillips et al., 1998], but recent analyses suggest that such dishlrbances were only localized [Bush et all2007]. In any case, given known rates of secondary succfssion [e.g., Hughes et al., 1999] and atmospheric CO2 7"'0lution [Nevle and Bird, 2008] recovering forest equilibrated in biomass telms within two centuries of the Spanish conquest (see Phillips et al. [2002a, 2002b] for more discussion).It is important to note that biomass increase is not the only structural change recorded in Amazonia. Across the 91 RAINFOR plots where we have tracked populations, there has also been a small increase in stem density between the first and last measurements, of 0.84 ± 0.77 stems ha- t a-I (Figure 2b; paired t test, t = 2.12, P = 0.037), an annual increase of 0.15 ± 0.13% [Phillips et al., 2004]. Across all plots, stem-change rates are approximately nOlmally distributed and slightly shifted to the right of zero (Figure 2b). The same test using the smaller set of 59 plots where we have tracked biomass shows a similar increase in stem density (0.16 ± 0.15% per year), while a longer-term subset of plots (50 plots fi'om Lewis et al., 2004b) shows a slightly larger increase (0.18 ± 0.12% per year). These increases in stem density, while propOliionally smaller than the biomass changes, run counter to expectations if the plots were in an advanced state of secondmy succession [e.g., Coomes and Allen, 2007]. They falsify the hypothesis that the generalized biomass increase across Amazonian plots can be explained as a result of disturbance recovelY. For practical reasons, the pan-Amazon sample is nonrandomly distributed. It is possible to test whether this spatial bias might be driving the result by assessing whether we have oversampled unusually heavily regions that happened to be gaining biomass and undersampled those that happened to lose biomass. At smaller scales, this appears unlikely, since the long-term mean net gain is almost identical whether the sampling unit is taken to be the "plot" (as here) or a larger unit such as a "landscape cluster of plots" [Phillips et al., 2009]. At larger scales, the climate- and soil-environmental space is well-covered, but the network still leaves large expanses of Brazilian Amazonia unsampled (Figure 1). Concerted monitoring efforts in these regions are clearly needed to reduce this source of uncertainty.
379
convention [Phillips and GentT)), 1994], we estimate stem turnover between any two censuses as the mean of annual mortality and recruitment rates for the population of trees ::::10 cm diameter. Second, we examine changes in biomass :fluxes of the forest, in terms of growth of trees and the biomass lost with mortality events. These stand-level rates of biomass growth and biomass loss should be approximately proportional to the rate at which surviving and recruiting trees gain basal area and the rate at which basal area is lost fi'om the stand through tree death [Phillips et al., 1994]. Among 50 old-growth plots across tropical South America with at least three censuses (and therefore at least two consecutive monitoringperiods that can be compared), we find that all of these key ecosystem processes, stem recruitment, mortality, and turnover, and biomass growth, loss, and turnover, are increasing significantly (Figure 3), between the first and second monitoring periods [Lewis et al., 2004b]. Thus, over the past two decades, these forests have become, on average, faster-growing and more dynamic. Notably, the increases in the rate of the dynamic :fluxes (growth, recruitment, and mortality) are about an order of magnitude larger than are the increases in the structural pools (aboveground biomass and stem density) [Lewis et al., 2004b]. These and similar results can be demonstrated graphically in a number of w~ys. In Figure 4, we plot the across-site mean values for stem recruitment and mortality as a function of calendar year. The increase is not the shOli-term result of a year with un~sual weather: recruitment rates have on average consistently exceeded mortality rates, and mortality
2.5 2.0 ?f!.
2' ~ ro :J C C
«
1.5 1.0 0.5
4.2. Dynamic Changes An alternative way of examining forest change is to look for changes in the processes (growth, recruitment, death), as' well as the structure (biomass, stem density): are these forests simply gaining mass, or are they becoming more dynamic too? We measured the dynamics offorests in two ways. First, we can examine changes in stem population dynamics. By
Figure 3. Annualized rates of stand-level basal-area growth, basalarea mortality, stem recruitment, and stem mortality from plots with two consecutive census intervals, each giving the mean from 50 plots with 95% confidence intervals. Paired t tests show that all of the increases are significant. The average mid-year of the first and second censuses was 1989 and 1996, respectively [from Lewis et al., 2004b].
380
PHILLIPS ET AL.
CHANGES IN AMAZONIA
nant macroecological gradient across Amazonia [Phillips et aZ., 2004; Vieira et aZ., 2004; tel' Steege et aZ., 2006]. Both ~ ~ 2.5 groups showed increased stem recruitment, stem mortality, Ul.§ stand basal-area growth, and stand basal-area mOliality, with c'l 2 Q) (j greater "absolute" increases in rates in the faster-growing ro ~ .............. and more dynamic sites than in the slower-growing and less ro::l >, 0 1.5 C:!: dynamic sites (Figure 5) [Lewis et aZ., 2004b], but proporern tional increases in rates that were similar and statistically :<15 1-h~~~~~--rT--rT--rT--rT-rr.---r-.-, indistinguishable among forest types [Lewis et aZ., 2004b]. E 1976 1981 1986 1991 1996 2001 Increasing growth, recruitment, and mortality has occurred \ Year across different forest types and geographically widespread Figure 4. Mean and 95% confidence intervals for stem recruitment areas. The simultaneous recent increases in plot dynamic rates, and mortality rates against calendar year, for plots arrayed across Amazonia. Rates for each plot were corrected for the effects of biomass, and stem density raise the following question: for differing census-interval lengths, for "site-switching" (changes how long has this been going on? Monitoring of Amazonian through time in the plots being measured), and for "majestic-for- plots only began in a concerted fashion around 1980. To go est bias" (potential avoiding of gaps when establishing plots). A much further back in time requires annual dating of growth detailed justification methodology for these corrections is given by rates of a large sample of individual trees from different spePhillips et al. [2004]; all trends hold if these corrections are not cies, something that to our knowledge has been only been applied. Black indicates recruitment, gray indicates mortality, solid done in Amazonia from two locations in tena firme [Vieira lines are means, and dots are 95% confidence intervals [from Philet aZ., 2005], using radiocarbon dating. Although the malips et al., 2004]. jority of trees tested did grow faster since 1960 than before 1960, the null hypothesis of no change in growth rate could appears to lag recruitment [Phillips et aZ., 2004; Lewis et aZ., not be rejected. This technique is complicated by potential ontogenetic variation in growth rates partly related to 2004b]. For the 50 plots which have two consecutive census in- changing light environments [e.g., Worbes, 1999] and could tervals, we can separate them into two groups, one fast- overestimate "stand-level" growth rates in the past because growing and more dynamic (mostly in western Amazonia), individual trees with slow- and declining growth are more and one slow-growing and much less dynamic (mostly in susceptible to mortality [Chao et aZ., 2008] and therefore eastern and central Amazonia), which reflects the domi- less likely to survive to the point at which they are dated.
4.3. FunctionaZ Composition Changes
....
L.
L.
3.0
3.0
2.5
2.5 ~
';f2.
T§ rn::l 1.5
T§ rn::l 1.5
:- 2.0
~ 1.0
oJ 2.0
c 1.0-
«c
0.5
0.5
0.0
0.0
(>. g(o\N\\\ 11\0(\al\\'1 (\.I\\l1\en\ 11\0(\al\\'1 c S\ano \3 s\ano \3(>. S\el1\ (e S\el1\
Changes in the structure and dynamics of tropical forests are likely to be accfmpanied by changes in species composition and forest ft\l!ction. Phillips et aZ. [2002a, 2002b] studied woody climbers (structural parasites on trees, also called lianas), which typically contribute 10-30% of forest leaf productivity, but are ignored in almost all monitoring studies except in most of our western Amazonian sites. Across the RAINFOR plots of western Amazonia, there has been a concerted increase in the density, basal area, and mean size of lianas (Figure 6) [Phillips et aZ., 2002b]. Over the last two decades of the twentieth centulY, the density of large lianas relative to trees increased here by 1.7-4.6% per year. This was the first direct evidence that intact tTopical forests are changing in terms of their functional composition. A long-term monitoring study from beyond Amazonia (Barro Colorado Island in Panama) has since reported a substantial increase in absolute and relative liana leaf-fall rates since the 1980s, indicating that lianas are both increasing and becoming more dominant there [Wright et aZ., 2004]. There is some experimental evidence [Granados and Korner, 2002] for tropicallianas to respond more strongly than trees to elevated atmospheric CO 2 concentrations. Finally, a large cluster of plots in central Amazonia shows consistent changes in tree species composition over the past two decades [Laurance et aZ., 2004]. Many faster-growing genera of canopy and emergent stature trees increased in basal area or density, whereas some slower-growing genera of subcanopy or understOly trees declined. Laurance et aZ.
Figure 5. Annualized rates of stand-level basal-area growth, basal-area mortality, stem recruitment, and stem mortality over consecutive census intervals forplots grouped into "slower growing less-dynamic" (left) and "faster growing moredynamic" (right) forests. Of the slower-dynamics group, 20 of 24 plots are from eastern and central Amazonia, whereas just two are from western Amazonia. Of the faster-dynamics group, 24 of 26 plots are from western Amazonia, with just one from central Amazonia. The remaining three plots are from Venezuela and outside the Amazon drainage basin. Changes have occurred across the South American continent, and in both slower- and faster-dynamic forests [from Lewis et al., 2004b].
[2004] provide evidence ofpervasive changes in central Amazonian forests: growth, mortality, recruitment all increased significantly over two decades (basal area also increased, but not significantly so), with faster-growing genera showing larger absolute and relative increases in growth, relative to slower-growing genera. Further studies are urgently needed to determine whether comparable shifts in tree communities are occurring throughout Amazonia. 5. WHAT IS DRIVING THESE CHANGES?
What could be causing the continent-wide changes in tree growth, recruitment, mOliality, stem density, and biomass? Many factors could be invoked, but there is only one parsimonious explanation. The results appear to show a coherent fingerprint of increasing growth [i.e., increasing net primary productivity (NPP)] across tropical South America, probably caused by a long-term increase in resource availability [Lewis et aZ., 2004a, 2004b]. According to this explanation, increasing resource availability increases NPP, which then increases stem growth rates. This accounts for the increase in stand basal-area growth and stem recruitment rates, and the fact that these show the clearest, most highly significant changes [Lewis et aZ., 2004b]. Because of increased growth, competition for limjting resources, such as light, water, and nutrients, increases. Over time, some of the faster-growing, larger trees die, as 40 some of the "extra" recruits (the accelerated growth perc61ates through the system). This accounts for the increased losses from the system: biomass-mortality and stem-mortality rates increase. Thus, the system gains biomass and stems, while the losses lag some years behind, causing an increase in aboveground biomass and stems. Overall, this suite of changes may be qualitatively explained by a long-term increase in a limiting resource. 18 The changes in composition can also be explained by inC\l creasing resource availability, as the rise in liana density may J: ...Q) 14 be either a direct response to rising resource supply rates ., .....:······· ..·.. ·11· c. or a response to greater disturbance caused by higher tree1/1 ..... .... . ... E 10 mOliality rates. The changing tree composition in central Q) o Amazonian plots [Laurance et aZ., 2004] is also consistent 1/1 AI. o C\l o with increasing resource supply rates, as experiments show I: 6 .. 0" C\l that faster-growing species are often the most responsive, in ::::i absolute terms, to increases in resource levels [Coomes and 2 Grubb, 2000], although others have argued [e.g., Korner, 1986 1981 1991 1996 2001 2004; J. Lloyd, personal communication, 2008] that the Figure 6. Five-year lUnning means (solid line) with 95% confi- greatest proportional response should be in understory seeddence intervals (dashed lines) of liana stem density per hectare - lings and saplings, which are likely to be close to carbon (:::10 cm diameter at breast height), with values plotted separately deficit due to shading; a small increase in photosynthetic for northern PelU (filled squares), southern PelU (filled triangles), rate here could therefore have a great proportional impact Bolivia (filled circle), and Ecuador (unfilled squares) (adapted from on carbon balance. There is some experimental evidence to [Phillips et al., 2002b]; see that paper for full details offield and support this view [e.g., Kerstiens, 2001; AidaI' et aZ., 2002]. analytical methodology).
-
,
(>. g(o'f'i\\\ o(\al\\'1 \.I\\l1\en\ 11\0(\al\\'1 c s\ano \3 s\ano \3(>.11\S\el1\ (e ( s\el1\
381
382
PHILLIPS ET AL.
CHANGES IN AMAZONIA
What environmental changes could be increasing the growth and productivity of tropical forests? While there have been widespread changes in the physical, chemical, and biolo~ical environment of tropical trees [Lewis et al., 2004a], only increasing atmospheric CO 2 concentrations [Prentice et al., 2001], increasing solar radiation inputs [Wielicki et al., 2002], rising air temperatures, and changing precipitation pattems [Malhi and Wright, 2004] have been documented across most or all of Amazonia and could be responsible for increased growth and productivity. For none of these changes, however, do we have overwhelming evidence that the driver has both certainly changed and that such a change is likely to accelerate forest growth [Lewis et al., 2004a]. The increase in atmospheric CO 2 is the primary candidate because of the undisputed long-term increase in CO 2 concentrations, the key role of CO 2 in photosynthesis, and the demonstrated positive effects of CO 2 fertilization on plant growth rates, including experiments on whole temperate-forest stands [Norby et al., 2002; Hamilton et al., 2002; Lewis et al., 2004a]. However, a substantial role for increased insolation [e.g., Nemani et al., 2003; Ichii et al., 2005], or aerosol-induced increased diffuse fraction of radiation [e.g., Oliveira et al., 2007], cannot be mled out. Elsewhere, we have discussed the candidate drivers in more detail [Lewis et a!., 2004a, 2006a, 2006b; Malhi and Phillips, 2004, 2005]. Here we do not revisit that discussion, but briefly discuss our philosophy and approach to forest ecology and inference, which has antecedents in peripatetic ecologists of the past from Darwin to Gentry. Ecological science is largely done at individual sites, some of which have become extraordinarily well-known. This is dangerous: there is a natural human tendency to generalize from rather limited personal experience, sometimes exacerbated by the pressure to publish rapidly and to exaggerate the global importance of local research findings. But Amazonia is a very big place indeed. The site-centric approach inevitably colors interpretations and means that the peculiar features of that site including fragmentation, atypical soil conditions, previous cyclones or fires, dominate researchers' findings and interpretations. Our results and those of a recent, parallel study [Chave et al., 2008] show that all sites, if studied intensively enough, will likely reveal strong, local idiosyncratic features which dominate their contemporary ecology, but that no individual site (or even handful of sites) can be satisfactorily used to test for the presence or absence of larger-scale processes. The synoptic challenge is to reveal general pattems that lie beyond the local idiosyncrasies. That will only come from a standardized, geographically distributed, tmly long-term, and intemationalized science. RAINFOR represents a positive step in that direction.
6. THE FUTURE: POTENTIAL SUSCEPTIBILITY OF AMAZON FOREST TO ENVIRONMENTAL STRESS AND COMPOSITIONAL CHANGES In sum, then, long-term observations indicate that Amazonia, the world's largest remaining h'act of tropical forest, has shown conceried changes in forest dynamics over the past two decades. Such unexpected and rapid alterations, regardless of the cause, were not anticipated by ecologists and raise concems about other possible surprises that might arise as global changes accelerate in the coming decades. On current evidence, tropical forests are sensitive to changes in incoming resource levels and may show large structural and dynamic changes in the future, as resource levels alter furiher, temperatures continue to rise, and precipitation pattems shift. The implication of such rapid changes for the world's most biodiverse region is unknown, but could be substantial. Old-growth Amazonian forests have evidently helped to slow the rate at which C02 has accumulated in the atmosphere, thereby acting as buffer to global climate change. The concentration of atmospheric CO 2 is rising at an annual rate equivalent to 3-4 Pg C; this would be significantly greater without the tropical South American biomass carbon sink of 0.5-0.9 Pg C a-I. This subsidy from nature could be a relatively short-lived phenomenon. Mature Amazonian forests may either (l) continue to be a "carbon sink" for decades [Chambers et al., 2001b; Cramer et al., 2001], or (2) soon become "neutral or a small carbon source" [Cramer et al., 2001; Phillips et al., 2002b; Korner, 2004; Laurance et al., 2004], or (3) become a "mega-carbon source" [Cox et al., 2000; Cramer et al., 2001]. Given that a 0.4% annual increase in Amazonian forest biomass roughly compensates for the entire fossil-fuel emissions of westem Europe (or the deforestation in Amazonia), a switch of mature tropical forests from a moderate carbon sink to even a moderate carbon source would have implications for global climate and human welfare. The ~0.4% annual sink represents the difference between two much larger values: stand-level growth (averaging ~2%) and moriality (averaging ~1.6%), so a s~all decrease in growth or small increase in moriality would be enough to shut the sink down. There are several mechanisms by which such a switch could occur, apart from the obvious and itmnediate threats posed by land use change and associated disturbances by fragmentation and fire.
rise in forest productivity over time, it is predicted that forests would remain a.carbon sink for decades [e.g., Lloyd and Farquhar, 1996]. However, the current increases in productivity, app~t'ently caused by continuously improving conditions for tJie growth, cannot continue indefinitely: if CO 2 is the cause, trees are likely to become C02 saturated (Le., limited by another resource) at some point in the future. More generally, whatever the driver for recently accelerated growth, forest productivity will not increase indefinitely, as other factors such as soil nutrients will limit productivity. Rising temperatures could also shrink the current forest sink or cause forests to become a carbon source in the future. Warmer temperatures increase the rates of virtually all chemical and biological processes in plants and soils (including the enhancement of any C02 fertilization effect), until temperatures reach inflection points where enzymes and membranes lose functionality. There is some evidence that the temperatures of leaves at the top of the canopy, on warm days, may be reaching such inflection points around midday at some locations [Lewis et al., 2004a]. Canopy-toair vapor deficits and stomatal feedback effects may also be paramount in any response of tropical forest photosynthesis to future climate change [Lloyd et a!., 1996]. The relationship between temperature changes and respiration is critical. The first global circulation model (GCM) to include dynamic vegetation and a carbon cycle that is responsive to these dynamic changes suggests that under the "business as usual" scenario of emissions, IS92a, atmospheric CO 2 concentrations are 900-980 parts per million by volume (ppmv) in 2100, compared to ~700 ppmv from previous GCMs [Cox et al., 2000, 2004]. These concentrations depend critically on (1) dieback of the eastem Amazonian forests, caused by climate change-induced drought, and (2) the subsequent release of C from soils. The release of C from soils is critically dependent on the assumed response of respiration to temperature and soil moisture and the modeling of soil carbon. Carbon losses fi'om respiration will almost certainly increase as air temperatures continue to increase. The key question is what form this relationship takes. Carbon gains fi'om photosynthesis cannot rise indefinitely and will almost certainly asymptote. Thus, the sink in intact tropical forests will diminish and eventually reverse. The major uncertainty is "when" this will occur. 6.2. Moisture Stress
6.1. Photosynthesis/Respiration Changes
Intact forests will remain a sink as long as carbon uptake associated with photosynthesis exceeds the carbon efflux from respiration. Under the simplest scenario of a steady
Climate change will alter precipitation pattems. There are critical thresholds of water availability below which tropical forests cannot persist and are replaced by savanna systems; currently, the threshold lies around 1300-1500 mm rainfall
383
per annum [Salzmann and Hoelzmann, 2005], but this could increase with rising temperatures. Thus, increasing temperatures and/or changing precipitation pattems may cause shifts in vegetation from carbon-dense tropical forests, to carbonlight savanna systems. The degree to which Amazonian forests mayor may not be ecophysiologically resilient to extreme temperatures is a subject of active research, reviewed by Lloyd et al. [this volume]. What is the evidence, so far, of drought impacting Amazonian forests? The temporal resolution of RAINFOR plots has generally been insufficient to allocate growth and mortality rates to individual years. Nevertheless, among the 10 longest-running plots (initiated in the 1970s or earlier), the severe 1982-1983 El Nino event apparently did not greatly affect forest dynamics [Phillips, 1995]. Where there are annual or higher-resolution records, there is some evidence of short-tenn stand-level rates responding to moisture stress, with growth decreasing markedly in the dry season near Rio Branco, Acre [Vieira et al., 2004] and mortality temporarily increasing during the 1997-1998 El Nino near Manaus [Williamson et a!., 2000]. However, the impact on growth rates of moderate dry conditions in Amazonia may not always be negative. There is some evidence from leaf and branch level [e.g., Graham et al., 2003] and at regional scales [Huete et al., 2006] to suggest that neotropical moist forests may be as light limited as they are moisture-limited. If so, while droughts reduce productivity and exacerbate fire risk in more marginal forest locations, more cloud-free rainless days would enhance productivity in some cloudier locations. In a separate study [Phillips et al., 2009], we repori the results of intensive recensusing following the 2005 drought to assess just how sensitive Amazonian forests are to drought across the whole basin. 6.3. Compositional Change
Biodiversity change has inevitable consequences for climate change because different plant species vary in their ability to store and process carbon. Yet most models that project the future carbon balance in Amazonia (and future climate-change scenarios) make no allowance for changing forest composition. Representation of composition is challenging, both because of the computational complexities in integrating ecological processes into ecophysiology-driven models and because the ecological data themselves are sparse. But representing composition better, and its potential for change, is important. Lianas, for example, ignored in all forest models, often contribute little to forest biomass but heavily to productivity [Schnitzer and Bongers, 2002], while killing h'ees [Phillips et al., 2005] and preferentially infesting denser-wooded species [van del' Heijden et al.,
384
(a)
CHANGES IN AMAZONIA
PHILLIPS ET AL.
Inilial stand-level mean wood density as a function of subsequent annual stem turnover rate; 127 lowland forest plots across Soulh America
0.80,--.:....lt....--:;:--'_--------------, '?
~ 0.70 t----'I'-~""
'"
~
.~ 0.60 t----~"_it__'__
'"
"0 "0
+
o~ 0.50 +-----~.--".:!"-~--._+------___l "0
y = O.7075x-o·1605
~
.<:
.2' 0.40
~
.,'" :u
ro
+---------------~=i~I4__-____l
0.30 - \ - - - - - - - - - - - - - - - - - - - - - - j
lJl
c;;'" 0.20 - \ - - - - - - - - - - - - - - - - - - - - - - j
.Q
>
.!!!
,;
:;; 0.10 f - - - - - - - - - - - - - - - - - - - - - - - - j
en
0.00 - ! - - - - - - r - - - - , - - - - , - - - - r - - - - , - - - - - - ! 4 5 6 3 o 2 Mean of mortality and recruitment, %a "\census-inlerval corrected
Initial biomass as a function of subsequent annual stem turnover rate; 127 lowland forest plots across South America
(b)
500,--.:--------------------
400
+--~-;;:-----------------____l
y
=349.43x .u·1955
~;., 300 -\--_-:--.r-='!..:L,:-;~~~~~-_-_K=2-"-'_U.''I_lI~i__~ .c
'"
::;;
~
+.
E
•
•
~ 200 +------------'.~---------~
•
100 f - - - - - - - - - - - - - - - - - - - - - - - - j
2008]; their recent increase suggests that the tropical carbon sink might shut down sooner than current models suggest. Large changes in tree communities could also lead to net losses of carbon from tropical forests [Phillips and Gently, 1994; Korner, 2004]. One way this could happen is a shift to faster-growing species, driven by increasing tree mortality rates and frequency of gap formation [Phillips and GentlY, 1994; Phillips et al., 2004]. Such fast-growing species generally have lower wood specific gravity, and hence less carbon [West et al., 1999], than shade-tolerant trees. Better effort to detect whether or not such changes are occurring is clearly a priority for future monitoring effOlis. The potential scope for such impacts of biodiversity changes on carbon storage is highlighted by Bunker et al. [2005], who explored various biodiversity scenarios based on the tree species at Barro Colorado Island: if slower-growing tree taxa are lost from an accelerated, liana-dominated forest, as much as one third of the carbon storage capacity of the forest could be lost. In Amazonia, a basin-wide annual decrease in mean wood specific gravity of 0.4% would cancel out the carbon sink effect. There is currently a ~20% difference in mean wood density of the faster foi'ests in the west, compared with slower forests in the east (Figure 7) [Baker et al., 2004b], and because these faster forests also have lower basal area, the differences in terms of biomass carbon stored are greater still (Figure 7) [Lewis et al., 2006b]. Concerted compositional changes driven by greater resource supply, increased mortality rates, and possible selection for faster-growing trees which escape lianas, may shut down the carbon sink function of tropical forests earlier than ecophysiological analyses predict. Acknowledgments. The results summarized here depended on contributions from numerous assistants and rural communities in Brazil, Bolivia, Ecuador, French Guiana, Peru, and Venezuela, and more than 50 grants from funding agencies in Europe and the United States of America aclmowledged in earlier publications. This paper was supported in particular by the Leverhulme Trust (0. L. Phillips), NERC grants NE1B503384/1 and NEIDOI025X11 (0. L. Phillips), a Royal Society University Research Fellowship (S. L. Lewis), University of Leeds (T. R. Baker and J.-K. Chao), and an Overseas Research Studentship (J.-K. Chao). We thank M. Alexiades, S. Almeida, L. Arroyo, S. Brown, J. Chave, I A. Comiskey, C. 1. Czimczik, A Di Fiore, T. Erwin, I Grace, T. Killeen, C. Kuebler, S. G. Laurance, W. F. Laurance, J. Lloyd, G. Lopez-Gonzalez, Y. Malhi, A Monteagudo, H. E. M. Nascimento, D. A Neill, P. Nunez Vargas, I Olivier, W. Palacios, S. Patino, I Peacock, N. C.
O-l---~--_,__--_,__--_,__--_,__-__I
o
2
3
4
5
Mean of mortality and recruitment, % a·"census·lnterval corrected
6
Figure 7. Biomass as a function of mean stand-level wood density, for 127 lowland forest plots across South America. Note that faster forests have lower wood density and much lower biomass.
A. Pitman, C. A Quesada, M. Saldias, J. N. M. Silva, .I. Terborgh, A Torres Lezama, R. Vasquez Martinez, and B. Vinceti for contributing data and/or dis'cussions.
REFERENCES Aidar, M. P. M., C. A. Martinez, A C. Costa, P. M. F. Costa, S. M. C. Dietrich, and M. S. Buckeridge (2002), Effect of atmospheric CO 2 enrichment on the establishment of seedlings of Jatoba, Hymenaea courabi/ L. (Leguminosae, Caesalpinioideae), Biota Neotropica, 2, 1-10, BN01602012002. Baker, T., C. E. Honorio, O. L. Phillips, G. van del' Heijden, I Martin, M. Garcia, and 1 Silva Espejo (2007), Low stocks of coarse woody debris in a south-western Amazon forest, Oecologia, 152(3), 495-504, doi:10.1007/s00442-007-0667-5. Baker, T. R., et al. (2004a), Increasing biomass in Amazonian forest plots, Phi/os. Trans. R. Soc. London, Ser. B, 359, 353-365. Baker, T. R, et al. (2004b), Variation in wood density determines spatial pattems in Amazonian forest biomass, Global Change BioI., 10,545-562. Bunker, D., F. De Clerck, I Bradford, R. Colwell, P. Garden, 1. Perfecto, O. L. Phillips, M. Sankaran, and S. Naeem (2005), Carbon sequestration and biodiversity loss in a tropical forest, Science, 310, 1029-1031. Bush, M. B., M. R. Silman, M. B. de Toledo, C. Listopad, W. D. Gosling, C. Williams, P. de Oliveira, and C. Krisel (2007), Holocene fire and occupation in Amazonia: Records ft'om two lake districts, Phil. Trans. R. Soc. Ser. B, 362, 209-218, doi: 10.1 098/rstb.2006.1980. Chambers, I Q., 1 Santos, R. 1 Ribeiro, and N. Higuchi (2001a), Tree damage, allometric relationships, and above-ground net primaty production in central Amazon forest, For. Ecol. Manage., 152, 73-84. Chambers, 1 Q., N. Higuchi, E. S. Tribuzy, and S. E. Tmmbore (2001b), Carbon sink for a century, Nature, 410, 429. Chao, K.-1, O. L. Phillips, E. Gloor, A Monteagudo, A TorresLezama, and R. Vasquez Martinez (2008), Growth and wood density predicts tree mOliality in Amazon forests, J. Ecol., 96, 281-292. Chao, K.-I, O. L. Phillips, T. R Baker, J. Peacock, G. LopezGonzalez, R. Vasquez Matiinez, A Monteagudo, and A. TonesLezama (2009), After trees die: Quantities and determinants of necromass across Amazonia, Biogeosci. Discuss., 6, 1979-2006. Chave, 1, et al. (2003), Error propagation and scaling for tropical forest biomass estimates, Phi/os. Trans. R. Soc. London, Ser. B, 359, 409-420. Chave, 1, et al. (2005), Tree allometry and improved estimation of carbon stocks and balance in tropical forests, Oecologia, 145, 87-99. Chave, I, et al. (2008), Assessing evidence for a pervasive altera-' tion in tropical tree communities, PLoS BioI., 6, e45. Clark, D. A. (2002), Are tropical forests an important carbon sink? Reanalysis of the long-term plot data, Ecol. Appl., 12, 3-7. Coomes, D. A, and R B. Allen (2007), Mortality and tree-size distributions in natural mixed-age forests, J. Ecol., 95, 27-40.
385
Coomes, D. A., and P. J. Grubb (2000), Impacts of root competition in forests and woodlands: A theoretical framework and review of experiments, Eeol. Monogr., 200,171-207. Cox, P. M., R A Betts, C. D. Jones, S. A Spall, and 1. I Totterdell (2000), Acceleration of global wanning due to carbon-cycle feedbacks in a coupled climate model, Nature, 408,184-187. Cox, P. M., et al. (2004), Amazonian forest dieback under climatecarbon cycle projections for the 21st century. Theor. Appl. Climatol., 78(1-3), 137-156. Cramer, W., et al. (2001), Global response of terrestrial ecosystem structure and function to CO 2 and climate change: Results from six dynamic global vegetation models, Global Change BioI., 7, 357-373. Crutzen, P. 1 (2002), Geology of mankind, Nature, 415, 23. Denman, K. L., et al. (2007), Couplings between changes in the climate system and biogeochemistry, in Climate Change 2007: The Physical Science Basis: Contribution of Working Group 1 to the Fourth Assessment Report ofthe Intergovernmental Panel on Climate Change, edited by S. Solomon et al." Cambridge Univ. Press, Cambridge, U. K. Food and Agriculture Organization (FAO) (2000), Global forest resources assessment 2000: Main report, FAO For. Pap. 140, Rome. Graham, E. A, S. S. Mulkey, K. Kitajima, N. I Phillips, and S. 1 Wright (2003), Cloud cover limits net CO 2 uptake and growth of a rainforest tr'ee during tropical rainy seasons, Proc. Natl. Acad. Sci. U. S. A., 100, ~72-576. Granados, I, and C. 'Korner (2002), In deep shade, elevated CO 2 increases the vigour of tropical climbing plants, Global Change Biol.,8,1109-11117. Groombridge, B., and M. D. Jenkins (2003), World Atlas ofBiodiversity, Univ. of Calif. Press, Berkeley. Hamilton, 1 G., E. H. DeLucia, K. George, S. L. Naidu, A C. Finzi, and W. H. Schlesinger (2002), Forest carbon balance under elevated CO 2, Oecologia, 131, 250-260. Higuchi, N., I dos Santos, 1 R Ribeiro, L. Minette, and Y. Biot (1998), Biomassa da parte aerea da floresta tropical umida de terra firme da Amazonia Brasileira, Acta Amazonica, 28, 153166. Huete,A R,K. Didan, Y. E. Shimabukuro, P. Ratana, S. R. Saleska, L. R Hutya, W. Yang, R. R. Nemani, and R. Myneni (2006), Amazon rainforests green-up with sunlight in the dry season, Geophys. Res. Lett., 33, L06405, doi:1O.1029/2005GL025583. Hughes, R F., I B. Kauffman, and V. 1 Jaramillo (1999), Biomass, carbon, and nutrient dynamics of secondary forests. in a humid tropical region of Mexico, Ecology, 80, 1892-1907. Ichii, K., H. Hashimoto, R Nemani, and M. White (2005), Modeling the interannual variability and trends in gross and net primaty productivity of tropical forests from 1982 to 1999, Global Planet. Change, 48, 274-286. Kerstiens, G. (2001), Meta-analysis of the interaction between shade-tolerance, light environment and growth response of 'Yoody species to elevated CO 2 , Acta Oecologica, 22, 61-69. Komer, C. (2004), Through enhanced tree dynamics carbon dioxide enrichment may cause tropical forests to lose carbon, Phi/os. Trans. R. Soc. London, Ser. B, 359, 493-498.
386
CHANGES IN AMAZONIA
governmental Panel on Climate Change, edited by S. Solomon Laurance, W. F., et a!. (2004), Pervasive alteration of tree comet a!., pp. 747-845, Cambridge Univ. Press, Cambridge, U. K. munities in undisturbed Amazonian forests, Nature, 428, Nemani, R. R., C. D. Keeling, H. Hashimoto, W. M. Jolly, S. C. 171-174. Piper, C. 1. Tucker, R. B Myneni, and S. W. Running (2003), Lewis, S. L,:J-(2006), Tropical forests and the changing earth sysClimate-driven increases in global terrestrial net primary protem, Phi/os. Trans. R. Soc. Ser. B, 361,195-210. duction from 1982 to 1999, Science, 300,1560-1563. Lewis, S. L., Y. Malhi, and O. L. Phillips (2004a), Fingerprinting Nevle, R. 1., and D. K. Bird (2008), Effects of syn-pandemic fire the impacts of global change on tropical forests, Phi/os. Trans. reduction and reforestation in the tropical Americas on atmoR. Soc. London, Ser. B, 359, 437-462. spheric CO 2 during European conquest, Palaeogeogr. PalaeoLewis, S. L., et a!. (2004b), Concerted changes in tropical forest climatol. Palaeoecol., 264, 25-38. structure and dynamics: Evidence from 50 South American long-term plots, Phi/os. Tran1. R. Soc. London, Ser. B, 359, Norby, R. 1., et al. (2002), Net primary productivity of a CO 2-enriched deciduous forest and the implications for carbon storage, 421-436. Ecol. Appl., 12, 1261-1266. Lewis, S. L., O. L. Phillips, and T. R. Baker (2006a), Impacts of Oliveira, P. H. F., P. A1iaxo, C. Pires, S. De Lucca, A. Procopio, B. global atmospheric change on tropical forests, Trends Ecol. Holben, J. Schafer, L. F. Cardoso, S. C. Wofsy, and H. R. Rocha Evol., 21,173-174. (2007), The effects of biomass burning aerosols and clouds on Lewis, S. L., O. L. Phillips, T. R. Baker, Y. Malhi, and 1. 1. Lloyd the CO 2 flux in Amazonia, Tellus, 59B, 338-349. (2006b), Tropical forests and atmospheric carbon dioxide, in Avoiding Dangerous Climate Change, edited by H. 1. Schell- Palace, M., M. Keller, G. P. Asner, 1. N. M. Silva, and C. Passos (2007), Necromass in undisturbed and logged forests in the Branhuber et a!., pp. 147-153, DEFRA, Cambridge Univ. Press, zilian Amazon, For. Ecol. Manage., 238, 309-318. New York. Lloyd, 1., and G. D. Farquhar (1996), The CO 2 dependence ofpho- Parmesan, C., and G. Yohe (2003), A globally coherent fingerprint of climate change impacts across natural systems, Nature, 421, tosynthesis, plant growth responses to elevated atmospheric CO 2 37-42. concentrations and their interaction with plant nutrient status, Peacock, J., T. R. Baker, S. L. Lewis, G. Lopez-Gonzalez, and O. Funct. Ecol., 10, 4-32. L. Phillips (2007), The RAINFOR database: Monitoring forest Lloyd, 1., 1. Grace, A. C. Miranda, P. Meir, S. C. Wong, H. S. Mibiomass dynamics, J. Veg. Sci., 18, 535-542. randa, I. R. Wright, J. H. C. Gash, and 1. A. MacIntyre (1996), A simple calibrated model of Amazon rainforest productivity Phillips, O. L. (1995), Evaluating turnover in tropical forests, Science, 268, 894-895. based of leaf biochemical propeliies, Plant Cell Environ., 18, Phillips, O. L., and A. H. Gentry (1994), Increasing turnover 1129-1145. through time in tropical forests, Science, 263, 954-958. Lloyd, 1., M. L. Goulden, 1. P. Ometto, S. Patino, N. M. Fyllas, and C. A. Quesada (2009), Ecophysiology offorest and savanna veg- Phillips, O. L., P. Hall, A. H. Gently, S. A. Sawyer, and R. Vasquez (1994), Dynamics and species richness of tropical forests, Proc. etation, Geophys. Monogr. Ser., doi:IO.l029/2008GM000740, Natl. Acad. Sci. U. S. A., 91, 2805-2809. this volume. Malhi, Y., and 1. Grace (2000), Tropical forests and atmospheric Phillips, O. L., et al. (1998), Changes in the carbon balance oftropical forest: Evidence fi'omlong-term plots, Science, 282, 439-442. carbon dioxide, Trends Ecol. Evol., 15,332-337. Malhi, Y., and O. L. Phillips (2004), Tropical forests and global Phillips, O. L., et al. (2002a), Changes in the biomass of tropical forests: Evaluating potential biases, Ecol. Appl., 12, 576-587. atmospheric change: A synthesis, Phi/os. Trans. R. Soc. London, Phillips, O. L., et a!. (2002b), Increasing dominance of large lianas Ser. B, 359, 549-555. in Amazonian forests, Nature, 418, 770-774. Malhi, Y., and O. L. Phillips (2005), Tropical Forests and Global Phillips, O. L., et al. (2004), Pattern and process in Amazon tl'ee Atmospheric Change, 260 pp., Oxford Univ. Press, New York. turnover, 1976-2001, Phi/os. Trans. R. Soc. London, Ser. B, 359, Malhi, Y., and 1. Wright (2004), Spatial patterns and recent trends 381-407. in the climate of tropical rainforest regions, Phi/os. Trans. R. Phillips, O. L., R. Vasquez Mmiinez, A. Monteagudo, T. Baker, Soc. London, Ser. B, 359, 311-329. and P. Nunez (2005), Large Iianas as hyperdynamic elements of Malhi, Y., et al. (2002), An intemational network to understand the tropical forest canopy, Ecology, 86, 1250-1258. the biomass and dynamics of Amazonian forests (RAINFOR), J. Phillips, O. L., et al. (2009), Drought sensitivity of the Amazon Veg. Sci., 13,439-450. rainforest, Science, 323, 1344-1347. Malhi, Y., et al. (2004), The above-ground coarse woody productivity of 104 neotropical forest plots, Global Change Bioi., 10, Prentice, 1. c., et a!. (200 I), The carbon cycle and atJnospheric carbon dioxide, in Climate Change 2001: The Scientific Basis: Contribu563-591. tion o/Worldng Group 1 to the Third Assessment Report a/the InMayaux, P., P. Holmgren, F. Achard, H. Eva, H.-J. Stibig, and A. tergovernmental Panel on Climate Change, edited by 1. Houghton Branthomme (2005), Tropical forest cover change in the 1990s et aI., pp. 183-237, Cambridge Univ. Press, Cambridge, U. K. and options for future monitoring, Phi/os. Trans. R. Soc., Ser. B, Retallack, G. 1. (2001), A 300-million-year record of atmospheric 360,373-384. carbon dioxide from fossil plant cuticles, Nature, 411,287-290. Meehl, G. A., et a!. (2007), Global climate projections, in Climate Change 2007: The Physical Science Basis: Contribution 0/ Rice, A. H., E. H. Pyle, S. R. Saleska, L. Hutyra, M. Palace, M. Keller, P. B. de Camargo, K. POliilho, D. F. Marques, and S. C. Working Group I to the Fourth Assessment Report a/the Inter-
PHILLIPS ET AL. Wofsy (2004), Carbon balance and vegetation dynamics in an old-growth Amazoni~11 forest, Ecol. Appl., 14, S55-S71. Salzmalill, U., and P. Jloelzma11l1 (2005), The Dahomey Gap: An abrupt climaticalJ,~ induced rain forest fragmentation in West Africa during th9'Iate Holocene, Holocene, 15, 190-199. Schnitzer, S. A., ahd F. Bongers (2002), The ecology of lianas and their role in forests, Trends Ecol. Evol., 17, 223-230. tel' Steege, H., et al. (2006), Continental-scale patterns of canopy tree composition and function across Amazonia, Nature, 443, 444-447. Van der Heijden, G., 1. Healey, and O. L. Phillips (2008), Infestation of trees by Iianas in a tropical forest in Amazonian Peru, J. Veg. Sci., 19, 747-756. Vieira, S., et a!. (2004), Forest structure and carbon dynamics in Amazonian tropical rain forests, Oecologia, 140,468-479. Vieira, S., S. Trumbore, P. B. Camargo, D. Selhorst, 1. Q. Chambers, N. Higuchi, and L. A. Mmiinelli (2005), Slow growth rates of Amazonian trees: Consequences for carbon cycling, Proc. Natl. Acad. Sci. U. S. A., 102,18,502-18,507. West, G. B., 1. H. Brown, and B. J. Enquist (1999), A general model for the structure and allometly of vascular plant systems, Nature, 400, 664-667. Wielicki, B. A., et a!. (2002), Evidence for large decadal variability in tropical mean radiative energy budget, Science, 295, 841-844. -
387
Williamson, G. B., W. F. Laurance, A. A. Oliveira, P. Delamonica, C. Gascon, T. E. Lovejoy, and L. Pohl (2000), Amazonian tree mortality during the 1997 EI Nino drought, Conserv. Bioi., 14, 1538-1542. Worbes, M. (1999), Annual growth rings, rainfall-dependent growth and long-term growth patterns of tropical trees fi'om the Caparo Forest Reserve in Venezuela, J. Ecol., 87, 391-403. Wright, S. J. (2005), Tropical forests in a changing environment, Trends Ecol. Evol., 20, 553-560. Wright, S. 1., O. Calderon, S. Hernandez, and S. Paton (2004), Are Iianas increasing in importance in tropical forests? A 17-year record from Panama, Ecology, 85, 484-489.
T. R. Baker, K.-1. Chao, S. L. Lewis, and O. L. Phillips, Earth and Biosphere Institute, School of Geography, University of Leeds, Leeds LS2 9JT, UK. ([email protected]) N. Higuchi, Laboratorio de Silvicultura Tropical, Instituto Nacional de Pesquisas da Amazonia, Manaus, Amazonas, Brazil. S. Vieira, Laboratorio de Ecologia Isotopica, USP, CENA, BR13400970 Piracicaba, SP, Brazil.
Ecosystem Carbon Fluxes and Amazonian Forest Metabolism Scott Saleska, l Humberto da Rocha,2 BaIi Kruijt,3 and Antonio Nobre 4 Long-term measurements of ecosystem-atmosphere exchanges of carbon, water, and energy, via eddy flux towers, give insight into three key questions about Amazonian forest function. First, what is the carbon balance of Amazon forests? Some towers give accurate site-specific carbon balances, as validated by independent methods, but decisive resolution of the large-scale question will also require integration of remote sensing techniques (to detect and encompass the distribution ofnaturally induced disturbance states across the landscape ofold growth forests) with eddy flux process studies (to characterize the association between carbon balance and forest disturbance states). Second, what is the seasonality of ecosystem metabolism in Amazonian forests? Models have historically simulated dry season declines in photo,synthetic metabolism, a consequence of modeled water limitation. Tower sites in equatorial Amazonian forests, however, show that photosynthetic metabolism increases during dly seasons ("green up"), perhaps because deep roots buffer trees from dry season water stress, while phenological rhythms trigger leaf flush, associated with increased solar irradiance. Third, how does ecosystem metabolism vary across biome types and laJ,id use patterns? As dry season length increases from equatorial forest, to drier southern forests, to savanna, fluxes show seasonal patterns consistent with increasing water stress, including a switch from dly season green up to "brown down." Land use change in forest ecosystems removes deep roots, artificially inducing the same trend toward brown down. In the final part, this review suggests that eddy tower network and satellitebased insights into seasonal responses provide a model for detecting responses to extreme interannual climate variations that can test whether forests are vulnerable to model-simulated Amazonian forest collapse under climate change.
IpIRE Program in Amazon-Climate Interactions, Department of Ecology and Evolutionary Biology, University ofArizona, Tucson, Arizona, USA. 2Depatiamento de Ciencias Atmosfericas, Universidade de Sao Paulo, Sao Paulo, Brazil. 3Alterra, Wageningen University and Research Center, Wageningen, Netherlands. 4Escritorio Regional do INPE, National Institute for Amazonian Research, Sao Jose dos Campos, Brazil. Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union, 10.1029/2008GM000728
1. INTRODUCTION Ecosystem-atmosphere exchanges of carbon, water, and energy are key integrated measures of ecosystem structure and function, as well as the means by which ecosystems are coupled to atmospheric processes and climate [Moorcroft, 2003]. Understanding the ecological, atmospheric, and climatic mechanisms that control these fluxes is intrinsically connected with understanding both ecosystems and the climate system, and, hence, with our ability to understand and pre'dict the future of these systems under increasing human pressure from both local development activities and global atmospheric change. 389
....-_--------------------------------------
390
ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLISM
Long-term measurements of ecosystem fluxes are critical to developing this understanding, and the Amazon basin, which by itself carries nearly 20% of the global freshwater discharge anti contains the largest intact tropical rainforests left on earth, is particularly impOliant to understand. Such long-term ecosystem flux measurements are now routinely made at hundreds of sites around the world [Baldocchi et al., 2001], primarily via eddy flux towers, which can use micrometeorological methods to integrate fluxes over many hectares at once, but most of these are in North America and Europe. A few early short-term measurements of ecosystem fluxes in Amazonian forest were made in the 1980s and 1990s [Shuttleworth et al., 1984; Fan et al., 1990; Grace et al., 1995], but only in the last few years has the scientific infi'astructure for measuring ecosystem fluxes in Amazonia reached a point where consistent long-term measurements are being made at a number of sites across the Amazon basin
--
-
SALESKA ET AL.
391
(Table 1 and Figure 1), and only now are these measurements are being integrated into a unified and widely available BrasilFlux network database [da Rocha et al., 2009; N. RestrepoCoupe et al., What controls the seasonality ofphotosynthesis across the Amazon basin? A cross-site analysis of eddy flux tower measurements from the Brasil flux network, manuscript in preparation, 2009 (hereinafter referred to as Restrepo-Coupe et al., manuscript in preparation, 2009)]. Here we will review the extent to which recent eddy flux measurements of carbon dioxide (C0 2) at Brasil-Flux sites (Table 1), together with remote sensing of forest vegetation indices, have given new insight into three key questions about Amazonian forest function: (1) What is the carbon balance of Amazonian forests? (2) What is the seasonality of ecosystem metabolism in equatorial Amazonian forests? (3) How does the seasonality of ecosystem metabolism vmy across gradients in climate and land use? Finally, we address
Table 1. Brasil Flux Sites With Eddy Flux Towers for Study ofAmazonian Carbon Dynamics ,75
Study Site Sao Gabriel Cachoeira, Amazonas Manaus, Amazonas (Reserva Cuieras) Santarem, Para (Tapaj6s National Forest)
Bananal Island, Tocantins Belem, Para (Caxiuana Nat'l Forest) Ii-Parana, Rondonia (Reserva Jan}) Mato Grosso Sinop Cotriguayu Pe-de-Gigante (Sao Paulo state)
Eddy Towers"
Precipitation (mma- 1)
DiY Season Length (months)
1. SGC: primaly wet forest (0028.8'N, 66°30'W)
started in 2005 b
4000+
0
K34: primmy rain forest (2°36.5'S, 60 012.5'W) C14: primary rain forest (ZF2) (2°35.4'S, 60 06.9'W) K67: primmy moist forest (2 051'S, 54°58'W) K83: selectively logged primmy forest W3'S, 54°56'W) 6. K77: pasture/agriculture (3.0l2°S, 54.537°W) 7. BAN: Seasonally flooded transitional forest
1999-presentb 1999-presentb•c 2001-2006 d 2000--2005 e,f
2200
2.4
1920
4.7
1500 2500
2.8
1999-2003i 1999-2003 1999-2002k
~2000
5
~2000
4
2004-20061
~1700
5
-70
-65
-60
203
-55
300
-50
o
-45
300
Kilometers
Figure 1. Map of Brazilian Amazonia showing annual average precipitation [New et ql., 2000], with flux tower locations
2. 3. 4. 5.
8. CAX: primary forest (l°43.06'S, 51°27.60'W) 9. RJA: primmy forest (lo o 4.7'S, 61°56.02'W) 10. FNS: pasture (lo o 45.7'S, 62°21.5'W) 11. SIN: transitional dly forest (II °24.75'S, 55°19.50'W) 12. COT: pasture (9°51.73'S, 58°13.81'W) 13. PEG: savanna (21°36'S, 47°36'W)
I999-present g 2004-presenth i
1999-present
"See footnotes below for principal investigators and references for each site. bA. O. Manzi!A. Nobre, INPA, Brazil. For Manaus, see Armijo et al. [2002] (1999-2001 data); or Malhi et al. [1998] (1995-1996 data). cPlus 1 year from Sept 1995 to Oct 1996 dS. C. Wofsy, Harvard University (2001-2007); S. R. Saleska, UofA, USA (2008-present); P. B. Camargo, CENAIUSP, Brazil. See Saleska et al. [2003], Hutyra et al. [2007]. eM. L. Goulden, UC Irvine, USA, H. R. da Rocha, USP, Brazil. See da Rocha et al. [2004], Goulden et al. [2004], Miller et al. [2004]. fNot including data gap when tower damaged by treefall. gD. R. FitzjmTald, SUNY, USA, O. L. Moraes, UFSM, Brazil. See Sakai et al. [2004]. hL. S. BOl'ma/E. Collicchio, UFT, Brazil; H. R. da Rocha, USP, Brazil; O. M. Cabral, EMBRAPA, Brazil. See EOIma et al. [2009]. iL. Sa/J. Cohen (Museu Goeldi, and UFPa-Belem, Brazil). See Carswell et al. [2002]. iA. O. Manzi, INPA, F. L. Cardoso (UFR), Brazil. See von Randall' et al. [2004]; Kruijt et al. [2004]. kN. Priante, UFMG, Brazil; G. L. Vourlitis, USA. See Priante-Filho et al. [2004], Vourlitis et al. [2001,2004,2005]. IH. R. da Rocha, USP, Brazil.
indicated by black points and site codes (see Table 1 for more details). Boxes identify Moderate Resolution Imaging Spectroradiometer (MODIS) remote-sensing transects: east-west equatorial zone (ext~nding from near the Atlantic coast to forests west of the K34 site near Manaus, Figure 4) and a north-south forest-savanna transition zone extending from central Amazonian forests near Santarem, past Sinop to savanna (Figure 7).
how ongoing long-term observations, including from eddy flux towers, may give us a powerful tool to address key aspects of the fundamentally important question: (4) What is the future of Amazonian forests under climate change? 2. WHAT IS THE CARBON BALANCE OF AMAZONIAN FORESTS? One of the motivations for eddy covariance flux measurements has been their potential utility for answering the question of whether intact forests are a source or sink for atmospheric carbon dioxide. This question, motivated by the hypothesis that excess atmospheric CO 2 should stimulate photosynthetic uptake [Lloyd et al., 1995], thereby enhancing ecosystem carbon storage, continues to be a central debate for tropical forests [Grace et al., 1995; Houghton, 2003; Ometto et al., 2005; Stephens et al., 2007]. Multiple methods in addition to the eddy flux technique are being used to quantify forest biomass and carbon balance, including plot-based biometric studies [Phillips et al., 1998, this volume], atmo-
spheric inversions based on changing CO 2 concentrations [Chou et al., 2002; Lloyd et al., 2007; see also Houghton et al., this volume], and satellite~based remote sensing [Lejsky et al., 2005; Saatchi et al., 2007]. We here review two general approaches by which eddy flux methods may contribute to estimates of ecosystem carbon balance: direct sampling at multiple sites and integrated process-based approaches combining tower measurements with remote sensing and modeling. The· first approach is to use eddy flux towers to directly sample Amazonian ecosystems. Initial eddy flux studies in the Amazon reported large annual carbon uptake [Grace et al., 1995; Malhi et al., 1998], but the approach of using eddy covariance methods to directly sample the carbon bal-ance of a large number of sites is not, in general, proving to be decisive for resolving the carbon-balance question in Amazonia. This is due, in large pmi, to uncertainties associated with measuring nighttime CO 2 losses, which make it difficult to achieve accurate estimates of annual carbon balances at some, though not all, Amazonian sites. The problem
392
ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLISM SALESKA ET AL.
arises because during calm nighttime conditions, turbulent exchange between the land surface and the atmosphere often tends toward zero, and other transpOli processes, such as subcanopy ~vection via cold-air drainage [Goulden et al., 2006; Acevedo et al., 2007; Tota et al., 2008], unmeasured by standard eddy flux configurations, can become significant [Goulden et al., 1996; Finnigan, 1999]. If these other processes are not accounted for, nighttime losses will be underestimated, leading to overestimation of whole-forest carbon uptake. This is a well-lrnown issue in eddy covariance flux measurements globally [Aubinet et al., 2000], but it may be more problematic in Amazonia than in temperate or boreal zones because of a greater frequency of calm nights [Malhi et al., 1998] and because the absolute magnitude of the annually integrated nighttime losses are larger, since there is no dormant season. Thus, the underestimated nighttime loss (proportional to the magnitude of nighttime fluxes) is potentially a much larger fraction of the net annual carbon balance at Amazonian forest sites [Miller et al., 2004]. In general, at eddy flux towers around the world, the problem of underestimation of nighttime emissions during calm periods is addressed by replacing measurements made at these times by some kind of independent estimate, either from models [Goulden et al., 1996], by scaling up data from chamber studies [Chambers et al., 2004], or by interpolation of eddy flux data from nighttime periods of more vigorous mixing [Gu et al., 2005]. In the absence of sufficient measurements to estimate the missing advective flows directly [e.g., Staebler and Fitzjarrald, 2004; Tota et al., 2008], underestimates in eddy flux-derived net ecosystem exchange are often corrected via a simple empirical method known as the "u* cOlTection." This method uses low friction velocity (or u *), a measure of the strength of mechanical turbulent mixing, to identify conditions when fluxes are likely underestimated, filters out flux measurements made at these times, and replaces the removed measurements by interpolation from time periods of more vigorous mixing [Goulden et al., 1996]. The u * correction method is not entirely satisfactOly because it provides only a rough index of when net ecosystem CO 2 exchange (NEE) fluxes are inadequate and can filter out a large fraction of the nighttime measurements (60% at the km 67 site, and 80% at the km 83 site, both near Santarem [Saleska et al., 2003]; and more than 90% at the k34 site near Manaus [Aratlj'o et al., 2002]), leading to large uncertainties and a risk of bias if the remaining fluxes deemed valid are not representative [Tota et al., 2008]. Nonetheless, the u * method has been well validated at some Amazonian sites by multiple independent methods (e.g., Hutyra et al. [2008] shows that u * -corrected fluxes at the Santarem km 67
site produce reliable estimates of carbon balance). Because of site-to-site differences, this method does not work equally well at all Amazonian sites. For example, as discussed by Kruijt et al. [2004], nighttime fluxes during low-u* periods at the Jam forest site near Ji-Parana in Rondonia are not significantly different to fluxes during high-u * periods, so the u * correction has little effect at this site. A number of different independent methods are now being used at Amazonian sites to test eddy flux estimates of carbon balance, including comparison to biometrically derived carbon balance [Saleska et al., 2003; Miller et al., 2004], comparison to sum of bottom-up flux components [Chambers et al., 2004], extrapolation of daytime NEE versus photosynthetically active radiation (PAR) curves [Hutyra et al., 2008], use of radon as a tracer [Martens et al., 2004], use of local boundmy-layer budget methods [Culj et al., 1999; Acevedo et al., 2004; Lloyd et al., 2007], and direct detection of the normally unmeasured advective fluxes [Tota et al., 2008]. These methods have both validated eddy-derived carbon budgets [Hutyra et al., 2008] and falsified them [Lloyd et al., 2007]. The larger obstacle to resolving the carbon-balance question in Amazonia, however, is not so much one of measurement accuracy, but of matching the scale at which measurements are made (whether via eddy flux towers or forest inventOly plots) to the scale of the processes that stmcture carbon gain and loss across the landscape. Oldgrowth forest landscapes are expected to be a mosaic of patches, a few of them recently disturbed (and losing carbon), with most long undisturbed (and gaining carbon or close to balance). If the dominant cause of disturbance is the mortality of individual trees, then the length scale of the resulting patches will be small, and there is a high chance that a single flux tower footprint will integrate over the whole spectmm of successional stages. On the other hand, ifmajor disturbance processes include large blowdowns, fire events, or regional-scale droughts, then we expect large patches and will require a much greater sampling effort to represent all the different stages [Moorcroft et al., 2001]. This s~aling problem leads to the second general approach, integrated process-based studies, which consist of combining the answers to two basic questions to quantify landscape-scale carbon balance: 1. What is the association between carbon balance and forest stmcture as a function of time since local disturbance? Ecological theories about forest succession [Odum, 1969] and gap phase regeneration [Brokmv, 1985; Hubbell et al., 1999] predict that increasing carbon storage (as initial large carbon losses from decaying dead wood gradually decline and are overtaken by increases in primmy production, Figure 2a) is intimately associated with changing forest stmcture (e.g.,
increasing canopy heights, Figure 2b) following disturbance, but empirical validation in real forests is critical. 2. What is the patcJx~ise distribution of disturbance states (time-since-disturb.~hce) and, hence, of patchwise carbon balances, across ,tile landscape? Time~since~disturbance is not known or observable for most old-growth forest patches in Amazonia, but it may be infelTed fi'om observations of forest stmcture indices. Evidence suggests that at least at large scales, disturbance events cluster to create length correlations larger than tower footprints [Nelson et al., 1994], so the answer to question 2 is unlikely to be obtained by random sampling by eddy flux towers or of forest inventOly plots [Feeley et al., 2007; Fishel' et al., 2008]. Remote sensing at sufficiently high resolution should be able to tell us something about the length scale of forest patchiness and its spatial variation and, hence, inform our strategy for answering this question [Jupp and Twiss, 2006]. For example, Hurtt et al. [2004] relied on aircraft-based lidar surveys to provide a direct measure forest stmcture, while Chambers et al. [2007] discussed satellite-based hyperspectral reflectance to detect variations in canopy photosynthetic greenness as an indicator of re. cently disturbed forest patches.
Question 1, however, can probably best be answered by validated long-term eddy flux measurements at sites that are known to have been recently disturbed, either from natural causes or from deliberate human disturbances such as selective logging [Figuiera et al., 2008], combined with biometric measurements of the components of carbon balance and forest stmcture. For example, at the km 67 tower site near Santarem, where the imbalance between live and dead wood pools and an initial observed net loss of forest carbon gives evidence of recent natural disturbance [Rice et al., 2004; Pyle et al., 2008], eddy flux observations show a trend fi'om source to sink (Figure 2c), and measurements of forest canopy heights over time show increasing canopy heights (Figures 2d and 2e). These trends are consistent with modeled disturbance-recovely dynamics (Figure 2), provide a critical test of such trajectories, and most impOliantly, a ground-based empirical basis for constraining models with remote sensing observations of disturbance state in order to extrapolate carbon balance to landscape scales. It is through these kinds of integrated studies, combining eddy flux measurements, models, and remote sensing, which the long-sought question about Amazonian forest carbon balance may be decisively addressed.
(A). Modeled Net Ecosystem Exchange (NEE = Rhet - NPP)
(8). Modeled Mean Canopy Height
f:k2:::J 00
20
40
60
Time since disturbance (years)
80
393
(0) 50 42
E
i
(E)
34 26
OJ
~ 18 10 2 I
I
I
0.0 0.04 0.08
-0.2
Fraction (2003)
density (m 2 m· 3 ) (2005 minus 2003)
0
0.2
1':1
Figure 2. Ecosystem demography model [Mool'cl'Ofi et al., 2001] providing a hypothesis for forest stmcture-carbon
~~l~nce relations: (a) simulated NEE and (b) canopy heights of a tropical forest gap following disturbance event. Large
Imtlal losses from dead wood are eventually overtaken by long-term regrowth as canopy heights increase. Measured carbon exchange and canopy height dynamics at Santarem km 67 tower site: (c) eddy-flux NEE trend from 2001 to 2005 (moving annual average) showing shift fi'om carbon source to sink [Hutyra et al., 2007] and portable canopy !idar measurements of (d) canopy height distl'ibution in tower footprint 'and (e) changes in vertical profile of canopy density between 2003 and 2005, showing increases in canopy height. Ground-based portable canopy !idar data provided courtesy of G. G. Parker.
394
SALESI(A ET AL.
ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLISM
3. WHAT IS THE SEASONALITY OF ECOSYSTEM METABOLISM IN EQUATORIAL lI" AMAZONIAN FORESTS? Measurements of fluxes that encompass the dominant timescales of variation in driving variables, diurnal, seasonal, and the multiyear oscillations associated with El Nifi(}-So~lthern Oscillation (ENSO) and also with other interhemispherical modes, proviqe a fundamental test of our understanding of ecosystem metabolism, since these fluxes are the consequence of the dominant metabolic activities of photosynthesis and respiration. We focus here on seasonality of these processes. Unlike in temperate zones, where the seasonality of ecosystem metabolism is obvious and straightforwardly dominated by an active growing season, which transitions to a dormant season in which the ecosystem substantially slows, the answer in the evergreen tropics is not so obvious. Eddy flux-derived gross ecosystem productivity (GEP, due to the total photosynthesis of the whole ecosystem) and ecosystem respiration (R eco ) provides direct insight into whole-forest metabolism. Eddy flux methods directly measure NEE of CO 2, the sum of the component photosynthetic and respiration fluxes. The component fluxes are typically separated by interpolating valid nighttime measurements (which represent respiration only) into the daytime, and then subtracting the daytime respiration from net exchange to derive photosynthesis [Hutyra et al., 2007]. We take advantage of this kind of separation (which necessarily accounts for both eddy fluxes measured above the canopy and fluxes due to changes in within-canopy storage of C02 in order to clearly distinguish nighttime and daytime net biological fluxes) to separately examine the metabolic processes of photosynthesis and respiration. 3.1. Photosynthetic Metabolism
Ecosystem and land surface models have historically represented ecosystems of Amazonia as water-limited, predicting dry season declines in photosynthesis and/or in the associated fluxes of water from transpiration by leaves [Dickinson and Henderson-Sellers, 1988; Nobre et al., 1991; Costa and Foley, 1997; Tian et al., 1998; Botta et al., 2002; Werth and AVissar, 2002; Lee et al., 2005]. An accumulating suite of direct observations, however, now suggests a different picture: in terra firme tropical humid forests in Amazonia, the photosynthetic metabolism of vegetation (as captured by photosynthetic and transpiration fluxes) is not water limited, at least up to seasonal timescales, but driven by a combination of available solar energy and intrinsic vegetation phenology. Support for this picture comes from both ground-based eddy
flux studies and from high-frequency remote sensing observations made possible by the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor on the TerTa satellite, validated by comparison to the ground-based studies. Surface-atmosphere water fluxes provide important insight into photosynthetic metabolism because transpiration by leaves is influenced by stomatal regulation. We briefly review the water flux findings here; for a fuller discussion evapotranspiration (ET), see da Rocha et al. [this volume]. Observations of whole-system water flux generally show a correlation with available energy (net radiation) across sites [Juarez et al., 2007; Hasler and AVissar, 2007; da Rocha et al., 2009], implying higher ET in the dry season, especially for sites near the equator, where net surface radiation is controlled primarily by cloudiness (since top-of-atmosphere solar radiation varies only moderately over the year). The ET-net radiation data suggest a distinction between equatorial sites, near Manaus [Shuttleworth, 1988; Armijo et al., 2002], Santarem [da Rocha et al., 2004; Hutyra et al., 2007] (Figure 3a), and Belem [Souza-Filho et al., 2005], which show no seasonal differences in the ET-net radiation relation and, hence, no evidence of seasonal water stress and southern forests, which have more intense dry seasons. The southern sites near Ji-Parana (semideciduous forest at JalU, tower RJA in Table 1), the transitional forest near Sinop (SIN in Table 1), and the ecotone (savanna-forest) site on Bananal Island at Tocantins Javaes (BAN in Table 1) show evidence of a weaker response to available energy in the dry season, a possible indication that sites with more intense dry seasons feel the onset of seasonal water stress, at least in some years [da Rocha et al., 2009]. Eddy flux towers, however, measure whole-system water flux (which includes evaporation from soils and leaf-intercepted rainfall), so observed patterns are not attributable to vegetation metabolism alone. Photosynthetic carbon fluxes are a direct indicator ofvegetation metabolism. The three seasonal equatorial sites (near Manaus, Santarem, and Belem) show little or no evidence of seasonal water limitation of the vegetation photosynthetic metabolism, as they all sustain high levels of GEP in the dry season (Figure 3) (Restrepo-Coupe et aI., manuscript in preparation, 2009) (an exception to this trend is a dry season decline in photosynthesis near the Manaus site, reported for the 1995 dry season by Malhi et al. [1998], although this same site later shows sustained dry season photosynthesis, as repOlied by Armijo et al. [2002]). Sustained (or increasing) dry season photosynthesis is consistent with the tower observations of sustained (or increasing) dry season ET, and neither supports the paradigm of an ecosystem operating the edge ofwater limitation (see also Meir et al. [this volume] for the effects of artificially imposed extreme drought).
A. Manaus (K34)
B. Santarem (mean of K67, K83)
395
C. Caxiuana (CM)
~
o
..
W
~ CSJ' ~
EE
LOSO I'--~E
(')
«::t
o~
e.t~
JFMAMJJASOND D. Ji-Parana Forest (RJA)
JFMAMJJASOND E. Ji-Parana Pasture
0:
JFMAMJJASOND F. Southern Savanna
LO
c::i ~~
~~~o
o
JFMAMJJASOND
JFMAMJJASOND
J FMAMJ,J ASOND
Figure 3. Composite seasonality of carbon fluxes and environmental drivers for each of six sites: (a-e) equatorial Amazon forest sites, (d) southern Amazon forest and (e) nearby converted pasture, and (f) a savanna site in Sao Paulo state (see Figure 1 and Table I). For each site, the top panel shows gross ecosystem productivity (GEP) (solid line), and ecosystem respiration (R eeo ) (dashed line), each expressed as a fraction of its annual maximum, while the bottom panel shows TRMM-derived precipitation (bars, with diy-season periods, < I00 mm month-I, highlighted by shaded gray panel), measured photosynthetical1y active radiation (PAR) (solid line), and estimated top-of-atmosphere radiation (R TOA ) (dashed line). Adapted from Restrepo-Coupe et aI., manuscript in preparation, 2009.
Remote sensing observations indicate that this general trend during the dry season (Figure 4). Especially strong dry picture, derived from point measurements at a few individ- season increases in forest photosynthetic activity were obual tower sites, also applies at broad scales across equatorial served over the portions of the eastern Amazonia that have Amazonia. The enhanced vegetation index (EVI) from the the longest dry season (5 months) and a minimum in annual MODIS is a composite of leaf area and canopy chlorophyll rainfall (in the vicinity of the Tapaj6s forest, 55°W), with content that does not saturate, even over dense Amazonian smaller increases in the western pOliion near Manaus where forests. Properly filtered to remove atmospheric aerosol and the dry season is shorter (~2 months) and rainfall greater. cloud effects, MODIS EVI tracks seasonal variations in can- In sum, in the forests of the central part of the equatorial opy photosynthesis [Huete et al., 2006]. The average seasonal . Amazon basin, those places with the greatest potential for profile of high-frequency (8-day composite) MODIS EVI drought stress (in terms of rainfall amounts and variability) observations, averaged over the years 2000-2005, and over also show the greatest tendency to increase photosynthea large swath of the equatorial Amazon basin (the primary sis a's water becomes limiting. A complete reversal in EVI forests from 50° to 65°W, location indicated in Figure 1), phenology (dry season "brown down") was observed east shows that these forests exhibit a large-scale "green-up" of Cauaxi, along the "arc of deforestation," where extensive
396 ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLISM pastures and secondary forests remain following deforestation [Roberts et al., 2003] (see question 3, below). The first-order picture of sustained or even increasing Amazonian ft)rest photosynthesis in the dry season also includes important second-order details that present a somewhat more nuanced picture of ecosystem metabolism than the water fluxes do. Perhaps the best insight is derived by analysis that combines meteorological observations from eddy flux towers with detailed biometric estimates of vegetation dynamics. In the Tapaj6s Forest near Santarem, for example, which has the longest dly season among equatorial Amazonian sites, ET shows a simple conelation with sunlight (Figure Sa ET versus PAR in Figure 5c), increasing in the dly season, in contrast to models representing waterlimited vegetation. Average GEP also exhibits dly season increases (Figure 5b), but with a more complex seasonal variation that is not simply correlated with either precipitation or sunlight. The GEP stmis to decline around April, but this cannot be due to drought stress because the decline begins well before the dly season begins, during a time when soil water supplies are high. Instead, the GEP decline appears driven by loss of photosynthetic capacity of the canopy (the forest's intrinsic ability, apart from seasonally varying light levels, to fix carbon, estimated by observing GEP at a given high light level) (Figure 5c, left axis). The loss in capacity is due to a presumed decline in leaf area index (LAI) from increasing litterfall, which accelerates in the late wet season [Goulden et aI., 2004; Rice et al., 2004; Hutyra et aI., 2007] (Figure 5c, right axis). Observed GEP, the consequence of the combined variations in photosynthetic capacity and variation in sunlight, reaches a minimum around the begil1lling of the dly season (just before the maximum litterfall), then begins to rise again, due to the combined effect of increases in light (due to fewer clouds), and an observed increase in photosynthetic capacity of the canopy [Goulden et al., 2004; Hutyra et al., 2007]. Goulden et al. [2004] suggest that these patterns result from an interaction between environmental variation and phenological rhythms of the forest canopy: photosynthesis decline late in the wet season is a result of the onset of leaf senescence, which is shortly followed by increases in litterfall. DIy season onset brings high light levels, followed by a flush of new leaves with high photosynthetic capacity, which combine to increase canopy photosynthesis. The conh'ols on the phenology of tropical leaf senescence and flushing are not well understood, but dly season flushing of new leaves has been well documented at the individual plant level when water is not limiting [Borchert, 1983; Wright and van Schaik, 1994]. DIy season LAI increases have been repOlied at the basin-wide scale with remote sensing [Myneni et al.,
2007]. Seasonal patterns in LAI, together with the seasonally changing distributions ofleaf age, drive the seasonal patterns of GEP in the Tapaj6s [Goulden et al., 2004; Doughty and Goulden, 2008]. Since leaf phenology is likely an adaptive response that anticipates seasonal climate variations, models which depend on direct responses to environmental driving variables may have difficulty predicting the response of forest photosynthetic metabolism to climate change. How are central Amazonian forests able to sustain high levels of metabolism during dry seasons, and why have so many model studies not captured this dynamic? Observations of exceptionally deep tree roots (up to 18 m) [Nepstad et al., 1994], hydraulic redistribution by roots [da Rocha et al., 2004; Oliveira et al., 2005], and seasonal shifts in the water supply for ET from shallow (3 m) to deep (>7 m) soil layers between wet and dly seasons [Bruno et al., 2006] suggest mechanisms by which these tropical forests maintain access to nontransient stores of deep soil water even when the surface becomes sufficiently dry to limit biological activity. Such mechanisms were missing from the early global models, many of which were built around more studied temperate and agricultural systems' and which typically represent forests as having 2-3 m rooting depths. Models (e.g., the CASA model) [Potter et al., 1998], which use empirical remote sensing metrics to drive vegetation phenology, rather than prognostically simulating vegetation phenological dynamics internally, should more closely follow the satellite-observed seasonal patterns. Indeed, the Potter et al. [1998] modeling study, driven by advanced very high resolution radiometer (AVHRR) remote sensing products, proposed hypotheses (including limitation ofAmazonianvegetationproductivityby solarradiationrather than by soil moisture because of deep roots that provided drought tolerance) that anticipate some of the later observations discussed here. Since EVI shows dly season green up (Figure 4), carbon cycle models driven by EVI also show this seasonality. Ecosystem model studies that have built on these findings have significantly improved seasonal phenology. For example, Costa and Foley used the same model (the integrated biosphere simulator), first [Costa and Foley, 1997], with a 2-m-deep forest root system, which simulated a wet season peak ET, but a second study [Costa and Foley, 2000] used 12-m-deep forest roots, and simulated a dly season peak in ET, similar to the eddy flux results. Stand-alone ecosystem models can be modified to represent seasonality well, as in the example Costa and Foley [2000] study cited above and other studies [Ichii et al., 2007; Fisher et al., 2007]. However, accurate simulation ofvegetation seasonality in fully coupled climate-carbon cycle models, which not only simulate dynamic vegetation but
SALESKA ET AL. generate the climate as well, remains a challenge. Inclusion of mechanisms such 3cs deep roots and hydraulic redistribution indeed improv~s general circulation model (GCM) perfOlmance, incre~~ing dly season vegetation activity and significantly movjrlg average annual model performance toward observation [Kleidon and Heimann, 2000; Lee et aI" 2005]. However, even 'the improved simulations, which show significant nonzero dly season metabolism, continue to show divergence from the observed seasonality of ecosystem metabolism (Figures Sa and 5b). More work is needed with land surface models embedded in GCMs to investigate how these results might be used to improve climate predictions. 3.2. Respiration Metabolism
Respiration has been represented across biomes globally as increasing functions of temperature and moisture (or precipitation) [Raich and Schlesinger, 1992; Lloyd and Taylor, 1994], with impOliant interacting effect of moisture seen even in relatively mesic systems like temperate forests [Davidson et al., 1998]. Even so, ecosystem respiratOly;netabolism emerges fi'om disparate complex processes (autotrophic and heterotrophic, aboveground and belowground, from a variety of substrates, including soil, litter, and decomposing dead wood, and from a variety of decomposing organisms
[see Metcalfe et aI., 2007] and may be expected to show different patterns than that of the comparatively unified processes underlying photosynthesis. Indeed, within equatorial Amazonian forests, flux towers show different and contrasting patterns in respiration by comparison to photosynthesis. Whereas ecosystem photosynthesis (GEP) in the central Amazonian sites of Manaus, Santarem, and Caxiuana (Figures 3a-3c) exhibits a common pattern (sustained or increasing during the dly season), ecosystem respiration (R eeo ) follows different patterns across these sites. In particular, we call attention to the conh'ast between primaly forests near Santarem, which show dly season declines in whole-system respiration with a minimum near the end of the dly season, and those near Manaus, where whole-system respiration increases throughout the dry season, reaching a seasonal peak shortly after the end of the dly season. The Santarem pattern of eddy flux-measured ecosystem respiration (Figure 3b), consistent with independently measured soil respiration in Santarem [Keller et al., 2005], indicates that water availability regularly limits respiration at these well-drained upland sites and especially during the relatively longer dly season in Santarem. The inverse precipitation sensitivities of photosynthesis (which increases during the dly season) and respiration (which declines) presumably arise from the contrasting availability of water for the two
Manaus
Fiona Caxiuana
1 45
Figure 4. MODIS-Enhanced Vegetation Index profile (200{}-2005 average) depicting spatial-seasonal variability across the equatorial Amazon transect of Figure 1, fi'om humid forests west of Manaus (65°W) to seasonally moist forests of eastem Amazonia (~500W) to the heavily convelted landscapes near the coast (45°-50 0W). [Huete et aI" 2006]. Note that the "gutter" at 50 W is an artifact of the Rio Tocantins, a tributmy of the Amazon. 0
397
398
SALESKA ET AL.
ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLISM
A. 4.5 ;;;:; 4.0 «l
:E 3.5 E 3.0
g
t:i
2.5 2.0 0.7 (;)
e
3.0
0.6
2.5'
0.5
2.0
• 0.4':
~ ~m 1.5
0.3 ~
1.0
0.2 .....
0.5
0.1 w
Q..
J:
c.
25
l:';:-In [C':' 20 III E
III
5-
:a :>
Oct
Jan Apr 2002
'u
u-o E 15 w::t
4. HOW DOES ECOSYSTEM METABOLISM VARY ACROSS GRADIENTS IN BlOME, CLIMATE, AND LAND USE?
Q..
Cl'-'
Figure 5. (a) (shaded band) Evapotranspiration in the equatorial Tapaj6s National Forest near Santarem: observed (mean ± SD across years 2002-2004 of eddy fluxes, K67 site), and (solid line) modeled by National Center for Atmospheric
Research (NCAR) GCM+CLM3.0 land model with deep roots and hydraulic lift [Lee et al., 2005] and (b) (shaded band) gross ecosystem productivity at the same site: observed (same period as in Figure Sa) and (solid line with points) modeled by modified NCAR CLM3.0 and (solid line) by integrated biosphere simulator [Botta et al., 2002]. (black squares) MODIS EVI (average 2000-2004) is plotted with GEP in Figure 5b. Models show dry season declines in contrast to observations from both satellite and eddy tower. (c) Mean annual cycle of canopy photosynthetic capacity (observed GEP when PAR;::; 800 Ilmol m-2 s-I), litterfall rates (plotted with reverse scale), and mean daily photosynthetically active radiation (PAR). (d) GEP and EVI in a pasture/agricultural area (K77 site) that has opposite seasonality from the nearby (~12 km distant) forest site in Figure 5b. Data are adapted from Saleska et al. [2003], Sakai et al. [2004], Huete et al. [2006], and HUlyra et al. [2007].
processes: deep tree roots can supply water for photosynthesis throughout the dry season, but most heterotrophic respiration arises from sources near the surface (soils, litter, and coarse woody debris) and is therefore inhibited by desiccation during the dry season [Saleska et al., 2003]. The Manaus pattem (Figure 3a), by contrast, suggests respiration may be oxygen-limited at this site which is a mosaic of upland forest plateaus and valleys, whose soils are saturated during much of the wet season [Armijo et al., 2002; Sotta, et al., 2004]. Detailed analysis of the pattems of ecosystem and soil respiration near Manaus [Chambers et al., 2004a, 2004b] generally supports this interpretation, showing that at the Manaus site: (l) annual average soil respira-
experiments, which show unexpectedly contrasting responses to simulated drought. In,the eastem Amazonian Caxiuana national forest, droughte6n sandy oxisols induced large (22%) reductions in soil re,is'piration [Sotta et al., 2007], while in the central-eastem fo!,Jst near Santarem, drought on clay oxisol smprisingly induced no net change in soil respiration [Davidson et al., 2008]. These differences may be due to shallower rooting depth and 'sandy soil texture, which may make the Caxiuana forest more subject to water limitation, but significant unceriainties remain (for a more detailed discussion, see Meir et al. [this volume] on the effects of drought). Pattems of soil respiration may also be related to the interaction between climate and topography. As with overall carbon balance, scaling soil respiration to the landscape may be aided by remote sensing-based landscape models of terrain and hydrology [Renno et al., 2008], which can then be used to analyze flux tower footprints at different sites [Nobre et al., 2008].
lion is lower in the valleys (where soils are more frequently saturated) and higher on plateaus; (2) soil respiration shows a negative relation with soil moisture, peaking at a volumetric soil water of ~0.20 m 3 m-3, and declining through most of the range of observed soil moisture; and (3) that the response of ecosystem respiration to precipitation pulses is context dependent: R eeo responds positively to precipitation pulses in the dry season, but decreases after precipitation pulses in the wet season. The similarity in observed pattems of photosynthesis, but difference in pattems of respiration between different sites, suggests that controls on respiration are complex, a conclusion further reinforced by ecosystem-scale water manipulation
We here address the impOliant question about how seasonal pattems change across space and thereby gain insight into the fundamental issue of how ecosystem stmcture and climate give rise to variations in ecosystem function. Eddy flux measurements are giving insight into at least three kinds of spatial variation in Amazonia: (1) within biome variations in uptake of CO 2 (assessed by comparing nearby towers, as, for example, located within the similar forests near Manaus); (2) changes in the seasonality of ecosystem metabolism with climatic variation (from the equatorial moist forests near the main stem of the Amazon River to the drier forests along the southem edge of Amazonia); and finally, (3) changes in seasonality caused by "anthropogenic" impacts (i.e., land use changes, primarily the conversion of forest to pasture or cropland). Our main focus here is on how seasonality varies with climate or biome, rather than variations in the magnitude of average ecosystem productivity (which are much less ceriainly known from tower measurements, see question I). Biometric assessments of wood productivity are likely a more appropriate indicator of large-scale variations in productivity across the Basin [Malhi et al., 2004, this volume]. 4.1. Within-Biome Variability
Because eddy flux towers are relatively expensive, towers are rarely installed near each other in forest considered to represent the same biome. However, the two towers near Manaus (at sites CI4 and K34) and two towers near Santarem (at
399
sites K67 and K83 of the Tapaj6s National Forest), in each case, installed in forests of the same biome, have provided an opportunity to investigate this assumption. Net ecosystem exchange of CO 2 at the two Manaus towers showed significantly more net carbon uptake at the CI4 site than at the K34 site, over the same time period in 1999-2000 [Armijo et al., 2002], suggesting that there may be more uptake on well-drained plateaus than in valleys, which typically have saturated soils, and that this may explain why the Cl4 tower, whose footprint encompasses over 25% more plateau areas within 1 km of the tower, also shows greater uptake. Ara~ijo et al. [2002] also do not mle out the possibility that there may be artifacts due to the difficulty of interpreting eddy flux measurements taken over complex terrain [Baldocchi et al., 2000]. By contrast, net carbon exchange measured at the K67 and K83 tower sites (before selective logging at the K83 site) was indistinguishable from each other after appropriate corrections for calm nighttime conditions [Saleska et al., 2003]. The Tapaj6s upland tower sites are unusually flat, with fairly unifOllli clay soils interspersed with patches of sand on a well-drained planaIto [Silver et al., 2000] and similar canopy stmcture, so one might expect that these two sites should exhibit similar carbon balance. The tenain at these tower sites more closely approtimates the ideal for the application of eddy flux methods,but it is also less representative of the broader landscape of most Amazonian forest. Since much of Arnazonia consists of dissected terrain like that seen around the eddy flux towers near Manaus, an important remaining challenge in using eddy flux methods to interpret Amazonian forest carbon balance lies in improved understanding of how the components pieces ofplateau-andvalley mosaics integrate to landscape-scale carbon balance. This requires study in two key areas: the micrometeorology of fluxes over complex terrain [e.g., T6ta et al., 2008] and biogeochemical ecology of plateau and valley forests. 4.2. Climate/Biome Variability: From Equatorial to Southern Forests to Savanna
Moving from equatorial Amazonian forests (Figures 3a3c) to drier southem Amazonian forests (Figure 3d) reveals a distinct shift in seasonal pattems: Reserva Jam, near JiParana, Rondonia, shows a distinct dry season decline in canopy photosynthesis (Figure 3d) [von Randow et al., 2004]. This is also similar to that exhibited by transitional . forest (lIAoS, 55AOW) near Sinop [Vourlitis et al., 2005]. This suggests that photosynthetic metabolism at these southern sites experiences some water limitation during the dry season. This is consistent with ET patterns at these sites, which also show dry season declines [da Rocha et al., 2009],
400
ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLISM
though it should be noted that dry season ET is still much higher than coincident precipitation inputs, indicating significant soil water storage and ability to mine deep soil water that is re-charged only by wet season inputs, even at these drier southern sites. This compares to a marked seasonal difference also seen in terra fume humid forest tree growth (as recorded by diameter increment) compared to the floodplain tropical forests, which are seasonally inundated by nearby rivers and which represent as much as 14% ofthe Amazon River basin. Tree growth reaches its maximum during the rainy season in the fonner according to biometry [Rice et al., 2004] and during the dry season in the latter, respectively [Dezzeo et al., 2003]. In the Tocantins ecotone-floodplain eddy flux site [Bmma et al., 2009], ecosystem productivity abmptly drops during inundation likely due to anoxia-induced stress (thus similar to the floodplain forests) and quickly recovers early in the dry season (thus similar to the equatorial tena finne and the floodplain forests). In addition, increasing productivity during the dry season is concunent with decreasing ET, hence, the ecotone site apparently behaves similarly to the equatorial forests for GEP seasonality and similarly to the savarmas for seasonal ET [da Rocha et al., 2009]. The trend from high or increasing dry season photosynthetic activity (in equatorial forests) to moderate dry season photosynthetic decline (in southern forests) reaches its extreme in the transition from forest to savanna (or cerrado) biome (Figure 3f). In comparison to forest vegetation, savanna photosynthetic metabolism exhibits a much larger range, experiencing an average reduction of 80% during the driest part of the dly season (Figure 3f). Analysis suggests that two factors are associated with these changing seasonal patterns of photosynthetic metabolism (Restrepo-Coupe et a1., manuscript in preparation, 2009): differences in the depth of soils and changes in the seasonality of smface solar energy input. First, for example, shallower soils of about ~3 mat Ji-Parana [von Randow et al., 2004] may induce water limitation that slows photosynthesis during the dry season. Second, as the environmental driver data across sites in Figure 3 shows, photosynthetically active radiation (PAR) at the surface does not increase at southern sites (Figures 3d-3f) during the dry season as it does in equatorial forests. In equatorial forests, with relatively constant top-ofatmosphere solar input, surface solar radiation, being limited primarily by clouds which are at their peak in wet seasons, achieves a strong maximum in the dly season. With increasing latitude, however, the seasonality of top-of-atmosphere radiation increases, with strong dry season minima emerging to counteract the illy season clear-sky effect (compare Figure 3a and 3b versus Figures 3d and 3e in top-of-atmosphere and surface PAR); the net result is less seasonal variability in ra-
diation reaching the surface at the southern sites. In addition, southern sites are subject to "fi'iagens," or cold periods, due to weather patterns circulating colder air from more southern latitudes into the margins of Amazonia [Oliviera et al., 2004]. In equatorial sites, there may be an adaptive advantage to organizing plant form (including allocation to deep roots) to maintain high levels of photosynthetic metabolism to take advantage of increased energy inputs during the dry [Goulden et al., 2004; Doughty and Goulden, 2008]. If so, this advantage may well be diminished at southern sites where energy input is not any higher in the dry season (RestrepoCoupe et a1., manuscript in preparation, 2009). With regard to respiration metabolism, the southern forest, pasture, and savanna sites all exhibit seasonal patterns that show consistency between photosynthesis and respiration: these sites all exhibit at least some degree of water limitation, as evidenced by dry season declines and/or dry season minima in whole-system respiration. The seasonality of the combined effects of photosynthesis and respiration, NEE, despite arising from distinct processes, nonetheless exhibits a convergent response across biomes: areas with longer dry seasons or less rainfall exhibit larger ranges in NEE seasonality (Figure 6), and this trend seems to hold, even as the processes dominating seasonality vary across precipitation gradients: wetter sites like Manaus showed little NEE seasonality, sites with moderate precipitation (like Tapaj6s and Sinop) showed moderate NEE seasonality dominated by variation in respiration, and drier savamla sites (like those in Brasilia and Sao Paulo state) showed large NEE seasonality dominated by variation in GEP (Figure 6). The net result is that the intensity of drought is a key control on the seasonality of carbon cycling, but that the ecophysiological process by which drought exerts this control varies across sites. This again points to the need for studies, which can generate a more predictive understanding of when and where the photosynthetic versus respiratory processes will be more sensitive to climatic variations. 4.3. Land ,Use Changes (Pasture and Agricultural)
The impacts of land use changes in Amazonian forests are one of the most directly visible effects of human development as large areas of Amazonia are deforested and transformed into pasture or other agricultural uses. Beyond the most apparent impacts, however, eddy flux measurements of carbon exchange, together with remote sensing measurements, are allowing deeper insight into the functional consequences of land use change for ecosystem metabolism. In central Amazonia, for example, the general dry season green-up pattern seen in the tower-measured photosynthetic metabolism of intact forests (Santarem K67 site, Figures 5a
SALESKA ET AL. 1800
p~~.,. ;1~;~~O~;lilY
(l)
Ol
c;--- 1500
,OE, phollJsynlhesls r!
l.'! ~b
wE 1200 w~
Zoo - .c
:go
900
°Ol (J)ol<:
600
ro~ Q)
if)
NEE seasonality dominated by respiration
",
. <J~'
hi'
\r4(~~
,
c
300 0
c2 = 0.67; ~ '" 0.005 1600
2000
2400
~K34
14(9)
2800
Annual average rainfall (mm)
Figure 6. Seasonal variation in NEE (maximum-minimum of monthly average) versus annual average rainfall. Symbols are (see Table 1 for more details): P, PEG (Sao Paulo savanna); B, Brasilia (savanna); S, SIN (Sinop transitional forest); K67, K83, Tapaj6s National Forest (equatorial moist forest); J, RJA (southem forest); 14(5), 14(9), Manaus C14 site (5 = 1995-1996; 9 = 1999-2000 in equatorial rainforest); C, CAX, Caxiuana National Forest (equatorial rainforest); K34, Manaus K34 site (equatorial rainforest). Adapted from VOllrlitis et at. [2005]. Copyright 2005 American Meteorological Society.
and 5b) is completely reversed to dry season brown down when the forest is converted to pasture/agricultural land (Santarem km77 site, Figure 5d). (These same contrasting seasonal patterns in forest versus pasture are also seen in the satellite-derived EVI of MODIS at these sites, Figure 5b versus Figure 5d.) Similarly, the large-scale green up in equatorial forests seen from satellite across central Amazonia (from 50° to 65°W, Figure 4) exhibits a complete reversal in phenology in observations east of Cauaxi (from 45° to 50°, Figure 4), along the "arc of deforestation," where extensive pastures and secondary forests remain following deforestation [Roberts et al., 2003]. In biomes and climate zones where intact forests are already drought-sensitive compared to the drought-resistant equatorial forests (e.g., southern forest in Reserva Jam, near JiParana, Figure 3d), conversion to pasture further exacerbates the seasonal sensitivity, as in the Ji-Parana-convelied pasture site at Fazenda Nossa Senhora (Figure 3e) [von Randow et al., 2004], where ecosystem photosynthesis suffers a 50% drop by the end of the dry season compared to the more moderate declines seen in the nearby Reserva Jam forest. 4.4. Summwy
An integrated depiction of different climate zones, biomes, and land use changes in terms of the seasonality of ecosystem photosynthesis (as measured at all eddy tower sites, Figure 7a, or as recorded via EVI from satellites, Figure 7b) sum-
401
marizes the differing effects discussed above. For example, shifts from equatorial forest, to southern forest, to savanna show seasonal patterns that correspond to distinctly different degrees ofwater stress (Figure 7a). Likewise, sufficiently long dry seasons eventually lead, along the north-south climate gradient of Figure 1 (which corresponds mostly to declines in precipitation), to brown down of vegetation in the dry season (Figure 7b). Large-scale land use change/deforestation can artificially induce the effect of the forest-savanna transition by converting forests from green-up to brown-down regions, presumably due to the removal of deep roots, which allow for access to stored water during dry periods. 5. WHAT IS THE FUTURE OF AMAZONIAN FORESTS UNDER CLIMATE CHANGE? Globally significant changes in Amazonian carbon and water cycles, including widespread forest collapse and conversion to savanna due to global warming-induced drought, are projected by some coupled carbon/climate models [Cox et al., 2000; Betts et al., 2004]. Other models imply forest persistence [Friedlingstein et al., 2006] (Figure 8a). These differences are not due only to different climate changes in different models, put importantly, to differences among models in representation of forest function and feedbacks to climate [Sitch et aJ., 2008]. Current knowledge is insufficient to determine which model representations ofvegetation function are most consistent with real forest ecosystems. Despite acknowledged shortcomings in model simulations (for example, virtually all of the climate models consistently simulate Amazonian precipitation that is too low, even under present climate) [Malhi et al., 2009], different model mechanisms thus provide hypotheses that can be combined with continuing observations from the network of eddy flux towers and frorn satellites, in order to provide key insights into forest-climate interactions and, hence, the potential future of Amazonia. For example, some of the mechanisms implicated in the Hadley Centre model-predicted forest collapse [Cox et al., 2000; Betts et al., 2004] should be testable by the right set of observations. In these simulations, modeled forest collapse is a consequence not only of climate change-induced drought, but of amplification by the physiological response of the forest: water-limited vegetation responds to initial drought by reducing transpiration (and photosynthesis), which in turn exacerbates the drought by intenupting the supply of water that would otherwise contribute to the recycled component ofprecipitation [Betts et al., 2004]. This modeled physiological feedback mechanism, implicated in long-term mechanisms leading to forest collapse under climate change,
402
ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLISM
SALESKA ET AL. Equatorial forest 1<34
>< co
1
'7 E 0.8 0W
1<83 CAY.
Southern forest RJA Pasture/Agriculture
(9
0.4
0W
"" 1<77 FNS
0.2
(9
Savanna
30 60 90 120 150 180 210 240 270 300 330 360
PDG
\
0.80 0.70 0.60
~
0.50
0040 0.30 0.20 0.10
o
32 64
96 128 160 192 224 256 288 320 352
Days since dry-season onset Figure 7. (a) Seasonal pattern in normalized GEP (relative to stali of illy season) across tower sites spanning equatorial forests, southern forest, convelied sites, and savanna (for site locations, see Figure 1). (adapted from Resh'epo-Coupe et a!., in review, 2009). (b) MODIS EVI seasonal profiles (relative. to stali of the local dly season) at sites progressively further south along Tapaj6s-Cuiaba transect (black vetiical box in Figure 1). (shaded region) As dly season length grows, mean annual EVI declines, and the magnitude of dry season "green up" diminishes, eventually reversing to "brown down" as vegetation transitions to cerrado (adapted from P. Ratana, et a!., Variability in Amazon phenology across the Transitional Rainforest-Cerrado Ecotone, submitted to Remote Sensing ofEnvironment, 2009).
(a) 12 I':R::-e-s:-id::-ua""'l-e-m7is-si'-on-s--",--:-::'"J'/'--j
EI Nino Drough
(Fos. Fuel minus /:::;;,., \ 8 ocean sink) ,fl Loss of I-Iadley u Center' /7 Amazon §.4 fares Nl\' ~ o
"D.
.~
~
~
o
E E
0
-5
-4 Terrestrial sink
Year: 1950
200 D. 100'~
IPSL -""""".>4" "\"","'4,'\ "'0/\1""'
'-
L--'------'------'-_'------'------'-_'------w0"-
2000
2050
2100
Years: -3
-2
-1
0
2
3
4
Figure 8. What is the future of Amazonia under climate change? (a) Models predict starkly different fuhlres for the terrestrial carbon sink: from sink enhancement (IPSL) [Dl!fi'esne el al., 2002] to switchover to carbon source (Hadley Centre) [Cox el al., 2000], a consequence in part of catastrophic collapse of the Amazon forest, triggered by onset of a semipermanent drought, critically exacerbated by physiological feedback, which amplifies the drought [Belts el aI., 2004] (see text). The importance of mechanisms implicated in forest collapse can be tested by observing forest response (fluxes, vegetation dynamics, and remotely sense indices) to real-world El Nifio-Southem Oscillation -like drought, and comparing to (b) simulated photosynthesis and precipitation in central Amazonia through years relative to El Nifio drought [Jones el a!', 2001].
should also cause short-term reductions in transpiration and photosynthesis during periods of drought under current climates (as shown infi'tgure 8b). The evidence presented in the previous sectionsJfocused on seasonal patterns of eddy flux observations. H)fwever, the key question for testing mechanisms implicated in simulated Amazonian forest collapse is not regular seasonal variations, but more extreme drought events, such as those arising from variation at climatically relevant timescales (interannual to decadal, to include strong droughts such as those associated with El Nino and the tropical Atlantic modes) [Nobre and Shukla, 1996]. Anticipated effects of extreme drought can be tested by experiments [see Nepstad et al., 2007; Fisher et al., 2007; Meir et al., this volume] and by observations that are of sufficient duration to include wide-scale periodic droughts. The Brasil-Flux tower network and the MODIS remote sensing satellites together now make such observational tests possible. Since viliually all of both the ground-based tower and satellite infi'astlucture were put in place since the last large ENSO-related drought in 1997/1998, we have a unique opportunity to detect, at the basin-scale and using independent methods, ecosystem metabolic and carbon cycle flux responses to large-scale real-world droughts like those expected under climate change. An example ofthe scientific opportunities was provided by the short but intense Amazon basin drought of2005, related not to El Nino but to the anomalies in tropical north Atlantic sea-surface temperatures that occurred in 2005 [Marengo et al., 2008]. The 2005 drought coincided with the onset of the dry season in the southern and western areas of the basin, a distinctly different pattern from El Nino droughts, which suppress rains in the wet season and prevent groundwater recharge, mostly in central and eastern Amazonia. Eddy towers in the central-eastern parts of the basin therefore did not sample the 2005 drought the way they likely would an El Nino drought, but analysis of MODIS EVI data from satellite observations of vegetation response did not show the expected decline (as in Figure 8b). Instead, an increase in forest photosynthetic capacity in droughted parts of the basin was observed [Saleska et al., 2007], perhaps due to increased light availability, suggesting that Amazonian forest metabolism may be more resilient than models project, at least to short-term drought (drought effects are discussed by Meir et al. [this volume]). Over a period of several years, which encompassed the 2005 drought, however, tree mortality reportedly increased in those forest plots that were in droughted palis of the basiri [phillips et al., 2009]. Whether these seemingly opposite responses (shOli-term increase in photosynthetic capacity dm1ng the peak months of drought, encompassed by excess mOliality integrated over several years) may be reconciled by account-
403
ing for the differing timescales and the effect of time lags, is the subject of ongoing study. Plant mOliality is a consequence of carbon deficit, in which respiration exceeds gross ptimaty productivity. Carbon deficit, however, may accumulate to fatal levels even in the face of short periods of elevated productivity, for example, if elevated temperatures simultaneously induce sufficient increases in respiration. However, relationships between controls on metabolism and on mortality are poorly understood and are an active research area, as sophisticated models of plant mortality are lacking (but see McDowell et al. [2008] for a review and discussion of possible ways fOlward). The conjunction of observations from space and from a ground-based network (including biometric plots and eddy flux towers) is thus a powerful empirical resource for improving understanding of overlapping patterns of forest resilience and vulnerability in response to climatic change. Continued observations are essential for obtaining robust observational tests of the ecosystem-scale mechanisms important to understanding the future of Amazonian forests. 6. CONCLUSION Amazonian eddy flux studies are providing new insights into the controls on ecosystem carbon balance, on the seasonality of photo~ynthetic and respiration metabolism in tropical systems, the different factors that influence metabolism in different ]Jiomes and climate zones, and the ways in which human changes to the landscape alter ecosystem functions along with the changes in stmcture. Significant work is still needed to address important outstanding issues about Amazonian forest function, including the following. 6.1. Large-Scale Amazonian Carbon Balance
Long-term eddy flux measurements can provide essential details and test models of forest disturbance dynamics. Integrated with remote sensing, models, and long-term plotbased biometric studies, they provide a powerful approach to resolving this long-standing question. Eddy flux-derived carbon balance has been independently validated at some individual sites, but many other sites still need to be evaluated before they can be reliably used with respect to this goal, due to ongoing uncertainties about accurate characterization of nighttime losses. 6.2. Source ofDifferences in the Seasonality Between Biomes and Processes
Current understanding of the controls on respiratOly metabolism is particularly limited. Existing studies suggest
404
ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLISM
intriguing hypotheses for these differences, but additional work to test these is needed, including more extensive coverage in regions of the basin where measurements are lacking (e.g., the very wet northwestern part of Amazonia, where new measurements from Sao Gabriel da Cachoeira should add new observations soon). 6.3. Future ofAmazonian Forests Under Climate Change
The network of Amazonian \eddy flux towers, together with satellite sensors like MODIS, provide a unique opportunity, not yet realized, to observe vegetation responses to ENSO-scale droughts and thereby test some of the model mechanisms that are critical to determining Amazonian forest response to future drought, including those implicated in the widely cited simulation of Amazonian forest collapse [Betts et aI., 2004]. In sum, the newly functioning network of Brasil-Flux towers has in the last few years provided fundamental new insights into mechanisms which control the cycling of carbon and the functioning of forest metabolism in Amazonia. These observations are transforming our understanding of how these forests work, but longer-tenn observations to capture the scales of response to interannual variability, especially if integrated with other kinds of data (remote sensing, plot-based biometlY), promise to open as many more windows of understanding in the future. REFERENCES Acevedo, O. C., et al. (2004), Inferring nocturnal surface fluxes from vertical profiles of scalars in an Amazon pasture, Global Change BioI., 10(5), 886-894. Acevedo, O. C., et al. (2007), Turbulent carbon exchange in velY stable conditions, Boundm)J Layer Meteorol., 125(1),49--61. Aralljo, A. C., et al. (2002), Comparative measurements of carbon dioxide fluxes from two nearby towers in a central Amazonian rainforest: The Manaus LBA site, J Geophys. Res., 107(D20), 8090, doi: 10.1 029/200IJD000676. Aubinet, M., A. et al. (2000), Estimates of the annual net carbon and water exchange of forests: The EUROFLUX methodology, Adv. Ecol. Res., 30,113-175. Baldocchi, D., 1. Finnigan, K. Wilson, K. T. PawU, and E. Falge (2000), On measuring net ecosystem carbon exchange over tall vegetation on complex terrain, Boundm)J Layer Meteorol., 96 (1-2),257-291. Baldocchi, D., et al. (2001), FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities, Bull. Am. Meteorol. Soc., 82(11), 2415-2434. Betts, R. A., P. M. Cox, M. Collins, P. P. Harris, C. Huntingford, and C. D. Jones (2004), The role of ecosystem-atmosphere interactions in simulated Amazonian precipitation decrease and
forest dieback under global climate warming, Theor. Appl. Climatol., 78(1-3), 157-175. Borchert, R. (1983), Phenology and control of flowering in tropical trees, Biotropica, 15(2),81-89. Borma, L. S., et al. (2009), Atmosphere and hydrological controls of the evapotranspiration over a floodplain forest in the Bananal Island region, Amazonia, J Geophys. Res., 114, GOI003, doi: 10.1 029/2007JG000641. Botta, A., N. Ramankutty and 1. A. Foley (2002), Long-term variations of climate and carbon fluxes over the Amazon basin, Geophys. Res. Lett., 29(9),1319, doi:lO.l029/200IGL013607. Brokaw, N. V. L. (1985), Gap-phase regeneration in a tropical forest, Ecology, 66, 682--687. Bruno, R. D., H. Rocha, H. C. Freitas, M. L. Goulden, and S. O. Miller (2006), Soil moisture dynamics in an eastem Amazonian h'opical forest, Hydrol. Processes, 20, 2477-2489. Carswell, F. E., et al. (2002), Seasonality in CO 2 and H 2 0 flux at an eastern Amazonian rain forest, J Geophys. Res., 107(D20), 8076, doi: I0.1 029/2000JD000284. Chambers, 1. Q., E. S. Tribuzy, L. C. Toledo, B. F. Crispim, N. Higuchi, 1. dos Santos, A. C. Araujo, B. Kruijt, A. O. Nobre, and S. E. Trumbore (2004), Respiration from a tropical forest ecosystem: Partitioning of sources and low carbon use efficiency, Ecol. Appl., 14(4 Suppl.), S72-S88. Chambers, 1. Q., G. P. Asner, D. C. MOlton, L. O. Anderson, S. S. Saatchi, F. D. B. Espirito-Santo, M. Palace, and C. Souza (2007), Regional ecosystem structure and function: Ecological insights from remote sensing of h'opical forests, Trends Ecol. Evol., 22(8), 41~23. Chou, W. W., S. C. Wofsy, R. C. Harriss, 1. C. Lin, C. Gerbig, and G.W. Sachse (2002), Net fluxes of CO2 in Amazonia derived fi'om aircraft observations, J Geophys. Res., 107(D22), 4614, doi: 10.1029/200 IJDOO 1295. Costa, M. H., and 1. A. Foley (1997), Water balance of the Amazon Basin: Dependence on vegetation cover and canopy conductance, J. Geophys. Res., 102(D20), 23973-23989. Costa, M. H., and 1. A. Foley (2000), Combined effects of deforestation and doubled atmospheric C02 concentrations on the climate of Amazonia, J Clim., 13, 18-34. Cox, P. M., R. A. Betts, C. D. Jones, S. A. Spall, and 1. 1. Totterdell (2000), Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model, Nature, 408(6809), 184-187. Culf, A. D:, G. Fisch, Y. Malhi, R. C. Costa, A. D. Nobre, A. D. Marques, 1. H. C. Gash, and 1. Grace (1999), Carbon dioxide measurements in the nocturnal boundmy layer over Amazonian forest, Hydrol. Earth Syst. Sci., 3(1), 39-53. da Rocha, H. R., M. L. Goulden, S. O. Miller, M. C. Menton, L. D. V. O. Pinto, H. C. de Freitas, and A. M. S. Figueira (2004), Seasonality of water and heat fluxes over a tropical forest in eastern Amazonia, Ecol. Appl., 14(4), S22-S32. da Rocha, H. R., et al. (2009), Patterns of water and heat flux across a biome gradient from tropical forest to savanna in Brazil, J Geophys. Res., 114, GOOB 12, doi: 10.1 029/2007JG000640. da Rocha, H. R., A. O. Manzi, and 1. Shuttleworth (2009), Evapoh'anspiration, Geophys. Monogr. Ser., doi: I0.1 029/2008GM000744, this volume.
SALESKA ET AL. Davidson, E. A., E. Belk, and R. D. Boone (1998), Soil water content and temperature as independent or confounded factors controlling soil respiratioji in a temperate mixed hardwood forest, Global Change Bioli!,'4, 217-227. Davidson, E. A., Francoise, and D. C. Nepstad (2008), Effects of an expet'fmental drought and recovelY on soil emissions of carbon dioxide, methane, nitrous oxide, and nitric oxide in a moist tropical forest, Global Change BioI., 14, 25822590. Dezzeo, N., M. Worbes, 1. Ishii, and R. Herrera (2003), Annual tree rings revealed by radiocarbon dating in seasonally flooded forest of Mapire river, a tributary of the lower Orinoco river, Venezuela, Plant Ecol., 168, 165-175. Dickinson, R. E., and A. Henderson-Sellers (1988), Modeling tropical deforestation-A study of GCM land surface parameterizations, Q. J R. Meteorol. Soc., 114,439-462. Doughty, C. E., and M. L. Goulden (2008), Seasonal patterns of tropical forest leaf area index and CO 2 exchange, J Geophys. Res., 113, GOOB06, doi: lO.l029/2007JG000590. Dufresne, 1.-L., P. Friedlingstein, M. Berthelot, L. Bopp, P. Ciais, L. Fairhead, H. Le Treut, and P. Monfi'ay (2002), On the magnitude of positive feedback between future climate change and the carbon cycle, Geophys. Res. Lett., 29(10), 1405, doi: 10.1029/ 200 lGLO 13777. Fan, S. M., S. C. Wofsy, P. S. Bakwin, D. 1. Jacob, and D. R. Fitzjan'ald (1990), Atmosphere-biosphere exchange of CO 2 and 0 3 in a central Amazon forest, J Geophys. Res., 95(DlO), 16,85116,864. Feeley, K. 1., et al. (2007), The role of gap phase processes in the biomass dynamics of tropical forests, Proc. R. Soc., Ser. B, 274(1627),2857-2864. Figuiera, A. M. S., S. D. Miller, C. A. D. de Sousa, M. C. Menton, A. R. Maia, H. R. da Rocha, and M. L. Goulden (2008), Effects of selective logging on h'opical forest tree growth, J Geophys. Res., 113, GOOB05, doi:10.1029/2007JG000577. Finnigan, J. 1. (1999), A cOlmnent on the paper by Lee (1998), On micrometeorological observations of surface-air exchange over tall vegetation, Agric. For. Meteorol., 97(1), 55-64. Fisher, 1. 1., et al. (2008), Clustered disturbances lead to bias in large-scale estimates based on forest sample plots, Ecol. Lett., 11(6),554-563. Fisher, R. A., et al. (2007), The response of an Eastern Amazonian rain forest to drought stress: Results and modelling analyses from a throughfall exclusion experiment, Global Change BioI., 13(11),2361-2378. Friedlingstein, P., et al. (2006), Climate-carbon cycle feedback analysis: Results from the (CMIP)-M-4 model intercomparison, J. Clim., 19(14), 3337-3353. Goulden, M. L., 1. W. Munger, S.-M. Fan, B. C. Daube, and S. C. Wofsy (1996), Measurements of carbon storage by long-term eddy correlation: Methods and a critical evaluation of accuracy, Global Change BioI., 2, 169-182. Goulden, M. L., S. D. Miller, H. R. da Rocha, M. C. Menton, H. C. Freitas, A. M. Figueira; and A. C. D. de Sousa (2004), Diel and seasonal patterns of tropical forest CO 2 exchange, Ecol. Appl., 14(4), S42-S54.
Y.l
405
Goulden, M. L., S. D. Miller, and H. R. da Rocha (2006), Nocturnal cold air drainage. and pooling in a tropical forest, J. Geophys. Res., 11 1, D08S04, doi: 10.1029/2005JD006037. Grace, 1., et al. (1995), Carbon-dioxide uptake by an undisturbed tropical rain-forest in southwest Amazonia, 1992 to 1993, Science, 270, 778-780. Gu, L. H., et al. (2005), Objective threshold determination for nighttime eddy flux filtering, Agric. For. Meteorol., 128(3-4), 179-197. Hasler, N., and R. Avissar (2007), What conh'ols evapotranspiration in the Amazon Basin, J Hydrometeorol., 8, 380-395. Houghton, R. A. (2003), Why are estimates of the terrestrial carbon balance so different?, Global Change BioI., 9(4), 500-509. Houghton, R. A., M. Gloor, J. Lloyd, and C. Potter (2009), The regional carbon budget, Geophys. Monogr. Ser., doi:1O.1029/ 2008GM000718, this volume. Hubbell, S. P., R. B. Foster, S. T. O'Brien, K. E. Harms, R. Condit, B.Wechsler, S. 1. Wright, and S. Loo de Lao (1999), Light-gap disturbances, recruitrnent limitation, and tree diversity in a N eotropical forest, Science, 283, 554-557. Huete, A. R., K. Didan, Y. E. Shimabukuro, P. Ratana, S. R. Saleska, L. R. Hutyra, W. Yang, R. R. Nemani, and R. Myneni (2006), Amazon rainforests green-up with sunlight in dly season, Geophys. Res. Lett., 33, L06405, doi:10.102912005GL025583. Hurtt, G. C., R. Dubayah, J. Drake, P. R. Moorcroft, S. W. Pacala, 1. B. Blair, and M. G. Fearon (2004), Beyond potential vegetation: Combining lidar remote sensing and a height-structured ecosystem model for imprbved estimates of carbon stocks and fluxes, Ecol. Appl., 14(3), 873-883. Hutyra, L. R., 1. w. Munger, s. R. Saleska, E. Gottlieb, B. C. Daube, A. L. Dunn; D. F. Amaral, P. B. de Camargo, and S. C. Wofsy (2007), Seasonal conh'ols on the exchange of carbon and water in an Amazonian rainforest, J Geophys. Res., 112, G03008, doi: 1O.1029/2006JG000365. Hutyra, L. R., 1. W. Munger, E. H. Pyle, S. R. Saleska, N. RestrepoCoupe, P. B. de Camargo, B. C. Daube, and S. C. Wofsy (2008), Resolving systematic errors in estimates of net ecosystem exchange ofCO 2 and ecosystem respiratioil in a tall-stature forest: Application to a tropical forest biome, Agric. For. Meteorol., 148, 1266-1279. Ichii, K., H. Hashimoto, M. A. White, C. Potter, L. R. Hutyra, A. R. Huete, R. B. Myneni, and R. R. Nemani (2007), Constraining rooting depths in tropical rainforests using satellite data and ecosystem modeling for accurate simulation of gross primaly production seasonality, Global Change BioI., 13, 67-77. Jones, C. D., M. Collins, P. M. Cox, and S. A. Spall (2001), The carbon cycle response to ENSO: A coupled climate-carbon cycle model study,J Clim., 14, 4113-4129. Juarez, R. 1. N., M. G. Hodnett, R. Fu, M. L. Goulden, and C. von Randow (2007), Control of dly season evapotranspiration over the Amazonian Forest as infelTed from observations at a Southem Amazon Forest site, J Clim., 20, 2827-2839. Jupp, T. E., and S. D. Twiss (2006), A physically motivated index of subgrid-scale pattern, J Geophys. Res., 111, 019112, doi: 10.1 029/2006JD007343. Keller, M., R. K. Varner, J. D. Dias, H. Silva, P. Crill, R. C. de Oliveira Jr., and G. P. Asner (2005), Soil-atmosphere exchange
406
SALESKA ET AI..
ECOSYSTEM CARBON FLUXES AND AMAZONIAN FOREST METABOLlSM
of nitrous oxide, nitric oxide, methane, and carbon dioxide in logged and undisturbed forest in the Tapajos National Forest, Brazil, Earth Interact., 9(23), EIl25, doi:lO.1175/EIl25.1. Kleidon, A., ~nd M. Heimann (2000), Assessing the role of deep rooted vegetation in the climate system with model simulations: Mechanism, comparison to observations and implications for Amazonian deforestation, Clim. Dyn., 16, 183-199. Kruijt, B., J. A. Elbers, C. von Randow, A. C. Araujo, P. J. Oliveira, A. Cui£, A. O. Manzi, A. D. Nobre, P. Kabat, and E. J. Moors (2004), The robustness of eddy correlation fluxes for Amazon rain forest conditions, Ecol. Appl., 14(sp4), 101-113. Lee, J. E., R. S. Oliviera, T. E. Dawson, and 1. Fung (2005), Root functioning modifies seasonal climate, Proc. Natl. Acad. Sci., V. S. A., 102(49),17,576-17,581. Lefsky, M. A., et al. (2005), Estimates offorest canopy height and aboveground biomass using ICESat, Geophys. Res. Lett., 32, L22S02, doi: 10.1 029/2005GL023971. Lloyd, J., and J. A. Taylor (1994), On the temperature dependence of soil respiration, Funct. Ecol., 8(3), 315-323. Lloyd, J., J. Grace, A. C. Miranda, P. Meir, S. C. Wong, B. S. Miranda, 1. R. Wright, J. H. C. Gash, and J. Mcintyre (1995), A simple calibrated model of Amazon rain-forest productivity based on leaf biochemical-properties, Plant Cell Environ., 18(10),1129-1145. Lloyd, J., et al. (2007), An airborne regional carbon balance for Central Amazonia, Biogeosciences, 4(5), 759-768. Malhi, Y., A. D. Nobre, J. Grace, B. Kmijt, M. G. P. Pereira, A. Culf, and S. Scott (1998), Carbon dioxide transfer over a Central Amazonian rain forest, J. Geophys. Res., 103, 31,59331,612. Malhi, Y., et al. (2004), The above-ground coarse wood productivity of 104 Neoh'opical forest plots, Global Change Bioi., 10(5), 563-591. Malhi, Y., I.. E. O. C. Aragao, D. Galbraith, C. Huntingford, R. Fisher, P. Zelazowskia, S. Sitch, C. McSweeney, and P. Meir (2009), Exploring the likelihood and mechanism of a climatechange induced dieback of the Amazon rainforest, Proc. Natl. Acad. Sci. V. S. A., doi:lO.1073/pnas.0804619106. Published online before print February 13,2009. Malhi, Y., S. Saatchi, C. Girardin, and I.. E. O. C. Aragao (2009), The production, storage, and flow of carbon in Amazonian forests, Geophys. Monogr. Ser., doi:lO.102912008GM000779, this volume. Marengo, J. A., C. A. Nobre, J. Tomasella, M. D. Oyama, G. S. de Oliveira, R. de Oliveira, H. Camargo, I.. M. Alves, and 1. F. Brown (2008), The drought of Amazonia in 2005, J. Clim., 21, 495-516. Martens, C. S., et al. (2004), Radon fluxes in tropical forest ecosystems of Brazilian Amazonia: Night-time CO 2 net ecosystem exchange derived from radon and eddy covariance methods, Global Change Bioi., 10,618-629. McDowell, N. G., et al. (2008), Tansley Review: Mechanisms of plant survival and mortality during drought: Why do some plants survive while others succumb?, New Phytol., 178, 719-739. Meir, P., et al. (2009), The effects of drought on Amazonian rain forests, Geophys. Monogr. Sa., doi:IO.l029/2008GM000882, this volume.
Metcalfe, D. J., et al. (2007), Factors controlling spatio-temporal variation in carbon dioxide efflux from surface litter, roots, and soil organic matter at four rain forest sites in the eastern Amazon, J. Geophys. Res., 112, G04001, doi: 10.1029/2007JG000443. Miller, S. D., M. I.. Goulden, M. C. Menton, H. R. da Rocha, H. C. de Freitas, A. M. E Silva Figueira, and C. A. D. de Sousa (2004), Biometric and micrometeorological measurements of tropical forest carbon balance, Ecol. Appl., 14(4 Suppl.), Sl14-S126. Moorcroft, P. R. (2003), Recent advances in ecosystem-atmosphere interactions: An ecological perspective, Proc. R. Soc. London, Sec. B, 270, 1215-1227. Moorcroft, P. R., G. C. Hurtt, and S. W. Pacala (2001), A method for scaling vegetation dynamics: The ecosystem demography model (ED), Ecol. Monogr., 71(4),557-585. Myneni, R. B., et al. (2007), Large seasonal swings in leaf area of Amazon forests, Proc. Natl. Acad. Sci. V. S. A., 104,4820-4823. Nelson, B. W., et al. (1994), Forest disturbance by large blowdowns in the Brazilian Amazon, Ecology, 75, 853-858. Nepstad, D. C., et al. (1994), The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures, Nature, 372, 666-669. Nepstad, D. C., I. M. Tohver, D. Ray, P. Montinho, and G. Cardinot (2007), Mortality of large trees and lianas following experimental drought in an Amazon forest, Ecology, 88(9), 2259-2269. New, M., M. Hulme, and P. Jones (2000), Representing twentiethcentury space-time climate variability. Part II: Development of 1901-1996 monthly grids of terrestrial surface climate, J. Clim., 13(13),2217-2238. Nobre, A. D., I.. A. Cuartas, C. Rellllo, J. Tomasella, M. Hodnett, M. Waterloo, and J. V. Soares (2008), Revealing hidden terra finne rainforest enviromnents in Amazonia, Amazon in Perspective: Integrated Science/or a Sustainable Future, LBA/GEOMAlPPBio Conference, Manaus, November 17-20, 2008. (Available at http://www.Ibaconferencia.org/lbaconf_2008/eng). Nobre, C. A., P. J. Sellers, and J. Shukla (1991), Amazonian deforestation and regional climate change, J. Clim., 4, 957-988. Nobre, P., and J. Shukla (1996), Variations of sea surface temperature, wind stress, and rainfall over the Tropical Atlantic and South America, J. Clim., 9, 2464-2479. Odum, E. P. (1969), The strategy of ecosystem development, Science, 164, 262-270. Oliveira, P. J., E. J. P. da Rocha, G. A. Fisch, B. Kruijt, and J. B. M. Ribeiro'(2004), Efeitos de um evento de friagemnas condiyoes meteorol6gicas na Amazonia: Um estudo de caso, Acta Amazonica, 34(4), 613-619. Oliveira, R. S., T. E. Dawson, S. S. O. Burgess, and D. C. Nepstad (2005), Hydraulic redistribution in three Amazonian trees, Oecologia, 145, 354-363. Ometto, J. P. H. B., A. D. Nobre, H. R. da Rocha, P. Artaxo, and I.. A. Martinelli (2005), Amazonia and the modern carbon cycle: Lessons learned, Oecologia, 143(4),483-500. Phillips, O. 1.., et al. (1998), Changes in the carbon balance of h'opical forests: Evidence from long-tenn plots, Science, 282, 439-442. Phillips, O. 1.., et al. (2009), Drought sensitivity of the Amazon rainforest, Science, 323, 1344-1347.
Phillips, O. 1.., N. Higuchi, S. Vieira, T. R. Baker, K.-J. Chao, and S. I.. Lewis (2009), Chimges in Amazonian forest biomass, dynamics, and composi,tron, 1980-2002, Geophys. Monogr. Sa., doi: 10.1 029/2008G}\t1000739, this volume. Potter, C. S., E. A/Davidson, S. A. Klooster, D. C. Nepstad, G. H. de Negreir6s, and V. Brooks (1998), Regional application of an ecosystem production model for studies of biogeochemistry in Brazilian AlTlazonia, Global Change Bioi., 4(3), 315-334. Priante-Filho, N., et al. (2004), Comparison ofthe mass and energy exchange of a pasture and a mature transitional tropical forest of the southern Amazon Basin during a seasonal transition, Global Change Bioi., 10, 863-876. Pyle, E. H., et al. (2008), Dynamics of carbon, biomass, and structure in two Amazonian forests, J. Geophys. Res., 113, GOOB08, doi: 10.1029/2007JG000592. Raich, J. W., and W. H. Schlesinger (1992), The global carbon dioxide flux in soil respiration and its relationship to vegetation and climate, Tellus, Ser. B, 44(2), 81-99. Renno, C. D., et al. (2008), HAND, a new terrain descriptor using SRTM-DEM: Mapping terra-finne rainforest environments in Amazonia, Remote. Sens. Environ., 112(9), 3469-3481. Rice, A. H., E. Hammond-Pyle, S. R. Saleska, I.. Hutyra, M. Palace, M. Keller, P. B. de Cannargo, K Portilho, D. Marques, and S. C. Wofsy (2004), Carbon balance and vegetation dynamics in an old-growth Amazonian forest, Ecol. Appl., 14(4), S55-S71. Roberts, D. A., M. Keller, and J. V. Soares (2003), Studies of landcover, land-use, and biophysical properties of vegetation in the Large Scale Biosphere Atmosphere experiment in Amazonia, Remote Sens. Environ., 87, 377-388. Saatchi, S. S., R. A. Houghton, R. C. D. S. Alvala, et al. (2007), Distribution of aboveground live biomass in the Amazon basin, Global Change Bioi., 13(4), 816-837. Sakai, R. K, D. R. Fitzjarrald, O. I.. L. Moraes, R. M. Staebler, O. C. Acevedo, M. J. Czikowsky, R. da Silva, E. Brait, and V. Miranda (2004), Land-use change effects on local energy, water and carbon balances in an Amazonian agricultural field, Global Change Bioi., 10(5), 895-907. Saleska, S. R., et al. (2003), Carbon in Amazon forests: Unexpected seasonal fluxes and disturbance-induced losses, Science, 302, 1554-1557. Saleska, S. R., K Didan, A. R. Huete, and H. R. da Rocha (2007), Amazon forests green-up during 2005 drought, Science, 318, 612, doi:IO.l 126/science.l 146663. ShuttlewOlih, W. J. (1988), Evaporation from Amazonian rain forest, Proc. R. Soc. London, Ser. B, 233,321-346. Shuttleworth, W. J., et al. (1984), Eddy-correlation measurements of energy partition for Amazonian forest, Q. J. R. Meteorol. Soc., 110(466), 1143-1162. Silver, W. 1.., J. Neff, M. McGroddy, E. Veldkamp, M. Keller, and R. Cosme (2000), Ecosystems, 3, 193-209. Sitch, S., et al. (2008), Evaluation of the terrestrial carbon cycle, fut\.lre plant geography and climate-carbon cycle feedbacks using five Dynamic Global Vegetation Models (DGVMs), Global Chqnge BioI., 14(9),2015-2039.
407
Sotta, E. D., P. Meir, Y. Malhi, A. D. Nobre, M. Hodnett, and J. Grace (2004), Soil CO 2 efflux in a tropical forest in the central Amazon, Global Change BioI., 10,601-617. Sotta, E. D., et al. (2007), Effects of an induced drought on soil carbon dioxide (C0 2) efflux and soil CO 2 production in an Eastern Amazonian rainforest, Brazil, Global Change BioI., 13(10), 2218-2229. Souza-Filho, J. D. C., A. Ribeiro, M. H. Costa, and J. C. P. Cohen (2005), Control mechanisms of the seasonal variation of transpiration in a northeast Amazonian tropical rainforest (in Portuguese), Acta AlIlazonica, 35(2), 235-241. Staebler, R. M., and D. R. Fitzjarrald (2004), Observing subcanopy CO 2 advection, Agric. For. Meteorol., 122(3-4),139-156. Stephens, B. B., et al. (2007), Weak northern and strong tropical land carbon uptake from vertical profiles of atmospheric CO 2, Science, 316(5832), 1732-1735. Tian H., J. M. Melillo, D. W. Kicklighter, A. D. McGuire, J. V. K Helfrich, B. Moore, and C. J. Vorosmarty (1998), Effect of interannual climate variability on carbon storage in Amazonian Ecosystems, Nature, 396, 664-667. T6ta, J., D. R. Fitzjarrald, R. M. Staebler, R. K. Sakai, O. M. M. Moraes, O. C. Acevedo, S. C. Wofsy, and A. O. Manzi (2008), Amazon rain forest subcanopy flow and the carbon budget: Santarem LBA-ECO site, J. Geophys. Res., 113, GOOB02, doi: 1O.l029/2007JG000597. von Randow, C., et al. (2004), Comparative measurements and seasonal variations ~n energy and carbon exchange over forest and pasture in South West Amazonia, Theor. Appl. Clilllatol., 78(1-3),5-26. Vourlitis, G. 1.., N. Pri~nte-Filho, M. M. S. Hayashi, J. S. Nogueira, F. T. Caseiro, and J. H. Campelo Jr. (2001), Seasonal variations in the net ecosysteni CO 2 exchange of a mature Amazonian transitional tropical forest (cerradao), Funct. Ecol., 15, 388-395. Vourlitis, G. 1.., J. S. Nogueira, N. P. Filho, W. Hoeger, F. Raiter, M. S. Biudes, J. C. Arruda, V. B. Capistrano, J. L. Brito, and F. A. Lobo (2005), The sensitivity of diel CO 2 and H2 0 vapor exchange of a tropical transitional forest to seasonal variation in meteorology and water availability, Earth Interact., 9(27), 1-23. Werth, D., and R. Avissar (2002), The local and global effects of Amazon deforestation, J. Geophys. Res., 107(D20), 8087, doi: 10.1029/200 Im000717. Wright, S. J., and C. P. van Schaik (1994), Light and the phenology of tropical trees,Am. Nat., 143(1),192-199. H. da Rocha, Departamento de Ciencias Atmosfericas, lAG, Universidade de Sao Paulo, Rua do Matao, 1226 Cidade Universitaria, Sao Paulo, SP CEP 05508-090, Brazil. B. Kruijt, Alterra, Wageningen University and Research Centre, P.O. Box 47, NL-6700 AA Wageningen, Netherlands. A. D. Nobre, Escritorio Regional do INPE, National Institute for Amazonian Research, Sao Jose dos Campos, Brazil. S" Saleska, PIRE Program in Amazon-Climate Interactions, University of Arizona, 1041 East Lowell Street, BioSciences West, Room 510, Tucson, AZ 85721, USA. ([email protected])
The Regional Carbon Budget R. A. Houghton Woods Hole Research Center, Falmouth, Massachusetts, USA
Manuel Gloor and Jon Lloyd School a/Geography, University a/Leeds, Leeds, UK
Christopher Potter Ecosystem Science and Technology Branch, NASA Ames Research Center Moffett Field, California, USA
A number of approaches have been used to infer whether Amazonia is a net source or sink for carbon. Top-down approaches based on inverse calculations with CO 2 concentrations and atmospheric transport models are, problematic because of a paucity of air samples and poor constraints on regional air transport. Direct measurements ofchanges in aboveground biomass suggest a ):let carbon sink in oldgrowth forests but remain controversial. Direct measurements of CO 2 flux with the eddy covariance technique indicate forests to be both sources and sinles of carbon, depending in part on when the last disturbance occurred. These flux measurements may be extrapolated through time and space with ecosystem models based on physiological processes, but many models fail to reproduce even the conect sign of carbon balance observed seasonally in some forests. Models based on changes in forest stmcture, driven by both anthropogenic (e.g., deforestation for pasture) and natural (e.g., fire) disturbances and recovery, consistently calculate net carbon emissions, emissions that may be offset by the increased biomass observed in longterm plots in old-growth forests. Aquatic systems are nearly neutral with respect to carbon, with inputs from seasonally flooded forests and grasslands accounting for the measured efflux. Taken together, these different approaches, which often consider different components of the region's carbon cycle, suggest that Amazonia has been, on average, nearly neuh'al with respect to carbon over the last decade, albeit a small net source during El Nino events.
1. INTRODUCTION Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10. 102912008GM0007 18
This chapter is concemed with two questions: What is the carbon balance of Amazonia, and why (i.e., What are the processes responsible for sources and sinks of carbon in the region)? The processes fall into two broad categories. The 409
410
HOUGHTON ET AL.
REGIONAL CARBON BUDGET
first category includes metabolic or physiological processes, should be possible to infer surface flux strength, location, photosynthesis, respiration, decomposition, water relations, and time course. Atmospheric transport can be simulated processes that respond to environmental drivers, such as fairly realistically by numerical integration of the transport light, moistul'e, temperature, C02 concentrations, and nutri- equation with data derived from regular worldwide observaents. The second broad category of processes includes dis- tions of the state of the atmosphere (e.g., winds, air masses). turbances and recovery, including both direct anthropogenic Nonetheless, these models are not perfect. Particularly effects (e.g., deforestation for pasture) and natural or indi- problematic is the representation of processes operating at smaller spatial scales than the scale used for the discretizarect anthl'opogenic effects (e.g., fire). In addressing whether Amazonia is a source or sink for car- tion of the transport equation [e.g., Peters et al., 2004; Gloor bon, the chapter looks first at results from two atmospheric- et al., 2007]. Atmospheric concentrations of CO 2 are currently meabased approaches: (1) inverse calculations based on spatial and temporal variations in atmospheric CO 2 concentrations sured at approximately 50 stations operated by NOAA's and models of atmospheric transport and (2) atmospheric CMD and at a smaller number of other measurement netair column carbon budgets based on vertical profiles of CO 2 works run by CSIRO Division (Australia), University of over the region. The results of these approaches are ambigu- Heidelberg (Germany), (University of Tokyo, National Inous with respect to whether Amazonia is a net source or sink stitute for Environmental Studies (NIES) (Tsukuba, Japan), Laboratoire des Sceinces du Climat et de l'Environnement for carbon. Subsequent sections consider ground-based measure- (France), and Max-Planck Institute for Biogeochemistty ments: measurements of CO 2 flux by eddy covariance, long- (Germany). Until the end of the year 2000, these regular term measurements of forest biomass on pennanent plots, measurements were predominantly in remote locations at and measurement of carbon dynamics in aquatic systems. the earih's surface. The station locations were chosen to Although the results of eddy covariance measurements are avoid large and rapidly varying fluxes, such as those due difficult to extrapolate over large regions, they are useful to photosynthesis and respiration on land and those due to for constructing process-based physiological models that, fossil fuel burning and cement manufacture. Another tradiin turn, are used to estimate sources and sinles of carbon in tionaI limitation of atmospheric concentration sampling has response to environmental variables. Models are also used been the lack of information in the vertical dimension, which to calculate the sources and sinks of carbon that result from is not only necessmy to characterize the CO 2 field but also essential to validate and calibrate model transpori. In recogdisturbances and recovery. In the last section, we bring together the results of the ob- nition of these shoricomings, dense vertical profile and tall servations and models to answer the two questions posed: tower continuous sampling have been initiated in the last few What is the net carbon balance? and What are the mecha- years over the United States, Europe, Japan, Russia, and, to nisms responsible? More specifically, we seek to identify a limited extent, over South America. As these data are very and quantify the sources and sinks of carbon attributable to recent, their implications are only now being investigated, as metabolic responses to environmental change, and those at- discussed below. While enormous progress has been made in expanding the number of air sampling stations, important tributable to disturbance and recovery processes. gaps in the network remain, most notably in tropical land regions where sampling is still very sparse. 2. CARBON BALANCE OF AMAZONIAN REGIONS: Generally, the weaknesses in the approach of atmospheric ESTIMATES FROM THE ATMOSPHERE inversions explaining the range in estimates are the following: (1) Flux estimates tend to be highly sensitive to biases 2.1. Global Estimates From Inverse Calculations: What Do and uncertainties in data and modeled transport. (2) The atAtmospheric Concentration Data Tell Us About the Carbon mosphere has been and still is under-sampled. (3) Data to Balance ofAmazonia? calibrate transport in models have been and still are sparse. There have been numerous shldies in which large-scale (4) The covariation oftransport and surface fluxes (the "reccarbon fluxes are inferred from atmospheric concentration tifier effect") [Denning et at., 1995] is a major uncertainty in data and inverse modeling of atmospheric transport, but the these analyses. Because tropical data are very sparse and thus resolution underlying inversion methods vmy substantially, and the results are conflicting. The basic principle of the approach is of different land regions in the tropics is hardly possible, to infer surface fluxes from the accumulation or depletion we focus primarily on estimates of flux for tropical land of CO 2 in the air above the region of interest. If air motion as a whole instead of the Amazon Basin (or tropical South acting on the CO 2 concentration field can be detennined, it America).
Early shldies by Keeling et al. [1989] and Tans et al. [1990] came to the conclusion that there is a substantial sink in Northern Hemispl1ere midlatitudes. Keeling et al. [1989] attributed the sinl~/to the oceans, while Tans etal. [1990] attributed it to land)Independent estimates ofair'-sea flux at that time were quit: uncertain. Subsequent shldies were inconclusive [Fan et al., 1998; Bousquet et al., 2000], and a community effort was initiated to assess the model dependence of inverse-based flux estimates. This effort (Atmospheric Tracer Transport Model Intercomparison Project (TRANSCOM3» resulted in an often cited publication [Gurney et al., 2002]. Followirrg the Fan et al. 1998, 1999 studies, air-sea flux priors (defined below) were based on compilations of partial pressure differences and a bulk gas exchange formulation [Takahashi et al., 1999]. While this was probably the best that could be done at the time, global bulk exchange parameterizations are inherently uncertain. The use of priors, or prior information, is one approach for reducing uncertainties in the approach. As long as prior guesses are based on data and uncertainty variance-covariance can be rationally estimated, this approach is reasonable. However, often, regularization using prior estimates has not followed these principles. ' Since the shldy of Gurney et al. [2002], the field has seen several advances. One was the estimation of air-sea gas fluxes and patterns based on ocean interior data and models [Gloor et al., 2003; Mikaloff-Fletcher et al., 2007]. A second initiated atmospheric sampling of the troposphere up to 8 km height over continents, mainly in Norihern Hemisphere midlatitudes (Norih American Carbon Plan, Aerocarb, NIES Tsukuba). Furthermore, in recognition of the rectification problem (see point 4 above) and because of interest in the interannual variation in the atmospheric growth rate, inverse modeling shldies started to resolve fluxes on monthly and shorier timescales [Rayner et al., 1999; Bousquet et al., 2000; Rodenbeck et al., 2003; Baker et al., 2006] (TRANSCOM3 level 2 intercomparison Shldy). The next advance in methodology was the analysis by Jacobson et al. [2007], which combined ocean interior and atmospheric data in a coupled annual mean inversion using the entire 12-model suite of atmospheric inverse models from TRANSCOM3. The final study we refer to here is that of Stephens et al. [2007], which assessed the implications of vertical profile measurements over the continents (predominantly Northern Hemisphere) using the TRANSCOM3 level 2 simulations and results. Based on these new data, Stephens et al. [2007] were able to select a subset of transport models from the' TRANSCOM3 level 2 model suite that simulated existing CO 2 vertical profiles most closely. The fluxes estimated by the subset of selected models differs from the TRANSCOM3 level 2 mean (Table 1). In our compilation (Table 1), we also
411
report the TRANSCOM3 level 1 results for the subgroup of models selected by the Stephens et al. shldy, results which achlally shed a somewhat different light on the published results of Stephens et al. [2007]. Before discussing the carbon balance estimates, themselves, we note that there is substantial interannual variation of fluxes between tropical lands and the atmosphere (Figure 1). It is noteworihy that South America seems to dominate the global interannual variability in the atmospheric growth rate, at least according to the Rodenbeck et al. [2003] study. Furthermore, these interannual flux "anomalies" are related to climate anomalies associated with El Nino-Southern Oscillation (ENSO), particularly over the Amazon Basin. Independent evidence for large variations in the fluxes from tropical land come from atmospheric CO, CH4, and H2 data [e.g., Langenfelds et at., 2002]. The large interannual variability of these fluxes, closely associated with the ENSO, implies that an adequate budgeting period should include several El NinolLa Nina phases. Unfortunately, this last requirement is only pariially fulfilled by the inversion studies discussed here (Table 1). All of these studies cover the period from JanualY 1992 to December 1996 (the Shldy of Rodenbeck et al. [2003] also covers the period from Janumy 1996 to December 1999). As shown in Figure 1, the pt)riod contains both an El Nino and a La Nina phase. Thus,' the estimates are likely not to be dominated by either oqhe two but rather provide a "climatologi; cal" mean. According to the study of Gurney et at. [2002], based on 12 different models and annual means, there is a substantial Norihern Hemisphere land sink, while tropical lands, including the Amazon Basin, release significant amounts of carbon to the atmosphere. The study of Jacobson et at. [2007] finds generally similar results, a strong Northern Hemisphere sink, and a statistically significant source fi'om h'opical and Southern Hemisphere land combined. This is not surprising as both studies used the same suite of 12 transpori models from TRANSCOM3 level 1 (there are three TRANSCOM3 levels; level 1 calculations use annual mean data to estimate annual mean fluxes; level 2 calculations use monthly data to estimate monthly fluxes; level 3 calculations are, except for the fixed time period, "open protocol," i.e., the inversion method and data selection are selected by the investigator). The flux from the study of Jacobson et al. [2007] may be considered to have more weight, however, as more realistic air-sea fluxes were used and no priors on land fluxes were employed. The other shldies come to somewhat different conclusions. The analysis by Stephens et al. [2007] described above, reduced the 12-model TRANSCOM3 suite to the 3 models that seem to exhibit the least transport biases. This model selection implies a slightly reduced source in the tropics and a
412
REGIONAL CARBON BUDGET
HOUGHTON ET AL.
Table 1. Breakup of Land-Atmosphere Flux Estimates Into Three Zonal Bands From a Range ofAtmospheric Transport Inversion Studies" Gurney et al. Jacobson et al. [2002] [2007]
1992-1996
S Hem «20S) Tropics N Hem (>20N)
S Hem «20S) Tropics NHem(>20N)
Total
1992-1996
12 Models T3Ll
12 Models T3Ll
-0.2 ± 1.1 (0.15) 1.1±1.3 (1.5) -2.3 ± 0.6 (-0.7) -1.4
-2.4 ± 2.0
-O.8±0.7 (0.0) 0.4 ± 0.7 (0.0) -1.1 ± 0.5 (0.0)
-1.4 ± 0.1
Rodenbeck et al. [2003]
1992-1996
Atmosphere Land Flux O.O± 0.20 0.1 ± 0.2 0
4.2 ± 2.7
-1.0 ± 0.4 0
-2.9 ± 1.0
-0.7 ± 0.2 0
-1.1
1996-1999
Transport Model TM3 TM3
-0.8 ± 0.4 (0.3) -0.4 ± 1.00
-1.8 -1.3 Atmosphere Ocean Flux -1.0 ± 0.1 0 -1.2 ± 0.20
Baker et al. [2006]
Stephens et al. [2007]
1992-1996
1992-1996
12 Models {TM3, UCI, JMA} T3Ll b T3L2 c T3L2
TM3 T3L2 c
-1.2
0.1 ± 1.1
1.6
0.7 ± 1.4
-0.1 ± 0.8
1.0 (0.2)
-2.7
-2.2± 0.6
-1.5 ± 0.6
-2.2 (0.8)
-2.3
-1.4
-0.7
-1.1 ± 0.7
0.8 ± 0.1
0.9 ± 0.2 0
1.1 ± 0.2 0
0.7
0.4 ± 0.7
-1.1 ± 0.1
-1.6 ± 0.1 0
-1.7 ± 0.1 0
-1.0
-0.7 ±0.6
-1.5
-1.7
-1.7
-1.8
-1.0
-1.4
-2.9
-2.8
-3.5
-3.3
-3.3
-2.9
"Values are given in Pg C a-I. The sign convention is that a positive flux is directed to the atmosphere. Numbers in brackets are a priori prescribed fluxes used in those inversions which use Bayesian priors to regularize the calculations. Empty parentheses indicate that prior fluxes have been used but cannot be inferred from the publication in question. TM3, UCI, and JMA stand for specific tracer transport models [see Gurney et al., 2002], T3Ll stands for the TRANSCOM level 1 experiment results (based on annual mean data), and T3L2 stands for TRANSCOM level 2 results (based on monthly mean data). Stephens et al. [2007] determined a subset oftransport models most compatible with recent vertical profile data not included in TRANSCOM inversions. We report on the results from this subset as well as for the transport model TM3 alone and with both T3Ll and T3L2. bCalculated by Gloor from Gurney et al. [2003]. cSouthem Hemisphere fluxes are missing because Stephens et al. [2007] report only land flux estimates for Tropics and Northem Hemisphere (see p. 1734).
reduced northern midlatitude sink when focusing on annual mean inversions (TRANSCOM3 level 1). In contrast, when applying the same selection criterion to the TRANSCOM3 level 2 model suite (which resolve fluxes monthly instead of annually), tropical lands were carbon neutral with the Northern Hemisphere land sink strongly reduced. This result is qualitatively similar to that found by the other inversion study that used monthly inversions [Rodenbeck et al., 2003], except the latter suggested a larger tropical land sink. It thus seems that results from annual mean inversions differ from inversions that solve for monthly fluxes with or without model selection based on new vertical profile data. One reason is likely the role played by the rectification term applied in annual mean inversions (while not applied in monthly inversions). In contrast, the difference between
results based on monthly versus annual mean inversions does not seem to be so much related to the specific transport model used. Annual inversions based on the transpOlt model TM3 differ markedly from the monthly mean inversion results based on the same model. In summaty, the balance of Northern Hemisphere versus tropical land sinks, and thus the Amazon carbon balance provided by the atmospheric concentration constraint, remains inconclusive. The recent new vertical profile data permit not only traditional inverse modeling but also a much simpler direct approach which makes more direct use of the main constraint provided by the atmosphere: the accumulation of constituents within an air volume above a surface source. This more direct approach simply balances inflow and outflow of air into the total air volume above the region in question
Tropical South America (200 S to 20° N)
413
Kuck et al., 2000]; by considering diurnal differences in the shape and integrated values of vertical CO 2 profiles fi'om airborne measurements, even when made on different days ...----I..... [Chou et al., 2002]; and by undertaking planned sequences >. of flights and analysing the vertical CO 2 profiles obtained in oOl conjunction with model-derived estimates of other important (L -- OH-;----,-------+-~----'- parameters known to influence the derived surface fluxes, >< :J such as the net vertical velocity of the air masses OCCUlTing IT: at the times of measurement and a consideration of potential "0 -1 c advective effects [Lloyd et aI., 2007]. ro .....I In all cases, analyses have been limited to periods of days .8 -2 or months with the validity of any conclusions also unclear ..... because of methodological considerations and a requirement to make assumptions of unknown validity. For example, 1~82 1986 1990 1994 1998 2002 Chou et al. [2002] interpreted both the overall magnitude Year AD and diurnal pattern of their CO 2 balance calculations fi'om a Figure 1. Estimate of fluxes to and from tropical South America reanalysis of historical aircraft measurements over Amazoto the atmosphere, based on atmospheric data and inverse model- nia as indicating a significant influence of net CO 2 emissions ing by Rodenbeck et al. [2003]. The different shadings correspond from wetlands, rivers, and inundated forest. Nevertheless, to estimates based on different station networks with records that that conclusion required the assumption that a uniform temcover the entire period for which fluxes are being estimated. poral pattern of daytime convection existed so as not to bias their interpretation of the diurnal changes in many different vertical CO 2 profiles observed over many different days. Nevertheless, convective activity is generally at its greatest in the afternooqJevening [Machado et al., 2002, 2004], [Crevoisier et al., 2006]. The method makes use of the flux- thereby often tending to "reset" profiles [Lloyd et al., 1996] integrating property of the atmosphere and is insensitive to allowing for a new pycle in atmospheric [C0 2] to commence vertical transport biases and rectification, unlike traditional from a different stait point to before the storm. Once a coninverse transport modeling. It has recently been applied to vective event has occurred (more or less dissipating any verNOlth America (Crevoisier et aL, Robust carbon balance of tical gradient in CO 2), the effects of that convective event contiguous North America 2004-2006 based on novel at- should continue to be observed in all CO 2 profiles observed mospheric data and methods, in preparation) as part of the for the same air mass over the remainder of the day (or until NOlth American Carbon Plan. In our view, this provides the next convective event). evidence that the flux integrating property ofthe atmosphere Likewise, the study of Kuck et al. [2000] ignored any poscan be successfully exploited. However, this is only hue if sible vertical velocity effects, and even when such estimates adequate vertical profile data are available. Some 5-10 regu- are made, these inevitably come from model output rather lar vertical profile sites across Amazonia, complemented by than direct measurements, with any errors in the values used surface reference stations along the western coastline, would having potentially large effects on the fluxes so derived be needed. [Lloyd et al., 2001, 2007]. Nevertheless, even after such errors in derived fluxes are 2.2. Regional Estimates taken into account, the airborne budgeting approach can prove useful, for example, in checking the validity of towerFrom repeated measurements of CO 2 concenh'ations based estimates of ecosystem carbon balances; the study of within and above the atmospheric boundary layer, it is pos- Lloyd et al. [2007], for example, showed that a significant sible to derive the net surface fluxes of reasonably large ar- underestimation of nighttime effluxes of carbon using the eas of order 1000 km2 [Wofty et al., 1988; Raupach et al., . eddy covariance technique was occurring for two towers lo1992; Raupach and Finnigan, 1995; Kuck et al., 2000; Lloyd cated above evergreen tropical forest near Manaus (see secet al., 2001, 2007; Laubach and Fristch, 2002]. Variants of tion 3.2). Kuhn et al. [2007] also applied aircraft data to help this technique have now been applied on four occasions to evaluate regional estimates of surface isoprene fluxes and obtain estimates of the regional carbon balance for Ama- associated atmospheric chemistry reaction parameters in the zonia: by using tethered balloon profiles [Culf et al., 1999; surface boundary layer.
«
414
REGIONAL CARBON BUDGET
HOUGHTON ET AL.
3. CARBON BALANCE OF INTACT FORESTS Several estimates of the carbon balance of Amazonian forest and dhrado have been made over the last 15 or so years with two primary methods employed: continuous and direct measurements of carbon fluxes using the eddy covariance technique [Grace et al., 1995a, 1995b, 1996; Miranda et al., 1996; Malhi et al., 1998; Carswell et al. 2002; Santos et al., 2003; Saleska et al., 2003; Hutyra et al., 2007] and estimates of rate of net aboveground biomass change from repeated censuses of permanent sample plots [Phillips et al., 1998; Baker et al., 2004; Lervis et al., 2006]. However, these two approaches to determining the net carbon balance of individual forest or savanna stands, both of which involve the sampling of an area of approximately 1 ha, have proved controversial. Concerns have been raised with respect to methodological issues [Clark, 2002; Saleska et al., 2003; Wright, 2005] and with the validity of extrapolating results of such shldies to estimate the carbon balance of the Amazon Basin as a whole [Saleska et al., 2003; Chambers and Silver, 2004; Chambers et al., 2004; Wright, 2005]. Central to the latter criticism is the notion that all forests studied are recovering either from some unspecific (and thus hypothetical) small scale but regular disturbance [Chambers et al., 2004] or from severe, widespread mortality events [Wright, 2005], such as may have occurred in the great Amazon drought of 1926 [Williams et al., 2005], or earlier [Keller et al., 1996]. 3.1. Quantification ofStand Level Carbon Balances
The net carbon balance of any given area of vegetation and soil can be written as
-dCp + -dCs dt
dt
= - NE = Gp - RE,
(1)
where CE is the ecosystem carbon density (typically Mg C ha- I or mol C m-2), Cp is the density of the live plant carbon pool, Cs is the density of the soil carbon pool which also includes the dead plant litter pool (including coarse woody debris), and NE is the rate of net ecosystem exchange, equal to the ecosystem (plant + soil) respiration rate (R E) less the rate of net CO 2 assimilation through photosynthetic processes, the latter often being referred to as gross primary productivity, Gp , or OPP. The sign of change in carbon density reflects the atmospheric convention, where emissions are positive and a net carbon uptake by vegetation is negative. The rate of change in plant carbon can be expressed as dCp dt
-
= Np-Lp
(2) '
where N p is the net primary productivity (NPP) of the ecosystem (defined as the rate of new plant growth), and Lp represents the losses from the live plant carbon pool, being equivalent to the sum ofthe rates oflitterfall, herbivory, and tree mortality. Likewise for soil carbon (3)
where RH is the rate of heterotrophic respiration. The equivalences in equation 1 arise because R E = Rp + R H with Rp being the rate of autotrophic (plant) respiration and with Np = Gp - R p • That is to say,
dCp dCs += Np - RH = Gp - Rp - RH = Gp - RE = - NE. dt dt (4)
-
In terms of the different methodologies applied, towerbased eddy covariance measurements attempt to determine stand level carbon balances by measuring NE continuously over a period of 1 year or more [Saleska et al., this volume]. On the other hand, biomass inventOly measurements provide a direct estimate of dCp/dt or, more precisely, the rate of change in the carbon stocks of aboveground woody biomass between two or more census dates which are typically separated by a period of 5 to 10 years [Phillips et al., this volume]. Thus, any long-term changes in leaf and root carbon stocks, as well as in soil and litter carbon stocks, are not usually taken into account using biomass inventmy approaches. 3.2. Tower-Based Eddy Covariance Approach Grace et al. [1995a] used a combination of direct eddy covariance measurement [Grace et al., 1995b, 1996] and modelling approach [Lloyd et al., 1995] to estimate the carbon balance of semievergreen tropical forest in southwest Amazonia, inferring a net carbon sink (dCE/dt) of about 1.0 Mg ha- 1 a-I. Although consistent with what might be theoretically expected if increasing atmospheric [C0 2] were stimulating growth of tropical forest trees [Taylor and Lloyd, 1992; Lloyd and Farquhar, 1996], the suggestion that old growth rain forests should not be steady state systems proved immediately controversial [Keller et al., 1996], with other authors arguing that both short- and long-term disturbance regimes needed to be taken into account [Clark, 2004, 2007; Chambers et al., 2004] (discussed flUiher in section 3.3). Nevertheless, subsequent work by the same group in an evergreen
415
tropical forest near Manaus suggested an even larger carbon about 0.1 of tropical forest Gp [less than 0.3 of Np in (2)J sink of 5.7 Mg ha~~/a-l [Malhi et al., 1998], which led Malhi ends up being allocated to wood production [Malhi et al., and Grace [2000l to suggest that tropical forests might be 1998; Saleska et aI., 2003; Aragao et al., 2009]. sequestering a~inuch as 2.0 Pg C a-Ion a global scale, effecIt thus seems to us that there is no single reliable or scientively accou9,ring for the bulk of the terrestrial carbon sink. tifically defensible diagnostic to indicate the correct magniSeveral subsequent studies above Amazonian forests have tude of any "corrections" to be applied under conditions of also inferred them to be substantial carbon sin1es [Carswell low turbulence in eddy covariance shldies, and we are thus et al., 2002; A;mijo et al., 2002], although recently dishlrbed left to conclude that stand level carbon balances estimated by forests may not be [Keller and Crill, 2000; Saleska et al., the eddy covariance method, whether quoted as being nega2003]. tive or positive, are not resolved with sufficient accuracy to It has become increasingly apparent, however, that the eddy allow meaningful estimates of dCE/dt to be made. Daytime covariance technique yields what seem to be erroneously CO 2 flux may also be biased because the eddy covariance low estimates of ecosystem carbon dioxide efflux rates on technique does not usually close the energy budget during calm nights, even when changes in the amount of CO 2 stored daytime [Wilson et al., 2002; Aranibar et al., 2006]. Taken within the canopy space are taken into account [Goulden et al., in conjunction with the need to consider other carbon fluxes 1996]. Thus, below a certain turbulence threshold, it is not measured by the technique, such as those associated now more or less common practice to substitute observed with emissions of volatile organic compounds [Kesselmeier data with data simulated from a model calibrated with et al., this volumeJ, fluxes of dissolved organic (DOC) and measurements made under conditions of stronger vertical inorganic (DIC) carbon into riverine systems through the mixing [Loescher et al., 2005]. Nevertheless, estimated soil system [Richey et al., this volumeJ, and nochlrnal cold annual carbon balances can be drastically altered depend- air drainage [Goulden et al., 2006; Armijo et al., 2007], the ing on the apparent turbulence threshold us~d [Miller et main potential of the eddy covariance technique would seem aI., 2004]. Thus, accounting for this effect is not straight- to lie most in understanding drivers of intra-annual and infOlward. For example, even though an apparent "flux loss" terannual variability in Gp, R p, and R H [e.g., Miranda et al., may be observed on calm nights, it has also been shown 1997; Saleska e( aI., 2003; Vow'litis et al., 2005J and also in that 24-h integrals may be independent of the nighttime the acquisition of data to aid the calibration and development turbulence regime [Kruijt et al., 2004]. In such a situation, of ecosystem-lyvel gas exchange models [e.g. Lloyd et al., corrections for nighttime flux losses might, in fact, not be 1995; Mercadd et al., 2006; Baker et al., 2008; Mercado appropriate. et al., 2009J. Saleska et al. [2003] suggest that nighttime flux losses Even then, the meaningful scaling of any derived parameshould always be "corrected" for. Their argument was ters to the basin-wide scale must remain problematical, as all based, in part, on the observation that correcting for night- eddy covariance towers in Amazonian forest are currently time fluxes caused near-identical net annual ecosystem car- confined to the relatively infertile oxisol and dystrophic ultibon balances to be calculated for their tower and another sol soil types (ferrasols and acrisols/alisols in the new World approximately 20 km distant. But such logic is at odds with Reference Base soil classification system). These occupy only another conclusion of Saleska et al. [2003] in that their for- about 0.65 of the Amazon Basin area, with other soil types, est was a significant source of CO 2 to the atmosphere be- most ofwhich are more fertile, occupying the remaining 0.35 cause of CO 2 emissions from the unusually high amounts of [Quesada et al., 2009]. Although such forests are well studied coarse woody debris (CWD) at that site. As noted by Rice et in terms of aboveground biomass dynamics [Phillips et al., al. [2004] the site sampled by Saleska et al. [2003] contained this volumeJ, we currently know nothing about the magniabout 50% more CWD than the site with which they were tudes and environmental modulation of rates of their cancomparing [Miller et al., 2004]. It can simply be estimated, opy gas exchange. However, it does seem likely that due other things being equal, that this should have caused a dif- to significantly higher foliar nutrient concentrations [Fyllas ference in overall carbon balances between the two sites of et al., 2009; Lloyd et al., this volume], the photosynthetic approximately 2 Mg ha- I a-I. propeliies of these forests should be substantially different Other justifications for applying any given magnitude of from their less feliile eastern and central Amazonian counnighttime correction to eddy flux measurements are also de- terparis [Mercado et al., 2009]. As discussed below, forests batable. For example, Saleska et al. [2003J suggested that growing on more fertile soils also seem to be accumulatsimilar estimates of the carbon balance from eddy covari- ·ing carbon at a higher rate than forests growing on the less ance measurements and aboveground woody biomass inven- fertile fenasols and acrisols/alisols of central and eastern tmy measurements provide another justification. But only Amazonia.
416
HOUGHTON ET AL.
REGIONAL CARBON BUDGET
3.3. Plot-Based Biomass Approach Phillips et al. [1998] first reported that Amazonian and other tropieal forests appeared to be increasing in aboveground biomass (AGB), a result that has since been expanded upon and refined by Baker et al. [2004], the latter study estimating an average rate of increase in the carbon stocks of Amazoliian forest to be 1.22 ± 0.43 Mg ha- 1 a-I. Baker et al. [2004] also showed that this increase in biomass was not uniform across the Amazon Basin, but rather seemed to be greatest for forests growing on the more fertile soils of westem Amazonia, with rates of AGB accumulation also being especially high for forests growing on Holocene floodplain and contemporary floodplain plots [see also Phillips et al., this volume]. This suggested an overall aboveground carbon sink in the Amazon Basin of 0.6-0.8 Pg a-I. Nevertheless, as for eddy covariance measurements, these results have proved controversial, with possible methodological errors and biases having been suggested as giving rise to this apparent AGB increase [Clark, 2004; Wright, 2005]. Many of these concems seem to have been addressed and/or refuted [Phillips et al., 2002; Baker et al., 2004; Lewis, 2006], and attention ofthose philosophically opposed to a tropical forest carbon sink has tumed more to the possibility that the permanent plot sampling network does not adequately capture significant carbon losses. This has been suggested to arise either as a consequence of small scale but regular disturbances (such as may occur over areas of order 0.1 to 1.0 ha on a timescale ofthe order of decades) not being adequately detected by the current network or because current rates of AGB accumulation merely reflect the recovelY of forests from significant past disturbances [Keller et al., 1996; Saleska et al., 2003; Chambers et al., 2004; Wright, 2005; Clark, 2007]. Examples of possible major events proposed by these "catastrophists" include the great drought of 1926 [Williams et al., 2005] or mega-Nino events occurring many centuries ago [Meggers, 1994]. Such arguments are, however, hard to justify on even simple quantitative grounds. For example, as is also acknowledged by Chambers et al. [2004], if forests are recovering from some sort of disturbance regime, then rates of new stem growth would be expected to be decreasing with time, as the ecosystem approaches a new equilibrium. But, as has been shown by Lewis et al. [2004], for a range of plots across Amazonia, the opposite is, in fact, observed: stem growth rates are continually increasing. Likewise, if rates of AGB accumulation are currently around 0.7 Pg C a-I across the basin [Baker et al., 2004], as a result ofrecovelY from some past mega-disturbance such as the 1926 drought [Guariguata and Ostertag, 2001], and if they have been accumulating biomass at that rate for 80 years, then the amount of carbon
released fi'om that drought must have been at a minimum 60 Pg C, which would equate to an increase of about 25 f!ri101 mol- 1 in global atmospheric [C02] if it had all remained in the atmosphere. Of course, if such an event had occurred, then much of the carbon released would have been relatively quickly taken up by the oceans, but nevertheless, such a massive release of carbon, even if spread over a decade, would still have remained discemable in the recent Southem Hemisphere ice-core record. No such signal is apparent [Etheridge et al., 1996], and thus the notion that many Amazonian forests must be recoveting from severe carbon losses as a consequence of the 1926 event is unsubstantiated. Baker et al. [2004] provide several further lines of argument as to why the observed increase in forest biomass is not just a simple artefact, and analysing the data set in more detail, Lewis et al. [2004] also concluded that increased rates of forest growth in response to continually increasing atmospheric [C0 2] or radiation were the most likely explanation for this AGB increase in Amazonian forests. Those authors also noted that, as for the increase in AGB, tree growth rates seemed to be being stimulated to a greater extent in the more fertile forests of westem Ama'zonia. Not all studies, whether inside or outside of Amazonia, find increased growth rates. Changes in living wood, including recruitment and mortality as well as growth, was near steady state in Manaus, but accumulating carbon (~1.5 Mg C ha- I) in Santarem and Rio Branco [Vieira et al., 2004]. The authors suggested the accumulations were related to previous disturbances, and similarly, Fisher et al. [2008] have recently used an area-based stochastic simulator of forest disturbance and recovery to suggest that current measurements of an increase in AGB in Amazonia may, indeed, just be simple sampling artefacts. Yet the data analysis protocol and some fundamental assumptions implicit in the latter analysis may be flawed [Lloyd et al., 2009a], and a more rigorous data-based analysis of results from the Amazon RAINFOR plot network [Phillips et al., this volume] has shown that, overall, the reported AGB increase for Amazon forests is almost certainly real [Gloor et al., 2009]. Nevertheless, only four out oftenplots (1 6-52 ha each) sampled throughout the tropics showed a significant increase in aboveground biomass, for a combined average rate of0.24 Mg C ha- 1 a-I, a value on average lower than reported for Amazonia; this being attributed to either an increase in resource availability (presumably light and/or CO 2) or a recovelY from past disturbance [Chave et al., 2008]. Interestingly, the latter explanation was based, at least in part, on increased growth rates being higher (in relative tenus) for slower growing late succession species. Yet it is quite possible that slower growing species with high respiratory requirements might respond propottionally more to increases in C02 than
faster growing ones [Lloyd and Farquhar, 1996,2000], and it is also the case thaJlate successional species growing under light-limited c0uditions should be more responsive to increases in CO::lthan larger h'ees exposed to higher light levels higher upAn the canopy stratum [Lloyd and Farquhar, 2008]. Thus, it is not necessary to include disturbance-based recovelY as an explanation for the results of Chave et al. [2008], as they are, in fact, consistent with what we might expect for an ecosystem level stimulation of growth by increasing [C02], Two of the large plots (50 ha) examined by Chave et al. [2008], one in Panama and one in Malaysia, were also reported to show declining rates of tree growth [Feeley et al., 2007]. The changes in growth were related to regional climate changes: minimum daily temperature, precipitation, and insolation, with increased carbon losses associated with higher respiration rates at warm temperatures in recent years being considered the most likely explanation. It has, however, been shown by Lloyd and Farquhar [2008] that, if higher plant respiration rates were accounting for the decline in growth observed by Feeley et al. [2007], the QlO for autoh'ophic respiration would be unreasonably high. They suggested that stomatal closure in respons~ to higher canopy-to-air vapor pressure deficits in dry years was a more likely explanation. For the Pasoh forest in Malaysia, at least, ongoing soil acidification as a consequence of increasingly high rates of nitrogen- and sulfur-based pollutants in the region [Lewis et al., 2004] is another potential cause. One interesting question is the extent to which the currently increasing AGB of Amazonian forests is associated with changes in other carbon pools. That is, the results of Baker et al. [2004] reflect only one component of the dCp/dt term in (1); with leaf and root biomass changes not considered. Moreover, aboveground coarse woody debris (CWD) and the dCs/dt term are totally ignored. It is unlikely that significant changes in leaf biomass are currently occurring for tropical forests because their leaf area indices (LAI) are already very high [Lloyd and Farquhar, 1996]. The shott tumover time for leaves argues against any significant longterm accumulation of carbon [Lloyd and Farquhar, 1996]. This should also be the case for fine roots [Lloyd, 1999], for which overall biomass in Amazonian forests is, in any case, rather small [Aragao et al., 2009]. Based on the data available [Cairns et al., 1997; Mokany et al., 2006], however, it does seem likely that belowground coarse (woody root) carbon stocks should be increasing at about 0.25 the rate of increase in AGB, i.e., around 0.3 Mg C ha-1 a-lor 0.2 Pg C a-Ion a basin-wide basis. On the other hand, aboveground CWD may not follow changes in AGB. In a forest near Santarem, Para, CWD was recently observed to decline, while AGB was increasing [Pyle et al., 2008]. The observation was believed to result
417
from a recent disturbance, which transfened a fi'action of AGB to CWD. The net flux of carbon at the site was a source to the atmosphere, despite increasing AGB, and the net source was expected to last 10-15 years before the annual uptake of carbon in recovering AGB exceeded the annual loss in decaying CWD. The net source of carbon from the ecosystem despite a net accumulation in AGB calls attention to the importance of full carbon accounting. If carbon is accumulating in AGB, it may also be accumulating in the soil carbon pool [the dCs/dt term in (1)] as the extra live carbon accumulating aboveground and belowground must eventually be transferred to the litter and humus pools [Lloyd and Farquhar, 1996]. From theoretical principles we know that the magnitude of this accumulation should be dependent on both the rate at which plant litterfall and mortality increases in response to stimulated growth, itself dependent on dCp/dt, as well as the complex spectrum of litter and soil carbon tumover times [Lloyd, 1999]. A shott tumover time for litter, for example, would preclude any significant long-term accumulation of carbon. Available data on the different soil carbon pools and their associated tumover times are rare, but Telles et al. [2003] used radiocarbon isotope analysis to calibrate a simple model of soil carbon dynamics for some relatively infeltile ferrasol and acrisol soils of central and eastem Amazonia. They concluded that associated with a rate of increase in AGB of ~0.5 Mg C ha- I a-I should be an accumulation of carbon in soil of about 0.2 Mg C ha- I a-I, yet they observed no measurable change in organic carbon stocks over the past 20 years. Their simulated rate should be higher for the more fertile forests of westem Amazonia, where AGB is accumulating carbon at a faster rate [Baker et al., 2004]. A second important factor contributing to greater rates of carbon accumulation in the more fertile soils of the westem portion of the basin may be the characteristic differences in clay mineralogy between the ferrasols/acrisols of central and eastem Amazonia and the cambisols and other more fertile soil types of the westem portion of the basin. Most of the former are dominated by kaolinite [Irion, 1984], including the actual sites sampled by Telles et al. [2003]. On the other hand, reflecting a different parent material as well as different weathering conditions (typically less well drained), the soils of westem Amazonia are typically dominated by 2:1 clays [Irion, 1984; Quesada et al., 2009], which are characterized by stronger mineralorganic matter associations and hence longer mean residence times for soil carbon [Wattel-Koekkeok et al., 2003]. This means that, relative to the rate of AGB accumulation, substantially more carbon may, in fact, be accumulating in these westem Amazonian soils, especially when it is also considered that many ofthese soils are still weathering and contain considerable amounts of noncrystalline secondmy iron and
418
REGIONAL CARBON BUDGET
aluminum oxide minerals (Lloyd and Quesada, unpublished data). The latter bond more extensively to soil organic matter than the more crystalline forms, such as hematite and kaolinite [Torn et al., 1997]. The carbon dynamics of the soils of the Amazon Basin are discussed in more detail by Trumbore et al. [this volume]. 4. CARBON BALANCE OF AQUATIC SYSTEMS The streams, rivers, and wetlands of Amazonia appear to be net sources of carbon, releasing an estimated 0.5 Pg C a-I to the atmosphere [Richey et al., 2002, this volume]. If this efflux were attributable to a displaced respiration from terra firma forests, it would offset some of the annual net sinks of CO 2 measured by eddy covariance at tower sites. But it is unclear what fraction of terra firma forest NPP (or what fraction of upland areas) is contributing to this aquatic respiration. It has generally been supposed that the productivity ofseasonally flooded forests should be relatively high due to their occurrence on generally fertile alluvial soils [Worbes, 1997; Parolin et al., 2004], and this seems to be the case for varzea (seasonally flooded white water forests), which are among the most productive in the Amazon Basin [Malhi et al., 2004]. Nevertheless, it would seem likely that igap6 (seasonally flooded black or clear water forests) have a lower productivity than varzea due to the overall lower nutrient levels of their supporting soils [Quesada et al., 2009]. Indeed, Worbes [1997] reported significantly higher litterfall rates for varzea than igap6. The ratio of fine litterfall to aboveground (stem) growth rates for varzea (about 1.6) was similar to that observed for Amazon terre firme forests [Aragao et al., 2009], with most leaf litterfall occurring during the flooding period [Worbes, 1997; Parolin et al., 2004] and thus directly into the aquatic system. From the above observations, a reasonable estimate for leaf litterfall input to rivers from seasonally flooded forests is 5 Mg C ha- I a-I, allowing calculation ofthe total (Amazon basin-wide) river input from seasonally flooded forest foliar litterfall as follows: Seasonally flooded areas for a study site of 1.8 x 10 6 km 2 in the central Amazon Basin [Hess et al., 2003] and a 2.2 x 106 km2 area of western Amazonia [Toivonen et al., 2007] accounted for about 0.17 of the areas. Assuming the fraction of seasonally flooded areas occupied by forests is ~0.7 [Hess et al., 2003] implies a total area of seasonally flooded forest in the Amazon Basin of 0048 x 106 kni or 4.8 x 107 ha. Multiplying this by the estimated leaf litterfall rate of 5 Mg C ha- I a-I gives a net direct input of ~ 0.25 Pg C a-I into the Amazon river system, almost all of which would be expected to be respired downriver and thus part of the Richey et al. [2002] measured efflux of approximately 0.5 Pg C a-I.
HOUGHTON ET AL.
Though occupying only about 0.1 the area of seasonally flooded forest, C4 aquatic grasses are a second potentially significant autochthonous source into the Amazon river system. The NPP ofthese grasses is very high, estimated to be of the order 50 Mg C ha- I a-I [Piedade et al., 1991], yet ecosystem respiration rates from such wetlands are low [Morison et al., 2000], suggesting that much of the carbon assimilated by aquatic grasses must be transported to, and respired in, the main river system, estimated from the above numbers as ~ 0.25 Pg C a-I. These estimates, combined with observations that the rates of loss of dissolved organic carbon [Waterloo et al., 2006] and dissolved CO 2 [Davidson et al., 2008] from terre finne forests to rivers are low, suggest that inputs into the Amazon river system from seasonally flooded forests and aquatic grasses may well be sufficient to account for the 0.5 Pg C a-I net efflux observed by Richey et al. [2002], This explanation is consistent with the very young age of the carbon respired from the Amazon River system and with a l3C isotopic composition indicative of a C4 source [Mayorga et aI" 2005]. It is also consistent with the review by Melack et al. [this volume] that flooded forests and aquatic macrophytes ac'count for 96% of the carbon inputs to the river. It is possible that the efflux of carbon fi'om wetlands is only a fraction of the inputs from terrestrial systems (that the aquatic sediments are a net carbon sink). We know that little carbon is exporied by the Amazon River to the ocean (0.07 Pg C a-I) [Richey et al., 1990], but of the inputs to aquatic systems (both allochthonous and autochthonous), the fractions respired, as opposed to accumulating, are unknown. In two watersheds in the United States, erosion and deposition were estimated to yield a net sink of the order of 1% ofNPP [Berhe et al., 2007]. If the same fraction ofNPP is sequestered in Amazonia, the sink would be on the order of 0.1 Pg C a-I (NPP is approximately 10 Pg C a-I) [Potter et al., 2004] (see section 5). It thus seems that the role of wetlands and rivers in the regional carbon balance is nearly neutral, with the aquatic efflux of 0.5 Pg C a-I balanced by authochthonous inputs from seasonally flooded forests and grasslands. 5. CARBON BALANCE OF AMAZONIAN REGIONS: ECOSYSTEM MODELS BASED ON PHYSIOLOGY Ecosystem carbon models have been used to estimate Amazonia-wide fluxes ofcarbon by scaling up measurements and processes observed at fine scales. Not surprisingly, both the models themselves and the estimates they provide vary. Furthermore, models have been used to address different components of the carbon balance. The distinction pettinent for this discussion is between those modeling studies
that have focused on the physiological processes governing natural fluxes of carbo!). and those that have focused on disturbance and recovl?PY processes, both anthropogenic and natural. The two types of studies are reviewed in this and the next section, resJ1i:ktively. Sometimes the same model has '" both types of processes, been used to addi'ess Terrestrial biogeochemical models have been used to calculate the fluxes of carbon fi'om photosynthesis (GPP), NPP, and heterotrophic respiration (RH ), including decay. To one degree or another, all of the models reviewed here attempt to compute rates ofplant carbon uptake and litter decomposition as influenced by environmental variables (radiation, temperature, precipitation, and nutrients), In some models, NPP is driven with environmental data alone; in others, it is determined by changes in leaf-area index, inferred fi'om NOAA's advanced vety high resolution radiometer (AVHRR) satellite or NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) satellite [Potter et al., 1999]. Several studies have examined how interannual variations in climate affect the carbon balance of the Amazon Basin [Kindermann et aI" 1996; Tian et al., 1998; Prentice and Lloyd, 1998; Asner et al., 2000; Potter et aI" 2001a, 2001b, 2001c; Foley et al., 2002]. All of the studies found that the net annual flux of carbon is significantly correlated with ENSO events. The Amazon Basin appears to be a significant carbon source during El Nino events and a sink during La Nina events, consistent with atmospheric inverse calculations (section 2.1) (Figure 1). Model estimates of the El Nino source varied from 0.17 to 004 Pg C a-I; estimates of the La Nina sink varied from -0.27 to -0.7 Pg C a-I [Ticl71 et al., 1998; Potter et aI" 2001a, 2001b, 2001c, 2004; Foley et al., 2002]. Most of these studies concluded that the major variations in regional carbon balance are related chiefly to changes in precipitation. Foley et al. [2002], using Integrated Biosphere Simulator (IBIS) [Foley et al., 1996; Kucharik et al., 2000], found that changes in carbon balance are largely driven by changes in ecosystem productivity, linked to changes in soil moisture and drought stress. Potter et aI, [2001a, 2001b, 2001c] also found that drought during El Nino years reduced NPP, Periods of relatively high solar surface irradiance combined with several months of adequate rainfall were required to sustain the forest carbon sink, Notwithstanding the observation that TEM [Melillo et al., 1993; McGuire et aI" 1995,2001; Tian et al., 1998], NASACASA [Potter et aI" 2001a, 2001b, 2001c, 2004], and IBIS [Foley et aI" 1996; Kucharik et al., 2000] all reproduced net annual sources of carbon during El Nino years and net annual sinks during La Nina years, they did a poor job reproducing seasonal carbon fluxes for the tropical evergreen forests near Santarem, Para, Brazil [Saleska et aI" 2003]. Photosynthesis
419
is apparently less water stressed, seasonally, than the models predict and more, light limited. The trees are able to acquire water from deep roots [Nepstad et al., 1994] or by hydraulic redistribution of soil water [Rocha et al., 2004; Oliveira et al., 2004]. Heterotrophic respiration in the upper soil layers, on the other hand, probably is limited by soil water. Thus, reduced respiration contributes to the greater net uptake of carbon during dty seasons [see Saleska et aI" this volume] for a more detailed discussion of seasonality, including apparent differences in the water stress of vegetation in different parts of Amazonia). The fact that ecosystem models failed to reproduce even the sign of the seasonal fluxes of carbon raises the question of whether they simulate accurately year-to-year variations in flux, The recent observation that Amazonian forests may have greater photosynthesis during the dty season than the wet season is also the reverse of what most models simulate [Saleska et aI" 2007]. It has recently been shown, however, that the simultaneous inclusion of water storage to large depth, the ability of roots to extract water from such depths as the surface soil dries, an allowance for hydraulic redistribution to occur allowing for more efficient water uptake during the wet season and moistening of the near-surface soil during drought, and the inclusion of a photosynthetic response to high light levels during the dry season, including all of these processes in model simulations, does allow such observations to be simulated [Baker et aI" 2008]. An analysis of the phY$iological mechanisms responsible in the models for variations in metabolism seems vital but is beyond to the scope of this review. Although NPP in the humid tropics has been observed to increase during the dty season [Saleska et al., 2003; Huete et al., 2006], it is unlikely that an extended drought would lead to an increased storage of carbon. For example, ENSO events seem to be associated with reduced sinks or larger sources of carbon. Nevertheless, it is unclear whether the larger sources during ENSO are physiological (differential changes in photosynthesis and respiration) or disturbancebased (fires). The relative importance of water, as opposed to light, in limiting photosynthesis is complicated by the effects of aerosols, which are at higher concentrations during the dty season as a result of fires. Aerosol loading increases photosynthesis initially because it increases diffuse radiation. At higher loadings, however, the lower total radiation reduces photosynthesis more than the diffuse radiation increases it [Oliveira et aI" 2007]. It seems likely that short-tenn responses to drought (or incr.eased radiation) are opposite to long-term responses. During protracted droughts, vegetation may eventually become water stressed [Nepstad et aI" 2007; Branda et al.,
420
REGIONAL CARBON BUDGET HOUGHTON ET AL.
2008]. In a simulation with IBIS, Botta et aZ. [2002] found that the Amazon Basin can have long-term, climate-induced variations in carbon balance. Considering only variations in climate (an~ not in CO 2 or land use), they found that Amazonia was almost neutral from the late 1930s to the late 1950s (-0.42 Pg Cover 1935-1957), a net carbon source during the 1960s (+1.98 Pg Cover 1958-1967), a net sink during the 1970s (-;-2.54 Pg Cover 1968-1978), and back to nearly neutral during the 1980s and 1990s (+0.61 Pg Cover 1979-1995). The recent neutral trend is not obtained with other analyses. Over the 18-year period 1982-1999, both the NASACamegie-Ames-Stanford approach (CASA) model [Potter et aZ., 2004] and global biome model-biogeochemical cycle (BIOME-BGC) [Nemani et aZ., 2003] predicted an increase in NPP. The long-term observation that cloudiness has decreased (net radiation increased) over this period led both models to calculate an increase in NPP, in the case of [Nemani et aZ., 2003] an increase in Amazonian NPP that accounted for 42% of the global NPP increase. That trend in radiation has recently been questioned, however [Evan et aZ., 2007]. The long-term trend in cloudiness may be an atiifact of the satellite record. Rates of disturbance, both anthropogenic and natural, may be as impOliant as climate variability in affecting ecosystem composition and carbon dynamics in the Amazon region [Botta and FoZey, 2002]. Using IBIS to examine the effects of climate variability (using either long-term average climate or actual historical variations in climate) and ecological disturbances, Botta and FoZey [2002] found that interannual climate variability and frequent disturbances both favor grasses over trees, causing large increases in the geographic extent of savanna in the south and east of the region. A more constant climate and less frequent disturbances both favor trees over grasses, causing forest to dominate most of the study area. Similar results, suggesting that changes in land use may lead to changes in regional climate with consequences for carbon storage, have been obtained with atmospheric general circulation models [Cox et aZ., 2000; Betts et aZ., 2004]. Oyama and Nobre [2003] showed, for example, that the loss of Amazonian forests could change the existing climatevegetation system to a drier equilibrium state, with savannization in parts of Amazonia and desertification in the driest area of Northeast Brazil. The conversion of forests to degraded pasture and soybean croplands increased surface air temperatures and decreased evapotranspiration and precipitation, especially in eastern Amazonia [Sampaio et aZ., 2007]. The continued expansion of cropland in Amazonia could have important consequences for the continued existence of the region's forests and, hence, for carbon storage [Nepstad et aZ., 2008].
6. SOURCES AND SINKS OF CARBON FROM LAND USE CHANGE, DISTURBANCE, AND RECOVERY Several analyses have calculated the net emissions of carbon from Brazilian Amazonia as a result of land use change, management, and disturbance. All of the analyses have considered deforestation and regrowth of secondaly forests within the moist forest zone. They differ with respect to whether they considered changes in soil carbon, whether they included cerrado, and whether they considered timber harvest, fires, or other disturbances. Even if all of these processes were included, it is important to recognize that the net flux attributable to disturbances is not necessarily the total net flux of carbon for the region. Nor is it equivalent to the fluxes estimated from the biogeochemical models discussed in section 5. The analyses discussed in this section do not include the fluxes in undisturbed ecosystems. All of the analyses calculate a net carbon source within the range of 0.15 to 0.35 Pg C a-I [Fearnside, 1997; Houghton et aZ., 2000; Potter et aZ., 2001c; DeFries et al., 2002; Hirsch et aZ., 2004; Ramankutty et aZ., 2007]. Unceliainties in biomass, deforestation rate, and rates of decomposition were estimated to account for 60%,25%, and 15% ofthe uncertainty in flux estimates [Houghton et aZ., 2000], and thus, the higher estimates were largely the result of higher estimates of biomass [Hirsch et aZ., 2004; Ramankutty et aZ., 2007]. The relatively small error from deforestation rate is unique to this region of the tropics. Annual rates of deforestation in Brazilian Amazonia are better documented than elsewhere. For all of Brazil, DeFries et aZ. [2002] estimated higher average emissions for the 1990s (0.28 Pg C a-I) than for the 1980s (0.15 Pg C a-I). Most of the net flux was attributable to burning and decay of vegetation and slash, with only a small uptake by secondaly (regrowing) forests. Sources or sinles of carbon in soils were ignored in these studies, as were logging and fire. Adding fires [see Longo et aZ., this volume], Potter et aZ. [2001c] used a version of the NASA-CASA model, together with Landsat-derived mapping of burned areas for the Legal Amazon [AZves, 1999], to estimate total fluxes of 0.2 to 1.2 Pg C a-I for the entire Legal Amazon. The variation depended strongly on annual rainfall pattems. Based on an analysis of fire counts during 1992-1993, Potter et aZ. [2001 c] calculated a net flux of 0.77 Pg C a-I, of which 0.71 Pg C a-I was from fires and only 0.056 Pg C a-I was from postburning (decomposition). Fmihennore, 75% of the fires were in the cerrado, outside of the moist forest area included in other studies. In comparison, Van del' Welfet aZ. [2003] used the CASA model with inputs of rainfall from the NASA Tropical Rainfall Monitoring Mission to calculate annual carbon emissions fi'om fires in the Legal Amazon of 0.2 to 0.5 Pg C a-I.
Adding logging [see Asner et aZ., this volume] to the analysis increased estimat~s of the net emissions of carbon. Asner et aZ. [2005] calcutll:ted a gross source of 0.08 Pg C a-I from decomposition qf roundwood, residual stumps, branches, foliage, and roqis left on site following wood harvest. The value is a grodflux because logged forests will presumably accumulate carbon as they regrow. However, many logged forests are not permitted to regrow. Approximately a third of the logged forests are cleared for agriculture before they recover [Asner et aZ., 2006]. The probability that logged forests will be cleared is four times greater than the probability that unlogged forests will be cleared. The finding is consistent with another study of the region, which found that the mean age ofsecondaty forests (4.4 to 4.8 years) had changed velY little between 1978 and 2002 [Neeff et aZ., 2006]. The average age did not increase because secondary forests were usually re-cleared. Adding soils [Trumbore et aZ., this volume] to the analysis also increased the emissions of carbon as a consequence of cultivation. Most of the lands deforested in Amazonia are not cultivated, however, but used for pasture. Cerri et aZ. [2007] repOli that the conversion of forest to well-managed pastures causes an initial decline in soil carbon stocks '(0-20 cm) followed by a slow rise to levels exceeding those under native forest. In degraded pastures, the carbon stocks may not recover. New data and analyses from LBA suggest that the next generation of emission estimates fi'om land use change and management are likely to have smaller enol's. For example, two new estimates of forest biomass at high spatial resolution [MaZhi et aZ., 2006; Saatchi et aZ., 2007; Phillips et aZ., this volume] and more detailed accounting for differences in wood density and allometIic equations [Nogueira et aZ., 2008]
should allow more precise estimates of carbon emissions fi'om deforestation. As mentioned above, unceIiain estimates of biomass contributed more than any other factor to the variability offlux estimates [Houghton et aZ., 2000, 2001]. Deforestation for new types of land use [Morton et aZ., 2006; AZves et aZ., this volume; WaZker et aZ., this volume] may also affect the emissions of carbon. A greater proportion of deforestation in Mato Grosso in recent years, for example, has been for soybean production rather than for pastures. This change in land use has at least two effects. One effect is to release more carbon more rapidly. Aboveground biomass and woody roots are removed rapidly and completely when the land is cultivated, as opposed to grazed. Cultivation leaves little forest biomass for decomposition and delayed emissions. Second, cultivation leads to a 25-30% loss ofsoil organic carbon from the top meter. 7. SYNTHESIS; WHAT DO WE KNOW? The net flux of carbon between Amazonia and the atmosphere is small relative to the stocks of carbon held in the vegetation and soils of the region's forests and small relative to the background, or natural, fluxes of carbon ammally taken up and released through photosynthesis and respiration. Thus, detem;tining the net flux is difficult. As a result, at least five different approaches have been used to estimate this flux or portiops of it: 1. Inverse methods based on atmospheric CO 2 and transport are not well constrained because atmospheric transport is poorly known, and CO 2 sampling stations are too sparse. Furthermore, the approach provides limited infonnation on the mechanisms involved or the long-term response to changes in climate.
Table 2. Different Approaches Used to Estimate the Net Sources and Sinks of Carbon for the Amazonian Regiona Flux (Mg C ha- 1 a-I) Inverse methods Vertical profiles Eddy covariance Pennanent plots Aquatic systems Process models Disturbance/recovery Fire Total C
Flux (Pg C a-I) +4.2 b
Credible range
-1.2 1.2 ± 0.3
421
-1.0 to Insufficient data available Insufficient data available -0.6 to -0.8 0.0 -0.6 to 0.2 0.1 to 0.5 0.1 to 0.3 -0.6 to +0.2
Chapter Section 2.1
2.2 3.2 3.3 4 5 6 6
aNegative values represent a terrestrial sink. bAll h·opics. If the net carbon balance for Amazonia were -in proportion to forest area, the range for Amazonia would be -0.3 to +1.4 Pg C a-I. CRange of the values (in italics) that, together, include all ecosystems and all processes.
422
REGIONAL CARBON BUDGET
2. Direct measurements of C02 flux at landscape level (eddy covariance) help elucidate the mechanisms important in short-term, metabolic fluxes of carbon but are difficult to extrapolate tOl1the entire Amazonian region. 3. Direct measurement of changes in carbon stocks (inventories) over years capture the longer-term changes but are difficult to attribute to particular mechanisms. The most likely explanation for an increase seems to be that resources have become more available for plant growth (e.g., solar radiation, atmospheric CO 2 concel'ltrations). 4. Ecosystem models, together with eddy covariance measurements, are good for predicting ecosystem responses to short-term variations in climatic factors, but whether they capture the important long-term responses is unclear. Furthermore, early versions of these models did not reproduce observed short-term (seasonal) changes in NPP and heterotrophic respiration. . 5. Disturbance, recovery, management, and changes m land use involve large changes in carbon stocks per unit area that are reasonably well known, but the carbon stocks of the forests actually deforested are still uncertain, as is the change in soil carbon stocks in response to grazing. Only two of these approaches (3 and 5) yield unambiguous information on the sources and sinles of carbon for the Amazonian region (Table 2). Changes in biomass on permanent plots in old-growth forests suggest that these forests have been accumulating carbon over the last ~20 years at a rate of approximately 0.6 to 0.8 Pg C a-I, Changes in land use, management, and fire, on the other hand, release carbon at rates of 0.2 to 0,8 Pg C a-I. Year-to-year variations are large, so that terrestrial ecosystems in Amazonia appear ~o be a net carbon sink in some years and a net carbon source m others, The data are too uncertain and too variable to specify whether the region has been a net sinle or source over the last decade, a conclusion similar to earlier summaries [Davidson
and Artaxo, 2004; Ometto et aI" 2005]. If the efflux of carbon from rivers includes part of the ecosystem respiration from forests, this efflux is implicitly included in process-based models and, thus, included in the modeled estimates of net carbon balance for intact ecosystems. But it is not included in either of the approaches that yield unambiguous estimates and, therefore, must be added to those estimates for determining a basin-wide estimate of carbon balance. Estimates of carbon inputs from seasonally flooded forests and grasslands suggest, however, that the efflux is largely balanced by autochthonous inputs. Carbon may be accumulating in aquatic sediments, but the magnitude of such an accumulation is not quantified. Thus, including rivers does not change the conclusion that the Amazonian carbon balance is neutral within the errors of measurement.
HOUGHTON ET AL. Summing the carbon sink in old-growth forests (-0.6 to -0,8 Pg C a-I) with the sources from disturbed lands (0.2 to 0.8 Pg C a-I) yields a net flux for the entire region of -0.6 to 0.2 Pg C a-I. Amazonia is essentially balanced with respect to carbon. The causes of the offsetting sources and sinks of carbon are not entirely clear. Rates of deforestation have been greater than rates of reforestation and afforestation, and thus, the net flux of carbon from anthropogenic disturbance and recovelY is clearly a net source of carbon to the atmosphere. The accumulation of biomass on permanent plots within oldgrowth forests, on the other hand, may be the result of a growth enhancement brought about by increased radiation or higher concentrations of CO 2 , that is, by physiological responses to changes in the environment. RecovelY, rather than metabolic, processes cannot be ruled out, however, because local disturbances are COlmnon, The fi'equency of such natural disturbances is not well quantified but is of critical impOliance because changes in coarse woody debris as a result of disturbance may offset the accumulations of carbon in aboveground biomass. If such offsets are the rule, the net flux of carbon in old-growth forests would be nearly zero, and the net flux for the region would be a source, REFERENCES Alves, D. S, (1999), Geographical pattems of deforestation in the 1991-1996 period, Proceedings of the 48th Annual Conference of the Center for Latin American Stndies, Patterns and Processes of Land Use and Forest Change in the Amazon, University of Florida, Gainesville, March 23-26, Alves, D. S" D. C, Morton, M, Batistella, D. A Roberts, and C. Souza Jr. (2009), The changing rates and patterns of deforestation and land use in Brazilian Amazonia, Geophys. Monogr. Ser" doi:l0.1029/2008GM000722, this volume. Aragao, L. E. 0, C" et al. (2009), Above- and below-ground net prima1y productivity across ten Amazonian forests on conh'asting soils, Biogeosci. Discuss., 6, 2441-2488. Aranibar, 1 N" 1 A. Beny, W. 1 Riley, D, E, Patakis, B, E. Law, and 1 R. Ehleringer (2006), Combining meteorology, eddy fluxes, isotope measurements, and modeling to understand environmerital controls of carbon isotope discrimination at the canopy scale, Global Change BioI" 12, 710-730, Arm\jo, A, C., et al. (2002), Comparative measurements of carbon dioxide fluxes from two nearby towers in a central Amazonian rainforest: The Manaus LBA site, J. Geophys, Res" 107(020), 8090, doi: 10.10291200 IJD000676. Aralljo, A C., B, Kruijt, A, D. Nobre, A. 1 Dolman, M, 1 Waterloo, E, 1 Moors, and 1 De Souza (2007), Noctural accumulation of CO 2 underneath a tropical forest canopy along a topographical gradient, Ecol, Appl" in press, Asner, G, P" A. R. Townsend, and B. H, Braswell (2000), Satellite observation ofEI Nino effects on Amazon Forest phenology and productivity, Geophys, Res, Lett., 27(7), 981-984,
Asner, G, P., D. E. Knapp, E. N. Broadbent, P, J. C, Oliveira, M. Keller, and 1 N, Silva (9005), Selective logging in the Brazilian Amazon, Science, 31Qi480-482. Asner, G, P" E. N, Btpadbent, P, 1 C, Oliveira, M, Keller, D, E. Knapp, and 1 N, Silva (2006), Condition and fate of logged forests in the Braiilian Amazon, Proc. Natl. Acad. Sci, U. S. A., 103,12,947-12,950. Asner, G. P., M. Keller, M, Lentini, F, Meny, and C. Souza Jr. (2009), Selective iogging and its relation to deforestation, Geophys. Monogr, Ser" doi:IO,102912008GM000723, this volume. Baker, D, F" et al. (2006), TransCom 3 inversion intercomparison: Impact oftransport model errors on the interannual variability of regional CO 2 fluxes, 1988-2003, Global Biogeochem, Cycles, 20, GB1002, doi:IO.l029/2004GB002439, Baker, 1. T., L. Prihodko, A S. Denning, M, Goulden, S. Miller, and H. R, da Rocha (2008), Seasonal drought stress in the Amazon: Reconciling models and observations, J. Geophys. Res" 113, GOOBOl, doi:IO.l029/2007JG000644. Baker, T, R., et al. (2004), Are Amazonian forest plots increasing in biomass? Phi/os, Trans, R, Soc, London, Ser. B, 359, 353-365, Berhe, A A., 1 Harte, 1 W. Harden, and M, S. Torn (2007), The significance of the erosion-induced terrestrial carbon sink, BioScience, 57, 337-346, Betts, R, A, P, M, Cox, M, Collins, P. P, Harris, C. Huntingford, and C. D, Jones (2004), The role of ecosystematmosphere interactions in simulated Amazonian precipitation decrease and forest dieback under global climate warming, Them', Appl. Climato!., 78, 157-175. Botta, A, and 1 A Foley (2002), Effects of climate variability and distnrbances on the Amazonian terresh'ial ecosystems dynamics, Global Biogeochem. Cycles, 16(4), 1070, doi:l0.l029/ 2000GB001338. Botta, A" N, Ramankutty, and 1 A Foley (2002), Long-term variations of climate and carbon fluxes over the Amazon basin, Geophys. Res, Lett., 29(9),1319, doi:l0,1029/2001GLOI3607. Bousquet, P., P, Peylin, P, Ciais, C, Le Quere, P. Friedlingstein, and P, P, Tans (2000), Regional changes in carbon dioxide fluxes ofland and oceans since 1980, Science, 290, 1342-1346, Brando, P. M., 0, C, Nepstad, E. A Davidson, S. E. Trumbore, D. Ray, and P, Camargo (2008), Drought effects on litterfall, wood production, and belowground carbon cycling in an Amazon forest: Results of a throughfall reduction experiment, Phi/os, Trans, R, Soc, Ser. B, 363, 1839-1848, Caims, M. A" S. Brown, E, H, Helmer, and G. A Baumgardner (1997), Root biomass allocation in the world's upland forests, Oecologia, 111,1-11, Carswell, F, E" et al. (2002), Seasonality in CO 2 and H 20 flux at an eastem Amazonian rain forest, J. Geophys, Res" 107(D20), 8076, doi: 10.1 029/2000ID000284. CelTi, E. P, C., et al. (2007), Simulating SOC changes in 11 land use change chronosequences from the Brazilian Amazon with RothC and Century models, Agric, Ecosyst, Environ" 122,46-57. Chambers, 1 Q, and W. L. Siver (2004), Some aspects of ecophysiological and biogeochemical responses of tropical forests to atmospheric change, Phi/os. Trans, R, Soc, London, Ser, B, 359, 463-476,
¥.
423
Chambers, 1 Q., N, Higuchi, L. M, Teixeira, 1 dos Santos, S. G, Lurance, and S, E. Trumbore (2004), Response of tree biomass and wood litter to disturbance in a Central Amazon forest, Oecologia, 141, 596-614, Chave, l, R. et al. (2008), Assessing evidence for pervasive alteration in tropical tree communities, PLoS BioI" 6(3), e45, doi: 10. 137 l/journal.pbio.0060045, Chou, W, W" S. C, Wofsy, R. C. Harriss, 1 C. Lin, C, Gerbig, and G, W, Sachse (2002), Net fluxes of CO 2 in Amazonia derivcd fi'om aircraft observations, J. Geophys, Res., 107(D22), 4614, doi: 10.1029/200 lIDOO 1295, Clark, D. A (2002), Are tropical forests an important global carbon sink?: Revisiting the evidence from long-term inventOly plots, Ecol, Appl" 12, 3-7, Clark, D. A (2004), Sources or sinks? The responses of tropical forests to current and futnre climate and atmospheric composition, Phi/os, Trans, R. Soc. London, Ser. B, 359, 477-491. Clark, D. A (2007), Detecting tropical forests' responses to global climatic and atmospheric change: Current challenges and a way forward, Biotropica, 39, 4-19. Cox, P. M., R. A Betts, C, D. Jones, S, A. Spall, and 1. 1 Totterdell (2000), Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model, Nature, 408, 184-187, Crevoisier, C" M. Gloor, E, Gloaguen, C, Sweeney, L. Horowitz, and P. Tans (2006), A direct carbon budgeting approach to infer carbon sources and sinks. Introduction, synthetic tests and application to complement NACP observations network, Tellus, Ser. B, 58(5), 366-375, doi: 10.111l/j.1600-0889.2006.00214,x. Culf, A D., G. Fisch, Y. Malhi, R. C. Costa., A D, Marques, 1 H. C. Gash, and:l Grace (1999), Carbon dioxide measurements in the noctnrhal boundmy layer over Amazonian forest, Hydrol, Earth Syst, Sci" 3, 39-53, Davidson, E, A, and P, Artaxo (2004), Globally significant changes in biological processes of the Amazon Basin: Results of the Large-scale Biosphere-Atmosphere Experiment, Global Change BioI., 10, 519-529, Davidson, E. A., R. de O. Figueiredo, D. Markewitz, and P, de S. Silva (2008), Dissolved CO 2 in small catchment sh'eams of eastern Amazonia. Amazon in Perspective-International Conference, Manaus, Brazil, November 17-21, 2008, Absh'act #525. (Available at http://www.lbaconferencia.org/lbaconC2008/pOlilindex.hhn). DeFries, R. S., R, A Houghton, M. C. Hansen, C. B. Field, D. Skole, and 1 Townshend (2002), Carbon emissions fi'om h'opical deforestation and regrowth based on satellite observations for the 1980s and 90s, Proc, Nat!' Acad. Sci. U. S. A., 99, 14,256-14,261. Denning, A S" I. Y. Fung, and D. Randall (1995), Latitndinal gradient of atmospheric CO2 due to seasonal exchange with land biota, Nature, 376, 240-243. Etheridge, D. M" L. P. Steele, R. L. Langenfelds, R. J. Francey, l-M. Barnola, and V. 1. Morgan (1996), Natnral and anthropogenic changes in atmospheric CO 2 over the last 1000 years from air in Antarctic ice and firn, J. Geophys. Res" 101,4115-4128. Evan, A T., A K. Heidinger, and D. 1 Vimont (2007), Arguments against a physical 10ng-telID trend in global ISCCP cloud amounts, Geophys, Res, Lett" 34, L04701, doi:l0.l029/ 2006GL028083.
424
REGIONAL CARBON BUDGET
an undisturbed tropical forest in south-west Amazonia, Global Fan, S., M. Gloor, J. Mahlman, S. Pacala, J. Sarmiento, T. TakaChange BioI., 1,1-12. hashi, and P. Tans (1998), A large terrestrial carbon sink In NOlih America implied by atmospheric and oceanic carbon di- Grace, J., Y. Mali, J. Lloyd, J. McIntyre, A. C. Miranda, P. Meir, and H. S. Miranda (1996), The use of eddy covariance to infer oxide daj:p and models, Science, 282, 442--446. the net carbon dioxide uptake of Brazilian rain forest, Global Fan S.-M.,\ L. Sarmiento, M. Gloor, and S. W. Pacala (1999), Change BioI., 2, 209-217. On the use of regularization techniques in the inverse modeling of atmospheric carbon dioxide, 1. Geophys. Res., 104, 21,503- Guiriguata, M. R., and R. Ostertag (2001), Neotropical secondaly sucession; changes in structure and functional characteristics, 21,512. For. Ecol. Manage., 148,185-206. FearnsiQe, P. M. (1997), Greenhouse gases from deforestation in Gurney, K. R., et al. (2002), Towards more robust estimates of CO2 Brazilian Amazonia: Net committed emissions, Clim. Change, fluxes: Control results from the TransCom3 inversion intercom35,321-360. parison, Nature, 415, 626-630. Feeley, K. J., S. J. Wright, M. N. N. Supardi, A. R. Kassim, and Gurney, K. R, et al. (2003), Transcom 3 CO 2 inversion intercomS. J. Davies (2007), Decelerating growth in tropical forest trees, parison: I. Annual mean control results and sensitivity to transEcol. Lett., 10,461--469. pOli and prior flux information, Tellus, Ser. B, 55, 555-579. Fisher, J. I., G. C. Hurtt, R. Thomas, and J. Q. Chambers (2008), Clustered disturbances lead to bias in large-scale estimates based Hess, L. L., J. M. Melack, E. M. L. M. Novo, C. C. F. Barbosa, and M. Gastil (2003), Dual-season mapping of wetland inundaon forest sample plots, Ecol. Lett., 11, 1-10. tion and vegetation for the central Amazon basin, Remote Sens. Foley, J. A., I. C. Prentice, N. Ramankutty, S. Levis, D. Pollard, Environ., 87, 404--428. S. Sitch, and A. Haxeltine (1996), An integrated biosphere model of land surface processes, terrestrial carbon balance, Hirsch, A. I., W. Little, R. A . Houghton, N. A. Scott, and J. D. White (2004), The net carbon flux due to deforestation and forest and vegetation dynamics, Global Biogeochem. Cycles, 10(4), re-growth in the Brazilian Amazon: Analysis using a process603-628. based model, Global Change BioI., 10, 908-924, doi: 10.1111/ Foley, J. A., A. Botta, M. T. Coe, and M. H. Costa (2002), EI j.l529-8817.2003.00765.x. Nino-Southern oscillation and the climate, ecosystems and rivers of Amazonia, Global Biogeochem. Cycles, 16(4), 1132, Houghton, R. A., D. L. Skole, C. A. Nobre, J. L. Hackler, K. T. Lawrence, and W. H. Chomentowski (2000), Annual fluxes of doi: 10.1 029/2002GBOO 1872. carbon from deforestation and regrowth in the Brazilian AmaFyllas, N. M., et al. (2009), Basin-wide variations in foliar properzon, Nature, 403, 301-304. ties of Amazon forest trees: Phylogeny, soils and climate, BioHoughton, R. A., K. T. Lawrence, J. L. Hackler, and S. Brown (2001), geosci. Discuss., 6, 3707-3769. The spatial distribution offorest biomass in the Brazilian Amazon: Gloor, M., N. Gruber, J. Sarmiento, C. L. Sabine, R. A. Feely, and A comparison of estimates, Global Change BioI., 7,731-746. C. Rodenbeck (2003), A first estimate of present and preindustrial air-sea CO 2 flux patterns based on ocean interior carbon Huete, A. R., K. Didan, Y. E. Shimabukuro, P. Ratana, S. R. Saleska, L. R. Hutyra, W. Yang, R. R. Nemani, and R. Myneni measurements and models, Geophys. Res. Lett., 30(1), 1010, (2006), Amazon rainforests green-up with sunlight in dly season, doi: 10.1029/2002GLO15594. Geophys. Res. Lett., 33, L06405, doi:l0.l029/2005GL025583. Gloor, M., E. Dlugokencky, C. Brenninkmeijer, L. Horowitz, D. F. Hurst, G. Dutton, C. Crevoisier, T. Machida, and P. Tans Hutyra, L. R., J. W. Munger, S. R. Saleska, E. Gottlieb, B. C. Daube, A. L. Dunn, D. F. Amaral, P. B. Camargo, and S. C. (2007), Three-dimensional SF 6 data and tropospheric transport Wofsy (2007), Seasonal controls on the exchange of carbon simulations: Signals, modeling accuracy, and implications for and water in an Amazonian rain forest, 1. Geophys. Res., 112, inverse modeling, 1. Geophys. Res., 112, Dl5112, doi:IO.1029/ G03008, doi: 10.1 029/2006JG000365. 2006JD007973. Gloor, M., et al. (2009), Does the disturbance hypothesis explain Irion, G. (1984), Clay minerals of Amazonian soils, in The Amazon, Limnology and Landscape Ecology of a Mighty Tropical biomass increase results found in basin-wide Amazon forest plot River and its Basin, edited by H. Sioli, Springer, Dordrecht. data?, Global Change BioI., in press. Jacobson, A. R, S. E. Mikaloff-Fletcher, N. Gruber, J. L. Sarmiento, Goulden, M. L., J. W. Munger, S. M. Fan, B. C. Daube, and S. C. and M. Gloor (2007), A joint atmosphere-ocean inversion for Wofsy (1996), Measurements of carbon sequestration by long surface fluxes of carbon dioxide: 1. Methods and global-scale term eddy covariance. Methods and a critical evaluation of acfluxes, Global Biogeochem. Cycles, 21, GBI019, doi:l0.l029/ curacy, Global Change BioI., 2, 169-182. 2005GB002556. Goulden, M. L., S. D. Miller, and H. R. da Rocha (2006), Nocturnal Keeling, C. D., S. C. Piper, andM. Heimann (1989), A three-dimencold air drainage and pooling in a tropical forest, 1. Geophys. sional model of atmospheric CO 2 transport based on observed Res., 111, D08S04, doi: 1O.l029/2005JD006037. winds: 4, Mean annual gradients and interannual variations, Grace, J., et al. (1995a), Carbon dioxide uptake by an undisturbed in Aspects of Climate Variability in the Pacific and the tropical rain forest in southwest Amazonia, 1992 to 1993, Sciern Americas, Geophys. Monogr. Sa., vol. 55, edited by D. H. ence, 270, 778-780. Peterson, pp. 305-363, AGU, Washington, D. C. Grace, J., J. Lloyd, J. McIntyre, A. C. Miranda, P. Meir, H. S. Keller, M., and P. Crill (2000), Site Scouting and Selection Miranda, C. R Nobre, J. Moncrieff, I. R. Wright, and J. H. C. the Tapaj6s National Forest and SantaremlBelterra, Para. Gash (1995b), Fluxes of carbon dioxide and water vapour over
HOUGHTON ET AL. published report, available at http://lba.cptec.inpe.br/lba/eng/ research/santaremJepoli/santarem2.html). Keller, M., D. A. CI~p<:, D. B. Clark, A. M. Weitz, and E. Veldkamp (1996), If ~ 'tree falls in the forest ... , Science, 273, 20 I, doi: 1O.1126/sciylice.273.5272.201. Kindenllann, J.vii. Wurth, G. H. Kohlmaier, and F.-W. Badeck (1996), Interannual variation of carbon exchange fluxes in terrestrial ecosystems, Global Biogeochem. Cycles, 10(4), 737-755. Kruijt, B., J. A. Elbers, C. von Randow, A. C. Araujo, P. 1. Oliveira, A. Culf, A. O. Manzi, A. D. Nobre, P. Kabat, and E. 1. Moors (2004), The robustness in eddy correlation fluxes for Amazon rainforest conditions, Ecol. Appl., 14, SI01-S113. Kucharik, C. J., J. A. Foley, C. Delire, V. A. Fisher, M. T. Coe, J. D. Lenters, C. Young-Moiling, N. Ramankutty, J. M. Norman, and S. T. Gower (2000), Testing the performance of a dynamic global ecosystem model: Water balance, carbon balance, and vegetation structure, Global Biogeochem. Cycles, 14(3), 795-825. Kuck, L. R., et al. (2000), Measurements oflandscape-scale fluxes of carbon dioxide in the Peruvian Amazon by vertical profiling through the atmospheric boundary layer, 1. Geophys. Res., 105, 22,137-22,146. Kuhn, U., et al. (2007), Isoprene and monoterpene fluxes fi'om Cenh'al Amazonian rainforest infened from tower-based and airborne measurements, and implications on the atmospheric chemishy and the local carbon budget, Atmos. Chem. Phys., 7,2855-2879. Langenfelds, R L., R. J. Francey, B. C. Pak, L. P. Steele, J. Lloyd, C. M. Trudinger, and C. E. Allison (2002), Interannual growth rate variations of atmospheric CO 2 and its 13 C, H 2, CH4 , and CO between 1992 and 1999 linked to biomass burning, Global Biogeochem. Cycles, 16(3),1048, doi:10.1029/200IGBOOI466. Laubach, J., and H. Fritsch (2002), Convective boundary layer budgets derived fi'om aircraft data, Agric. For. Meteorol. 111, 237-263. Lewis, S. L., et al. (2004), Concerted changes in tropical forest stmcture and dynamics: Evidence from 50 South American long-term plots', Phi/os. Trans. R. Soc. Ser. B, 359, 421--436. Lewis, S. L., O. L. Phillips, and T. R Baker (2006), Impacts of global atmospheric change on h'opical forest, Trends Ecol. Evol., 21,173-175. Lloyd, J. (1999), The CO2 dependence of photosynthesis, plant growth responses to elevated CO 2 concentrations and their interactions with soil nutrient status II. Temperate and boreal forest productivity and the combined effects of increasing CO 2 concentrations and increased nitrogen deposition at a global scale, Funct. Ecol., 13, 439--459. Lloyd, J., and G. D. Farquhar (1996), The CO 2 dependence ofphotosynthesis, plant growth responses to elevated atmospheric CO 2 concentrations and their interaction with plant nutrient status, Funct. Ecol., 10, 4-32. Lloyd, J., and G. D. Farquhar (2000), Do slow-growing species and nutrient-stressed plants consistently respond less to elevated CO2 ? A clarification of some issues raised by Poolier (1998), Global Change BioI., 6, 871-876. Lloyd, J., and G. D. Farquhar (2008), Effects of rising temperatures and [C0 2] on the physiology of h'opical forest trees. Trans. R. Soc. London, Ser. B., 363,1811-1817.
o
425
Lloyd, J., J. Grace, A. C. Miranda, P. Meir, S.-C. Wong, H. S. Miranda, I. R. Wright, J. H. C. Gash, and J. A. MacIntyre (1995), A simple calibrated model of Amazon rainforest productivity based of leaf biochemical propeliies, Plant Cell Environ., 18, 1129-1145. Lloyd, J., et al. (1996), Vegetation effects on the isotopic composition of atmospheric CO 2 as local and regional scales: Theoretical aspects and a comparison between rainforest in Amazonia and a boreal forest in Siberia, Aust. Plant Physiol., 23,371-399. Lloyd, J., et al. (2001), Vertical profiles, boundary layer budgets, and regional flux estimates for CO2 and its 13 C/ 12 C ratio and for water vapor above a forest/bog mosaic in central Siberia, Global Biogeochem. Cycles, 15(2), 267-284. Lloyd, J., et al. (2007), Airborne estimates of the Amazon carbon balance, Biogeosciences, 4, 759-768. Lloyd, J., E. U. Gloor, and S. L. Lewis (2009a), Are the dynamics of tropical forests dominated by large and rare disturbance events?, Ecol. Lett., in press. Lloyd, J., M. L. Goulden, J. P. Ometto, S. Patino, N. M. Fyllas, and C. A. Quesada (2009b), Ecophysiology offorest and savanna vegetation, Geophys. Monogr. Ser., doi:l0.102912008GM000740, this volume. Loescher H. W., B. E. Law, L. Marui, D. Y. Hollinger, 1. Campbell and S. C. Wofsy (2006), Uncertainties in, and interpretation of, carbon flux estimates using the eddy covariance technique, 1. Geophys. Res., 111, D21S90, doi:10.102912005JD006932. Longo, K. M., S. R! Freitas, M. O. Andreae, R. Yokelson, and P. Artaxo (2009), Biomass buming in Amazonia: Emissions, longrange transpOli of smoke and its regional and remote impacts Geophys. Monogr. Ser., doi:l0.1029/2008GM000847, this vol~ ume. Machado, L. A. T., H. Laurent, and A. A. Lima (2002), Diurnal march of the convection observed during TRMM-WETAMCILBA, 1. Geophys. Res., 107(D20), 8064, doi:IO.1029/200IJD000338. Machado, L., H. Laurent, N. Dessay, and I. Miranda (2004), Seasonal and diurnal variability of convection over the Amazonia: A comparison of different vegetation types and large scale forcing, TheaI'. Appl. Climatol., 78,61-77. Malhi, Y., and J. Grace (2000), Tropical forests and atmospheric carbon dioxide, Trends Ecol. Evol., 15,332-337. Malhi, Y., A. D. Nobre, J. Grace, B. Kmijt, M. G. P. Pereira, A. Culf, and S. Scott (1998), Carbon dioxide transfer over a Central Amazonian rain forest, 1. Geophys. Res., 103, 31,593-31,612. Malhi, Y., et al. (2004), The above-ground coarse wood productivity of 104 Neotropical forest plots, Global Change BioI., 10, 563-591. Malhi, Y., et al. (2006), The regional variation of aboveground live biomass in old-growth Amazonian forests, Global Change BioI., 12,1107-1138. Mayorga, E., A. K. Aufdenkampe, C. A. Masiello, A. V. Kmshe, J. I. Hedges, P. D. Quay, J. E. Richey, and T. A. Brown (2005), Young organic matter as a source of carbon dioxide outgassing from Amazonian rivers, Nature, 436, 538-541. McGuire, A. D., J. M. Melillo, D. W. Kicklighter, and L. A. Joyce (1995), Equilibrium responses of soil carbon to climate
.r.
426
HOUGHTON ET AL.
REGIONAL CARBON BUDGET
change: Empirical and process-based estimates, J. Biogeogr., 22, 785-796. Meggers, B. 1. (1994), Archeological evidence for the impact of mega-Nino eyents on Amazonia during the past two millennia, Clim. Change, 28, 321-338. Melack, 1. M., E. M. L. M. Novo, B. R. Forsberg, M. T. F. Piedade, and L. Maurice (2009), Floodplain ecosystem processes, Geophys. Monogr. Ser., doi:lO.l029/2008GM000721, this volume. Melillo, J. lVI., A. D. McGuire, D. W. Kicklighter, B. Moore, C. J. Vorosmarty, and A. L. Schloss (1993), Global climate change and terrestrial net primary production, Nature, 363, 234-240. Mercado, L. M., 1. Lloyd, F. Carswell, Y. Malhi, P. Meir, and A. D. Nobre (2006), Modelling Amazonian forest eddy covariance data: A comparison of big leaf versus sun/shade models for the C-14 tower at Manaus I. Canopy photosynthesis, Acta Amazonica, 36, 69-82. Mercado, L. M., 1. Lloyd, A. 1. Dolman, S. Sitch, and S. Patino (2009), Modelling basin-wide variations in Amazon forest productivity I. Model calibration, evaluation and upscaling functions for canopy photosynthesis, Biogeosci. Discuss., 6, 2965-3030. Mikaloff Fletcher, S. E., et al. (2007), Inverse estimates of the oceanic sources and sinks of natural CO 2 and the implied oceanic carbon transport, Global Biogeochem. Cycles, 21, GB101O, doi: 1O.1029/2006GB002751. Miller, S. D., M. L. Goulden, M. C. Menton, H. R. Rocha, H. C. Freitas, A. M. Figueira, and C. A. D. Sousa (2004), Biometric and micrometoro10gica1 measurements of tropical forest carbon balance, Ecol. Appl., 14, Sl14-S126. Miranda, A. C., H. S. Miranda, 1. Lloyd, J. Grace, R 1. Francey, P. Riggan, and 1. Brass (1996), Fluxes of carbon dioxide and water vapour over cerrado vegetation in Central Brazil. An analysis using eddy cOlTe1ation and stable isotope techniques, Plant Cell Environ, 20, 315-328. Mokany, K., R. J. Raison, and A. S. Prokushkin (2006), Critical analysis of root:shoot ratios in terrestrial ecosystems, Global Change Bioi., 12, 84-96. Morison, 1. 1. L., M. T. F. Piedade, E. Muller, S. P. Long, W. 1. Junk, and M. B. Jones (2000), Very high productivity of the C4 aquatic grass Echinochloa polystachya in the Amazon floodplain confirmed by net ecosystem CO 2 flux measurements, Oecologia, 125(3),400-411, doi: 10.1007/s004420000464. MOlion, D. C., R. S. DeFries, Y. E. Shimabukuro, L. O. Anderson, E. Arai, F. del Bon Espirito-Santo, R. Freitas, and 1. Morisette (2006), Cropland expansion changes deforestation dynamics in the southem Brazilian Amazon, Proc. Nat!. Acad. Sci. U. S. A., 103(39), 14,637-14,641. Neef, T., R. M. Lucas, J. R. dos Santos, E. S. Brondizio, and C. C. Freitas (2006), Area and age of secondmy forests in Brazilian Amazonia 1978-2002: An empirical estimate, Ecosystems, 9, 609-623. Nemani, R. R., C. D. Keeling, H. Hashimoto, W. M. Jolly, S. C. Piper, C. 1. Tucker, R B. Myneni, and S. W. Running (2003), Climate-driven increases in global terrestrial net primmy production from 1982 to 1999, Science, 300, 1560-1564. Nepstad, D. C., C. 1. R. d. Carvalho, E. A. Davidson, P. Epp, P. A. Lefebvre, G.H.d. Negreiros, E. D. da Silva, T. A. Stone, S. E.
Trumbore, and S. Vieira (1994), The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures, Nature, 372, 666-669. Nepstad, D., I. Tohver, D. Ray, P. Moutinho, and G. Cardinot (2007), Mortality of large trees and lianas following experimental drought in an Amazon forest, Ecology, 88, 2259-2269. Nepstad, D. C., C. M. Stickler, B. Soares-Fi1ho, and F. Meny (2008), Interactions among Amazon land use, forests and climate: Prospects for a near-term forest tipping point, Phi/os. Trans. R. Soc. Ser. B., 363, 1737-1746, doi: 10.1098/rstb.2007.0036. Nogueira, E. M., P. M. Fearnside, B. W. Nelson, R. 1. Barbosa, and E. W. H. Keizer (2008), Estimates of forest biomass in the Brazilian Amazon: New allometric equations and adjustments to biomass from wood-volume inventories, For. Ecol. Manage., 256(11), 1853-1857. Oliveira, P. J., E. 1. P. da Rocha, G. A. Fisch, B. Kruijt, and 1. B. M. Ribeiro (2004), Efeitos de um evento de friagem nas condiyoes meteorol6gicas na amazonia: Um estudo de caso, Acta Amazonica, 34(4), 613-619. Oliveira, P. H. F., P. Artaxo, C. Pires, S. de Lucca, A. Procopio, B. Holben, 1. Schafer, L. F. Cardoso, S. C. Wofsy, and H. R. Rocha (2007), The Effects of biomass burning aerosols and clouds on the CO 2 flux in Amazonia, Tellus, Ser. B, 59, 338-349. Ometto, J. P. H. B., A. Nobre, H. R. Rocha, P. Ariaxo, and L. A. Martinelli (2005), Amazonia and the modern carbon cycle: Lessons learned, Oecologia, 143, 483-500, doi:10.1007/s00442005-0034-3. Oyama, M. D., and C. A. Nobre (2003), A new climate-vegetation equilibrium state for Tropical South America, Geophys. Res. Lett., 30(23), 2199, doi:10.1029/2003GL018600. Pm'olin, P., et al. (2004), Central Amazonian Floodplain Forests: Tree adaptations in a pulsing system, Bot. Rev., 70(3),357-380. Peters, W., M. C. Kro1, E. J. Dlugokencky, F. 1. Dentener, P. Bergamaschi, G. Dutton, P. v. Velthoven, 1. B. Miller, L. Bruhwiler, and P. P. Tans (2004), Toward regional-scale modeling using the two-way nested global model TM5: Characterization of transport using SF 6 , J. Geophys. Res., 109, D19314, doi:lO.l029/ 2004JD005020. Phillips, O. L., et al. (1998), Changes in the carbon balance of tropical forest: Evidence from long-term plots, Science, 282, 439-442. Phillips, O. L., et al. (2002), Changes in growth of tropical forests: Evaluating potential biases, Ecol. Appl., 12, 576-587. Phillips, O. L., N. Higuchi, S. Vieira, T. R. Baker, K.-J. Chao, and S. L. Lewis (2009), Changes in Amazonian forest biomass, dynamics, and composition, 1980-2002, Geophys. Monogr. Ser., doi:lO.l029/2008GM000739, this volume. Piedade, M. T. F., W. J. Junk, and S. P. Long (1991), The productivity of the C4 grass Echinochloa polystachya on the Amazon floodplain, Ecology, 72,1456-1463. Potter, C. S., S. A. Klooster, and V. Brooks (1999), Interannual variability in terrestrial net primary production: Exploration oftrends and controls on regional to global scales, Ecosystems, 2,36-48. Potter, C., S. Klooster, C. R. de Carvalho, V. B. Genovese, A. Torregrosa, 1. Dungan, M. Bobo, and 1. Coughlan (2001a), Modeling seasonal and interannual variability in ecosystem carbon
cycling for the Brazilian Amazon region, J. Geophys. Res., 106, 10,423-10,446. Potter, C. S., E. A. Dav\~§on, D. C. Nepstad, and C. R. Carvalho (200 1b), Ecosystem)ihodeling and dynamic effects of deforestation on trace gas flJixes in Amazon tropical forests, For. Eco!. Manage., 152,97""117. Potter, C., V. Brooks-Genovese, S. Klooster, M. Bobo, and A. Torregrosa (2001c), Biomass burning losses of carbon estimated from ecosystem modeling and satellite data analysis for the Brazilian Amazon region, Atmos. Environ., 35,1773-1781. Potter, C., S. Klooster, M. Steinbach, P. Tan, V. Kumar, S. Shekhar, and C. Carvalho (2004), Understanding global teleconnections of climate to regional model estimates of Amazon ecosystem carbon fluxes, Global Change Bioi., 10, 693-703. Prentice,1. C., and 1. Lloyd (1998), C-quest in Amazon Basin, Nature, 396, 619-620. Pyle, E. H., et al. (2008), Dynamics of carbon, biomass, and structure in two Amazonian forests, J. Geophys. Res., 113, GOOB08, doi:1 0.1029/2007JG000592. Quesada, c. A., 1. Lloyd, L. O. Anderson, M. Schwarz, and C. 1. Czimczik (2009), Soils of Amazonia with particular reference to the RAINFOR sites, Biogeosci. Discuss., 6, 3851-3921. Ramankutty, N., H. K. Gibbs, F. Achard, R. DeFries, 1. A. Foley, and R A. Houghton (2007), Challenges to estimating carbon eJ1lissions fi'om tropical deforestation, Global Change Bioi., 13, 51-66. Raupach, M. R., and J. J .Finnigan (1995), Scale issues in boundmylayer meteorology: Surface energy balances in heterogeneous terrain, Hydrol. Processes, 9, 589-612. Raupach, M. R., O. T. Denmead, and F. X. Dunix (1992), Challenges in linking atmospheric CO 2 concentrations to fluxes at local and regional scales, Aust. J. Bot., 40, 697-716. Rayner, P., 1. Enting, R Francey, and R. Langenfelds (1999), Reconstructing the recent carbon cycle from atmospheric CO 2, ol3 C and 02IN2 observations, Tellus, Ser. B, 51,213-232. Rice, A. H., E. H. Pyle, S. R. Saleska, L. Hutyra, P. B. Carmargo, K. Portilho, D. F. Marques, and S. F. Wofsy (2004), Carbon balance and vegetation dynamics in an old-growth Amazonian forest, Eeal. Appl., 14, 855-871. Richey, 1. E., et al. (1990), BiogeochemiSl1y of carbon in the Amazon River, Limnol. Oceanogr., 35, 352-371. Richey, 1. E., 1. M. Melack, A. K. Aufdenkampe, V. M. Ballester, and L. L. Hess (2002), Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric C02, Nature, 416,617-620. Richey, 1. E., A. V. Krusche, M. S. Johnson, H. B. da Cunha, and M. V. Ballester (2009), The role of rivers in the regional carbon balance, Geophys. Monogr. Ser., doi:1O.1029/2008GM000734, this volume. Rocha, H: R, M. L. Goulden, S. D. Miller, M. C. Menton, L. D. V. O. Pinto, H. C. de Freitas, and A. M. S. Figueira (2004), Seasonality of water and heat fluxes over a tropical forest in eastern Amazonia, Ecol. App!., 14(4), S22-S32. Rodenbeck, C., S. Houweling, M. Gloor, and M. Heimann (2003), C02 flux histOly 1982-2001 inferred from atmospheric data using a global inversion of atmospheric transport, Atmos. Chem. Phys., 3, 1919-1964.
427
Saatchi, S. S., R. A. Houghton, R. C. dos Santos Alvah'!, 1. V. Soares, and Y. Yu(2007), Distribution of aboveground live biomass in the Amazon Basin. Global Change Bioi., 13, 816-837. Saleska, S. R., et al. (2003), Carbon in Amazon forests: Unexpected seasonal fluxes and disturbance-induced losses, Carbon in Amazon forests: Unexpected seasonal fluxes and disturbanceinduced losses, Science, 302, 1554-1557. Saleska, S. R., K. Didan, A. R Huete, and H. R. da Rocha (2007), Amazon forests green-up during 2005 drought, Science, 318, 612. Saleska, S., H. da Rocha, B. Kruijt, and A. Nobre (2009), Ecosystem carbon fluxes and Amazon forest metabolism, Geophys. Monogr. Ser., doi:10.1029/2008GM000728, this volume. Sampaio, G., C. Nobre, M. H. Costa, P. SatyamUliy, B. S. SoaresFilho, and M. Cardoso (2007), Regional climate change over eastem Amazonia caused by pasture and soybean cropland expansion, Geophys. Res. Lett., 34, Ll7709, doi:10.1029/2007GL030612. Santos, A. J. B., G. T. da Silva, H. S. Miranda, A. C. Miranda, and 1. Lloyd (2003), Effects of fire on surface carbon, energy and water vapour fluxes over campo sujo savanna in Central Brazil, Funct. Eeal., 17,711-719. Stephens, B. 8., et al. (2007), Weak northem and sl1'ong tropical land carbon uptake from vertical profiles of atmospheric CO 2, Science, 316,1732-1735. Takahashi, T., R. H. Wanninkhof, R. A. Feely, R F. Weiss, D. W. Chipman, N. Bates, J. Olafsson, C. Sabine, and S. C. Sutherland (1999), Net sea-air CQ2 flux over the global oceans: an improved estimate based on the sea-air CO 2 difference, in Proceedings of the 2nd International Symposium, CO2 in the Oceans, edited by Y. Nojiri, pp. 9-14, <J:enter for Global Environnlental Research, National Institute for Environmental Studies, Tsukuba. Tans, P. P., 1. Y. Fung, and T. Takahashi (1990), Observational constraints on the global atmospheric carbon dioxide budget, Science, 247,1431-1438. Taylor, 1., and J. Lloyd (1992), Sources and sinks of CO 2, Aust. J. Bot., 40, 407-418. Telles, E. de C. C., P. B. de Camargo, L. A. Martinelli, S. E. Trumbore, E. S. da Costa, 1. Santos, N. Higuchi, and R. C. Oliveira (2003), Influence of soil texture on carbon dynamics and storage potential in tropical forest soils of Amazonia. Global Biogeochem. Cycles, 17(2),1040, doi:l0.1029/2002GB001953. Tian, H., 1. M. Melillo, D. W. Kicklighter, A. D. McGuire, 1. V. K. Helfi'ich, B. Moore, and C. 1. Vorosmarty (1998), Effect of interannual climate variability on carbon storage in Amazonian ecosystems, Nature, 396, 664-667. Toivonen, T., S. Maki, and R. Kalliola (2007), The riverscape of Western Amazonia-A quantitative approach to the fluvial biogeography of the region, J. Biogeogr., 34,1374-1387. Tom, M. S., S. E. Trumbore, O. A. Chadwick, P. M. Vitousek, and D. M. Hendricks (1997), Mineral control of organic carbon storage and turnover, Nature, 389, 170-172. Tmmbore, S., and P. B. de Camargo (2009), Soil carbon dynamics, Geophys. Monogl'. Ser., doi: 10. 1029/2008GM000741 , this volume. Van der Werf, G. R, 1. T. Randerson, G. 1. Collatz, and L. Giglio (2003), Carbon emissions from fires in tropical and subtropical ecosystems, Global Change Bioi., 9, 547-562.
428
REGIONAL CARBON BUDGET
Vieira, S., et al. (2004), Forest stmcture and carbon dynamics in Amazonian tropical rain forests, Oecologia, 140,468--479. Vourlitis, G. L., 1. de Souza Nogueira, N. Priante Filho, W. Hoeger, F. Raiter, JvIl: Sacardi Biudes, 1. C. Armda, V. Buscioli Capistrano, 1. L. Brito de Faria, and F. de Almeida Lobo (2005), The sensitivity of diel C02 and H20 vapor exchange of a tropical transitional forest to seasonal variation in meteorology and water availabiFty, Earth Interact., 9, 27. Walker, R., R. DeFries, M. del C. Vera-Diaz, Y. Shimabukuro, and A. Venturieri (2009), The expansion of intensive agriculture and ranching in Brazilian Amazonia, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000735, this volume. Waterloo, M. J., et al. (2006), EXpOli of organic carbon in mn-off from an Amazonian rainforest blackwater catchment, Hydrol. Processes, 20, 2581-2597. Wattel-Koekkoek, E. 1. W., P. Buunnan, 1. van der Plicht, E. Watter, and N. van Bremen (2003), Mean residence time of soil organic matter associated with kaolinite and smectite, Eur. J. Soil Sci., 54, 269-278. Williams, E., A. Dall' Antonia, V. Dall' Antonia, J. M. Almeida, F. Suarez, B. Liebmann, and A. C. M. Malhado (2005), The drought of the century in the Amazon Basin: An analysis of the
regional variation of rainfall in South America in 1926, Acta Amazonica, 35, 238-238. Wilson, K. B., et al. (2002), Energy partitioning between latent and sensible heat flux during the warm season at FLUXNET sites, Water Resoll/', Res., 38(12), 1294, doi: 10.1 02912001WR000989. Wofsy, S. C., R. C. HalTiss, and W. A. Kaplan (1988), Carbon dioxide in the atmosphere over the Amazon Basin, J. Geophys. Res., 93, 1377-1387. Worbes, M. (1997) The forest ecosystem ofthe floodplains, in The Central Amazon Floodplain. Ecology ola pulsing system, edited by W. 1. Junk, pp. 187-206, Springer Verlag, Berlin. Wright, 1. (2005), Tropical forests in a changing environment, Trends Ecol. Evol., 20, 553-555.
M. Gloor and J. Lloyd, School of Geography, University of Leeds, Leeds LS2 9JT, UK. R. A. Houghton, Woods Hole Research Center, 149 Woods Hole Road, Falmouth, MA 02540, USA. ([email protected]) C. Potter, Ecosystem Science and Technology Branch, NASA Ames Research Center, MS 242-4, Moffett Field, CA 94035, USA.
The Effects of Drought on Amazonian Rain Forests P. Meir,l P. M. Brando,2,3,4 D. Nepstad,5 S. Vasconcelos,6 A. C. L. Costa,7 E. Davidson,4 S. Almeida,8 R. A. Fisher,9 E. D. Sotta,1O D. Zarin,2 and G. Cardinot 11
The functioning ofAmazonian rain forest ecosystems during drought has become a scientific focal point because of associated risks to forest integrity and climate. We review current understanding of drought impacts on Amazon rain forests by summarising the results fi'om two throughfall exclusion (TFE) experiments in old-growth rain forests at Caxiuana and Tapaj6s National Forest Reserves, and an irrigation experiment in secondary forest, near Castanhal, Brazil. Soil physical properties strongly influenced drought impacts at each site. Over years 1 to 3 of soil moisture reduction, leaf area index declined by 20-30% at the TFE sites. Leaf physiology and tree mOliality i'esults suggested some species-based differences in drought resistance. Mortality was initially resistant to drought but increased after 3 years at Tapaj6s to 9%, followed by a decline. Transpiration and gross primary production were reduced under TFE at Caxiuana by 30--40% and 1213%, respectively, and the maximum fire risk at Tapaj6s in,creased from 0.27 to 0.47. Drought reduced soil CO 2 emissions by more than 20% at Caxiuana and Castanhal but not at Tapaj6s, where N 20 and CH4 emissions declined. Overall, the results indicate shOli-term resistance to drought with reduced productivity, but that increased mortality is likely under substantial, multiyear, reductions in rainfall. These data sets from field-scale experimental manipulations uniquely complement existing observations from Amazonia and will become increasingly powerful if the experiments are extended. Estimating the long-term (decadal-scale) impacts of continued drought on Amazonian forests will also require integrated models to couple changes in vegetation, climate, land management, and fire risk.
ISchool of Geosciences, University of Edinburgh, Edinburgh, UK. 2Department of Botany, University of Florida, Gainesville, Florida, USA. 3Instituto de Pesquisa Ambiental da Amazonia, Belem, Brazil.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2009GM000882
4Woods Hole Research Center, Falmouth, Massachusetts, USA. 5Gordon and Betty Moore Foundation, Palo Alto, California, USA. 6EMBRAPA-Amazonia Oriental, Belem, Brazil. 7Centro de Geociencias, Universidade Federal do Pani, Belem, Brazil. 8Museu Paraense Emilio Goeldi, Belem, Brazil. 9Los Alamos National Laboratory, Los Alamos, New Mexico, USA. lOEMBRAPA-Amapa, Macapa, Brazil. llInstituto de Pesquisa Ambiental da Amazonia, Brazil. 429
430
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
1. INTRODUCTION Over the last decade, the possible impacts of drought have become toudl"stone issues for environmental science and governance in Amazonia. The geographical size, biophysical properties, and species diversity of Amazonian forests have led to analysis of their role as providers of key environmental service's at all scales. Increased moisture limitation in the region is likely this century, and some of the services provided by forests, including the storage and sequestration of carbon, evapotranspiration, and the maintenance of species diversity are potentially at risk. Human natural resource use intersects with these roles because of rapid land use change and tends to magnify the likelihoods of drought and forest degradation through fire. There is uncertainty with respect to all of these outcomes, governed by two core questions: (1) How likely, widespread, and severe is future drought? (2) What is the likely impact of drought on forest ecosystem properties? In this chapter, we address the latter question considering alterations to the carbon balance, transpiration, tree mOliality, species-based differences in drought responses, and vulnerability to fire. Evidence of the response by forests to drought is often based on observation of forest processes during natural droughts. However, unusually for any region, three large-scale soil moisture manipulation experiments have been implemented in Amazonia during the Large-Scale Biosphere-Atmosphere (LBA) Experiment in Amazonia. These experiments are providing new tests of the modeled response by rain forest ecosystems to levels of drought that are beyond the bounds of recent climatic variation, but in line with some future climate scenarios. A comparison and analysis of the first results emerging from these experiments forms the core of this chapter and is used to summarize a view of the risks to, and responses by, Amazonian rain forests experiencing drought during the twenty-first century. 2. BACKGROUND 2.1. Future Drought?
Two main lines of evidence suggest that Amazonian drought may become more fi'equent and more severe during this century. First, episodic drought has been associated with the occurrence ofthe El Nino-Southern Oscillation (ENSO), caused by the warming of the tropical eastern Pacific Ocean, and more recently in 2005, with abnormal warming of the northern tropical Atlantic Ocean, relative to the south [Cox et al., 2008; Marengo et al., 2008]. Future increases in greenhouse gas concentrations coupled with reductions in global aerosol emissions may increase the likelihood of 2005-like
drought events [Cox et al., 2008] and possibly also ENSO events [Timmermann, 1999]. More generally with respect to climatic change over the twenty-first century, the multimodel data set (MMD) used in the IPCC Fourth Assessment Report projected an annual mean warming in Amazonia of 1.8 0 -5.1 DC over this century, with rainfall reductions in parts of central and eastern Amazonia likely as a result, especially in the dry season [IPCC Working Group I, 2007]. In a recent analysis involving 23 MMD models, Malhi et al. [2008] reported 20-70% agreement among models in the prediction of substantial dry season reductions in precipitation across Amazonia, with the greatest likelihood of drought in the east of the region. Second, land use change is likely to exacerbate the effects of climatic warming. Widespread forest conversion to pasture and agriculture is expected to reduce rainfall in the region through differential effects on latent and sensible heat transfer [Werth and AVissar, 2002; Chagnon and Bras, 2005; Costa et al., 2007; Nobre et al., 1991] and increased regional atmospheric aerosol loading could cause widespread reductions in precipitation [IPCC Working Group 1,2007]. Overall, these results imply an increased fi'equency of extreme events at the seasonal and interannual timescale, and a secular shift to drought at decadal timescales, the strength of any shift strongly influenced by land use change. 2.2. Modeling Drought Impacts
Estimates of the impact of drought on Amazonian forests have historically been made with limited field data, and this has contributed to uncertainty in estimates of future global atmospheric CO 2 concentrations [Meir et al., 2006; 1PCC Working Group 1,2007; Huntingford et al., 2009]. Recent analysis of the bioclimatic distribution of current natural Amazonian vegetation and the predictive output from 19 global circulation models (GCMs) suggests that twenty-first centmy climate change is most likely to lead to drier conditions more appropriate for seasonal forest in eastern Amazonia [Malhi et al., 2009], although edaphic conditions may in reality favor a transition to degraded forest or savanna. In contrast, smaller impacts are likely on vegetation in western Amazonia [Malhi et al., 2009]. Driving a vegetation model with a variety of climate scenarios, Sampaio et al. [2007] predicted a similar but more extreme range of forest-to-savanna switches following climatic warming and drying. In both cases, these decadal-centUlY timescale scenarios were based on some form of "equilibrium" vegetation response to drought. In reality, the actual vegetation response will be determined (rapidly or gradually) by multiple ecological and physical processes, as acknowledged by the same authors.
MEIR ET AL.
By contrast, process-based dynamic vegetation models have the stmcture to c,pture the relevant ecology, enabling land surface-atmosphere interactions to be modeled on a continuous basis. ~owever, they are computationally expensive and ultir}Jl{tely require observation-based, ecological parameterization [Prentice and Lloyd, 1998; Meir et al., 2008]. The first predictions of substantial Amazonian dieback in response to warming and drought emerged fi'om a global dynamic vegetation model (DGVM) framework and were made at the decadal-century timescale, with relatively simplistic representations ofcanopy stmcture, soil processes, and functional diversity among species [White et al., 1999; Cox et al., 2000]. Large differences in parameterization and mathematical description of core ecosystem processes have since been identified among DGVMs [Dufresne et al., 2002; Meir et al., 2006; Friedlingstein et al., 2006; B. Poulter et a1., Managing uncertainty of tropical Amazon dieback, submitted to Global Change Ecology, 2009; D. Galbraith et a1., Quantifying the contributions of different environmental factors to predictions of Amazonian rain forest dieback in tlu'ee dynamic global vegetation models (DGVMs), submitted to Global Change Biology, 2009], emphasizing the need for improvements using field-based measurement; and experimentation. Nondynamic, but process-based, vegetation and biogeochemical models have also been used to advance the representation of Amazonian forest functioning [Lloyd et al., 1995; Williams et al., 1998; Potter et al., 2004], although differences in process description have been apparent [Tian et al., 1998; Prentice and Lloyd, 1998; Foley et al., 2002; Zeng et al., 2005; Meir et al., 2008]. Only recently have detailed validations been made of the modeled response to drought and with some success over seasonal to interannual timescales [Fisher et al., 2007; Baker et al., 2008]. The validation of longer-tenn vegetation dynamics is more difficult because it requires data sets at the scale of years to decades, representing multiple ecological processes. The multiyear soil moisture manipulation experiments described here provide unique insight into some of the relevant physiological and ecological processes. Also, they indicate the potential first steps in quantifying natural lags in vegetation change as a function of tree species' drought resistance and regeneration characteristics. However, vegetation change under climatic drying is influenced strongly by fire and deforestation. The area of burnt understory forest during the ENSO of 1998 (2.9 x 10 6 ha) was more than 10 times greater than during an average rainfall year and twice the area of annual deforestation [Nepstad et al., 1999; Alencar et al., 2006], while a burnt area up to 2800 km2 was attributed to fire leakage alone in the 2005 drought [Aragao et al., 2007]. Fire is more probable in forest that has burned
431
previously [Cochrane, 2003], and the fragmentation effects of deforestation tend to increase fire risk strongly [Uhl and Kmiffman, 1990; Cochrane alld LClUrallce, 2002]. Deforestation has repeatedly been shown to affect regional climate, though the effects vary with scale [Werth alld Avissar, 2002; Chagnon and Bras, 2005]. At large scale, deforestation scenarios result in modeled reductions in precipitation and relative humidity, and increases in temperature [e.g., Hof!mann et al., 2003; Costa et al., 2007]. Strong increases in fire risk (20-120%) have thus been associated with scenarios of partial and complete deforestation [Cardoso et al., 2003]. While we need to understand the responses to drought by Amazonian rain forest, the interactions between drought and increased fire risk must also be considered, as fire may well be the agent that, in the context of secular or episodic climatic drying, triggers the switch from a forest to a savanna [Hutyra et al., 2005; Aragao et al., 2007]. 2.3. Basin-Wide Observations ofthe Forest Carbon Balance During Drought
Inversion studies have provided the largest scale of observation-based information on drought effects on Amazonia. Despite the limitations of this method [see Houghton et al., this volume], a net ~mission of CO 2 from the region of up to 1.5 Pg C a-I has been reported for dry and warm periods during strong ENSQs [Bousquet et al., 2000, Rodenbeck et al., 2003, Zeng et cil., 2005, see also Houghton et al., this volume], although it is unclear if these higher emissions are principally the result of changes in ecosystem carbon cycling or increased fire occurrence [Langenfelds et al., 2002; Meir et al., 2008]. Results from biogeochemical modeling of the forest carbon cycle have been consistent with the observation of net regional CO 2 ,emissions during ENSO [e.g., Tian et al., 1998; Foley et al., 2002; Zeng et al., 2005], but the mechanisms underpinning these modeled results have tended to overemphasize the role of the temperature sensitivity of soil respiration during drought [Meir et al., 2008]. Phillips et al. [2009] recently reported basin-wide long-term observations of Amazonian forest productivity, estimating an overall negative impact of 1.2-1.6 Pg C in old growth forests during the 2005 drought, which was driven by lower growth and (spatially patchy) increases in tree mortality. This outcome is consistent with the notion that gross primary production (GPP) declines under drought, but without ,measurements of concurrent soil processes (e.g., CO 2 efflux from soil), it does not resolve the question of the short-term impact of the 2005 drought on net ecosystem productivity (NEP). [NEP is the difference between carbon that is photosynthesised and respired by an ecosystem. NEP = GPP ReeD, (ReeD = total ecosystem respiration).]
432
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
In summary, analysis of basin-wide atmospheric and tree growth data has highlighted the importance of understanding NEP at different timescales, and identified large changes in possible ec6system functioning during interannual-scale drought, including reductions in aboveground productivity, increases in mOltality in response to drought, and differential tree species' survival [cf. Engelbrecht et al., 2007]. These data will be invaluable to validate estimates of tree growth from DGVMs or finer-scale vegetation models, but alone, they do not pinpoint the mechanisms determining large-scale CO 2 exchange observed during intermittent severe drought [e.g., Rodenbeck et al., 2003]. In particular, C02 emission processes have been hard to pin down, first because respiration terms are often poorly quantified [Meir et al., 2008], and second, because the increases in fire occurrence identified during ENSO or the 2005 drought [Nepstad et al., 1999; Aragiio et al., 2007] are hard to quantify in terms of CO 2 emissions [van del' Welf et al., 2004]. 2.4. Observations at the Stand Scale
Some of the first eddy covariance measurements in Amazonia, quantifying the forest water and carbon cycles at the scale of a few square kilometers, suggested little seasonality in rain forest carbon exchange capacity [Grace et al., 1995]. However, although not universal, seasonal drought effects on NEP have since been observed at some sites, with observed changes in gross photosynthesis, respiration, and transpiration across sites differing in seasonality yielding new insights into the functional basis of the response [Malhi et al., 1998; Carswell et al., 2002; Saleska et al., 2003; Vourlitis et al., 2005; see da Rocha et al., this volume; Saleska et al., this volume; da Rocha et al., 2009]. Uniquely, experimental field manipulation ofsoil moisture enables the separation of the effects of othelwise naturally cov81ying environmental drivers (i.e., edaphic and atmospheric variables) ofthe processes determiningNEP and has the potential to provide powerful insight into process-level responses to short-term and extended drought. Three such field experiments have been performed within LBA, all in drought-threatened eastem Amazonia. Strong soil moisture deficit has been imposed over 1 ha of old-growth forest at two experimental sites, in the National Forest Reserves of Caxiuana (near Portel, State of Pal'll) and Tapaj6s (near Santan':m, State of Pant) [Nepstad et al., 2002; Fisher et al., 2007; Meir et al., 2008], and at a third site, dIy-season moisture stress has been reduced though the ilTigation of replicated 0.04-ha plots in second81Y regrowth forest near Castanhal, State of Pant [Vasconcelos et al., 2004]. The remainder of this chapter is concemed primarily with how the results of these experiments inform our understand-
ing of drought impacts on Amazonian rain forest, mostly in relation to the cycling of carbon and water, but also including the emission of other trace gases such as CH4, NO, N20, and isoprene. Where possible, we make a first comparison of the still-emerging experimental results. We ask three questions: (1) What have we discovered about the impact of drought at seasonal to interannual timescales? (2) What have we discovered about the impact of drought at multiyear to decadal timescales? (3) How does the combined impact of fire and drought influence the risk of widespread loss of forest in favor of smaller-stature vegetation types, such as savanna or degraded forest? 3. SOIL MOISTURE MANIPULATION EXPERIMENTS AT CAXIUAN.A, TAPAJ6S, AND CASTANHAL Species diversity is substantial (>150-200 species ha- 1) in the old-growth rain forests at both Caxiuana and Tapaj6s, while only four tree species (=70% of stems) dominate the regrowth forest at Castanhal. Standing biomass is much larger at the rain forest sites (approximately 300 t ha- 1 at Caxiuana, 240 t ha- 1 at Tapaj6s, and 50 t ha- 1 at Castanhal) [Baker et al., 2004; Vasconcelos et al., 2004; Brando et al., 2008]. Rain forest canopy heights are similar at 30-40 m, while the height of the second81Y forest at Castanhal is approximately 5 m [Coelho et al., 2004]. The soils at Caxiuana and Tapaj6s are highly weathered Oxisols, and at Castanhal, the soil is a dystrophic yellow Latosol. At Caxiuana, the soil composition is a sandy loam (70-83% sand) with the water table at 15-20 m depth; at Tapaj6s, the composition is more clay-rich and the soil profile is much deeper (60-80% clay; >100 m in depth); and at Castanhal, the soil is shallow and concretion81y, with a high sand content (20% clay, 74% sand). Annual precipitation at Caxiuana, Tapaj6s, and Castanhal is approximately 2300, 2000, and 2500 mm a-I, respectively. Further site descriptions are provided elsewhere [Davidson, 1992; Nepstad et al., 2002; Ruivo et al., 2003; Fisher et al., 2007; Vasconcelos et al., 2004]. The method ofphysically excluding rainfall that penetrates the canopy ("throughfall exclusion" (TFE)) was replicated at Caxiuana and Tapaj6s using approximately six thousand 4.5 m2 plastic panels and guttering placed at 2 m above the ground. The infrastructure removed approximately 50% of incoming precipitation [Nepstad et al., 2002; Fisher et al., 2007] (Table 1) and was installed at the beginning of 2000 at Tapaj6s and 2002 at Caxiuana. Each experiment comprised 1 ha ofTFE forest and 1 ha of undisturbed (nonmanipulated) "control" forest. The large scale ofthe manipulation was neceSS81Y because of substantial lateral extension of the surface roots of large trees. Treatment replication at both sites was limited by financial resources, but pretreatment calibration
MEIRETAL.
433
Table L Total Incoming Precipitation for Each Year of the Drought Treatment by DiY (Panels Off) and Wet/Season (Panels On)a Precipitation, mm
1'1
2000 2001 2002 2003
Period of Exclusion
Total Incoming
Panels Off
I Feb to 8 July 1 Jan to 31 July 1 Jan to 31 July 21 Jan to 14 Aug Totals
2517 1882 1958 1690 8047
830 171 292 394 1687
Panels On (Throughfall Excluded) 1687 (844) 1711 (856) 1665 (833) 1295(648) 6358 (3179)
aApproximately 50% of incoming precipitation was excluded when the panels were on; those volumes are presented in parentheses.
measurements were made in all plots to enable replication over time [Davidson et al., 2004], and the method follows the design of other unreplicated large-scale ecosystem manipulation experiments [e.g., Likens et al., 1970], whose strength is acknowledged, especially where large treatment effects are expected [Hurlbert, 2004]. The perimeter of the TFE plots was trenched to 1-2 m depth to prevent the horizontal ingress of water from adjacent normally watered soil, and the control plot perimeters were also trenched t~ avoid confounding treatment effects. Soil and plant measurements were made more than 20 m inside the perimeter of each plot to further eliminate confounding treatment effects. Management of the experiments was similar, except that at Tapaj6s, the panels were removed during the peak dry season, while at Caxiuana, this procedure was not followed because of the prevailing late dry season storm risk [Carswell et al., 2002]. Litterfall was manually retumed to the soil, where it fell on the TFE paneling. Full canopy access was provided using 40 m towers in all plots, also enabling the installation of on-site automatic weather stations. Principal access to the soil was provided by four soil shafts per plot, excavated to a maximum depth of 10 m at Caxiuana and 14 m at Tapaj6s. The irrigation experiment at Castanhal was designed to remove dly season moisture stress and began in 1999 [Vasconcelos et al., 2004]. Plot size was 20 m x 20 m (0.04 ha), and four replicate plots were used to contrast irrigated and undisturbed vegetation; adjacent plots were placed 10 m or more ap81t and a nested 10 m x 10 m plot in each 0.04 ha main plot was used for measurements. An additional treatment removing litter was also implemented at Castanhal, but is not discussed here. Ground-level irrigation was provided using perforated microtape installed at 2-m spacings; water was applied at a rate of approximately 5 mm d- 1 for 30 min during the dly season. The amount of irrigation was selected to approximately replace daily evapotranspiration estimated regionally (660-790 mm) [Jipp et al., 1998, Vasconcelos et al., 2004]. Surface soil moisture availability was measured
in all plots, and the relatively small stature of the vegetation allowed lower-canopy access for leaf water potential measurements [Fortini et al., 2003]. 4. SEASONAL TO INTERANNUAL DROUGHT IMPACTS 4.1. Soil Moisture and Its Supply to Plants
The change in soil moisture, in relation to adjacent undisturbed (control) fprest, was the main experimentally manipulated parameter in each experiment. The TFE infrastructure at Caxiuana:and Tapaj6s resulted in reductions in plant available water '(PAW) of 80-200 mm in the top 3 m of soil (Table 1 and Figure 1a) [Fisher et al., 2007, 2008; Brando et al., 2008], while the irrigation at Castanhal nearly completely removed the dry season moisture constraints [Vasconcelos et al., 2004]. A strong seasonality was evident in PAW at the TFE experiments, and the rate of soil moisture drawdown was larger over the first 1-3 years at Caxiuana than at Tapaj6s. The impact on PAW of manipulating precipitation input to the soil was strongly modified by the soil properties at each site. At Caxiuana, the sandy-loam composition created a relatively high moisture holding capacity per unit volume [CG/:s'well et al., 2002; Fisher et al., 2008], while the deep clay-rich soil at Tapaj6s potentially held substantial water reserves mainly because of its exceptional soil volume [Nepstad et al., 2002; Belk et al., 2007]. Detailed measurements of soil hydraulic properties [e.g., Tomasella and Hodnett, 1997] remain rare in Amazonia, but variation in the key soil parameters determining PAW, soil water potential, hydraulic conductivity, and moisture volume content, can cause large and basin-wide differences in the supply of water to plants' under moisture stress [Fisher et al., 2008]. Analogously, the shallow concretion81Y soil structure at Castanhal likely contributes to moisture limitation of root and micro-
434
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
MEIR ET AL. • Control - Tapaj6s o TFE - Tapaj6s Control - Caxiuana TFE - Caxiuana
o o
"
o
o (')
,
\ \
G~0"\
,
, \
\
\
\
\
o~_\_---Q
o
, !\
o
0
c1a
§m
~~
"'F
;¥-
a
~ J, .Let
'" I
I
'" '" ceo~
",'C'
~~ m"§
0'" l(!~ "'~ '"..iiI ~
1
",'C'
m-g
:3"
g~
l-
c!!:-
;¥-
c1ij
iM
a
g91
'" 10
'"
I
cal
a~
(a) 400
(j) (,,)
t: «l
n; .c
200
l!:!
::l
'~E E.§.
0
'(5 C/)
~ t:
'a::ro
-200 _
Monthly Rainfall Monthly Soil Water Balance ____ Annual Soil Water Deficit
~
-400 Jan 99
Jul 99
Jan 00
JulOO
Jan 01
Jul01
Jan 02
Jul 02
Jan 03
Jul03
Jan 04
Date
(b) Figure 1. (a) Plant available water (PAW) over wet and dly seasons in the conh'ol (filled symbols) and throughfall exclusion (TFE) plots (open symbols) for both experiments Tapaj6s (circles) and Caxiuana (triangles). The experiment in Caxiuana began in 200 I, and in Tapaj6s, it started in 1999 (pretreatment). PAW is the water available to plants between field capacity and a soil moisture potential of -1500 kPa. The arrow shows when TFE infi'astructure was installed at each site. The data points are derived from 48 formightly measurements made per soil access shaft, four access shafts per plot [Branda et al., 2008; Fisher et ar, 2007]. (b) Monthly natural rainfall, soil water balance, and annual soil water deficit at Caxiuana, 1999-2004. Maximum PAW at Caxuiana is ~400 mm water (a) [Fisher et al., 2008]; under normal rainfall, soil moisture supply is sufficient to support transpiration, but under strong drought, moisture limitation is likely.
bial activity during the natural dty season [Vasconcelos et al., 2004]. More soil hydraulics data from across the region are required to make satisfactory PAW calculations for the modeling ofvegeta~bn activity. The supply of 9'fbisture from soil to leaf is also affected by site differences in rooting properties. Rooting depth can be substantial, enabling increased access to soil water: roots have been detected' at depths> 14 m at Tapaj6s and at Caxiuana, to the approximate maximum depth of excavation, 9 m [Nepstad et al., 2004; Fisher et al., 2007]. As well as enabling fuller exploration of the soil profile, deep rooting can also facilitate hydraulic redistribution, thus further helping to maintain rhizosphere moisture availability and plant function under reduced rainfall, as observed at Tapaj6s in some instances [Oliveira et a1., 2005]. Initial data have not identified major changes in root dynamics at Tapaj6s in the TFE treatment [Brando et al., 2008]. However, responses in the modes of root growth at Caxiuana are partially consistent with theory [Joslin et al., 2000; Schymanski et al., 2008]. As the drought progressed at Caxiuana, surface roots (0-30 cm) tended to increase in length per unit mass, thus increasing the explored soil volume [Metcalfe et al" 2008], but it is less clear how plastic root growth responses were to changes in the vertical distribution of soil moisture availability within the soil profile, and answering this question will have implications for modeling resilience to drought. Species variation in these root propeliies differentially affects water acquisition among species and between sites, and variation in resistance to drought was also observed in the foliage. Leaves generally experience maximum daily moisture limitation at high atmospheric moisture deficit, soon after midday. Early aftemoon measurements of the minimum leaf water potential tolerated by tree species at Caxiuana, Tapaj6s, and Castanhal demonstrated differences among species (Table 2), although the minimum value measured in both TFE plots was similar (-3.2 and -2.7 MPa, respectively), indicating a possible maximum tolerance to moisture limitation in eastem Amazonian rain forest trees. Stem hydraulic conductance did not appear to be the major constraint to leaf water supply at low soil PAW, but species-based differences in this parameter were also observed at Caxiuana [Fisher et al., 2006] potentially further influencing species-based differences in resistance to soil moisture limitation [cf. Franks et al., 2007], In summary, substantial differences in soil propeIiies at each experimental site strongly influenced the storage ofwatel' in soil and its supply to plant roots under drought. The TFE results fi'om Tapaj6s and Caxiuana also highlighted mechanisms conferring tolerance to drought in terms of root, stem, and leaf hydraulic propeIiies. Variation among tree species in these responses to drought, including addi-
435
Table 2. Minimum Leaf Water Potential (Min 'VI) Measurements for Tree Species at the Caxiuana and Tapaj6s TFE Experiments and the Castanhallrrigation Experiment" Site Species Min 'Vl b, MPa T T T T C C C C Cs
Aparisthmium cordatum Astrocatyum gynacanthum Coussarea racemosa Poecilanthe effusa Licaria ameniaca HiJiela bicornis Lecythis confertiflora Swmizia racemosa Miconia ciliate
-1.5 -1.5 -2.7
-2.4 -2.2 -2.1 -2.9 -3.2 -3.0
"Abbreviations are C, Caxiuana; T, Tapaj6s; Cs, Castanhal. bYalues for T and C are taken from trees in the TFE plots, and values for Cs are from an understorey species (data from Fisher et al. [2006], Fortini et al. [2003], and G. Cardinot et al., manuscript in review, 2009).
tional evidence of direct uptake of dew in two species at the Tapaj6s experiment (G. Cardinot et aI., manuscript in review, 2009) suggest likely differences in survivorship under drought at multiyear or decadal time scales.
4.2. Canopy Structurf! and Productivity Although some savanna tree species have a deciduous or brevi-deciduous phehology [Furley et a1., 1992], this strategy is relatively unusual in rain forests, where leaf area index (LAI, m 2 1eaf area per m 2 unit ground area) is maintained under nOlmal climatic variation. There is some ground-based evidence for dIy-season increases in LA! [Cars,vell et al., 2002] and albedo [Culf et al., 1995], but irrespective of how this may affect forest functioning, it seems that PAW can often be maintained by forest trees through extensive and sometimes exceptionally deep rooting systems [Nepstad et al., 1994; Bruno et al., 2006]. However, under moisture constraints in excess of normal climatic variation, we have limited understanding ofthe limits to moisture access by forest trees. The TFE experiments thus provide a direct way of determining the thresholds in PAW that may lead to changes in canopy productivity, LAI, and fire vulnerability. The LAI at both sites was resistant to strong artificial soil moisture deficit for about a year following installation of the TFE infrastmcture, remaining at 5-6 m 2 m-2. After ,12 months, and a reduction in PAW of about 150-200 mm, LAI declined to 70-80% of the control (and original) values at each site (Figure 2), and this reduction was maintained subsequently following three more years of TFE (Brando et al., 2008, D. B. Metcalfe et aI., Impacts of experimentally imposed drought on leaf respiration and morphology
436
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
MEIR ET AL.
" I
437
A - Castanbal
Tapaj6s Caxiuana
to
0
f'-.
B - Tapaj6s
:>-LO '", ;''<1"
o
::z;
::::-C'>
{g
Q)(\j
to
0L-.-_ _- .
. -_ _-.c"'" 0'" Kl~
n £-
c"'" 0'" l3~ ~~ i::'-
0
,--_ _-.c@
~~
~§
U
,--_ _- . - - l
<1>'0
~~
~~ on,
o~
~c
'",0 c i::'lil
c@
c@
~:c
g
o~
~"E
:3 0
0
Figure 2. Variation in leaf area index (LAI) during the TFE experiments at Caxiuana and Tapaj6s. LAI is expressed as the quotient of (TFE plot LAI) / (Control plot LAI). LAI in undisturbed forest at both sites is 5-6 m2 m-2. The arrow denotes when the TFE infrastructure was installed at each site. (LAI was measured using a Li2000 (Licor, USA) at fixed 100 points [Branda et al., 2008; Fisher et al., 2007; Metcalfe et aI., submitted manuscript, 2009]).
f'-.
C - Caxiuana
to '
-r:-
(\j
0
in an Amazon rain forest, submitted to Functional Ecology, 2009). However, litter production patterns differed among sites, perhaps pariially reflecting differences in tree community responses to moisture limitation. At Caxiuana, litterfall in the TFE plot declined within the first 12 months following installation of the TFE and remained lower than undisturbed forest over the following dry seasons. At Tapaj6s in the TFE plot, and at Castanhal under inigation, litterfall did not change significantly and tracked the control forest litter flux rates closely in the first 2-3 years of rainfall manipulation (Figure 3). The reduction in LAI and subsequent litterfall at Caxiuana is consistent with reduced leaf regrowth, while the maintenance oflitterfall at Tapaj6s and Castanhal suggests a stronger limitation on leaf replacement. However, observed increases in leaf mass per unit area during experimental drought also contributed substantially to changes in LAI at Caxiuana and Tapaj6s [cf. Wright et al., 2006; Metcalfe et aI., submitted manuscript, 2009; Tohver et aI., unpublished data, 2007]. Finally, changes in leaf turnover rates, and the temporary impacts of mortality events after three years or more (data not shown) may help explain litterfall patterns over the longer term at Tapaj6s and Castanhal [Brando et al., 2008; Vasconcelos et al., 2008].
In summary, initial resistance to change in LAI during the first 12 months of soil drought at both TFE experiments was followed by substantial reductions in LAI of 20-30% over the following 2 years, and this was maintained subsequently. DIy season litterfall declined at Caxiuana relative to undistm'bed forest, but artificial droughting and inigation had only small effects on litterfall at Tapaj6s and Castanhal. Drought impacts on forest canopy physiology and stmcture have also been examined using remote sensing data products [Asner et al., 2004; Saleska et al., 2007; Huete et al., 2008, see also Salesk'a et al., this volume]. It remains unclear if these data can be used to reliably quantify changes in productivity during drought, but the TFE experiments provide a way of specifying models to test such estimates. The consequences of reduced LAI for alterations in gas exchange capacity and fire vulnerability are considered below. 4.3. Trace Gas Emissions From Soil
The flux of CO 2 from soil ("soil respiration," R s ), comprises the largest single respiratOly flux in the terrestrial carbon cycle and derives fi'om the combined respiration of heteroh'ophic (microbial, faunal) and autotrophic (root)
gia'
~~
b£
0-
§m l3F
cZ' 0'" l3,J!?
itt ;::-
~ ~ C:-Q. 0-
ill.!,
,
c"'" 0'"
l3~
n
<1>-
;::
Figure 3. Seasonallitterfall (±SE) at Caxiuana, Tapaj6s, and Castanhal from 1999 to 2004. The al1'0WS denote when the TFE or irri.gation treatments were installed. The data are derived from 20 litter traps (I m2) per plot; litterfall was collected fortmghtly or monthly at each site [Branda et al., 2008; Vasconcelos et al., 2008; Meir et al., manuscript in preparation, 2009].
components of the soil biological community [Trumbore, 2006; see also Trumbore and de Camargo, this volume]. R s has been reported for all three experiments, and additional measurements of NO, N 20, and CH4 have also been made at Tapaj6s and Castanhal. Consistent with biophysical expectation [Howard, 1979; Meir et al., 2008], seasonal declines in R s under reduced soil moisture have been observed widely in Latin American rain forests [Davidson et al., 2000; Schwendenmann et al., 2003]. While the experimental manipulation of soil moisture lowered R s strongly during the first 2 years at Caxiuana (by >20%, equivalent to >2 t C ha- I a-I; Figure 4) [Sotta et al., 2007], the response of R s to temperature was small and non-
significant [Sotta et al., 2007]. After 3-4 years, dry season R s in the Caxiuana TFE plot remained lower than for soil in undroughted forest, although overall between-plot differences in R s were smaller, perhaps because of increased wet-season root respiration rates in the TFE plot [Metcalfe et al., 2007]. At Castanhal, the difference in R s between irrigated and undisturbed soil reached maxima during the dry seasons, and annual R s in inigated plots was 13-27% larger than on undisturbed (drier) plots [Vasconcelos et al., 2004]. In contrast, at Tapaj6s, although the gross fluxes were not unusual in magnitude, R s was similar between the TFE and control plots throughout the experiment, even after 5 years ofTFE h'eatment [Davidson et al., 2008]. This outcome was
438
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
MEIR ET AL.
Castanhal
---.
(\J
IU) CjI
E
o E
(\J
0
:J
co
"'-"
X
::t=
'
oo
(\J
:J
W
'0 (f) ..-
L--------r;-:::~::;;;:_--------I Caxiuana
(\J
'7:; 5
0 ..-
"E 4
CO
8
~2 .2
'
~
3
l'51
0 0
2 log{-qt)·kPa
-10
0
10
20
Months from start of the treatment Fi ure 4. Soil respiration (C0 2 efflux from soil, R s) at Caxiuana, Tapaj~s, and CastanhaI. TFE or itTigate~ pl~t~ (gray tri:ngles) are plotted against control plots (black filled circles) at each site. were mflaled " ntly I'llstalled collars Data are presented as monthly means ±SE (n - 20 at CaxlUana and Tapa] , I ana YSls usmgpermane . " , hOI' I I f h' n - 12 ~t Castanhal) with the experimental soil moisture manipulations begmnmg at mont . nset. t le re a IOns Ip ' tu re' po tent'la I and R s from Caxiuana . Data are fi'om Vasconcelos et al. . [2004], Satta between SOl' 1 mOiS , ,. kw IIet al. [2007], and Davidson et ai, [2008]. Inset from Satta et ai, [2007], reprinted with penmsslOn from Wl1ey-Blac e.
Measuremen~
ma~e b~
~ss
surprising given the seasonal variation in PAW ~t Tapaj6s, but was not detected using the time domain reflectometry and the close tracking of R s with seasonal PAW In the TFE instruments installed there [Nepstad et al., 2002; Oliveira plot at Caxiuana (Figures 1 and 4). The similari~ in R s ~e et al., 2005], then autotrophic and heterotrophic respiration tween TFE and control plots at Tapaj6s is consIstent WIth may have continued in the roots and rhizosphere, respecthe similarity in total litterfall and the radiocarbon-derived tively, maintaining relatively high Rs even at. the low PAW ages of respired carbon from both plots. This has led to the measured in the TFE plot, although how (or If) Rs and GPP suggestion that the unusually large soil and rooting depths at were differentially affected by such levels of rhizosphere Tapaj6s explains the unexpected maintenance of nor~al R s moisture availability remains unexplored at Tapaj6s. Emissions of NO, N20, and CH4 were also measured at fluxes under drought, although increased root mortalIty an.d decomposition in the TFE plot may also have been res??nsr- Tapaj6s and Castanhal. Moisture limitation effects on each ble [Brando et at., 2008; Davidson et al., 2008]. ~n ad~ltI~nal followed biophysical expectation [Forster et aI" 2007], alpossible explanation is that the role of hydra.ulI~ red~stnbu though NO emissions were more resistant to chang.e ~han tion has been underestimated, For example, rflllghttIme re- expected. At the Tapaj6s TFE experiment, N20 er~llSSlO?S covely ofrhizosphere moisture content occurred at Tapaj6s, declined, and CH4 consumption increased at low sorl mOIS-
ture availability. Consistent with this, irrigation at Castanhal had the reverse effectsyon fluxes of both trace gases; the thin concretionaty soils }Iud secondaty regrowth histOly of this site were associate~ with lower overall fluxes [Vasconcelos et al., 2004]. Emii~ions of NO were not altered by irrigation at Castanhal, and they did not increase substantially at the Tapaj6s TFE until soil moisture availability was veIY low, a result also partially attributed to the soil texture at this site [Vasconcelos et al., 2004; Davidson et al., 2008]. Following pellnanent removal of the TFE treatment after 5 years at Tapaj6s, Davidson et al. [2008] reported a return of all trace gas soil fluxes to pre-TFE treatment levels and argued that the observed effects of the drought treatment at Tapaj6s most likely reflected changes in soil aeration rather than substrate supply. In summaty, experimental manipulation quantified the strong influence of soil moisture on Rs in two of the three experiments (at Caxiuana and Castanhal). These results demonstrated the primaty importance of moisture limitation over temperature on respiration processes during drought, consistent with observations elsewhere in the region and, contradicting earlier, widely employed, modeling ,assumptions [Tian et al., 1998; Zenget al., 2005;Peylin etal., 2005]. The maintenance of relatively high R s at low soil moisture content observed in the TFE plot at Tapaj6s, possibly explained by an exceptionally deep soil profile, serves to highlight the existence of spatial variation in drought responses across the basin, although other drought-related trace gas emissions of NO, N 20, and CH4 responded to reduced soil moisture availability as expected, with temporary reductions in Cf4 and N20 emission and increases in NO production under severe moisture limitation.
439
developed from experimental fires conducted in the vicinity of the Tapaj6s TFE experiment [Ray et aI" 2005], we calculated daily probabilities of a fire spreading in both control and TFE plots for the Tapaj6s experiment, from July 2000 to December 2004 [equation (1), Figure 5]. We ran two simulations: one in which both precipitation and LAI varied and another in which precipitation was set to zero while LAI was not constrained. P = 1- 1+ e 5.35-0.3*cwp-0,131*CH-0,36*LAl'
(1)
where cwp is the sum of precipitation in time (t) in days and the cumulative precipitation in time (t - 1) x 0.5; CH is canopy height (kept constant); and LAI is monthly leaf area index. During the dry season of 2002, the maximum probability of fire spread (P) in the control forest was 0.27, while in the TFE, plot P was 0.47. This increase in fire susceptibility occuned especially after the pulse in tree mOliality, 3 years following installation of the TFE infrastructure. By simulating forest flammability based only on understOly microclimate, we assumed that fine fuel loads would not limit fire spread, although we note that the increase in coarse woody debris resulting from higher mortality would have increased fire intensity.
Control
- -
TEF Max Prob - Control Max Prob - TEF
4.4. Fire Risk One of the most obvious ways in which drought affects the flammability of tropical forests is through temporary changes in the understOly microclimate: drier and warmer conditions increase the risk of fire [Alencar et al., 2004, 2006]. But less obvious are the indirect and lasting effects of drought on forest flammability through reduced PAW [Nepstad et at., 1999,2001,2004]. As PAW reaches deficits large enough to induce leaf shedding, solar radiation penetrating the forest canopy increases and leads to higher air temperatures near the forest floor [Uhl and Kauffman, 1990]. This 2001 2002 2003 2004 2005 Year process speeds the rate at which the dlying of fine fuel (e.g., leaves and small twigs) occurs, one of the best proxies of Figure 5. Probability offire spread for the Tapaj6s TFE (gray) and forest flammability [Hoffinann et al., 2003]. contrql (black) plot forests, Dashed black and gray lines represent The observed LAI reductions in both TFE experiments where weighted precipitation was set to zero, so LAI is driving forwere sufficient to increase forest flammability in the TFE est flammability. Fire spread probability is calculated using equaplots. Based on an LAI and precipitation-driven model tion (1) and data from the TFE experiment [Ray et al., 2005].
440
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
In summary, the experimental drought h'eatment substantially increased the susceptibility of the forest to fire, even over the short term (Figure 5), converting it from a state where fire wa§ unlikely to one where fire was probable in the presence of appropriate ignition sources. The risk of fire thus increases markedly under extended drought, and this has implications for long-term carbon storage and emissions [van del' Wer! et al., 2004]. The decline of LAI in both TFE experiments, and the particularly high mortality at Tapaj6s, signify that the positive feedbacks between drought and fire may be stronger than previously hypothesized. 5. MEDIUM- AND LONG-TERM DROUGHT IMPACTS: RESPONSES IN PHYSIOLOGY AND MORTALITY Physiological responses to drought observed over the short term have impacts over the long term, most notably through their negative impacts on growth and mortality. Tree death delivers carbon to the decomposer pool, committing future emissions of CO 2 to the ahnosphere, and in the absence of replacement through recruitment and regrowth, overall net primary production and transpiration decline, while the risk to fire rises. McDowell et al. [2008] review proposed plant physiological responses to drought [Tardieu and Simonneau, 1998] and distinguish between plants under moisture limitation that exert strong stomatal control to maintain leaf water potential above a minimum value ("isohydry") and those that exert less stomatal control of transpiration, relying on continued water supply from the soil ("anisohydry"). Under strong moisture limitation (and high tension), the transpiration stream may cleave (embolize), and if this happens to a large extent, hydraulic failure can occur, leading to plant death [Tyree and Speny, 1988; West et al., 2008]. Anisohydric plants exert minimal stomatal control under moisture limitation and thus risk mmtality by hydraulic failure under severe drought unless this risk is reduced, for example, through the construction of resistant xylem vessels. By contrast, isohydric plants reduce the risk of hydraulic failure through stomatal closure, although this risk is not entirely avoided, and other resistance terms in the soil-to-atmosphere hydraulic path may also change [Franks et at., 2007]. Stomatal closure can lead to high leaf temperatures, to reduced photosynthetic carbon gain, and under extended drought to increased risk of mortality through carbon starvation of metabolism (mainly respiration) and/or increased susceptibility to pathogen attack. Leaf water potential measurements from the Caxiuana TFE experiment were consistent with trees responding isohydrically to drought [Fisher et al., 2006]. This proposed mode of response is further consistent with: (1) declines in
GPP and stem growth rates following TFE treatment [Fisher et al., 2007; Brando et al., 2008]; (2) the initial resistance to mortality observed over the first 2 years of experimental drought at Caxiuana and Tapaj6s [Nepstad et al., 2007; A. C. L. Costa et aI., manuscript in preparation, 2009]; and (3) pantropical observations of small average increases in mortality following ENSO events over the last three decades (the median increase in mortality following an ENSO event across 45 tropical forest plots was 1.2%) [Meir et al., 2008]. Further consistent with the notion that isohydry may be common in rain forest trees that are not adapted to long-term and severe droughts, resistance in mortality rates to the TFE treatment gave way after 2-3 years to substantial mortality increases [Nepstad et al., 2007; Brando et al., 2008], possibly as a result of carbon starvation. The Tapaj6s TFE experiment revealed highly variable background rates ofmmtality in both the TFE and control plots (1-3%), and this was then followed by exceptionally high mortality (9% in trees with dbh> 10 em) after 3 years of experimental drought [Nepstad et al., 2007; Brando et al., 2008]. In years 4 and 5 of the experimental drought at Tapaj6s, mortality declined to just above (the relatively high) background levels, but 1 year after removal of the TFE infrastructure at Tapaj6s, mortality rose again to 7%, suggesting longer term and possibly speciesspecific impacts. Indications of a correlation between species differences in stomatal control and mortality at Tapaj6s [Ehleringer et al., 2004] further support the notion of species-specific variation in mortality risk during drought [cf. Fisher et al., 2006] (see Table 2), and increased regrowth during years 4 and 5 by understory species released via mortality impacts on the canopy will also have influenced the range of species-based responses at Tapaj6s [Brando et al., 2008]. The immediate influence of increased mortality on CO 2 emissions is likely to be small during drought because of the desiccation constraint on organic matter breakdown [Meir et al., 2008], although the effect of such strong pulses of tree death on ecosystem-level GPP is less clear because mortality reduces LAI while simultaneously increasing radiation availability within the canopy. However, over the long term, mortality clearly commits substantial carbon to the atmosphere through the breakdown of additional necromass. If the high mortality pulse observed at the Tapaj6s TFE experiment after 3 years of drought (5.4 Mg C ha-1) [Brando et al., 2008] occurred widely over the region, it would represent a large net committed emission to the atmosphere. Although mortality at Tapaj6s declined to 2-4% under the following 2 years ofTFE treatment [Brando et al., 2008], the live biomass removed during this single mortality event represented up to 8.5 years of aboveground growth under normal rainfall [Nepstad et al., 2007].
MEIR ET AL.
441
The findings from the TFE experiments are consistent with 6. MODELING TWENTY-FIRST CENTURY DROUGHT IMPACTS ON AMAZONIAN RAIN FORESTS recent observational ~:~ports of natural drought effects on Amazonian forest t1i~es. For example, following the severe drought event of 2105, tree growth observations in 55 1-ha 6.1. Short-Term Effects: Seasonal to Interannual plots distributedAtcross· Amazonia demonstrated droughtNotwithstanding the possible discovery of new long-term induced reductions in stem growth (especially in larger trees) and significant, but spatially patchy, increases in mortality drought responses, correct attribution of physiological pro[Phillips et al., 2009]. The 2005 drought in Amazonia was cesses at seasonal or interannual tirnescales is needed to inless severe or prolonged than the soil moisture limitation terpret the effects of climate anomalies, such as ENSO or imposed in the TFE experiments and comprised additional the 2005 drought in Amazonia, and to provide the basis for climatic impacts on temperature, atmospheric humidity, and robust model predictions. The TFE experiments simulated rainfall reductions similar radiation, yet Phillips et al. [2009] estimated an overall reduction in aboveground growth of 5.3 Mg ha-1, in addition to to that ofa severe ENSO, such as the 1997/1998 event [Meir a substantial increase in mortality-committed carbon emis- et al., 2008], but the results are only beginning to be incorsions. The regional-scale spatial variation in mortality ob- porated into vegetation modeling frameworks. Fisher et al. served during 2005 was dependent on differences in climate, [2007] successfully simulated the effects of the Caxiuana soil-type, and species, with a tendency for higher mmtality TFE manipulation on GPP, specifying a detailed multilayer in species with lower density wood [Phillips et al., 2009]. soil and canopy physiological model [Williams et al., 1996] Such spatial variation in the response to drought was also using measurements fi'om the TFE experiment of soil hyevident in a recent pantropical survey of recent ENSO im- draulic properties, leaf biochemical photosynthetic capacity pacts on mmtality [Meir and Grace, 2005; Meir et al., 2008]. and LAI, and meteorological data (Figure 6). Gas exchange Taken together with evidence of species differences in pho- was validated at leaf and canopy scales using independent tosynthesis, growth, and reproduction from the TFE experi- leaf-scale stomatal conductance [Fisher et al., 2006] and ments [Ehleringer et al., 2004; Fisher et al., 2006; Brando et tree-scale sap flux [Fisher et al., 2007] measurements. The al., 2006], these results imply initial, but spatially variable, analysis demonstrated that GPP and h'anspiration at Caxresistance to shmt-term soil moisture limitation, followed by iuana are not constrained by moisture supply under normal increased mortality and likely alterations in tree community climatic variation [s;ee Saleska et al., this volume], but that composition as the severity of drought deepens and extends. more severe moisture limitation imposed strong constraints The longer term (multiyear to decadal) effects of drought upon h'anspiration and GPP. Transpiration declined by 30on NEP are not well constrained by any current data sets, 40% (300-418 mm a-I) and GPP by 13-14% (4.0-4.3 t C but may contain surprises. As observed in the TFE experi- ha- 1 a-r) during the first 2 years of experimental drought at ments and elsewhere, short-term reductions in GPP and R s Caxiuana [Fisher et al., 2007]. Changes in LAI and hydrauare likely, and increased mortality and fire incidence will in- lic (rather than biochemical) properties were the principal crease losses of carbon to the atmosphere. However, recent determining parameters: maximum foliar stomatal conductobservations of significant medium-term (5 year) increases ance declined by more than 50% and belowground hydrauin leaf respiration at the Caxiuana TFE experiment [Meir et lic resistance increased more than 10-fold. Combined with al., 2008; Metcalfe et aI., submitted manuscript, 2009] sug- changes in heterotrophic and autotrophic respiration [Meir et gest unexpected additional foliar emissions of CO 2 during al., 2008], the impact of drought on NEP is probably finely drought, even after correcting for temperature. Although balanced and closer to zero than the large carbon emissions previously not considered in NEP calculations, this effect predicted by earlier model analyses of the effects on NEP of has been reported for other trees experiencing low rainfall the 1997/1998 ENSO drought [e.g., Tian et al., 1998; Zeng [Turnbull et al., 2001; Wright et al., 2006] and, as well as et al., 2005; Peylin et al., 2005]. Coarser-scale models have also been used to simulate reducing NEP, may also increase mortality risk through excessive. metabolism of carbon reserves. Other unexpected Amazonian drought impacts. Potter et al. [2004] used the drought response processes may need to be considered in Carnegic Ames Stanford Approach (CASA) model to quanthe future, including the potential for changes in isoprene . tify rain forest ecosystem functioning during drought, paremissions, cUlTently a small component of the carbon cycle tially driving simulations with remotely sensed data. In this [pegoraro et al., 2004], in pathogen attacks, known to be work, Rs was more moisture sensitive than in earlier model substantial during drought stress in other forest ecosystems analyses, and in a subsequent development, the spectral sig[Ayers and Lombardero, 2000] and in soil fungal activity nal fi'om the Tapaj6s canopy to drought was also successfully incorporated [Asner et al., 2004], offering the future [Meir et al., 2006].
442
MEIR ET AL.
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
Apr
July
2001
Oct
Jan
Apr
July
2002
Oct
Jan
Apr
2003
July
Oct
Time
Figure 6. The modeled response to normal rainfall and TFE treatment by GPP and canopy (stomatal) conductance to water vapor, at Caxiuana: (top) incident rainfall, (middle) stomatal conductance at canopy scale, and (bottom) GPP. Open circles denote TFE treatment forest; solid circles denote control forest. From Fishel' et al. [2007], reprinted with pennission from Wiley-Blackwell.
prospect of detecting drought stress from space. Recently, in a third-generation development of the SiB ecosystem model [Sellers et al., 1986], Baker et al. [2008] incorporated new soil depth and rooting properties observed at Tapaj6s [Oliveira et al., 2005; Lee et al., 2005], together with a seasonal soil moisture response function for R s observed at a separate site near the Tapaj6s TFE experiment [Saleska et al., 2003]. The new model formulation successfully replicated measurements ofthe seasonal variation in forest-atmosphere carbon exchange made using eddy covariance at Tapaj6s. Only with the combination ofseveral newly observed mechanisms was it possible to simulate seasonality in carbon exchange correctly with this model [Baker et al., 2008]. However, getting water supply mechanistically correct should ideally start with correctly parameterized soil hydraulic properties, as highlighted by Harris et al. [2004] for a central Amazonian site near Manaus [cf. Fisher et al., 2008]. Whether the successful reformulation of SiB by Baker et al. [2008] included a representation of site-corrected soil hydraulic parameters is unclear, but this could strongly influence modeled PAW and gas exchange at low soil moisture contents, and underlines the importance of ongoing data-model validation efforts.
6.2. Longer-Term Effects
At longer (decadal-to-centmy) timescales, it has been neceSSalY to derive the physiological impacts on forest carbon metabolism of twenty-first centmy increases in drought stress, temperature, and atmospheric CO 2 concentration from physiological principles, as well as from measurements made over shOlier time periods [Betts, 2004]. In a recent review, Lloyd and Farquhar [2008] argued that the positive effect of increased atmospheric CO 2 concentration on photosynthesis ,is likely to outweigh any negative impacts of concutTent warming, and this will probably balance in favor of a positive impact on NEP. In addition, the tendency in most plants to reduce stomatal conductance at high atmospheric CO 2 concentration makes possible reductions in water loss through transpiration without diminishing carbon acquisition, thus further increasing the resistance of vegetation to climatic drying. However, there remains uncertainty in transpiration estimates, as under drought and/or warming, a drier atmosphere will impose a bigger atmospheric demand on evaporation potentially leading instead to higher rates of evapotranspiration [Salazar et al., 2007] irrespective of reductions in stomatal conductance. Plant respiration [e.g.,
Meir et al., 2001] now seems likely to acclimate at higher temperatures [Atkin ~fjd Tjoelker, 2003; Atkin et al., 2008], and this would confer drought resistance through reduced use of plant carbq;{ reserves. But whether stomatal conductance responses,tJ increased atmospheric CO 2 concentration over the long tenD remain similar to short-term measurements is unceliain, and in any case, the carbon economy of trees and ecosystems may be further influenced by changes in LAI and the drought response in leaf respiration [Atkin and Macherel, 2009; Meir et al., 2008]. Of course, over such longer timescales, any resistance to drought based on plant physiology may also be strongly and negatively impacted by pest or pathogen attack [Ayers and Lombardero, 2000; Meir et al., 2006], or by an increase in the frequency of extreme weather events [JPCC Working Group J, 2007]. Incorporating many of these responses into vegetation models operating over the time periods required to simulate vegetation change is still at an early stage [Meir et al., 2006; Os tIe et al., 2009]. However, analysis ofthe drought response mechanisms specified in different dynamic global vegetation models (DGVMs) has identified the need for corrections to some process representations. The mortality ri*s recently quantified for natural and more severe drought [Nepstad et al., 2007; Meir et al., 2008; Phillips et al., 2009] have not yet been incorporated into current modeling frameworks and require specific model structures or parameterization to do so [Moorcroft, 2006]. Furthermore, differences exist among different DGVMs in the allocation of fixed carbon to aboveand belowground ecosystem components, and this has a significant impact on the response in R s to climatic warming and hence to changes in NEP [Dufresne et al., 2002]. Recent new insight into carbon allocation processes [see Malhi et al., this volume) should inform this issue further. More surprisingly, Galbraith et al. (submitted manuscript, 2009), analyzing three widely used DGVMs [Cox et al., 2000; Levy et al., 2004; Sitch et al., 2003], have shown that the modeled "dieback response" in Amazonian vegetation [Cox et al., 2000; Sitch et al., 2008] is more strongly dependent on the specified temperature responses in respiration and photosynthesis than on the direct effects of moisture limitation, despite observational evidence elsewhere of acclimation to temperature in plant respiration [Atlan and TJoelker, 2003], and the effects of drought summarized here. Thus, the challenge now is to incorporate the range of observed moisture limitation effects cotTectly into DGVMs and other vegetation modeling fi·ameworks. Getting the balance right between the drought-buffering effects of above- and belowground ecosystem components and representing them at the correct scale will require a two-way interaction between data providers and modelers. This work will improve the modeling of vegetation-atmosphere interactions during drought, but to
443
understand the overall effects of drought on Amazonian rain forests, fuliher outward links to models of fire risk, and land use change are also needed. 6.3. Drought and Fire
Switches in vegetation cover under climatic dlying and wanning have been predicted using both dynamic and equilibrium vegetation model analyses [Oyama and Nobre, 2003; Salazar et al., 2007; Sitch et al., 2008], but none of these simulations is likely to be realistic without the inclusion of fire risk. Lags in the development of natural vegetation under climatic change outside the fundamental niche of many tree species may occur because trees are large, resistant, and long-lived organisms. However, fire has the potential to switch forest to savanna or grassland, short-circuiting these lags and rapidly accelerating natural rates of climate-driven vegetation change. The networks of positive feedbacks among climatic warming and dlying, deforestation, forest fragmentation, and fire have been described in detail previously [Nepstad et al., 1999,2001; Soares-Filho, 2006], and the southeastern sector of Amazonia seems most vulnerable to forest loss as high drought risk and high rates of deforestation overla; [Malhi et al., 2008]. However, dynamic integrated models of climate, fire risk, vegetation, and deforestation have not yet been developed Vtlry far [Nepstad et al., 2008]. In one such early study, GoldiJlg and Betts [2008] demonstrated substantially increased fire risk across the region by 2020, rising to a "high" risk of fire across 50% of the region by 2080. This analysis superimposed deforestation scenarios [SoaresFilho et al., 2006; van Vuuren et al., 2007] and a simple fire model [a forest fire danger index (FFDI), parameterized in Australia] [Noble et al., 1980; Ho.Oinann et al., 2003] on an ensemble climate model analysis using HADCM3 that incorporated the vegetation response in a simplified way through altered GCM parameter sets [Golding and Betts, 2008]. The next steps in this process will be to incorporate the flammability estimates and vegetation responses from the TFE and other observational data into a fully functioning GCM-DGVM-fire vulnerability framework. The Australia-derived FFDI model used by Betts [2008] does not consider fire vulnerability in the forest understOly, a frequent precursor to subsequent full-canopy fire events in Amazonian forests [Nepstad et al., 2001]. Hence, the results from the TFE experiments (Figure 5) probably indicate a higher vulnerability to fire than specified by the FFDI model: under scenarios of stronger dlying or greater deforestation, the biophysical component of the risk to forest loss estimated by Betts et al. [2008] may prove conservative. However, future fire risk is also strongly dependent on the
444
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS MEIR ET AL.
nature of environmental governance in forested regions because of its impact on deforestation and other modes of land use change [Malhi et al., 2009]. In this regard, there may be some roonY for optimism. In particular, the potential to mitigate the risk to fire and forest degradation through local and regional governance mechanisms is growing rapidly in parts of Amazonia [Nepstad et al., 2006; Soares-Filho et al., 2006]. The added possibility of national and international agreements that may permit and help finance sustainable land use based on payments for forest ecosystem services [Mitchell et al., 2008; Daily et al., 2009] may further reduce the risk of drought-related forest loss, thus delaying or avoiding some of the more extreme "dieback" scenarios that have been modeled for the region. 7. CONCLUSIONS Drought in Amazonia cannot be represented as a single parameter, and cunent modeling trends are moving toward the representation of the suite of changes in climate, vegetation, soil, fire, and land use that map to this term. The LBA observational network has enriched our capacity to specify such modeling frameworks. The manipulative experiments described here form part of this network and comprise an impOliant means by which we can test the mechanistic basis of the relevant ecological processes, as well as validate measured gross ecosystem fluxes, and thus provide confidence in model predictions of future scenarios. Validating process representation in models at multiyear or longer timescales is becoming increasingly important, and the value of long-term experimental data is increasingly recognized [Sitch et al., 2008]. Combining the experimental results summarized here with observational data from across the basin, a picture of the impacts of drought on Amazonian vegetation is beginning to emerge. This understanding needs to be incorporated into the next generation of DGVMs and can cunently be summarized as follows. (l ) Resistance to seasonal drought in rain forest functioning is likely at most locations, especially climatically marginal sites. (2) Even under severe drought, such as that imposed in the TFE experiments, a surprising degree of resistance is probable at first, although spatially patchy increases in mortality were observed during the 2005 drought. In the short term, while strong responses in GPP, transpiration, respiration, and LAI can occur during periods of strong moisture limitation of up to 24 months, overall NEP may change only slightly. (3) As severe drought extends to 3-5 years and beyond, mortality increases markedly, and species differences may emerge in terms of loss, survival, reproduction, and regrowth, with substantial negative impacts on NEP and transpiration, and substantial increases in vulnerability to fire. (4) At longer timescales,
our ability to tightly constrain the tempo and mode of any drought-driven tipping point in forest function is limited by data availability, but the increased risk offire associated with drought means that switches in vegetation type at decadal or greater timescales will rely as much on human activities as on climate. New results are still emerging fi'om the three soil moisture manipulation experiments at Caxiuana, Tapaj6s, and Castanhal. The rainfall exclusion at Tapaj6s has now been halted, and the processes governing recovery from drought are under analysis [e.g., Davidson et al., 2008]; at Caxiuana and Castanhal, the experimental treatments continue. The possibility onong-term data sets offered by these experiments represents a uniquely powerful way by which we can begin to understand how multiyear and decadal-scale drought will impact the species composition, vulnerability, and gas exchange properties of Amazonian vegetation. The standard science funding cycle of 3 or 5 years is insufficient to fully address ecological questions of this SOli, and although this issue has been recognized within LBA and elsewhere [Hobbie et al., 2003], the case for supporting long-term ecological studies is ul:gent and needs to be made more widely. As the experiments, data, and model analyses are extended, some of the key emerging science challenges are likely to include at least some of the following questions: 1. What is the minimum set of rooting and soil depth properties required to model vegetation and soil function adequately during drought [Woodward and Osborne, 2000; Bruno et al., 2006; Metcalfe et al., 2008]? 2. How can plant hydraulic and biochemical sensitivities be represented accurately at large scale, including their role in affecting tree mortality [Fisher et al., 2006, 2007; McDowell et al., 2008]? 3. How can respiration in plants and microbes be represented over short and long timescales under climatic warming and drying? [Trumbore, 2006; Meir et al., 2008] 4. Can mortality risks and alterations to reproductive output be modeled to predict change in vegetation properties and species composition under drought [Branda et al., 2008; Meir et al., 2008; Phillips et al., 2009]? 5. What are the sensitivities in the components of NEP and the allocation of net primary production to PAW and temperature [Branda et al., 2008; Galbraith et al., submitted manuscript, 2009]? 6. Can we model recovery from drought, and how long does recovery take [Branda et al., 2008]? 7. How will twenty-first century land use change, fire incidence, and drought interact to affect rain forest functioning or a transition from rain forest to different vegetation types [Soares-Filho et al., 2006; Betts et al., 2008]?
Acknowledgments. We ~ould like to thank several funding bodies for the initiation and/~ontinuation of research funding support at all three sites from 1999 to 2009' LBA Brazil' NERC UK E·U I ' " " 5th Framework Prog~amme; NASA, USA; NSF, USA; CNPq, Brazil. We also thank 9cfe institutes responsible for maintenance of the reserves at each sIte and for providing permission and support for this work: MPEG, Belem, Pani; Santarem Para' UFRA Castanhal Para. A large number ofBrazilian students 'have been trained at B.S.: M.S., and Ph.D. level during the running of the three experiments, and we thank LBA for making this training support available.
REFERENCES Alencar, A A C., L. A Solorzano, and D. C. Nepstad (2004), Modeling forest understOly fires in an eastem Amazonian landscape, Ecol. Appl., 14, SI39-S149. Alencar, A A C., D. Nepstad, and M. D. V Diaz (2006), Forest understOly fire in the Brazilian Amazon in ENSO and non-ENSO years: Area bmned and committed carbon emissions, Earth Interact., 10(6), EIl50, doi:IO.1l75/EIl50.1. Aragiio, L. E. O. C., Y. Malhi, R M. Roman-Cuesta, S. Saatchi, L. O. Anderson, and Y. E. Shimabukuro (2007), Spatial patterns and fire response of recent Amazonian droughts, Geophys. Res. Lett., 34, L07701, doi:IO.l029/2006GL028946. Asner, G. P., D. Nepstad, G. Cardinot, and D. Ray (2004), Drought stress and carbon uptake in an Amazon forest measured with spacebome imaging spectroscopy, Proc. Nat!. Acad. Sci. U. S. A., 101, 6039-6044. Atkin, O. K., and D. Macherel (2009), The crucial role ofplant mitochondria in orchestrating drought tolerance, Ann. Bot., 103(4), 581-597, doi:l0.l093/aob/mcn094. Atkin, O. K, and M. G. Tjoelker (2003), Thermal acclimation and the dynamic response ofplant respiration to temperature, Trends Plant Sci., 8, 343-351. Atkin, O. K, L. J. Atkinson, R. A Fisher, C. D. Campbell, J. Zaragoza-Castells, J. W. Pitchford, F. I. Woodward, and V. Hurry (2008), Using temperature-dependent changes in leaf scaling relationships to quantitatively account for thermal acclimation of respiration in a coupled global climate-vegetation model, Global Change BioI., 14,2709-2726. Ayers, M. P., and M. J. Lombardero (2000), Assessing the consequences of global change for forest disturbances for herbivores and pathogens, Sci. Total Env., 262, 263-286. Baker, I. T., L. Prihodko, A S. Denning, M. Goulden, S. Miller, and H. R. da Rocha (2008), Seasonal drought stress in the Amazon: Reconciling models and observations, J. Geophys. Res., 113, GOOBOI, doi:l0.1029/2007JG000644. Baker, T. R, et al. (2004), Variation in wood density determines spatial patterns in Amazonian forest biomass, Global Change BioI., 10,1-18, doi:IO.1llllj.l529-8817.2003.0075I.x. Belk, E. L., D. Markewitz, T. C. Rasmussen, E. J. M. Cmvalho, D. C. Nepstad, and E. A. Davidson (2007), Modeling the effects of throughfall reduction on soil water content in a Brazilian Oxisol under a moist tropical forest, Water Resow'. Res., 43, W08432, doi: I0.1 029/2006WR005493.
445
Betts, R. A. (2004), Global vegetation and climate: Self-beneficial effects, climate forcings and climate feedbacks, J Phys. IV, 121, 37-60. Bousquet, P., P. Peylin, P. Ciais, C. Le Quere, P. Friedlingstein, and P. P. Tans (2000), Regional changes in carbon dioxide fluxes ofland and oceans since 1980, Science, 290, 1342-1346. Brando, P., et al. (2006), Effects of partial throughfall exclusion on the phenology of Coussarea racemosa (Rubiaceae) in an eastcentral Amazon rainforest, Oecologia, 150, 181-189. Brando, P. M., D. C. Nepstad, E. A Davidson, S. E. Trumbore, D. Ray, and P. Camargo (2008), Drought effects on litterfall, wood production and belowground carbon cycling in an Amazon forest: Results of a throughfall reduction experiment, Phi/os. Trans. R. Soc., Ser. B, 363, 1839-1848. Bruno, RD., H. R. da Rocha, H. C. De Freitas, M. L. Goulden, and S. D. Miller (2006), Soil moisture dynamics in an eastern Amazonian tropical forest, Hydrol. Processes, 20, 2477-2489. Cardoso, M. F., G. C. Hurtt, B. Moore, C. A Nobre, and E. Prins (2003), Projecting futme fire activity in Amazonia, Global Change BioI., 9, 656-669, doi:l 0.1046/j.l365-2486.2003.00607.x. Carswell, F. E., et al. (2002), Seasonality in CO 2 and H2 0 flux at an eastern Amazonian rain forest, J. Geophys. Res., 107(D20), 8076, doi: 10.1029/2000JD000284. Chagnon, F. J. F., and R. L. Bras (2005), Contemporary climate change in the Amazon, Geophys. Res. Lett., 32, L13703, doi:l0.1029/2005GL022722. Cochrane, M. A (200~), Fire science for rainforests Nature 421 913-919.· , " Cochrane, M. A, an~ W. F. Laurance (2002), Fire as a largescale edge effect in Amazonian forests, J. Trap. Ecol., 18, 311-325. Coelho, R F. R, D. G. Zarin, 1. S. Miranda, and J. M. Tucker (2004), Analise floristica e estrutural de uma fioresta em diferentes estagios sucessionais no municipio de Castanhal, Para, Acta Amazonica, 33, 563-582. Costa, M. H., S. N. M. Yanagi, P. J. O. P. Souza, A Ribeiro, and E. J. P. Rocha (2007), Climate change in Amazonia caused by soybean cropland expansion, as compared to caused by pastmeland expansion, Geophys. Res. Lett., 34, L07706, doi:l0.l029/ 2007GL02927I. Cox, P. M., R A. Betts, C. D. Jones, S. A Spall, and 1. J. Totterdell (2000), Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model, Nature, 408, 184-187. Cox, P. M., P. P. Harris, C. Huntingford, R. A. Betts, M. Collins, C. D. Jones, T. E. Jupp, J. A. Marengo, and C. A Nobre (2008), Increasing risk of Amazonian drought due to decreasing aerosol pollution, Nature, 453, 212-215. Culf, A. D., G. Fisch, and M. G. Hodnett (1995), The albedo of Amazonian forest and ranchland, J. Clim., 8,1544-1554. Daily, G. C., et al. (2009), Ecosystem services in decision making: Time to deliver, Front. Ecol. Environ., 7, 21-28, doi: 10.1890/080025. da Rocha, H. R, et al. (2009), Patterns ofwater and heat flux across a biome gradient fi'om tropical forest to savanna in Brazil, J. Geophys. Res., 114, GOOBI2, doi:l0.l02912007JG000640.
MEIR ET AL. 446
447
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
daRocha, H. R., A. O. Manzi, andJ. Shuttleworth (2009), Evapotranspiration, Geophys. Monogr. Ser., doi: 10.1 02912008GM000744, this volume. Davidson, E.ll'A. (1992), Sources of nitric oxide and nitrous oxide following wetting of dly soil, Soil Sci. Soc. Am. J., 56, 95-102. Davidson, E. A., L. V. Verchot, J. H. Cattanio, 1. L. Ackerman, and J. E. M. Carvalho (2000), Effects of soil water content on soil respiration in forests and cattle pastures of eastern Amazonia, BiogeochemistlJI, 48, 53-69. Davidson, E. A., F. Y. Ishida, an~ D. C. Nepstad (2004), Effects of an experimental drought on soil emissions of carbon dioxide, methane, nitrous oxide, and nitric oxide in a moist tropical forest, Global Change BioI., 10,718-730. Davidson, E. A., et ai. (2008), Trace gas emissions Tapajos 5 years, Global Change BioI., 14,2582-2590. Dufresne, J. L., P. Friedlingstein, M. Berthelot, L. Bopp, P. Ciais, L. Fairhead, H. Le Treut, and P. Monfray (2002). On the magnitude of positive feedback between future climate change and the carbon cycle, Geophys. Res. Lett., 29(10), 1405, doi: I 0.10291 200lGL013777. Ehleringer, J. R., et ai. (2004), The Third International Conference of the Large-Scale Biosphere Atmosphere Program, Available at ftp:/llba.cptec.inpe.br/presentationsILBA-III-ConferenceJuly2004-Brasilia/July28,2004/S 15/12_7%20Ehleringer%20Sta ble%20isotope%20analyses.pdf Engelbrecht, B. M. J., L. S. Comita, R. Condit, T. A. Kursar, M. T. Tyree, B. L. Turner, and S. P. Hubbell (2007), Drought sensitivity shapes species distribution patterns in tropical forests, Nature, 447, 80-82. Fisher, R. A., M. Williams, R. Lobo do Vale, A. C. Lola da Costa, and P. Meir (2006) Evidence from Amazonian forests is consistent with isohydric control of leaf water potential, Plant Cell Environ., 29, 151-165. Fisher, R. A., M. Williams, A. L. Costa, Y. Malhi, R. F. da Costa, S. Almeida, and P. Meir (2007), The response of an Eastern Amazonian rain forest to drought stress: Results and modelling analyses from a throughfall exclusion experiment, Global Change BioI., 13, 2361-2378. Fisher, R. A., M. Williams, M. de Lourdes Ruivo, A. L. Costa, and P. Meir (2008), Evaluating climatic and soil water controls on evapotranspiration at two Amazonian rainforest sites, Agric. For. Meteoro!', 148, 850-861. Foley, J. A., A. Botta, M. T. Coe, and M. H. Costa (2002), El Nifio--Southern oscillation and the climate, ecosystems and rivers of Amazonia, Global Biogeochem. Cycles, 16(4), 1132, doi: 10.1 029/2002GBOOI872. Fortini, L. B., S. S. Mulkey, D. J. Zarin, S. S. Vasconcelos and C. R. J. D Carvalho (2003), Drought constraints on leaf gas exchange by Miconia ciUata (Melastomataceae) in the understOly of an eastern Amazonian regrowth forest stand, Am. J. Bot., 90,1064-1070. Forster, P., et ai. (2007), Changes in atmospheric constituents and in radiative forcing, Climate Change 2007: The Physical Science Basis, edited by S. Solomon et aI, pp. 129-234, Cambridge Univ. Press, Cambridge, U. K. Franks, P. J., P. L. Drake, and R. H. Froend (2007), Anisohydric but isohydrodynamic: Seasonally constant plant water gradient
explained by a stomatal control mechanism incorporating variable plant hydraulic conductance, Plant Cell Environ., 30,19-30. Friedlingstein, P., ct ai. (2006), Climate-carbon cycle feedback analysis: Results from the C4MIP model intercomparison, J. CUm., 19, 3337-3353. Fudey, P. A., J. Proctor, and J. A. Ratter (1992), Nature and Dynamics ofForest-Savanna Boundaries, CRC Press, London. Golding, N., and R. Betts (2008), Fire risk in Amazonia due to climate change in the HaclCM3 climate model: Potential interactions with deforestation, Global Biogeochem. Cycles, 22, GB4007, doi: 10.1029/2007GB003166. Grace, J., J. Lloyd, J. Mclntyre, A. C. Miranda, P. Meir, H. S. Miranda, C. Nobre, J. Moncrieff, J. Massheder, and Y. Malhi (1995), Carbon dioxide uptake by an undisturbed tropical rain forest in southwest Amazonia, 1992 to 1993, Science, 270, 778. Harris, P., C. Huntingford, P. M. Cox, J. H. C. Gash, and Y. Malhi (2004), Effect of soil moisture on canopy conductance of Amazonian rainforest, Agric. For. Meteorol., 122, 215-227. Hobbie, J. E., S. R. Carpenter, N. B. Grimm, J. R. Gosz, and T. R. Seastedt (2003), The US Long Term Ecological Research program, BioScience, 53, 21-32. Hoffmann, W. A., W. Schroeder, and R. B. Jackson (2003), Regional feedbacks among fire, climate, and tropical deforestation, J. Geophys. Res., 108(D23), 4721, doi:10.102912003JD003494. Houghton, R. A., M. Gloor, J. Lloyd, and C. Potter (2009), The regional carbon budget, Geophys. Monogr. Ser., doi:10.1029/ 2008GM000718, this volume. Howard, P. J. A. (1979), Respiration of decomposing litter in relation to temperature and moisture, Oikos, 33,457--465. Huete, A., K. Didan, W. van Leeuwen, T. Miura, and E. GleruI (2008), MODIS vegetation indices, Land Remote Sensing and Global Environmental Change: NASA's Earth Observing System and the Science ofASTER and MODIS. Huntingford, c., J. A. Lowe, B. B. B. Booth, C. D. Jones, G. R. Harris, and P. Meir (2009), How large is the effect of model uncertainty in the global carbon cycle compared to model uncertainty in thermal properties of the Earth system when predicting temperature increases by year 2100?, Tellus, Ser. B, 61, 355-360. Hurlbeli, S. H. (2004), On misinterpretations of pseudoreplication and related matters: A reply to Oksanen, Oikos, 104, 591. Hutyra, L. R., J. W. Munger, C. A. Nobre, S. R. Saleska, S. A. Vieira, 'and S. C. Wofsy (2005), Climatic variability and vegetation vulnerability in Amazonia, Geophys. Res. Lett., 32, L24712, doi: 10.1 029/2005GL02498I. IPCC Working Group I (2007), Regional Climate Projections. in Climate Change 2007: The Physical Science Basis. Contribution o.lWorking Group I to the Fourth Assessment Report ofthe Intergovernmental Panel on Climate Change, edited by S. Solomon et aI., pp. 847-940, Cambridge Univ. Press, Cambridge, U. K. Jipp, P. H., D. C. Nepstad, D. K. Cassel, andC. Reis de Carvalho (1998), Deep soil moisture storage and transpiration in forests and pastures ofseasonally-dlY Amazonia, CUm. Change, 39, 395--412. Joslin, J. D., M. H. Wolfe, and P. J. Hanson (2000), Effects of altered water regimes on forest root systems, New Phytol., 147, 117-129.
Langenfelds, R. L., R. J. Francey, B. C. Pak, L. P. Steele, J. Lloyd, C. M. TlUdingyr, and C. E. Allison (2002), InteralUlUal growth rate variatiSl1s of atmospheric COz and its .s 13 C, Hz, CH4, and CO bet'ifeen 1992 and 1999 linked to biomass burning, Global Bio/~ochem. Cycles, 16(3), 1048, doi: 10.10291 200 lGBOO 14613:' Lee, J. E., R. S. Oliveira, T. E. Dawson, and 1. Fung (2005), Root functioning modifies seasonal climate, Proc. Nat. Acad Sci. U. S. A., 102, 17,576-i7,581. Levy, P. E., M. G. R Cannell, and A. D. Friend (2004), Modelling the impact of future changes in climate, COz concentration and land use on natural ecosystems and the telTestrial carbon sink Global Environ. Change, 14,21-30. ' Likens, G. E., F. H. Bormann, N. M. Johnson, D. W. Fisher, and R. S. Pierce (1970), Effects of forest cutting and herbicide treatment on nutrient budgets in the Hubbard Brook watershedecosystem, Ecol. Monogr., 40(1), 23--47. Lloyd, J., and G. D. Farquhar (2008), Effects of rising temperatures and [COz] on the physiology oftropical forest trees, Phi/os. Trans. R. Soc. Ser. B, 363,1811-1817. Lloyd, J., J. Grace, A. C. Miranda, P. Meir, S. C. Wong, B. S. Miranda, 1. R. Wright, J. H. C. Gash, and J. McIntyre (1995), A siInpIe calibrated model of Amazon rain forest productivity based on leaf biochemical propeliies, Plant Cell Environ., 18, 1129-1145. Malhi, Y., A. D. Nobre, J. Grace, B. Kruijt, M. G. P. Pereira, A. Culf, and S. Scott (1998), Carbon dioxide transfer over a Central Amazonian rain forest, J. Geophys. Res., 103(D24), 31,593-31,612. Malhi, Y., J. T. Roberts, R. A. Betts, T. J. Killeen, W. Li, and C. A. Nobre (2008), Climate change, deforestation, and the fate ofthe Amazon, Science, 319, 169. Malhi, Y., et ai. (2009), Exploring the likelihood and mechanism of a climate-change-induced dieback of the Amazon rainforest. Froc. Natl. Acad. Sci. U. S. A., doi: 10.1073/pnas.0804619106. Malhi, Y., S. Saatchi, C. Girardin, and L. E. O. C. Aragao (2009), The production, storage, and flow of carbon in Amazonian forests, Geophys. Monogr. Ser., doi: IO.l029/2008GM000779, this volume. Marengo, J. A., C. A. Nobre, J. Tomasella, M. D. Oyama, G. S. De Oliveira, R. De Oliveira, H. Camargo, L. M. Alves, and 1. F. Brown (2008), The drought of Amazonia in 2005, J. Clim., 21, 495-516. McDowell, N., et ai. (2008), Mechanisms of plant survival and mortality during drought: Why do some plants survive while others succumb to drought?, New Phytol., 178, 719-739. Meir, P., and J. Grace (2005). The response to drought by tropical rain forest ecosystems, in Tropical Forests and Global Climate Change, edited by Y. Malhi and O. Phillips, pp. 71-80, Oxford Univ. Press, Oxford, U. K. Meir, P., J. Grace, and A. C. Miranda (2001), Leaf respiration in two tropical rain forests: constraints on physiology by phosphorus, nitrogen and temperature, Funct. Eeal., 15, 378-387. Meir, P., P. Cox, and J. Grace (2006), The influence of terrestrial ecosystems on climate, Trends Ecol. Evol., 21,254-260. Meir, P., D. B. Metcalfe, A. C. L. Costa, and R. A. Fisher (2008), The fate of assimilated carbon during drought: Impacts on respiration in Amazon rainforests, Philos. Trans. R. Soc. Ser. B, 363,1849-1855.
Metcalfe, D. B., P. Meir, L. E. O. C. Aragao Y. Malhi, A. C. L. da Costa, A. Braga, P. H. L. Gonc;alves, J. de Athaydes, S. S. de Almeida, and M. Williams (2007), Factors controlling spatio-temporal variation in carbon dioxide efflux from smface litter, roots, and soil organic matter at four rain forest sites in the eastern Amazon, J. Geophys. Res., 112, G04001, doi:10.1029/ 2007JG000443. Metcalfe, D. B., P. Meir, L. Aragao, A. C. L. da Costa, A. P. Braga, P. H. L. Gouc;alves, 1. de Athaydes Silva Jr., S. S. de Almeida, L. A. Dawson, and Y. Malhi (2008), The effects of water availability on root growth and morphology in an Amazon rainforest, PlantSoi/, 311,189-199. Mitchell, A. W., K. Secoy, N. Mardas, M. Trivedi, and R. Howard (2008), Forests now in the fight against climate change, Forest Foresight Report 1. v3, Global Canopy Programme, Oxford. Moorcroft, P. R. (2006), How close are we to a predictive science of the biosphere?, Trends Eeal. Evol., 21, 400--407. Nepstad, D. C., C. R. de Carvalho, E. A. Davidson, P. H. Jipp, P. A. Lefebvre, G. H. Negreiros, E. D. da Silva, T. A. Stone, S. E. Trumbore, and S. Vieira (1994), The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures, Nature, 372, 666-669. Nepstad, D. C., et ai. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505-508. Nepstad, D. C., G. Carvalho, A. C. Barros, A. Alencar, 1. P. Capobianco, J. Bishop, P. Moutinho, P. Lefebvre, U. L. Silva, and E. Prins (200 I), Road,raving, fire regime feedbacks, and the future of Amazon forests,' For. Ecol. Manage., 154, 395--407. Nepstad, D. C., et aL(2002), The effects of partial throughfall exclusion on canopy iprocesses, aboveground productiou and biogeochemistlY of an Amazon forest, J. Geophys. Res., 107(D20), 8085, doi:10. 1029/2001JD000360. Nepstad, D. C., P. Lefebvre, U. L. Da Silva, 1. Tomasella, P. Schlesinger, L. Solorzano, P. Moutinho, D. Ray, and J. G. Benito (2004), Amazon drought and its implications for forest flammability and tree growth: A basin-wide analysis, Global Change BioI., 10, 704-717. Nepstad, D., et ai. (2006), Inhibition of Amazon deforestation and fire by parks and indigenous lands, Conserv. BioI., 20, 65-73. Nepstad, D. C., 1. Tohver, D. Ray, P. Moutinho, and G. Cardinot (2007), Mortality of large trees and Iianas following experimental drought in an Amazon forest, Ecology, 88, 2259-2269. Noble, 1. R., A. M. Gill, and G. A. V. Bary (1980), McArthur's fire danger meters expressed as equations, J. Ecol., 5, 201-203. Nobre, C. A., P. 1. Sellers, and 1. Shukla (1991) Amazonian deforestation and regional climate change, J. CUm., 4, 957-988. Oliveira, R. S., T. E. Dawson, S. S. O. Burgess, and D. C. Nepstad (2005), Hydraulic redistribution in three Amazonian trees Oecologia, 145, 354-363. ' Ostle, N. 1., et ai. (2009), Integrating plant-soil interactions into global carbon cycle models, J Eeal., 97, 851-863. Oyama, M. D., and C. A. Nobre (2003), A new climate-vegetation equilibrium state for tropical South America, Geophys. Res. Lett., 30(23), 2199, doi:1O.1029/2003GLOI8600. Pegoraro, E., et ai. (2004), Effect of drought on isoprene emission rates from leaves of Quercus virginiana Mill, Atmos. Environ., 38(36),6149-6156.
448
EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
Peylin, P., P. Bousquet, C. Le Quere, s, Sitch, P. Friedlingstein, Sitch, S., et al. (2008), Evaluation of the terrestrial carbon cycle, future plant geography and climate-carbon cycle feedbacks usG. McKinley, N. Gruber, P, Rayner, and P. Ciais (2005), Muling five Dynamic Global Vegetation Models (DGVMs), Global tiple constraints on regional CO 2 flux variations over land and Change BioI" 14,25, oceans, GloJal Biogeochem, Cycles, 19, GBI011, doi:lO.l0291 Soares-Filho, B. S., D, C. Nepstad, L. M. CUlTan, G. C. Cerqueira, 2003GB002214, R. A. Garcia., C. A. Ramos, E. Voll, A. McDonald, P. Lefebvre, Phillips, O. L., et al. (2009), Drought sensitivity of the Amazon and P, Schlesinger (2006), Modelling conservation in the Amarainforest, Science, 323, 1344-1347. zon basin, Nature, 440, 520-523, Potter, C" et al. (2004), Understanding global teleconnections of climate'to regional model estimates of Amazon ecosystem car- Sotta, E. D., E. Veldkamp, L. Schwendenmann, B. R. Guimadies, R. K. Paixao, M. L. P, Ruivo, A. C. L Costa, and P. Meir bon fluxes, Global Change BioI., 10, 693-703. (2007), Effects of an induced drought on soil C02 efflux and Prentice, C., and 1 Lloyd (1998), C-quest in the Amazon, Nature, soil CO 2 production in an Eastern Amazonian rainforest, Brazil, 396,619. Global Change BioI., 13(10), 2218-2229, doi:l0.11ll/j.1365Ray, D., D. Nepstad, and P, Moutinho (2005), Micrometeorologi2486.2007.01416.x. cal and canopy controls offire susceptibility in a forested AmaTardieu, F" and T. Simonneau (1998), Variability among species zon landscape, Ecol, Appl., 15,1664-1678, of stomatal control under fluctuating soil water status and evapoRodenbeck, C., S. Houweling, M. Gloor, and M. Heimann (2003), rative demand: Modelling isohydric and anisohydric behaviours, CO 2 flux history 1982-2001 inferred from atmospheric data usJ. Exp. Bot., 49, 419-432. ing a global inversion of atmospheric transport, Atmos. Chem. Tian, H., 1 M. Melillo, D. W. Kicklighter, A. D. McGuire, 1 V, K. Phys, , 3,1919-1964. Helfrich, B. Moore, and C. J. Vorosmarty (1998), Effect of inRuivo, M., S. B, Pereira, B. Quanz, and P. Meir (2003), Characteriterannual climate variability on carbon storage in Amazonian zation and classification of soils of the LBA experimental sites, ecosystems, Nature, 396, 664-667, Caxiuana, PA, Revista de Ciencias Agrarias, 75-85. Salazar, L. F., C. A. Nobre, and M, D. Oyama (2007), Climate Timmermann, A. (1999), Detecting the nonstationary response of ENSO to greenhouse warming, J. Atmos. Sci" 56, 2313-2325. change consequences on the biome distribution in tropical South America, Geophys, Res, Lett., 34, L09708, doi:l0.l0291 Tomasella, 1, and Hodnett, M. G. (1997), Estimating unsaturated hydraulic conductivity of Brazilian soils using soil-water reten2007GL029695. tion data, Soil Sci., 162, 703. Saleska, S. R., S. D. Miller, D. M, Matross, M, L. Goulden, S. C. Wofsy, H. R. da Rocha, P. B de Camargo, P. Crill, B. C. Daube, Tmmbore, S. U. S. A, (2006), Carbon respired by terrestrial ecosystems-Recent progress and challenges, Global Change BioI. and H. C, de Freitas (2003), Carbon in Amazon forests: Unex12,141-153. pected seasonal fluxes and disturbance-induced losses, Science, Trumbore, S., andP. B. de Camargo (2009), Soil carbon dynam302, 1554-1557. ics, Geophys. Monogr, Ser., doi:l0.1029/2008GM000741, this Saleska, S. R., K. Didan, A. R. Huete, and H. R, da Rocha (2007), volume. Amazon forests green-up during 2005 drought, Science, 318, Turnbull, M. H., et al. (2001), Responses ofleafrespiration to tem612. perature and leaf characteristics in three deciduous tree species Saleska, S" H. da Rocha, B. Kruijt, and A. Nobre (2009), Ecovary with site water availability, Tree Phys., 21,571-578. system carbon fluxes and Amazon forest metabolism, Geophys. Tyree, M. T., and J. S. Speny (1988), Do woody plants operate Monogr. Ser" doi: 10.1029/2008GM000728, this volume. near the point of catastrophic xylem dysfunction caused by dySampaio, G., C. Nobre, M. H. Costa, P. Satyamurty, B. S. Soaresnamic water stress-Answers from a model, Plant Phys" 88, Filho, and M. Cardoso (2007), Regional climate change over 574-580. eastern Amazonia caused by pasture and soybean cropland expansion, Geophys, Res. Lett., 34, LI7709, doi: 10.10291 Uhl, C" and J. B. Kauffman (1990), Deforestation, fire susceptibility, and potential tree responses to fire in the eastern Amazon, 2007GL030612. Ecology, 71,437-449. Schwendenmatrn, L., E. Veldkamp, T, Brenes, 1 1 O'Brien, and 1 Mackensen (2003), Spatial and temporal variation in soil CO 2 van der Werf, G. R, 1 T. Randerson, G. 1 Collatz, L. Giglio, P. S. Kasibhatla, A. F. Arellano, S. C. Olsen, and E. S. Kasischke efflux in an old-growth neotropical rain forest, La Selva, Costa (2004), Continental-scale partitioning of fire emissions during Rica, Biogeochemistl)l, 64, 111-128, the 1997 to 2001 EI NinolLa Nina period, Science, 303, 73-76. Schymanski, S. 1, M. Sivapalan, M. L. Roderick, J. Beringer, and L. B. Hutley (2008), An optimality-based model of the coupled van Vuuren, D. P., P. L. Lucas, and H. Hilderink (2007), Downscaling drivers of global environmental change: Enabling use of soil moisture and root dynamics, Hydrol. Earth Syst, Sci" 12, global SRES scenarios at the national and grid levels, Global En913-932. viron. Change, 17(1), 114-130, doi:l0.1016/j,gloenvcha.2006. Sellers, P. J., Y. Mintz, Y. C. Sud, and A. Dalcher (1986), A simple 04.004. biosphere model (SiB) for use within general circulation models, Vasconcelos, S. S., et al. (2004), Moisture and substrate availability J. Atmos. Sci" 43, 505-531. constrain soil trace gas fluxes in an eastern Amazonian regrowth Sitch, S., et al. (2003), Evaluation of ecosystem dynamics, plant geforest, Global Biogeochem, Cycles, 18, GB2009, doi:lO.l0291 ography and terrestrial carbon cycling in the LPJ dynamic global 2003GB002210. vegetation model, Global Change BioI" 9,161-185,
MEIR ET AL. Vasconcelos, S, S., D. 1 Zarin , M. M. Araujo, L. G. T RangelVasconcelos, C. 1 .It. De Carvalho, C. L. Staudhammer, and F. A. Oliveira (20,98), Effects of seasonality, litter removal and dry-season irrig~tion on litterfall quantity and quality in eastern Amazonian for~t regrowth, Brazil, J. Trop. Ecol., 24, 27-38. Vourlitis, G. L.,<1. de Souza Nogueira, N. P. Filho, W. Hoeger, F. Raiter, M. S. Biudes, 1 C. Arruda, V. B. Capistrano, 1 L. Brito de Faria, and F., de Almeida Lobo (2005), The sensitivity of diel CO 2 and H 20 vapor exchange of a tropical transitional forest to seasonal variation in meteorology and water availability, Earth Interact,,9, 1-23, doi:1O.1175/EIl24.1. Werth, D., and R. Avissar (2002), The local and global effects of Amazon deforestation, J. Geophys, Res., 107(D20), 8087, doi: 10.1029/200 IJD000717, West, A. G., K. R. Hultine, 1 S. Sperry, S. E. Bush, and 1 R Ehleringer (2008), Transpiration and hydraulic strategies in a pinonjuniper woodland, Ecol. Appl., 18, 911-927. White, A., M. G, R. Cannell, and A. D. Friend (1999), Climate change impacts on ecosystems and the terrestrial carbon sink: A new assessment, Global Environ. Change, 9, S21-S30. Williams, M., et al. (1996), Modelling the soil-plant-atmosphere continuum in a Quercus-Acer stand at Harvard forest: The regulation of stomatal conductance by light, nitrogen and soil/plant hydraulic properties, Plant Cell Environ" 19, 911-927. Williams, M" Y. Malhi, A. D. Nobre, E. B. Rastetter,). Grace, M. G. P. Pereira (1998), Seasonal variation in net carbon exchange and evapotranspiration in a Brazilian rainforest: A modelling analysis, Plant Cell Environ., 21,953-968. Woodward, F. I., and C. P, Osborne (2000), The representation of root processes in models addressing the responses of vegetation to global change, New Phytol., 147,223-232. Wright, I. 1, P. B. Reich, O. K. Atkin, C. H. Lusk, M. G. Tjoelker, and M. Westoby (2006), Irradiance, temperature and rainfall
449
influence leaf dark respiration in woody plants: Evidence from comparisons across 20 sites, New Phytol., 169, 309-319. Zeng, N., A. Mariotti, and P. Wetzel (2005), Tel'l'estrial mechanisms of interannual CO 2 variability, Global Biogeochem, Cycles, 19, GBI016, doi:1O.1029/2004GB002273.
S. Almeida, Museau Paraense Emilio Goeldi, Av. Perimetral 1901, Terra Firme, CEP 66077-830, Belem, PA, Brazil. P. M. Brando and D. Zarin, Department of Botany, University of Florida, Gainesville, FL 32611, USA. G. Cardinot, Instituto de Pesquisa Ambiental da Amazonia, Av. Nazare 669, 66035-170, Belem, PA, Brazil. A. C. L. Costa, Centro de Geociencias, Universidade Federal do Para, Belem, CP 1611 66017-970, Brazil. E. Davidson, Woods Hole Research Center, Falmouth, MA 02540, USA. R. A. Fisher, Los Alamos National Laboratory, Los Alamos, NM 87545, USA. P. Meir, School of Geosciences, University of Edinburgh, Edinburgh EH8 9XP, UK. (pmeir@ed,ac.uk) D. Nepstad, Gordon and Betty Moore Foundation, Palo Alto, CA 94304, USA. ' E. D. Sotta, EMBRAPA Amapa, Rod. Juscelino Kubitscheck km 05, 2600, CEP p8903-419, Macapa, AP, Caixa Postal 10 CEP 68906-970, Brazil. . S. Vasconcelos, EMBRAPA-Amazonia Oriental, Belem, PA CEP 66095-100, Brazil.
Soil Carbon Dynamics Susan Tmmbore Department ofEarth System Science, University of California, Irvine, California, USA
Plinio Barbosa de Camargo Laboratof)1 ofIsotope Ecology, CENAIUSP, Piracicaba, Brazil
The amount of organic carbon (C) stored in the upper meter of mineral soils in the Amazon Basin (~40 Pg C) represents ~3% of the estimated global store of soil carbon. Adding surface detrital C stocks and soil carbon deeper than 1 m can as much as quadmple this estimate. The potential for Amazon soil carbon to respond to changes in land use, climate, or atmospheric composition depends on the form and dynamics of soil carbon. Much (~30% in the top ~ 10 cm but >85% in soils to 1 m depth) of the carbon in mineral soils of the Oxisols and Ultisols that are the predominant soil types in the Amazon Basin is in forms that are strongly stabilized, with mean ages of centuries to thousands of years. Measurhble changes in soil C stocks that accompany land use/land cover change occur in the upper meter of soil, although the presence of deep roots in forests systems drives an active C cycle at depths > 1 m. Credible estimates of the potential for changes in Amazon soil C stocks with future land use and climate change are much smaller than predictions of aboveground biomass change. Soil organic matter influences fertility and other key soil properties, and thus is important independent of its role in the global C cycle. Most work on C dynamics is limited to upland soils, and more is needed to investigate C dynamics in poorly drained soils. Work is also needed to relate cycles of C with water, N, P, and other elements.
I. INTRODUCTION: WHY IS SOIL CARBON IMPORTANT? Globally, soils store at least twice as much carbon (C) as the atmosphere. Hence, changes in soil C stores can potentially play an important role in intera1ll1ual to decadal variationsin the global C cycle, and management of C during
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000741
land use change could be significant in tenns of regional or national C commitments. In the Amazon Basin, uncertainties about the degree to which stores of soil C in the large areas of intact forest will change with climate and elevated CO 2 or with future land cover or land use change are part of what limits our ability to estimate future feedbacks between ecosystem C stocks and atmospheric CO 2 levels [Cox et al., 2000; Fl'iedlingstein et al., 2006]. Intact Amazon tropical forests have been suggested to be current carbon sinks to a degree that nearly offsets deforestation losses [Stephens et al.,' 2007], though the role of soils is uncertain in this balance. Some models predict large soil C losses from intact forests under scenarios of future climate change [Falloon et 451
452
SOIL CARBON DYNAMICS
al., 2007], and land clearing also has the potential to destabilize soil C. It is, thus, necessary to examine the capacity for Amazon soils to store or lose carbon, which, in turn, requires knowing (l~how much C is stored in soils and (2) how rapidly it can be stabilized or destabilized. Soil organic matter controls key soil properties, in particular those associated with overall nutrient availability, water holding <;apacity, and fertility that make it important beyond its role in the global C cycle. Decomposition processes in soils are major sources or sinks of important non-C0 2 trace gases (methane, oxides of nitrogen, carbon monoxide). Further, the presence of Amazon "black earth" soils, anthropogenically altered soils with large stocks of stabilized organic matter and high fertility, suggest the potential to manage even soils generally considered unfertile to store carbon [Woods, 2003]. This chapter will focus on current understanding of the dynamics of carbon in Amazonian soil organic matter and comment on the potential role of Amazon soils in the perturbed global carbon cycle. It will summarize the key factors needed to determine soil C dynamics: (1) the total inventory of organic C in soils; (2) the fluxes of C into and out of soils; and (3) the age ofC stored in and respired from soil. The potential for soil organic matter to behave as a source or sink of CO 2 can then be assessed from estimates of how C fluxes into and out to the soil pool are altered by climate, vegetation, or land use change and the time C resides in soil organic matter. Each ofthese soil properties, in turn, is related to soil fmming factors: climate, organisms (vegetation and soil fauna), parent material, topographic position, and time [Jenny, 1947]. Studies of Amazon soil carbon and its dynamics have tended to focus on old, highly weathered soils developed under tena firme tropical forest or cerrado vegetation that cover most of the area of the Basin. Much of the attention of these studies has been on the influence ofland use and its effects on C and nutrient cycling and ultimately soil fertility [see Tiessen et al., 1994; Lehmann et al., 2001; Cerri et al. 2007a, 2007b; Paul et al., 2008]. In contrast, this review will focus on the dynamics and cycling of carbon in intact forest soils and their role in the overall ecosystem carbon cycle. It will also focus on upland, well-drained soils; further studies are needed to understand C dynamics in seasonally flooded and wetland soils.
2. ORGANIC CARBON STOCKS IN AMAZON SOILS Most estimates of the amount of carbon stored as organic matter in Amazonian soils are based on profile data collected by the Radambrasil Project (1973-1986) and compiled in the SOTER database [Batjes, 2005; Batjes and Dijkshoorn, 1999]. Researchers attempting to make basin-wide estimates
TRUMBORE AND CAMARGO
of C storage from these profile data have used different assumptions of how to fill in missing data for bulk density, depths between samples, and to extrapolate spatially from limited profile data; summaries can be found in Table 1 of Batjes [2005] or Table 2 of Cerri et al. [2007b]. Estimates of Amazonian soil C stocks to 1 m depth range from 41 to 47 Pg C for the ~500 Mha of the Brazilian Amazon [Cerri et al., 2007b; Moraes et al., 1995]; Batjes and Dijkshoorn [1999] report a total storage of 66.9 Pg C for the top 1 m of the total ~ 700 Mha of the Amazon Basin. Some 44-67% of the C stored in the top meter is present in the 0-30 cm depth interval [Bernoux et al., 2006]. Most (50-75%) of the area of the Amazon basin is represented by two soil orders: Acrisols (Ultisols, US Taxonomy or Podz61ico, Brazilian taxonomy) and Ferralsols (Oxisols or Latossolo); other important soil orders include Gleysols and Leptosols (each about 8% of the area) [Batjes and Dickshorn, 1999; Cerri et al., 2000, 2007b]. C stocks in Oxisols and Ultisols soils are on average similar, from 44 Mg C ha- 1 (Ultisols) to 55 Mg C ha- I (Oxisols) in the top 30 cm and 85 (Ultisols) to 100 Mg C ha- I (Oxisols) in the top 100 cm, respectively [Batjes and Dijks,horn, 1999]. Carbon to nitrogen ratios for 0-30 cm are ~10 (Ultisols) to 13 (Oxisols), decreasing to val\.Jes of 9-11 by ~ 1 m depth [Batjes and Dijkshorn, 1999]. Soils within the same Order tend to have lower C stocks in Acre State (western Amazon) than their counterparts in the central and eastern Amazon [Melo, 2003]. For example, average C stocks in Acre state are estimated at 38 (0-30 cm) and 66 Mg C ha-1 (0-100 cm) [Melo, 2003]. In general, dystrophic soils (more highly weathered, with lower base saturation and lower fertility) tend to have higher C stocks than eutrophic soils (less weathered, higher base saturation, and more fertile soils). Soils in the western Amazon tend to be younger and more eutrophic, which may explain lower C stocks observed by Melo [2003]. Soil C stocks vary locally with factors like topography and land management, and more regionally with soil parent material and underlying geology [Cerri et al., 2004; Holmes et al., 2004, 2006]. In local studies of soil C and N stocks along topographic gradients near Manaus and Santarem, soils graded from Oxisols on plateaus, to Ultisols on slopes, and Spodosols in valleys (Arenosols in the Brazilian classification) [Luiziio et al., 2004; Telles et al., 2003]. Spodosols in periodically flooded lowlands that dissect plateaus at sites near Manaus and Santarem have distinctly higher sand and lower clay contents, with lower soil C stocks and higher C: N ratios in the upper 10--40 cm of soil [Luiziio et al., 2004; Telles et al., 2003]. The published estimates for soil C stocks do not include superficial litter [Batjes and Dijkshorn, 1999], which can add
a substantial amount ofrapidly cycling carbon to these totals (Table 1). Leaf litter ~nventories vary seasonally but contain from 3 to 7 Mg C h~i in primary forests [Dantas and Phillipson, 1989]. Stoclgloflitter can be larger in secondary growth forests and are Jgenerally much smaller in agroforests and pastures. Estiriates of coarse (> 10 cm diameter) woody debris on in mature forest floorrange from ~1O to 35 Mg C m-2 (assuming biomass is 50% C) [Chambers et al., 2001b; Keller et al., 2004; Rice et al., 2004]. Other litter components, branches, fruits, etc., can make up an additional 15 Mg C ha- 1 [Chambers et al., 200 Ib], though this component is not often quantified. Hence, detritus (also referred to as necromass) on the forest floor contains roughly the same amount of C as is found in the upper 30 cm of mineral soil. Soil carbon stocks calculated to I m depth may be considered a minimum for total soil C storage, as many Amazonian forests are deep-rooting [Nepstad et al., 1994]. Even though the concentrations of C in soils decrease dramatically with depth, the large volumes of soil mean that the total amount of C stored from 1 to 8 m depth can equal what is stored in the top 30 cm [Trumbore et al., 1995; Camargo et al., 1999]. Root standing stocks belowground are also seldom considered in estimates of soil carbon. Telles et al. [2003] found that fine root biomass was 2-2% of total C stocks in
Table 1. Representative Carbon Stocks in Unmanaged Terra Finne Forest (Oxisols) Inventory (Mg C ha- I )
Stock Surface litter Fine
3-7
Medium
15
Coarse
10-35
Root biomass a 0-10 em 10-100 em 100-800 em Soil organic matter 0-30 em 30-100 em 100-300 em 300-800 em
Reference Dantas and Phillipson [1989], Selva et al., [2007] Chambers et al. [2001b] Chambers et al. [2001b], Keller et al. [2004], Rice et al. [2004]
0.2-0.5, 0.6-1.2 0.5,0.7 0.6,0.7
Silver et al. [2004] Trumbore et al. [2006]
44-55
Batjes and Dijkshorn [1999]
28-47 ~70
Camargo et al. [1999]
~100
aLive biomass is shown in roman; dead biomass is shown in italic.
453
the upper meter of primary forest soils (after sieving to remove objects >2 mm), and roughly a third of that in the 0-10 cm layer. Live and dead fine roots clearly drive active C and water cycles deeper than 1 m [Nepstad et al., 1994; Oliveira et al., 2005; Trumbore et al., 2006; Fisher et al., 2007]. Surface soil (0-20 or 0-30 cm) C stores in Amazon soils are generally low compared to temperate or boreal soils [Sanchez and Buol, 1975]. The 40-50 Pg C in the top meter of Brazilian Amazon soils (66 Pg C for the whole basin) represents only about 2-5% of global mineral soil C stocks to 1 m depth, though the Amazon represents roughly 14% ofland area. Including necromass (surface detritus) and soil depths greater than 1 m more than double estimates of total C stored in intact forest soils of the Amazon (Table 1), making soil C stocks roughly equivalent to aboveground carbon stocks for mature forests. The question of most interest for understanding the role ofAmazonian soil C in the global carbon budget, however, is not how much C is stored, but what fraction of that organic carbon is in fmms that can accumulate or be released on timescales of the next decades to centuries. 3. CARBON FLUXES INTO AND OUT OF AMAZON FOREST SOILS The C added annually to soils includes fine and coarse litterfall, tree mortality, and root mmiality. Only a small fraction of this added carl;lOn ends up stored as soil organic matter; the vast majority lis decomposed to CO 2 [Luiziio and Schubart, 1987; Parton et al., 2007]. Relatively few measures of C inputs are available in the literature for Amazonian forests (Table 2a), and this is a major limitation to understanding forest C dynamics and how it varies in space and time. Surface litter fluxes range from 2 to 4 Mg C ha- 1 a-I (summarized by Cattiano et al. [2004D, some 40-67% by mass of which is leaves [Martius et al., 2004; Selva et al., 2007]. Fewer estimates of belowground productivity exist; recent estimates are ~1 Mg C ha- I a- 1 [Silver et al., 2005; Trumbore et al., 2006]. Vieira et al. [2005] used dendrometer measurements in the western, central, and eastern Amazon to estimate that C allocated to woody stem growth is ~2 Mg C ha-1 a-i. Assuming a steady state forest, this provides an average estimate for woody debris inputs. For the Tapajos forest, stem growth increment is in accord with the estimates of coarse (> 10 cm diameter) necromass production by Palace et al. [2008]. Palace et al. [2008] also estimate medium (5-10 cm diameter) and fine (2-5 cm diameter) branch and woody necromass inputs in an intact forest as adding an additional ~ I Mg C ha- 1 a-I. Allocation of net primmy production (l'TPP) by the estimates in Table 2a (leaf litterfall: stem! branch growth:root production of ~ 3--4:2-3: 1) contrast with estimates built in to ecosystem models of roughly equal C
454
SOIL CARBON DYNAMICS
TRUMBORE AND CAMARGO
Table 2a. Representative C Fluxes in Unmanaged TelTa Finne Forests Reference
Flux Litterfall <2cm 2-10 cm >10 cm Fine root production
3.0-6.2 1.0 2.3 2.0-2.2" 0.65-1.0
Export as POC
0.0176
Export as DOC
0.0315
Cattiano et al. [2004] Palace et al. [2008] Palace et al. [2008] Vieira et al. [2005] Silver et al. [2005], Trumbore et al. [2006] Johnson et al. [2006], Selva et al. [2007] Johnson et al. [2006]
"More than 10 cm inputs derived from estimates of annual woody biomass increment and the assumption of steady state. Table 2b. Fluxes of Carbon in Soil Respiration, Litterfall, and Total Belowground Carbon Allocation in Selected LBA Sites"
Paragominas (PA) Tapajos Forest (PA)
Soil Respiration
Litterfall
TBCAb
Reference
20.0
4.3
15.7
12.8
5.7-6.3
6.57.1
Davidson et al. [2000] Brando et al. [2008], Rice et al. [2004] Chambers et al. [2004] Salimon et at. [2004]
Manaus (AM)
12.1
3.3
8.8
Acre (AC)
17.0
4-6 c
11-13
"Fluxes are in Mg C ha- 1 a-I. bTBCA, total belowground C allocation equal to soil respiration minus litterfall. CLitterfall estimated from older secondary forests in Acre.
allocation of NPP to leaf litter, stem growth, and root productivity [e.g., Fung et at., 1997]. Soil respiration rates (Table 2b) integrate CO 2 derived from plant as well as microbial respiration sources and, hence, are larger than estimated soil C inputs. A proposed measure of total belowground C allocation (TBCA) [Ryan, 1991] is determined as soil respiration rates minus surface (leaf) litterfall. Using repOlied soil respiration and litterfall estimates from several studies, greater belowground C allocation occurs in sites with extended seasonal drought (Paragominas and Acre) compared to the central Amazon (Manaus), although the data for the Tapaj6s forest near Santarem does not follow this pattern (see Table 2b). Given low root productivity (Table 2a), high TBCA values also indicate that the large amounts of C allocated belowground are allocated to forms other than root growth, e.g., root respiration or transfer to the soil through symbiotic fungi or root exudates.
EXpOli ofcarbon from soils in dissolved inorganic (DIC) or organic forms (DOC) is an important source of C to streams, but these fluxes are small compared to C inputs or soil respiration rates. Richey et at. [2002] estimated that most of the C respired by the Amazon River system originates in soils and that this export could be large enough to explain the difference between C uptake reported for eddy covariance studies and what was observed accumulating in vegetation and soils in the footprints of the flux towers. However, estimates of fluxes from upland forests are small (~0.03% oflitterfall) and suggest upland litter export does not supply large amounts of particulate C to streams compared to riparian and seasonally flooded areas (Table 2a) [Setva et at., 2007]. Soil pore space COz is likely the largest contributor to stream DIC [Johnson et at., 2006; Setva et at., 2007]. Dissolved organic C fluxes from upland soils are smaller than DIC fluxes. Remington et at. [2007] demonstrated lower sorption capacity of sandier lowland soils compared to upland clay-rich Oxisol soil material, supporting the idea that much of the in-stream source of DOC and even DIC may be in the riparian zone [McCtain et at., 1997]. In terms of the net C balance of upland soils, DIC and DOC fluxes are smalL compared to other terms in the annual soil C budget. 4. DYNAMICS OF SOIL C IN INTACT FORESTS Carbon in soils is not homogeneous; it consists of a wide range of chemical compounds that differ in their intrinsic rates of decomposition and the degree to which they may be stabilized through interactions with mineral surfaces or inclusion in aggregates. Soil biogeochemistry models like Century or RothC parameterize this by identifying pools of carbon, metabolic and resistant plant residues, microbial biomass, and "active," "slow," and "passive" pools, with decomposition rates that vaty from years or less for "active" pools to millennia for "passive" pools. A major and continuing challenge has been to estimate the amount of C in each of these pools from observable characteristics of the soils themselves. Therefore, predicting the response of soil C stocks to changes in land cover or climate requires an understanding of the rates at which stored C is replenished, and the rates at which C may be rapidly stabilized or destabilized when conditions change. Several methods have been employed for determining the rate at which C cycles through terrestrial ecosystems. The first of these compares the stocks of C stored in soil organic matter with the rate at which new C is added to soils as detrital material or lost by decomposition or leaching. This approach, if applied uniformly across all soil organic matter types, offers a long-term average that likely overestimates the shOli-term response of soil C to a change [Trumbore,
2000]. For example, an estimate based on the C inventories in Table I and inputs in Table 2a will Vaty depending on what soil depth (pL30, 0-100, 0-800 cm) is considered. A second approacl'l'is to observe changes in C stocks after some disturbance/though this depends on the availability of appropriate chr6nosequences for the study of dynamics of C on longer timescales. Carbon isotopes, including l3C as well as cosmogenic and bomb-produced radiocarbon, provide some of our best information on the rates of accumulation and loss of organic matter from soils. Even in situations where C loss or gain over decades can be measured directly, such as in long-term cultivated plots, isotopes give important clues as to what kind of C is lost and how long it might take to accumulate. Measures of C dynamics using stable C isotopes rely on differences in the fractionation of l3C and 12C during the fixation ofCO z during photosynthesis. For example, in tropical pastures where plants with predominantly C3 photosynthetic pathways are replaced with C4 grasses, the large difference in the isotopic signature of C can be used to distinguish the rate of loss offorest soil carbon from the rate of accumulation of grass-derived C. Changes in l3C have proved especially useful in studying tropical soils that have been 'converted from C3 forests to pastures dominated by C4 grasses; see reviews by Batesdent et at. [1987, 1998,2000]; Bernoux et at. [1998]; Votkoff and Cerri [1987]. However, there are a number of potential complications in applying the stable isotope methods, including unceliainties in the stable isotope signature of the end-member vegetation sources [Vetdkamp and Weitz, 1994]. Degraded pastures often include not just C4 grasses but a mix of shmbs and grasses which makes identification of an "end-member" isotopic signature difficult. Plant roots and leaves, and the different tissues present within them, may have different isotopic signatures and contribute disproportionately to stabilized soil organic matter pools. Further, many models interpreting changes in stable isotopes assume that the turnover time of forest C3 and grass C4 inputs is the same, which is not necessarily true given the different tissue chemistly for these plant types [Wynn and Bird, 2007]. Stable isotopes cannot be used to assess changes in C dynamics compared to intact forests experiencing no vegetation change. Radiocarbon may be used to study C dynamics on two timescales. Prior to 1950, the radiocarbon age can be used to infer the dynamics of C cycling on centUly to millennial timescales [e.g., Paut et at., 1997]. Atmospheric testing of nuclear weapons (which largely took place between 1960 . and 1964) nearly doubled the amount of 14C in atmospheric CO2 and produced a global isotopic tracer for organic matter dynamics [Trumbore, 2000, 2006]. As with stable isotopes, there are complications to the interpretation of radiocarbon
455
data in terms of the dynamics of the carbon in soils. First, radiocarbon provides a measure of the time elapsed since the C in organic matter was first fixed from the atmosphere by plants; it thus includes the time spent in living plant tissues . . th e estimate III 0 f 14C" age. " For example, wood detritus from a century-old tree might decompose within a few decades, but the "age" of the decomposing C will integrate both timescales; any attempt to infer decomposition rates from radiocarbon must account for this effect. Also, unless multiple samples are available from various points in time since 1950 [e.g., Telles et at., 2003], several different models of C dynamics can be used to explain the same set of observations of radiocarbon content. Interpretation of radiocarbon data should report the sensitivity of model-derived turnover times to such unceliainties. Changes in the 14C of Amazonian soil organic matter since 1960, combined with observations of the rate of change of soil C stocks and l3C signatures in disturbed soils, show definitively that C in soils has several intrinsic timescales of accumulation and decomposition and that modeling all soil organic matter as a homogeneous pool with a single turnover time is clearly overestimating response on decadal to centulY timescales [Telles et at., 2003]. For example, initially rapid changes in the amount and l3C signatures of organic carbon in surface ~oils following conversion to C4 grassdominated pasture demonstrate the presence of fast-cycling organic matter pools. However, the persistence of SaM that is hundreds to thousands of years old and derived from C3 plant sources even in decade-old pastures [Tiessen et at., 1994; Camargo et at., 1999] signifies that a large fraction of SaM cycles much more slowly. Attempts to physically or chemically separate SaM into fractions that cycle on intrinsically different timescales have met with limited success. Nonetheless, some generalizations can be made. Turnover times are fastest for low-density «2 g cm-3) or sand-sized (>63 /lm) organic matter that mostly represents relatively fresh litter and root detritus [Lehmann et at., 2001; Paut et at., 2008], while the oldest C in soils is strongly associated with clay mineral surfaces [Telles et at., 2003]. While sources of low-density organic matter (e.g., dead roots) do not show large trends with soil depth [TrulIIbore et at., 2006], other fractions do increase in age with depth (Figures 1 and 2). In forest soils, Telles et at. [2003] identified three different components of soil C that cycled on different timescales: (1) light (density <2 g cm-3) fraction, particulate organic material that could still be identifiedas plantdetritus with more depleted ol3 C signatures, and radiocarbon ages of decades or less; (2) milwral-associated C that is solubilized in acids and bases, l3C-enriched, with radiocarbon ages of decades at the surface to millennia at depths >20 cm, and (3) nonhydrolyzable,
456
TRUMBORE AND CAMARGO
SOIL CARBON DYNAMICS 70 60
' t c
50 -
o years I!iI decades-centuries l!iI>centuries
40 ~ 30
20 10 • 0 u:=:J - ----'-------------'-~ Litter
Q.10em
1Q.1 OOcm 100-300em 30Q.500cm
Figure 1. Age distribution of carbon stored in soils with depth. Data are from Telles et al. [2003] and Trumbore et al. [1995]. Ages ofC are derived from radiocarbon data in soil organic matter fractions and infelTed from modeling C fluxes and isotopic signatures.
l3C-depleted C strongly associated with clay surfaces with ages of hundreds ofyears (in the 0- to lO-m layer) to >20,000 years (deeper than 20 cm). The amount of C in each of these fractions is shown in Figure 1. Soil texture, in this case clay content, exerts a major control on the amount of slowly cy-
........ E
1
2
3
-30
0
0
20
20
t.> -.......-
40
.s:: .....
40
60
60
80
80
100
100
Q.. Q)
0
0
........
E
~
.s:: ...... Q..
1
-21
-800 -600 -400 -200 0
I
........... Llght
fraction _ _ Heavy Fraction
_ _ Hydrolysis ResIdue
-e-27
-24
Hydrolyzed
20 40 60 80 100
-21
-800 -600 -400 -200
0
0
20
20
20
40
40
40
60
Degraded Pasture
80
60
60
80
80
0
1
2
-30
3
-27
-21
-800 -600 -400 -200 0
0
0
0
20
20
20
.s:: .....
40
40
40
60
60
60
0
80
80
80
100
100
........ E
t.> -.......Q.. Q)
100
Managed Pasture
0
200
100
100
100
200
0
-30
3
-24
-27
0
Q)
0
2
Ll14C (%0)
813 C (%0)
C (%) 0
cling carbon and therefore influences the storage and dynamics of carbon in tropical forest soils. Telles et al. [2003] also demonstrated predictable relationships between the l3C and 14C content of soil organic matter and soil clay content that are potentially useful for scaling relationships among Oxisol and Ultisols with similar soil age and vegetation. As shown in Figure 2, ofthe total~lOO Mg Cha- I in the top meter of an Oxisol in an intact forest, ~5% is in forms with ages of several years or less, ~28% in forms fixed from the atmosphere decades-centuries ago (the half of this in the upper few centimeters is in the form of decades, the remainder centuries), and the rest (>65%) in forms with ages averaging many thousands of years. By contrast, the mean "turnover time," one would calculate for soil C to 20-30 cm depth, defined as the inventory of carbon divided by the rates of C additionorloss would be~40 Mg Cha-I/~7 MgC ha- I a-I or~6 years (Tables 1 and 2). Increasing the depth interval to 100 cm would more than double the C inventory without increasing C inputs significantly (~100 Mg C ha-I/~7 Mg C ha- l a-I or ~ 14 years). The mean radiocarbon ages of carbon for the
200
CD
Figure 2. Comparison of C in soil fractions versus depth for three ecosystem types: forest, degraded pasture (~20 years after conversion) and managed (fertilized, planted with productive grass) pasture developed ~5 years previously on degraded pasture. Data are from sites near Paragominas, PA and reported by Camargo et al. [1999] and Telles et al. [2003]. Note low 14C content of the hydrolysis residue (dark solid line) compared to hydrolyzed carbon (dashed line) and changes in 13C for all fractions in pasture management near the surface.
upper 20 cm of soil range from 200~380 years for clay-rich soils. All of these me~sures show that C cycles on different intrinsic timescales;,the mean "turnover time" averages over components withrfImch younger and older ages. The use of the mean "turno/ver time" would overestimate short-term (i.e., decadal) tisponses to disturbance [Telles et al., 2003]. The above results are for well-drained upland soils with high clay content. Radiocarbon data are rare for seasonally flooded Spodosols or wetlands. Telles et al. [2003] reported data for a Spodosol near Manaus that showed that the majority of the C to 40-m depth was fixed in the post-1963 period. Given the hydromorphic nature ofthis material, it is likely that these soils are not representing steady state conditions, more work is needed to understand the dynamics of these soils. 5. RATES OF SOIL C CHANGE WITH LAND MANAGEMENT Forest clearing for pasture or agricultural use results in changes in the soil physical, biological, and chemical environment. Soil physical structure is altered, along with soil temperature and moisture regimes. Allocation patterns above- and belowground are altered, and thus· so are the quality, quantity, and vertical distribution of litter inputs to soil. Changes in C stocks are largest and most rapid in sandsized organic matter (equivalent to low-density or pm-ticulate organic matter), followed by silt-sized fractions, with little or no change in the organic material associated with clay-sized particles [Shang and Tiessen, 1997; Lehmann et al., 2001]. Figure 2 shows typical differences in fractionated soil C and isotopes between intact forest (top panel), a l7-year-old degraded pasture (middle panels), and a managed pasture (lower panel). The fertilized and managed pasture has gained carbon, and the degraded pasture lost carbon, compared to the forest soil. These C gains or losses are pronounced only in the surface ~20 cm and are accompanied by changes in l3C associated with the C4 pasture grasses. Largest changes in l3C and C content are observed in hydrolyzable components (organic matter associated with sesquioxides and weakly bound to clay minerals), compared to nonhydrolyzable residues. Although changes in C with depth are attenuated and difficult to detect with depth, changes in rooting depth and root production rates between forest and pasture grasses can be accompanied by significant additional C gains above ~2 m (for very productive grasses) and losses below ~2-3 m [Trumbore et al., 1995; Camargo et al., 1999]. These changes will occur over decades, and there may be a delay associated with root lifetimes that can range up to a decade or more [Trumbore et al., 2006]. Similarly, models of C increase may have to take into account the time lag required
457
for shorter-lived grass roots to increase inputs of dead root material to the upper few meters of soil; this may explain why Camargo et al. [1999] predicted larger-than-observed changes in the l3C of SaM in their managed pasture site. On conversion fi'om forest to pasture, C inventory may increase, decrease, or stay about the same [Shang and Tiessen, 1997; Neill and Davidson, 1999; Holmes et al., 2006]. Summaries of detailed chronosequence studies [Neill and Davidson, 1999] and more spatially extensive data sets [Holmes et al., 2006] demonstrate that the overall direction and magnitude of change in SOC following forest clearing is broadly predictable from the original forest soil carbon content and pH. Decreases in C occur when initial forest C content is high and replaced by less productive pasture or agricultural vegetation. Increases in C occur when initial forest surface soil C content is low, and vegetation is replaced with highly managed and fertilized productive grasses. Rates of initial change in C stocks can be rapid (~5% per year) but decline over time. For sites with declining C content, reductions in the amount of fast-cycling C mean overall reduction in the supply of soil nutrients derived from mineralization of organic matter and loss of feliility without additional fertilization [Tissen et al., 1994]. Using an ecosystem C model (CentulY) where the turnover time of C in active slow and passive pools is controlled by litter quality, clim~te, and soil texture, Sehimel et al. [1994] showed the predicted average residence time for SaM in the upper 20 cm 9f soils in tropical regions can range from <14 years up to several decades. However, data in Figure 2 suggest that the model underestimates the residence time of C in passive C pools (the inventOly-weighted age of C for the soil in Figure 2 is several hundred years). Such differences between models and observations affecting vety long timescales may not be impOliant for predictions of changes on timescales of a few decades. For example, Cerri et al. [2007a] have used the Centmy model to successfully reproduce observed changes in C inventOly and l3C in several pasture sites following conversion in Rondonia. Studies of soil C change with land use at specific sites are useful for understanding the dynamics and local controls on C cycling. However, spatial extrapolations require assumptions about the areal extent of land management practices. For the state of Rondonia, Holmes et al. [2006] were able to take advantage of a large data set of C large regional soil profile database and remotely sensed classification ofland cover to estimate the net C storage or loss associated with land cover changes across a range of original soil properties and management regimes. Although individual locations could have either large C gains or losses, the spatially averaged result was a net small loss of C for the state of Rondonia. Hence, while site data are impOliant for managing individual
458
SOIL CARBON DYNAMICS
plots for soil fertility, uncertainties in net C storage or loss at the landscape scale may depend more on the distribution of management across the landscape. Cerri et alft [2007b] ~nd Falloon et al. [2007] have predicted future soil C stock changes based on scenarios ofland cover and climate change for the Amazon Basin. Cerri et al. [2007b] predict a ~ 7% decline in Amazon soil C stocks (0- to 30-1)1 depth) between 2000 and 2030, balancing losses in newly cleared areas with increases in areas abandoned or used for secondaly forest regrowth. Falloon et al. [2007] \predicted very large net loss of C with climate change alone, I due to drying and decrease of forest C productivity. However, the magnitude of their estimated loss (a decline from 45 Pg C to 23 Pg C between 2000 and 2100) is too large given that >70% of the soil C in the 0- to 30-cm layer is in soil organic matter fractions with tumover times longer than centuries. The large predicted losses likely result from the use of a single pool model for soil C that overpredicts C changes in the short term [Knorr et al., 2005]. Using more realistic models of C dynamics, losses from soils will be small to minimal compared to changes in aboveground biomass that occur with deforestation and reforestation. It is the fate of tree biomass that will determine the overall magnitude of tropical land use change as a source of C to the atmosphere over the next decades. 5. J. Anthropogenic Soils
Soils with the greatest C storage in the Amazon are the "terra preta do Indio" or Indian Black Earth soils. These soils, known as Antrhoposols, or human-generated soils, are associated with areas inhabited by indigenous peoples from 500 to 2500 years ago and abandoned after European arrival [Woods, 2003]. Organic C is enriched to 1 to 2 m depth compared to adjacent soils, with much of the stabilized C thought to be in the form of charred or "black" C; [Glaser et al., 2001]. The carbon stabilized in these soils remains hundreds to thousands of years after they were abandoned. Indian Black Earths are also high in phosporous, CEC, pH, and base saturation and, consequently, are more fertile than surrounding soils. For a review, including the potential for people to manage soils to sequester carbon, in essence to recreate these soils, see Lehmann et al. [2003]. 6. TIME LAGS BETWEEN PHOTOSYNTHESIS AND RESPIRATION AND THE ESTIMATION OF C SINK POTENTIAL IN INTACT (UNMANAGED) FORESTS Undisturbed tropical forests have been proposed as a potentially large sink for anthropogenic carbon based on inver-
TRUMBORE AND CAMARGO
sions of regional atmospheric CO 2 concentration variations [Stephens et al., 2007]. What role might soils play in a regional C sink? The capacity for an ecosystem component such as soil to serve as a net sink of carbon may be estimated from the magnitude of gross fluxes and the time C resides in each cascading C pool [Fung et al., 1997; Thompson et al., 1996]. Because of the heterogeneous nature of soil carbon, the age of C respired from soils is nearly always younger than the age of bulk C stored in soils [TrulIlbore, 2000]. This is due to the fact that most of the C being respired is from pools that cycle slowly, while pools with the longest residence times have the largest C stocks. An estimate of the time lag between photosynthesis and respiration based on radiocarbon data [Trumbore et al., 2006; Telles et al., 2003], stemwood production [Vieira et al., 2005] and respiration fluxes [Chambers et al., 2004] in a central Amazon forest near Manaus is illustrated in Figure 3. Even though much of the C fixed is respired autotrophically and over a relatively short time period, the remaining C either stays in living leaf and root tissues for an average 1-3 years (leaves), or longer for the ~50% oflitterfall that is not leaves, and 5-10 years (roots), then decomposes rapidly. Overall even though only ~20% of the total ecosystem respiration is estimated to be from microbial decomposition Total ecosystem respiration -30 (3 -7 yr) Total autotrophic respiration -23.7 (0.01-1 yr)
Total heterotrophic respiration -6.3 (15-29 yr)
litter 3.3 (2-3 yr) Wood 2.0 (40-80 yr*) Root/SOM 1.0 (8-18 yr)
Figure 3. Calculation of the mean age of ecosystem respired CO2, Fluxes are based on Chambers el al. [2004]; mean age of decomposing wood based on Vieira el al. [2005] and wood decomposition rates of Chambers el al. [200Ib]; age of C derived from dead root decomposition from Tl'llmbare el al. [2006]; mean age oflitter respired C02 from Branda el al. [2008] and unpublished data of P. B. Camargo and S. Trumbore (2008).
of plant residues and soil organic matter, the time lags involved are significant1/Results in Figure 3 generally agree with those published in Fung et al. [1997], which estimated the mean age of h~terotrophicallyrespired C to be 24 years for broad-leave¥vergreen tropical forests. Models of C dynamics indicate vegetation and soils under conditions of increased forest productivity should remain sinles for as long as inputs outstrip the increases in decomposition or mortality rates. For example, Chambers et al. [200la] used an individually based model of tree growth and mortality, forced with a 25% increase in productivity over a period of 50 years (an increase of 0.25%/year) to estimate maximum C sequestration rates of ~0.5 Mg C ha- I year- I in woody biomass in the period over which forcing was applied. Models of C dynamics in soil organic matter calibrated using observations of l4C [Telles et al., 2003] used the same forcing scenario (assuming all inputs would increase at the same rate as overall NPP with no additional time lag) and showed that rates of net C accumulation in soil (assuming decomposition rates remained unchanged) would not be larger than about 0.1 Mg C ha- I a-I in the upper 40 cm of soil. According to Trumbore et al. [1995], this rate could potentially be doubled if changes in root producti~ity below 40 cm were taken into account. Hence, the overall predicted rate ofC sequestration would be in the order of 0.7 Mg C ha- I a-I, with the majority C storage in aboveground biomass. Given that the increased NPP derived from C02 feltilization would not be expected to increase quite as rapidly, a more realistic estimate, using a simple box model that forces increases NPP in a tropical forest (fluxes in Figure 3) with a ~ factor of 0.2 is represented in Figure 4 [see also Chambers and Silver, 2005]. Soil C storage lags vegetation storage and is responsible for a much smaller fi'action of the total estimated C sinle for the early 2000s (at ~ 13% of a total ecosystem C sink estimated at ~0.12 Mg C ha- I a-I). However, soil C storage will continue for decades after NPP increases cease because ofthe time lags in living vegetation. Multiplying an estimated sink of 0.12 Mg C ha- I a-I times the area of the Brazilian Amazon (~5 x 108 ha) would result in a net C sink associated with CO 2 fertilization of 0.07 Pg C a-I, 13% of which would be in soil. A sink ofthis magnitude is not large enough to balance regional C losses from deforestation and is not easily detectable given current methods for determining forest C balance. Recent measurements of permanent plots in Panama and Malaysia [Feely et al., 2007] have documented declines in stem wood increment at .the stand level, rather than the increases that might be expected· with C02 fertilization. Long-term analysis ofpermanent plot data in the Brazilian Atlantic forest have similarly shown a tendency for rapid C loss associated with sudden mortality events, followed by periods of more rapid tree growth
459
0.10
0.09
-Vegetation sink
0.08 ~L
0.07 ~>, 0.06 Co 0.05 .c
<6,
-Soil sink
0.04
::2: 0.03 0.02
::::=,.--
0.01 0.00 .~n~_"""":'..:::, 1900
1930
1960
----.-~ 1990
Figure 4. Estimated C sink for the forest ecosystem pictured in Figure 3, assuming a 8 factor of 0.2 for CO 2 fertilization, the record of measured atmospheric C02 from 1800 to 1990 and a linearly increasing rate of CO 2 increase from 1990 to 2010 that matches observations through 2007 (C0 2 concentrations in 2010 are estimated at 395 pap. The soil sink lags the vegetation sink and is only ~13% ofthe total sink: of 0.12 Mg C ha- 1 a-I estimated for the year 2007.
[Rolim et al., 2005]. Even if productivity increases, it will likely be associated with either changes in vegetation/litter quality or allocatioJ;l among leaves, stems, and roots (particularly involving rooting depth). As such, changes are likely to occur over decaqes; it is difficult to predict the overall impact on predicted d stores in soils. What we can conclude is that changes in soil C stocks have limited potential to offset current deforestation sources; again it is the more dynamic vegetation C pools that will dominate any response. Variations in NPP from 1 year to the next are larger than the long-tenn trends calculated above. For example, interannual variations in litterfall and tree biomass increment of order ~1 Mg C ha- I a-I each have been repOited [Rice et al., 2004; Vieira et al., 2006]. Given a deviation of2 Mg C ha- l a-I (out of ~30) and a time lag of ~3 years between photosynthesis and ecosystem respiration (75% respired in same year, 25% with average of 12 years), the expected interannual variation in net C storage/loss can reach fO.25 Mg C ha- I a-I. If such variations are spatially coherent variations across the Amazon Basin (e.g., like those associated with ENSO climate anomalies), they have the potential to contribute significantly to observed interannual variations in global CO 2 accumulation in the atmosphere.
7. OUTSTANDING QUESTIONS The major conclusion of this review is that changes in soil C in response to land use or climate change, or even CO 2 fertilization will be minimal compared to changes in
460
SOIL CARBON DYNAMICS
TRUMBORE AND CAMARGO
aboveground C pools over the next centuly. Why then argue Balesdent, 1, C. Chenu, and M. Balabane (2000), Relationship of soil organic matter dynamics to physical protection and tillage, for more research in Amazonian soil C? Most importantly, Soil Till. Res., 53, 215-230. soil organic matter largely determines the fertility and sustainability 01' pastures and agricultural land [Tiessel1 et al., Batjes, N. H. (2005), Organic carbon stocks in the soils of Brazil, Soil Use Manage., 21, 22-24. 1994]. More studies are needed to elucidate the processes Batjes, N. H., and 1 A Dijkshoorn (1999), Carbon and nitrogen stocks stabilizing C in soils and links between how C is stabilized in the soils ofthe Amazon Region, Geoderma, 89, 273-286. and how long it remains in the soil. Cycling of organic mat- Bernoux, M, C. C. Cerri, C. Neill, and J. F. L. de Moraes (1998), ter is inexorably linked to biogeochemical cycling of water, The use of stable carbon isotopes for estimating soil organic matnitrogen, phosphorous, and other elements, which are keys ter turnover rates, Geoderma, 82, 43-58. to soilfertility and agricultural'sustainability. Research link- Bernoux, M., et al. (2006), Cropping systems, carbon sequestration and erosion in Brazil, a review, Agron. Sustain. Dev., 26, ing C and nutrient dynamics is clearly needed. 1-8. While the overall dynamics of C in the major soil orders found in the Amazon basin have been explored in a prelimi- Brando, P. M., D. C. Nepstad, E. A. Davidson, S. E. Tmmbore, D. Ray, and P. Camargo (2008), Drought effects on litterfall, wood naly way, the amount of data, especially for intact forests, production and belowground carbon cycling in an Amazon forremains small. Recent studies have demonstrated some maest: Results of a throughfall reduction experiment, Philos. Trans. jor surprises in how intact forests cycle carbon, especially: R. Soc. Ser. B, 363,1839-1848. (1) the low carbon use efficiency ofthese forests [Chambers Camargo, P. B., S. E. Trumbore, L. A. Martinelli, E. A. Davidson, et al., 2004] and (2) the longevity of live fine root biomass D. C. Nepstad, and R. L. Victoria (1999), Soil carbon dynamics [Trumbore et al., 2006]. However, these conclusions are in regrowing forest of eastern Amazonia, Global Change Bioi., based on data from very few sites and may not be representa5,693-702. tive of the Amazon basin as a whole. In particular studies of Cattiano, 1 H., A. B. Anderson, 1 S. Rombold, and D. S. Nepstad (2004), Phenology, growth, and root biomass in a tidal floodplain C stocks and fluxes in wetland and seasonally flooded soils forest in the Amazon estuary, Rev. Bras, Bot., 27, 703-712. need to be augmented with isotopic measures of C dynamics Cerri, C. C., M. Bernoux, D. An'ouays, B. 1 Feigl, and M. C. Picon decadal and longer timescales. colo (2000), Carbon stocks in soils of the Brazilian Amazon, in Some of the C in soils is very old (>25,000 years), and Global Climate Change and Tropical Ecosystems, edited by R. retains evidence of C4 vegetation sources likely dating from Lal, 1 M. Kimble, and B. A. Stewart, pp. 33-50, CRC Press, the last glacial period [Sal1iotti et al., 2002]. Little is unBoca Raton, Fla. derstood of the processes that can store C for so long-and Cerri, C. E. P., M. Bernoux, V. Chaplot, B. Volkoff, R. L. Victoria, whether it is in the form of char ("black C") or other fOllliS. J. M. Melillo, K. Paustian, and C. C. Cerri (2004), Assessment Amazon landscapes changed dramatically over the last glaof soil property spatial variation in an Amazon pasture: A basis cial cycle, and some of the properties we observe in soils for selecting an agronomic experimental area, Geoderma, 123, today may reflect conditions at that time. 51-68. Another area where more research is needed is in the tools CeITi, C. E. P., et al. (2007a), Simulating SOC changes in 11 land use change chronosequences from the Brazilian Amazon with RothC for extrapolating soil C stocks and dynamics from point and Century models, Agric. Ecosyst. Environ., 122,46--57. measurements to landscapes. Holmes et al. [2006] demonCerri, C. E. P., et al. (2007b), Predicted soil organic carbon stocks strate the impOliance of spatial approaches in determining and changes in the Brazilian Amazon between 2000 and 2030, what the important factors for assessing C at different spatial Agric. Ecosyst. Environ., 122, 58-72. scales are. It is paliicularly important to build an understandChambers, 1 Q., and W. L. Silver (2005), Ecophysiological and ing of the key processes involved in stabilizing and destabibiogeochemical responses to atmospheric change, in Tropical lizing carbon, in particular, the role of minerals, soil fauna, Forests and Global Atmospheric Change, edited by O. Phillips and aggregates, and how these may ValY regionally and in and Y. Mahli, pp. 57-65, Oxford Univ. Press, Oxford, U. K. response to land cover change. Chambers, 1 Q., N. Higuchi, E. S. Tribuzy, and S. E. Trumbore
REFERENCES Balesdent, 1, A. Mariotti, and B. Guillet (1987), Natural C-13 abundance as a tracer for studies of soil organic-matter dynamics, Soil Bioi. Biochem., 19,25-30. Balesdent, 1, E. Besnard, D. Arrouays, and C. Chenu (1998), The dynamics of carbon in particle-size fractions of soil in a forestcultivation sequence, Plant Soil, 201,49-57.
(2001a), Carbon sink for a centulY, Nature, 410, 429. Chambers, 1 Q., J. P. Schimel, and A. D. Nobre (2001b), Respiratio.n from coarse wood litter in central Amazon forests, Biogeochemistry, 52, 115-131. Chambers, J. Q., et al. (2004), Respiration from a tropical forest ecosystem: Partitioning of sources and low carbon use efficiency, Ecol. Appl., 14, S72-S88. Cox, P. M., R. A. Betts, C. D. Jones, S. A. Spall, and 1. 1 Totterdell (2000), Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model, Nature, 408,184-187.
Dantas, M., and 1 Phillipson (1989), Litterfall and litter nutrient content in primaly and secondary Amazonian terra finne rainforest, J Trop. Eco}, 5, 27-36. Davidson, E. A., L 11V. Verchot, 1 H. Cattanio, 1. L. Ackerman, and .T. E. M. CarvaVlo (2000), Effects of soil water content on soil respiration inA'orestsand cattle pastures of eastern Amazonia, BiogeochemistrJI, 48, 53-69. Falloon, P., et al. (2007), Climate change and its impact on soil and vegetation carbon storage in Kenya, Jordan, India and Brazil, Agric. Ecosyst. Environ., 122, 114-124. Feeley, K. 1, S. 1 Wright, M. N. N. Supardi, A. R. Kassim, and S. .T. Davies (2007), Decelerating growth in tropical forest trees, Bcol. Lett., 10,461-469. Fisher, R. A., M. Williams, A. L. Da Costa, Y. Malhi, R. F. Da Costa, S. Almeida, and P. Meir (2007), The response of an Eastern Amazonian rain forest to drought stress: Results and modelling analyses from a throughfall exclusion experiment, Global Change Bioi., 13,2361-2378. Friedlingstein, P., et al. (2006) Climate-carbon cycle feedback analysis: Results from the (CMIP)-M-4 model intercomparison, J Clim., 19, 3337-3353. Fung, 1., et al. (1997) Carbon 13 exchanges between the atmosphere and biosphere, Global Biogeochem. Cycles, 11,507-533. Glaser, B., L. Haumaier, G. Guggenberger, and W. Zech (2001), The 'Terra Preta' phenomenon: a model for sustainable agriculture in the humid tropics, Naturwissenscha.ften, 88, 37-41. Holmes, K. W., D. A. Roberts, S. Sweeney, 1. Numata, E. Matricardi, T. W. Biggs, G. Batista, and O. A. Chadwick (2004), Soil databases and the problem of establishing regional biogeochemical trends, Global Change Bioi., 10, 796-814. Holmes, K. W., O. A. Chadwick, P. C. Kyriakidis, E. P. S. de Filho, .T. V. Soares, and D. A. Roberts (2006), Large-area spatially explicit estimates of tropical soil carbon stocks and response to land-cover change, Global Biogeochem. Cycles, 20, GB3004, doi: 10.1 029/2005GB002507. Jenny, H. (1947), Factors of Soil Formation, 241 pp., McGrawHill, New York. Johnson, M. S., 1 Lehmann, E. C. Selva, M. Abdo, S. Riha, and E. G. Couto (2006), Organic carbon fluxes within and streamwater exports from headwater catchments in the southern Amazon, Hydrol. Processes, 20, 2599-2614. Keller, M., M. Palace, G. P. Asner, R. Pereira, and 1 N. M. Silva (2004), Coarse woody debris in undisturbed and logged forests in the eastern Brazilian Amazon, Global Change Bioi., 10, 784795. Knorr, W., 1. C. Prentice, J. 1. House, and E. A. Holland (2005), Long-term sensitivity of soil carbon tumover to wanning, Nature, 433, 298-301. Lehmann, 1, M. Silva Cravo, and W. Zech (2001), Organic matter stabilization in a Xanthic Ferralsol of the central Amazon as affected by single trees: Chemical characterization of density; aggregate, and paIiicle size fi'actions, Geodel'lna, 99, 147168. Lehmann, J., D. C. Kern, B. Glaser, and W. 1. Woods (2003), Amazonian Dark Earths: Origin, Properties, Management, 523 pp., Kluwer Acad., Dordrecht, Netherlands.
461
Luizao, F. 1, and H. O. R. Schubart (1987), Litter production and decomposition in a terra-finne forest of central Amazonia, Experientia, 43, 259-265. Luizao, R. C. C., F. 1 Luizao, R. Q. Paiva, T. F. Monteiro, L. S. Sousa, and B. Kl1lijt (2004), Variation of carbon and nitrogen cycling processes along a topographic gradient in a central Amazonian forest, Global Change Bio!., 10, 592-600. Martius, C., H. Hofer, M. V. B. Garcia, 1 Rombke, and W. Hanagarth (2004), Litter fall, litter stocks and decomposition rates in rainforest and agroforeshy sites in cenh'al Amazonia, NutI'. Cycl. Agroecosyst., 68, 137-154. McClain, M. E., 1 E. Richey, .T. A. Brandes, and T. P. Pimentel (1997), Dissolved organic matter and terrestrial-lotic linkages in the central Amazon basin of Brazil, Global Biogeochem. Cycles, 11,295-311. Melo, A. W. F. (2003), Avalia<;:ao do estoque e composi<;:ao isot6pica do carbono do solo no Acre, M.S. thesis, 74 pp., University of Sao Paulo, Piracicaba, SP. Moraes, 1 L., C. C. Cerri, J. M. Melillo, D. Kicklighter, C. Neill, D. L. Skole, and P. A. Steudler (1995), Soil carbon stocks of the Brazilian Amazon basin, Soil Sci. Soc. Am. J, 59, 244-247. Neill, C., and E. A. Davidson (1999), Soil carbon accumulation or loss following deforestation for pasture in the Brazilian Amazon, in Global Climate Change and Tropical Ecosystems, edited by R. Lal, 1 M. Kimble, and B. A. Stewart, pp. 197-212, CRC Press, Boca Raton, Fla. Nepstad, D. C., et ~l. (1994), The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures, Nature, 372, 666-669. Oliveira, R. S., L. Ji!ezerra, E. A. Davidson, F. Pinto, C. A. Klink, D. C. Nepstad, imd A. Moreira (2005), Deep root function in soil water dynamics in cerrado savannas of central Brazil, Funct. Ecol., 19, 574-581. Palace, M., M. Keller, and H. Silva (2008), Necromass production: Studies in undisturbed and logged Amazon forests, Ecol. Appl., 18,873-884. Parton, W., et al. (2007), Global-scale similarities in nitrogen release patterns during long-tenn decomposition, Science, 315, 361-364. Paul, E. A., R. F. Follett, S. W. Leavitt, A. Halvorson, G. A. Peterson, and D . .T. Lyon (1997), Radiocarbon dating for detennination of soil organic matter pool sizes and dynamics, Soil Sci. Soc. Am. J., 61,1058-1067. Paul, S., H. Flessa, E. Veldkamp, and M. Lopez-Ulloa (2008), Stabilization of recent soil carbon in the humid h'opics following land use changes: Evidence from aggregate fi'actionation and ,stable isotope analyses, Biogeochemisty, 87, 247-263. Remington, S. M., B. D. Strahm, V. Neu, 1 E. Richey, and H. B. da Cunha (2007), The role of sorption in control of riverine dissolved organic carbon concentrations by riparian zone soils in the Amazon basin, Soil Sci., 172, 279-291. Rice, A. H., et al. (2004), Carbon balance and vegetation dynamics in an old-growth Amazonian forest, Eco!. App!., 14, S55-S71. Richey, .T. E., 1 M. Melack, A. K. Aufdenkampe, V. M. Ballester, 'and L. L. Hess (2002), Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric CO 2 , Nature, 416,617-620.
462
SOIL CARBON DYNAMICS
Rolim, S. G., R. M. Jesus, H. E. Nascimento, H. T. Z. do Couto, and J. Q. Chambers (2005), Biomass change in an Atlantic h'opical moist forest: The ENSO effect in permanent sample plots over a 22-year peri~, Oecologia, 142,238-246. Ryan, M. G. (1991), A simple method for estimating gross carbon budgets for vegetation in forest ecosystems, Tree Physiol., 9,255-266. Salimon, C. I., E. A. Davidson, R. L. Victoria, and A. W. F. Melo (2004), COz flux from soil in pastures and forests in southwestern Amazonia, Global Change BioI., 10, 833-843. Sanchez, P. A., and S. W. Buol (1975), Soils of the tropics and world food crisis, Science, 188, 598-603. Saniotti, T. M., L. A. Martinelli, R. Victoria, S. E. Trumbore, and P. B. Camargo (2002), Past vegetation changes in Amazon Savannas determined using carbon isotopes of soil organic matter, Biotropica, 34, 2-16. Schimel, D. S., B. H. Braswell, E. A. Holland, R. McKeown, D. S. Ojima, T. H. Painter, W. J. Parton, and A. R. Townsend (1994), Climatic, edaphic, and biotic controls over carbon storage and turnover in soils, Global Biogeochem. Cycles, 8, 279-293. Selva, E. C., E. G. Couto, M. S. Johnson, and J. Lehmann (2007), Litterfall production and fluvial expOli in headwater catchments of the southem Amazon, J. Trop. Ecol., 23, 329-335. Shang, C., and H. Tiessen (1997), Organic matter lability in a tropical oxisol: Evidence from shifting cultivation, chemical oxidation, particle size, density, and magnetic fractionations, Soil Sci., 162, 795-807. Silver, W. L., A. W. Thompson, M. E. McGroddy, R. K. Vamer, J. D. Dias, H. Silva, P. M. Crill, and M. Keller (2005), Fine root dynamics and trace gas fluxes in two lowland tropical forest soils, Global Change BioI., 11, 290-306. Stephens, B. B., et al. (2007), Weak nOlihern and strong tropical land carbon uptake fi'om vertical profiles of atmospheric COz, Science, 316,1732-1735. Telles, E. D. C., P. B. de Camargo, L. A. Maliinelli, S. E. Trumbore, E. S. da Costa, J. Santos, N, Higuchi, and R. C. Oliveira (2003), Influence of soil texture on carbon dynamics and storage potential in tropical forest soils of Amazonia, Global Biogeochem. Cycles, 17(2), 1040, doi:lO.l029/2002GBOOI953. Thompson, M. V., J. T. Randerson, C. M. Malmstrom, and C. B. Field (1996), Change in net primaly production and heterotrophic respiration: How much is necessalY to sustain the terrestrial carbon sink?, Global Biogeochem. Cycles, 10,711-726.
Tiessen, H., E. Cuevas, and P. Chacon (1994), The role of soil organic-matter in sustaining soil fertility, Nature, 371, 783-785. Trumbore, S. (2000), Age of soil organic matter and soil respiration: Radiocarbon constraints on belowground C dynamics, Ecol. Appl., 10, 399-411. Trumbore, S. (2006), Carbon respired by terrestrial ecosystemsRecent progress and challenges, Global Change BioI., 12, 141153. Trumbore, S. E., E. A. Davidson, P. B. de Camargo, D. C. Nepstad, and L. A. Martinelli (1995), Belowground cycling of carbon in forests and pastures of eastern Amazonia, Global Biogeochem. Cycles,9,515-528. Trumbore, S., E. S. da Costa, D. C. Nepstad, P. B. de Camargo, L. A. Martinelli, D. Ray, T. Restom, and W. Silver (2006), Dynamics of fine root carbon in Amazonian tropical ecosystems and the contribution of roots to soil respiration, Global Change BioI., 12, 217-229. Veldkamp, E., and A. M. Weitz (1994), Uncertainty analysis of the l3C method in soil organic matter studies, Soil BioI. Biochem., 26(2), 153-160. Vieira, S., et al. (2004), Forest structure and carbon dynamics in Amazonian tropical rain forests, Oecologia, 140,468-479. Vieira, S., S. Trumbore, P. B. de Camargo, D. Selhorst, J. Q. Chambers, N. Higuchi, and L. A. Martinelli (2005), Slow growth rates of Amazonian trees: Consequences for carbon cycling, Proc. Natl. Acad. Sci. U. S. A., 102,18,502-18,507. Volkoff, B., and C. C. CelTi (1987), Carbon isotopic fractionation in subtropical Brazilian grassland soils-Comparison with tropical forest soils, Plant Soil, 102,27-31. Woods, W. I. (2003), Development ofanthroposol research, inAmazonian Dark Earths: Origin, Properties, Management, edited by J. Lehmann et aI., pp. 3-14, Springer, Dordrecht, Germany. Wynn, J. G., and M. I. Bird (2007), C4-derived soil organic carbon decomposes faster than its C3 counterpart in mixed C3/C4 soils, Global Change BioI., 13, 2206-2217.
P. B. de Camargo, Laboratory ofIsotope Ecology, CENAlUSP, Piracicaba, SP 13416-000, Brazil. ([email protected]) S. Trumbore, Department of Earth System Science, University of California, Irvine, CA 92697-3100, USA. ([email protected])
Ecophysiology of Forest and Savanna Vegetation 1. Lloyd,l M. L. Goulden,2 J. P. Ometto,3 S. Patifio,4 N. M. Fyllas, l and C. A. Quesada 5
Ecophysiological characteristics of forest and savanna vegetation are compared in an attempt to understand how physiological differences within and between these vegetation types relate to their geographical distributions. A simple ordination first shows that although precipitation exerts a key effect on Amazonian vegetation distributions, soil characteristics are also impmiant. In patiicular, it is found that under similar precipitation regimes, deciduous forests tend to occur on more fertile soils than do cenado vegetation types. A high subsoil clay content is also impmiant in allowing the existence of semievergreen forests at only moderate rainfall. Such observations are consistent with biome specific physiological characteristics. For example, deciduous trees have higher nutrient requirements than do evergreen ones which also tend to have characteristics associated with severe water deficits such as a low specific leaf area. Nutrient contents and photosynthetic rates are lower than for savanna than for forest species with several eco,system characteristics suggesting a primary limitation of nitrogen on savanna productivity. By contrast, phosphorus seems to constrain the productivity of most Amazonian forest types. Differentiation is made between the fast-growing, high-nutrient-requiring forest types of western Amazonia and their counterpatis in eastern Amazonia, which tend to occupy infertile but deeper soils of high water-holding ability. On the basis of observed physiological characteristics of the various vegetation fmills, it is argued that, should Amazonian precipitation decline sharply in the future, the slower growing forests of eastern Amazonia will transform directly into an evergreen cerrado type vegetation but with the more fertile western Amazonian forests being replaced by some form of drought-deciduous vegetation.
I School of Geography, University of Leeds, Leeds, UK. zEarth System Science, University of California, Irvine, California, USA. 3Centro de Ciencias do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos, Brazil. 4Grupo de Ecologia de Ecosistemas Terrestres Tropicales, UniversidiJd Nacional de Colombia, Sede Amazonia, Instituto Amazonico de Investigaciones-Imani, Leticia, Colombia. 5Institito Nacional de Pesquisas da Amazonia, Manaus, Brazil.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000740
1. INTRODUCTION
In this chapter, we first consider the various vegetation types of Amazonia and the underlying factors influencing their distribution. We then look at their contrasting physiological characteristics in some detail, first at the leaf and plant and then at the whole stand level. Finally, in the spirit of Schimper [1903], we speculate on the extent to which observed differences between the various vegetation types stu,died as part of Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) reflect adaptations to the different soil and climatic environments in which they occur. 463
464
LLOYD ET AL.
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
2. CONTRAST OF AMAZONIAN VEGETATION TYPES 11-
Although dominated by tropical evergreen and semievergreen forest, here we adopt the terminology of Eyre [1963] to delineate the various tropical forest types; see also section 2.3. Amazonia also contains significant amounts of other vegetation'types of a contrasting physiology, especially toward its southern and eastern margins. These include large areas of both well-drained\savanna, often referred to ascerrado, along with seasonally flooded savannas such as those that occur in the Pantanal (Brazil) and the "Llanos de Moxos" (Bolivia and Colombia). Also, along the southern fringe, especially in northern Bolivia, large areas of (semi) deciduous seasonal forest occur (Chiquitano). Savannas are also scattered along the northern borders of the basin, for example, in Roraima State in Brazil and in the La Sabana region of Venezuela. An overview of the lowland tropical vegetation types of South America, including Amazonia, is given by Daly and Mitchell [2000). In this review, we focus on telTe firme forest and cerrado only. 2.1. Primmy Determinants ofthe Different Vegetation Types It has long been apparent that the main factor delimiting tropical forest and savanna is rainfall and its seasonality. Schimper [1903] and Nix [1983] developed seven criteria to predict the OCCUlTence of tropical savannas throughout the world, four of which involved precipitation, two of which involved temperature, and one of which involved incoming radiation. Recently, Huytra et al. [2005] extended the water availability notion, arguing that frequency of drought is also an excellent predictor of the forest savanna boundary, overall supporting the notion of Oyama and Nobre [2003] that seasonality of soil moisture is a critical factor in determining forest-savanna boundaries. Schimper [1903] expressed this point himself, saying that savannas generally are found where there are clearly defined wet and dry seasons, the latter being characterized by very dry soil and a very dry atmosphere, with wet seasons OCCUlTing eveIY year and rarely broken by significant dry periods. Malhi et al. [2009] also developed a simple model of Amazonian vegetation distribution based on rainfall and its seasonality: their savanna zone being defined as where the mean annual precipitation was less than 1500 mm a-I and with a mean cumulative water deficit (MCWD) as defined by Malhi et al. [2009] as the maximum climatological water deficit (CWD) attained over a year with CWD calculated using a "bucket model" and with no feedback effect ofCWD on evaporation, greater than 300 mm. Malhi et al. [2009] also differentiated between "rain
forest" and "seasonal forest," the latter apparently being most prevalent where rainfall is reasonably high (between 1300 and 2000 mm) but with moderate MCWD (350--450 mm) also occurring. Although precipitation and its seasonality are no doubt important factors controlling the relative distributions of forest and savanna across Amazonia, it is also clear that additional factors must be involved. This is because within the Amazon forest domain itself, there are many occurrences of savanna vegetation despite high rainfall (>2000 mm) with unusually adverse physical and/or chemical conditions for tree growth apparently responsible [Beard, 1953; Anderson, 1981; Brown, 1987). Examples of this include the savannas of the relict sandstone cover that once made up much of the land portion of the Guyanas [van Donselaar, 1969]; these include the savannas of Roraima (Brazil) as well as the Rupunini savannas of Guyana [Myers, 1936] and the Sipalwini savanna of southern Suriname [van Donselaar, 1968], the Grand Sabana area of Venezuela [Dezzeo et al., 2004] as well as scattered "islands" across the northeast of Brazilian Amazonia [Andreae Lima, 1959;, Egler, 1960; Ratter et al., 2003]. Although in some cases the presence of such "Amazonian savannas" can clearly be attributed to water-logging [e.g., Huber, 2006], in other cases, the generally poor nutrient status of the sandy soils seems to be the primary cause [Beard, 1953; Anderson, 1981; Brown, 1987). The possibility of poor water-holding capacity of the sandy soils associated with many of the "dIY" savannas giving rise to unusually severe water deficits in the dly season preventing the establishment of forest does not seem to have been investigated. In areas usually associated with (semi)evergreen forest, savannas may also occur on unusually shallow and/or rocky soils [Reatto et al., 1998). 2.2. Variation in Savanna Structure
Within the savanna biome itself, considerable variation exists and explanations for the various physiognomic forms, especially the degree of woodiness, have concentrated on the degree to which variations in the density of woody plants is a function of degradation due to fire and human activity [Rizzini, 1963; Coutinho, 1990; Bond et al., 2005] as opposed to variations in soils such as effective rooting depth, waterlogging, and fertility [Eiten, 1993]. Nevertheless, on balance, there seems little to suggest that burning and other human activities account for large-scale variations in savanna form, although this is no doubt the case under certain circumstances in Brazil as elsewhere [Eiten, 1983; Sarmiento, 1983; Cavelier et al., 1998; Dezzeo et al., 2004). That factors other than fire are important in determining woodiness of cerrado vegetation is also suggested by
plant ordination studies [Ribeiro and Tabarelli, 2002; Miranda et al., 2003] anqipoints to soil fertility being a major determinant of woodifiess in Brazilian savanna at the land11 scape scale [Lopes/and Cox, 1977a, 1977b). For example, it is well establi~~d that some woodland types, cerradao, tend to occur only on unusually fertile soils [Furley et al., 1988; Morriera, 2000; Chapuis-Lardy et al., 2001], and on the basis of soil pH and exchangeable cation measurements, a distinction is sometimes made between "mesotrophic facies cerradao" and "dystrophic facies celTadao," which are characterized by different species compositions. Nevertheless, despite these differences, both vegetation forms seem to be characterized by relatively high soil-soluble phosphorns concentrations [Furley and Ratter, 1988]. Whether the grassland cerrado forms such as campo sujo occur because of soils with an exceptionally low nutrient status is less clear [Alvim and Araujo, 1952; Askew et al., 1970; Goodland and Pollard, 1973; Lopes and Cox, 1977a, 1977b; Ho'ley and Ratter, 1988; Furley, 1992; Ruggiero et al., 2002]. Fire frequencies must also be important. What is clear in any case is that the Brazilian celTado occupies an area much ofwhich on the basis of climate alone would be expected to be pccupied by forests [Bond et al., 2005], and one significant factor accounting for the actual vegetation is the relatively infertile soil there [MontgomelY and Askew, 1983). It is also interesting to note that the seasonally dry tropical forest, which occurs on the southern edges of Amazonia, is considered to exist because of the more fertile soils occurring there, with areas of celTado vegetation in the same regions occurring where soils are of the more infertile type typically associated with Amazonia [Prado and Gibbs, 1993; Prado, 2000; Oliveira-Filho and Ratter, 2002).
465
seasonally dry tropical forest
ECEC(top soil)
L~ semi-evergreen tropical forest
~-----I&t:;;;;;;i~'-----=::"""'Fl------"
CCA-I closed savanna
Texttlre (stlb soil) evergreen tropical forest
Anntlalpmipitatiotl Figure 1. Partial canonical correspondence analysis of Amazonian vegetation types in terms of climate and soil properties. The arrow for soil texture (right to left) goes from fine (clay) to coarse (sand) textured soils, with topsoil being defined as 0.0- to 0.2-m depth and subsoil as 0.21- to 0.50-m depth. For more details, see section 2.3.
vided vegetation into "closed" and "open: on the basis of woodiness: "open" savannas being grassland with or without scattered trees ~nd shrubs (campo limpo, campo sujo, and campo celTado) and "closed" savannas consisting of the tree savanna and woodland forms commonly refelTed to as cerrado (sensu strictu) and celTadao. To remove any spatial autocolTelation structures in the data, latitude and longitude of the centers of the various polygons describing the vari2.3. A Basin- Wide Ordination ofForest and Savanna ous different landfonlls within Amazonia within Cochrane Vegetation Types et al. [1985] have been taken as covariates [Legendre and Legendre, 1998). In an effort to formalize the above vegetation/climate/soil Figure 1 uses "Type 2 Scaling" [tel' Braak, 1994; Legende relationships, we have undeliaken a constrained ordination and Legendre, 1998], where the distances among the various of Amazonian vegetation types from the spatially explicit centroids approximate their chi-square distances in the ordidatabase of Cochrane et al. [1985], which also includes both nation space. The ranking of any vegetation type along any soil chemical and physical characteristics, also investigating environmental variable consists ofprojecting (at right angle) effects of temperature and precipitation, these coming from those vegetation type centroids onto the alTOW representing New et al. [2000]. Results from this partial canonical cor- that variable. This gives an approximation for the weighted respondence analysis are shown in Figure 1. Here as in Co- average for the vegetation type with respect to environmenchraneet al. [1985], we have divided the forest vegetation tal variables and, as an aid, is shown explicitly for annual into three types; "evergreen," "semievergreen," and "sea- . precipitation. The three environmental/edaphic variables sonally dry" following the broad definition of Eyre [1963). shown have been chosen from a group of 26 taken from CoSemievergreen forest is considered to consist of a mixture chrane et al. [1985] and New et al. [2000] using forward seof evergreen and deciduous trees, whereas seasonally dry lection [tel' Braak and Smilauer, 2002] and together account forests consist mainly of species which lose all their leaves for 0.38 of the variation. In addition to annual precipitation, in the dry season. Within the savanna biome, we have di- the other two variables that emerge as important are the
466
LLOYD ET AL.
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
effective cation capacity of the topsoil (ECEC) and subsoil texture, both these parameters coming from Cochrane et al. [1985], In the work of Cochrane et al. [1985], "topsoil" is defined as sdtI fi'om 0.00-0.20 m depth and "subsoil" as soil from 0.21-0.50 m depth. Figure 1 suggests that although precipitation amount is a key environmental variable influencing vegetation distributions aCross Amazonia, soil fertility and texture are also important. In pmiicular, the dish'ibution of closed savmma versus semievergreen h'opical forest seems to be conh'olled as much by topsoil ECEC and subsoil texture as by precipitation and with seasonally dly tropical forest also being associated with feIiile soils as well as a lower than average precipitation. This confirms and, to some extent, formalizes many of the statements in sections 2.1 and 2.2. For example, where soils are unusually infertile and/or subsoils are coarsely textured, savanna type vegetation can occur, even when precipitations are reasonably high. Likewise, the main differentiation between seasonally dry h'opical forests and savanna vegetation types is soil fertility. The key question which we try to answer in this chapter is to what extent these differences in "niche optima" are reflected by differences in plant physiological characteristics for these various vegetation types. 3. PHYSIOLOGY OF FOREST AND SAVANNA VEGETATION 3.1. Structural Aspects
The amount and distribution of biomass for any vegetation type provides a first indication of both physiological strategy and likely limitations on plant function imposed by soil and/or environment. For example, with the Amazonian tropical forest biome as a whole, there is a clear tendency for aboveground biomass (AGB) to decline with increasing dry season length [Malhi et al., 2006; Saatchi et al., 2007, 2009]. This suggests that either carbon resources for growth become more limiting as dly season length increases or that an increased allocation of carbon belowground occurs as precipitation declines [Cairns et aI., 1997; Mokany et aI., 2006]. A third possibility is that more severe soil water deficits for the driest forests result in higher long-term rates of mortality (i.e., an increased frequency disturbance regime) as has recently been suggested by Quesada etal. [2009a]. Although there is no enough evidence yet to clearly differentiate these three possibilities, generally speaking, rootshoot ratios are higher for seasonally dIy-deciduous forests than for evergreen tropical forests [Fittkau and Klinge, 1973; Mwphy and Lugo, 1986; Castellanos et al., 1991; Jipp et aI" 1998; Mokany et al., 2006], and it is clear that significant variations in rootshoot between forest and cerrado exist. For
example, Mokany et al. [2006] cite globally average root shoot ratios of 0.24 and 0.64 for tropical forest and savanna, respectively, and this is consistent with the veIY high belowground biomass values of up to 53 Mg DW ha- I found by Castro and Kaufinann [1998] for cerrado vegetation near Brasilia. This OCCUlTed despite a relatively low AGB of about 17 Mg DW ha- 1. That AGB value may be an underestimate, however, because an allometric equation suitable for forest as opposed to savanna trees was used. Nevertheless, even when more appropriate calculations are applied [Abdala et al., 1998], AGB for dense cerrado vegetation still only amounts to 20-40 Mg DW ha- I [Haridasan, 2000; Quesada et al., 2008]. Although belowground biomass values for Amazonian evergreen forests are typically less than for woody cerrado vegetation, 20-40 Mg DW ha- 1 [Jipp et al., 1998; Metcalfe et al., 2007], aboveground biomass values are much higher, typically 200-300 Mg DW ha- 1 [Malhi et al., 2006; Saatchi et al., 2007]. Although it is tempting to account for these differences in rootshoot solely in terms of physiological adjustments associated with the large differences in soil water regime, which may be characteristic of the two biome types (section 4.1), low rates of soil feIiility (sections 3.6 and 4.2) and the characteristic fire regime of savanna ecosystems, itself associated with the presence of grasses [Miranda et al., 2002], may also be important. Studies with seedlings, in particular, have shown higher rootshoot ratios for savanna as opposed to forest species [Paulilo and Felippe, 1998; Hoffmann et al., 2004]. This allows for greater belowground carbohydrate reserves to be present for young savanna trees, thus facilitating a more rapid growth recovelY after fire than is the case for those from the forest [Hoffinann et al., 2004]. Savanna trees often also have additional physiological and anatomical characteristics associated with fire resistance, such as unusually thick bark [Gignollx et al., 1997] and an ability to resprout from dormant or adventitious buds [Hoffmann and Moreira, 2002].
Schaik et al., 1993; Goulden et al., 2004]. Satellite observations have confirm,~d dly season leaf flushing, showing increases in both pormalized difference vegetation index (NDVI) and enha9ted vegetation index (EVI) late in the dry season [Potter eral., 2001; Huete et al., 2006; Xiao et al., 2006]. Myneni et al. [2007] used the Moderate Resolution Imaging Spectroradiometer (MODIS) LAI product to infer that leaf out creates a large, transient increase in Amazonian forest LAI during the dry season and subsequent decrease in the wet season. However, an alternative hypothesis, that tropical forests simply exchange leaves in the dly season, and LAI remains nearly constant or is reduced during this exchange [Goulden et aI" 2004], remains viable. For example, it is not clear that the MODIS LAI product fully accounts for the increased NIR reflectance by young leaves [Roberts et al., 1998], and this raises the possibility that the observed dly season increases in NDVI and EVI are simply a result of seasonal shifts in mean leaf age and leaf level spectral reflectance. The few in situ time series observations of LAI that have been reported are either ambiguous or too short to fully address this issue [Carswell et al., 2002; Asner et aI., 2004], and longer-term, in situ observations of LAI are still needed for evergreen and semideciduous forest The issue of seasonality in cerrado and seasonally dry tropical forest is more certain. Both cenado trees and seasonal dly tropical forest show the expected seasonality of LAI with significantly lower values in the dly season than in the wet [Vollrlitis et al., 2004; Hoffinann et al., 2005a, 2005b], though with many cerrado trees beginning a new flush of vegetative growth prior to the wet season commencing [Lenz and Klink, 2006]. Cerrado grasses and herbs show sharp reductions in green leaf biomass/LAI toward the end of the wet season with commensurate increases in the amount of dead material present [Miranda et al., 1997; Santos et al., 2003; Hoffinann et al., 2005a; Quesada et aI., 2008]. 3.3. Nutrients and LeafStructure
3,2, LeafArea and Light Interception
Tropical forests have among the highest leaf areas per unit ground area (leaf area index (LAI)) of any biome [Asner et al., 2003] with values for Amazonia typically ranging from 4 to 8 [McWilliam et al., 1993; Carswell et aI" 2002], significantly greater than cerrado vegetation for which the peak LAI (trees and grasses) typically varies from less than 1 to around 2.5 [Miranda et al., 1997; Hoffinann et aI" 2005a]. Quantitative, in situ observations of the timing ofleafproduction are difficult given the height of most h'opical forest canopies, but qualitative and anecdotal observations suggest that new leaf production often occurs in the dly season [van
Although with a high LAT, nutrient contents of the leaves of tropical forest and trees can be surprisingly low compared to trees of the temperate zone, this being especially the case for phosphorus [Reich and Oleksyn, 2004], but only on highly weathered and infertile ferralsol and acrisollalisol soils. (Here we use the new World Reference Base (WRB) for Soil Resources classification system [IUSS Worldng GroupWRB, 2006]. For the soils of the Amazon Basin, equivalences with the USDA system are given by Quesada et al. [2009b].) Such soils are typical of eastern Amazonia and for some forests on the Guinean and Brazilian shields [Quesada et al., 2009b] with the forests growing on them typically
467
having foliar phosphorus of ~0.6 mg g-l DW [Townsend et al., 2007; Fyllas et al., 2009]. In western Amazonia (and generally closer to the Andes), soils are generally younger and more feIiile and with the forests there having higher levels of available soil phosphorus [Quesada et al., 2009c], also reflected in significantly higher foliar phosphorus concentrations, typically 1.0-1.6 mg g-1 DW [Fyllas et al., 2009]. This reduction in soil phosphorus availability as soils age is consistent with soil pedogenic theory [Walker and Syers, 1976], with Quesada et al. [2009a] showing that the higher productivity of forests in western Amazonia [Malhi et al., 2004] is almost certainly attributable to higher levels of available phosphorus being present there. This is consistent with phosphorus being a key determinant of tropical forest productivity, as was first suggested by Vitousek [1984]. Despite these differences in foliar phosphorus concentration, foliar nitrogen concentrations of Amazonian forests are much less variable, being comparable to temperate zone forests [Reich and Oleksyn, 2004], typically averaging 25 mg g-1 DW, although values may be only half of that on white sand (arenosol) or podzol soils [Fyllas et al., 2009]. These relatively high foliar nitrogen contents are consistent with the notion that nitrogen may be available in excess for many tropical forests [Martinelli et al., 1999] with the lower [N] for forests growing on white sand soils, also being consistent with the idea that nitrogen may indeed be limiting for these systems as ittdicated by plant and soil 15N/14N ratios [Martinelli et al., i999; Mardegan et al., 2008; Quesada et al., 2009c] and the dominance of ectomyconhizal associations for the trees growing there [Alexander and Lee, 2005]. Despite the abundance of leguminous trees in Amazonia, it now appears that many of these do not fix nitrogen, even though they clearly have this ability [Nardoto et al., 2008] (see also section 4.2). Most recently, Davidson et aI, [2007] have suggested that the nitrogen availability in terrestrial ecosystems can be ephemeral and eventually disrupted by disturbance; thus, periods of low nutrient availability might arise, for instance, due to limited litter decomposition during the dly season [Saleska et al., 2003], even though on longer-teilli timescales nitrogen may still be relatively abundant. The observations by Davidson et al. [2007] rely on a study canied out in a forest succession following agricultural abandonment in eastern Amazonia using biogeochemical and isotopic parameters, and the patterns of nitrogen and phosphorus cycling, during the succession over decadal time scales, are considered to be compared to nitrogen and phosphorus cycling patterns during primary succession as soils age over thousands and millions of years. When compared to tropical forests, cerrado trees tend to have significantly lower foliar nitrogen and phosphorus
468
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
concentrations (dry weight basis) [Franco, 2002; Hoffinann et al., 2005b], with foliar nitrogen concentrations for both C3 and C4 gra~s species even lower [Miranda et al., 1997]. Although not known at this stage, it seems reasonable to assume that, as is the case elsewhere [Prior et al., 2004], leaves of dry-deciduous forests within Amazonia would be both thinner (i.e., higher specific leaf area (SLA)) and with higher foliar nutrient contents, consistent with their tendency to occur on more feliile soils (section 2.3). 3.4. LeafPhotosynthetic Characteristics
Compared to broadleaf temperate zone tree species, photosynthetic rates of tropical forest and savanna species are relatively low, typically ranging from 5 to 15 Ilmol m-2 S-1 [Medina and Klinge, 1982; Franco, 2002; Carsewell et al., 2002; Domingues et al., 2005; Franco et al., 2005; Miranda et al., 2005; Domingues et al., 2007]. As pointed out by Meir et al. [2002], such low rates occur, despite foliar nitrogen concentrations being similar to temperate zone broadleaf tree species whose photosynthetic rates are typically significantly higher, 25-40 Ilmol m-2 S-I; a result confilmed and expanded by the recent global survey of Kattge et al. [2008], who showed that this effect (i.e., low photosynthetic rate per unit leaf nitrogen for tropical tree species) was most pronounced for trees growing on the relatively infertile felTalsol soils. As there is also now increasing evidence that plants growing on such soils generally have lower foliar phosphOlUS concentrations than those growing on more feliile soils and with this occulTing despite similar foliar N concentrations [Townsend et al., 2007; Fyllas et al., 2009], this raises the interesting possibility that phosphorus rather than nih'ogen may limit photosynthetic rates on such soils. Nevertheless, even on these characteristically phosphorus-poor soils, some analyses have continued to concentrate solely on nih'ogen as the underlying nuh'ient limiting photosynthesis [e.g., Coste et al., 2005; Domingues et al., 2005], even though it is well documented that phosphorus rather than nitrogen can limit photosynthetic rates under some circumstances [Brookes et al., 1984; Campbell and Sage, 2006]. Considerable differences between species in photosynthetic rates can be observed for both savanna and forest [Reich et al., 1994; Prado and De Moraes, 1997; Turner, 2001]. For example, a simple literature survey has shown that forest trees leaves of "shade-intolerant" trees typically have higher photosynthetic rates than "light demanders" (h'ees that can tolerate shade, but require light to express their true growth potential), which in turn are higher than "shadetolerant" plants [Turner, 2001]. Likewise, deciduous tropical trees tend to have higher photosynthetic rates than their evergreen counterparts in both the cerrado [Prado and De
Moraes, 1997; Franco et al., 2005] and elsewhere [Sobrado, 1991; Prior et al., 2004]. Such differences are readily accountable for in terms of physiological trade-offs associated with differing growth strategies [Turner, 2001] and are considered in more detail in section 3.6. Despite their low foliar nih'ogen contents (section 3.2), h'opical C4 grasses are typically capable of higher photosynthetic rates than their C3 counterparts [Pearcy and Ehleringer, 1984; Anten et al., 1998] and with massively higher photosynthetic nitrogen and phosphorus use efficiencies; for a summalY, see discussion section of Mantlana et al. [2008a]. Although not yet measured for cerrado grasses to our knowledge, a simple comparison can be made of leaf level gas exchange and nutrient data of Domingues et al. [2005] for semievergreen tropical forest with that of Anten et al. [1998] for a C4 grass (Hypharrhenia rufa) growing in a central Venezuelan savanna. The highest foliar [N] observed for H. rufa was about 50 rnrnol m-2, this being associated with net CO 2 assimilation rate of around 20 Ilmol m-2 S-I. By contrast, none of the forest leaves studied by Domingues et al. [2005] had [N] less than 70 rnrnol m-2 with these leaves having net C02 assimilation rates of less than 5 Ilmol m-2 S-I. From what little we cUlTently know, cerrado h'ees seem to be intelmediate between these two contrasts but, not unexpectedly, much closer to forest trees. For example, from the study of Franco et al. [2005], the lowest foliar [N] repOlied was around 120 rnrnol m-2 with these leaves having net CO 2 assimilation rates of approximately 10 Ilmol m-2 s-l. These differences are considered in more detail in section 3.6. Seasonal variations in photosynthetic capacity for either forest or savanna trees have yet to be studied in any great detail, although Domingues [2005] reported little evidence of reduced photosynthetic capacities during the dry season for h'ees growing in a semievergreen forest near Santarem. By contrast, Miranda et al. [2005] observed significant reductions in maximum photosynthetic capacities during the dry season for several species growing in a semideciduous (h'ansitional) forest near Sinop, and Franco [1998] and Franco et al. [2005] also reported reduced photosynthetic rates for cerrado tree species during the illy season. These differences between biomes are also being reflected in the strong seasonality of stand level carbon fluxes considered in section 4.2. 3.5. Plant Water Relations
Much work done over the last 10 years as part of the LBA project has confirmed earlier observations of Nepstad et al. [1994] and Hodnett et al. [1995, 1996] that water uptake from considerable soil depths during the dry season allows for the continued functioning of the semievergreen forests
LLOYD ET AL.
469
of eastern and southwestern Amazonia. In particular, Jipp et did not exhibit a similar reverse flow. Moreover, consistent al. [1998] and Bruno e~al. (2006] demonstrated significant with their ability to cany out hydraulic lift, the deciduous water uptake at depth/Uuring the dly season, with the latter and brevideciduous species had both shallow and deep roots study also showing ~fgnificant water uptake at depths greater (i.e., dimorphic root systems), whereas the evergreen species than 10 m for suc!;vtorests. had mostly deep roots only. Evidence was also found that Work by Oliveira et al. [2005a, 2005b] has also shown deciduous and brevideciduous cerrado species may utilize that some forest trees growing in such seasonal environments more water for processes such as nocturnal transpiration. are capable ofhydr:mlic redistribution: that is, the nocturnal The picture that emerges then, is that, for cerrado species, transfer of water from moist to dry regions of the soil profile, there is a trade-off between year-round access to nutrients with this distribution being upward in the dly season (i.e., in the upper soils (the integrity of surface roots for decidufrom moister soil at depth to the drier layers closer to the sur- ous and brevideciduous species being maintained through face) with flow OCCUlTing in the opposite direction in the wet hydraulic lift) versus a greater access to deeper and more season (i.e., from moist surface layers to myer subsurface reliable water during the dry season for evergreen species. layers). This can be interpreted as an indication that plants Lee et al. [2005] observed during the Amazonian dly seagrowing in such regions may utilize hydraulic redistribution son that when forest plants are allowed to redistribute soil to help alleviate drought stress during dry periods by creat- water through hydraulic lift, photosynthesis and transpiraing a readily available pool of water in the superficial soil tion rates were significantly increased. According to these where most of the trees' fine roots are located. Nevertheless, authors, the hydraulic redistribution increases the dry season questions remain. For example, as pointed out by Ludwig et transpiration by 40% over Amazonia, establishing a direct al. [2004], interplant competition can reduce the facilitative link between plant root functioning and climate. effects of hydraulic lift, as it is not only the plants investing Despite this clear ability ofboth forest and savanna h'ees to in deep roots and allowing the uplift of water that are likely utilize water from considerable soil depth during the dly seato benefit from the increased water availability of soil water son, this being potentially aided by hydraulic lift at least for close to the surface. Thus, it might be expected that not all some species, semievergreen forest trees utilizing such deep trees growing in such an environment would exhibit such water still show cle~r evidence of the effects of soil water a phenomenon, with some species simply being "parasitic" deficits during the dry season as evidenced by significantly users of the upper layer soil water transferred from depth more negative midqay leaf water potentials [Domingues, by others. Hydraulic lift has also been observed to occur for 2005; Fisher et al., 2006]. By contrast, only relatively minor some celTado species during the dry season [Scholz et al., differences in midday leaf water potentials between wet and 2002; Moreira et al., 2003] with that study also showing that dry season are observed for celTado trees [Meinzer et al., smaller seedlings growing near the larger trees actually trans- 1999; Bucci et al., 2005; Franco et al., 2005; Goldstein et porting the water also had access to the water made available al., 2008]. Presumably, these differences arise as a consethrough hydraulic lift. Similarly, it is also clear that celTado quence of the different phenological pattems observed for trees can extract water from depth during the dry season, at the two vegetation types, as even evergreen cerrado trees least when growing on deep and highly weathered ferralsol lose some of their leaves in the dry season [Hoffinann et al., soils [Jackson et al., 1999; Oliveira et al., 2005a, 2005b; 2005a], allowing for leaf-specific hydraulic conductances Quesada et al., 2008]. Nevertheless, Dawson et al. [2007] (conductance for water flow from soil to leaf expressed per have observed, for several plant species growing under dif- unit leaf area) to be maintained or even increased during ferent climate conditions, that many plants still transpire at the dry season [Bucci et al., 2005]. As pointed out by Bucci night, especially when slight soil water deficits happen or et al. [2005], this apparent "isohydric" behavior (tendency immediately after a rain event, consistent with isotopic data for leaf-water potentials to remain constant during the day discussed in section 4.3. If the night transpiration exists, it and across seasons) is also facilitated by strong stomatal would reduce the efficacy of the hydraulic lift or hydraulic responses in leaf-to-air water vapor mole fraction differdistribution processes; but as discussed by Goldstein et al. ences (D) in celTado trees, as was also reported by Miranda [2008], nocturnal transpiration, as observed for celTado trees et al. [1997] and Meinzer et al. [1999]. Similarly, Navesby Bucci et al. [2004], may be an adaptive trait allowing . Barbieiro et al. [2000] showed a strong stomatal control enhanced nutrient uptake from nutrient-poor savanna soils. of transpiration for two evergreen species from different This picture was further clarified by Scholz et al. [2008] who vegetation types in the Brazilian cenado. Although somenoted that the OCCUlTence of reverse sap flow for deciduous times construed as being indicative of some "mechanistic" and brevideciduous cerrado species during the dry season stomatal response [Williams et al., 1998], a recent analysis was consistent with hydraulic lift, but that evergreen species of stomatal and whole plant hydraulics suggests "isohydric
470
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
controls" of leaf water potential may actually be much more apparent than real [Buckley, 2005]. From the above discussion, significantly more negative leaf water <potentials for trees growing in semievergreen forests during the dry season can be interpreted as a consequence of the leaf area of such trees not declining during the dly season (see section 4.5). Second, there is also some evidence from leaf exchange experiments ofDomingues [2005] that for some plant species growing on a deep and highly weathered ferralsol near Santar'em, stomata are relatively insensitive to changes in leaf-to-air water vapor mole fraction deficit, suggesting a much less tight control of transpiration rates by stomata than is the case for the cerrado. Carswell et al. [2002] also observed that canopy conductances for a semievergreen forest growing on a deeply weathered ferralsol at Caxiuana were actually higher in the dly season than the wet season. This was despite higher D during the dly season. By contrast, working in the Rebio Jam semievergreen forest site in southwest Amazonia, strong stomatal responses to D have been observed at both the individual leaf [McWilliam et ai" 1996] and whole canopy level [Grace et al., 1998]. McWilliam et al. [1996] also observed that leaf water potentials tended to be similar or even less negative during the dly season than for the wet season at this site, a result also repOlied for late-stage canopy trees growing in French Guiana [Bonal et al., 2000a]. Examining the relationship between stand level latent heat fluxes (AE) and net radiation (R n ), Hasler and Avissar [2007] also noted that, as would be expected from the above, propOliionally greater reductions in the AEIRn occurred for the Rebio Jaru site compared to the Santarem and Caxiuana sites, attributing this to shallower rooting depths at the fOlIDer. Although this is conceivably also the case for the French Guiana site mentioned above [Bonal et aI., 2000b], water extraction during the dry season celiainly occurs at depths beyond 2.4 m for this site [Bonal et ai" 2008] and below at least 3.4 m for Rebio Jaru [Negron Juarez et ai" 2007]. In conclusion, where soils are old and weathered, they are almost invariably also deep [Quesada et ai" 2009c], and this means that for much of Amazonia, both forest and savanna species have access to water at depths greater than 3.0 m. This allows for continual woody plant functioning through the dly season, but for cerrado, where this dly season is of a longer duration than for semievergreen or evergreen forest, significant reductions in leaf area also occur. For both forest and savanna, there is good evidence for hydraulic lift occurring, at least for deciduous and brevideciduous species, this facilitating continued functioning of surface roots throughout the dly season. It would also be interesting to see if other characteristics, for example, the presence of xylem pits, which are geneti-
cally associated and thought to influence the ease of transport of water [Jansen et al., 2004], are also characteristic of the cerrado and dry deciduous forest species found in the drier environments, as there is now good evidence that species-dependent differences in drought sensitivity are important in shaping species distributions along rainfall gradients [Engelbrecht et al., 2007]. UnfOliunately, little attention seems to have been paid to the water relations and physiological functioning of cerrado grasses, but it is clear from studies on other continents that considerable reductions in stomatal conductance occur during the dly season as a consequence of dramatic reductions in soil water availability and much higher leaf-to-air vapor pressure differences [e.g" Mantlana et al., 2008b]. 3.6. Integration ofPhysiological Characteristics and Fundamental Trade-Offs It is now well established that plant physiological characteristics do not vary independently of each other. But rather, they tend to covary in a systematic manner in accordance with what is often considered a multivariate ecological "strategy" dimension [Westoby et al., 2002]. Although the underlying source of such variation is often considered to be solely genetic, arising as difference in phylogenetic background [e.g., Wright et al., 2006; Swendon and Enquist, 2007], it is also clear for tropical trees that even within a given species, systematic and significant variations occur according to growth conditions [Patino et al., 2009] and that the nature of the covariation between key plant physiological attributes varies according to the underlying cause of any contrasts in characteristics observed [Fyllas et ai" 2009]. Indeed, analyzing 1040 individual trees positioned in 63 plots across Amazonia, Fyllas et al. [2009] found some foliar characteristics, such as SLA, [C], [N], and [Mg], to be highly constrained by the taxonomic affiliation of tree species but others, such as [P], [K], [Ca] and (PC, to be more strongly influenced by site growing conditions. By removing the environmental contribution to trait variation, they found that intrinsic values of most trait pairs coordinate, although different species (characterized by different trait suites) were found at discrete locations along a common axis of coordination. Species that tend to occupy higher fertility soils, such as those that typically occur in the western pati of the Amazon Basin [Quesada et al., 2009b], are characterized by intrinsically higher SLA and have an intrinsically higher [N], [P], [K], [Mg], and Ol3 C than their lower fertility counterpalis, generally OCCUlTing on the more heavily weathered soils toward the east. Despite this consistency, different scaling patterns were observed between low and high fertility
LLOYD ET AL.
sites showing that, for Amazonian trees, trait interrelationships are substantially modified by growth environment. Soil fertility was found M'be the most important environmental influence, affectinlall leaf nutrient concentrations and Ol3 C composition angAhcreasing SLA. One of the longest established conelations is between photosynthetic capacity and stomatal conductance [Wong et aI., 1979], and hei'e Amazonian forest trees are no exception [McWilliam et aI., 1996; Carswell et al., 2000; Domingues et al., 2005, 2007]. Nevertheless, as outlined in section 3.3, what is still unclear though is the extent to which nitrogen, as opposed to phosphorus, constitutes the key limiting nutrient for photosynthetic activity. As for other plant species [Wright et aI., 2004], SLA of tropical trees tends to scale positively with photosynthetic capacity (dlY weight basis) and with dly weight-based foliar nitrogen phosphorus concentrations when measured [Prado and De Moraes, 1997; Franco et al., 2005; Domingues et aI., 2005], although Hoffinann et al. [2005b] showed that even when controlling for phylogenetic differences and going in similar environments, savanna trees tended to have lower SLA and higher NIP ratios than their rainforest counterpalis, the opposite result to that seen when comparing across biomes (section 3.3). As is discussed in section 5, this suggests that nitrogen may be intrinsically more limiting in savanna than in rainforest environments. It might also be expected that the lower SLA typical of evergreen savanna trees (section 3.2) would be associated with a longer leaf life time, on average, than is the case for evergreen tropical forest trees, although this has not yet been established. Nevertheless, as expected from theOly [Givnish, 2002], it is now well documented that deciduous savanna trees tend to have higher SLA and higher photosynthetic rates and/or nitrogen contents when expressed on a dly weight basis [Prado and De Moraes, 1997; Franco et al., 2005], and such a distinction presumably also exists for tropical forest trees, both within and across forest types. Although deciduous leaves tend to have higher photosynthetic rates on a dly weight basis than do evergreen leaves, photosynthetic rates on an area basis are typically lower [Prado and De Moraes, 1997; Prior et ai" 2003; Franco et at., 2005] and, because of their shorter lifetime, generally result in a lower return on their carbon and nutrient investment than do evergreen leaves [Chabot and Hicks, 1982; Givnish, 2002l Thus, the dominance of evergreen trees in the cerrado vegetation with relatively long-lived schlerophyllous type, leaves can best be interpreted as a possible adaptation to the low soil nutrient status [Franco, 2002] with, as discussed in section 3.5, deciduous cerrado trees saving water through dry season leaf losses and with a greater emphasis on high rates of nutrient uptake through the presence of dimorphic
471
root systems with the integrity of nutrient-acquiring surface roots maintained throughout the dry season, As well as there being an intercorrelation between the various leaf traits as discussed above, correlations with whole plant hydraulic parameters can also be expected to exist. For example, deciduous trees in both forest and savanna typically have higher specific hydraulic conductivities (conductance to water flow per unit stem cross-sectional area, K s) than do their evergreen counterparts [Sobrado, 1993; Choat et aI., 2005], and these should be associated with lower wood densities, D w [Hacke et at., 2001] and an increased susceptibility to xylem cavitation [Sobrado, 1997]. Leaf specific conductivities, KL , such as those that can also be estimated fi'ommeasurements of transpiration rate and leaf! soil water potentials [Mencuccini, 2003] can be expressed as KL = KsAdA s, where AdA s represents the leaf area per unit twig cross-sectional area (the inverse of the so called "Huber value"), and working for a range of trees from a semievergreen forest in Panama, Santiago et at. [2004] found an excellent correlation between KL and leaf level photosynthetic rates, but no correlation between photosynthetic rate and leaf nitrogen concentrations. UnfOliunately, they did not test for leaf phosphorus concentrations, as was also the case for the study ofMeinzer et at, [2008] who, also working in Panama, fuliher showed tha~, despite K L and photosynthetic capacity per unit leaf nitrogen both declining with increasing branch wood density, AdAs and leaf nitrogen also simultaneously increased. Also, tl'tis increase in ALIAs and nitrogen was not sufficient to offset the costs of producing denser wood, also with the increased AdA s and presumably higher photosynthetic capacity considered to exacerbate the negative effect of increasing wood density on branch hydraulic and leaf water stahlS. Meinzer et at. [2008] did, however, find that, across the range of species examined, SLA declined sharply with increasing wood density. As discussed above, this would be expected to result in a greater longevity for leaves of high wood density species, a factor which also needs to be taken into account in the calculation of any carbon-related trade-offs. Although it is generally theorized that a low wood density and a high SLA with associated higher nutrient contents of a dry weight basis should all be generally associated with higher rates of tropical tree growth [Wright et at" 2006; Poorter et at., 2008], strong negative correlation between these two traits, as reported by Meinzer et at. [2008], for Panamanian forest species, and also by Bucci et at. [2004], for cerrado trees, has not always been observed [Wright et at., 2006], This is perhaps a consequence of differences between studies in the methods of wood density determination, with studies across different sites also being complicated by the fact that wood density seems to be a more "plastic" trait
472
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
than may have been appreciated up until now [Patino et aI., 2009]. Moreover, when comparing across sites, it is important to note that the genetically dependent trait interrelationships betweei'i SLA and leaf nutrient concentrations may be substantially modified in accordance with variations in soil fertility [Fyllas et al., 2009]. As detailed in section 3.1, one would also expect increased allocation belowground in response to both increased soil water deficits and low soil nutrient status. Working with long-term nitrogen and phosphorus fertilization in the cerrado, Bucci et al. [2006] observed a decline in midday leaf water potential for woody species. These authors argued that those species apparently have the capacity to exploit changes in nutrient availability by allocating resources to maximize carbon gain and enhance growth, with cost of increased allocation to leaf area relative to water transport capacity considered to have resulted in a higher total water loss per plant and a decrease in minimum leaf water potentials. 4. PHYSIOLOGY AT THE STAND LEVEL SCALE 4.1. Geographical Variations
The pioneering work on Amazonian soils of Sombroek [1966] in Brazil, Cochrane [1973] in Bolivia, and Sanchez and Buol [1974] in Peru was soon followed by large scale studies such as RADAMBRASIL, which when integrated together revealed considerable variations in the soil types of Amazonia [Sombroek, 1984, 2000; Cochrane et al., 1985], a recent summary of which is provided by Quesada et al. [2009b]. In ShOli, these studies reveal a large-scale gradient in soil feliility running broadly from nOliheast to southwest, with soils becoming considerably more fertile as one approaches the Andes. As has also been pointed out by Quesada et al. [2009c], it is not only soil fertility that changes, but also soil physical conditions, with the prevalence of shallower soils with more potential physical constraints on plant productivity also increasing toward the Andes. Superimposed upon this gradient in soil physical characteristics is a second large-scale gradient in rainfall, which increases more or less from the southeast to the northwest [Malhi and Wright, 2004]. Given that these two large gradients are key drivers of plant physiological processes vmying more or less orthogonally, it is not surprising that largescale variations in plant physiological processes occur, with broad-scale changes in stand level wood density, tree dynamics, and aboveground growth rates all associated with soil fertility variations [Baker et al., 2004a, 2004b; Malhi et al., 2004; Phillips et al., 2004; Quesada et al., 2009a]. The variations in both soil and stand characteristics are also associated with variations in plot level nutrient status with fo-
liar concentrations of phosphorus, in pmiicular, being much higher in the faster growing forests of western Amazonia [Fyllas et al., 2009]. Nevertheless, direct causal linkages remain to be firmly established. For example, as has already been discussed by Malhi et al. [2004], effects of soil fertility on aboveground forest net primary productivity (NPP) could be due to nutrient effect differences in allocation above- and belowground or, altematively, to higher rates of ecosystem photosynthesis (often referred to as gross primmy productivity (GPP)) associated with the higher soil fertilities in westem Amazonia. We currently lack the necessary ecosystem and leaf level physiological measurements to allow these different possibilities to be considered. However, examining detailed measurements of above- and belowground productivities for 10 Amazonian forests across a range of different soil types, Aragiio et al. [2009] found no differences between the fraction of NPP allocated above- versus belowground. This suggests that a higher GPP, perhaps associated with the higher foliar phosphorus for western Amazonian forests discussed above [Fyllas et aI., 2009], may be the main driver for the geographical differences in wood productivity. This is also suggested by the strong'relationship between wood productivity and the appropriate measures of soil available phosphorus [Quesada et al., 2009a]. 4.2. Seasonal Patterns
Several dendrometer data sets oftree stem diameter change at monthly resolution have been published for Amazonian forest. Most of these records indicate that stem increment is comparatively large in the wet season and small in the dly season [Vieira et al., 2004; Rice et al., 2004; Goulden et al., 2004]. This pattern might be attributable to a reduction in wood production with drought stress or even to changes in stem water content, though a more detailed examination indicates increased diameter increment at the end of the dly season that often precedes the onset of heavy rain. Consequently, the seasonality of wood growth may be associated with factors such as the production of new xylem associated with leaf flushing, rather than the direct effects of drought. Such observations also imply that the seasonal patterns of tropical forest-atmosphere exchange do not solely reflect the direct effect of the physical environment on physiology and that many tropical trees follow genetically programmed phenological patterns [van Schaik et al., 1993; Goulden et al., 2004]. In situ observations of the seasonality of root production are extraordinarily difficult measurements given the depth of root penetration in tropical forest [Nepstad et al., 1994; Bruno et al., 2006], although first advances are now being made in this respect, at least for surface roots. For forests
LLOYD ET AL.
growing on reasonably high clay content soils, Silver et al. [2005] and Jimenez (Jl al. [2009] found maximum rates of fine root growth dUl'ing the wet season, although Jimenez et al. [2009] obsf-ved the opposite pattern for a forest in the relatively high precipitation Colombian Amazon region growing on a podzol soil. They attributed the lack of fine root growth observed during the wet season for this forest to waterlogging, a consequence of the high rainfall and an impermeable ortsteinic horizon (a horizon consisting of cemented sesquioxides and organic matter; see fUSS Worldng Group WRB [2006]) in this case located at about l.l-m soil depth [Quesada et al., 2009b]. Several micrometeorological records of CO 2 and water vapor exchange were collected during LBA, building on the earlier work of Grace et al. [1995] in Rondonia, Malhi et al. [1998] near Manaus, and the ABRACOS project [Gash et al., 1996]. As is discussed in detail elsewhere [da Rocha et al., this volume; Saleska et al., this volume], seasonal patterns of daytime gross CO 2 uptake (canopy photosynthesis) and canopy conductance to water vapor vmy fi'om forest to forest. Some researchers have repOlied that canopy photosynthesis and canopy conductance decrease in the dly season [Malhi et al., 1998,2002; Vourlitis et al., 2004], while other researchers have reported that canopy photosynthesis remains nearly constant year round [Carswell et al., 2002; Araujo et al., 2002; Saleska et al., 2003] or increases moderately toward the end of the dly season [Goulden et al., 2004; da Rocha et al., 2004]. In a broad sense, these results are consistent with the leaf level obsel-vations mentioned in sections 3.3 and 3.4. That is to say, there seems little evidence of direct water stress effects on either photosynthetic capacity or stomatal conductances for such forests. Nevertheless, differences exist, which probably reflect contrasts between sites or years, though the controls on tropical forest seasonality and physiological activity remain poorly understood. Progress toward a more mechanistic understanding of seasonality requires that researchers (1) draw a clear distinction between seasonal changes that are genetically programmed and those that are a direct effect of limitations imposed by the physical environment on physiology, and between the proximate and ultimate causes of seasonal activity [van Schaik et al., 1993]; (2) recognize that the controls on plant activity obsel-ved at interannual time scales may not apply to seasonal time scales; and (3) recognize the diversity of tropical forest and the possibility that the controls on tropical forest seasonality may differ from forest to forest, or year. to year. As is also considered in more detail by Phillips et al. [2009], there is, however, evidence that extreme drought directly impacts tropical forest production. Nepstad et al. [2002] reported that prolonged rainfall exclusion from an evergreen
473
forest decreased primary production and increased large tree mortality. Satellite observations combined with the CASA model suggest a decrease in NPP during dry El Nino periods [Potter et al., 2001]. It appears that many tropical forests growing on the deeper soils avoid drought stress during average rainfall years and that seasonal patterns observed largely reflect genetically programmed phenological patterns. On the other hand, it appears likely that more severe dry periods deleteriously impact tropical forest NPP and GPP by exerting a direct effect on physiology. The challenge for researchers is to quantitatively model and predict the point at which a drought becomes so severe that it results in plant mortality [Phillips et al., 2009] with a potential lasting impact on landatmosphere exchange and vegetation structure. In contrast to semievergreen and evergreen forests, transitional (semideciduous) forest and cerrado vegetation show marked seasonalities in both LAI and surface fluxes [Miranda et al., 1997; Santos et al., 2004; Vourlitis et al., 2004], these also being associated with large changes in soil water status [Quesada et aI., 2004, 2008] and consistent with the leaf level results reported in sections 3.2, 3.3, and 3.4. Also consistent with the idea that hydraulic lift can sel-ve to maintain the integrity and function of the surface roots of deciduous and brevideciduous savanna trees during the dly season, Quesada f(t al. [2008] found that, almost immediately upon the commencement of wet season rains, the main region of cerrado ecosystem level root water uptake shifted I fi'om depths greater than 2.0 m to the upper soil surface layers. 4.3. Insights From Isotopes
Stable isotopes can provide good integrative measurements of variations in nutrient, water cycling, and assimilation capacity in telTestrial ecosystems, for example, allowing a determination of the relative contributions of C3 trees versus C4 grasses to the productivity of savanna ecosystems [Lloyd et al., 2008]. According to Ometto et al. [2006], the isotope ratio data for three distinct regions in the forested region of Amazonia are consistent with the CUlTent understanding of the roles oflight, water availability, recycling of soil-respired CO 2 and also consistent with the understanding that an open nitrogen cycle can lead to high o15N values, despite a significant number of legumes in the vegetation, and as already mentioned (section 3.3), the relatively negative o15N of plants and soil for white sand forests in Amazonia provides some strong indications that these forests may be nitrogen rather than phosphorus limited. Although Nardoto [2005] observed an inverse cOlTelation of the 015N with the length ofthe dly season for nonwhite sand forests, suggesting more open N cycling as one goes to more humid regions of
474
LLOYD ET AL.
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
Amazonia, an alternative hypothesis accounting for variations in (515N across the full spectrum of Amazonian forests was put forward by Quesada et al. [2009c). They argued that as soils age o"il geological time scales, phosphorus becomes progressively more limiting and nitrogen progressively more in excess (increasing soil and plant (5 15N) and with the structural and physiological characteristics of the vegetation also changing, there being a tendency toward species with slower potential growth rates, lower intrinsic SLA and nutrient requirements, and higher levels of structural defenses on older more heavily weathered soils (see also section 3.6). During this phase of ecosystem development, leaf and soil 015N progressively increases, this reflecting continual losses of nitrogen from the ecosystem. Nevertheless, eventually a point is reached where nitrogen, rather than phosphorus, becomes limiting for plant productivity. This is hypothesized to arise primarily as a consequence of changes in leaf litter chemistry with high levels of tannins and lignin in leaves of slow growing species inhibiting enzymes involved in nitrogen mineralization, though with several other factors also involved. In support of their theory, Quesada et al. [2009c] noted that not only are unusually negative 015N observed on the arenosols and podzols, but that they also occur for forests on the most heavily weathered ferralsol and acrisol soil types. Bustamante et al. [2004] found a wide range of variations in cerrado tree foliar 015N, which they related to differences between tree species and individuals in nitrogen uptake characteristics, precipitation seasonality, fire frequency, also observing that in contrast to the forest, many cerrado Fabaceae species seem to be actively fixing atmospheric N 2. Nevertheless, even for non-N2-fixers, 015N was, on average, significantly less than is observed for Amazonian forest species growing on similar soils [Nardoto, 2005; Nardoto et al., 2008). This supports the general view that cerrado productivity may be nitrogen rather than phosphorus limited [Bustamante et aI., 2006]. Water cycling in plant and ecosystems can also be approached using the stable isotope signature for the oxygen on the water molecule enclosed in these compartments and fluxes within. Distinct pattern in anatomical structure of various plant species in Amazonia results in complex pathways of water flow within a leaf causing large differences in the diel fluctuation of the oxygen isotope ratios (0 18 0) in the leaf water [Lai et al., 2008]. The implications of these findings for regional water and carbon balance are related to the isotopic signal of the ecosystem-respired CO 2 [Ometto et al., 2005] and to the terrestrial contribution to the seasonal fluctuations in the 0 18 0 of atmospheric CO 2 [Friedli et al., 1987). According to Lcd et al. [2008], the leaf water turnover calculated for nighttime was consistently 2-1 0 times greater
than those during daytime. This prolonged turnover time is the reason for the considerable nonsteady state effect on the leaf water 018 0 enrichment at night [Cel'l1usak et aI., 2002). Lower isotopic 0 180 on leaf water compared to stem (soil) water in understory plants suggests exchange of leaf water with vapor water [Lcd et al., 2008], which might be an important contributor to the water balance in environment under high humidity and plants with open stomata (J. Beny, personal connnunication, 2007). Water uptake by leaves was also observed in a controlled dry-out experiment in Brazilian Amazonia by Cardinot [2007]. Furthermore, Doughty et al. [2006] have recently proposed that the photosynthesis gas exchange in 65% of studied species in Brazilian Amazonia function under circadian rhythms once they have closed stomata and photosynthetic rates during the night, even under continuous and constant light, and resuming the active gas exchange during the normally light period. The importance of lianas for the ecology of the tropical forest has been pointed out elsewhere. Usually ignored in forest inventories, lianas exert an important ecological effect in the forest representing, for instance, less than 5% of the forest biomass, but 40% of lea{productivity [Phillips et al., 2002). According to Ometto et al. [2006], the lianas tend to show higher o13 C values and lower ratios of intercellular to ambient CO 2 concentration, suggesting that this group is more conservative in its water use, in agreement with relatively lower stomatal conductance when compared to upper canopy tree species [Domingues et aI., 2007]. Lianas tend to have the highest leaf water turnover time due to their relatively smaller conductances. Grasses and shrubs in the pasture have significantly lower turnover time at night compared to the overstOly trees and lianas in the forest, which contributes to the relatively smaller nonsteady state leafwatel' enrichment [Lai et al., 2008). 5. CONCLUDING COMMENTS AND SYNTHESIS As outlined in the introduction, one main aim of this review was to investigate the extent to which contrasts in the physiology of the various vegetation types found across Amazonia were correlated with broad-scale patterns of their distributions. We have been partially successful in this respect. For example, it has been concluded that semievergreen forests can persist despite an extended dry season through the ability to develop roots and transport water from considerable soil depths (section 3.4). Likewise, from Figure 1, we can also reasonably conclude that this is facilitated by the presence of clay-rich, yet well-drained, subsoils in these areas, also with relatively high water-holding capacities [Quesada et aI., 2009a, 2009c] effectively allowing high amounts of wet season rainfall to be stored in the soil profile
and utilized during th~ dry season. The importance of this should not be under5,stimated. For example, although generally more fertilep,'lnany soils of western Amazonia have physical restrictiqns present at depths shallower than 2 m [Quesada et al.;2009c]. Were such soils to exist in eastern or southern Amazonia, where, even in forested areas, the rainfall is generally much less than in the western part of the basin [Malhi and Wright, 2004], it is highly unlikely that semievergreen forest could exist. Similarly, as is summarized in section 3.5, the strong presence of evergreen trees in the cerrado is most likely also a consequence of their ability to extract water from considerable depths during the dry season. This can occur because, although most cerrado soils are old, strongly weathered and thus infertile, as a consequence of this extreme weathering, they are also velY deep and with good water-holding characteristics. In section 3.5, it was also noted that deciduous and brevideciduous trees also coexist with evergreen trees; this coexistence being possible through their employment of an alternative ecophysiological strategy, namely, a reduction in water requirements through having a leafless period in the dly season. This allows a greater allocation of resources toward the presence of surface roots, which, in turn, allows for higher rates of nutrient uptake to occur: Such high nutrient uptake rates being required to support the deciduous habit. In similarly low rainfall regions where soils are more fertile, but also more shallow, and with a lower water-holding capability, the evergreen habit would not be favored, this probably being the main factor defining the distribution of drought-deciduous forests as the dominant vegetation type in lower rainfall regions, where more fertile soils occur (sections 2.2 and 2.3). Other ecophysiological characteristics related to different growth environments across Amazonia have also been noted. For example, the species which occupy the more fertile forests of western Amazonia have intrinsically higher requirements for nutrients, higher SLA, and lower wood densities than their slower growing counterparts in eastern Amazonia (sections 3.3 and 3.6). Moreover, foliar nutrient concentrations, especially nitrogen, seem to be lower for cenado trees than for (semi) evergreen forest trees in eastern Amazonia (section 3.3), with this also being associated with variations in other leaf properties such as SLA and leaf longevity. Indeed, as is summarized in section 4.3, evidence from 015N studies strongly suggests, in contrast to most Amazonian forests, that the productivity of cerrado h'ees and grasses is. nitrogen limited. How can such a stark contrast between two ecosystems occur? Overall, there does not seem to be any great difference between eastern Amazonian cerrado and forest soils, both often being rather infertile but with good water-holding
475
characteristics [Motta et al., 2002; Quesada et al., 2009b]. Nevertheless, asis outlined by Bustamante et a1. [2006] and Nardoto et al. [2006], slow rates of nitrogen mineralization during the dry season and significant fire-associated losses of nitrogen could both be contributing factors to the apparent low nitrogen availability of cenado as opposed to forest ecosystems in Amazonia. Yet we suspect that is not the whole story. It seems quite likely that the physiological characteristics of cerrado trees and grasses also contribute to maintaining a low nitrogen status ecosystem through mechanisms similar to that outlined as giving rise to eventual nitrogen limitation for the oldest forest ecosystems in section 4.3. That is to say, associated with the schlerophyllous characteristics of the leaves of (in particular, evergreen) cerrado trees, namely, a low SLA and low nutrient contents (section 3.3), are also relatively high levels of lignin and phenols [Varanda et aI., 1997], which should themselves serve to impose significant constraints on nih'ogen mineralization. As discussed by Quesada et al. [2009c], this would occur through nitrogen being incorporated into the lignin fi'action of the litter during humification and through the direct inhibition of enzymes involved innitrogen mineralization by high levels of tannins, lignin, and associated compounds. This, along with factors such as fire [Miranda et al., 2Q02] probably serves to make the cenado ecosystem, to some extent, self-sustaining, and this provides one explanation asI to why the last expansion of cerrado into previously forested areas during the brief dly episode of the Holocene has not been reversed, with Amazonian forests apparently failing to regain their original area, even though rainfall has apparently returned to previous levels over the last few thousand years [Ledru, 1993]. These general ideas are illustrated in Plate 1, when it is suggested that the vegetation changes associated with any spatial or temporal variation in precipitation should depend fundamentally on soil characteristics, as well as the precipitation regime itself. For the forest areas in western Amazonia, consisting ofrelatively fertile but often shallow soils, it is suggested that, at reduced precipitation regimes, high nutrient-requiring drought-deciduous forest should prevail. On the other hand, for eastern Amazonia, where soils are often old and extremely weathered and infertile, they are also often deep and thus capable of storing considerable amounts of water. This favors evergreen cerrado-type species with low nutrient requirements but also capable of extracting water from depth and maintaining their leaves over the dly season. Also shown in Plate 1 is a feedback loop, representing effects of leaf schlerophyllous physiological properties and also potentially fire (as discussed in the paragraph above) in maintaining a savanna-type ecosystem through a tightening of the nitrogen cycle.
476
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
LLOYD ET AL.
c:
Fast growing (semi) evergreen forest
Slow growing (semi) evergreen forest
:+:i
High nutrient requirements
Low nutrient requirements
'i5..
Open nitrogen cycle
Open nitrogen cycle
0
.....CO
"[5
~
a.
!
1 •
Slower growing droughtdeciduous forest High nutrient requirements Open nitrogen cycle
!
0>
c:
'en CO (l.) , .... (,) (l.)
0
Slower growing savanna
......
Low nutrient requirements Closed nitrogen cycle
Of course, much more work has to be done to validate and/or develop upon Jllis general scheme, Nevertheless, understanding the i~Je;'actions between precipitation regime and soil chemicalll1ld physical characteristics in influencing Amazonian vegetation structure and function will be of considerable importance in understanding and predicting effects of climate change on Amazonia, especially if significant declines in precipitation occur as predicted by some Global Climate Models [see Marengo et a/, , this volume], For example, according to Malhi et a/, [2009], any future reduction in precipitation in eastern Amazonia, as a consequence of climate change, should give rise to drought-deciduous forest rather than celTado. We disagree: this is because much of the forest in the areas predicted to be most severely affected by drought are in the southeastern part of the basin [Salazar et al., 2007] where nutrient-poor but deep ferralsol and acrisol soils dominate [Quesada et a/, , 2009b]. According to our understanding, and as shown in Plate 1, we consider it much more likely that the semievergreen forests there would undergo a direct h'ansition to an evergreen-dominated cerrado-type vegetation. On the other hand, should southwestern Amazonia be most affected by future declines in precipitation as was the case for the 2005 Amazon drought [Phillips et a/, , 2009], then it is more likely that the fast-growing and dynamic forests situated there would h'ansfonn to drought-deciduous forest, the soils of this part of Amazonia typically being rather fertile, but also often of limited depth and thus also of a much lower water-holding capability [Quesada et a/, , 2009b]. Acknowledgments, We thank Augusto Franco for useful comments on an earlier version of the manuscript. This work has been supported through the UK Natural Environment Research Council "TROBIT" and "QUERCC" consortia projects,
REFERENCES
Increasing pedogenic age Plate 1. Proposed scheme for transitions between forest and savanna vegetation types as affected by soil age and precipitation, As is described in the text, the feedback loop for savanna (cerrado) vegetation represents a closure of the whole ecosystem nitrogen cycle initiated through the physiological characteristics of woody and herbaceous savanna species as well as fire, (top left) Evergreen forest in southwest Amazonia (Porongaba, Brazil), (top right) Semievergreen forest in eastern Amazonia (Caixuana, Brazil), (bottom left) Drought-decidious (Chiquitano) forest in southern Amazonia (Tucavaca, Bolivia), (bottom right) Cerrado in southern Amazonia (Los Fierros, Bolivia), Photos credits: T, Baker (Brazil) and J, Lloyd (Bolivia),
Abdala, G. C., L. S, Caldas, M. Haridasan, and G. Eiten (1998), Above and be10wground organic matter and rootshoot ratio in cerrado in central Brazil, Braz, J. Ecol, , 2,11-23. Alexander, 1. 1., and S, S. Lee (2005), Mycorrhizas and ecosystem processes in tropical rain forest Implications for diversity, in Biotic Interactions in the Tropics, edited by D. Burs1em, M. Pinard, and S. Hartley, pp. 165-203, Cambridge Univ. Press, Catnbridge, u. K. Alvim, P. de T, and W. A Araujo (1952), El sue10 como factor eco10gico en el desarollo de la vegetaci6n en el centro-oeste dei Brasil, Turrialba, 2,153-160, Anderson, A B, (1981), White-sand vegetation of Brazilian Amazonia, Biotropica, 13,199-210, Andreae Lima, D, (1959), Viagem aos campos de Monte Alegre, Para, Bol. Tecn, Inst, Agron, Norte, 36, 99-149.
477
Anten, P, R., M, 1. A Weger, and E. Medina (1998), Nitrogen distribution and leafarea indices in relations to photosynthetic nitrogen use efficiency in savanna grasses, Plant Eco!., 13, 63-75, Aragao, L. E" et a1. (2009), Above- and below-ground net primary productivity across ten Amazonian forests on contrasting soils, Biogeosci. Discuss" 6, 2441-2488. Aralljo, A c., et a1. (2002), Comparative measurements of carbon dioxide fluxes fi'om two nearby towers in a central Amazonian rainforest: The Manaus LBA site, J. Geophys, Res., 107(D20), 8090, doi: 10.1 029120011D000676, Askew, G, P., D. 1. Moffatt, R. F, Montgomery, and P. L. Searl (1970), Interrelationships of soils and vegetation in the savannaforest boundary zone of north-eastern Mato Grosso, Geogr, J., 136,370-376. Asner, G. P., 1. M. O. Scurlock, and 1. A Hicke (2003), Global synthesis of leaf area index observations: Implications for ecological and remote sensing studies, Global Ecol. Biogeogr" 12, 191-205. Asner, G. P" D. Nepstad, G. Cardinot, and D, Ray (2004), Drought stress and carbon uptake in an Amazon forest measured with spaceborne imaging spectroscopy, Proc, Natl. Acad Sci, U. S. A., 101, 6039-6044, Baker, T" et a1. (2004a), Increasing biomass in Amazonian forest plots, Philos. Trans. R. Soc. London, Ser, B, 359, 353-365. Baker, T., et a1. (2004b), Variation in wood density determines spatial patterns in Amazonian forest biomass, Global Change BioI" 10,545-562, Beard, 1. S. (1953), The savamla vegetation of northern tropical America, Ecol. MrJl1ogr., 23,149-215, Bonal, D., T. S. Barigah, A Granier, and 1-M, Gueh1 (2000a), Late-stage canopx tree species with extremely low Ol3 C and high stomatal sensitivity to seasonal drought in the tropical rainforest of French Guiana, Plant Cell Environ" 23,445-459. Bona1, D., C. Atger, T. S. Barigah, A Ferhi, J.-M. Guehl, and B. Ferry (2000b), Water acquisition patterns of two wet tropical canopy trees of French Guiana as inferred from H 2 18 0 profiles, Ann. For. Sd., 57, 717-724., Bonal, D" et a1. (2008), Impact of severe dry season on net ecosystem exchange in the Neotropica1 rainforest of French Guiana, Global Change BioI" 14,1917-1933. Bond, W, 1, F.r. Woodward, and G. F, Midgley (2005), The global distribution of ecosystems in a world without fire, New Phytol" 165,525-538, Brookes, P. c., D. S. Powlson, and D, S, Jenkinson (1984), Phosphorus in the soil microbial biomass, Soil BioI. Biochem" 16, 169-175, Brown, K. S. (1987), Soils and vegetation. I, in Biogeography and Quatel'l/m)1 Histol)1 in Tropical America, edited by T. C. Whitmore and G. T. Prance, pp. 19-45, Clarendon, Oxford, V.K. Bruno, R. D., H. R. da Rocha, H. C. de Freitas, M. L. Goulden, and S. D. Miller (2006), Soil moisture dynamics in an eastern Amazonian tropical forest, Hydrol. Processes, 20, 2477-2489. Bucci, S. J., F. G. Scholz, G. Goldestein, F. C. Meizer, 1 A. Hinojasa, W. A Hoffmann, and A C. Franco (2004), Processes preventing nocturnal equilibration between leaf and soil water
478
LLOYD ET AL.
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
potential in tropical savanna woody species, Tree Physiol., 24, 1119-1127. Bucci, S. J., G. Goldestein, F. C. Meizer, A. C. Franco, P. I. Campanello, anMF. G. Scholz (2005), Mechanisms contributing to seasonal homeostasis of minimum leaf water potential and predawn disequilibrium between soil and plant water potential in neotropical savanna trees, Trees, 19,296-304. Bucci, S. J.., F. G. Scholz, G. Goldestein, F. C. Meizer, A. C. Franco, P. I. Campanello, R. Villa Lobos-Vega, M. Bustamante, and F. Miralles-Wilhelm (2006), ~utrient availability constrains the hydraulic architecture and water relations of saVaIma trees, Plant Cell Environ., 29, 2153-2167. Buckley, T. N. (2005), The control of stomata by water balance, Nell' Phytol., 168,275-292. Bustamante, M. M. C., L. A. Martinelli, D. A. Silva, P. B. Camargo, C. A. Klink,T. F. Domingues, and R. V. Santos (2004), 15N natural abundance in woody plants and soils of central Brazilian savannas (cerrado), Ecol. Appl., 14, S200-S213. Bustamante, M. M. C., E. Medina, G. P. Asner, G. B. Nardoto, and D. C. Garcia-Montiel (2006), Nitrogen cycling in tropical and temperate savannas, Biogeochemistl)!, 79,209-237. Cairns, M. A., S. Brown, E. H. Helmer, and G. A. Baumgardner (1997), Root biomass allocation in the world's upland forests, Oecologia, 111,1-11. Campbell, C. D., and R. F. Sage (2006), Interactions between the effects of atmospheric CO2 content and P nutrition on photosynthesis in white lupin (Lupin us albus L.), Plant Cell Environ., 29, 844-853. Cardinot, G. K. (2007), Tolerancia it Seca de Especies Arb6reas de uma F10resta Tropical: Resultados de um Experimento em Larga Escala de Exclusao Artificial de Chuvas, Ph.D. thesis, 107 pp., Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil. Carswell, F. E., P. Meir, E. V. Wandelli, I. C. M. Bonates, B. Kruijt, E. M. Barbosa, A. D, Nobre, J. Grace, andP, G, Jarvis (2000), Photosynthetic capacity in a central Amazonian rain forest, Tree Physiol" 20, 179-186, Carswell, F, E" et a1. (2002), Seasonality in C02 and H20 fiux at an eastern Amazonian rain forest, J. Geophys. Res" 107(D20), 8076, doi: 10,1 029/2000JD000284, Castellanos, J" M. Maass, and J. Kummerow (1991), Root biomass of a dry deciduous tropical forest in Mexico, Plant Soil, 131, 225-228, Castro, E, A" and J, B. Kauffman (1998), Ecosystem structure in the Brazilian cerrado: A vegetation gradient of aboveground biomass, root biomass and consumption by fire, J. Trop. Ecol., 14, 263-283, Cavelier, J., T. M, Aide, C, Santos, A. M, Eusse, and J. M. Dupuy (1998), The savannization of moist forests in the Sierra Nevada de Santa Marta, Colombia,J. Biogeogr" 25, 901-912, Cernusak, L. A., J. S, Pate, and G, D. Farquhar (2002), Diurnal variation in the stable isotope composition of water and dry matter in fruiting Lupinus angus!(olius under field conditions, Plant Cell Environ" 25, 893-907, Chabot, B" and D, Hicks (1982), The ecology of leaf life spans, Annu, Rev, Ecol. Syst" 13, 229-259.
Chapuis-Lardy, L., M, Brossard, and H, Quiquampoix (2001), Assessing organic phosphorus status of cerrado oxisols (Brazil) using P-31-NMR spectroscopy and phosphomonoesterase activity measurement, Can, J. Soil Sci" 81, 591-691. Choat, B" M. C. Ball, J. G, Luly, and J. A. M, Holtum (2005), Hydraulic architect1ll'e of deciduous and evergreen dry rainforest tree species from north-eastern Australia, Trees, 19, 305311. Cochrane, T. T, (1973), El Potencial Agricola del uso de la Tierra en Bolivia. Un Mapa de Sistemas de Tierras, 826 pp" Ministry of Overseas Development, London, U. K" and Edit, Don Bosco, La Paz, Bolivia, Cochrane, T. T" L. G. Sanchez, L. G. de Azevedo, J. A, Porras, and C. L. Carver (1985), Land in Tropical America, vol!., 144 pp, Centro Internacional de Agricultura Tropical, Cali, Colombia, Coste, S., J, C. Roggy, P. Imbert, C, Born, D. Bonal, and E. Dreyer (2005), Leaf photosynthetic traits of 14 tropical rain forest species in relation to leaf nitrogen, Tree Physiol" 25, 1127-1137. Coutinho, L. M. (1990), Fire in the ecology of the Brazilian cerrado, in Fire in the Tropical Biota-Ecosystem Processes and Global Challenges, edited by J. G, Goldammer, Ecological Studies, vol. 8A, pp, 82-105, Springer, Berlin. Daly, D. C., and J. D. Mitchell (2000), Lowland vegetation of tropical South America-An overview, in Impel/Ext Balance: Landscape Transformations in the pre-Columbian Americas, edited by D. Lentz, pp. 391-454, Columbia Univ. Press, New York. da Rocha, H. R., L. Coutinho, L. M, L. Goulden, S ,D. Miller, M, C. Menton, L. D. V. 0, Pinto, H. C. de Freitas, and A. M, E. S. Figueira (2004), Seasonality of water and heatfiuxes over a tropical forest in eastern Amazonia, Ecol, Appl" 14, S22-S32, da Rocha, H. R" A. O. Manzi, and J. Shuttleworth (2009), Evapotranspiration, Geophys. Monogr, Ser" doi:10.1029/2008GM000744, this volume. Davidson, E. A., et a1. (2007), Recuperation of nitrogen cycling in Amazonian forests following agricultural abandonment, Nature, 447, 995-998, Dawson, T. E" S. S. 0, Burgess, K. P, Tu, R. S. Oliveira, L. S, Santiago, J. B, Fisher, K. A. Simonin, and A. R. Ambrose (2007), Nighttime transpiration in woody plants from contrasting ecosystems, Tree Physio!., 27, 561-575, Dezzeo N., N, E, Chacon, E, Sanoja, and G. Picon (2004), Changes in soil properties and vegetation characteristics along a forestsavanna gradient in southem Venezuela, For, Ecol. Manage., 200, 183-193, Domingues, T. F. (2005), Photosynthetic gas exchange among ecosystem compartments in forest and pasture ecosystems, Ph,D. thesis, Department of Biology, University of Utah, Domingues, T. F., J. A. Berry, L. A. Martinelli, J. P, Ometto, and J. R. Ehleringer (2005), Paramaterisation of canopy gas exchange and leaf-level gas exchange for an eastern Amazonian tropical rain forest (Tapaj6s National Forest, Para, Brazil), Earth Interact" 9(17), EIl49, doi: 10,1175/EIl49.1. Domingues, T, F" L. A. Martinelli, and J, R, Ehleringer (2007), Ecophysiological traits of plant functional groups in forest and
pasture systems from eastern Amazonia, Brazil, Plant Ecol" 191,103-112, Doughty, C. E" M. L/Goulden, S. D, Miller, and H, R. da Rocha (2006), Circadiali!rhytluns constrain leaf and canopy gas exchange in an Alpazonian forest, Geophys. Res, Lell" 33, Ll5404, doi: 10.1 029/2006GL026750. Egler, W. A, (1960), Contribuyoes ao conhecimento dos campos da Amazonia I. Os campos do Ariramba, Bol, lYIus. Paranese E. Goeldi ser 2, Bot" 43, 36 pp, Eiten, G, (1983), Class!fica~'Go da Vege/(u;:Go do Brasil, 305 pp., Conselho Nacional de Desennvolvimento Cientifico e Techo16gico, Brasilia, Brazil. Eiten, G. (1993), Vegetayao do celTado, in Cerrado: Caracterizat;:GO, Ocupat;:Go e Perspectivas, edited by M, N. Pinto, pp. 17-25, Editora Universidade de Brasilia, Brasilia, Brazil. Engelbrecht, B, M, J., L. S, Comita, R. Condit, T. A. Kursar, M. T, Tyree, B, L. Turner, and S. P. Hubbell (2007), Drought sensitivity shapes species distribution patterns in tropical forests, Nature, 447, 80-82. Eyre, S, R. (1963), Vegetation and Soils. A World Picture, 324 pp" Edward Arnold, London, Fisher, R. A., M, Williams, R. L. do Vale, A. L. da Costa, and P, Meir (2006), Evidence from Amazonian forests is consistent with isohydric control of leaf water potential, Plant Cell Environ,,29,151-165. Fittkau, E. J., and H. Klinge (1973), On biomass and tropic structure of the central Amazonian rain forest ecosystem, Biotropica, 5,2-14. Franco, A. C. (1998), Seasonal patterns of gas exchange, water relations and growth of Roupala Montana, an evergreen savanna species, Plant Ecol" 136,69-76. Franco, A. C, (2002), Ecophysiology of celTado woody plants, in The Cerrados ofBrazil: Ecology and Natural HistOl)' ofa Neotropical Savanna, edited by O. Paulo and R, Marquis, pp, 178197, Columbia Univ. Press, New York. Franco, A. C" M. Busamante, L. S, Caldas, G, Goldstein, F, C. Meinzer, A. R. Kozvotis, P. Rundel, and V, T. R. Cordain (2005), Leaf functional traits of neotropical savanna trees in relation to seasonal water deficit, Trees, 19, 326-335. Friedli, H., U, Siegenthaler, 0, Rauber and H, Oeschger (1987), Measurements ofconcentration, l3C/12C and 18 0/ 16 0 ratios oftropospheric carbon dioxide over Switzerland, Tellus, 39B, 80-88. Furley, P, A. (1992), Edaphic changes at the forest-savanna boundary with particular reference to the neotropics, in Nature and Dynamics ofForest-Savanna Boundaries, edited by P. A. Furley, J. Proctor, and J. A. Ratter, pp. 91-117, CRC Press, London. Furley, P. A" and J. A. Ratter (1988), Soil resources and plant communities of Central Brazilian cerrado and their development, J. Biogeogr" 15,97-108. Furley, P. A" J, A. Ratter, and D. R. Gifford (1988), Observations on the vegetation of Eastern Mato Grosso III. Woody vegetation and soils of Morro de Fumaya, Torixoreu, Brazil, Proc. R, Soc, London, Ser, B, 203,191-208. Fyllas, N. M" et a1. (2009), Basin-wide variations in foliar properties of Amazon forest trees: Phylogeny, soils and climate, Biogeosci. Discuss., 6, 3707-3769.
479
Gash, J. H, c., C. A Nobre, J. M. Roberts, and R. L. Victoria (1996), Amazonian Deforestation and Climate, 638 pp., John Wiley, Chichester, U. K. Gignoux, J., J. Clobert, and J.-c. Menaut (1997), Alternative fire resistance strategies in savanna trees, Oecologia, 110, 576-583, Givnish, T. J. (2002), Adaptive significance of evergreen vs, deciduous leaves: Solving the triple paradox, Silv, Fenn" 36, 703-743, Goldstein, G., F, Meinzer, S, J. Bucci, F. G, ScholZ, A. C. Franco, and W, A. Hot1lnaIm (2008), Water economy of Neotropical savanna trees: Six paradigms revisited, Tree Physiol" 28, 395404, Goodland, R, J. A., and R. Pollard (1973), The Brazilian cerrado vegetation: A fertility gradient, J. Ecol" 61, 219-224. Goulden, M, L., S, D, Miller, H. R. da Rocha, M, C. Menton, H. C. de Freitas, A. M. E. S. Figueira, and C. A. D. de Sousa (2004), Diel and seasonal patterns of tropical forest CO 2 exchange, Ecol. Appl" 14, S42-S54, Grace, J., et al. (1995), Carbon dioxide uptake by an undisturbed tropical rain forest in southwest Amazonia, 1992 to 1993, Science, 270, 778-780. Grace, J., J. Lloyd, A. C. Miranda, and J. H, Gash (1998), Fluxes of water vapour, carbon dioxide and energy over a C4 pasture in south-western Amazonia, Aust, J. Plant Physiol" 25, 519530, Hacke, U., G., J. S. Sperry, W. T. Pockman, S, D. Davis, and K. A. McCulloch (2001), Trends in wood density and structure are linked to preventi9n of xylem implosion by negative pressure, Oecologia, 126,457-461. Haridasan, M, (2000), Nutriyao mineral de plantas natives do cerrado,R. Bras, Fisz'ol, Veg., 12, 54-64, Hasler, N., and R. Avissar (2007), What controls evapotranspiration in the Amazon Basin?, J. Hydrometeoro!., 8, 380-395, Hodnett, M. G., L. Pimental da Silva, H. R, da Rocha, and R. CruzSena (1995), Seasonal soil water changes beneath central Amazonian rainforest and pasture, J. Hydro!., 170, 233-254, Hodnett, M, G" J. Tomasella, A, de O. Marques Filho, and M, D. Oyama (1996), Deep soil water uptake by forest and pasture in central Amazonia: Predictions from long-term daily rainfall data using a simple water balance model, in Amazonian Deforestation (md Climate, edited by J. H. C, Gash et aI., pp. 79-99, John Wiley, Chichester, U. K. Hoffmann, W, A., and A. G, Moreira (2002), The role of fire in population dynamics of woody plants, in The Cerrados ofBrazil: Ecology and Natural Histol)1 ofA Neotropical Savanna, edited by P, S, Oliveira and R. J. Marquis, pp. 159-177, Columbia Univ. Press, New York. Hoffman, W, A., B. Orthen, and A. C. Franco (2004), Constraints to seedling success of savamla and forest trees across the savanna-forest boundary, Oecologia, 140,252-260, Hoffman, W. A., E. R. da Silva Junior, G, C. Machado, S. J, Bucci, F. G. Sholz, G. Goldstein, and F. C. Meinzer (2005a), Seasonal leaf dynamics across a tree density gradient in a Brazilian savanna, Oecologia, 145, 306-315, Hoffman, W, A., A, C. Franco, M, Z. Moreira, and M. Haridasan (2005b), Specific area explains differences in leaf traits between congeneric savanna and forest trees, Funct. Ecol" 19, 932-940,
480
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETAnON
Huber, O. (2006), Herbaceous ecosystems on the Guyana Shield, a regional overview, J. Biogeogr., 33,464-475. Huete, A. R., K. Didan, Y. E. Shimabukuro, P. Ratana, S. R. Saleska, L. R<'!" Hutyra, W. Yang, R. R. Nemani, and R. Myneni (2006), Amazon rainforests green-up with sunlight in dry season, Geophys. Res. Lett., 33, L06405, doi:1O.1029/2005GL025583. Huytra, L. R., 1. W. Minger, C. A. Nobre, S. R. Saleska, S. A. Vieira, aqd S. C. Wofsy (2005), Climatic variability and vegetation vulnerability in Amazonia, Geophys. Res. Lett., 32, L24712, doi: 1O.1029/2005GL024981. \ IUSS (International Union of Soil Science) Working Group WRB (2006), World reference base for soil resources 2006: A framework for international classification, correlation and communication, World Soil Resources Report 103, FAO, Rome. Jackson, P. C., F. C Meinzer, M. Bustamante, G. Goldstein, A. Franco, P. Rundel, L. Caldas, E. Igler, and F. Causin (1999), Partitioning of soil water among tree species in a Brazilian Cerrado ecosystem, Tree Physiol., 19, 717-724. Jansen, S., P. Baas, P. Gasson, F. Lens, andE. Smets (2004), Variation in xylem stmcture from tropics to tundra. Evidence from vestured pits, Proc. Natl. Acad. Sci. U. S. A., 101, 8833-8837. Jimenez, E. M., F. H. Moreno, 1. Lloyd, M. C. Penuela, and S. Patino (2009), Fine root dynamics for forests on contrasting soils in the Colombian Amazon, Biogeosci. Discuss., 6, 3133-3453. Jipp, P. H., D. C. Nepstad, D. K. Cassel, and C. Reis de Carvalho (1998), Deep soil moisture storage and transpoltation in forests and pastures ofseasonally-dlY Amazonia, Clim. Change, 39, 395-412. Kattge, J., W. Knorr, T. Raddatz, and C. Wirth (2008), Quantifying photosynthetic capacity and its relationship to leaf nitrogen content for global-scale terrestrial biosphere models, Global Change BioI., 15, 976-991. Lai, C.-T., J. P. H. B. Ometto, L. A. Martinelli, J. A. Beny, T. F. Domingues, and 1. R. Ehleringer (2008), Life form-specific variations in leaf water oxygen-18 enrichment in Amazonian vegetation, Oecologia, 157, 197-210. Ledm, M. P. (1993), Late Quaternary environmental and climatic changes in central Brazil, Quat. Res., 39, 90-98. Lee,1. E., R. S. Oliveira, T. E. Dawson, and I. Fung (2005), Root functioning modifies seasonal climate, Proc. Natl. Acad. Sci. U. S. A., 102,17,576-17,581. Legendre, P., and L. Legendre (1998), Numerical Ecology, 853 pp., Elsevier, Amsterdam. Lenz, E., and C. Klink (2006), Comportamento fenol6gico de especies lenhosas em um cerrado sentido restrito de Brasilia, DF, Revista Brasi!. Bot., 29, 627-638. Lloyd, J., M. J. Bird, L. Yellen, A. C. Miranda, E. M. Veenendaal, G. Djagbletey, H. S. Miranda, G. Cook, and G. Farquhar (2008), The relative contributions of woody and herbaceous vegetation to tropical savanna ecosystem productivity: Field studies in Australia, Brazil and Ghana and a quasi-global estimate, Tree Physiol., 28, 451-468. Lopes, A. S., and F. R. Cox (1977a), A survey of the fertility status of surface soils under Cerrado vegetation in Brazil, Soil Sci. Soc. Alii. J., 41, 742-747. Lopes, A. S., and F. R. Cox (1977b), Cerrado vegetation in Brazil: An edaphic gradient, Agric. J., 69, 828-831.
Ludwig, F., T. E. Dawnson, H. H. T. Prins, F. Berendse, and F. Kroon (2004), Below ground competition between trees and grasses may ovelwhelm the facilitative effects of hydraulic lift, Ecol. Lett., 7,623-631. Malhi, Y., and J. Wright (2004), Spatial patterns and recent trends in the climate of tropical rainforest regions, Philos. Trcms. R. Soc. London, Ser. B, 359, 311-329. Malhi, Y., A. D. Nobre, J. Grace, B. Kruijt, M. G. P. Pereira, A. Culf, and S. Scott (1998), Carbon dioxide transfer over a Central Amazonian rain forest, J. Geophys. Res., 103,31,59331,612. Malhi, Y.,E. Pregararo, A. D. Nobre,M. G. P. Perieira,J. Grace, A. D. Culf, and R. Clement (2002), The energy and water dynamics of a central Amazonian rain forest, J. Geophys. Res., 107(D20), 8061, doi: 10.1 029/200 IJD000623. Malhi, Y., et al. (2004), The above-ground coarse wood productivity of 104 Neotropical forest plots, Global Change Bio!., 10, 563-591. Malhi, Y., et al. (2006), The regional variation of aboveground live biomass in old-growth Amazonian forests, Global Change BioI., 12, 1107-1138. Malhi, Y., L. E. O. C. Al'agao, D. Galbraith, C. Huntingford, R. Fisher, P. Zelazowski, S. Sitch, C. McSweeney, and P. Meir (2009), Exploring the likelihood' and mechanism of climatechange-induced dieback of the Amazon rainforest, Proc. Natl. Acad. Sci. U. S. A., doi:1O.1073/pnas.0804619106. Mantlana, K. B., A. Arneth, E. M. Veenendaal, P. Wohland, P. Wolski, O. Kolle, M. Wagner, and J. Lloyd (2008a), Inter-site and species specific differences in photosynthetic properties of C4 plants growing in an African savanna/wetland mosaic, J. Exp. Bot., 59, 3941-3952. Mantlana, K. B., A. Al'neth, E. M. Veenendaal, P. Wohland, P. Wolski, O. Kolle and 1. Lloyd (2008b), Seasonal and interannual photosynthetic response of representative C4 species to soil water content and leaf nitrogen content across a tropical seasonal floodplain, J. Trop. Eco!., 24, 201-213. Mardegan, S. F., G. B. Nardoto, N. Higuchi, M. Z. Moreira, and L. M. Martinelli (2008), Nitrogen availability patterns in whitesand vegetations of Central Brazilian Amazon, Trees, 23, 479488 doi: 10.1007/s00468-008-0293-9. Marengo, J., C. A. Nobre, R. Betts, P. Cox, G. Sampaio, and L. Salazar (2009), Global warming and climate change in Amazonia: Cllmate-vegetation feedback and impacts on water resources, Geophys. Monogr. Ser., doi:l0.1029/2008GM000743, this volume. Martinelli, L. A., M. C. Piccolo, A. R. Townsend, P. M. Vitousek, E. Cuevas, W. McDowell, G. P. Robertson, O. C. Santos, and K. Treseder (1999), Nitrogen stable isotopic composition ofleaves and soil: Tropical versus temperate forests, BiogeochemistlJI, 46, 45-65. McWilliam, A. L.-C., J. M. Roberts, O. M. R. Cabral, andM. V. 8. R. Leitao (1993), Leaf-area index and aboveground biomass of terra-finne rainforest and adjacent clearings in Amazonia, Funct. Ecol., 7,310-317. McWilliam, A. L. C., O. M. R. Cabral, B. M. Gomes, J. L. Esteves, and J. M. Roberts (1996), Forest and pasture leaf-gas exchange
LLOYD ET AL. in south-west Amazonia, in Amazonian Deforestation alld Climate, edited by 1. Gash et al., pp. 265-285, Jolm Wiley, Chichester, U. K. Medina, E., and H. /llinge (1982), Productivity of tropical forests and woodlands,jncycl. Plant Physio!., 12,281-303. Meinzer, F. C., G. Goldstein, A. C. Franco, M. Bustamante, E. Igler, P. Jackson, L, Caldas, and P. W. Rundel (1999), Atmospheric and hydraulic limitations on transpiration in Brazilian cerrado woody species, Funct. Eco!., 13,273-282. Meinzer, F. c., P. A. Campanello, J.-C. Domec, M. G. Gatti, G. Goldstein, R. Villalobos-Vega, and D. R. Woodmff (2008), Constraints on physiological function associated with branch architecture and wood density in tropical forest trees, Tree Physiol.,28, 1609-1617. Meir, P., B. Kmijt, M. Broadmeadow, E. Barbosa, O. Kull, F. Carswell, A. D. Nobre, and P. G. Jarvis (2002), Acclimation ofphotosynthetic capacity to irradiance in tree canopies in relation to leaf nitrogen and leaf mass per unit area, Plant Cell Environ., 25,343-357. Mencuccini, M. (2003), The ecological significance of longdistance water transport: Short term regulation, long term acclimation and the hydraulic costs of stature across plant life forms, Plant Cell Environ., 26,163-182. Metcalfe, D. B., M. Williams, L. E. O. C. Aragao, A. C. L. da Costa, S. S. de Almeida, P. Braga, P. H. L. Goncalves, J. de Athaydes Silva Junior, Y. Malhi, and P. Meir (2007), A method for extracting plant roots from soil facilities rapid sample processing without compromising measurement accuracy, Ne,v Phytol., 301, 279-288. Miranda, A. C., H. S. Miranda, 1. Lloyd, J. Grace, J. R. 1. Francey, J. A.Mcintyre, P. Meir, P. Riggan, R. Lockwood, and 1. Brass (1997), Fluxes of carbon, water and energy over Brazilian cerrado, an analysis using eddy covariance and stable isotopes, Plant Cell Environ., 20, 315-328. Miranda, E. J., G. L. Vourlitis, N. Priante Filho, P. C. Prianate, J. H. Campelo Jr., G. S. Suli, C. L. Fritzen, F. de A. Lobo, and S. Shiraiwa (2005), Seasonal variation in the leaf gas exchange of tropical forest trees in the rain forest-savanna transition of the southern Amazon Basin, J. Trop. Ecol., 21, 451-460. Miranda, H. S., M. M. C. Bustamante, and A. C. Miranda (2002), The fire factor, in The Cerrados ofBrazil: Ecology and Natural HistOlJI of a Neotropical Savanna, edited by O. Paulo and R. Marquis, pp. 51-68, Columbia Univ. Press, New York. Miranda, H. S., M. M. C. Bustamante, and A. C. Miranda (2002), The fire factor, in The Cerrados ofBrazil: Ecology and Natural HistOIJI ofa Neotropical Savanna, edited byP. S. Oliveira and R. J. Marquis, pp. 51-68, Columbia Univ. Press, New York. Miranda, I. S., M. L. Absy, and G. H. Rebelo (2003), Community stmcture of woody plants of Roraima savannahs, Brazil, Plant Ecol., 164,109-123. Mokany, K., R. J. Raison, and A. S. Prokushkin (2006), Critical analysis of root:shoot ratios in terrestrial ecosystems, Global Change BioI., 12, 84-96. Montgomery, R. F., and G. P. Askew (1983), Soils of tropical saValmas, in Ecosystems ofthe World, vol. 13, Tropical Savannas, edited by F. Bourliere, pp. 63-78, Elsevier, Amsterdam.
481
Moreira, A. G. (2000), Effects of fire protection on savanna structure in Central Brazil, J. Biogeogr., 27, 1021-1029. Moreira, M., F. Scholz, S. Bucci, S. L. Sternberg, G. Goldstein, F. Meinzer, and A. Franco (2003), Hydraulic lift in a neotropical savanna, Funct. Ecol., 17, 573-581. Motta, P. F., N. Curi, and D. P. Franzmeier (2002), Relation of soils and geomorphic surfaces in the Brazilian cerrado, in The Cerrados ofBrazil: Ecology and Natural HistolJI ofa Neotropical Savanna, edited by O. Paulo and R. Marquis, pp. 13-32, Columbia Univ. Press, New York. Murphy, P. G., and A. E. Lugo (1986), Structure and biomass of a subtropical d1y frest in Peurto Rico, Biotropica, 18, 89-96. Myers, J. G. (1936), Savalma and forest vegetation of the interior Guiana plateau, J. Ecol., 24,162-184. Myneni, R. B., et al. (2007), Large seasonal swings in leaf area of Amazon rainforests, Proc. Nat!' Acad. Sci. U. S. A., 104,48204823. Nardoto, G. B. (2005), Abundancia natural de 15N na Amazonia e Cerrado - implicacoes para a ciclagem de nitrogenio, Ph.D. thesis, ESALQIUSP, Piracicaba, Brazil. Nardoto, G. B., M. M. Da Cunha Bustamante, A. S. Pinto, and C. A. Klink (2006), Nutrient use efficiency at ecosystem and species level in savanna areas of Central Brazil and impacts of fire, J. Trop. Ecol., 22, 191-201. Nardoto, G. B., J. P. H. B. Ometto, J. R. Ehleringer, N. Higuchi, M. M. da C. Bustamante, and L. A. Martinelli (2008), Understanding the influelfces ofspatial patterns on N availability within the Brazilian Amazon forest, Ecosystems, 11, 1234-1246. Naves-Barbiero, C. C., A. C. Franco, S. J. Bucci, and G. Goldestein (2000), Fluxo de seiva e conduHincia estomMica de duas especies lenhosas sempre-yerdes no campo sujo e cerradao, Braz. J. Plant Physio!., 12,119-134. Negr6n Juarez, R. I., M. G. Hodnett, R. Fu, M. L. Goulden, and C. von Randow (2007), Control of dry season evapotranspiration over the Amazon forest as inferred from observations at a southern Amazon forest site, J. Clim., 20, 2827-2839. Nepstad, D. C., C. R. de Carvaiho, E. A. Davidson, P. H. Jipp, P. A. Lefebvre, G. H. Negreiros, E. D. da Silva, T. A. Stone, S. E. Trumbore, and S. Viera (1994), The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures, Nature, 372, 666-669. Nepstad, D. C., et al. (2002), The effects of partial throughfall exclusion on canopy processes, aboveground production, and biogeochemistry of an Amazon forest, J. Geophys. Res., 107(D20), 8085, doi: 10.1 029/200 IJD000360. New, M., M. Lister, M. Hulme, and I. Makin (2000), A highresolution data set of surface climate over global land areas, Clim. Res., 21,1-25. Nix, H. A. (1983), Climate of tropical savannas, in Ecosystems of the World, vol. 13, Tropical Savannas, edited by F. Bouliere, pp. 37-62, Elsevier, New York. Oliveira, R. S., T. E. Dawson, S. O. Burgess, and D. C. Nepstad (2005a), Hydraulic redistribution in thee Alnazonian trees, Oecologia, 145, 354-363. Oliveira, R. S., L. Bezerra, E. A. Davidson, F. Pinto, C. A. Klink, D. C. Nepstad, and A. Moreira (2005b), Deep root function in
482
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
soil water dynamics in cerrado savannas of central Brazil, Funet. Ecol., 19, 574-581. Oliveira-Filho, A, and 1. A. Ratter (2002), Vegetation physiognomies and wodtdy flora of the Cerrado biome, in The Cerrados oj' Brazil: Ecology and Nalural Hislol)' ofa Neolropical Savanna, edited by O. Paulo and R. Marquis, pp. 91-120, Columbia Univ. Press, New York. Ometto, 1. p. H. B., L. B. Flanagan, L. A Mmiinelli, and 1. R. Ehleringer (2005), Oxygen isotope ratios of waters and respired CO 2 in Amazonian forest and pa~ture ecosystems, Ecol. App!., 15,58-70. Ol11etto, 1. P. H. B., 1. R. Ehleringer, T. F. Domingues, 1. A Berry, F. Y. Ishida, E. Mazzi, N. Higuchi, L. B. Flanagan, G. B. Nardoto, and L. A. Martinelli (2006), The stable carbon and nitrogen isotopic composition of vegetation in tropical forests of the Amazon Basin, Brazil, Biogeochemisll)l, 79,251-274. Oyama, M. D., and C. A Nobre (2003), A new climate-vegetation equilibrium state for tropical South America, Geophys. Res. Lell., 30(23), 2199, doi:l0.l029/2003GLOI8600. Patino, S., et al. (2009), Branch xylem density variations across the Amazon Basin, Biogeosci. Discuss., 6, 545-568. Paulilo, M. T. S., and G. M. Felippe (1998), Growth ofthe shrub-tree flora ofthe Brazilian cerrados: A review, Trop. Eco!., 39, 165-174. Pearcy, R. W., and 1. Ehleringer (1984), Comparative ecophysiology ofC 3 and C4 plants, Planl Cell Environ., 7, 1-13. Phillips, O. L., et al. (2002), Increasing dominance of large lianas in Amazonian forests, Nalure, 418, 770-774. Phillips, O. L., et al. (2004), Pattern and process in Amazon forest dynamics, 1976--2001, Philos. Trans. R. Soc. London, Ser. B, 359,381--407. Phillips, O. L., et al. (2009), Drought sensitivity of the Amazon rainforest, Science, 323,1344-1347. Poorter, L., et al. (2008), Are functional traits good predictors of demographic rates? Evidence from five Neotropical forests, Ecology,98,1908-1920. Potter, C., S. Klooster, C. R. de Carvalho, V. B. Genovese, A Torregrosa,1. Dungan, M. Bobo, and 1. Coughlan (2001), Modelling seasonal and interannual variability in ecosystem carbon cycling for the Brazilian Amazon region, J. Geophys. Res., 106, 10,423-10,446. Prado, C. H. B. A, and J. A. P. V. De Moraes (1997), Photosynthetic capacity and specific leaf mass in twenty woody species of cerrado vegetation under field conditions, Photosynlhelica, 33, 103-112. Prado, D. E. (2000), Seasonally dly forests of tropical south America: From forgotten ecosystem to a new phytogeographic unit, Edinb. J. Bot., 57, 437--461. Prado, D. E., and P. E. Gibbs (1993), Patterns of species distributions in the dry seasonal forests of South America, Ann. Mo. Bal. Gard., 80(4), 902-927. Prior, L. D., D. Eamus, and D. M. 1. S. Bowman (2003), Leaf attributes in the seasonally dly tropics: A comparison of four habitats in northern Australia, Funct. Ecol., 17, 504-515. Prior,L. D., D. Eamus, andD. M. 1. S. Bowman (2004), Tree growth rates in north Australian savmma habitats: Seasonal patterns and correlations with leaf attributes, Allsi. J. Bot., 52, 304-314.
Quesada, c. A, A C. Miranda, M. G. Hodnett, A 1. B. Santos, H. S. Miranda, and L. M. Breyer (2004), Seasonal and depth variation of soil moistllre in a burned open savanna (campo sujo) in central Brazil, Ecol. Appl., 14, S33-S41. Quesada, C. A, M. G. Hodnett, L. M. Breyer, A 1. B. Santos, S. Andreae, H. S. Miranda, A C. Miranda, and 1. Lloyd (2008), Seasonal variations of soil moisture in two woodland savannas of central Brazil with different fire histOly, Tree Physiol., 28, 417--424. Quesada, c. A, et al. (2009a), Regional and large-scale patterns in Amazon forest structure and function are mediated by variations in soil physical and chemical properties, Biogeosci. Discuss., 6, 3993--4057. Quesada, C. A, 1. Lloyd, L. O. Anderson, N. M. Fyllas, M. Schwarz, and C. 1. Czimczik (2009b), Soils of Amazonia with particular reference to the RAINFOR sites, Biogeosci. Discuss., 6,3851-3921. Quesada, C. A., et al. (2009c), Chemical and physical properties of Amazonian forest soils in relation to their genesis, Biogeosci. Discuss., 6, 3923-3992. Ratter, 1. A, S. Bridgewater, and 1. F. Ribeiro (2003), Analysis of the floristic composition of the Brazilian cerrado vegetation III: Comparison of the woody vegetation of 376 areas, Edinb. J. Bot., 60, 57-109. Reatto, A, 1. R Correia, and S. T. Spera (1998), Solos de bioma cerrados: Aspectos pedologicos, in Cerrado: Ambiente e Flam, edited by S. M. Sano and S. P. Almeida, pp. 47-86, Empresa Bras. de Pesqui. Agropecu., Cent. de Pesqui. Agropecu. dos Cerrados, Brasilia. Reich, P. B., and J. Oleksyn (2004), Global patterns of plant leafN and P in relation to temperature and latitude, Proc. Nail. Acad. Sci. U. S. A., 101,11,001-11,006. Reich, P. B., M. B. Walters, D. S. Ellsworth, and C. Uhl (1994), Photosynthesis-nitrogen relations in Amazonian tree species. 1. Patterns among species and communities, Oecologia. 97, 62-72. Ribeiro, L. F., and M. Tabarelli (2002), A struchlral gradient in cerrado vegetation in Brazil: Changes in woody plant density, species richness, life history and plant composition, J. Trop. Ecol., 18, 775-794. Rice, A. H., E. H. Pyle, S. R. Saleska, L. Hutyra, P. B. Carmargo, K. Portilho, D. F. Marques, and S. F. Wofsy (2004), Carbon balance and vegetation dynamics in an old-growth Amazonian forest, Ecol. Appl., 14, 855-871. Rizzini, C. T. (1963), A flora do cerrado. Analaise floristica das savmmas centrais, in Simposio Sabre a Cerrado, edited by M. G. Ferri, pp. 125-177, Univ. of Sao Paulo, Sao Paulo. Roberts, D. A, B. W. Nelson, 1. B. Adams, and F. Palmer (1998), Spectral changes with leaf aging in Amazon caatinga, Trees, 12, 315-325. Ruggeiro, P. G. C., M. A Batahla, V. R. Pivello, and S. T. Meirelles (2002), Soil-vegetation relationships in cerrado (Brazilian savanna) and semi-deciduous forest, Southeastern Brazil, Planl Ecol., 160,1-16. Saatchi, S. S., R A. Houghton, R. C. Dos Santos Alvala, J. V. Soare, and Y. Yu (2007), Distribution of aboveground live
LLOYD ET AL. biomass in the Amazon basin, Global Change BioI., 13, 816837. Saatchi, S., et al. (2009), Mapping landscape scale varations offorest structure, bior~ihss, and productivity in Amazonia, Biogeosci. Discuss., 6, 546/-5505. Salazar, L. F., C A Nobre, and M. D. Oyama (2007), Climate change consequences on the biome distribution in tropical South America, Geophys. Res. Lett., 34, L09708, doi:IO.10291 2007GL029695. Saleska, S., H. da Rocha, B. Kruijt, and A Nobre (2009), Ecosystem carbon fluxes and Amazon forest metabolism, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000728, this volume. Saleska, S. R, et al. (2003), Carbon in Amazon forests: Unexpected seasonal fluxes and disturbance-induced losses, Science, 302, 1554-1557. Sanchez, P. A, and S. W. Buol (1974), Properties of some soils of the Amazon Basin of Peru, Soil Sci. Soc. Am. Proc., 42, 771-776. Santiago, L. S., G. Goldstein, F. C. Meinzer, 1. B. Fisher, K. Machado, D.Woodurff, and T. Jones (2004), Leafphotosynthetic traits scale with hydraulic conductivity and wood density in Panamanian forest canopy trees, Oecologia, 140, 543-550. Santos, A 1. B., G. T. da Silva, H. S. Miranda, A C. Miranda, and J. Lloyd (2003), Effects of fire on surface carbon, energy and water vapour fluxes over campo sujo savanna in Central Brazil, Funcl. Ecol., 17, 711-719. Santos, A. 1. B., C. A Quesada, G. T. Silva, 1. F. Maia, H. S. Miranda, A. C. Miranda, and J. Lloyd (2004), High rates of net ecosystem carbon assimilation by Bracchiara pasture in the Brazilian Cerrado, Global Change Bio!., 10,877-885. Sarmiento, G. (1983), The savannas of Tropical America, in Ecosystems oflhe World, vol 13, edited by F. Bourliere, pp. 245288, Elsevier, Amsterdam. Schimper, A F. W. (1903), Plant Geogmphy Upon A Physiological Basis, translated by W. R. Fisher, Gustav Fischer, Jena. Scholz, F. G., S. 1. Bucci, G. Goldstein, F. C. Meinzer, and A C. Franco (2002), Hydraulic redistribution of soil water by neotropical savanna trees, Tree Physiol., 22, 603-612. Scholz, F. G., S. 1. Bucci, G. Goldstein, M. Z. Moreira, F. C. Meinzer, 1.-c. Domec, R. Villalobos-Vega, A C. Franco, and F. MirallesWilhelm (2008), Biophysical and life-histOlY determinants of hydraulic lift in Neotropical savanna h'ees, Funcl. Ecol., 22, 773-786. Silver, W. L., A W. Thompson, M. E. McGroddy, R. K. Varner, J. D. Dias, H. Silva, P. M. Crill, and M. Keller (2005), Fine root dynamics and trace gas fluxes in two lowland tropical forest soils, Global Change Bioi., 11, 290-306. Sobrado, M. A (1991), Cost benefit relationships in deciduous and evergreen leaves of tropical illy forest species, Funet. Eco!., 5, 608-616. Sobrado, M. A (1993), Trade-off between transport efficiency and leaflife-span in a tropical dly forest, Oecologia, 96, 19-23. Sobrado, M. A (1997), Embolism vulnerability in droughtdeciduous and evergreen species of a tropical dly forest, Acla Oecol., 18, 383-391. Sombroek, W. G. (1966), Amazon Soils. A Reconnaissance oflhe Soils of the Brazilian Amazon Region, Centre for Agricultural Publications and Documentation, Waginingen, Netherlands.
483
Sombroek, W. G. (1984), Soil of the Amazon region, in The Amazon. Limnology and Landscape Ecology of a Mighty Tropical Rivei' and Its Basin, edited by H. Sioli, pp. 521-536, Springer, Boston. Sombroek, W. G. (2000), Amazonian landforms and soils in relations to biological diversity, Acta Amazonica, 30, 81-100. Swendon, N. G., and B. J. Enquist (2007), Ecological and evolutionary determinants of a key plant functional trait: Wood density and its community wide variation across latitude and elevation, Am. J. Bot., 94, 451--459. ter Braak, C. 1. F. (1994), Canonical commllility ordination. part I: Basic theOly and linear methods, Ecoscience, I, 127-140. ter Braak, C. 1. F., and P. Smilauer (2002), CANOCO Reference 1I1anual and CanoDrawfor Windows User's Guide: Software for Canonical Community Ordination (Version 4.5), Microcomputer Power Ithaca, New York. Townsend, A R., C. C. Cleveland, G. P. Asner, and M. M. C. Bustamante (2007), Controls over foliar N:P ratios in tropical rain forests, Ecology, 88,107-118. Turner, I. M. (2001), The Ecology of Trees in the Tropical Forest, Cambridge Univ. Press, Cambridge, U. K. van Donselaar, 1. (1968), Phytogeographic notes on the savanna flora of southern Surinam (South America), Acta Bot. Neerl., 17, 202-214. van Donselaar, 1. (1969), Observations on savanna vegetationtypes in the Guianas, J. Veg. Sci., 17, 271-312. van Schaik, C. P., J.; W. Terborgh, and S. 1. Wright (1993), The phenology of tropical forests-Adaptive significance and consequences for primmy consumers, Annu. Rev. Eco!. Syst., 24, 353-377. . Varanda, E. M., C. V. Ricci, and 1. M. Brasil (1997), Especies congeneres da mata e do cerrado: Teor de proteinas e compostos fen61icos, Bal. Bot. Univ. S. Paulo, 17, 25-30. Vieira, S., et al. (2004), Forest stmcture and carbon dynamics in Amazonian tropical rain forests, Oecologia, 140,468--479. Vitousek, P. M. (1984), Litterfall, nutrient cycling and nutrient limitation in tropical forest, Ecology, 65, 285-298. Vourlitis, G. L., N. Priante-Filho, M. M. S. Hayashi, 1. De Sousa Nogueira, F. Raiter, W. Hoegel, and 1. H. Campelo (2004), Effects of meteorological variations on the CO 2 exchange ofa Brazilian transitional tropical forest, Ecol. App!., 14, S89-SI00. Walker, T. W., and 1. K. Syers (1976), The fate ofphospholUs during pedogenesis, Geoderma, 15, 1-19. Westoby, M., D. S. Falster, A T. Moles, P. A. Vesk, and 1. J. Wright (2002), Plant ecological strategies: Some leading dimensions of variation between species, Annu. Rev. Eco!. Syst., 33, 125-159. Williams, M., Y. Malhi, A D. Nobre, E. D. Rastetter, J. Grace, and M. G. P. Pereira (1998), Seasonal variation in net carbon exchange and evapotranspiration in a Brazilian rain forest: A modelling analysis, Plant Cell Environ., 21, 953-968. Wong, S. C., 1. R. Cowan, and G. D. Farquhar (1979), Stomatal conductance correlates with photosynthetic capacity, Nalure, 282,424--426. Wright, 1. 1., et al. (2004), The worldwide leaf economics spectrum, Nalure, 428, 821-827.
484
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
Wright, 1. J., et al. (2006), Relationships among ecologically impOl·tant dimensions of plant trait variation in seven neotropical forests, Ann. Bot., doi: 10.1093/apb/mv I066. Xiao, X. M., 3': Hagen, Q. Y. Zhang, M. Keller, and B. Moore (2006), Detecting leaf phenology of seasonally moist tropical forests in South America with multi-temporal MODIS images, Remote Sens. Environ., 103,465-473.
\
N. Fyllas and J. Lloyd, School of Geography, University of Leeds, Leeds LS2 9JT, UK. ([email protected])
M. Goulden, Earth System Science, University of California, Irvine, 3319 Croul Hall, Irvine, CA 92697-3100, USA. J. P. Ometto, Centro de Ciencias do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Avenida dos Astronautas 1758, Sao Jose dos Campos, SP CEP 12227-010, Brazil. S. Patino, Grupo de Ecologia de Ecosistemas Terrestres Tropicales, Universidad Nacional de Colombia, Sede Amazonia, Instituto Amaz6nico de Investigaciones-Imani, Kilometro 2, via Tarapaca, Leticia, AM, Colombia. C. A. Quesada, Institito Nacional de Pesquisas da Amazonia, Avenida Andre Araujo 2936, Aleixo, Manaus, AM CEP 69060001, Brazil.
Surface Waters in Amazonia: Key Findings and Perspectives John M. Melack Bren School ofEnvironmental Science and Management and Department ofEcology, Evolution, and Marine Biology, University of California, Santa Barbara, Cal(fornia, USA
Reynaldo L. Victoria Centro de Energia Nuclear na Agricultura, Piracicaba, Brazil
Javier Tomasella Centro de Ciencia do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, Brazil
The four chapters representing studies of surface waters in Amazonia span spatial scales from very small catchments to the whole lowland Amazon basin. Hydrological aspects are examined with detailed field studies, remote sensing, and modeling. Organic carbon at all scales, and nitrogen and phospholUs in small catchments are emphasized. This introductOly chapter highlights key fin~ings reported in the four chapters and offers synthetic perspectives.
The central question for the Large-Scale BiosphereAtmosphere Experiment in Amazonia (LBA) studies of surface waters as framed by Richey et al. [1997] is stated in the LBA Concise Science Plan as, "How do the pathways and fluxes of organic matter, nutrients and associated elements through river corridors (riparian, floodplain, channels and wetlands) change as a function of land cover?" The four chapters representing these studies span multiple scales from very small catchments to mesoscale catchments to regional floodplains to the whole lowland Amazon basin. Hydrological aspects of these systems are examined with detailed field studies, remote sensing, and modeling. Chemical aspects considered emphasize organic carbon at all scales and include nitrogen and phosphorus in small catchments. This introductory chapter highlights key find-
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10. I029/2009GM000876
ings reported in the four chapters starting at the small scale and offers synthetic perspectives. Tomasella et at. [this volume] summarize studies of hydrology and nutrient fluxes in small catchments and responses to conversion offorest to pasture. Overall, the studies indicate that the conversion of forest to pasture produces an increase of discharge, an increase of stormflow, and a reduction of evaporation. Changes of soil hydraulic conductivity with depth are fundamental to stormflow generation and are influenced by the intensity of soil disturbance. Land use histmy before abandonment of deforested areas is clUcial to understand the potential effects of clearance on soil hydraulic properties and, in particular, how rapidly saturated hydraulic conductivity recovers. Catchment studies in various LBA sites indicate that the groundwater is impmiant in the generation of base flow and stormflow. Recent work in eastern Pant by Moraes et al. [2006] measur~d saturated hydraulic conductivity near the surface of a pasture of about 4 mm h- I , compared to 230 mm h- I in a paired forested catchment. In paired catchments in central Amazonia, Trancoso [2006] found a reduction of recession 485
484
ECOPHYSIOLOGY OF FOREST AND SAVANNA VEGETATION
Wright, 1. J., et al. (2006), Relationships among ecologically important dimensions of plant trait variation in seven neotropical forests, Ann. Bot., doi: 10.1093/apb/mvl066. Xiao, X. M., S>'t Hagen, Q. Y. Zhang, M. Keller, and B. Moore (2006), Detecting leaf phenology of seasonally moist tropical forests in South America with multi-temporal MODIS images, Remote Sens. Environ., 103,465-473.
, N. Fyllas and J. Lloyd, School of Geography, University of Leeds, Leeds LS2 9JT, UK. ([email protected])
M. Goulden, Earth System Science, University of California, Irvine, 3319 Croul Hall, Irvine, CA 92697-3100, USA. J. P. Ometto, Centro de Ciencias do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Avenida dos Astronautas 1758, Sao Jose dos Campos, SP CEP 12227-010, Brazil. S. Patino, Grupo de Ecologia de Ecosistemas Terrestres Tropicales, Universidad Nacional de Colombia, Sede Amazonia, Instituto Amaz6nico de Investigaciones-Imani, Kilometro 2, via Tarapaca, Leticia, AM, Colombia. C. A. Quesada, Institito Nacional de Pesquisas da Amazonia, Avenida Andre Aralljo 2936, Aleixo, Manaus, AM CEP 69060001, Brazil.
Surface Waters in Amazonia: I(ey Findings and Perspectives John M. Melack Bren School ofEnvironmental Science and ]Y[clllagement and Department ofEcology, Evolution, and ]y[arine Biology, University ofCalifomia, Santa Barbam, Califomia, USA
Reynaldo L. Victoria Centro de Energia Nuclear na Agricultura, Piracicaba, Brazil
Javier Tomasella Centro de CiI!ncia do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, Brazil
The four chapters representing studies of stu'face waters in Amazonia span spatial scales from very small catchments to the whole lowland Amazon basin. Hydrological aspects are examined with detailed field studies, remote sensing, and modeling. Organic carbon at all scales, and nitrogen ana phosphorus in small catchments are emphasized. This introductory chapter h~ghlights key findings reported in the four chapters and offers synthetic perspectives, .
The central question for the Large-Scale BiosphereAtmosphere Experiment in Amazonia (LBA) studies of surface waters as framed by Richey et al. [1997] is stated in the LBA Concise Science Plan as, "How do the pathways and fluxes of organic matter, nutrients and associated elements through river corridors (riparian, floodplain, channels and wetlands) change as a function of land cover?" The four chapters representing these studies span multiple scales from very small catchments to mesoscale catchments to regional floodplains to the whole lowland Amazon basin. Hydrological aspects of these systems are examined with detailed field studies, remote sensing, and modeling. Chemical aspects considered emphasize organic carbon at all scales and include nitrogen and phosphorus in small catchments. This introductory chapter highlights key find-
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2009GM000876
ings reported in the four chapters starting at the small scale and offers synthetic perspectives. Tomasella et al. [this volume] summarize studies of hydrology and nutrient fluxeS in small catchments and responses to conversion offorest to pasture. Overall, the studies indicate that the conversion of forest to pasture produces an increase of discharge, an increase of stonnflow, and a reduction of evaporation. Changes of soil hydraulic conductivity with depth are fundamental to storrnflow generation and are influenced by the intensity of soil disturbance. Land use history before abandonment of deforested areas is crucial to understand the potential effects of clearance on soil hydraulic properties and, in particular, how rapidly saturated hydraulic conductivity recovers. Catchment studies in various LBA sites indicate that the groundwater is impOliant in the generation of base flow and storrnflow. Recent work in eastern Para by Moraes et al. [2006] measured saturated hydraulic conductivity near the surface of a pasture of about 4 mm h- I , compared to 230 mm h- 1 in a paired forested catchment. In paired catchments in central Amazonia, Trancoso [2006] found a reduction of recession 485
486
SURFACE WATERS IN AMAZONIA: KEY FINDINGS AND PERSPECTIVES
time and time of peak discharge as well as a significant increase in the stormflow in the pasture compared with forested catchment; annual evaporation in the pasture was estimated as''t876 mm, while it was 1277 mm in the forest. A study in the Asu catchment by Tomasella et al. [2008] showed how storage in the groundwater system modulated seasonal climate anomalies from 1 year to the next and that the deep unsaturated zone played a key role in reducing most of the intraseasonal variability. A new terrain descriptor using information fi'om the Shuttle Radar Topographic Mission was applied to the region around the Asu catchment and indicated that waterlogged environments and valley bottoms occupied 43% of the area [Renno et al., 2008]. Shldies before and during LBA have reported low dissolved inorganic nitrogen concentrations in stream water, which are much lower than in soil solutions [Williams and Melack, 1997; Neill et al., 2001; Markewitz et al., 2004]. However, the mechanisms of nitrogen reduction, presumably in the riparian zone, remain uncertain. Elevated nitrate concentrations in streams do occur right after deforestation [Williams et al., 1997]. Once inorganic nitrogen enters small forested streams, recent work indicates that it has the potential to travel long distances [Neill et al., 2006]. After conversion to pashlre, if grasses infill the stream channel, nitrogen transport is modified. Phosphorus concentrations in soil solution and in forest streams are typically very low [Lesack, 1993; Neill et aI., 200 I; Markewitz et aI., 2004]. Deforestation appears not to increase soil solution and groundwater phosphate concentrations, at least at the few sites with measurements, and this is consistent with the high sorption capacity of the iron and aluminum oxides that predominate in the lowland Amazon basin. Uptake lengths for phosphate in pasture streams are short, suggesting rapid uptake by benthic and attached plant and microbial cOlllinunities. Richey et al. [this volume] outlined progress during LBA toward an understanding of processing of the multiple fractions of organic carbon fi'om uplands through streams and small to medium rivers. In very small headwater streams, Johnson et al. [2007, 2008] documented groundwater discharge of CO 2 and its subsequent evasion as a conduit for terrestrially respired carbon. By combining estimates of groundwater fluxes with soil pC0 2, they calculated CO 2 evasion from first-order streams draining the uplands of the Amazon as 114 ± 10 Tg C a-I. To extrapolate estimates of gas evasion from small- to medium-sized rivers requires information on their surface area, and an extrapolation fi'om an analysis done in the Ji-Parana basin to the entire Amazon for third- to fifth-order rivers was done by RaseI'[[ et al. [2008]. In light of the multiple scales and heterogeneity of the Amazon basin, deciphering the sources of organic carbon
that fuel the production of carbon dioxide resulting in almost all the waters of the basin being supersahlred with respect to equilibrium with the atmosphere is a significant challenge. Based on a survey of carbon isotopes in organic and inorganic fractions throughout mountain and lowland rivers of the Amazon basin, Mayorga et al. [2005] found that the primary source of respired CO 2 in the lowlands was <5 years old on average and that pC0 2 was commonly isotopically distinct from coincident organic carbon fi'actions [fine particulate organic (FPOC) and coarse particulate organic carbon]. However, recent measurements of respiration by Ellis (unpublished data) revealed a positive correlation with FPOC in lowland rivers and streams. Furthermore, examination of the 0 13 C of respired CO 2 revealed that this CO 2 in the Rio Negro and in small, shaded streams in Acre was consistent with carbon originating from C3 plants, while downstream in the Rio Solimoes, temporal variability in the organic source fueling respiration was detected. During early rising water, the material being respired had a 0 13 C of -22.9%0, indicating an important contribution from C4 aquatic macrophytes. Melack et al. [this volume] examine floodplains with an emphasis on hydrological and biogeochemical processes and remote sensing analyses that allow regionalization of fluxes of carbon. Though basin-scale models with floodplain inundation assume a horizontal water surface approximately equal to the levels in the main river channels and operate at moderate resolution, recent results indicate considerable spatial and temporal variations in elevations of water surfaces across Amazon floodplains [Alsd07:f et aI., 2000]. In the first application of a two-dimensional hydrodynamic model to a large reach of Amazon floodplain, Wilson et al. [2007] found that more than 40% of the total river flow was routed through the floodplain near the confluence ofthe rios Purus and Solimoes. During LBA, floodplain ecosystems to the north, west, and east of the central area near Manaus, where most preLBA research was conducted, were examined in terms of their ecology, hydrology, and biogeochemistry (e.g., Jau basin [Rosenqvist et al., 2002]; upper Negro interfluvial savannas [Belger, 2007]; Balbina reservoir [Kemenes et aI., 2007]; Lake Curuai [Maurice-Bourgoin et al., 2007; Bonnet et al., 2008]; and Lake Amana [Silva, 2005; Rodriques, 2007]). Based on a methodology for classification of Amazonian wetlands using JERS-l radar mosaics [Hess et al., 2003], Melack and Hess [2009] mapped 14% of the lowland Amazon basin (the region less than 500 m above sea level) as floodable at 100 m resolution, of which about 76% is represented by floodable woody vegetation and 8% by open water. Floodplains play an important role in the organic carbon balance of the Amazon basin and are sites of high rates
MELACK ET AL. of aquatic plant production and a major source of methane to the troposphere. C~lculations done by Melack et al. [this volume] estimated ,bobl net production attributed to flooded forests (excluding!wood increments), aquatic macrophytes, phytoplankton, .atrd periphyton for the 1.77 million km2 characterized by Hess et al. [2003] as about 300 Tg C a-I. Flooded forests wyre estimated to account for 62% of the total, aquatic macrophytes for 34% with the remaining 4% associated with periphyton and phytoplankton. Approximately 10% of the total value equals the export of organic carbon by the Amazon River [Richey et al., 1990J, methane emission is about 2% to 3% [Melack et al., 2004], and a similar percent is likely to be buried in sediments. The remaining portion is close to being sufficient to fuel the respiration that results in the degassing of 210 ± 60 Tg C a-r as carbon dioxide from the rivers and floodplains [Richey et al., 2002]. Costa et al. [this volume] review shldies that examine the effects of climate variability and changes in land cover on river flow throughout the Amazon basin. Both empirical and modeling results are examined. For example, based on analyses of river discharge from stations distributed through the basin, Marengo [1995] and Marengo et al. [1998] detected a spatial component to ENSO correlated variations in river flow with low discharges concentrated in the northwestern portions of the basin during El Nifios, and detected a decadal-scale variability with increased discharge from 1945 to 1960 and decreased discharge fi'om 1970 to 1990. Furthermore, Coe et al. [2002, 2007] modeled monthly mean river discharge using long-term records of precipitation and noted years of relatively high discharge clustered in the 1940s-1950s and the 1970s and low-discharge years clustered in the 1960s and 1980s-1990s. This pattern, although spatially variable in magnihlde, occurred throughout the basin, and ENSO events are evident, embedded within a 28year mode ofvariability. Though most of the interannual and interdecadal variability in river discharges appears related to climatic variability, increased deforestation appears to be contributing in some areas, such as the Tocantins basin [Costa et al., 2003]. Not unexpectedly, as more locations over longer time periods are investigated throughout the Amazon basin, the regional heterogeneity and temporal variability are becoming more apparent as are the influences of human- and climate-induced changes. Hence, to advance understanding of Amazon surface waters will require further integration of the multiscale data and models. Remote sensing will continue to . play an important role. For example, the combination offield measurements at numerous sites, innovative remote sensing, and hydrologic modeling has led to recognition of the regional significance of carbon dioxide evasion fi'om small streams, mesoscale rivers, floodplains, and major river chan-
487
nels. Experimental shldies have refined parameterizations of gas transfer coefficients. The organic carbon fueling the respiration that leads to the supersahlration of carbon dioxide is now known to vary by scale, e.g., streams are dominated by terrestrial sources, and among ecosystems, e.g., floodplains have large local sources, and isotopic analyses indicate large river channels have mixed sources. REFERENCES Alsdorf, D. E., J. M. Melack, T. Dunne, L. A. K. Mertes, L. L. Hess, and L. C. Smith (2000), Interferometric radar measurements of water level changes on the Amazon floodplain, Nalure, 404,174-177.
Belger, L. (2007), Fatores que infiuem na emissao de CO 2 e CILt em areas alagaveis interfluviais do medio Rio Negro, Ph.D. thesis, Univ. Federal do Amazonas and Instituto Nacional de Pesquisas da Amazonia, Manaus, Brazil. BOlmet, M. P., et al. (2008), Floodplain hydrology in an Amazon floodplain lake (Lago Grande de Curuai), J. Hydro!., 349, 18-30. Coe, M. T., M. H. Costa, A. Botta, and C. M. Birkett (2002), Longtenn simulations of discharge and floods in the Amazon Basin, J. Geophys. Res., 107(020), 8044, doi:lO.l029/200IJD000740. Coe, M. T., M. H. Costa, and E. Howard (2007), Simulating the surface waters of t~e Amazon River Basin: Impacts of new river geomorphic and dynamic flow parameterizations, Hydro!. Processes, 21,2542-2553, doi: IO.1002/hyp.6850. Costa, M. H., A Bott'a, and J. A Cardille (2003), Effects oflargescale changes in I~nd cover on the discharge of the Tocantins River, Southeastern Amazonia, J. Hydro!., 283, 206-217. Costa, M. H., M. T. Coe, and J. L. Guyot (2009), Effects of climatic variability and deforestation on surface water regimes, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000738, this volume. Hess, L. L., J. M. Melack, E. M. L. M. Novo, C. C. F. Barbosa, and M. Gastil (2003), Dual-season mapping of wetland inundation and vegetation for the central Amazon basin, Remote Sens. Environ., 87, 404-428. Johnson, M. S., M. Weiler, E. G. Couto, S. J. Riha, and J. Lehmann (2007), Storm pulses of dissolved CO 2 in a forested headwater Amazonian stream explored using hydrograph separation, WaleI' Resow'. Res., 43, W1l201, doi:l0.l02912007WR006359. Johnson, M. S., J. Lehmaml, S. Riha, A V. Krusche, J. E. Richey, J. P. H. B. Ometto, andE. G. Couto (2008), CO 2 efflux from Amazonian headwater streams represents a significant fate for deep soil respiration, Geophys. Res. Let., 35, L17401, doi:10.1029/ 2008GL034619. Kemenes, A, B. R. Forsberg, and J. M. Melack (2007), Methane release below a tropical hydroelectric dam, Geophys. Res. Let., 34, Ll2809, doi:l0.l02912007GL029479. Lesack, L. F. W. (1993), Export ofnutrients and major ionic solutes from a rain forest catclill1ent in the central Amazon basin, Water Resow'. Res., 29(3), 743-758. Marengo, J. A (1995), Variations and change in South American streamflow, CUm. Change, 21,99-117,
488
SURFACE WATERS IN AMAZONIA: KEY FINDINGS AND PERSPECTIVES
Marengo, J. A, J. Tomasella, and C. R. Uvo (1998), Trends in streamflow and rainfall in tropical South America: Amazonia, eastern Brazil, and northwestern Peru, J Geophys. Res., lO3(D2), 17$5-1783. Markewitz, D., E. A. Davidson, P. Moutinho, and D. Nepstad (2004), Nutrient loss and redistribution after forest clearing on a highly weathered soil in Amazonia, Ecol. App!., 14, Sl77-S199. Maurice-B~JUrgoin,L., M. P. Bonnet, J. M. Martinez, P. Kosuth, G. Cochonneau, P. M. Turcq, J. L. Guyot, P. Vauchel, N. Filizola, and P. Seyler (2007), Temporal d>,namics of water and sediment exchanges between the Curuai floodplain and the Amazon River main stream, Brazil, J Hydrol., 335,140-156. Mayorga, E., A K. Aufdenkampe, C. A Masiello, A V. K.1Usche, J. 1. Hedges, P. D. Quay, J. E. Richey, and T. A. Brown (2005), Young organic matter as a source of carbon dioxide outgassing from Amazonian rivers, Nature, 436, 538-541. Melack, J. M., and L. L. Hess (2009), Remote sensing of the distribution and extent of wetlands in the Amazon basin, in Amazonian Floodplain Forests: Ecophysiology, Ecology, Biodiversity and Sustainable lYIanagement, edited by W.J. Junk and M. Piedade, Springer, Berlin, in press. Me1ack, J. M., L. L. Hess, M. Gastil, B. R. Forsberg, S. K. Hamilton,!. B. T. Lima, and E. M. L. M. Novo (2004), Regionalization of methane emissions in the Amazon Basin with microwave remote sensing, Global Change BioI., 10, 530-544. Melack, J. M., E. M. L. M. Novo, B. R. Forsberg, M. T. F. Piedade, and L. Maurice (2009), Floodplain ecosystem processes, Geophys. Monogr. Ser., doi: 10.1 029/2008GM00072I , this volume. Moraes, J. M., A. Schuler, T. Dunne, R. O. Figueiredo, and R. L. Victoria (2006), Water storage and runoff processes in linthic soils under forest and pasture in eastern Amazon, Hydrol. Processes, 20, 2509-2526. Neill, c., L. A Deegan, S. M. Thomas, and C. C. Cerri (2001), Deforestation for pashlre alters nitrogen and phosphorus in soil solution and streamwater of small Amazonian watersheds, Ecol. Appl., 11,1817-1828. Neill, C., L. A Deegan, S. M. Thomas, C. L. Haupert, A V. K.rusche, V. M. Ballester, and R. L. Victoria (2006), Deforestation alters channel hydraulic and biogeochemical characteristics of small lowland Amazonian streams, Hydrol. Processes, 20, 2563-2580. Rasera, M., M. V. R. Ballester, A. V. Krusche, C. Salimon, L. A. Montebelo, S. R. Alin, R. L. Victoria, and J. E. Richey (2008), Estimating the surface area of small rivers in the southwestern Amazon and their role in CO 2 outgassing, Earth Interact., 12, 1-16, doi: 10. 1l75/2008EI257. 1. Renn6, C. D., AD. Nobre, L. A. Cuartas, J. V. Soares, H. G. Hodnett, J. Tomasella, and M. J. Waterloo (2008), HAND, a new terrain descriptor using SRTM-DEM: Mapping terra-firme rainforest environments in Amazonia, Remote Sens. Environ., 112, 3469-3481. Richey, J. E., J. 1. Hedges, A. H. Devol, P. D. Quay, R. Victoria, L. Martinelli, and B. R. Forsberg (1990), Biogeochemistly of carbon in the Amazon River, Linmo!. Oceanogr., 35, 352-371. Richey, J. E., S. R. Wilhelm, M. E. Mcclain, R. L. Victoria, J. M. Melack, and C. Araujo-Lima (1997), Organic matter and nutri-
ent dynamics in river corridors of the Amazon Basin and their response to anthropogenic change, Cienc. Cuft., 49, 98-110. Richey, J. E., J. M. Melack, A K. Aufdenkampe, V. M. Ballester, and L. L. Hess (2002), Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric CO 2, Nature, 416,617-620. Richey, J. E., A V. Krusche, M. S. Johnson, H. B. da Cunha, and M. V. Ballester (2009), The role of rivers in the regional carbon balance, Geophys. MO/logr. Ser., doi: 10.1 029/2008GM000734, this volume. Rodrigues, R. (2007), Diversidade floristica, estruhlra da comunidade arb6rea e suas relayoes com variaveis ambientais ao longo do lago Amana, RDSA, Amazonia Central, masters thesis, Instituto Nacional de Pesquisas da Amazonia, Manaus, Brazil. Rosenqvist, A, B. R. Forsberg, T. Pimentel, Y. A Rauste, and J. E. Richey (2002), The use of spaceborne radar data to model inundation patterns and trace gas emissions in the central Amazon floodplain, I/lt. J Remote SeIlS., 23,1303-1328. Silva, R. M. (2005), Fisicoquimica e macr6fitas no lago Amana, masters thesis, Instituto Nacional de Pesquisas da Amazonia, Manaus, Brazil. Tomasella, J., M. G. Hodnett, L. A Cuartas, A. D. Nobre, M. J. Waterloo, and S. M. Oliveira (2008), The water balance of an Amazonian micro-catchment: TI;e effect of interannual variability of rainfall on hydrological behaviour. Hydro!. Processes, 22, 2133-2147. Tomasella, J., C. Neill, R. Figueiredo, and A D. Nobre (2009), Water and chemical budgets at the catchment scale including nutrient exports from intact forests and disturbed landscapes, Geophys. Monogr. Ser., doi:lO.l02912008GM000727, this volume. Trancoso, R. (2006), Mudanyas na coberhlra da terra e alterayoes na resposta hidrol6gica de bacias hidrograficas da Amazonia, masters thesis, Instihtto Nacional de Pesquisas da Amazonia, Manaus. Williams, M. R., and J. M. Melack (1997), Solute export from forested and partially deforested catchments in the central Amazon, Biogeochelllistl)!, 38, 67-102. Williams, M. R., T. R. Fisher, and J. M. Melack (1997), Solute dynamics in soil water and groundwater in a central Amazon catchment undergoing deforestation, Biogeochemistl)!, 38, 303-335. Wilson, M., P. Bates, D. Alsdorf, B. Forsberg, M. Horritt, J. Melack, F. Frappart, and J. Famglietti (2007), Modeling large-scale inundation 'of Amazonian seasonally flooded wetlands, Geophys. Res. Lett., 34, Ll5404, doi:10.1029/2007GL030156.
J. M. Melack, Department of Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, CA 93106, USA ([email protected]) J. Tomasella, Centro de Ciencia do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, SP 12630-00, Brazil. ([email protected]) R. L. Victoria, Centro de Energia Nuclear na Agricultura, Piracicaba, SP 13400-000, Brazil. ([email protected])
The Role of Rivers in the Regional Carbon Balance Jeffrey E. Richey, I Alex V. K.rusche,2 Mark S. Johnson,3 Hillandia B. da Cunha,4 and Maria V. Ballester 2 Through the evolution of the Large-Scale Biosphere-Atmosphere Experiment in Amazonia, fluvial systems evolved from being regarded as ecologically interesting, but not necessarily relevant to the carbon budget, to important systems outgassing a volume of CO 2 roughly equal to the carbon sequestered by the forest. Resolving the role of fluvial systems in the carbon balance of the Amazon basin is a problem in scaling, from small seeps and springs to streams to larger rivers. Groundwater discharge of CO 2 and its subsequent evasion is a significant conduit for terrestrially respired carbon in tropical headwater catchments. Hydrologic transport of dissolved CO 2 was equivalent to nearly half the gaseous CO2 contributions from deep soil (>2 m) to respiration at the soil surface. At larger scales, the dominant feature was a clear relation between discharge and biogeochemical concentrations, with systematic variance among sites. Seasonal distributions ofpC0 2 rose and fell almost exactly with the discharge hydrograph, while pH d~creased and dissolved organic carbon increased. This suggests a constancy of prqcesses across systems. Gas exchange is greater than previously thought, primarily dlle to greater outgassing in smaller streams than expected. No single organic matter source consistently fuels respiration; instead, the 8 13 C of respiration-derived CO 2 varies with time and space. Photochemical production of labile bioavailable compounds would appear to be limited to clear water conditions. Based on these results, the original Richey et al. (2002) estimate of outgassing of 1.2 ± 0.3 Mg C ha- I a-I is conservative; the true value is likely higher.
ISchool of Oceanography, University of Washington, Seattle, Washington, USA 2Laborat6rio de Analise Ambiental e Geoprocessamento, Centro de Energia Nuclear na Agcricultura, Piracicaba, Brazil. 3Instihlte for Resources, Environment and Sustainability and Department ofEarth and Ocean Sciences, University of British CoImnbia, Vancouver, British Columbia, Canada. 4Coordenyao de Pesquisas em Clima e Recursos Hidricos, Instituto Nacional de Pesquisas da Amazonia, Manaus, Brazil. . Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000734
1. INTRODUCTION A long-standing paradigm of river networks is that they are minor components in the global carbon cycle, passively connecting the land and ocean reservoirs. As quoted by Cole et al. [2007], Leopold et al. [1964] described rivers as the "gutters down which flow the ruins of continents." The major biogeochemical role ofriver systems is typically considered to be the "carbon leakage" (in the sense of Malhi and Grace [2000]), primarily the fluvial expOli of total organic carbon (TOC) and dissolved inorganic carbon (DIC) to the ocean of ~0.4 Gt C a-I, respectively [Degens et al., 1991; Stallard, 1998]. While these fluxes are significant compared
489
490
RICHEY ET AL.
ROLE OF RIVERS IN THE REGIONAL CARBON BALANCE
to the net oceanic uptake of anthropogenic CO 2 of ~2 Gt C a-I [Sarmiento and Sundquist, 1992], they are small components of the global carbon cycle. These view§" influenced the initial design for the inclusion of water chemistry within the Large-Scale BiosphereAtmosphere Experiment in Amazonia (LBA) project. Richey et al. [1997] reviewed the state of understanding ofthe fluvial dynamics 6fthe Amazon basin at that time. With regard to carbon cycling, they posed a series of questions, addressing the nature ofthe transfer of organic matter from terresh'ial to river environments as well as the riverine fate of organic material: (1) How does the evolution of CO 2 and CH4 change with increasing proximity to the riparian zone and sh'eams? (2) How will the composition of organic matter entering sh'eams be altered under different land use change scenarios? (3) To what extent do intact riparian zones buffer sh'eams against changes due to anthropogenic activities in slllTounding uplands? (4) How far do sediments, nUh'ients, and organic matter travel before they are taken up, decomposed, temporarily stored, permanently buried, or degassed to the atmosphere? Richey et al. [1997] noted that the nature of the research required to address these questions crossed multiple scales, from the individual stream to the mouths of major tributaries. Small watershed studies where the export of water, sediments, nutrients, and organic matter can be quantified are as important to understanding "horizontal" carbon fluxes between land and water, as instrumented towers are to veliical fluxes between land and the atmosphere. Sampling transects of soil solution and trace gas fluxes from uplands through riparian zones into streams, in which the partitioning of organic matter and nutrients into gaseous, liquid, and particulate phases is tracked, can help identify the sequence of subsh'ate and oxidation/reduction conditions, which control the partitioning and degradation pathways. How, then, do the inputs from small streams and their watersheds translate downstream? This is a "mesoscale" question, where it is necessary to understand the distribution of moishlre regimes and biogeochemical processes at scales of ~ 10,000-100,000 km 2 . Within mesoscale areas, a significant problem in analyzing the impacts of land use change on aquatic systems is evaluating how far downstream local changes are detectable. The corollary is that the cumulative impact of change in a series of low-order streams may be manifested in higher-order streams in a significant, but nonadditive manner. Working across scales thus requires integrating a range of approaches. Finding paired basins that are forested/deforested for comparison is difficult. Rather, chemical tracers and computer models are useful. This understanding needs to be evaluated across different land covers and land uses, dominant climatological regimes, and topographic regimes.
For the integration of multiple mesoscale regions into regional, to whole basin perspectives, a functional basin scale is that of major tributaries, e.g., the downstream segments of the major tributaries, such as the Rio Madeira, Rio Tapaj6s, Rio Xingll, or Rio Tocantins. At this scale, the dynamics of factors controlling basin output from smaller scales would be aggregated, and the subsequent output would be comparable to the whole-basin scale of the hydrology and energy balance shldies [Richey et al., 2004]. Time series measurements of dissolved and particulate fractions are necessary to establish the basic composition regime of the river. Emphasis should include those tracers of terrestrial-scale controls that persist through the river system. Basin-scale modeling would be based on hydrologic routing modeling and process-level understanding. The objective is to predict the output hydrographs of water and chemical constiments under different conditions of land use. These perspectives were then translated into the original Concise Science Plan of LBA, where the central question for the study of surface water chemistry was the following: How do the pathways and fluxes of organic matter, nutrients, and associated elemellts through river corridors (riparian, floodplain, channels, and wetlands) change as a function of land cover? The perspective was that "because the river corridors of a region express the integrated interaction of hydrological processes and the land surface, understanding how the organic matter and nutrient composition of river corridors respond to forest conversion is an essential precursor to assessing the impact of land use change on the ecological functioning and sustainability ofthe region." The proposed research plan was to first, study lower-order streams, then to use modeling to consider elemental budgets at larger scales. An important conclusion from the initial sehlp of LBA is the fluvial system was regarded as an important ecological entity unto itself, but not a priori ofthe overall carbon cycle of Amazonia. What have we learned since? Our intent in this chapter is to outline the progress toward developing a quantitative understanding of the sequence of carbon processes from small streams to the Amazon main stem The focus on flowing water here complements the focus of Melack et al. [this volume] on inundated areas, to jointly place fluvial systems in the overall LBA context of the role of rivers in the regional carbon cycle of the Amazon basin. 2. A HEURISTIC MODEL OF CARBON FLOW THROUGH A RIVER SYSTEM To put our analysis into context, it is useful to evaluate the elements of the work done under LBA in relation to a heuristic river model. Richey [2004] formalized a concephIal
model of river corridors from Richey et al. [1997]. Understanding the processes/that control the pathways from initial source to final minerafization of riverine organic matter is impOliant on both re&,l~nal and global scales. At regional scales, river basins are l)J('fural integrators of surficial processes (Figure 1). Large rivers owe their flow and chemical loads to a much denser network of small rivers and streams bordered by areas of periodicaily inundated land, so that upland areas are dissected by corridors of wet soils and flowing water. Hence, understanding the hydrological and chemical patterns observed at the mouths of major rivers requires delineating the sequences ofbiogeochemical processes operating across multiple scales in time and space. The central premise of a river basin model is that the constihlents of river water provide an integrated record of upstream processes whose balances vary systematically depending upon changing interactions of flowing water with the landscape and the interplay ofbiological and physical processes [Karlsson et al., 1988; Billen et at., 1991]. That is, the chemical signahlres of riverine materials can be used to identify different drainage basin source regions and can be tied to landscape-related processes such as chemical weathering and nutrient retention by loc,al vegetation [Meyer et al., 1988]. Because of the dynamic namre and abrupt moismre gradients of river corridors, the cumulative signal fi'om a series of low-order streams may be manifested by higher-order rivers in a nonadditive manner. Within this framework, there are three primaly forms of carbon that are mobilized from land and transported into and through fluvial systems, each with characteristic pathways (Figure 2). Particulate organic carbon (POC) enters rivers
Atmosphere
Riparian/Floodplain
491
Surface Water
>
Figure 2. Grid-based view of land surface processes transferring water and its dissolved and particulate load to streams, where these constituents are subsequently processed and routed downstream.
from the erosion of soils (typically older materials) and as leaflitter (typically newly produced). Dissolved organic carbon (DOC) is produced through soil organic carbon being rendered soluble and entering streams via groundwater, surface runoff, and subsurface stormflow. Both POC and DOC also come from autochthonous production in rivers and associated floodplain environments. Atmospheric CO 2 fixed through photosynthesis and released into soils via microbial and root respiratioll is dissolved in soil water. This process sequesters atmospheric CO 2 via weathering of carbonate and silicate rocks, establishes the alkalinity, and influences the pH of water, which governs the subsequent partitioning of DIC between pC0 2 , bicarbonate, and carbonate ions. The dynamics of carbon in fluvial systems are not defined solely by the export fluxes of bulk carbon. Rather, they are defined as a complex interplay of multiple carbon fractions; each exhibits distinct dynamics and compositional traits that hold over broad ranges of geological, hydrological, and climatic conditions [Hedges et at., 1994]. 3. PATHWAYS OF RIVER CARBON THROUGH THE AMAZON
Figure 1. Schematic illustration of the major reservoirs and pathways in fluvial systems. Inputs from land occur directly or pass through the riparian zone. Streams coalesce to form larger rivers that exchange with their floodplain. Rivers can pass directly to the coastal zone or be restrained behind dams. Dotted lines indicate ex-' change with the atmosphere, grounded areas indicate sinks, arrows within boxes indicate internal transformations (adapted from the work of Richey [2004]). Copyright 2004 Scientific Committee on Problems of the Environment. Reproduced by permission ofIsland Press, Washington, DC.
3.1. Small Springs and Seeps A major unknown is what happens in small streams and seeps. How does terrestrial production via litterfall get transported into stream discharge? McClain et al. [1997] estimated that for terra finne streams near Manaus, 20% to 40% (in' Oxisols and Spodosols, respectively) of the dissolved organic matter (DOM) could derive from a combination of wetland seepage and in-channel leaching ofPOC. Remington
492
ROLE OF RIVERS IN THE REGIONAL CARBON BALANCE
et al. [2007] evaluated processes across a toposequence responsible for partitioning of DOC produced from litter between particles and what is available for export to streams. DOC was sorbedl'il1ore rapidly on the plateau and slope than in the valley bottom, as a function of both soil organic carbon content and mineral surface area. Johnson et al. [2006] examined the forms and quantities of organic cal'boh fluxes at the soil surface, and organic carbon exports from four small (1-2 ha) headwater catchments in the Juruena watershed, in the upper'Rio Tapaj6s watershed. Litterfall carbon at the soil surface was 43 times greater than the DOC flux in throughfall, with the highest rates of carbon deposition during the dry season. For watershed exports, however, the form and timing of organic carbon was reversed, where DOC comprised 59% of the annual TOC export, and exports were greatest during the 4-month rainy season (63% of total alillual exports). As is seen in largerrivers, fine particulate organic carbon (FPOC) in stream water represented a substantially larger flux than coarse particulate organic carbon (CPOC), with 37% and 4% of total annual organic carbon exports, respectively, and with POC mobilized primarily in the rainy season and strongly connected to StOlll1 events. In the rainy season, over 90% ofFPOC exports were transported by stormflow, while only 32% of DOC exports were expOlied by stolll1flow. Stream water DOC concentrations were found to increase linearly with increasing terrestrial litterfall during the dry season, indicating that instream processing of allochthonous litterfall is an important source of DOC during the dly season. Waterloo et al. [2006] examined organic carbon export dynamics in the blackwater Igarape Asu rainforest catchment ofthe Rio Negro. They found that DOC concentrations in rainfall, 1~2 mg L-I, were similar to those measured in rainfall elsewhere in the Amazon basin, producing annual atmospheric DOC deposition estimates of2-5 g m-2 a-I. Daily average DOC concentrations in runoff ranged from 8 mg L- 1 under low flow conditions to 27 mg L- 1 during large quick flow events. POC (> 10 f!m) averaged 28% of OC, with a median concentration of 4.1 mg L-I. EXpOlis associated with large storms were much higher than average daily export. Export of carbon during the wet seasons amounted to 70% of the total. Annual exports in river water were different between the years because of differences in runoff, from 26 g C m-2 in 2002 (transported by 1362 mm of runoff) to 11.7 g C m- 2 in 2003 (transported by 780 mm of runoff). Organic carbon exports were dominated by DOC, with exports in sediment constituting 6-8% of the total. Organic carbon export in stream water varied from 23 to 9 g m-2 of watershed area in 2002 and 2003, respectively. The annual average organic carbon export of 19 g m-2 a-I, as the sum of dissolved and particulate fractions over the 2 years ofthe study, was about
5-6% of the rainforest net primary production of 300~00 g m-2 a-I, estimated from eddy covariance measurements [see TO/llasella et al., this volume]. Neither of these studies included pC0 2 in their evaluations. Krllsche et al. [2009] found concentrations ranging from 5000 ~latm to over 20,000 ~latm in blackwater streams not far from Igarape Asu. While these concentrations are roughly that of DOC in these streams, they are 10-100 times supersaturated relative to the atmosphere. Johnson et al. [2008] found that 77% of carbon transported by water from the landscape was as terrestrially respired CO 2 dissolved within soils, over 90% of which evaded to the atmosphere within headwater reaches via turbulent mixing due to streambed roughness. Hydrologic transport of pC0 2 was equivalent to nearly halfthe gaseous CO 2 contributions fi'om deep soil (>2 m) to respiration at the soil surface. The pC0 2 in emergent groundwater was isotopically consistent with soil respiration and demonstrated agreement with deep soil CO 2 concentrations and seasonal dynamics. Deep soil (2-8 m) CO 2 concentration profiles during wet seasons indicated gaseous diffusion to deeper layers, thereby enhancing CO 2 drainage to streams. Hence, groundwater discharge of CO 2 and its subsequent evasion is a significant conduit for terrestrially respired carbon in tropical headwater catchments. This subsurface transport of soil CO 2 to tropical headwater streams was an order of magnitude greater than for temperate headwater catchments. Hence, tropical headwaters comprise a zone of rapid biogeochemical transformation, where emergent groundwater fuel base flow with water having a ratio of DIC to DOC nearly 30 times greater than the ratio for large Amazonian rivers. Following development of a method for real-time direct in situ measurement of pC0 2 in surface water and emergent groundwater [Johnson et al., 2009], several previously unobserved features of carbon cycling in headwater catchments became apparent. The CO 2 concentration ofhydrological flow paths were evaluated, and their relative contributions during base flow and storm events were explored empirically [Johnson et al., 2006] and through hydrograph separation of event water [Johnson et al., 2007]. Base flow continually delivers groundwater discharge to streams that is highly supersaturated in CO 2. The excess CO 2 results from equilibration of soil water with the high CO 2 concentrations of the soil atmosphere derived from autotrophic and heterotrophic respiration within soils. Groundwater discharge occurs in both focused and diffusive forms, where the fOlll1er represents groundwater emergence in springs and seeps, and the latter occurs via groundwater discharge across the streambed along the length of stream reaches that gain in discharge. The pC0 2 concentration in focused groundwater discharge, prior to interactions with the atmosphere, was
RICHEY ET AL.
generally above 50,000 ~latm (~20 mg COrC L- 1) [Johnson et aI., 2008]. Diffusive gl;olmdwater discharge contributes to maintaining stream pC02 well above equilibrium, even as r' excess CO 2 evades frpm the stream surface due to turbulent mixing in the headJY'ter channels. During storm events, the pre-event component of stormflow (e.g., water that is stored in catchments prior to precipitation events and released to streams during storms) was found to dominate total stormflow, although quick flow derived from direct precipitation and surface runoff contributes water that is at or near atmospheric concentration for C02 [Johnson et al., 2007]. Modeling of the hydrologic flow paths via hydrograph separation found that a slower eventwater flow path (e.g., subsurface stormilow) contributes "preevent CO 2'' via event-water peak, while a faster event-water flow path delivers low CO 2 water to streams. That is, water entering streams via surface runoff and direct precipitation is low in CO 2 , while water newly input to soils during a storm event dissolves and translocates soil CO 2 to streams. As the subsurface stormflow event-flow path is later-arriving than quicker flow paths (direct precipitation and surface runoff), the subsurface stormflow results in a pulse of CO 2 oqserved on the falling limb of storm hydrographs. The event-water CO 2 concentration peaks during the CO 2 pulse, and its concentration (~25,000 ppm) is consistent with that of soil CO 2 in the upper 50 cm of soil [Johnson et al., 2008], although it is only about half the concentration of CO 2 in emergent groundwater [Johnson et al., 2006]. The mechanism of equilibration ofsoil water with soil CO 2 during percolation and subsequent transport of terrestrial respiration products to streams allowed Johnson et al. [2008] to model potential headwater CO 2 evasion based on basin-wide published spatial data sets of hydrologic variables and soil properties. In this approach, groundwater concentrations of dissolved CO 2 were estimated based on carbonate equilibrium reactions for CO 2 in soil solution in relation to subsoil pH for the simplest case of pure water in equilibrium with soil CO 2 [McBride, 1994], which is valid for soils without exposure to strong acid inputs such as from acid rain [McBride, 1994]. Thus, soil pH determines the minimum pC0 2 of soil solution and groundwater, although the pC0 2 of soil air can increase significantly due to biological activity. However, in order to provide a minimum (e.g., conservative) estimate of dissolved CO 2 delivered to streams via CO 2-supersaturated groundwater flow paths, the model was structured to only consider equilibration reactions based on basin-wide spatial data sets. Soil pH was derived from the lower soil horizon (30-100 cm) of a global soils database [Batjes, 2005], which repOlis soil reaction class based upon the predominant soils of each pixel. Johnson et al. [2008] assigned soil pH values for each class (Table 1). In a very few cases (less than
493
Table 1. Soil Reaction Classes and pH Ranges From Digital Soils Database, and Soil pH Used for Computing Soil pC02" Soil Reaction Class
I 2 3
4 5 6 7 8
Soil pH Range pH < 5.5 5.5 8.5 (complex) pH < 6.5 (complex) 5.5 < pH < 8.5 (complex) pH> 7.5 (complex)
pH Used
Modeled Soil pC0 2 , %
4.5 6.0 7.0 8.0 9.0 5.0
0.04 0.04 0.04 0.04 0.51
7.0
0.04
8.0
0.04
5.13
aS ee Ba(jes [2005]. Atmospheric CO 2 concentration (0.04%) was used as a minimum condition when resulting pC0 2 values were less than atmospheric.
20 pixels), classes were reassigned due to missing data in the underlying database, which led to skewed soil pH values and isolated pixels of differing soil pH relative to soil carbonate contents from the same database [Batjes, 2005]. Previous estimates of DIC fluxes into groundwater using soil pH utilized a pH value of 4.0 for the most strongly acidic class of soils [l\{essler and Harvey, 2001]. Johnson et al. [2008] used a more conservative pH value of 4.5 together with an updated soil database with additional soil reaction classes compared to tHe earlier version [Kessler and Harvey, 2001]. This approach yielded modeled values of soil pC0 2 (Table 2) that agreed with, or were conservative relative to, theoretical [Brady and Weil, 1999; McBride, 1994] and observed values for eastern Amazonia for forest and pasture soils at 8 m [Davidson and TrlllJ1bore, 1995], for forested soils at 8 m in southern Amazonia [Johnson et a1., 2008] and for central Amazonian forest soils at 11 m under normal and drought conditions [Davidson et al., 2004]. The spatial distribution of groundwater fluxes was determined by Johnson et al. [2008] as the difference between long-term averages of annual precipitation [New et al., 1999] and actual evapotranspiration [AIm and Tateishi, 1994], which was scaled to the 1976-1996 mean annual discharge for the Amazon [Costa and Foley, 1999] to determine the water balance for each 0.1 pixel for average years. One standard deviation from the 1976-1996 mean annual water balance [Costa and Foley, 1999] was used to calculate wet and dly year water balances. The groundwater flux was set to zero for pixels where evapotranspiration exceeded precipitation. Johnson et al. [2008] excluded more than 300,000 k1112 of large rivers, wetlands, and seasonally inundated areas in central Amazonia, and an additional 500,000 km2 of permanently or seasonally inundated areas in other regions 0
494
ROLE OF RIVERS IN THE REGIONAL CARBON BALANCE
RICHEY ET AL.
495
Table 2. Measured and Modeled Soil pCO z for Amazonian Soils"
Location lie Eastern Amazon 2 0 59'S 47°31'W Southern Amazon lO o 25'S 58°46'W Central Amazon 2.897°S 54.952°W
Station !Db
Deepest Layer Reported, m
Measured Soil pCO z of Deepest Layer (Site Description)
Modeled pCO z, %
Reference
8
6.9% (forest) 6.3% (pasture)
5.13
Davidson and Tl'lIl11bore [1995]
2
8
4.9% (forest)
5.13
Johnson et al. [2008]
3
11
5.5% (control) 6.4% (throughfall exclusion)
5.13
Davidson et al. [2004]
"Measured values are means of data that include both dry season and wet season measurements for the deepest layer reported. bStation ID corresponds to the numbering oflocations in Plate 1.
of the Amazon basin based on remote sensing of high and low water periods [Melack and Hess, 2009]. The resulting mean annual groundwater discharge to headwater streams, 833 mm a-I, was significantly less than published estimates of groundwater drainage fluxes for the Amazon basin (1250 mm a-I) [Richey et al., 2002], but near estimates for headwater contributions to river network discharge in the Amazon [McClain and Elsenbeer, 2001]. The potential headwater COz outgassing flux for the terra finne portion ofthe Amazon basin was then calculated as the product of the groundwater flux and soil pCO z. The resulting COz evasion flux fi'om first-order streams for the terra finne portion of the Amazon basin was estimated as 114 (±10) x 1OIZ g C a-I [Johnson et al., 2008] (Plate 1). The areas with the highest potential COz evasion flux are those where the water flux (as precipitation minus evapotranspiration) is highest, and soils are acidic. These areas exhibit close correspondence with those areas with the highest soil COz emissions [Raich and Potter, 1995], which provides an independent check of the headwater COz evasion flux model, as the soil respiration data of Raich and Potter [1995] were not used to drive the model. Variability in the annual water balance for 1976 through 1996 [Costa and Foley, 1999] translates to a wet-year augmentation of the headwater outgassing flux by 1.0 x 10 13 g C a- 1 for a wet year and an equivalent decrease for a dry year. The influence ofland use change on this flux is more difficult to asceliain. Deforestation in the Brazilian component of the Amazon basin through 2004 has resulted in conversion of 16% ofthe original forest to other uses [Ometto et al., 2005]. However, since maximum COz concentrations in deep (>2 m) soil have not been shown to differ significantly between Amazonian forests and pastures [Davidson and TrUll/bore, 1995], the main influence on the headwater outgassing flux
is likely due to factors affecting the water balance following forest conversion. Although reduced evapotranspiration could increase the COz drainage flux, this term of the water balance is complicated by large uncertainties [Williams et al., 1997]. In order to maintain 'a conservative estimate of the headwater COz outgassing flux, Johnson et al. [2008] confined their analysis to reduced infiltration following land use conversion. Their simplifying assumptions that (1) stormflow increases by 100% following forest conversion [Williams et al., 1997], and (2) all stonnflow in the pertmbed system occms as overland flow and does not transport soil COz, resulted in a reduction of the total headwater outgassing flux by 1.7 x 10 1Z g C a-I. Overall, groundwater discharge is the predominant hydrologic flow path contributing inorganic carbon to headwater streams, while base flow and quick flow contribute approximately equivalent organic carbon fluxes. Total fluvial carbon fluxes in the headwaters are dominated by COz derived from terrestrial soil respiration, which is largely lost within the headwater reaches via gaseous evasion to the atmosphere. Quick flow delivers the majority of bioavailable DOC and pac to streams [Johnson et al., 2006], and while DOC and pac are biogeochemically and ecologically significant to downstream fluvial network, they are a lesser component of the carbon mass balance of headwater catchments compared to COz evasion. 3.2. Mesoscale Distributions ofRiver Carbon Fractions The first problem in understanding fluvial carbon dynamics at scales larger than relatively easily measmed discrete streams is determining the spatial and temporal distributions across a range of environments. Because of the vast and remote nature of the Amazon basin, the logistics of establish-
Plate 1. Potential headwater COz evasion flux for the Amazon basin modeled using available spatial data sets for hydrologic variables and soil pCO z calculated fi'om carbonate equilibrium reactions (adapted from the work of Johnson et al. [2008]). Gray areas within the basin correspond to pixels where precipitation is less than evapotranspiration and do not contribute to the evasion flux. White areas within the basin represent more than 800,000 km z of rivers, wetlands, and seasonally inundated areas that were excluded from headwater evasion calculations. Tpe basin-wide headwater COz evasion flux was estimated as 1.14 x 10 14 g C a-I after excluding permanently and seasonally inundated areas and accounting for land use change impacts, with a mean modeled COz efflux fi'om headwater strean1s of 195 kg C ha- I a-I. The numbered symbols refer to study locations of deep soil COz: 1, Davidson and Tl'lIlIIbol'e [1;995], 2, Johnson et al. [2008], and 3, Davidson et al. [2004], respectively. Additional study location details are presented in Table 2.
ing a comprehensive and representative sampling network are considerable. It is simply not feasible to send out a team from a central university or laboratOly and expect it to collect data with sufficient resolution in space and time. Accordingly, a sampling (and education and training) network was established, the "Rede Beija Rio," wherein each node is occupied by a researcher or a team of researchers from that site (Figme 3). An initial analysis of the data shows a series of provocative trends, demonstrating a high degree of spatial and temporal coherence in the distributions of carbon fi'actions across mesoscale rivers of the Amazon basin (J. Richey and A. Krusche, unpublished data, 2009). As illustrated in Figme 4, the dominant feature across all scales was a clear relation between discharge and biogeochemical concenh'ations, with systematic variance among sites. For example,pCO z is positively cOlTelated with discharge and tracks the hydrograph at all sites. The pCO z at low water ranges from 500 /-latm in the Rio Araguaia and Rio Ii-Parana to 1000 ~latm in the Rio Solimoes, to 2000 /-latm in the Rio Negro. High water concentrations exhibit a broader range and higher magnitude,
from 3000 /-latm (Rio Pimento Bueno) to 5000 ~lahn in the Rio Solimoes and 7000 /-latm in the lower Rio Negro. Interestingly,pCO z in the far upper Rio Negro, where pH is in the 3.5 range, had lower concentrations than at the mouth. The highest observed values were 20,000 /-latm in Campinas, a blackwater tributaty of the Rio Negro. The pH is inversely associated with the hydrograph and with pCO z. Rivers draining most directly from the Andes have considerable ranges in pH, such as the Rio Pmlls (7-8.4), Rio Madeira (6.3-7.5), and Rio Solimoes (6.3-7.5). Rivers of more lowland origin averaged ~5 to 7, with low levels (to pH 3) in the blackwatel'S of the Rio Negro. DOC is strongly and positively correlated with the stage of the hydrograph, trackingpCO z [cf. Moreira-Turcq et al., 2003]. The lowest increases in DOC concentrations of 3 to 6 mg L-1 were seen along the Rio Solimoes and Rio Madeira, whereas broader increases were observed in the Rio Negro and its tributaries to the blackwater rivers of the Rio Negro. The Rio Teles Pires and Rio Araguaia showed lower increases, ranging from 2 to 6 mg L-1. These general patterns are reflected in major and minor ions, nutrients, sediments and sediment composition.
496
ROLE OF RIVERS IN THE REGIONAL CARBON BALANCE
RICHEY ET AL. Solimoes
Madeira
497
Ji-Parana
40 30 20
0
10
30
8
~
6
...
4
::l
'"c:> 0
W
N
U
2
Q.
i :
.
:
i
i
!
f
2
..
~-~~------,--. ,~-
7.5
7.5
:t:
Mil i I i
6.5
6.5
5.5
5.5
5.5
15
15
6.5
Q.
Figure 3. Sites of the Rede Beija Rio sampling network.
~
oS
Overall, the relations between hydrograph and chemical species are consistent among sites. Concentrations of chemical species are maintained fi'om smaller tributaries to large rivers, which suggest a constancy of processes at work. The next step will be to analyze some of the key processes maintaining these patterns. 3.3. Gas Exchange
In considering gas exchange between the hydrosphere and the atmosphere, it is first necessary to understand the physical processes controlling the exchange of gases between water and air. Because C02 is often supersaturated in the waters of the Amazon relative to the atmosphere, there is an outgassing. The question then is, how much, and what controls it? Quantifying gas exchange correctly is a critical and difficult element in defining the overall carbon balance in fluvial systems. Gas exchange between water surfaces and the atmosphere is controlled by microscale water-side turbulence [Bane/jee and MacIntyre, 2004], and can be described by the simple equation, F C02 = k S (L'1pC0 2), where k is the gas transfer velocity, s is CO 2 solubility, and L'1pC0 2 represents the air-water CO 2 concentration gradient. In the fast-flowing river environments ofAmazonia, the turbulence controlling the parameter k is induced by a dynamic combination of river currents, wind, and rainfall. The parameter k is difficult to measure accurately and has been the subject
of different techniques. Eddy covariance has been utilized in marine systems and sparsely tested on large rivers (e.g., S. Miller, on the Amazon, unpublished data, 2004). However, it is not possible to deploy the open-air eddy covariance method in smaller-scale aquatic environments because of the contribution of the terrestrial environment to the gas flux signal. Injection of dual tracers has commonly been utilized on small streams, but the approach is not realistic across a spectrum of larger rivers. S. R. Alin et a1. (Environmental controls on carbon dioxide flux, transfer velocity, and partial pressure in the Amazon and Mekong river basins (Brazil, Southeast Asia), manuscript in preparation, 2009) used the eddy accumulation approach, using floating chambers attached to a portable CO 2 analyzer. Although this technique has been criticized [e.g., Belanger and Korzum, 1991], they have been found by numerous authors to provide consistent results under certain conditions, namely, low to moderate wind «8-10 m s-1) and wave conditions and when the air-water L'1pC02 is >300 Ilatm [Kremer et al., 2003; Borges et al., 2004]. These conditions are routinely met in Amazonian rivers, as winds rarely exceed 5 m s-), and air-water pC0 2 gradients are generally much steeper than 300 Ilatm, frequently exhibiting gradients on the order of thousands of microatmospheres. Guerin et al. [2007] got comparable results between the chambers and eddy covariance in lake environments, where both techniques could be simultaneously applied. Further, the floating
!
9
~
u 0 0
7.5
'
:
Ii!
3 '04
I
,
'05
'06
•
'07
".M\ ''A/ViVi~i~
,,
\ftJ\) '04
'05
'06
'07
'04
'05
'06
'07
Figure 4. The 2004-2007 discharge hydrographs (Q) and chemical hydrographs forpC0 2 , pH, and DOC, from the Rede Beija Rio sampling network, illustrated for the main stem (Solimoes at Manacapuru), the Rio Madeira at Porto Velho (tributary to the Amazon main stem), and the Ii-Parana (tributmy to the Madeira). Vertical dashed lines indicate hydrograph peak, for comparison to chemistry.
chamber approach can be consistently applied across a wide range of environments, whereas other approaches are more restricted. Bin-averaged data for lake, stream, and river environments show clear differences among these broad environment types, which correspond to significant differences in water turbulence regimes (Alin et aI., manuscript in preparation, 2009). Gas transfer velocities measured in streams and rivers are higher than lake and bay values due to the effects of water velocity and bed fi'iction on turbulence. In the Amazon and Mekong rivers, water velocities are frequently in the range of 100-300 cm s-), suggesting that observations of. elevated k values may be explained by the greater contribution of water current velocity to the turbulence controlling gas transfer. A consideration of equal importance to the exchange coefficient for regional evaluation of outgassing is the river sur-
face area exposed. Analyses to date [e.g., Richey et al., 2002] used satellite observations to define the river network, but that teclmology is applicable only to channels greater than 100 m in width. Considering that 92% of the Amazon River network [Mayorga et al., 2005] is composed of rivers with channels less than 100 m wide, these areas, although still not wellcmapped, must play an important role as CO 2 sources to the atmosphere. The surface area extent of small to medium size tropical rivers can have large variations through the hydrological cycle, affecting the size of the air-water interface and therefore gas evasion. Rasera et al. [2008] used a geographic information systems-based method to calculate the extent of river water for the Ji-Parana basin, a mesoscale river (75,400 km2) located in western Amazonia, then computed outgassing from these areas couple with local measurements of gas fluxes. CO 2 outgassing was the main carbon export pathway from the Ji-Parana, totaling 289 Gg C a-),
498
RICHEY ET AL.
ROLE OF RIVERS IN THE REGIONAL CARBON BALANCE
about 2.4 times the amount of carbon exported as DIC and forested Rio Negro soils export contemporary l3C-depleted 1.6 times the dissolved organic carbon export. Applying this CO 2 to streams, which strongly resembles the l3C signature model to the entire Amazon River network of channels less of C3 plants. In contrast, human-impacted pasture streams than 100 m wide (third to fifth order), Rasera et al. [2008] in Rondonia export young but highly l3C-em-iched C 0 2, calculated that the surface area of small rivers is 0.3 ± 0.05 resembling C4 vegetation. Finally, lowland watersheds in million km2, and it is potentially evading to the atmosphere western Amazonia appear to include significant carbonate 170 ± 42 Tg C a-I as CO 2. Therefore, these ecosystems play outcrops, leading to substantial exports of 14C-depleted and l3C-em-iched DIC from carbonate dissolution. While bulk an important role in the regional carbon balance. CO 2 is not the only form of gaseous carbon present in the dissolved and pac appear to be isotopically decoupled from rivers of Amazonia. Throughout the basin, redox conditions CO 2 in Amazonian rivers, they remain key components of favor the existence of methane [Devol et al., 1988; Bartlett river ecosystems and exports to the ocean. They are actively et al., 1990]. Melack et al. [2004] estimated that methane processed in the river system, albeit at slower rates. emissions accounted for another 6.8 Tg C a-I (±1.3 Tg C a-I) for the same quadrant of central Amazonia analyzed by 3.5. Factors Controlling Aquatic Respiration and Its Role Richey et al. [2002]. When extrapolated to the whole basin in Fueling CO 2 Outgassing area below the 500 m contour, this results in emissions of A distinguishing feature of Amazonian waters is the eleapproximately 22 Tg C a-I. It should be noted that, although vated concentration of pC02 relative to the atmosphere. The these figures might be modest for the carbon balance of Amcentral ecological question is what maintains pC0 2 superazonia, the photochemical properties of methane make it a saturation in these waters? In situ respiration of organic matfar more potent greenhouse gas, with roughly 20 times the ter derived from adjacent upland ecosystems [see Trumbore greenhouse wanning potential of C02. et al., this volume] and flushed into rivers is thought to be the primmy source of C02 saturation [Mayorga et al., 2005]. 3.4. Composite Tracers ojLandscape and in Situ Processes Total basin-wide inputs of DOC, FPOC, and CPOC are inA significant challenge is how to deconvolve signals of ex- adequate to support in situ oxidation occurring on the Amaternal sources from internal processes. Beyond its concentra- zon main stem by at least a factor of two, suggesting that tion, a molecule found in a parcel of water bears the imprint, an unmeasured pool of labile organic matter exists that is or signature, of its history. Ifthat signature can be interpreted, rapidly consumed [Richey et al., 1990]. As a contribution to important insight can be gained as to the biogeochemical our knowledge of how metabolism varies both spatially and factors that influenced that molecule. Using an extensive temporally across water types and of factors driving metabosurvey of carbon isotopes in organic and inorganic fractions lism, E. E. Ellis et al. (Factors controlling aquatic respiration throughout mountain and lowland rivers of the Amazon ba- and its role in fueling C02 gas evasion in rivers ofthe central sin, Mayorga et al. [2005] showed that the primmy source and southwestem Amazon Basin, submitted to Limnology of respired CO 2 in the lowlands was <5 years old on average and Oceanography, 2009) analyzed the spatial and temporal and that pC02 was commonly isotopically distinct from co- extent and the dynamics of in situ water-column metabolism, incident organic carbon fractions (DOC, FPOC, and CPOC). in rivers and streams of the central and southwestern Amazon Measured organic carbon fractions range in age from a de- basin. Using O 2 consumption for measuring respiration and the cade to thousands of years and are evelywhere depleted in stable oxygen isotopes ratios of dissolved oxygen to estil3C relative to pC02. The isotopic composition of outgassed mate the ratio of gross primary production (GPP) to respiraC02 may not reflect in situ respiration sources evelywhere tory uptake of dissolved oxygen (P:R) (following Quay et al. and varies significantly among regions, with progressive evo[1995]), Ellis et al. (submitted manuscript, 2009) found that lution of carbon isotopic values downstream, as aged pC 0 2 respiration rates spanned three orders of magnitude, with fi'om upstream weathering sources is outgassed. Elemental, values from 0.034 to 1.78 ~mol O2 L- I h- I . isotopic, and biochemical compositions of riverine organic Oxygen isotopes indicated that some sites are net hetmatter evolve from Andean source waters to large lowland erotrophic, while gross photosynthesis exceeds respirarivers, providing further evidence for upstream to down18 tion at other sites. Minimum and maximum values of 0 0 stream changes in composition, supporting the importance of sorptive processes implicated in previous experimental values ranged from 20.8%0 (in a small tributmy of the Rio studies in natural systems [Aujdenkampe et al., 200 I, 2007]. Acre) to 27.6%0 (in the main stem Rio Solimoes), with a Regional heterogeneity is observed in the isotopic composi- fi'action of dissolved oxygen saturation of 1.00 and 0.54, tion of pC02 exported fi'om upland soils to streams. Sandy, respectively. The ratio of gross photosynthesis:respiration
(P:R) ranged from 0.38 in the Solimoes River to 1.1 in the Rio PUrllS in the st~~e of Acre. GPP ranged from 0.02 to 1.5 ~llnol O 2 L -I h~~:'These ratios (P:R) are higher than those repOlied previou&jly, which ranged fi'om 0.26 to 0.67 and were primarily,J;llected along the Amazon main stem and in the mouths of major tributaries [Quay et al., 1995]. What leads to, such variations? In a dual isotope study (O l3 C and the ,i\.14C), Mayorga et al. [2005] suggested that a young (less than 5 years old), isotopically enriched pool of OC is fueling respiration in lowland Amazonian rivers, compared to the bulk size fi·actions. Ellis et al. (submitted manuscript, 2009) investigated the relationship between water-column respiration rates and environmental variables. Out of the bulk size fractions studied, respiration rates were most positively correlated with FPOC. In terms of size fractions of carbon, respiration rates were also correlated with the percentage (but not the concentration) of low molecular weight (LMW) DOC «5 kDa). They were not correlated with the concentration of DOC or CPOC. The lability of FPOC is consistent with the decrease in the 13C composition of the FPOC fi'om the Andes (where most sediment originates), to the lowlands, where it is identicaL to that of lowland-derived OM [Mayorga et a1., 2005; Quay et al., 1995]. This indicates almost complete remineralization of Andean-derived FPOC and replacement with lowlandderived organic carbon, likely through preferential sorption of fresh organic matter [Aujdenkampe et al., 2001]. Respiration is not controlled solely by the bioavailability ofFPOC. Ellis et a!. (submitted manuscript, 2009) found that pH was highly cOlTelated with respiration, with pH ranging from 4.6 to 8.6 across the rivers studied. pH may indirectly affect respiration rates by controlling bacterial abundance. Bacterial abundance alone explained 78% of the variation in respiration rates, but the relationship between respiration rates and abundance was no longer significant once the effects of pH had been controlled for. FUlihermore, the relationship between respiration rates and LMW DOC was no longer significant after accounting for pH. While phytoplankton production is well documented in floodplain waters [e.g., Novo et al., 2006], its role as a carbon source is generally considered to be minimal in rivers ofthe Amazon basin. However, Ellis et a!. (submitted manuscript, 2009) found that in situ photosynthetic production calculated from the 018 0 of dissolved oxygen at sites with pH >7 during low water contributed to the high respiration rates. These sites were more saturated with dissolved. oxygen and had a significantly greater ratio of P:R than low pH sites or measurements reported by Quay et al. [1995] along the Amazon main stem and in the mouths of major tributaries [Quay et al., 1995]. The reason for the increased productivity of these high pH rivers is that they were all
499
tributaries of the Rio Pums that were sampled at low water during the study period. Because the average depth was only 1.3 m, these rivers were likely not limited by light, enabling planktonic autochthonous production to occur. These results demonstrate that autochthonous material can be a significant source of labile carbon during the low-water period in upstream tributaries. Thus, algal material is a labile substrate that contributes to the high respiration rates observed in some rivers. The question then remains how far downstream an algal signal might persist. The organic matter sources available to fuel microbial respiration throughout the basin consist of material ultimately derived from C3 plants, C4 macrophytes, and algae. Ellis et a!. (submitted manuscript, 2009) examined the Ol3 C ofrespired CO 2 to determine the mix of sources being oxidized at anyone time and place. They found that respiration-derived CO 2 in the Rio Negro and two small, shaded, streams in Acre were consistent with carbon originating from C3 plants (O l3 C of respired C02 ranged between -28.3%0 and-30.l %0). The Ol3 C of respired CO 2 in shallow Purus tributaries (with high P:R ratios) were consistent with the oxidation of both algal and C3 plant sources: the Ol3 C was -33.2%0 and -31.2%0, in the Acre and Pums rivers, respectively. Downstream, the Solimoes main stem demonstrates considerable temporal variability in the of'ganic source fueling respiration. During early falling water, organic matter with a 0l3C of -32.6%0 is fueling respirationi This suggests that tissues derived from I both C3 plant material and algae are respired at this site. Because the oxygen isotope data from the Solimoes provides little evidence of autochthonous production, algal material produced in fringing floodplains and upstream tributaries (which were at low water) likely contributes to respiration. However, during early rising water, the material being respired was -22.9%0, consistent with the results of Quay et al. [1995] during the low/early rising water stage. These results are consistent with 34-40% of the respiration fueled by C4 aquatic macrophytes and the remainder coming from C3 plants. Given that riverine respiration has been hypothesized to be the primary source of CO 2 saturation in fi'eshwater, how much ofthe C02 evasion flux can be accounted for by watercolumn respiration? Depending on the size of the river and the water type (e.g., black- versus whitewater rivers), Ellis et a!. (submitted manuscript, 2009) suggest that depth-integrated respiration rates account for less than 1% to as much as 100% ofthe outgassing flux. The outgassing flux is largely supported by sources other than water-column respiration in small streams (the depth-integrated respiration rate is less tha'n 4% of the outgassing flux for all streams), most likely benthic respiration. The contribution of water-column respiration to outgassing appears to be affected by water type
500
RICHEY ET AL.
ROLE OF RIVERS IN THE REGIONAL CARBON BALANCE
in the large, lowland rivers of Amazonia. Between 66% and 128% of outgassing C02 is accounted for by water-colmllil respiration in large white-water rivers (Amazon, Madeira, and Solimoes;; whose gas evasion fluxes ranged from 0.8 to 3.7 g C m-2 d- l . In the Rio Negro, the areal respiration flux accounted for only between 15% and 34% ofthe outgassing flux. Additional sources of CO 2 production in blackwater rivers could be due to photomineralization (the complete oxidation of organic carbon to C02 by light). Remington [2008] estimated that photominei'alization was on the order of7-8%, but cautioned that this percentage is likely too low due to methodological reasons. The Amazon River has been proposed to be in a "dynamic equilibrium," or quasi-steady state, with respect to CO2 and 02 [Devol et al., 1988; Quay et al., 1995], based on evidence that inputs of CO 2 via respiration are balanced by the rate of outgassing CO 2 [Devol et al., 1988]. Furthermore, a dynamic equilibrium with respect to (5 l3 C has been hypothesized as the 13 (5 l3 C ofrespired CO 2, has been shown to equal the (5 C of outgassing C02 of the Amazon River [Quay et al., 1995]. Consequently, other studies have assumed that the l3C of C02 gas in solution in carbonate-free lowlands is equal to that of respired C02 without measuring the (5 13 C of respired CO 2 [Mayorga et al., 2005]. However, Ellis et al. (submitted manuscript, 2009) found that the (5 l3 C of respired C02 was not equal to that of dissolved free CO 2 at some of the sites. Potential explanations for this discrepancy vmy between sites, but it is generally due to the long equilibration time for (5 l3 C, long turnover times of DIC by respiration, gas exchange, and alternative sources of CO 2 production. Overall, it would seem that these isotopic signals measured to date are likely reflective of the transition to a steady state value of -28%0 starting from the isotopic signature set in the Andes and modified by respiration of organic matter derived fi'om C3 vegetation, with C4 vegetation becoming more important in the lowlands at celiain stages of the hydrograph. Overall, Ellis et al. (submitted manuscript, 2009) demonstrate that no single organic matter source consistently fuels respiration; instead, the (5 l3 C of respiration-derived CO 2 varies with time and space. In most cases, the respired carbon is isotopically similar to the bulk carbon, contrmy to that reported by Mayorga et al. [2005]. The respiration of organic matter fi'om adjacent terrestrial ecosystems is hypothesized to be the primmy source sustaining C02 satmation in Amazonia [Richey et al., 2002]. However, this research has provided evidence that, in some cases, C4 macrophytes and algae are being respired in addition to terrestrially derived C3 sources. The role of macrophytes in fueling respiration changes seasonally and photosynthetic production occurs in shallow whitewater rivers during low water. Therefore, it is necessary to measure both temporal and spatial changes
of sources of organic matter that are fueling respiration in Amazonia and further identify alternative sources of CO 2 production in order to accurately resolve the terrestrial carbon budget of Amazonia. 3.6. Photooxidation
Net ecosystem production is typically thought of as representing the net (biological) fluxes of 02 and pC02. However, another process may also contribute under certain conditions: photooxidation and its metabolites or breakdown products [Amon and Benner, 1996]. As past research suggests that bulle organic carbon in these large rivers is largely unavailable for bacterial consumption during transpOli [Ertel et al., 1986; Hedges et al., 1986, 1994], a small, rapidly turning-over, pool may be responsible [Richey et al., 1990]. Low molecular weight organic acids (LMWOAs) are a rapidly cycling, little-studied pool of biologically labile organic compounds [Kaplan and Newbold, 2003] and are produced by photochemical degradation of aquatic humic substances [Miller and Moran, 1997]. To examine the possible role of photooxidation in producing these compounds, Remington [2008] measured photochemical production rates of DIC and two LMWOAs (acetic and formic acid) from DOM in the Solimoes and Negro rivers. Depth-integrated photochemical DIC flux was 25% of the measmed CO 2 flux (Alin et al., manuscript in preparation, 2009) in the Rio Negro, whereas the combined acetic and formic acid production rate was 4% of the C02 flux. It should be noted that this rate is for depth integration (50 m); volumetric surface rates were much higher and could be considered as more representative of shallower rivers. There was no statistically significant production of any of these compounds from DOM in the Rio Solimoes. Based on these data, Remington [2008] hypothesized that photolytic production of DIC, acetic and formic acids, and other biologically labile compounds are significant CO 2 sources to river channels with humic-rich and clear, low-sediment waters ofthe Amazon basin. Photochemical production of these and other biolabile compounds may become even more significant at smaller scales with open canopies and shallower water. 4. EFFECTS OF ANTHROPOGENIC DISTURBANCE ON RIVER CARBON FLUXES The clearest evidence of the consequences of land cover and land use changes in the biogeochemistly of carbon in the rivers of Amazonia come from studies conducted in small basins «100 km2). In streams of the Nova Vida Farm, in central Rondonia, conversion of forests into pastures altered the functioning of these systems [see Tomasella et al., this
volume; Thomas et al., 2004; Neill et al., 2006]. The absence offorest canopy cov:Ffor the pastme watershed allowed extensive growth of naturally occurring Paspalllllil grasses on the margins and \6side stream channels. The resulting increase in organ)/ matter loading promoted higher respiration rates, resulting in increased CO 2 evasion and a shift of the oxic conditions observed in the forest streams to almost anoxia in the pastme ones [Neill et al., 2006]. The isotopic composition (5I3C) of both particulate and dissolved organic fractions in the forest and pastme streams reflected the predominance of C3 and C4 plants as sources for the riverine carbon in forest and pasture catchments, respectively. However, downstream from the confluence of these first-order streams that have only one vegetation type (e.g., forest or pasture), and where the larger watershed presents patches of forest within the predominant pasture cover, these isotopic signatures resemble again those of the forest stream. In the larger Rio Ji-Parana, where the tributaries Rolim de Moura and Urupa have more than 50% oftheir basins covered with pastmes [Ballester et al., 2003], a similar pattern is found, and both particulate and dissolved forms of carbon resemble the isotopic signature of soils, which cany a C3 signal [Bernardes et al., 2004]. Other evidence of the consequences of forest to pasture conversion was obtained in a detailed study offlow paths and elemental fluxes in zero-order streams in Rondonia (Rancho Grande). At this scale, a small fraction of total precipitation appears in stream flow (0.8% in the forest and 17% in the pasture), but most of the increased flow in the pastme appears as overland flow (60% of total sources), whereas in the forest, this flow path was negligible [Chaves et al., 2009]. Mainly as a result of this increased overland flow, pasture streams exported on an annual basis almost 20 times more dissolved organic carbon than forests (from 4.23 to 72.61 kg ha- l ). The analysis of rain events at this site, occurring from the beginning to the middle of the rainy season, showed that the forest retains most of the DOC entering the system from precipitation. In the pastme, however, net losses of carbon to the streams occurred throughout the year with the exception of the beginning of the rainy season [Gouveia-Neto, 2006]. At the larger scale, i.e., the Ji-Parana river basin, higher DOC concentrations were related to higher values of total suspended sediments (TSS) during the wet season [Ballester et al., 2003]. TSS originated in pasture areas, where soil compaction led to less infiltration and higher surface runoff, . promoting leaching of soil superficial layers carrying more DOC to the stream, while areas covered by forest had higher infiltration deeper into the soils. The percentage of the basin area covered by pasture was a good predictor for DOC concentrations during the wet season.
501
5. RELATION OF RIVER FLUXES TO REGIONAL CARBON BALANCES The Richey et al. [2002] estimate of outgassing of 1.2 ± 0.3 Mg C ha- l a-I 11'0m Amazonian rivers and wetlands offered arguments to complement early and intriguing results of the LBA experiment, suggesting that forests were large sinks of carbon [Malhi and Grace, 2000]. Since there was no evidences of carbon accumulation in the basin to support uptake rates of that rnagnitude, the export in riverine systems would comprise an unknown (at the time) but significant pathway ofthe regional carbon cycle, as a source to the atmosphere equivalent to lower estimates of forest sequestration. At the same time, this would imply large transfers of carbon from land to water, with outgassing over 10 times the fluvial export of organic carbon to the ocean. Our understanding since then has increased. First, how accurate are the nurnbers? The Richey et al. [2002] estimate was based prirnarily on data fi'om the Amazon main stem and floodplain, and mouths of major tributaries. Data collected since then allows us to re-evaluate this number. As reported by Alin et al. (manuscript in preparation, 2009), more extensive measurements of gas exchange showed that for the large rivers, fluxes were comparable to the Richey et al. [2002] result$, but gas transfer velocities measured in rivers less than 100111 wide were considerably higher. Rasera et al. [2008] comptlted that outgassing fi'om rivers <100 m was more than twi6e that repOlied by Richey et al. [2002]. Johnson et al. [2008] estimated that CO 2 outgassing of terresh'ially respired carbon fi'om headwater stl'eams represents a carbon flux to the atmosphere of more than 100 Tg C a-I for the Amazon basin, which is decoupled from and in addition to CO 2 evasion from larger rivers and wetlands. We now believe that CO 2 outgassing from the hydrosphere to the atmosphere varies locally with current velocity, water depth, and time of day. Cumulatively, CO2 evasion tenllS exceed the value of 1.2 Mg C ha- 1 a-I and represent a land-water-atlllosphere transfer comparable to that of terrestl'ial sequestration. The heterotrophic nature (R> P) ofthe waters ofthe Amazon basin and the areal extent offlooding making these systems a significant source of carbon to the atmosphere raises critical ecological questions, such as where does the carbon come from that ultimately is evaded, and how are carbon sources affected by land use change and climatic variability? From estimates of potential source strengths, we hypothesized that evasion is driven primarily by in-stream respiration of organic carbon fixed originally on land and to a lesser extent along river margins and mobilized into flowing waters and not just by dissolved CO 2 in groundwater. This presents us with a problem. Despite negligible measured downstream gradients in dissolved and particulate organic
502
ROLE OF RIVERS IN THE REGIONAL CARBON BALANCE
matter and constituent biochemicals that appear, in general, quite refractory, measured respiration rates in Amazonian waters are sufficiently large to recycle essentially all organic matter in a pit'cel of river water well before it reaches the ocean [Richey et al., 1990; Hedges et al., 2000]. The magnitude of and physical controls on gas transfer velocity appear to operate similarly across river basins, but to vary substantially with physical scale. Important transitions in the environmental controls on gas exchange appear to occur with the physical scale of tl~e river channel, as well as between standing and running water environments. Furthermore, the large variability seen in the small rivers highlights the critical importance of collecting numerous spatially distributed measurements of gas exchange variables rather than going to great effort to hone values at a few study sites that mayor may not represent the drainage network as a whole. An argument against the interpretation of the outgassing as a "separate" flux is that the outgassing is detected by the eddy covariance towers and hence was already included in the respiration term of the forest. The use of eddy covariance measurements to evaluate net ecosystem exchange of CO 2 and ecosystem respiration in the tall-stature, tropical forest is more problematic than originally thought, and is a topic of active research [see Saleska et al., this volume, 2005; Acevedo et al., 2007; Hutyra et al., 2007; Malhi and Aragao, 2007]. It is most likely that these towers do not detect any of the fluvial signals, including those of smaller streams and seeps. The fluvial fluxes, while small relative to photosynthesis and ecosystem respiration, are large relative to the magnitude of net ecosystem exchange ascribed to the forest using eddy covariance measurements. Hence carbon processing through the fluvial systems of the Amazon basin indicates the strength of land-water coupling and, ultimately, the importance offluvial systems in the regional carbon budget of the tropics. Aclmowledgments. We thank especially Reynaldo Victoria (CENA, SP) and the members of the Rede Beija Rio sampling network, Cleber Salimon (UFAC, AC), Beatriz Gomes (UNIR, RO), Kelli Munhoz (FAPEMAT, MT), Laura Borma (UFT, TO), Maria Emilia Sales (MPEG, PA), Roosevelt Barbosa (ESASGC, AM) Alexandra Montebelo, and Gustavo Baldi (CENA, SP), and the numerous students and employees who helped in the field. The research reported has been supported by NASA LBA, CNPq, FAPESP, and NSF grants. This is CAMREX Publication 145.
REFERENCES Acevedo, O. C., O. Moraes, D. R. Fitzjarrald, R K. Sakai, and L. Mahrt (2007), Turbulent carbon exchange in very stable conditions, BoundGl)1 Layer Meteorol., 125,49-61.
Ahn, C. H., and R. Tateishi (1994), Monthly potential and actual evapotranspiration and water balance, United Nations Environment Programme/Global Resource Information Database, Date/set GNV183. Amon, R. M. W., and R. Benner (1996), Bacterial utilization of different size classes of dissolved organic matter, Limnol. Oceanogr., 41, 51. Aufdenkampe, A K., J. 1. Hedges, J. E. Richey, A V. Kmsche, and C. Llerena (2001), Sorptive fractionation of dissolved organic nitrogen and amino acids onto fine sediments within the Amazon Basin, Limnol. Oceanogr., 46(8),1921-1935. Aufdenkampe, A K., E. Mayorga, J. 1. Hedges, C. Llerena, P. D. Quay, J. Gudeman, A. V. Krusche, and J. E. Richey (2007), Organic matter in the Peruvian headwaters of the Amazon: A comparison to Bolivian tributaries and the lowland Amazon mainstem,Org. Geochem., 38, 337-364. Ballester, M. V. R., D. Victoria, A. V. Krusche, R. Coburn, R. L. Victoria, J. E. Richey, M. G. Logsdon, E. Mayorga, and E. Matricardi (2003), Land use/cover of the Ji-Parami river basin: Building a GIS-based physical template to support the understanding ofthe biogeochemistly of surface waters in a meso-scale river in Western Amazonia, Remote Sens. Environ., 87, 429-445. Banerjee, S, and S. MacIntyre (2004), The air water interface: Turbulence and scalar exchange.' Chapter 5, in PIV and Water Waves, edited by J. Grue, P. Liu, and G. Pedersen, World Scientific Press. Bartlett, K. B., P. M. Crill, J. A Bonassi, J. E. Richey, and R. C. Harriss (1990), Methane flux from the Amazon River floodplain: Emissions dming rising water, J. Geophys. Res., 95, 16,77316,788. Batjes, N. H. (2005), ISRIC-WISE global data set of derived soil properties on a 0.5 by 0.5 degree grid (Version 3.0), /SRICWorld Soil b1!ormation, Wageningen. Belanger, T. V., and E. A Korzun (1991), Critique of floatingdome technique for estimating reaeration rate, J. Environ. Eng., 117,144-150. Bernardes, M. C., et al. (2004), Riverine organic matter composition as a function ofland use changes, Southwest Amazon, Ecol. Appl., 14, S263-S279. BilIen, G., C. Lancelot, and M. Meybeck (1991), N, P, and Si retention along the aquatic continuum from land to ocean, in Ocean Margin Processes in Global Change, edited by R. F. C. Mantoura, J.-M. Martin, andR. Wollast, pp. 19-44, Jolm Wiley, New York. Borges, A. V., et al. (2004), Gas transfer velocities of CO 2 in three European estuaries (Randel'S Fjord, ScheIdt, and Thames), Limnol. Oceanogr., 49, 1630-1641. Brady, N. C., and R. R. Weil (1999), The Nature and Properties of Soils, 881 pp., Prentice-Hall, Upper Saddle River, N. J. Chaves, J., C. Neill, H. Elsenbeer, A. I(rusche, S. Germer, and S. Gouveia Neto (2009), Magnitude and origin of channel flows in Amazon forest and pasture watersheds, Hydrol. Processes, in press. Cole, J. J., et al. (2007), Plumbing the global carbon cycle: Integrating inland waters into the terrestrial carbon budget, Ecosystems, 10,171-184.
RICHEY ET AL.
503
Costa, M. H., and J. A Foley (1999), Trends in the hydrologic cycle Johnson, M. S., M. F. Billett, K. J. Dinsmore, M. Wallin, K. E. Dyson, ofthe Amazon basin, .l;.Geophys. Res., 104(Dl2), 14,189-14,198. and R S. Jassal (2009), Direct and continuous measurement of Davidson, E. A, and S,(E. Trumbore (1995), Gas diffusivity and dissolved carbon dioxide in freshwater aquatic systems-Method production of CO:q;ln deep soils of the eastern Amazon, Tellus, and applications, Ecohydrology, in press. Ser. B, 47(5), 55gl565. Kaplan, L. A, and J. D. Newbold (2003), The role of monomers Davidson, E. A, F. Y. Ishida, and D. C. Nepstad (2004), Effects in stream ecosystem metabolism, in Aquatic Ecosystems: /nof an experimental drought on soil emissions of carbon dioxide, teractivity o.lDissolved Organic Matter, pp. 97-113, Elsevier, methane, nitrous oxide, and nitric oxide in a moist tropical forLondon. est, Global Change Bio!., 10, 718-730. Karlsson, G., A Grimvall, and M. Lowgren (1988), River basin Degens, E. T., S. Kempe, and J. E. Richey (1991), Summaly: Bioperspective on long-term changes in the transport of nitrogen geochemistly of major world rivers, in Biogeochemistl)1 ofMaand phosphorus, Water Res., 22, 139-149. jor World Rivers, edited by E. T. Degens, S. Kempe, and J. E. Kessler, T. J., and C. F. Harvey (2001), The global flux of carbon Richey, pp. 323-347, John Wiley. dioxide into groundwater, Geophys. Res. Lett., 28(2), 279-282. Devol, A H., J. E. Richey, W. A. Clark, S. L. King, and L. A. Mar- Kremer, J. N., et al. (2003), Technical note: Conditions for using tinelli (1988), Methane emissions to the troposphere from the the floating chamber method to estimate air-water gas exchange, Amazon floodplain, J. Geophys. Res., 95, 1583-1592. Estuaries, 26, 985-990. Ertel, J. R., J. 1. Hedges, A. H. Devol, and J. E. Richey (1986), Dis- Krusche, A V., et al. [2009], 0 papel dos ciclos evasivos de CO 2 solved humic substances of the Amazon River system, Limnol. de rios da Amazonia no balan90 regional e global de carbono, Oceanogr., 31, 739-754. FAPESP Thematic Project Final Report, 16 pp., CENAlUSP, Gouveia-Neto, S. C. (2006), Concentracoes e balancos de Carbono Piracicaba. organico dissolvido em duas bacias do estado de Rondonia: Uma Leopold, L. B., M. G. Wolman, and J. P. Miller (1964), Fluvial Proccomparacao entre floresta e pastagem, M.S. thesis, 55 pp., Uniesses in GeomOJphologJl, 2nd ed., 522 pp., Dover, New York. versity of Sao Paulo. Malhi, Y., and L. Aragao (2007), Internal carbon dynamics of Guerin, F., G. Abril, D. Serca, C. Delon, S. Richard, R. Delmas, Amazonian forest systems, Large-Scale Biosphere-Atmosphere A. Tremblay, and L. Varfalvy (2007), Gas transfer velocities of in Amazonia (LBA-Eco) 11th Science Team Meeting, Salvador, CO 2 and CH4 in a tropical reservoir and its river downstream, Brazil. J. Mal'. Syst., 66,161-172. Malhi, Y., and J. Grace (2000), Tropical forests and atmospheric Hedges, J. 1., J. R. Ertel, P. D. Quay, P. M. Grootes, J. E. Richey, carbon dioxide, TREE, 15, 333-337. A H. Devol, G. W. Farwell, F. W. Schmidt, and E. Salati (1986), Mayorga, E., A. K. Aufdenkampe, C. A Masiello, A V. Krusche, Organic carbon-14 in the Amazon River system, Science, 231, J. 1. Hedges, P. D. Quay, J. E. Richey, and T. A. Brown (2005), 1129-1131. Young organic ma~ter as a source of carbon dioxide outgassing Hedges, J. 1., G. L. Cowie, J. E. Richey, P. D. Quay, R Benner, and from Amazonian rivers, Nature, 436, 538-541. M. Strom (1994), Origins and processing of organic matter in the McBride, M. B. (1994), Environmental Chemistl)1 o/Soils, 416 pp., Amazon River as indicated by carbohydrates and amino acids, Oxford Univ. Press, Oxford. Limnol. Oceanogr., 39, 743-761. McClain, M. E., and H. Elsenbeer (2001), Terrestrial inputs to Hedges, J. 1., et al. (2000), Organic matter in Bolivian tributaries of Amazon streams and internal biogeochemical processing, in The the Amazon River: A comparison to the lower mainstem, LimBiogeochemistl)1 o.lthe Amazon Basin, edited by M. E. McClain, nol. Oceanogr., 45, 1449-1466. R. L. Victoria, and J. E. Richey, pp. 185-208, Oxford Univ. Hutyra, L. R, et al. (2007), Resolving systematic errors in estimates Press, Oxford. of net ecosystem exchange of CO2 and ecosystem respiration McClain, M. E., J. E. Richey, J. A Brandes, and T. P. Pimentel in a tall-stature forest: Applications to a tropical forest biome, (1997), Dissolved organic matter and terrestrial-Iotic linkages in Large-Scale Biosphere-Atmosphere in Amazonia (LBA-Eco) the central Amazon basin of Brazil, Global Biogeochem. Cycles, 11 th Science Team Meeting, Salvador, Brazil. 11,295-311. Jolmson, M. S., J. Lehmann, S. J. Riha, J. P. Novaes Filho, and Melack, J. M., and L. L. Hess (2009), Remote sensing of the distriE. G. Couto (2006), DOC and DIC in flowpaths of Amazonian bution and extent of wetlands in the Amazon basin, in Amazoheadwater catchments with hydrologically contrasting soils, Bionian Floodplain Forests: EcophysiologJl, EcologJl, Biodiversity geochemisll)l, 81,45-57. and Sustainable Management, edited by W. J. Junk, and M. PieJohnson, M. S., M. Weiler, E. G. Couto, S. J. Riha, and J. Lehmal1l1 dade, Springer, New York, in press. (2007), Storm pulses of dissolved CO 2 in a forested headwater Melack, J. M., L. Hess, M. A Gastil, B. R. Forsberg, S. K. Ham" Amazonian stream explored using hydrograph separation, Water ilton, 1. B. T. Lima, and E. Novo (2004), Regionalization of Resow'. Res., 43, W11201, doi: 1O.1029/2007WR006359. methane emissions in the Amazon Basin with microwave remote Johnson, M. S., J. Lehmalm, S. J. Riha, A V. Krusche, J. E. Richey, sensing, Global Change Bioi., 10, 1-15, doi:1O.1l111j.1529J. P. H. B. Ometto, and E. G. Couto (2008), CO 2 efflux from Am8817.2003.00763.x. azonian headwater streams represents a significant fate for deep Melack, J. M., E. M. L. M. Novo, B. R. Forsberg, M. T. F. Piedade, soil respiration, Geophys. Res. Lett., 35, L17401, doi:l0.l029/ and L. Mamice (2009), Floodplain ecosystem processes, Geo2008GL034619. phys. Monogr. Ser., doi: 10. 1029/2008GM00072 I, this volume.
504
ROLE OF RlVERS IN THE REGIONAL CARBON BALANCE
Meyer, 1. L., W. McDowell, T. Bott, J. W. Elwood, C. Ishizaki, J. M. Melack, B. L. Peckarsky, B. J. Peterson, and P. A. Rublee (1988), Elemental dynamics in streams, 1. North Am. Benthol. Soc., 7, 410--4'32. Miller, W. L., and M. A. Moran (1997), Interaction ofphotochemical and microbial processes in the degradation of refractOly dissolved organic matter from a coastal marine environment, Limnol. Opeanogr., 42,1317-1324. Moreira-Turcq, P., P. Seyler, 1. L. Guyot, and H. Etcheber (2003), Exportation of organic carbon from the Amazon River and its main tributaries, Hydrol. Processes', 17, 1329-1344, doi: 10.1002/ hyp.l287. Neill, C. ,L. A. Deegan, S. M. Thomas, C. L. Haupert, A. V. Krusche, V. M. Ballester, and R. L. Victoria (2006), Deforestation alters the hydraulic and biogeochemical characteristics of small lowland Amazonian streams, Hydrol. Processes, 320, 25632580. New, M. G., M. Hulme, and P. D. Jones (1999), Representing 20th centmy space-time climate variability. I: Development of a 1961-1990 mean monthly terrestrial climatology, J. Clim., 12, 829-856. Novo, E. M. L. M., C. C. F. Barbosa, R. M. Freitas, Y. E. Shimabukuro, 1. M. Melack, and W. P. Filho (2006), Seasonal changes in chlorophyll distributions in Amazon floodplain lakes derived from MODIS images, Limnology, 7, 153-161, doi: 10.1 007/s1020 1-006-0179-8. Ometto,1. P., A. D. Nobre, H. R. Rocha, P. Aliaxo, and L. A. Martinelli (2005), Amazonia and the modern carbon cycle: Lessons learned, Oecologia, 143,483-500. Quay, P. D., D. O. Wilbur, 1. E. Richey, A. H. Devol, R. Benner, and B. R. Forsberg (1995), The 18 0: 160 of dissolved oxygen in rivers and lakes in the Amazon Basin: Determining the ratio of respiration to photosynthesis rates in freshwaters, Limnol. Oceanogr., 40, 718-729. Raich, 1. W., and C. S. Potter (1995), Global patterns ofcarbon dioxide emissions from soils, Global Biogeochem. Cycles, 9, 23-36. Rasera, M., M. V. R. Ballester, A. V. KlUsche, C. Salimon, L. A. Montebelo, S. R. Alin, R. L. Victoria, and J. E. Richey (2008), Estimating the surface area of small rivers in the southwestern Amazon and their role in CO 2 outgassing, Earth Interact., 12, 1-16, doi:IO.1175/2008EI257.1. Remington, S. M. (2008), Sources and fate of dissolved organic matter in the Amazon River basin, Ph.D. thesis, 123 pp., Univ. of Washington. Remington, S. M., B. D. Strahm, V. Neu, J. E. Richey, and H. Brandao (2007), The role of sorption in control of riverine DOC concentrations by riparian zone soils in the Amazon basin, Soil Sci., 172,279-291. Richey, J. E. (2004), Pathways of atmospheric CO 2 through fluvial systems, in The Global Carbon Cycle, TO'ward CO 2 Stabilization: Issues, Strategies, and Consequences, A SCOPEIGCP Rapid Assessment Project, edited by C. Fields, and M. R. Raupach, pp. 329-340, Island Press, Washington, D. C. Richey, 1. E., 1. 1. Hedges, A. H. Devol, P. D. Quay, R. Victoria, L. Matiinelli, and B. R. Forsberg (1990), Biogeochemistry of carbon in the Amazon River, Limnol. Oceanogr., 35, 352-371.
Richey, 1. E., S. R. Wilhelm, M. E. Mcclain, R. L. Victoria, 1. M. Melack, and C. Araujo-Lima (1997), Organic matter and nutrient dynamics in river corridors of the Amazon Basin and their response to anthropogenic change, Cienc. Cult., 49, 98-110. Richey, 1. E., J. M. Melack, A. K. Aufdenkampe, V. M. Ballester, and L. L. Hess (2002), Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric CO 2, Nature, 416(6881),617-620. Richey, 1., R. Victoria, E. Mayorga, L. Matiinelli, and R. Meade (2004), Integrated analysis in a humid tropical region-The Amazon Basin, in Vegetation, Water, Humans, and the Climate, pp. 415-428, Springer, Berlin. Saleska, S., H. da Rocha, B. Krllijt, and A. Nobre (2009), Ecosystem carbon fluxes and Amazon forest metabolism, Geophys. Monogr. Ser., doi:10.1029/2008GM000728, this volume. Sarmiento, 1. L., and E. T. Sundquist (1992), Revised budget for the oceanic uptake of anthropogenic carbon-dioxide, Nature, 356, 589-593. Selva, E. C., E. G. Couto, M. S. Johnson, and 1. Lehmann (2007), Litterfall production and fluvial expOli in headwater catchments of the southern Amazon, 1. Trop. Ecol., 23, 329-335. Stallard, R. F. (1998), Terrestrial sedimentation and the carbon cycle: Coupling weathering and erosion to carbon burial, Global Biogeochem. Cycles, 12, 231-257. Thomas, S. M., C. Neill, L. A. Deegan, A. V. Krusche, R. Victoria, and M. V. Ballester (2004), Influences of land use and stream size on particulate and dissolved materials in a small Amazonian stream network, Biogeochemistl)l, 68, 135-151. Tomasella, J., C. Neill, R. Figueiredo, and A. D. Nobre (2009), Water and chemical budgets at the catchment scale including nutrient exports fi'om intact forests and disturbed landscapes, Geophys. Monogr. Ser., doi:10.l029/2008GM000727, this volume. Tmmbore, S., and P. B. de Camargo (2009), Soil carbon dynamics, Geophys. Monogr. Ser., doi:IO.1029/2008GM000741, this volume. Waterloo, M. J., et al. (2006), Export of organic carbon in run-off from an Amazonian rainforest blackwater catchment, Hydrol. Processes, 20, 2581-2597. Williams, M. R., T. R. Fisher, and 1. M. Melack (1997), Solute dynmnics in soil water and groundwater in a central Amazon catchment undergoing deforestation, Biogeochemisl7y, 38(3), 303-335.
M. V. Ballester and A. V. KlUsche, Laborat6rio de Analise Ambiental e Geoprocessamento, Centro de Energia Nuclear na Agcricultura, Piracicaba, SP 13400-970, Brazil. ([email protected]) H. B. da Cunha, Coordenyao de Pesquisas em Clima e Recursos Hidricos, Instituto Nacional de Pesquisas da Amazonia, Manaus, AM CEP 69060-001, Brazil. M. S. Johnson, Institute for Resources, Environment and Sustainability and Department of Ealih and Ocean Sciences, University of British Columbia, Vancouver, BC, Canada V6T lZ4. 1. E. Richey, School of Oceanography, University of Washington, Seattle, WA 98195, USA.
Water and Chemical Budgets at the Catchment Scale Including Nutrient Exports FrOlll Intact Forests and Disturbed Landscapes Javier Tomasella,l Christopher Neill,2 Ricardo Figueiredo,3 and Antonio D. Nobre 4 The objective of this chapter is to summarize current understanding of the hydrological function and nutrient dynamics of Amazonian forest derived from work in microcatchments and how these processes are affected by land use and land cover changes, mainly the conversion offorest to pasture. Our conclusions are based on field observations in catchments located in different regions ofAmazonia. This chapter is divided into sections that provide (I) a general overview of small catchment research in LBA and then address (2) mnoff and water budgets, (3) the influences of soil, vegetation, and riparian zones on stream chemistty and element budgets, and (4) the potential influence of catchment scale on the hydrological and biogeochemical processes thafcontrol water and element budgets. The first section provides a background on the principle sites where microcatchments have been studied as part of LBA and the questions that have driven ,'esearch at these sites. The second section reviews intensive studies of runoff, streamflow, and catchment water balance and how these processes are altered by clellring of tropical forest for pasture. The third section synthesizes what is known a~out the processes that control the concentrations and expOli of materials that reach streams via different hydrological flow paths inAmazonian forest and how these processes and flow paths are altered by deforestation and land use change. The fOUlih section summarizes what we know about how hydrological and biogeochemical processes change with scale and how this understanding can be used to both predict catchment response to land use change and manage Amazonian landscapes to maintain valuable hydrological and biogeochemical functions.
lCentro de Ciencia do Sistema Terrestre, INPE, Cachoeira Pau1. OVERVIEW OF MICROCATCHMENT WORK lista, Brazil. INLBA 2Marine Biological LaboratOly, The Ecosystems Center, Woods Hole, Massachusetts, USA. 1.1. Backgroulld 3Embrapa Amazonia Oriental, Belem, Brazil. 4Instituto Nacional de Pesquisas da Amazonia, Sao Jose. dos A change in land use from forest to nonforest can have imCampos, Brazil. . Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000727
pOl'tant consequences for the hydrological and biogeochemical function in small watersheds. Although the potential impacts of land use and land cover change on hydrological behavior are relatively well understood in hydrology, before LBA, there were few experimental studies in Amazonia. 505
506
TOMASELLA ET AL.
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
Most of the studies have been focused on the Barro Branco basin (1.3 km 2), in central Amazonia, and only for the periods 1976-1977 and 1981-1983 [Franken and Leopoldo, 1984, 1987; Nortclit/and Thomes, 1978, 1981, 1984; Leopoldo et al., 1982, 1984, 1985, 1995]. In addition, estimates ofnutrient export were made in the drainage basin of Lake Calado by Lesack [1993b], also in central Amazonia. Hodnett et al. [1996a, 1996b], Tomasella and Hodnett [1996], and Hodnett et a I. [1997a, 1997b] also presented results from a transect (zero order) in a central Amazonian pasture site. In other portions of Amazonia, Ross et al. [1990] and NortclifJ et al. [1990] presented results for Roraima (northern Amazonia), Elsenbeer and Vertessy [2000] studied a microcatchment in Pemvian Amazonia, and Elsenbeer et al. [1999] analyzed the implications ofland use and land cover change on mnoff generation in Rondonia (southwestern Amazonia). This lack of detailed information of the impact of land use and land cover changes on hydrological functioning in Amazonia and its sparse geographic coverage is also the case for the rest of the tropics. Andreasian [2004] analyzed the potential impact of forests on the hydrological cycle in 137 basins, only 10 experiments were from the tropics, none of them were from Amazonia. The popular perception is that in a forest, the soil, roots, and litter act as a sponge, absorbing water during the rainy season and releasing it gradually in the dry season. Forest conversion to pasture produces soil compaction by heavy machinery, loss of organic matter by oxidation and erosion [La I, 1987], thereby reducing the streamflow during the dry season [Bruijnzeel, 2004]. Equally, there is evidence indicating the opposite effect after reforestation. In general, forest clearance in the tropics leads to increased sh'eamflow during the first 3 years [Bruijnzeel, 1990]. This change is related to the reduction of evapotranspiration and an increase of stonn mnoff. Differences in the initial response to forest clearance between basins depends on a series of factors such as rainfall, elevation, maritime influence, catchment steepness and soil depth, clearance methods, soil fertility, among others. As several authors have pointed out, additional process studies are needed to fully understand the effect of forest conversion [Bruijnzeel, 1990, 2004; Bonell and Balek, 1993]. The limitations of comparative process studies in paired catchments are well documented in the scientific literature. The effects of land use and land cover change of catchments will first depend on the geology, which detennines the relative importance of the groundwater system, and on the interannual variability, which can also result in misleading conclusions regarding water yield [Bruijnzeel, 2004]. Because a multiplicity of factors can affect the hydrological response to forest clearance, comparison of results ob-
tained in different LBA catchments should be analyzed in the light of those limitations. Considering the sheer size of the Amazon Basin, where LBA catchments are located (Figure 1), such restrictions are particularly severe. Therefore, the purpose of this section is to provide the reader with some background infonnation regarding the general ecological, climatic, and geological characteristics of the different LBA catchments, where processes studies have been conducted.
507
. . . r ... l
2
.
.
-51
-45
1.2. Eastern Amazonia
Two different sites were studied in eastern Amazonia. The first site is the Cumaru catchment, located 110 km northeast of Belem at 1°11 'S, 47°34'W. This site includes three firstorder microcatchments of 25 ha each [Wickel, 2004; Wickel et al., 2008]. The climate of the region is humid tropical with an average temperature of 26°C and a dry season with less than 60 mm of rainfall during the driest month (Am, following the classification of Koppen). The average annual rainfall amounts to about 2500 mm, of which typically 60% falls during the wet season between January and April. The stratigraphy of the region is dominated by the Alter do Chao (Barreiras formation). The Alter do Chao formation is topped by a Quaternary series of variable thickness, of sands and clays known as the "Post-Barreiras" formation, consisting of unconsolidated, sandy sediments with fine and coarse quartz grains with clayey layers of varying colors. The Barreiras and post-Barreiras fOlmations are generally separated by a lateritic cmst of varying thickness. The soils in the region are predominantly classified in the Brazilian Soil System as Argissolo Amarelo (Ultisols in the Soil Taxonomy, Food and Agriculture Organization (United Nations), Rome (FAO) Acrisols), Latossolos Amarelos (Soil Taxonomy Oxisols, FAO Ferralsols), and Neossolos Quartzenicos (Soil Taxonomy Quartzipsamments, FAO Arenosols). The landscape is characterized by a rolling topography, covered with a heterogeneous patchwork of agricultural fields, fallow areas, and pasture, which is dissected by streams and fringed with a strip of riparian wetland forest. Deforestation of the primary forest in the region began after the colonization of the region known as Zona Bragantina. This deforestation accelerated following the constmction of a railroad between Belem and Braganya in 1883. Nowadays, agriculture in this region is dominated by smallholdings and farms of typically less than 50 ha in size. These farms typically consist of a patchwork of secondary/fallow vegetation and small agricultural fields planted with "traditional" crops such as cassava, corn, beans, rice, pepper, and passion fmit. Land is usually prepared using a slash-and-burn crop rotation system. The only areas not suitable for farming are the riparian wetlands; these are continuously waterlogged. In
10
-75
-69
-63
-57
Figure 1. Location of the LBA experimental catchments: AS, Asu catchment; qu, Cumaru Catchment; FV, Fazenda Vit6ria; JU, Jumena Catchment; RG, Rancho Grande catchment. The gray lines ihdicate national and state boundaries, while the black thick line indicates the Amazon Basin boundaty.
2001, about 12% was classified as riparian forest [Wickel, 2004], while approximately 50% was considered secondaly vegetation. The second eastern Amazonian site, Fazenda Vit6ria, is a cattle ranch, 6 km north of the town of Paragominas, at 2 0 59'S, 47°3l'W [Schuler, 2003; Moraes et al., 2006]. This site includes the Igarape Cinqiienta e Quatro catchment. The region was settled in the early 1960s, and the ranch is made up of a fi'agment of mature forest, pasture, logged forest, and capoeira (secondary-growth forest on abandoned land). The forest at Fazenda Vit6ria has a leaf area index (LAI) of 5.0-5.5 (aboveground biomass of 300 t ha- 1), with canopy height of 30 m and emergent trees of 45 m. The pasture was originally planted with Panicum maximum and later partially replaced by Braclliaria humidicola. The capoeira had been regenerating for about 12 years. Mean cattle density in the pasture was 0.8 head ha- I . Mean annual rainfall of the region is 1760 mm, with sh'ong seasonal variability (the dry season mns from June to September). Estimates of mean annual evaporation are 1515, 1370, and 1480 mm year- 1 for
the mature forest, pasture, and capoeira, respectively [Jipp et al., 1998]. The regional drainage system is located on the side of a Pleistocene river terrace. On the central part of this slope, two first-order catchments, one forested (0.33 ha), the other covered with pasture (0.72 ha), were selected. The catchments are overlain by 10 m deep clayey soils from the colluvium Belterra formation. The Belterra clay covered the Alter do Chao (Barreiras Group) formation. Between both formations, there is a deposit velY rich in iron (plinthic layer), which is a buried Paleosol [Schubert, 2005]. The soil in the forested microcatchment was classified in the Brazilian soil system as very clayey Latossolo Amarelo (Soil Taxonomy Oxisols, FAO Ferralsols), while in the pasture, it was classified as clayey Plintossolo (Soil Taxonomy Plinthic, FAO haply Plinthosol). The plinthite depth varies from 9 to 10 m at the plateau to 0.4 to 0.6 m downslope. According to Moraes et al. [2006], the hydraulic characteristics of both sites are similar, in spite ofthe differences in the soil type.
508
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
1.3. Southern Amazonia
Four adjacent, small (I-2 ha) headwater catchments in an undisturbed forest near Jumena, Mato Grosso, were selected in southern Amazonia (I0025'S, 58°46'W, 230-250 m above sea level (masl» in a landscape ofthe gently undulating hills typical of the Brazilian shield [Johnson et al., 2006]. Annual rainfall in the study area is 2379 mm, with 70% concentrated in the wet season. Temperature varies from 16° to 32°C in the forest. Soils in the catchments in the Brazilian System are Latossolo Amarelo (Soil Taxonomy Oxisols, FAO Ferralsols) and Argissolo Vermelho-Amarelo (Soil Taxonomy Ultisols, FAO Acrisols).
1.4. Central Amazonia In 2001, a second-order stream at the Asu catchment (6.58 k:m2) was gauged, located approximately 84 Ian NNW of Manaus (03°08'S, 60 07'W). It collects the discharge of five first-order streams and includes most of the common landscape forms that occur in the region [Waterloo et al., 2006; Tomasella et al., 2008]. By 2005, a paired pasture microcatchment was established at the Fazenda Colosso, which is about 89 Ian NNE of Manaus. This catchment has a drainage area of 1.22 km 2, at 02°25'S, 59°53'W. At the same time, and in order to allow comparisons with the forested Asu catchment, one of the first-order streams of the Asu was gauged to a drainage area of 1.16 k:m2 [Trancoso, 2006]. In the Manaus area, the regional basal geology is Precambrian clystalline rock. In addition, three stratigraphic units can be identified. The deepest is the Trombetas formation, which is not exposed. This is overlain by the kaolinite rich Alter-do Chao teliiaty formation (Barreiras Group), which is in some places covered by quaternaty sediments [Dias et al., 1980]. In the area, the local relief has been formed by the dissection of a plateau of tertimy sediments by valleys of various dimensions [Bravard and Righi, 1989]. The plateaus are generally flat or have gentle slopes «7%), with altitude vatying between 90 and 120 m. The valley floors are generally broad and almost flat, at an altitude ranging fi'om 45 to 55 m. The water table is maintained close to the valley floor up to 100 m from the stream, and these areas often contain swampy pools. These swampy areas of the valley floor are known locally as baixios. The plateau soils are classified as Latossolos amarelos Micos, textura argilosa [Ranzani, 1980; Chauvel, 1982; Chauvel et al., 1987a, 1987b] equivalent to the Oxisols of the Soil Taxonomy and the Fenalsols in FAO soil classification. These soils are clayey at the surface (6575% in the top 30 cm) and velY clayey below about 1 m depth (80-90%). On the intermediate surface (slope), the soils vmy from the yellow Latosols to the eluviated soils known as Ar0
gissolos Vermelho Amarelo Alicos [U.S. Department of Agriculture, Washington, D. C. Ultisols, FAO Acrisols]. These soils usually show a clay content of about 30% in the top 30 cm, increasing to 45% below 1.5 m. The valley bottom is dominated by very sandy Espodossolos (Soil Taxonomy Spodosols, FAO Podzols), with clay contents below 5%. The Colosso pasture microcatchment is covered by kikuio (Brachiaria humidicola), and it is about 45 years old. The area is still exploited with a cattle density of 1.5 head ha- 1, and it is burned yearly. In the Asu catchment, on the other hand, vegetation is dense humid evergreen tropical forest, with marked spatial variability depending on the position on the landscape. Over the sandy soils of the valley, the forest is classified as campinarana [Ranzani, 1980], which is a dense ShOli forest, dominated by palms, shmbs, and lichens. The canopy height of this forest type varies between 12 to 15 m. Over the slopes, where the argissolos occur, the forest is classified as mata arenicola, with canopy height up to 25 m. The tree density is high, with about 2500 trees ha- I with a diameter >5 cm. On the clayey soils of the plateau, the prevailing forest formation is the Mata argilicola, with a canopy height 30 to 35 m and a density of about 1500 trees ha- I with a diameter >5 cm. Although tree density is lower on the plateau than on the slope, the trees are bigger in diameter and height. The climate of the region is Afi according to Koppen classification, with an average annual temperature of 26°C (minimum 19°C and maximum 39°C). Relative humidity ranges from 77% to 88%, with an annual average of 84% [Leopolda et al., 1987]. Annual rainfall varies from 1800 to 2800 mm, with a rainy season from December to May. In March and April, the mean monthly rainfall exceeds 300 mm. 1.5. Southwestern Amazonia
Rancho Grande is located in the center of the state of Rondonia (IooI8'S, 62°52'W, 143 masl). Several studies have been conducted at this research site, including a comparison between forested (1.37 ha) and pasture (0.73 ha) paired catchments [Chaves et al., 2008]. Mean air temperature is about 27°C, and the mean annual rainfall was 2265 mm from 1984 to 1995 [Elsen beer et al., 1999]. The prevailing geology is the Precambrian gneiss, which has been weathered to a low relief landscape, smooth, convex slopes, with ridges of more than 500 m amsl. The area belongs to a morphostmctural unit known as Southern Amazon Dissected Highlands, which is characterized by a pronounced topography with altitudinal differences up to 150 m, which are the remains of ridges of Precambrian rocks of granites and gneisses of the Xingu Complex [Zimmermann et al., 2006], and separated by valleys ofvarious widths. Soil orders
TOMASELLA ET AL.
associated with this morphostmctural unit are Latossolo Vermelho Amarelo qistr6fico argissolico (Soil Taxonomy Oxisols, FAO Ferralsols), Argissolo Vermelho Amarelo (Soil Taxonomy 41'tisols, FAO Acrisols), Cambissolos (Soil T.axonomy Inc97hsols, FAO Cambisols), and Neossolos Quartzenicos (Soil Taxonomy Quartzipsamments, FAO Arenosols) on steep slopes and along streams, respectively [Germer et at., 2006]. The residual ridges and a portion of the lowland of Rancho Grande Research site are covered by rainforest, and the remaining area by capoeira, pasture (Brachiaria brizantha, Brachiaria humidicola, and Brachiaria decumbens), a small teak (Teca grandis) plantation, and an abandoned cacao-banana plantation [Godsey and Elsenbeer, 2002], referred to here as banana. The primary vegetation is open tropical forest (Floresta Ombr6fila Aberta) with a large number of palm trees. It is characterized by a discontinuous upper canopy of up to 35 m height with emergent trees up to 45 m tall, permitting the sunlight to reach the understOly and thereby facilitating dense undergrowth. Roberts et at. [1996] detennined a LAI of 4.6 for an open tropical rainforest at the ecological reserve Reserva Jarll about 100 Ian east of Rancho Grande, compared to a LAI of 6.1 for a dense tropical rainforest measured 60 Ian north of Manaus. For trees with diameter at breast height (DBH) >5 cm, the tree density is 813 ha- I including 108 palms, and 520 ha- I for DBH >10 cm, including 81 palms [Germer et a I., 2006]. This site has been used to study differences in mnoff mechanisms, with emphasis on saturated hydraulic conductivity. The pasture was cleared in 1985, planted with pasture (Brachiaria humidicola) in 1986, and has been actively grazed since then at about 1 head ha- I [Chaves et al., 2008]. The other research site in Rondonia is located at Fazenda Nossa Senhora, a cattle ranch at lO o45'S, 62°22'W. The sampling site is a microcatclunent that drains 14.5 km 2 of pasture (Brachiaria brizantha) between 18 and 25 years old, grazed at an average cattle density of2.1 head [Biggs et al., 2006]. The upslope portion of the catchment is dominated by the Argissolo Vennelho Amarelo (Soil Taxonomy Ultisols, FAO Acrisols) formed on the gneissic basement of the Brazilian craton. Soil texture is 50-85% sand and 7-15% in the upper 0-15 cm of soil. Clay content increases to 15-30% at about 35-50 cm depth. The near stream zone is loamy at the surface, changing to sandy loam at 25-35 cm depth. Bedrock underlies the soil at a depth of70-120 cm. 2. WATER BALANCE, RUNOFF GENERATION, AND THE EFFECTS OF DEFORESTATION Understanding the mechanisms that lead to mnoff generation was one of the main objectives of small catchment
509
hydrological research within LBA. The runoff caused by rainfall that exceeds the soil infiltration capacity is identified as Hortonian overland flow. Saturation overland flow includes only the runoff produced by precipitation that falls on saturated (or almost saturated) parts of the near stream zone (riparian). The exfiltration of soil water from upslope, when subsurface flow is unable to remove infiltrated water, or even rainfall falling on saturated areas upslope, is considered return/low. Finally, groundwater flow that drains from the upslope to the stream through a relatively thin layer (0-2 m thick in the riparian zone) is referred here as base flow. 2.1. Mechanisms ofStreamflow Generation in Forest Elsenbeer [2001] proposed a conceptual fi'amework ofhillslope hydrological response patterns and their environmental control in Amazonian forests. Elsenbeer [2001] reviews several field experiments in tropical basins and focused his analysis on Acrisols and Ferralsols, which covered 60% of the humid tropics. Most of the analysis is based on the range of measured saturated hydraulic conductivity and, in particular, its variability down the soil profile, essentially the concept of a "throttle" layer suggested by Bonell [1993]. The hypothesis emerging from Elsenbeer's review was that the most impOliant mechanism of runoff generation in forested ACl'isol is overland flow, but this is not the case on Ferralsol landscape. It is interesting to note that Elsenbeer stressed the need !for additional testing on Fenalsol landscapes, considering the scarcity of detailed data for hydraulic parameters such as saturated hydraulic conductivity. The lack of detailed soil infonnation of Amazonian soils, patiicularly with respect to the hydraulic parameters, resulted in a cmde generalization of the hydrological functioning of forested soils. Since most of the hydrological studies have been concentrated on central Amazonia, there is a widely accepted belief within the scientific commlmity that rainforest soils are free of overland flow [Elsenbeer et al., 1999]. This is due to a relatively weak decrease of saturated hydraulic conductivity with depth in the Oxisols and Ferralsols developed on the Barreiras formation in central Amazonia. Therefore, it is ofparticular interest that the results of Elsenbeer et at. [1999] from the catchment in Rondonia, demonstrated a pronounced vertical anisotropy (decrease of hydraulic conductivity) of those soils and suggesting a potential for the occurrence of lateral flow and a perched water table. Early results from the La Cuenca catchment, located in eastern Pem, revealed strong vertical anisotropy [Elsen beer and Vertessy, 2000]: the values of saturated hydraulic conductivity varied from 316 mm h- I for 0-0.1 m, to 0.14mm h- I for 0.3-0.4 m depth. A hydrological discontinuity at a depth of 0.1-0.2 m (indicated by an abmpt decrease of saturated
510
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
hydraulic conductivity, to a value of6.9 mm h- 1) is the main mechanism for controlling the runoff generation. The data of La Cuenca catchment suggested that return flow is the main mechanfsm for stormflow generation, in contrast to the results of central Amazonia [Franken, 1979; Nortel!ff et al., 1979; Lesack, 1993b] in soil with relatively weak reduction of saturated conductivity with depth. The existence of a perched water table with depth at La Cuenca catchment reinforced the idea of a significant proportion of return flow and was cOIToborated by the fact' that overland flow was extensively detected in the catchment. Working in a forested microcatchment at the Fazenda Vit6ria, eastern Pant, Moraes et al. [2006] analyzed the hydrological functions of plinthic soils, which are the main type of imperfectly drained soils in Amazonia. In those soils, saturated hydraulic conductivity decreased sharply with depth: from 200 mm h- 1 near the surface to 0.7 mm h- 1 at 0.8-0.9 m in the forest site [Moraes et al., 2006]. The strong vertical gradients are different from those values reported for central Amazonia [Franken and Leopodo, 1986; Norteliff and Thomes, 1989; Lesack, 1993b; Tomasella and Hodnett, 1996]. The strong vertical gradient on plinthic soils of eastern Amazonia explains the main mechanism of runoff generation on such basins. For this reason, Moraes et al. [2006] detected a significant number of days with a perched water table. In spite of the low conductivity at a depth of 1 m in the Fazenda Vit6ria forest microcatchment, the main mechanism for runoff generation was saturation overland flow. Although the soils have low conductivity, even above the plinthite, the contribution of subsurface flow to stormflow is relatively small: the forest catchment at Fazenda Vit6ria produced 5.3% of the total rainfall in the form of saturation overland flow and about 0.8% as subsurface flow on a 3-year average. Previous work indicated that saturation excess overland flow from valley floors with shallow water tables is the most important process of stormflow generation in central Amazonian catchments [Norteliff and Thomes, 1981; Lesack, 1993b;Hodnett et al. 1997a, 1997b]. A recent paper by M. G. Hodnett et al., (Subsurface hydrological flow paths in a Ferralsol (Oxisol) landscape in central Amazonia, manuscript in preparation, 2008), using data from the Asu catchment, partly contradicts previous beliefs regarding the hydrological functioning of central Amazonian landscapes. Although the dominant role of saturation overland flow is recognized, field data suggest that, after sufficiently large rainfall events, the throttle effect is likely to cause rapid throughflow in the conductive macroporous layer. When throughflow accumulates in concavities on the slope, it will produce rehlrn flow. Field data suggest that rehlrn flow seems to be the only process of stormflow generation in headwater catchments of a
few hectares, while deep drainage is almost certainly controlling most of the discharge. 2.2. Effects ofLand Use Change on RunoffMechanisms
To assess the effects of land use changes in hydrology, it is necessary to understand the interactions between the soil physical and rainfall characteristics, which determine those flow paths that will be activated in response to a given rainfall event. Between rainfall events, flow paths are predominantly vertical; during rainfall events, on the other hand, the horizontal movement of water (lateral flow) may become dominant. The changes of soil hydraulic conductivity with depth play a fundamental role in determining which flow paths will be activated and ultimately in all modes (except sahlration overland flow) of stormflow generation [Elsenbeer, 2001]. Several physical and biological factors determine the degree of vertical soil anisotropy. Among those factors, land use and land cover changes playa decisive role because they alter, with different intensity, vertical soil anisotropy, and hence water pathways during rainfall events. Again, the degree ofperturbation of soil anisotropy due to land use change depends not only on the intensity of soil disturbance, but on the soil susceptibility to such perhlrbations. Therefore, field studies, to asses the impact of forest conversion, are based on comparing the vertical distributions of soil hydraulic propeliies on primaly forest and on dishlrbed plots. In earlier work at Rancho Grande, Rondonia, Eisenbeer et al. [1999] suggested that Hortonian overland flow occurs quite frequently in pashlre. Their results indicated that sahlrated hydraulic conductivities at the surface were relatively low at the pashlre site, against the general evidence of high conductivities at the soil surface in forested areas. This result was attributed to hydrophobic pashlre soils, since the experiment was conducted at the end of a prolonged dry season. In agreement with previous knowledge, E. B. Safran and T. Dunne (unpublished data, 1995), using a sprinkler infiltrometer in a forest near Porto Velho, estimated surface infiltration cap~city of 150-180 mm h-1, compared to 18-20 mm h- I in a 10-year-old pashlre nearby. A more recent and detailed Shldy on the same Rancho Grande site by Zimmermann et al. [2006] showed that saturated hydraulic conductivity (KsaD decreased from an average of 200 mm h- 1 at a depth of 12.5 em in the forest, to a mean of 69 mm h- 1 in the teak plantation, and to an average of 26 mm h- 1 in the pasture. The shldy compared Ksal in three other land uses (capoeira, pashlre, abandoned banana plantation and forest). Interestingly, measurements of Ksal in capoeira showed values close to the forest: 200 mm h- 1 at 12.5 em. They concluded that a depth of 20 cm, most of the effect of clearance disappeared, which is in agreement
TOMASELLA ET AL.
with the previous results of Tomasella and Hodnett [1996] from central AmaZOtlia. Measurements showed lower soil conductivity in the capoeira and banana plantations but similar K sat compared/fo the primmy forest [Zimmermann et al., 2006]. This lea>!" to the conclusion that the capoeira and banana land uses have little, if any, consequences for overland flow generation. This conclusion has an important implication for large-scale hydrological models, suggesting that, at least on these soils, capoeira should be treated as primary forest, from the hydrological point of view. Results at Rancho Grande indicate that Hortonian overland flow is generated mainly on heavily exploited pashlre sites. Microcatchment data, combined with model results in Fazenda Nossa Senhora in Rondonia [Biggs et al., 2006], indicated that stormflow accounted for 17-22% oftotal runoff and about 8% of annual precipitation. In telms of the predominant runoff mechanism, Biggs et al. [2006] estimated that sahlration overland flow, subsmface and return flow, and Hortonian overland flow accounted for, respectively, 5-18%, 3-16 %, and 17-22% of the total discharge. The remaining percentage (55-66%) was estimated as groundwater flow. It should be noted, however, that Biggs et al. [2006] included both sahu'ation excess overland flow and rehlrn flow from upslope areas in their estimation of HOIionian overland flow. Therefore, it is sensible to assume that saturation overland flow could be underestimated in their calculations. In spite of these limitations, and consideling that measured infiltration capacities at Rondonia suggest that infiltration-excess overland flow rarely occurred in forest areas [Elsenbeer et al., 1999], it is possible to conclude that the conversion of forest to pasture has dramatically changed the runoff generation mechanism. In forest and pasture catchments at Rancho Grande, Germer et al. [2009] and Chaves et al. [2008] found that runoff was <1% of rainfall in the forest catchment but 17-18% of rainfall in the pasture catchment. While quick flow made up the vast majority (96-98%) of runoff in these velY small catchments in both land uses, the 20-fold greater amount in the pashlre compared with the forest clearly indicated that forest conversion significantly altered the hydrological runoff mechanisms. Based on tracer analysis, Chaves et al. [2008] estimated that during the rainy season, throughfall provided 57% of streamflow, groundwater provided 24%, and shallow soil water contributed with 19% oftotal streamflow. In the pashlre catchment, overland flow supplied 60% of stl'eamflow, groundwater supplied 35%, and soil water contributed with the remaining 5%. Therefore, the most significant changes in terms of streamflow composition be~ tween forest and pashlre catchments were in the soil water and groundwater components, despite the velY large differences in the total amount of overland flow that contributed to pasture streamflow.
511
Results fi'om the Shldy of the paired forest and pasture microcatchments at Fazenda Vit6ria in eastern Amazonia [Moraes et al., 2006] showed that sahlrated hydraulic conductivity near the surface of pashlre was about 4 mm h-I, compared to 230 mm h- 1 in the forest. At a depth of 0.8-0.9 m, Ksat decreased to 0.05 mm h- 1, against 0.7 nun h- 1 at the forest. It is interesting to note that the median Ksat at a 12year-old capoeira at Fazenda Vit6ria decreased from 14 mm h- 1 at the surface to 0.3 mm h- 1 at 0.8-0.9 depth, a velY different result to that found at Rancho Grande. In ternlS of the runoff generation mechanism, the paired catchment study at Fazenda Vit6ria indicated that overland flow in the pasture was 19% of the total rainfall, 40% in the form of infiltration excess (Hortonian overland flow), and 60% as sahlration overland flow. Results indicated that the forest conversion to pashlre increased overland flow to 19% of total throughflow, which is four times greater than in the forest. Subsurface flow in the pasture was 1.4%, slightly higher than in the forested catchment. Three events in the Fazenda Vit6ria catchments revealed a dramatic reduction in the hydrograph peak, the time of rise, and the centroid lag. In eastern Amazonia, in the catchment with mixed vegetation studied by Wickel [2004] and Wickel et al. [2008], field measurements indicated an average value ofK sat of 161 mm h -1. Based on mqasured rainfall intensity, they concluded that rainfall never exceeded infiltration rates. Those high infiltration rates 'Yere attributed to preferential flow occurring along the existent root system of the fallow vegetation. Wickel [2004] estimated that 98% of the total streamflow corresponds to groundwater flow, while the remaining 2% was considered stonnflow, which consists entirely of saturation overland flow. On a paired microcatchment shldy in central Amazonia, Trancoso [2006] found clear evidences of the impacts offorest conversion to pashlre. The analysis of 40 hydrological events in ~ 1 km 2 paired basins indicated a significant reduction of concentration time, recession time, and time of peak in the pashlre catchment. The events also revealed a significant increase in the stormflow volumes and the value of the peak in the pasture compared with the forested catchment, indicating that the runoff mechanisms has been severely affected by land use change. 2.3. The Annual Water Balance It is clearly recognized that the most important hydrological effect of forest conversion to pasture is related to evapotranspiration changes [Zhang et al., 2001]. The reduction of evapotranspiration over pashlre sites is related to the combination of several factors such as the decreasing of interception capacity, surface roughness, deep soil water uptake,
512
TOMASELLA ET AL.
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
and the increase of albedo. A reduction in evaporation has a profound impact on the radiation balance and, consequently, on soil-vegetation-atmosphere interactions. Interception~depends not only on the physical stmcture of the vegetation, but also on rainfall characteristics. LBA hydrological studies quantified interception in several locations and provide evidence on how the interception varies across different regions ofAmazonia and how interception is strongly affected by the interannual variability of rainfall. Schubert [2005] measured interception l6sses between 11.6% and 20% of rainfall at Fazenda Vit6ria forest site between 2000 and 2003. The average measured value was 15.5%, and the interannual variability of interception was attributed to rainfall characteristics. In agreement with those results, Cuartas et al. [2007] showed that in years with normal (near average) rainfall, interception was 13.3%, compared to 22.6% in a dry year within the Asu microcatchment in central Amazonia. The interception difference is explained by a comparison of mean intensity and duration of rain events in a normal year (8.77 mm h- 1 and 1.88 h) versus the driest year (5.36 mm h- 1 and 2.32 h). Interception loss for the whole period accounted for 16.5% of the gross rainfall, with throughfall 82.9% and stemflow 0.6%. More interestingly, interception models (particularly, that proposed by Gash et al. [1995]) succeeded in capturing the variability associated with the variability in the characteristics of precipitation. Germer et al. [2006] repOlied a throughfall of 89.6% of gross rainfall for Rancho Grande in southwestem Amazonia. Although this estimate is smaller than the studies in central Amazonia, it should be noted that the forest in Rondonia had a significantly lower LAI and, consequently, a reduced canopy capacity. A paired catchment study in central Amazonia [Trancoso, 2006] indicated significant differences in water balance components. Arumal evaporation in the pasture was estimated as 876 mm, while it was calculated as 1277 mm for the forest. This means that evaporation was 1.1 mm d- I lower in the pasture compared with the forest catchment. In terms of annual discharge, water yield in the pasture was 686 mm compared with 328 mm in the forest (0.98 mm d- 1 higher in the pasture than in the forest). It is interesting to note that the increase in discharge was almost equivalent to the corresponding decrease in the estimated evaporation. In addition, data from the paired catchment study of central Amazonia indicated that stonrrflow was 15.3% within forest and 26.4% within pasture. Therefore, mnoff coefficients (relation QIP) were 0.17 and 0.32 for forest and pasture, respectively. Moreover, flow duration curves showed irregularities in streamflow components in the pasture catchment, indicating that forest replacement has significant impacts on the ability to reduce floods during the wet season and droughts in dry seasons.
The Asu catchment Shldy [Tomasella et al., 2008] also demonstrated the effect of large inh'aseasonal and interannual variability of rainfall on the components of the water balance. Using 3 years of data, and based on measurements of soil moisture, rainfall, evaporation, and groundwater table, the authors showed a strong memory effect in the groundwater system that carried over seasonal climate anomalies from 1 year to the next and affected the hydrological response well beyond the time span of the anomaly. In addition, the deep unsahlrated zone was found to playa key role in reducing most ofthe intraseasonal variability and also affected the groundwater recharge. This memOlY effect is crucial for sustaining streamflow and evaporation in years with rainfall deficiency. The memory effect caused by storage in both groundwater and unsaturated systems may also prevent the closure of annual large-scale water balances, which assume that storage retums to a standard state each year. Data from Fazenda Vit6ria showed that, in annual terms, the discharge from the forested catchment was 46 and 43 mm for the years 2002 and 2003, respectively. In the pashIre, almual discharge varied from 320 to 319 mm for the same period. Moraes et al. [2006] calculated the drainage through the plinthic layer assuming free drainage and, by means of the water balance, estimated an evaporation of 1629 and 1316 mm for the forest (91-95% of annual rainfall) and 1419-1038 mm (70-75% of annual rainfall) for the pashlre. Because of the uncertainties of the drainage estimations, plus the effect of deep root uptake below the plinthic layer (particularly in the forest), these estimates of evaporation could be higher than other values reported for Amazonia and should be considered with caution. In terms of the mnoff coefficient (ratio between discharge and rainfall), the pasture showed a relationship of 17.3%, while in the forest, it was 3.2% of throughfall. The increase of discharge in the pashlre could be attributed to the reduction of evaporation in the pasture and changes in the rainfall runoff mechanisms. These numbers are in the range ofthose reported by Johnson et al. [2006] for four adjacent small catchments in Mato Grosso (so~lthem Amazonia). On an annual basis, they found that discharge was about 5.9% of total rainfall. In eastem Amazonia, and using the Penman Monteith method, Wicke I [2004] estimated an evaporation of about 59% of total rainfall. This number was close to the value derived by catchment water budgets. Runoff coefficient varied from 38% to 41 %. Biggs et al. [2006] estimated that the groundwater flow at Fazenda Nossa Senhora pasture catchment varies from 19% to 30% of annual rainfall, while evaporation was about 53-65% of annual rainfall. These results suggest that total discharge (groundwater and stonrrflow) represents 35% to 47% of annual rainfall, and stonrrflow was about 17% of total rainfall. Measurements in the forested and pasture
catchments in Rancho Grande [Chaves et al., 2008] showed that water yield was ~i7% of total precipitation in the pasture compared to 0.8% iIi'the forest. In summary, d",fa from small catchments in Amazonia indicate that the y6nver sion of forest to pasture produces an increase in discharge and a reduction in evaporation. In addition, there is an increase of stormflow in the pasture mainly because of an increase in infiltration excess (Hortonian) overland flow and even an increase of subsurface stormflow due to the reduction of saturated hydraulic conductivity, which in him increases the risks of erosion and sedimentation. In the forest, on the other hand, the main mechanism for stOlmflow generation is saturation overland flow in the riparian zone. As suggested by field data, this conclusion may not be completely valid in headwater catchments, where stormflow could be dominated by retum flow due to the throttle effect (Hodnett et aI., submitted manuscript, 2008). Overall, results suggested that land use history, in terms of duration and intensity, prior to abandonment of deforested areas, is cmcial to understand the potential effects of clearance on soil hydraulic properties and, particularly, how rapidly infiltration and saturated hydraulic conductivity Jecovered [Zimmermann et al., 2006]. Of all land uses, it seems clear that pasture with intensive grazing produces the most drastic impact on the hydrological response. 3. NUTRIENT DYNAMICS AND THE EFFECTS OF DEFORESTATION The conversion of Amazon forest to nonforested land uses has impOliant consequences for biogeochemical functioning and nutrient balances in small watersheds. These processes are controlled by a number of factors that interact with land use in potentially important ways. For example, changes in land use can alter the underlying biogeochemistty of soils by changing element cycling rates and the forms and availability of solutes in soils and groundwater. These solutes then interact with the stmcture of hydrological flow pathways, or flow paths, which move water off the terrestrial landscape and deliver water to streams. These flow paths include precipitation, throughfall, overland flow, veliical water movement into groundwater, lateral groundwater movement, and water movement through streamside riparian zones. Dominant soil-landscape units, or "soilscapes," also exert an important control on these flow paths because soil characteristics, such as horizons of reduced hydraulic conductivity, control generation of surficial "quick" flow paths versus deep "slow" flow paths [Elsen beer, 2001]. Changes in land cover, such as the clearing of forest for pashll'e, can influence nutrient transformation, retention, and loss in catchments by altering the relative impOliance of different flow paths and by
513
altering solute transformations that take place within soil-tostream hydrological flow paths and within stream channels themselves. Here we review insights into important features of the stmchlre of hydrological flow paths and the cycling of N, P, and cations that play important roles in catchment nutrient dynamics. 3.1. Nitrogen Dynamics in Forest
Despite relatively low rates of inorganic atmospheric N deposition to Amazonian forest, in the range of 1 to 4 kg N ha [Holland et al., 1997; Hauglustaine et al., 2004; Bouwmann et al., 2002; Filoso et al., 1999; Germer et al., 2007], most lowland Amazonian forest soils show high rates ofN release from mineralization of soil organic matter and high rates of nitrification [Neill et al., 1997; Neill et al., 1999; Verchot et al., 1999; Luiziio et al., 2004]. Amazonian forest foliage contains relatively low C:N ratios [Vitousek, 1984], and forest biomass and soils contain relatively high stocks ofN [Vitousek, 1984; Markewitz et al., 2004; Davidson et al., 2004], although 80% to 90% ofN in deep soils is passive and cycles very slowly [Trumbore et al., 1995]. These characteristics indicate high N availability and absence of N limitation for forest trees. Amazonian forest soils show several other indications of high rates of N cycling and lack of conservative N cycles. Soil emissions of both nitrous oxide (NzO) and nitric oxide (NO) are high [Verchot et al., 1999; Melillo et al., 2001; Davidsbn et al., 2002], soil solution concentrations of N0 3- and ratios of dissolved inorganic to dissolved organic nitrogen (DIN:DON) are high, indicating relatively rapid and complete nitrification of inorganic N [Neill et al., 2001; Markewitz et al., 2004; Neill et al., 2006a], and forest foliage 15N concentrations are high, suggesting a relatively open N cycle with preferential loss of 14N [Martinelli et al., 1999]. N oxide emissions from forest are in the range of 3 to 4 kg N a-I and are high compared to many other forests [Davidson et al., 2004; Neill et al., 2006a]. Forest soil solution inorganic N concentrations of > 100 Ilmol L, strongly dominated by N0 3-, occur in most places, where they have been measured [Neill et al., 2001; Markervitz et al., 2004; Neill et al., 2006a; Chaves et al., 2008]. Despite high soil solution N0 3- concentrations and apparent abundant stocks of N in vegetation and soils, stt'eam water dissolved N concentrations are low and typically two orders of magnitude less than concentt'ations in soil solution [Lesack, 1993a; McClain et al., 1994; Williams and Melack, 1997; Neill et al., 2001; Markewitz et al., 2004]. This suggests high N removal in hydrological flow paths leading from soils to streams. Sharp decreases in soil solution N0 3in gradients from upland to streams also support arguments for high N removal as water passes through riparian zones
514
TOMASELLA ET AL.
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
[McClain et al., 1994; Williams et al., 1997]. Annual catchment DIN export via groundwater and stream water is also low and has not been measured to be higher than about 1 kg N ha- 1 [LesacJt, 1993a; Davidson et al., 2004, Germer et al., in press]. Shallow lateral flow paths account for a relatively insignificant portion of total stream flow, even in locations with restricted soil hydraulic conductivity [Moraes et al., 2006, Chat1es et al., 2008], this suggests high rates of N removal in flow paths that lead from soil solution, where N03concentrations are high, to streaths, where N03- concentrations are quite low. Patterns of decreasing concentrations ofN03- in groundwater along toposequences from terra firme forest to saturated streamside floodplains has been interpreted to indicate high N03- losses in riparian zones [McClain et al., 1994; Williams et al., 1997]. While Neill et al. [2001] suggested that contribution of even a small percentage of soil solution N0 3- could increase N03- concentrations observed in streams, isotopic evidence suggests that most stream N03is derived from in-stream nitrification and not delivery from the catchment [Brandes et al., 1996]. Extremely low N0 3concentrations «1 I1mol L) observed in terra firme groundwater from forest in Rondonia suggest high N removal between soil solution rather than concentration in riparian zones. The groundwater N03- concentrations measured in Rondonia, however, were from Ultisols and were much lower than the approximately 30-50 I1mol L- 1 concentrations measured in wells at similar depths in the upland portions of small catchments in central Amazonia [McClain et al., 1994; Brandes et al., 1996; Williams et al., 1997] or in deep soil solution in the eastern Amazonia [Markewitz et al., 2004]. This difference suggests that soil flow path structure and lower within-flow path N loss or retention on Oxisols will lead to more N0 3- in the regional groundwater. These findings have important implications for our understanding of land-water N movement and the importance of soils and riparian zones as N transformers in Amazonian landscapes. They are also critical for developing models to predict land-water N movement and riparian function at larger scales. In the "riparian removal" conceptualization, N0 3- generated in forest soils arrives at streams, where it either is or is not removed depending on the presence of a saturated riparian zone. If this is true, land-water N fluxes at the river network and regional scales will depend on catchment riparian zone structure and the presence of flow paths that move water through riparian zones. Stream order and catchment size are likely to be important to these functions because they control the size of riparian zones and the structure of riparian flow paths. In the "soil removal" conceptualization, N03- is removed in soil vadose zones or at the interface of soil solution with groundwater. This
transformation occurs dispersed across catchments and results in groundwater that is nearly devoid of N03-. If this is true, flow paths that move water from soils to groundwater are more important controls of catchment N retention, but flow paths that move N through riparian zones are less critical. Stream order and structure of riparian zones would not be important because water arrives at near-stream zones already stripped ofN03-. These differences are also important potential determinants of how catchment N dynamics will respond to land use change. New work on N dynamics and transport in stream channels suggests that once inorganic N enters small forested streams, it has the potential to travel long distances. Additions of NH/ and N03- coupled with conservative tracers indicate that Amazonian forest streams have relatively low transient water storage and long NH4 + and N03- uptake lengths [Neill et al., 2006b]. Tracer additions of 15N to a second-order Rondonia forest stream revealed that approximately 67% of added NH4 + traveled through a 200-m reach unchanged, 8% of the added NH/ tracer was nitrified in the stream channel, and only 1% of added 15N was recovered in the stream organic matter compartments. Deegan et al. (submitted manuscript, 2009) also found no evidence for N0 3- uptake, indicating that although processes in catchments form highly effective barriers against N movement from land to streams, once inorganic N reaches forested stream channels, it has the capacity to travel distances that exceed many kilometers. 3.2. Effects ofLand Use Change on Nitrogen Dynamics
The clearing of tropical forest sets in motion a number of changes that have the potential to alter, both the concentrations of solutes that move from land to streams in small watersheds and the concentrations of solutes in stream channels. These processes are controlled in impOliant ways by the changes to soil biogeochemistry and the structure of stream channels that follow deforestation. Forest clearing can lead to. rapid but relatively shOli-lived increases in rates of soil N cycling, these then lead to increased concentrations of inorganic N in the form of nitrate in soil solution. In Rondonia, Neill et al. [2006a] found that forest clearing for pasture increased soil solution concentrations by more than 10-fold during the first year after clearing. Soil solution concentrations of ammonium and dissolved organic N seem to be much less influenced by clearing. This immediate postdisturbance elevated nitrate concentration has been reported at other Amazonian locations in San Carlos, Venezuela by Uhl and Jordan [1984] and from near Manaus in central Amazonia by Piccolo et al. [1994a] and may be a relatively general feature ofthe N cycling response of tropical forest on
weathered soils. There are suggestions that this elevated soil solution nitrate concytitration immediately following clearing leads to increasl"td nitrate export in streams. Williams et al. [1997] found llevated nitrate export in a small central Amazonian catyrlmentundergoing deforestation. However, whole-watershed experiments that examine soil-to-stream solute dynamics and watershed export during the clearing process have not yet been conducted in Amazonia. Studies fi'om both eastern and western Amazonia show that conversion of tropical forest to pasture leads to a decline in the rates of soil N cycling as reflected by lower rates of net N mineralization in soils and lower soil solution concentrations of nitrate in established pastures compared with forest [Piccolo et al., 1994b; Neill et al., 1997; Verchot et al., 1999; Markewitz et al., 2004]. This pattern appears to be widespread, and typical "pasture" patterns of N cycling generally become established within 1 to 3 years. While the links between lower rates of N cycling in soils and lower emissions of N-oxides from pasture soils appear clear [Verchot et al., 1999; Melillo et al., 2001], it is less clear whether lower soil solution nitrate concentrations in pastures lead directly to lower concentrations of nitrate in stream water draining watersheds that are dominated by pasture. Neill et al. [2001] found lower concentrations of nitrate in streams draining established pasture than in streams draining intact forest and argued that lower availability of inorganic N in soils and lower soil solution N concentrations in pastures was linked to lower concentrations of nitrate in small streams draining pasture compared to streams draining forest. New evidence points to two mechanisms that linle patterns of soil N cycling in forest to forest soils, to forest stream water nitrate concentrations. First, Chaves et al. [2009] repOli low nitrate concentrations «111M) under intact forest watersheds in Rondonia, suggesting that groundwater is not a primaty source of nitrate to small streams. Instead, nitrate delivered to ephemeral streams is derived primarily from a combination of more surficial flow paths, including both near-surface flows (that resemble throughfall in their chemistry) and soil solution. In pasture, neither groundwater nor these more surficial flow paths had median nitrate concentration that exceeded 111M, hence the capacity of any flow path to deliver nitrate to pasture streams was minimal. The key difference appears to be the combination ofthepresence of both nitrate in soil solution and the operation of near-surface flow paths capable of transporting that. nitrate in forest and the absence of any flow path capable of delivering nitrate to streams in pasture. Markewitz et al. [2001] argued that a shift of water sources to more surficial flow paths in pasture was responsible for higher stream water calcium concentrations in a partial deforested watershed
515
in Paragominas because surficial flows could transport calcium, deposited by forest clearing and burning, from shallow soil depths directly to stream water. The physical conditions in small pasture stream channels can also influence the oxidation-reduction conditions that in turn modify stream water solute concentrations. This was clearly shown in Rondonia, where infilling of streams by aquatic grasses transformed stream habitats from benthic cover of predominantly sand, leaf packs, and fine organic matter in forest, to predominantly grass with less sand and fine organic matter in pasture [Neill et al., 2006b]. This increased water residence time in transient storage zones, decreased oxygen concentrations, and led to very low «111M) stream water nitrate concentrations, presumably largely a result of high rates of denitrification under hypoxic conditions [Neill et al., 2006b]. Where it occurs, this channel infilling by pasture grasses appears to playa major role in determining how small pasture streams function as conduits for N transport in stream networks. The uptake distances of ammonium and nitrate (the number of meters the average molecule travels before it is taken up by the sediments) measured in Rondonia forest streams is long. Using in situ injection of 15N into stream channels, Deegan et al. (submitted manuscript) found ammonium uptake lengths of greater than 1 km and no evidence of nitrate uptake in forest streams. In contrast, ammonium uptake lengths in pasture streams were some 300 to 600 m. While 59% of 15N 'added as ammonium to forest streams was exported as armnonium downstream, 53% of 15NH4+ added to pasture streams was recovered in streamside grasses (Deegan et al., submitted manuscript), indicating that plant uptake may be an impOliant mechanism of nitrate removal fi'om stream water. Because pasture streams .that become infilled by grasses also accumulate large stocks of organic matter, infilling appears to be a mechanism that retains N in small watersheds and prevents downstream movement of inorganic N in stream water. Infilling of pasture channels appears prevalent in some areas (e.g., central Rondonia), while much less widespread and potentially less important in others, but the extent and controls on infilling remain largely unknown. Furthermore, the extent and importance of grass infilling appears to be greatest in small (first and second order) streams. For example, Thomas et al. [2004] found stream water oxygen and nitrate concentrations in third-order pasture streams to be more similar to those in small forest streams than in smaller-order pasture reaches directly upstream. Deegan et al. (submitted manuscript) also found that 15N retention in a third-order pasture stream more closely resembled the low N retention found in forest streams, compared to the high N retention found in small pasture streams, suggesting that
516
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
land use will cause the greatest change in stream N cycling in small, lowland headwater streams. 3.3. Phosphorus Dynamics
The deposition of phosphOlUS to Amazonian forest is very low. Gradients of P inputs to large forested regions may be maintained by long-distance transport of atmospheric dust, which reaches Amazonia largely from Nmih Africa and decreases westward [Swap et al., 1992; Okin et al., 2004]. While these inputs may play an important role in maintaining long-term forest productivity, they do not appear to play an important role in ammal soil P cycles. Little or no weatherable primary mineral P is available on the deep, highly weathered Ultisols and Oxisols that cover most of lowland Amazonia. Large total soil stocks of P are often present in vegetation and soils [Markewitz et al., 2004], but this P is predominantly associated with iron and aluminum in "occluded" forms or as relatively stable organic P [Townsend et al., 2002; Garcia-Montiel et al., 2000; McGrath et al., 2001]. P is widely thought to be the primary limiting element for tree growth. Because total soil P pools are high, enough P cycles through these occluded and organic pools to support the relatively high biomass of Amazonian forest [Johnson et al., 2003]. Phosphorus is highly conserved in forest vegetation, and Amazonian forest trees tend to reabsorb P from leaves and have root mychorrizal associations that facilitate P uptake [Herrera et al., 1978; Vitousek, 1984]. Phosphorus concentrations in soil solution are typically velY low and are controlled primmily by the rapid adsOlption offree P043- to secondmy minerals in soils with high clay content [Johnson et al., 2003]. There is no evidence ofsignificant movement of inorganic P in soil solution [Markewitz et al., 2004], concentrations ofP043- in forest sh'eam water are velY low [Lesack, 1993a; Neill et al., 2001; Biggs et al., 2002; Saunders et al., 2006], and estimated sh'eam water inorganic P exports are less than 0.01 kg P ha- 1 a-I [Markewitz et al., 2004]. Ratios of dissolved inorganic nitrogen to dissolved inorganic phosphorus (DIN:DIP) in sh'eam water in small Amazonian forest streams are high and suggest P is the most important nutrient limitation to primary production, though shaded conditions typically lead to velY low overall rates of primary production [Fittkau, 1967]. Phosphorus additions to forest streams show P is rapidly taken up compared with inorganic N [Neill et al., 2006b]. This suggests the concept of P limitation to production of forest vegetation on weathered Amazonian soils likely applies to control of stream algal production and that small streams act as relatively efficient barriers to downstream inorganic P movement at the low concentrations encountered in forest soil solution and stream water.
3.4. Effects ofLand Use Change on Phosphorus
Controls on the land-water movement of phosphorus is subjected to a potentially different set of mechanisms. Forest clearing for pashlre leads to a brief period of elevated P availability in soils [Garcia-Montiel et al., 2000]. The most available soil inorganic P remains velY low across a wide range of tropical forest and pashlre soils [McGrath et al., 2001]. The most significant shift in pattems of soil P availability following deforestation is a gradual shift toward higher stocks of organic P in pasture compared to forest soils [Garcia-Montiel et al., 2000; Townsend et al., 2002]. It is likely that this organic P plays an important role in regulating P availability in tropical soils [Johnson et al., 2003], but this has not been carefully examined following land use change in the Amazon. Deforestation appears to have little influence on soil solution and groundwater phosphate concentrations. While these have not been measured in a large number of locations, measurements that exist suggest concentrations are extremely low, consistent with the high sorption capacity of the iron and aluminum oxides that dominate highly weathered soils that predominate' in the lowland Amazon Basin [Neill et al., 2001; Markewitz et al., 2004]. Two mechanisms may exeli important influence on the phosphate concenh'ations in small sh·eams. First, pasture sh'eam infilling that leads to hypoxic conditions appears to be associated with elevated phosphate concentrations relative to those in forest sh'eams, presumably because of dissolution of iron and aluminum phosphates in stream sediments under reducing conditions [Neill et al., 2006b]. Second, erosion during overland flow during heavy precipitation is an important mechanism that delivers P to small stream channels in pasture. Biggs et al. [2006] found that overland flow dominated total P export from a pasture hillslope largely because total P concenh'ations in other potential flow paths were very low. Low delively of inorganic N to pasture streams, coupled with potential release ofP from sh'eam sediments, can lead to higher ratios of dissolved N:P in pasture sh'eam water and the development ofN limitation stream algae [Neill et al., 2001]. However, uptake lengths for phosphate in pasture streams are short, suggesting that high light and low absolute concentrations of both DIN and inorganic P lead to rapid uptake by benthic and attached plant and microbial communities. 4. PHYSICAL ANALYSIS OF INTERACTIONS OF LAND USE AND SCALE 4.1. Hydrological Perspective
Catchment studies in the various LBA sites clearly show that the groundwater system plays an important role in the
TOMASELLA ET AL.
generation of both base flow and stonnflow. Microcatchments in Amazonia C\re characterized by the presence of valley floors of variotis dimensions, where the water table 1'1 remains close to tife surface almost all the year. Although the replacement pl forest with pasture and agriculture is associated with the reduction of near surface sahlrated conductivity and a significant increase in Hortonian flow, field evidence indicates that sahlration overland flow remains as the main mechanism of stormflow generation. If the valley floor is the main source of StOlID runoff, it is clear that the proportion of rainfall leaving the catchment as stonnflow will be related to the proportion of the catchment area occupied by valley floor. Perhaps the most clear indication of the valley floor effect can be found in central Amazonia, where most of the catchment shldies are concentrated. Published streamflow data fi'om the 1.3 km 2 Barro Branco catchment [Leopoldo et al., 1995] showed that base flow and stonnflow were 91 % and 9% oftotal discharge, respectively. Lesack [1993b] made detailed hydrological measurements in a 0.23 km2 catchment, draining into Lake Calado, near Manaus. This catchment has similar soils and geology to the Barro Branco. The maximum percentage of eveJ1t rainfall leaving this catchment as stonnflow was 5%. Hodnett et al. (submitted manuscript, 2008) estimated that the stonnflow contribution of the Asu basin varies between 31 % and 45 % of the total discharge. Chauvel et al. [1987a, 1987b] show a block diagram ofthe Bacia Modelo catchment in the area to the north of Manaus. Examination of this diagram shows that as catchment size increases, the width of the valley floor and the proportion of the total catchment area occupied by valley floor also increase. Examination of satellite images of the area also reveals the gradual downstream increase in valley floor area. This implies that the percentage of rainfall leaving catchments as stormflow will increase with catchment size. Small catchments have a mainly V-shaped cross-section, with only a small area of flat valley floor in their lower reaches that produces a small percentage of stormflow. Therefore, in those catchments, with greater drainage area, the proportion of saturation overland flow tends to be higher because the valley floor is larger (U-shaped cross-section). These conclusions should not be considered a "new" contribution of hydrological science, since this observations follows those processes described by Dunne et al. [1975] for tempel:ate humid basins in undishlrbed forests. The phenomenon is, however, unique in the sense of being observed in . tropical forests only on highly permeable clayey soils, with significant rainfall intensities. Therefore, when examining the results of LBA catchment studies, it should be noted that, beside climate and geological considerations, the area of the basin is crucial for un-
517
derstanding the impact of deforestation on the hydrological response. The following table summarizes main findings in terms of the streamflow and stonnflow response in several LBA sites. Figure 2 shows the results of the analysis in terms of the runoff coefficient (the relationship between the annual rainfall and the annual discharge). Disregarding a vast list of factors such as interannual variability, differences in the geology, landscape form, effect of different land use changes on the soil conductivity, and periods of analysis, Figure 2 seems to indicate that the differences in response of a small catchment varies as a function of the area. Since the values in Table I were derived for the whole time of the conesponding study period, they should be analyzed in telIDS of their implication on the long-term water balance (section 2.3), rather than in terms of the overall effect of land use changes on hydrological flow paths. On an annual basis, runoff coefficient provides an indirect indication of catchment evaporation, although Tomasella et al. [2008] indicated the limitations in assuming that the state of the system retums to the original conditions (in central Amazonian basins). Therefore, the more likely explanation for the differences in the mnoff coefficient between deforested and forested catchments should be related to reduction of evaporation. Although, the results of small basins provide clear evidence that deforestation affects water yield after replacing forest by pasture ot· agriculhlre, such effects are sometimes more difficult to detect in larger catchments [Bruijnzeel, 2004]. Costa et al. [2003] detected an increased discharge
45
o
IJ
40
o
35 30
25 20 D
A
15 10
5 0.001
0.01
0.1
10
Drainage Area, km2 ODeforested
A Foresled
Figure 2. Variation ofmnoff coefficient with drainage area in LBA catchments.
518
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
TOMASELLA ET AL.
Table 1. Summary of Results in LBA Catchments in Terms of the Hydrological Response
Basin
Alf-ea
(km 2)
Location
Land Cover
Eastern Amazonia Eastern Amazonia \
Pasture
Rainfall (mm a-I)
Total Discharge (mm a-I)
Runoff Coefficient (%)
5000
Base Flow (%)
~
Source
Deforested
Fazenda Vitoria Cumam WS3
0.0072
Cumam WSI
0.358
Colosso
1.22
Nossa Senhora Rancho Grande
0.122
14.5 0.0073
Eastern Amazonia Central Amazonia SW Amazonia SW Amazonia
1769
296
2253
857
17.3 38.0
Moraes et al. [2006] 98.4
Fallow vegetationAgriculture Fallow vegetationAgriculture Pasture
2253
1602
689
43.0
73.4
Pasture
1918
671-894
35.0--46.6
55.0-66.0
Pasture
2184
378
17.3
Wickel [2004], Wickel et al. [2008]
40.9
98.4
Wickel [2004], Wickel et al. [2008] Trancoso [2006] Biggs et al. [2006] Chaves et al. [2008]
4500
~
2 000 'C\!
4000 3500
(])
~ 3000 ~ ~
0 .......
921
2500
"c\!
~~
(])
J 500
~ . 1000 Q ..d u r:n ......
2500
~ 2000
+-> r:n (]) I-;
519
I
...... ......
1500
~ (]) 1000 Q
~ ~ .......
500
~
500 0
0 1978 1980 1983 1984 1985 [986 [987 1988 1991 1992 1993 1994 1995 1996 1997 1998 [999 2000 2001
Forested
Jumena Oxisol Jutuena Ultisol Fazenda Vitoria AsuMirim
0.0095
Asu
6.56
Rancho Grande
0.0137
0.0194 0.0033 1.26
Southern Amazonia Southern Amazonia Eastern Amazonia Central Amazonia Central Amazonia
Forest
1394
97.5
Johnson et al. [2006]
Forest
1755
96.8
Johnson et al. [2006]
SW Amazonia
Forest
1628
48
3.2
Forest
1582
329
21.0
84.8
Trancoso [2006]
Forest
2514
1071
42.6
40.0
Forest
2184
16.7
0.8
Hodnett et al. (submitted manuscript, 2008) Chaves et al. [2008]
in the Rio Tocantins (400,812 km2), which was attributed to the deforestation in its headwater, an area originally covered to cerrado (savannah type vegetation) but converted to pasture and agriculture. On the other hand, Trancoso [2006] showed that the trends detected in selected stations (with at least 25 years of data) on the Xingu (483,397 km2), Tapaj6s (486,367 lan2 ), and Madeira (3,354,222 km 2 ) rivers can be explained by changes in the same direction of rainfall. This correlation indicates that those changes are not related to a reduced evaporation due to forest conversion (the sources of those rivers are in the severely disturbed area of Amazonia known as the deforestation arc). Perhaps the most significant result is observed in the Ji-Parana Basin (Figure 3) [adapted from Linhares, 2005], which is one of the most impacted mesoscale basins in Amazonia (60% of its original vegetation has been removed), where no trends either in rainfall or streamflow can be detected.
Moraes et al. [2006]
Unlike small catchments, large basins have a variety of land use types associated with intense temporal changes. In Amazonia, specifically along the deforestation are, the remaining forest is confined to an increasingly fragmented area of secondary forests, abandoned agricultural land, and primary remnants [Laurance and Bierregaard, 1997]. As shown by Giambelluca et al. [2003], the magnitude and spatial pattern ofh'anspiration in forest patches is strongly influenced by the conditions in surrounding clearings. According to those authors, transpiration enhancement can occur not only at forest edges, but also well within the patch for trees whose canopies are exposed to advection. They concluded that "fragmentation of remaining forested areas would partly offset the reduction in regional evaporation due to deforestation." This is generally attributed to the input of sensible heat energy to the sUlTounding clearings and secondary younger vegetation. In addition, Holscher et al. [1997] and Sommer
Figure 3. Annual variation of deforestation rates (vertical bars), rainfall and discharge at the Ji-Parana Basin [adapted from Linhares, 2005].
et al. [2002] measured evaporation rates similar to those obtained in mature forests in 2- and 3.5-year-old secondmy vegetation. Thus, the vigorous growth of secondary tropical can cause a rapid return of streamflow totals to predisturbance levels [Bruijnzeel, 2004]. Based on this evidence, a plausible explanation for the fact that the differences in the signal between forested and deforested basins cannot be detected at larger scales could be related to the aggregation of the evaporation processes in a fragmented landscape. Figure 4 presents the proportion of base flow to total rainfall as a function of drainage area. Because of the geomorphological characteristic of the dissected Amazonian valleys, the increase of scale is associated with an increase in the relative contribution of the valley/riparian area (Hodnett et aI., submitted manuscript, 2008), which is the most important area for stormflow generation. This explains why there is an increase of stormflow contribution with area, both in forested and deforested catchments. Because the compaction of soils, particularly in heavily grazed areas, Hortonian over- . land flow is more likely to occur in areas of low infiltration [Zimmerllllllan et al., 2006], which explains the differences between forested and deforested areas. In conclusion, there is clearly an incomplete knowledge of the functioning offorested patches, both from the hydrologi-
cal and meteorological point of view. The influence of the degree of fragment~tionon physical processes such as evaporation and surface runoff are not fully understood yet. There is an open question on the potential effect of deforestation on hydrological response, in pmiicular, how different spatial
0 0~
040 35 30
;jj! c:
25
~
20
q:: Ql
15
'"
10
'~ro 0
1Il
(ll
0 0 A
A
5 0.001
0.01
0.1
10
Drainage Area, km 2
Figure 4. Percentage of base flow as a function ofthe drainage area in LBA catchments.
520
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
patterns of deforestation, in relation to topography, affect the overall hydrological response [Giambelluca, 2002].
4.2. Nutrient j))!llamic Perspective At larger scales, in river basins ofhundreds to > I0,000 km2 , the patterns that appear in small watersheds largely disappear. Bigg§ et a1. [2004] in Rondonia found a relatively weak positive association between the dry season nitrate concentration in river water and the percentage ofwatershed deforested but no relationship in the wet season. They also found higher total dissolved P in river water in deforested watersheds, but only in watersheds that were >65% deforested. Ballester et a1. [2003] also found a relationship between the percentage of watershed in pasture and phosphorus concentration in river water among 14 major tributaries ofthe Rio Ii-Parana in Rondonia. At scales slightly larger than the first- and secondorder basins that are largely the focus of this review, higher water velocities and greater channel incising in larger pasture streams may be responsible for this pattern, but this has not been examined in detail. Both Biggs et al. [2004] and Ballester et al. [2003] found that soil base cation status played a large role in influencing river water chemistty and, in many cases, was a greater influence than land use. The inclusion of urban areas also plays an impOliant role in driving total dissolved Nand P concentrations at the scale of larger rivers.
TOMASELLA ET AL.
Future studies should be focused more on heterogeneous landscapes, particularly on the effects of secondaly vegetation and different land uses. Therefore, an improved land cover representation ofthe surface is urgently needed. These new products should have not only higher spatial resolution, but also include information regarding land use histOly, age of pasture/agriculture/secondary vegetation, and the relationship between topography and land cover patterns. There is a need for research to assess the downstream effects of land use change in mesoscale and macroscale catchments
[Bruijnzeel,2004]. On the modeling side, there is a clear need for more detailed hydrological and atmospheric models, able to represent the nonlinear interactions between land patches and forest. Finally, it is important to increase the integration between traditional paired catchments studies with modeling effort and physically based models. New methodologies for upscaling local hydrological processes in macroscale models are urgently needed. In addition, it is crucial to develop techniques for parameter estimations in ungauged basins, considering the lack of data in extensive areas of the Amazon Basin. Any critical assessment of land conversion basinwide could be considered reliable only if the modeling tools are based on sound and verifiable physical principles.
5. CONCLUSIONS
REFERENCES
Studies conducted during LBA provided new insights into the mechanisms of runoff generation in tropical basins, and how these mechanisms are altered by land use and land cover changes. The most impOliant characteristic of small Amazonian catchments is, undoubtedly, the role of the groundwater system. It is clear that base flow is the most impOliant component of streamflow produced by small streams. The influence of base flow is more impOliant in headwater catchments, and it is reduced gradually with the increase of the scale, closely related to topographic features (width of the riparian zone). Regarding stormflow generation, it is clear that saturation overland flow is the most important process in forested catchments. However, new evidence was found in headwater catchments indicating that return flow can be crucial in producing stormflow during the most intense rainfall events, probably related to the throttle layer effect. There are several open issues related to how most ofhydrological processes measured at small scales are aggregated at larger scales (from thousands to million k:m2 ). There is preliminaly evidence indicating that effects clearly detectable at small scales are "diluted" at large scales, probably related to the influence of forest fragmentation.
Andreassian, V. (2004), Waters and forests: From historical controversy to scientific debate, J. Hydrol., 291,1-27. Ballester, M. V. R., D. C. Victoria, A V. K.rusche, R. Coburn, R. L. Victoria, I. E. Richey, M. G. Logsdon, E. Mayorga, and E. Matricardi (2003), A remote sensing/GIS-based physical template to understand the biogeochemistly of the Ji-Parana river basin, Remote Sens. Environ., 87, 429-445. Biggs, T. W., T. Dunne, T. F. Domingues, and L. A Martinelli (2002), Relative influence of natural watershed properties and human disturbance on stream solute concentrations in the southwestern Brazilian Amazon basin, Water Resoll/·. Res., 38(8), 1150, doi:lO.l029/2001WR000271. Biggs, T. W., T. Dunne, and L. A Martinelli (2004), Nahlral controls and human impacts on stream nutrient concentl·ations in a deforested region of the Brazilian Amazon basin, BiogeochemistlJI, 68, 227-257. Biggs, T. W., T. Dmme, and T. Muraoka (2006), Transport ofwatel', solute and nutrients from a pashlre hillslope, southwestern Brazilian Amazon, Hydrol. Processes, 20, 2527-2548. Bonell, M. (1993), Progress in the understanding of runoff generation in forests,J. Hydrol., 150, 217-275. Bonell, M., and I. Balek (1993), Recent scientific developments and research needs in hydrological processes in the humid tropics, in Hydrology and Water Management in the Humid Tropics,
edited by M. Bonell, M. M. Hufschmidt, and J. S. Gladwell, pp. 167-260, Cambridge l)hiv.Press, Cambridge. Bouwmann, A F., D. P/van Vuuren, R. G. Derwent, and M. Posch (2002), A global ij;llalysis of acidification and eutrophication of terrestrial ecosys~ms, Water Air Soil POllllt., 141, 349-382. Brandes, I. A, M:E. McClain, and T. P. Pimentel (1996), 15N evidence for the origin and cycling of inorganic nitrogen in a small Amazonian catchment, BiogeochemistlJ', 34, 45-56. Bravard, S., and D. Righi (1989), Geochemical differences in an Oxisol-Spodosol toposequence of Amazonia, Brazil, Geoderma, 44,29-42. Bruijnzeel, L. A (1990), Hydrology of Moist Tropical Forest and Effect ofForest Conversion: A State o.l'Knowledge Review, 226 pp., UNESCO, Paris and Vrije University, Amsterdam, The Netherlands. Bruijnzeel, L. A (2004), Hydrological functions of tropical trees: Not seeing the soil for the trees? Agric. Ecosyst. Environ., 104, 185-228. Chauvel, A (1982), Os latossolos amarelos, alicos, argilosos dentl·o dos ecossistemas das bacias experimentais do INPA e da regiao vizinha, Acta Amazonica, 12,47-60. Chauvel, A, J. L. Guillaumet, and H. O. R. Schubart (1987a), Importance et distribution des racines et des etres vivants dans un latossol argileux sous foret amazonienne, Rev. Ecol., BioI. Sol, 24,19-48. Chauvel, A, Y. Lucas, and R. Boulet (1987b), On the genesis of the soil mantle of the region of Manaus, centl'al Amazonia, Brazil, Experientia, 43,234-241. Chaves, I., C. Neill, S. Germer, S. G. Neto, A. Krusche, and H. Elsenbeer (2008), Land management impacts on runoff sources in small Amazon watersheds, Hydrol. Processes, 22, 17661775. Chaves, J., C. Neill, S. Germer, S. Gouveia Neto, A V. Krusche and H. Elsenbeer (2009), Evidence of nitrogen loss in Amazon forest and pasture soils, Ecosystems, in press. Costa, M. H., A Botta, and J. A. Cardille (2003), Effects of largescale changes in land cover on the discharge of the Tocantins River, southeastern Amazonia, J. Hydrol., 283, 206-217. Cuartas, L. A, I. Tomasella, A D. Nobre, M. G. Hodnett, M. I. Waterloo, and J. C. M11llera (2007), Interception water-partitioning dynamics for a pristine rainforest in Centl'al Amazonia: Marked differences between normal and dry years, Agric. For. Meteorol., 145,69-83. Davidson, E. A., M. C. C. Bustamante, and A de S. Pinto (2002), Emissions of nitrous oxide and nitric oxide from soils of native and exotic ecosystems of the Amazon and cerrado regions of Brazil, in Optimizing Nitrogen1l1anagemellt in Food and Energy Production and Environmental Protection, edited by J. N. Galowayet aI., pp. 312-319, Balkema, Rotterdam, Netherlands. Davidson, E. A, C. Neill, A V. Krusche, M. V. R. Ballester, D. Markewitz, and R. de O. Figueiredo (2004), Loss of nutrients' from terrestrial ecosystems to streams and the atmosphere following land use change in Amazonia, in Ecosystems and Land Use Change, Geophys. Monogr. Ser., vol. 153, edited by R. DeFries, G. Asner, and R. H. Houghton, pp. 147-158, AGU, Washington, D. C.
521
Dias, A C. C. P., A D. S. Neves, and R. C. M. Barbosa (1980), Levantamento de solos da Estayao Experimental Rio Negro, Boletim Tecnico da CEPLAC, 71, 1-13. Dunne, T., T. R. Moore, and C. H. Taylor (1975), Recognition and prediction of runoff-producing zones in humid regions, Hydro!. Sci. Bull., 20, 305-327. Elsenbeer, H. (2001), Hydrologic flowpaths in tropical forest soilscapes-A review, Hydro!. Processes, 15,1751-1759. Elsenbeer, H., and R. A Vertessy (2000), Stonnflow generation and flowpath characteristics in an Amazonian rainforest catchment, Hydro!. Processes, 14,2367-2381. Elsenbeer, H., B. E. Newton, T. Dunne, and I. M. de Moraes (1999), Soil hydraulic conductivities of latosols under pasture, forest and teak in Rondonia, Brazil, Hydrol. Processes, 13, 14171422. Filoso, S., M. R. Williams, and J. M. Melack (1999), Composition and deposition ofthroughfall in a flooded forest archipelago (Negro River, Brazil), BiogeochemistlJI, 45,169-195. Fittkau, E.-I. (1967), On the ecology ofAmazonian rain-forest stl'eams, Atlas do Simposio sabre a Biota Amazonica, Rio de Janeiro, 3, 97. Franken, W. (1979), Untersuchungen im Einzugsgebiet des Zentralamazonischen Urwaldbaches 'Barro Branco auf del' 'terra finne.' 1. Abflussverhalten des Baches, Amazoniana, 6, 459-466. Franken, W., and P. R. Leopoldo (1984), Hydrology of catchment areas of Central-Amazonian forest streams. Ch 19, in: The Amazon, Limnology and Landscape Ecology 0.1' a Mighty Tropical River and Its Bashr, edited by H. Sioli, pp. 501-519, Springer, Dordrecht, The Netherlands. Franken, W., and P. ~. Leopoldo (1986), Relayoes entre fluxos de agua subterranea ~ superficial em bacia hidrografica caJ'acterizada pOl' cobertura.florestal Amazonica, Acta Amazonica, 16/17, 253-262. Garcia-Montiel, D., C. Neill, I. M. Melillo, S. M. Thomas, P. A. Steudler, and C. C. Cerri (2000), Soil phosphorus transformations after forest clearing for pashlre in the Brazilian Amazon, Soil Sci. Soc. Am. J., 64,1792-1804. Gash, I. H. c., C. R. Lloyd, and G. Lachaud (1995), Estimating sparse forest rainfall interception with an analytical model, J. Hydrol., 170, 79-86. Germer, S., H. Elsenbeer, and I. M. Moraes (2006), Throughfall and temporal trends of rainfall redistl'ibution in an open tropical rainforest, south-western Amazonia (Rondonia, Brazil), Hydrol. Earth Syst. Sci., 10, 383-393. Germer, S., C. Neill, A V. Krusche, S. Gouveia Neto, and H. Elsenbeer (2007), Seasonal and within-event dynamics of rainfall and throughfall chemistly in an open tropical rainforest in Rondonia, Brazil, BiogeochemistlJI, 86, 155-174. Germer, S., C. Neill, T. Vetter, I. Chaves, A V. Krusche, and H. Elsenbeer (2009), Implications of long-term land-use change for the hydrology and solute budgets of small catchments in Amazonia, J. Hydrol., 364, 349-363. Giambelluca, T. W. (2002), Hydrology of altered tropical forest, Hydro!. Processes, 16,1665-1669. Giambelluca, T. W., A D. Ziegler, M. A Nullet, D. M. Truong, and L. T. Tran (2003), Transpiration in a small tropical forest patch, Agric. For. Meteoro!., 117, 1-22.
522
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
Godsey, S., and H. Elsenbeer (2002), The soil hydrologic response to forest regrowth: A case study from southwestern Amazonia, Hydrol. Pmcesses, 6, 1519-1522. Hauglustaine, nt. A., F. Hourdin, L. Jourdain, M.-A. Filiberti, S. Walters, I.-F. Lamarque, and E. A Holand (2004), Interactive chemistry in the Laboratoire de Meteorologie Dynamique general circulation model: Description and background tropospheric chemistry evaluation, J. Geophys. Res., 109, D04314, doi: 10.1 029/2003JD003957. Herrera, R., T. Merida, N. Stark, and f.=. Jordan (1978), Direct phosphorus tJ'ansfer fromleaflitter to roots, Natul'lvissenschajten, 65, 208-209. Hodnett, M. G., M. D. Oyama, I. Tomasella, and A de O. Marques Filho (I 996a), Comparisons oflong-term soil water storage behaviour under pasture and forest in three areas of Amazonia, in Amazonian Deforestation and Climate, edited by J. H. C. Gash et al., pp. 57-77, John Wiley, Chichester, U. K. Hodnett, M. G., I. Tomasella, A de O. Marques Filho, and M. D. Oyama (1996b), Deep soil water uptake by forest and pasture in central Amazonia: Predictions from long-term daily rainfall using a simple water balance model, in: Amazonian Deforestation and Climate, edited by I. H. C. Gash et al., pp 79-99, John Wiley, Chichester, U. K. Hodnett, M. G., I. Vendrame, A de O. Marques Filho, M. D. Oyama, and I. Tomasella (1997a), Soil and groundwater behaviour in a catenmy sequence beneath forest in Central Amazonia: I. Comparisons between plateau, slope and valley floor, Hydm/. Earth Syst. Sci., 1,265-277. Hodnett, M. G., I. Vendrame, M. D. Oyama, A de O. Marques Filho, and J. Tomasella (1997b), Soil water storage and groundwater behaviour in a catenmy sequence beneath forest in Central Amazonia. II. Floodplain water table behaviour and implication for streamflow generation, Hydrol. Earth Syst. Sci., 1, 279290. Holland, E. A, et al. (1997), Variations in the predicted distribution of atmospheric nitrogen deposition and their impact on carbon uptake by terrestrial ecosystems, J. Geophys. Res., 102, 15,849-15,866. Holscher, D., T. D. de A Sa, T. X. Bastos, M. Denisch, and H. FoIster (1997), Evaporation from young secondary vegetation in eastern Amazonia, J. Hydrol., 193, 293-305. Jipp, P. H., D. C. Nepstad, D. K. Cassel, and C. R. Carvalho (1998), Deep soil moisture storage and transpiration in forests and pashlres of seasonally dry Amazonia, Clim. Change, 39, 395-412. Johnson, A H., I. Frizano, and D. R. Vann (2003), Biogeochemical implications of labile phosphorus in forest soils determined by the Hedley fi'actionation procedure, Gecologia, 135, 487-499. Johnson, M. S., J. Lehmann, E. C. Selva, M. Abdo, S. Riha, and E. G. Couto (2006), Organic carbon fluxes and streamwater exports from headwater catchments in the Southern Amazon, Hydm/. Processes, 20, 2599-2614. Lal, R. (1987), Tropical Ecology and Physical Edaphology, 732 pp., Wiley, New York. Laurance, W. F., and R. O. Bierregaard Jr. (1997), Preface: A crisis in the making, in Tropical Forest Remnants: Ecology, lYIan-
agement and Conservation ofFragmented Communities, edited by W. F. Laurance and R. O. Bierregaard Jr., Univ. of Chicago Press, Chicago, Ill. Leopoldo, P. R., W. Franken, E. Matsui, and I. Salati (1982), Estimativa de evapotranspira<;ao de floresta amazonica de terra finne, Acta Amazonica 12(Suppl), 23-28. Leopoldo, P. R., W. Franken, and E. Matsui (1984), Hydrological aspects of the tropical rainforest in Central Amazon, Interciencia, 9, 125-131. Leopoldo, P. R., W. Franken, and E. Matsui (1985), Hydrological aspects of the tropical rainforest in the central Amazon. Ch. 7, in Change in the Amazon Basin, Vol 1. NIall 's Impact on Forests and Rivers, edited by J. Hemming, 222 pp., Manchester Univ. Press, Manchester, U. K. Leopoldo, P. R., W. Franken, E. Salati, and M. N. G. Ribeiro (1987), Towards a water balance in Central Amazonian region, Experientia, 43, 222-233. Leopoldo, P. R., W. Franken, and N. A Villa Nova (1995), Real evapo-transpiration and transpiration through a tropical rainforest in central Amazonia as estimated by the water balance method, For. Ecol. Manage., 73, 185-195. Lesack, L. F. W. (1993a), Export of nutrients and major ionic solutes from a rain forest catchment in the central Amazon Basin, Water Resow'. Res., 29(3), 743-758. Lesack, L. F. W. (1993b), Water balance and hydrologic characteristics of a rain forest catchment in a central Amazon basin, Water Resow'. Res., 29(3), 759-773. Linhares, C. A (2005), Influencia do desflorestamento na diniimica da resposta hidrol6gica na bacia do Rio Ii-Parana IRO, tese de doutorado em Sensoriamento Remoto, INPE. Luizao, R. C., F. J. Luizao, R. Q. Paiva, T. F. Monteiro, L. S. Sousa, and B. Kruijt (2004), Variation of carbon and nitrogen cycling processes along a topographic gradient in a central Amazonian forest, Global Change Bioi., 10, 592600. Markewitz, D., E. A. Davidson, R. O. Figueiredo, R. L. Victoria, and A. V. Krusche (2001), Control of cation concentrations in stream waters by surface soil processes in an Amazonian watershed, Nature, 410, 802-805. Markewitz, D., E. A Davidson, P. Moutinho, and D. Nepstad (2004), Nutrient loss and redistribution after forest clearing on a highly weathered soil in Amazonia, Ecol. Appl., 14, S177S199. . Martinelli, L. A, M. C. Piccolo, A R. Townsend, P. M. Vitousek, E. Cuevas, W. McDowell, G. P. Robertson, O. C. Santos, and K. Treseder (1999), Nitrogen stable isotopic composition of leaves and soil: Tropical versus temperate forests, Biogeochemistry, 46, 45-65. McClain, M. E., I. E. Richey, and T. P. Pimentel (1994), Groundwater nitrogen dynamics at the terrestrial-Iotic interface of a small catclunent in the Central Amazon Basin, Biogeochemistl)', 27,113-127. McGlynn, B. L., and J. I. McDonnell (2003), Quantifying the relative contributions of riparian wand hillslope zones to catchment runoff, Water Resow'. Res., 39, 1310, doi: 10.10291 2003 WR002091.
TOMASELLA ET AL. McGrath, D. A, C. K. Smith, H. L. Gholz, and F. de A Oliveira (2001), Effects of land"use change on soil nutrient dynamics in Amazonia, Ecosystelfl~', 4, 625-645. Melillo, J. M., P. A ~teudler, B. I. Feigl, C. Neill, D. Garcia, M. C. Piccolo, C. c. C;Jh and H. Tian (2001), Nitrous oxide emissions from forests and pastures of various ages in the Brazilian Amazon, J. Geophys. Res., 106, 34,179-34,188. Moraes, I. M., A Schuler, T. Dunne, R. O. Figueiredo, and R. L. Victoria (2006), Water storage and runoff processes in linthic soils under forest and pasture in eastern Amazon, Hydm/. Pmcesses, 20, 2509-2526. Neill, C., M. C. Piccolo, C. C. Cerri, P. A. Steudler, I. M. Melillo, and M. Brito (1997), Net nitrogen mineralization and net nitrification rates in soils following deforestation for pashlre across the southwestern Brazilian Amazon Basin landscape, Gecologia, 110, 243-252. Neill, c., M. C. Piccolo, I. M. Melillo, P. A Steudler and C. C. Cerri (1999), NitJ'ogen dynamics in Amazon forest and pasture soils measured by 15N pool dilution, Soil Bioi. Biochem., 31, 567-572. Neill, c., L. A Deegan, S. M. Thomas, and C. C. Cerri (2001), Deforestation for pashtre alters nitrogen and phosphOlus in soil solution and streamwater of small Amazonian watersheds, Ecol. Appl., ll, 1817-1828. Neill, C., M. C. Piccolo, C. C. CelTi, P. A Steudler,.and J. M. Melillo (2006a), Soil solution and nitrogen oxide losses during clearing of lowland Amazon forest for cattle pashlre, Plant Soil, 281,233-245. Neill, C., L. A Deegan, S. M. Thomas, C. L. Haupeli, A V. K.lUsche, V. M. Ballester, and R. L. Victoria (2006b), Deforestation alters channel hydraulic and biogeochemical characteristics of small lowland Amazonian streams, Hydrol. Processes, 20, 2563-2580. Nortcliff, S., and I. B. Thornes (1978), Water and cation movement in a tropical rainforest environment: I. Objectives, experimental design and preliminary results, Acta Amazonica, 8, 245-258. Nortcliff, S., and J. B. Thomes (1981), Seasonal variations in the hydrology of a small forested catchment near Manaus, Amazonas, and its implications for management. Ch. 2.2, in Tmpical Agricultural Hydrology, edited by R. Lal and E. W. Russell, pp. 37-57, Jolm Wiley, New York. Nortcliff, S., and I. B. Thornes (1984), Floodplain response of a small tropical stream. Ch. 5, in Catchment Experiments in Fluvial GeomOlphology, edited by T. P. Burt and D. E. Walling, pp. 73-85, Geo Books, Norwich, U. K. Nortcliff, S., and I. B. Thornes (1989), Variations in soil nutrients in relation to soil moisture stahts in a tropical forested ecosystem, in Mineral Nutrients in Tropical Forest and Savanna Ecosystems, edited by I. Proctor, pp. 43-54, Oxford Univ. Press, Oxford, U. K. Nortcliff, S., I. B. Thornes, andM. I. Waylen (1979), Tropicalforest systems: A hydrological approach, Amazoniana, 6, 557-568. Nortcliff, S., S. M. Ross, and I. B. Thornes (1990), Soil moisture, . runoff and sediment yield from differentially cleared tropical rainforest plots. Ch. 25, in Vegetation and Erosion, edited by I. B. Thornes, pp. 419-436, Jolm Wiley, Chichester, U. K. Okin, G. S., N. Mahowald, O. A Chadwick, and P. Artaxo (2004), Impact of desert dust on the biogeochemistly of phosphorus in
523
terrestrialecosystems, Global Biogeochem. Cycles, 18, GB2005, doi: 10.1 029/2003GB002145. Piccolo, M. C., F. Andreux, and C. C. Cerri (l994a), Hydrochemistiy of soil solution collected with tension-free lysimeters in native and cut-and-burned tropical rain forest in central Amazonia, Geochim. Brasilien., 8,51-63. Piccolo, M. c., C. Neill, and C. C. Cerri (1994b), Net nitrogen mineralization and net nitrification along a tropical forest-to-pasture chronosequence, Plant Soil, 162,61-70. Ranzani, G. (1980), Identifica<;ao e caracteriza<;ao de alguns solos da Esta<;ao Experimental de Silviculhu'a Tropical do INPA, Acta Amazonica, 10, 7-41. Roberts, I. M., O. M. R. Cabral, I. P. da Costa, A-L. C. McWilliam, and T. D. A Sa (1996), An overview ofthe leaf area index and physiological measurements during ABRACOS, in Amazon Deforestation and Climate, edited by J. H. C. Gash et al., pp. 287-306, Jolm Wiley, Chichester, U. K. Ross, S. M., I. B. Thornes, and S. Nortcliff (1990), The Maraca rainforest Project, II. Soil hydrology, nutrient and erosional response to the clearance of terra firme forest, Maraca Island, Roraima, northern Brazil, Geogr. J., 156, 267-282. Saunders, T. J., M. E. McClain, and C. A Llerena (2006), The biogeochemistry of dissolved nitrogen, phospholUs and organic carbon of terrestrial-aquatic flowpaths in a small montane catchment ofthe Peruvian Amazon, Hydm/. Processes, 20, 2549-2562. Schuler, A E. (2003), Fluxos hidrol6gicos em microbacias com floresta e pastagem na Amazonia Oriental, Paragominas, Para, Tese apresentada ap CENAIUSP, 119 pp. Sommer, R., T. D. de A Sa, K. Vielhauer, A Carioca de Araujo, H. FOIster, and P. ,L. G. Vlek (2002), Transpiration and canopy conductance of secondmy vegetation in the eastern Amazon, Agric. For. Meteom/., ll2, 103-121. Swap, R., M. Garstang, S. Greco, R. Talbot, andP. Kallberg (1992), Saharan dust in the Amazon Basin, Tellus, Ser. B, 44,133-149. Thomas, S. M., C. Neill, L. A. Deegan, A V. K.l.usche, R. Victoria, and M. V. Ballester (2004), Influences of land use and stream size on particulate and dissolved materials in a small Amazonian stream network, Biogeochemistl)', 68, 135-151. Tomasella, I., and M. G. Hodnett (1996), Soil hydraulic propeliies and van Genuchten parameters for an Oxisol under pasture in central Amazonia, in: Amazonian Deforestation and Climate, edited by I. H. C. Gash et al., pp. 101-124, John Wiley, Chichester, U. K. Tomasella, I., M. G. Hodnett, L. A Cuartas, A D. Nobre, M. I. Waterloo, and S. M. Oliveira (2008), The water balance of an Amazonian micro-catchment: The effect of interannual variability of rainfall on hydrological behaviour, Hydrol. Processes, 22, 2133-2147. Townsend, A. R., G. P. Asner, C. C. Cleveland, M. E. Lefer, and M. M. C. Bustamante (2002), Unexpected changes in soil phosphorus dynamics along pashlre clu'onosequences in the humid tropics, J. Geophys. Res., 107(D20), 8067, doi:IO.10291 2001JD000650. Trancoso, R. (2006), Mudan<;as na cobertura da terra e altera<;oes na resposta hidrol6gica de bacias hidrograficas da Amazonia, disserta<;ao apresentada ao PPGBT-INPAIUFAM, 139 pp.
524
WATER AND CHEMICAL BUDGETS AT THE CATCHMENT SCALE
Trumbore, S. E., E. A. Davidson, P. B. de Camargo, D. C. Nepstad, and L. A. Martinelli (1995), Below-ground cycling of carbon in forests and pastures of eastern Amazonia, Global Biogeochem. Cycles, 9, 5f5~528. Uhl, C., and C. F. Jordan (1984), Succession and nutrient dynamics following forest cutting and burning in Amazonia, Ecology, 65, 1476-1490. Verchot, I,. V., E. A. Davidson, 1. H. Cattilnio, 1. L. Ackerman, H. E. Erickson, and M. Keller (1999), Land-use change and biogeochemical controls of nitrogen oxide emissions from soils in eastern Amazonia, Global Biogeochem. Cycles, 13, 31-46. Vitousek, P. M. (1984), Litterfall, nutrient cycling and nutrient limitation in tropical forests, Ecology, 65,285-298. Waterloo, M. 1., et al. (2006), Export of organic carbon in run-off from an Amazonian rainforest blackwater catchment, Hydral. Processes, 20, 2581-2597. Wickel, A. 1., N. C. van de Giesen, and T. D. A. Sa (2008), Stormflow generation in two headwater catchments in eastern Amazonia, Brazil, Hydrol. Processes, 22, 3285-3293. Wickel, B. (2004), Water and nutrient dynamics of a humid tropical agricultural watershed in Eastern Amazonia, Ecology and Development Series, 21, 120 pp., Cuvillier, G6ttingen.
Williams, M. R., and J. M. Melack (1997), Solute export from forested and partially deforested catchments in the central Amazon, Biogeochemistl)', 38, 67-102. Williams, M. R., T. R. Fisher, and 1. M. Melack (1997), Solute dynamics in soil water and groundwater in a central Amazon catchment undergoing deforestation, Biogeochemistl)l, 38, 303-335. Zhang, L., W. R. Dawes, and G. R. Walker (2001), Response of mean annual evapotransportation to vegetation changes at catchment scale, Water ResoU/·. Res., 37(3),701-708. Zimmermann, B., H. Elsenbeer, and 1. M. Moraes (2006), The influence of land-use changes on soil hydraulic properties for runoff generation, For. Ecol. Manage, 222, 29-38. R. Figueiredo, Embrapa Amazonia Oriental, Trav. Dr. Eneas Pinheiro sin°, Belem, PA 66095-100, Brazil. C. Neill, Marine Biological LaboratOly, The Ecosystems Center, 7 MBL Street, Woods Hole, MA 02543, USA. A. D. Nobre, Instituto Nacional de Pesquisas da Amazonia, Escrit6rio Regional, Avenida dos Astronautas, 1758, Sao Jose dos Campos, SP 12227-010, Brazil. 1. Tomasella, Centro de Previsao do Tempo e Estudos ClimMicos, 1NPE, Rod. Presidente Dutra km 39, Cachoeira Paulista, SP 12630-00, Brazil. ([email protected])
Floodplain Ecosystem Processes John M. Melack, I Evlyn M. L. M. NoVO,2 Bruce R. Forsberg,3 Maria T. F. Piedade,3 and Laurence Maurice4 Floodplains represent a major component of the central Amazon Basin and influence the hydrology, ecology, and biogeochemistry. Hess et al. (2003) used a classification of synthetic aperture radar data with 100 m resolution for a 1.77 million km 2 quadrat in central Amazonia and identified 17% as wetland most of which was inundated a portion of each year. Total net production attributed to flooded forests (excluding wood increments), aquatic macrophytes, phytoplankton, and periphyton for the 1.77 million km2 quadrat was estimated to be about 300 Tg C a-I. Flooded forests accounted for 62% of the total, aquatic macrophytes accounted for 34%, and the remaining 4% was associated with periphyton and phytoplankton. Approximately lO% of the total is the amount of organic carbon exported annually by the Amazon River according to Richey et al. (1990), methane emission is about 2.5% according to Melack et al. (2004), and a similar percent is estimated to be buried in sediments. The remaining poiiion is close to being sufficient to fuel the respiration that results in the degassing of 21 0 ± 60 Tg C a-I as carbon dioxide from the rivers and floodplains according Ito Richey et al. (2002). Variations in the distribution and inundation of floodplain habitats playa key role in the ecology and production of many commercially important freshwater fish. A significant relationship exists between maximum inundated area lagged by 5 years and annual yield of omnivores.
1. INTRODUCTION Floodplains are important components of the biogeochemistlY, ecology, and hydrology of the lowland Amazon I Bren School of Environmental Science and Management and Department of Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, California, USA. 2Instituto Nacional de Pesquisas Espaciais, Sao Jose dos Campos, Brazil. 3Instituto Nacional de Pesquisas da Amazonia, Manaus, Brazil. 4LMTG, Universite de Toulouse, Toulouse, France.
Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000721
Basin. Amazon floodplains contain thousands of lakes and associated wetlands linked to one another and to the many rivers and streams of the basin. These floodplains modify hydrology, influence carbon and nutrient biogeochemistly, emit carbon dioxide and methane to the atmosphere, and support highly diverse ecosystems and productive fisheries. Much of the progress made during the last 50 years toward understanding the ecology of Amazon floodplains is summarized by Sioh [1984], Junk and Piedade [1997], Melack and Forsberg [2001], and Junk and Piedade [2009]. Recent efforts to develop sustainable management options for these environments are covered by Junk et al. [2000]. Our purpose is to review and synthesize recent results largely derived from research associated with the Large-Scale BiosphereAtmosphere Experiment in Amazonia with an emphasis on biogeochemical and related hydrological processes and the results ofremote sensing analyses that permit regionalization 525
526
FLOODPLAIN ECOSYSTEM PROCESSES
distinct polygons, the mask was edited interactively to eliminate nonwetlands features such as hilly terrain, which were spectrally similar to wetlands but could be correctly identified by an interpreter based on geographical context, spatial pattern, and cross-referencing with Landsat scenes and topographic maps. The wetland pOliion of the coregistered high- and low-water mosaics was mapped as vegetative2. REGIONAL EXTENT OF FLOODPLAINS hydrologic classes using a rules-based classifier. The pixelAND AQUATIC HABITATS based classifier was based on backscattering coefficients. The Floodplains and other wetlands' in the Amazon Basin have classification system included five vegetation classes (nonpreviously been estimated to cover approximately 1 million vegetated, herbaceous, shrub, woodland, and forest) and two square kilometers and include seasonally inundated forests hydrologic classes (flooded and nonflooded). Both the wetand savannas, riparian zones bordering streams and rivers, land mask and vegetation classes were rigorously validated backwater swamps, and high-elevation bogs [Junk, 1997]. using airborne digital videography [Hess et al., 2002]. Seventeen percent of the central quadrat was identified Large seasonal variations in depth and extent of inundation and complex hydrology are characteristic of Amazon flood- as wetland, of which 96% was inundated at high water and plains, and, as water levels vary, the proportion of aquatic 26% was inundated at low water [Hess et al., 2003]. While habitats changes considerably. Characterization of the areal flooded forest occurred in nearly 70% of the wetlands at extent and temporal changes of inundation and wetland veg- high water, aquatic habitats varied regionally as a function etation on local, regional, and basin-wide scales is now pos- of geomorphology and environmental conditions (Plate 1). sible because of the availability of optical and microwave Extending the mapping to the whole lowland Amazon Badata from sensors on aircraft and spacecraft and recently sin (the region less than 500 m' above sea level), Melack developed algorithms for data analysis [Melack, 2004]. and Hess [2004, 2009] mapped 14% of the 5,822,000 la112 Application of spectral mixture analysis to Landsat multi- as floodable at 100-m resolution (Figure 1), of which about spectral scmmer and thematic mapper imagely has permit- 76% is represented by floodable woody vegetation and 8% ted calculation of concentrations of suspended sediments in is represented by open water. Thematic accuracy for the surface waters of the Amazon River [Mertes et al., 1993] whole lowland area was evaluated based on aerial videoand discrimination of patterns in aquatic vegetation [Mertes graphy [Hess et al., 2002] for most of the Brazilian Amazon. et al., 1995; Novo et al., 1997]. Monthly variations in re- For regions where the vegetation but not flood timing was gional-scale inundation can be derived with mixing models likely to be represented by the videography or for regions applied to passive microwave data [Sippel et al., 1994]. De- where neither vegetation nor flood timing was likely to be lineation of flooding status and vegetation, with accuracies well represented in the videography, qualitative evaluations greater than 90%, has been demonstrated using multifre- were done based on published vegetation surveys. To assess quency, polarimetric synthetic aperture radar (SAR) data for the timing of the JERS mosaics relative to typical flooding floodplains in central Amazonia [Hess et al., 1995], and the patterns, the seasonality of precipitation and river discharge theoretical basis for the algorithms is supported by modeling were compared with the acquisition dates ofthe scenes composing the mosaics. [Wang et al., 1995]. To estimate the monthly variations in inundation for the A key data set is the L band HH-polarized SAR imagely of the entire Amazon Basin acquired by the National Space De- 1.77 million la112 quadrat in central Amazonia, Richey et al. velopment Agency of Japan's JERS-l satellite [Rosenqvist et [2002] combined analyses of the JERS-l mosaic for rivers al., 2000]. The acquisitions were timed to low water in 1995 wider than about 100 m [Hess et al., 2003], passive microand high water in 1996 for the main stem of the Amazon wave data for the main stem Amazon and its floodplain [Sipin the central basin. A mosaic of individual scenes for each pel et al., 1998], and mean monthly river stage data from period was developed as digital data sets with 3~arc sec (ap- tributaries to approximate the temporal flooding patterns for proximately 100 m) resolution [Siqueira et al., 2000]. Hess these rivers. To account for river corridors less than 100 m in et al. [2003] developed a methodology for classification of width, an area-stream density function was computed from Amazonian wetlands using JERS-I radar mosaics for a 1.77 a digital river network and extrapolated to the smaller rivers million la11 2 quadrat (18° x 8°) in central Amazonia. The hy- [Mayorga et al., 2005b]. The summation for all rivers and brid machine-based and human-interpreted method first gen- floodplains plus estimates for streams results in the region erated a mask of floodable versus nonfloodable areas. After being most flooded in May with 350,000 km2 inundated, or a segmentation algorithm automatically generated spectrally 20% of the quadrat.
MELACK ET AL.
of fluxes of carbon. Complementary material is included in the Richey et al. [this volume] and Costa et al. [this volume] chapters and the review by McClain and Naiman [2008] of Andean influei'ices on the biogeochemistry and ecology of the Amazon.
527 4' N
a
16' S
Figure 1. Floodable area (black) for Amazon basin below 500-m contour derived from JERS synthetic aperture radar mosaic. Floodable areas are not all inundated simultaneously and may include areas not floodable that do not appear on the scale of the image.
b
and stage heights in nearby rivers were used to extend the records of inundation for nearly a centmy for the Amazon main stem and for several decades for the other floodplains. Time series ofSAR data at 3- or 6-week intervals are available for several s\lbregions within the Amazon Basin and
c Water Bare or herbaceous, non-flooded Herbaceous, flooded Shrub, non-flooded
Shrub, flooded Woodland, flooded Forest, non-flooded Forest, flooded
Plate 1. (a) Wetlands mask for the central Amazon. Wetlands (white) occupy 17% of the total area. Mapping of vegetation and inundation at (b) low-water stage (September-October 1995) and (c) high-water stage (May-June 1996). Adapted from Hess et al. [2003], reprinted with permission from Elsevier.
Satellite-borne passive microwave sensors provide a record of seasonal inundation in four large floodplains in the Amazon Basin: main stem Amazon floodplain in Brazil, Llanos de Moxos in Bolivia, Bananal in Brazil, and Roraima savannas in Brazil and Guyana [Hamilton et al., 2002, 2004]. The maximum areas subject to inundation (including open watel; in lakes and rivers) during an 8-year period (l979~ 1987) were as follows (in km2): main stem Amazon, 97,400; Moxos, 92,100; Bananal, 58,600; and Roraima, 16,500. Total extent of inundation of these floodplains varied considerably seasonally and interannually, with the greatest relative variation in maximum extent of inundation in the Bananal and Roraima (Figure 2). Regressions between flooded area
Phlte 2. Flood duration map for interfluvial wetland in upper Negro Basin. For each pixel, the longest span of continuously flooded dates was used to establish the duration of inundation.
528
FLOODPLAIN ECOSYSTEM PROCESSES
MELACK ET AL.
00" 59' 60" S September 2003
02" 59' 30" S < 20
20 -56
92 -110
Amazon River JERS mask
> 110
Other
56- 92
Chi
mgm·3
Plate 3. Chlorophyll distribution in the open waters derived from
Moderate Resolution Imaging Spech'oradiometer images. From Novo et at. [2006].
can be used to generate high-resolution maps of inundation and its seasonal variations. The three examples described here provide information relevant to biogeochemical results presented in subsequent sections, A time series of L band JERS-l data obtained for the Rio Jall, a tributary of the Rio Negro, in combination with in situ measurements of river stage were used to map the spatial variation of flood duration
in the largely forested floodplain [Rosenqvist et al., 2002]. A time series of C band RADARSAT data permitted development of a similar set of maps for the interfluvial savanna wetlands of the upper Negro Basin [Belger, 2007] (Plate 2). Martinez and Le Toan [2007] used a time series of JERS-l SAR images to map temporal variations in inundation (distinguished as never, occasionally, or always flooded) and the spatial distribution of vegetation (distinguished as pastures or clear cuts, savannas or pioneer fonnations, or forests) for the Curuai floodplain near Santarem, Brazil (see section 3). Barbosa [2005] circumvented, in part, the limitations of optical remote sensing in Amazonia by showing that temporal variations in the surface area and optical features on the Amazon floodplain were reCUlTent and mostly dependent on stages in the flood pulse, making it possible to use a time series of images acquired in different years at defined river stages to represent the seasonal variations. Intensive ground data collection was conducted during each of four conditions (low, rising, high, and falling water) in order to characterize the composition and spectral behavior of the water masses in the Curuai floodplain. This information was used in a supervised classification of Landsat thematic mapper images to identify water masses according to the abundance of the chlorophyll, inorganic particles, and dissolved organic matter. Several recent applications of optical and microwave remote sensing to floodplain ecology have employed new combinations of sensors or data. Novo et al. [2006] used a time series of Moderate Resolution Imaging Spectroradiom-
and early and late successional floodplain forests. Thieme et al. [2007] extended this approach to the whole Rio Madre de Dios basin and other watersheds in southwestern AmaZOllla.
3. INUNDATION HYDROLOGY J F MA MJ J A 90
SON
D
r;-,."....-----------.- 250 Moxos
60
The water balance of a floodplain can be expressed by the equation
200
150
!1S=P+ R±L ±H± G~ E.
100
30
50
o +-..-......-,...--.-.--.--.-..;.,-....--...-......-+ 0 J A SON D 60.,-----------.,.300 J
F M A M J
:§ c
'ro
Bananal 40
200
20
100
o
E E.0::
0 J
F M A M
J A SON D
18
400 300
12
200
6
100
J F M A M J J A SON D Month
Figure 2. Mean monthly flooded area in the four regions over the period of satellite observations (solid line) and mean monthly rainfall based on long-term records (dashed line). Monthly means of flooded area are based on passive microwave data and include open water areas in rivers and permanent lakes. Vertical bars indicate the range of the monthly area estimates. From Hamiltoll et at. [2002].
Plate 4. Curuai floodplain. From Maurice Bourgoin et at. [2007], reprinted with permission from Elsevier, Landsat image
(Landsat 7 Enhanced Thematic Mapper, 28 October 2002) corresponds to an absolute water level at the CUlUai gauge station of 4.74 m. Depicted are the four gauging stations (flags, daily record), the water quality monitoring stations (stars, monthly sampling), and the suspended sediment monitoring stations (circles, sampled every 10 days).
529
eter images to produce maps of chlorophyll distribution in a floodplain reach from Parintins to Almeirim in spite of the coarse spatial resolution and nonideal bands (Plate 3). RudOTff et al. [2007] utilized hyperspectral data from the EO1 Hyperion sensor to distinguish water masses in a swath across the Amazon floodplain near Santarem. Hamilton et al. [2007] employed Landsat Enhanced Thematic Mapper data with JERS-l SAR data and C band interferometrically derived topographic data to identify geomorphic and vegetation classes in a subregion of the Rio Madre de Dios basin (Peru). By using an image segmentation procedure validated with ground surveys, they distinguished open waters in rivers and lakes, seasonally exposed river bars, palm swamps,
!1S is the change in volume of water, P is rainfall onto water surfaces, R is inflowing upland runoff, L is exchange ofwater in either direction through cOllllections to adjacent floodplains, H is exchange of water in either direction through channels to an adjacent river, G is exchange with the surrounding groundwater system, and E is evaporation; units are m 3 [Lesack and Melack, 1995; Mertes, 1997]. Each of the terms will change on multiple time scales as a function of meteorological and hydrological conditions and will vary among floodplains as a function of differences in basin morphometty, catchment area, and local geomorphology. The first detailed analysis of inundation hydrology for an Amazon floodplain lake was performed at Lake Calado located near Manacapuru on the Rio Solimoes [Lesack, 1993, 1995; Lesack and ¥elack, 1995]. River water entered the lake at the start of rising level in the Rio Solimoes, but by midrising water, lake water steadily flowed from the lake into the river, while river levels continued to rise. By the end of the water year, Lake Calado had experienced a 1O-m range in water level, and local runoff had contributed 57% of the total water input, river inflow had contributed 21 %, rainfall directly onto the lake had contributed 11 %, inflow from an adjacent lake had contributed 6% and seepage had contributed 4%. On the basis of data acquired from 1997 through 2003, Maurice Bourgoin et al. [2007] and Bonnet et al. [2008] examined the inundation hydrology of the Lake Curuai floodplain (see description in section 3) in a manner similar to Lesack and Melack [1995]. The Amazon River dominated the inputs of water to the lake year-round, accounting for about 77% of the annual total, on average. Rainfall and runoff accounted for about 9% and 10%, respectively, while seepage from groundwater accounted for 4%. The differences between Curuai and Calado in the importance of riverine versus upland inputs reflect, in part, differences in lake basin morphologies and in upland catchment area (CA) relative to the floodplain area (FA): Calado CAlFA = 7; Curuai CAlFA = 2. To extend analyses of floodplain inundation to regional scales requires a combination of modeling and remote sensing. Basin-scale models that include floodplain inundation
530
MELACK ET AL.
FLOODPLAIN ECOSYSTEM PROCESSES
typically assume the water surface is horizontal and equal to the levels in the main river channels [e.g., Richey et al., 1989] and operate at moderate resolution (e.g., ~9 km [Coe et al., 2002, 2007]). However, recent results indicate considerable spatial and temporal changes in elevations of water surfaces across Amazon floodplains. AlsdOif et al. [2000, 2001] performed interferometric processing of synthetic aperture radar data collected during the space shuttle imaging radar mission to demonstrate the possibility of detecting centimeter-scale stage changes across the floodplain. These observations indicated that drops in water level diminished with increasing distance from the main stem river during recessional flow. AlsdOif [2003] suggested that errors in floodplain storages of 30% can result from assuming horizontal water surfaces. Further, AlsdOif et al. [2007] used interferometric SAR measurements from the JERS-l satellite to document the passage of a flood through a large, topographically complex floodplain at the confluence of the rios Pums and Solimoes. They noted abrupt differences in stage changes coincident with floodplain channels and other localized variations in water level in a scene from mid-rising water but less heterogeneity in water levels in a scene at high water. An important biogeochemical implication of these findings is that flow paths and residence times on floodplains are dynamic in space and time. By combining interferometric SAR measurements of stage changes and a continuity equation, AlsdOifet al. [2005] developed a linear diffusion model of floodplain drainage that captures the composite behavior of flow through channels, lakes, and aquatic vegetation. This fairly simple method of detennining storage changes requires floodplain connectivity to be parameterized by a description of floodplain topography and temporal changes in water levels on the floodplain. In the first application and validation of a two-dimensional hydrodynamic model to a large reach of Amazon floodplain, Wilson et al. [2007] found that more than 40% of the total river flow was routed through the floodplain near the confluence of the rios Purus and Solimoes. They applied the LISFLOOD-FP model [Bates and DeRoo, 2000] at a 270-m resolution over a ~260 Ian reach of Rio Solimoes with its ~40-km-wide floodplain for a 22-month period. By a combination of ground surveys of tree heights, remote sensing of vegetation types, and spatial aggregation, they reduced the noise in the Shuttle Radar Topographic Mission (SRTM) digital elevation data to <1.8 m. Channel topography was obtained by sonar surveys from boats. The model matched well with inundation extent estimated from JERS-l data [Hess et al., 2003] at high-water levels but overpredicted inundation at low water because the small channels were below the spatial resolution of the aggregated topographic data so the floodplain drained insufficiently.
Seasonal variations in the water stored on the floodplains of the Amazon are sufficiently large to cause anomalies in the Earth's gravity field detectable from space with Gravity Recovery and Climate Experiment (GRACE) [Tapley et aI., 2004]. Han et al. [2005] applied an improved algorithm to GRACE data for the Amazon and presented spatial patterns in water storage in terms of equivalent water heights at monthly intervals during 2002 and 2003 and demonstrated the possibility of monitoring water storage variations at submonthly intervals. By combining data from spaceborne SAR and altimetry sensors with hydrographic observations, Frappart et al. [2005] estimated the volume of water stored in the floodplains of the Rio Negro during 1995~1996. These two approaches are independent and, if applied to the same period, can be compared as a step toward their validation. While these recent remote sensing and modeling results are valuable and promising, limitations ofthe available measurements continue to constrain models of inundation dynamics. Gauges ofriver stage are widely spaced, Amazon floodplains remain ungauged except during field studies at a few locations, and the gauges are not accurately leveled relative to the geoid. Satellite-borne altimeters have wide spacing between tracks [Birkett et al., 2002], passive microwave sensors have coarse spatial resolution [Hamilton et al., 2002], and SAR systems have limited temporal [Hess et al., 2003] or spatial coverage [Rosenqvist et al., 2002]. Topographic data from the SRTM require considerable work to remove vegetation. Bathymetric data are available for few floodplains (e.g., Calado [Lesack and Melack, 1995] and Curuai [Bourgoin et al., 2007; Barbosa et al., 2006]). Ongoing work is begillliing to alleviate several of these issues, e.g., use of highly accurate and precise global positioning systems to level gauges, sonar surveys with recording systems combined with analyses of time series of SAR imagery to generate regional bathymetry, and new SAR sensors such as the ALOS PALSAR that offer repeated regional coverage at high resolution. These results, in tum, can be compared to GRACE-derived estimates of variations in water storage at regional scales. Costa et al. [this volume] summarized modeled and remotely sensed seasonal and interannual variations in inundation extent as a function of climatic conditions. They conclude that, while ENSO strongly influences variability in discharge, a 28-year mode in precipitation variability explains most of the interarlliual differences in extent of inundation. Rainfall anomalies are known to be associated with El Nino conditions and sea surface temperatures in the equatorial Pacific and tropical Atlantic oceans [Ronchail et al., 2002, and references therein]. One consequence of the low rainfall and reduced period of inundation during El Nino conditions is wider tree rings in floodplain trees [Schongart et al., 2005]. Hence, Schongart et al. [2004] were able to
develop a dendroclimatic reconstruction for the western Amazon spanning tl:\f1ast two centuries that indicates an increasing severity ~tEI Nino conditions. Schongart and Junk [2007] utilized tl)e relations among flooding, the Southem Oscillation Ind@~, and sea surface temperatures to demonstrate the possibility of forecasting maximum water levels 4 months in advanGe. 4. REPRESENTATIVE FLOODPLAIN ECOSYSTEMS 4.1. Central Amazon Floodplain
Much of the research on Amazonian floodplain ecosystem processes has occurred along the Solimoes/Amazon River and lower reaches of its major tributaries in Brazil. The fringing floodplain along the 2600-lGn reach ofthe Solimoes/ Amazon River from 52.5°W to 70.5°W contains about 6500 lakes and associated floodplains, and the lower 400 km of four major tributaries (Japuni, Punts, Negro, and Madeira) contain an additional 2320 lakes with associated floodplains [Sippel et al., 1992]. Seasonal changes in stage result in variations in inundated area (excluding river channels) from 19,000 lGn2 to 81,000 Ian 2 along the 2600-km main stem reach based on analyses of passive microwave data obtained from 1979 to 1987 [Sippel et al., 1998]. Several intensive investigations during the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) focused on floodplains in the lower reach near Santarem and complemented earlier work concentrated in the middle reach near Manaus and recent studies in the upper reach near Tefe. 4.2. Lower Reach
Lago Grande de Curuai, a complex system of about 30 shallow, intercoill1ected lakes linked to the Amazon River by several channels, is representative of the floodplains in the lower reaches and has been the subject of intensive study by several groups (Plate 4). The open water of the complex ranges from 760 1Gn2 to 1377 lon2 , flooded meadows can cover 570 1Gn2 , flooded savanna can cover 450 1Gn2 , and, during high water, flooded forests extend over 560 km 2 [Martinez and Le Toan, 2007]. Barbosa [2005] illustrated the large temporal and spatial variations in suspended chlorophy:Il and inorganic particles throughout the complex. The fluvial transport and storage of sediments within channel-floodplain systems is an important process in these systems, and Maurice Bourgoin et al. [2007] examined this aspect in the Cumai floodplain. Their analyses were derived from hydrological data, a monitoring network operated between 1999 and 2003; total suspended solids concentrations acquired near Obidos [Guyot et al., 2005]; multitemporal
531
remote sensing images and radar altimetry data; and the hydrological analyses of Bonnet et al. [2008]. Although the suspended sediment inputs associated with the flux of Amazon waters into the floodplain move through the different channels located around the system, the suspended sediment outputs from the floodplain into the mainstream river occurred via the two widest and deepest channels located on the southeastern side of the floodplain. Because of the amount and the timing of the local mnoff, the output of water from the floodplain was greater than the input of Amazon water to the system. The maximum sediment fluxes outflowing from the Curuai floodplain into the mainstream occurred from July to October. For the three simulated years, the annual sediment balance (input minus output) was positive, confirming that this Amazonian floodplain acted as a sediment trap, as observed in one connecting channel by Moreira-Turcq et al. [2004]. In 2000~2001 and 2002-2003, the annual volume of sediment trapped in the floodplain was of the same order of magnitude as the mean annual sediment fluxes outflowing from the floodplain into the Amazon River. Within the floodplain lakes, sedimentation was dismpted by the resuspension induced by wind waves favored by a large fetch especially during low water. Suspended sediment concentrations in three lakes ranged from 4 mg L-I during the flood peak to high values during the dly season (1600 mg L-I), more than 6 times the maximum concentrationmea~ured in the Amazon River during the same period. ' For the period 2000-2003, the mean specific sedimentation rate was 517 (±23%) t lGn-2 a-1. Expressed in linear kilometers of the Amazon River, this rate is far less than rates of 0.67-0.89 Mt lGn- 1 a-I reported for a reach of the Rio Solimoes by Mertes [1994], for the whole Amazon River by Dunne et al. [1998], and for the reach between Manacapuru and Obidos (0.25 Mt lan- I a-I [Laraque et al., 2005]). However, average sediment deposition rates published for long reaches cannot be readily applied to specific floodplain systems because tectonic activity affects the form and the behavior of the channel and floodplains, and geomorphology and local hydrology control the sediment exchanges between the mainstream and floodplain. 4.3. Middle Reach
The lakes and floodplains within about 100 km of Manaus along the rios Solimoes, Amazon, and Negro have received considerable attention largely because of their proximity to Brazilian National Institute of Amazonian Research (Institufo Nacional de Pesquisas da Amazonia). Much of this research is summarized by Sioh [1984], Junk [1997], Melack [1996], and Melad: and Forsberg [2001]. Two lakes have
532
MELACK ET AL.
FLOODPLAIN ECOSYSTEM PROCESSES
been the focus of many limnological, hydrological, and ecological studies: Lake Calado, a dendritic lake situated on the north bank of the Rio Solimoes, about 80 km west of the confluence of the rios Solimoes and Negro; and Lake Camaleao, a narrow lake located on Marchantaria Island at the confluence of the rios Solimoes and Negro.
4.4. Upper 'Reach The lakes and floodplains near 'the confluence of the rios Japura and Solimoes represent a wide range of morphologies [Mertes et al., 1995; Sippel et al., 1992], and the region has received increasing scientific attention because ofthe Mamirawi project based in Tefe. Flooded forests are the dominant habitat on floodplains in the upper reaches [Hess et al., 2003], and their distribution in relation to environmental conditions is becoming increasingly well understood [Haugaasen and Peres, 2005; Junk and Piedade, 2009; Kalliola et al., 1991; Kvist and Nebel, 2001; Wittmann et a!., 2006]. Wittmann et al. [2002] examined the spatial distribution of different types of floodplain forests within the Mamirawi Sustainable Development Reserve (MSDR) through a combination of inventories, aerial photography, and Landsat thematic mapper imagery. The focal area of the MSDR, approximately 70 lU11 northwest of Tefe, is mostly closedcanopy forest inundated annually with water from the rios Japura and Solimoes. At an average flood level of3 m, Wittmann et al. [2002] detected changes in species diversity and architecture, and they used these differences to categorize the floodplain forests as low or high varzea forest and were able to distinguish early and late successional stage low varzea forest. Wittmann and Junk [2003] determined the floristic composition and stmcture of saplings in three successional stages in the same area and compared the saplings to mature communities. They found that sapling distribution and species richness were associated with gradients in flooding and irradiance with highest richness in the high varzea forest. The zonation of mature communities depended on inundation and gradients in sediment deposition and soil texture [Wittmann et a!., 2004]. One system in the vicinity of Tefe is Lake Amana, located in the 23,500 km2 Reserva de Desenvolvimento Sustentavel Amana. The large lake receives sediment-laden waters via Parana Amana during high water and black water from small upland streams throughout the year. Silva [2005] investigated these dynamics in relation to the aquatic macrophyte composition and distribution. Measurements of limnological conditions and the occurrence of aquatic macrophytes, with a focus on Echinochloa polystachya, were made along longitudinal transects. Echinochloa polystachya proved to be a good indicator of the nutrient supply, as it formed pro-
lific stands within 8 km of the inputs of white waters. Other species, such as Pistia stratiotes, Eichhol'llia crassipes, Ludwigia densiflora, Salvinia auriculata, and Paspalum repens, grew well in the same areas. Aquatic macrophytes are important to Amazon manatees (Trichechus inunguis) that inhabit this region, and Arraut et al. [2007] have applied ground surveys with remote sensing to relate the seasonality of macrophyte growth with the movements of the manatee. Rodrigues [2007] inventoried the trees in 17 sites on a gradient through Lake Amana and found a relation between the decline in phosphorus toward the upper part of the lake and some tree species associated positively or negatively with this element. In addition, he found that pioneer species (e.g., Salix Martiana and Alchol'llea castaneifolia) associated with white-water influence occurred within the first 10 km, while species (e.g., Licania apetala, Licania micrantha, and Swartzia polyphylla) associated with black water occurred at distances greater than 13 km from parana influence. 5. REGIONALIZATION OF CARBON BIOGEOCHEMISTRY Floodplains play an important role in the organic carbon balance of the Amazon Basin and are sites of high rates of aquatic plant production and a major source of methane to the troposphere. Melack and Forsberg [2001] summarized data for the central Amazon floodplain on primary production, allochthonous inputs, heterotrophic activity, evasion of carbon dioxide and methane, sedimentation, and the overall carbon balance. Melack and Engle [2009] summarized the organic carbon budget for Lake Calado, the most complete budget available for an Amazon floodplain lake. Additional measurements and analyses of several aspects of the carbon biogeochemistry of Amazon floodplains occurred during LBA and permit more extensive regionalization than possible by Melack and Forsberg [2001]. 5.1. Organic Carbon Inputs to the Floodplain
Phytoplankton, herbaceous macrophytes, flooded forests, and periphytic algae contribute to autochthonous primaly production in floodplains (Table 1). Phytoplankton is limited generally to the open waters of lakes and some rivers, such as the Tapaj6s, where underwater light is available. Net primary production by phytoplankton averages about 200 Mg C km- 2 a-I [Melack and Forsberg, 2001, and references therein]. Herbaceous macrophytes are especially abundant in floodplains associated with rivers laden with nutrient-rich sedi-
Table 1. Net Primary Productivity for Major Types of Vegetation in a 1.77 million km 2 R~gion in the Central Amazon Basin Characterized by Hess et at/12003]"
533
bined above- and below-water biomass ranged from 31 % to 75% and averaged 49%. If this loss rate is generally applicable, it would result in increased estimates ofNPP for herIi Areal NPP l High-Water Total NPP baceous macrophytes, since Junk and Piedade [1993] and (Mg C km- 2 a-I) /Aquatic Vegetation Area (ktn 2) (Tg C a-I) Costa [2005], used lower rates of loss. NPP for herbaceous macrophytes in Lake Calado for the 7-month period from 1150 flooded forest b 184 160,000 FebmalY to August (assuming 50% carbon content) was cal2500 macrophytes 100 40,000 200 phytoplankton d 20,000 4 culated to be 6720 Mg C km- 2 . 100 periphyton (forest)e 8 160,000 Floodplain forests can occupy significant areas of Amazon 110 periphyton 2 40,000 floodplains especially west and north of Manaus, and their (macrophytes)C distributions are generally restricted to the shallower porTotal 298 tions [Junk and Piedade, 1997]. Worbes [1997] estimated "NPP is net primary productivity. Refer to text or references cited the combined production offine litter and large woody detrifor sources of values. tus to vary from 800 to 1250 Mg C la11- 2 a-I (assuming a dty bFlooded forest includes 16,000 km 2 of woodland and shrub cat- weight carbon content of 50%) and estimated root producegory. tion to be 30% ofthe live wood increment in well-developed cMacrophytes include 14,000 ktl12 of woodland and slmlb categovarzea forests. Average net production rate can be estimated ry because these habitats usually support floating macrophytes. The as 1I50 Mg C 1u11-2 a-I (excluding wood growth and aerial NPP value used is lower than the upper values reported by Costa grazing losses). Nebel et al. [2001] measured wood biomass [2005] and the 7-month value calculated by Engte et at. [2008]. dphytoplankton is assigned 20,000 km2 of open water as lake production and fine litter fall in three floodplain forests along the lower Rio Ucayali in Pem. In all three forests, fine litter habitat. 2 cperiphyton are assumed to grow on forest and macrophyte sub- fall was around 700 Mg C 1011- a-I and peaked in associastrata only about half the year when these habitats are inundated. tion with flooding. The average period of inundation estimated by Hamilton et at. Periphytic algae require solid substrata and are generally [2002] would reduce periphyton productivity associated with forfound near the water surface attached to the submerged porests by about half. tions of macrophytes and flooded trees [Engle and Melack:, 1993]. Production yalues given by Doyle [1991] for macrophyte habitat and Putz [1997] for varzea forest can be comments [Junk and Piedade, 1997]. The complex spatial and bined with reasonable estimates of inundation periods for temporal variations of herbaceous macrophytes make it dif- forests and macrophytes to calculate all11ua1 values. Net proficult to estimate their contribution to floodplain production duction ofperiphyton associated with floodplain forests was that must incorporate the cumulative, sequential production estimated as 100 Mg C km-2 a-I, and production of periphyof tenestrial, semiaquatic, and aquatic plant assemblages. ton associated with herbaceous macrophytes was estimated Junk and Piedade [1993] estimated the cumulative biomass as 110 Mg C km-2 a-I [Melack and Forsberg, 2001]. increase of three successive macrophyte assemblages growTo compute regional, net production rates requires incorpoing under favorable conditions on the central Amazon flood- rating estimates of the area occupied by open water, flooded plain to be 1500 Mg C km -2 a-I, and, assuming a dty weight forest, and macrophytes. The recent results derived from recarbon content of 50% and a monthly biomass loss of 10- mote sensing (summarized above) permit improvement of 25% during the growing season, they estimated net annual Melack and Forsberg's [2001] estimates and enlargement primaly production to be 2500 Mg C la11-2 a-I. Costa [2005] of the area to the 1.77 million km 2 region characterized by combined SAR images with field measurements to estimate Hess et al. [2003]. While providing the best documented and net primary productivity of aquatic macrophytes throughout up-to-date values (Table 1), it is important to aclO1owledge a large, shallow lake near Santarem, Brazil, and reported that calculations are based on a small to moderate number of values in different parts of the lake ranging from <900 Mg C measurements in only a subset of the geographical area and km- 2 a-I to >5000 Mg C lan- 2 a-I. Ground-based measure- habitats and probably do not represent basin-wide averages. ments of species composition, plant growth rates, plant den- . In Amazon floodplains and savanna wetlands, daily water sity, and areal biomass were combined with low-elevation column respiration usually exceeds planktonic photosynvideography to estimate community net primary productiv- thesis; hence dissolved oxygen is usually undersaturated ity (NPP) and to compare expected versus observed biomass [M'elack and Fisher, 1983], and dissolved carbon dioxide is at monthly time steps during the aquatic growth phase in supersaturated [Hamilton et al., 1995; Richey et al., 1988, Lake Calado [Engle et a!., 2008]. MOilthly loss rates of com- 2002]. The predominance of respiration in these systems is C
534
FLOODPLAIN ECOSYSTEM PROCESSES
due to the input of significant quantities of allochthonous carbon fi'om riverine and upland sources and to aerial photosynthesis and aquatic respiration by macrophytes and flooded fore~ts as well as aquatic decomposition of macrophytes and flooded forest litter. Melack and Forsbelg [2001] approximated riverine inflows and upland runoffto a reach of central Amazon floodplain in combination with measurements of organic carbon fractions to roughly estimate allochthonous inputs to be less than 5% of their estimates of primary production on the floodplain. Melack and Engle [2009] estimated that 9% of the organic carbon inputs to Lake Calado came from rain, streams, groundwater, and riverine inflows. Attempting to quantify all these sources of organic carbon and carbon dioxide in the Amazon floodplains, basin-wide, is outside the scope of our synthesis and awaits information not currently available on hydrological and biogeochemical fluxes. Organic carbon that is produced in or transported into floodplains is potentially available to heterotrophic organisms living there. A series of studies, based largely on isotopes of carbon, indicate general patterns of carbon flow in Amazon floodplains (summarized by Melack and Forsberg [2001 J). Nutritious organic matter, derived predominantly from algae, tree fruits, and seeds, provides most of the organic carbon to aquatic herbivores and detritivores. Nearly all vascular plant biomass from C-3 and CA plants decomposes and releases organic carbon, predominantly in the dissolved fonn, to the water column. The labile component of this DOC, dominated by CA plant carbon, is consumed by heterotrophic bacteria, and much is released as CO 2 or CH4 to the atmosphere. On the basis of 13C evidence, Quay et al. [1992] estimated that 40% of the organic matter being respired in the Amazon and its major tributaries is CA plant material. Recently, MayOlga et al. [2005a] reported that an organic carbon fraction less than 5 years old and disproportionately of CA plant origin is fueling respiration in Amazonian rivers. 5.2. Carbon Dioxide and Methane Evasion
Carbon dioxide emissions from floodplain habitats include respiratOly losses fi'om living plants and animals but are dominated by the metabolic losses from bacterial communities consuming dead organic matter. On the basis of a regional survey along the central Amazon floodplain during high water, Devol et al. [1988] reported CO 2 emission rates measured in open water within aquatic macrophyte beds, flooded forests, and lakes. On the basis of samples fi'om 13 expeditions over a 2000-km reach and a 10-year record at one site, Richey et al. [2002] repOlied the CO 2 evasion from the Amazon and its major tributaries and floodplains in the
1.77 million km 2 region characterized by Hess et aI. [2003] as 830 ± 240 Mg C km-2 a-I, based on an annual mean flooded area of250,000 km2. Methane is produced predominantly in anoxic environments associated with flooded habitats. Melack et al. [2004] combined the temporally varying extent of inundation and vegetation, derived from passive and active microwave data (see section 2), with habitat-specific field measurements [Devol et aI., 1990; Engle and Melack, 2000] to calculate regional rates of methane emission. Uncertainties in the regional emission rates were determined by Monte Carlo error analyses that combined error estimates for the measurements of emission and calculations of inundation and habitat areas. Methane emission, calculated using the total annual emission rate and mean annual flooded area of the Solimoes/Amazon main stem floodplain, was reported as 30.4 ± 7 Mg C km-2 a-I. On the basis of the work of Rosenqvist et al. [2002], a similar calculation for the Rio Jall basin yields 23 Mg C km-2 a-I, probably reflecting nutrient-poor conditions in this black water floodplain. Although seasonally flooded savannas cover extensive areas of the Amazon Basin, no measurements of methane emission from savannas were available to Melack et al. [2004], hence they used a mean value for aquatic macrophytes in central Amazonia (71 Mg C km-2 a-I) to estimate emissions from the Moxos, Roraima, and Bananal savannas. Extensive areas of interfluvial and riverine smral1ll1a lands and palm swamps occur in the upper Negro determine the atmosphere-water exchanges otl~arbol:I d,loXlde and methane in these environments, diffusive emissions of CO 2 and CH4 were measured mCI11tl1Iy 2005 [Belger, 2007]. CO 2 emission was C m-2 d- I. CH4 was consumed in dry envirol1lments of 3.7 ± 4.9 mg C m-2 d- I and emitted ments at a rate of 48 ± 109 mg C m-2 d- I , a used by Melack et al. [2004] for sav·am1as. Hydroelectric reservoirs release sigmtlc31ilt C02 and CH4, and while most studies emissions from reservoir surfaces, emISSllO of dams are likely to be important. tel' required to sustain methane and from reservoirs often include Imlcr
MELACK ET AL.
an~ eb.ullitive emissi09s were estimated, as well as methane OXl?atIOn downstre~J;l1 from the dam. An inundation model
denve~ fr?m a ba11~metric map was used for the spatial and
ten~p~Ial ll1terpo~~tIOn of reservoir emissions. Downstream elll1SSIOn accomfted for 55% of the total methane emission but. o~ly 4% of total CO 2 emissions. While total reservoir e~ll1s.sIOns are a small fraction of basin-wide fluxes, the consld~Iable gree~house gas emissions associated with Amazoman' reserVOirs (Balbina' Samuel ' as we II . , c uand nT lu l, as Petit Saut 1~1 French Guyana) indicate that hydroelectric power generatl?n can have climate w31ming effects similar to thermoelectnc power plants [Kemenes et aI., 2008]. 5.3. Sedimentation
Particula~e. organic carbon incorporated into sediments ~nd not OXIdIzed by detritivore and bacterial communities IS sequestere.d through burial. Sedimentation rates in Amazon floodpl~111s va.lY considerably in time and space [Aalto et al., 2003, Morelra-Turcq et al., 2004] ,and fiew measure' ments of carbon . . accumulation based on 210Pb c11ronoI ogIeS o~ recent. c?ndltlons are available. In an analysis of carbon dl~gen~SlS ll1 the pelagic sediments of lakes Jacaretinga and Cl~s.taI1110, Dev0lo et al. [1984] estimated the burial rate of particulate orga11lc carbon to be 44 g C m-2 -I d 28 C -2 - I . a an g m a ,.respectIvely. Smith et al. [2003J determined burial of orga11l? ca~'bon based on two cores collected fi'om each 0!1three SItes m Lake Calado to vary fi'om 18 to 62 g C m-2 a . In the large, sha~low Cmuai floodplain, Moreira- Turcq et al. [2004J determmed organic carbon accumulation was t generally about 100 g C m-2 a-I but COllld be h' h d I' " Ig er a some ept 1S m th~lr smgle core. It is difficult to extrapolate these few and vanable values to the whole basin. 5.4. Comparison ofNet PrimCllY Production With Evasion ofCarbon Dioxide and Methane
The ~otal net ~roduction attributed to flooded forests (excludmg wood l~lcrements), aquatic macrophytes, phyto~lankton, and penphyton for the 1.77 million km2 charactel~zed by Hess et al. [2003J is about 300 Tg C a-I. Flooded fOlests account for 62% of the total, aquatic macrophytes acc?unt for 34%, and the remaining 4% is associated with penphyton and phytoplankton. Approximately 10% of the total v.alue equals the export of organic carbon by the Ama~05no/,Rlver [Richey et al., 1990], methane emission is about b' b 0 .~Me:ack e~ al., 2004J, and a similar percent is likely to . e uued.m sedIments. The remaining portion is close to bel1:g suffiCient to fuel the respiration that results in the degassmgof21O±60T C -I b " . g a as car on dIOXIde from the rivers and floodplams [RicheJlJ et al., 2002J • Howe ver, conSl'der-
535
able terrestrial sources of DOC and POC I . '" . " a so support the aquatic hetelOtlOpluc actIVIty [Richey et al thi' I J o b b ., s vo ume ~e pro a. Ie expla~ation for the apparent excess of organi~ carbon beu:g supplIed to the rivers and floodplains is that the outgassll1g rates for carbon dioxide are underestimates as sugges~ed by R~chey e~ al. [2002J and as indicated by recent m~asurements 111 a var.lety of aquatic habitats [Richey et al., ~hlS volume]. AlternatIvely, the estimates of net productivIty could be too high, on average for the regiOl} H . . . 1 ' . oweveI, a.legIOna carbon balance for central Amazonia b d t' ase on airb orne es Imates. of CO2 exchanges does not require large CO 2 fluxes from nvers to balance the budget [L1olJd t I 2007]. J ea., 5.5. Next Steps
To advance predictive capability and understanding of how th.e carbon balance of Amazon floodplains will respond t? .e~vlronm~ntal changes requires several coordinated actiVIties. An Improved version of the LISFLOOD d I com~lemented by high-quality topography and bathy:~tI~' preCIsely level~~ ga~ges in floodplains, and measurement~ of water velOCIties 111 floodplain channels should b . t . t d . h b' , e m egia e WIt IOgeochemical models of processes. An upland runoff model that includes inputs of DOC and POC . t' I Add' . IS es~en Ia . Itlonal m. easurements of net primaly prod t' b d' . ; UC IVty ~ ,car on IOxlde and methane evasion, and carbon burial Iate~ are needed from undersampled or unsampled habitats. A~ Important further challenge is incorporating impacts of c~l1nate cha~ge on the hydrology, sediment dynamics, and bIOgeochemlstly of Amazon rivers and floodplains. 6. RELATIONS BETWEEN FLOODPLAIN HABITATS INUNDATION, AND FISHERIES '
':ariati?ns in the distribution and inundation of floodplam habItats playa key role in the ecology and production of many comme~'cially important fish in Amazonia. Most ~mazon fish denve their energy from food chains beginmng from algae or from tree seeds and fi1.lits [Forsbelg et al., 1993; Melack and FO/'Sberg, 200lJ . Th e d eve Iopment of these l:es~urces ~nd the growth patterns of fish are generally synchlOmzed w~th. seas~nal variations in flooding. Seas?nal vanatIOns ll1 the distribution and inundation of habItats used by fish were investigated on Murutu Islan~, : small fluvial island in the main channel of the Rio SolImoes .near 2004J . The na tura I I1a b'1. Manaus [Corredor, tats on, thIS Island included open water, grasslands, and seasonally flooded forests. A significant part of the island defOl·.ested .and used for agricultural crops. At peak h7;~ wateI, the Island was completely inundated, while at low
536
MELACK ET AL.
FLOODPLAIN ECOSYSTEM PROCESSES
water only a few open water environments remained flooded. This spatial and temporal variation in inundation controlled the production dynamics of planktonic and periphytic algae as well as th~ phenological development of tree fruits and seeds. The distr'ibutions and feeding habitats of herbivorous and onmivorous fish were closely linked to these patterns. To characterize this complex pattern of habitat dynamics, a temporal series of high-resolution JERS-l L band radar images was analyzed, and the seasonal changes in habitat cover were associated with fish' community structure using principal components analysis. The cardinal tetra (Paracheirodon axelrodi) is the most important aquarium fish exploited in Brazilian Amazonia, accounting for more than 80% of regional exports. A small species endemic to the upper Negro Basin, the cardinal migrates between two types of wetland and uses two distinct food webs during its short life cycle. The cardinal spends the initial phase of its life cycle in shallow savanna-like interfluvial swamps that occur at the headwaters of many tributaries during the rainy season. During the drier periods, these swamps are reduced in size, and the cardinal migrates down the tr'ibutaries to the alluvial forests and stream channels associated with the Rio Negro. Temporal series of C band radar images were used to characterize the flooding patterns in the interfluvial swamps and to link these to the seasonal migration patterns of the cardinal. Stable isotope analyses were used to investigate the associated changes in cardinal's autotrophic carbon source [Marshall et al., 2008]. The lowest 0 13 C values for the cardinal were encountered during the falling-water period, after the fish had spent several months in the interfluvial swamps feeding predominantly on invertebrates in a periphyton-based food chain. The 013 C values were significantly higher during the rising-water period indicating a switch to a tree leafbased food chain. The conservation of both of these wetland habitats is clearly needed for the effective management of this regionally important commercial fishery. The annual fisheries yield in several important Afi'ican river systems has been showed to depend on the maximum inundated area with a time lag [Welcomme, 1979]. These relationships indicate the role of floodplains in sustaining riverine fisheries and the importance of the maximum flooded area in determining the level of aquatic production and cohOlt size in any specific year. They are useful for predicting the expected variation in fisheries yield due to natural factors and for evaluating the effects of different management practices. The maximum annual flooded area of Brazilian whitewater floodplains was compared to the total annual fish yield aggregated for all states in Brazilian Amazonia during the period of 1980-1998 (B. Forsberg, unpublished data,
2002). Fisheries data were obtained from annual repOlts published by Instituto Brasileiro de Geograjia e Estatistica [1980-1990] and Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renovaveis [1993-1998]. Flooded areas were estimated from monthly series of passive microwave images acquired between 1978 and 1987 following the methodology described by Sippel et al. [1998]. The total maximum flooded area for each year was obtained by aggregating the maximum annual values determined for the central Amazon floodplain between 70 and 52.5°W latitude and the floodplains of the rios Jurua, Purus, and Madeira. A linear regression between maximum annual flooded area and the maximum river stage height measured in Manaus was used to determine the flooded area in years not covered by the satellite time series. No significant relationship was found between maximum total flooded area and total annual fish yield, independent of the lag time used between variables. The lack of a clear relationship in this case was attributed to the commercial fisheries in Amazonia being selective and tending to catch fish of varying sizes, ages, and trophic levels. The lag between floodplain production (flooding) and fisheries yield is thus expected to be variable among species. Hence, the yield results for adult fish of different sizes and trophic levels were analyzed separately, using longer lag times for larger species and species from higher trophic levels. Some fish groups (e.g., herbivores) had gradual changes in yield over time, presumably reflecting variable fishing pressure. Nonlinear regression was used in these cases to describe the anthropogenic trend, and residual variation was assumed to represent natural interannual variation in fish yields. When this was done, several significant relationships were found between fish yield and flooding (Figure 3). The relationships encountered for small species at lower trophic levels generally had short lag times (0-1 years), while those for large species at higher trophic levels had considerably longer lag times (3-5 years). While much of the fisheries data used in this analysis were based on crude sampling methods, the good relationships obtained suggest that the general approach is correct. The Provarzea Project recently reorganized and standardized field measurements offisheries yield in Brazilian Amazonia, and these data should provide an excellent basis for evaluating the influence of flooding on fish yield. 0
7. FUTURE DIRECTIONS AND HUMAN DIMENSIONS OF FLOODPLAIN USE AND MANAGEMENT In light of human- and climate-induced changes in the Amazon, the spatial and temporal dynamics detectable by remote sensing will play an increasingly important role in the understanding and management ofthe system. On a basin
a
ij
4000
'>, ~
o
2000
>
:a Q> .c
o
b
en -2000 "iii ::>
"0
'iii -4000 (I) ~
-6000
I.-..~--~---~----------'
70000
80000
90000
100000
110000
Maximum flooded area same year, km 2 45000
..--~-----~----~----,
111
en c o :;;;. 40000 W
':;;"
.c en I;::
~ .~c
35000
E
o
30000 l . " " - _.......... 22000
~
24000
......._
___I
26000
Maximum flooded area 5 years before, km 2
Figure 3. (a) Relationship between the residuals of herbivorous fish yield and maximum inundated area of the Brazilian Amazon, measured in the same year. (b) Relationship between omnivorous fish yield and maximum flooded area in the Brazilian Amazon, measured 5 years earlier.
scale, models ofthe hydrological and ecological systems will benefit from inclusion of the extent and variations of inundated surfaces and aquatic vegetation. Models of production and evasion of carbon dioxide and methane must incorporate wetlands [Cole et al., 2007]. The spatial and temporal variety of floodplains is linked to the high species diversity and biological adaptations associated with these habitats [Hamilton et al., 2007; Junk, 1997] and their importance to human populations [Junk et al., 2000]. On a regional scale, significant deforestation can be detected in floodplains, which, when associated with cattle grazing, could lead to the eutr'ophication of floodplain lakes [Ajfonso et al., 2007; Novo et al., 2007]. On a local scale, improved information on floodplain conditions will contribute to efforts for community-based ecosystem management [McGrath et al., 2005] and mitigation of pollution [Melack, 2005].
537
Brazil and other South American countries committed to developing Amazonia continue building roads, navigation channels, and ports to facilitate the flow of people and products and hydroelectric reservoirs to provide energy to the growing regional economy. The "Plan for Accelerated Growth," presented by the Brazilian federal government in 2007, outlined plans for investing in logistic and energy infrastructure in Amazonia. This ambitious proposal called for the construction of three large hydroelectr'ic reservoirs, six small reservoirs, 480 km of new highway, four major power transmission lines, and one major gas pipeline, the expansion of 19 fluvial ports, and paving of four additional major highways by 2010. All of these public works could potentially affect floodplains areas, and their individual and collective impacts should be assessed. The impacts of proposed and existing hydroelectric dams and reservoirs are being evaluated in both socioeconomic and environmental terms. River impoundment has created extensive wetland areas. Recent results indicating large releases of greenhouse gases (carbon dioxide and methane) from these areas are increasingly relevant as the impacts of climate warming are becoming evident. When degassing as the water exits the turbines is combined with downstream and within-reservoir evasion, greenhouse gas emissions per megawatt-hour of power generation fi'om existing Amazon reservoirs can equal or exceed emissions from fossil fuelbased thermoelectric plants [Kemenes et al., 2008]. While the rivers and ripalJian areas released carbon and methane prior to the construction of the reservoirs and further study is needed to evaluate this precondition, our current understanding of the carbon biogeochemistry of Amazon floodplains and lakes suggests that considerably more evasion of carbon and methane will occur after construction of a hydroelectric reservoir that inundates tropical rain forest. As petroleum exploration and transpOlt in the Amazon region continues to accelerate, the risk of impacts to floodplains is likely to increase. The high productivity and hydrological complexity of these systems makes them especially vulnerable to these impacts. Accurate models of floodplain hydrology and habitat dynamics are needed to predict oil spill flow paths and minimize the impacts of pipeline construction. The expansion of POlt facilities and dredging of river channels to facilitate fluvial transpOlt and commerce could increase the risk of urban pollution and alter the exchange of water, nutrients and sediments between rivers and , floodplains [Hamilton, 1999]. The expansion of cattle and soybean production and construction of roads will lead to increased deforestation that could have significant effects on the regional hydrological cycle [Malhi et al., 2008]. The dryer climates possible with increasing deforestation will be exacerbated by global
538
FLOODPLAIN ECOSYSTEM PROCESSES
warming leading to Amazonian drying during the 21 st centmy. The result of these impacts could well be a significant reduction in wetland area and major changes in biogeochemical processls linked to inundation. Changes in the characteristics of the flood pulse could have significant effects on both the biogeochemistty and productivity of floodplain environments.
Ac1olOwledgmellts. We thank eSpecially Claudio Barbosa, Marie-Paule Bonnet, Jean-Michel Martinez, Pascal Kosuth, Gerard Cochonneau, Mary Gastil-Buhl, and Laura Hess for contributions to the results repOlied, NASDA/JAXA for imagelY, and Steve Hamilton and an anonymous reviewer for useful comments on an earlier draft. The research reported has been suppOlied by several NASA, CNPq, FAPESP, and IRD grants.
REFERENCES Aalto, R., L. Maurice-Bourgoin, T. Dunne, D. R. Montgomery, C. A. Nittrouer, and J.-L. Guyot (2003), Episodic sediment accumulation on Amazonian floodplains influenced by EI Nino/ Southern Oscillation, Nature, 425, 493-497. Affonso, A. G., E. M. L. M. Novo, J. M. Melack, and L. L. Hess (2007), Identificac,;ao e quantificac,;ao do desflorestamento nas areas alagaveis nos municipios a margem do Rio Solimoes/Amazonas nos estados do Para e Amazonas, in Allais XIII Simposio Brasilerio de Sellsoriamellto Remoto, pp. 3235-3242, Inst. Nac. de Pesqui. Espaciais, Florian6polis, Brazil. Alsdorf, D. E. (2003), Water storage of the central Amazon floodplain measured with GIS and remote sensing imagely, AIIII. Assoc. Am. Geogr., 93, 55-66. Alsdorf, D. E., J. M. Melack, T. Dunne, L. A. K. Mertes, L. L. Hess, and L. C. Smith (2000), Interferometric radar measurements of water level changes on the Amazon floodplain, Nature, 404, 174-177. Alsdorf, D. E., L. C. Smith, and J. M. Melack (2001), Amazon floodplain water level changes measured with interferometric SIR-C radar, IEEE TrailS. Geosci. Remote Seils., 39, 423-431. Alsdorf, D., T. Dunne, J. Melack, L. Smith, and L. Hess (2005), Diffusion modeling of recessional flow on central Amazonian floodplains, Geophys. Res. Lett., 32, L21405, doi:l0.1029/ 2005GL024412. Alsdorf, D., P. Bates, J. Melack, M. Wilson, and T. DUillle (2007), Spatial and temporal complexity of the Amazon flood measured from space, Geophys. Res. Lett., 34, L08402, doi: 10.1029/ 2007GL029447. Arraut, E. M., J. E. Mantovani, and E. M. L. M. Novo (2007), Quanto alimento M para 0 Peixo-boi Amazonico? Tecnicas de processamento digital de imagens para estimar a dimensao de bancos de macr6fitas aquliticas, in Allais ,:ITII Simposio Brasilerio de Sellsoriamellto Remoto, pp. 6609-6614, Inst. Nac. de Pesqui. Espaciais, Florian6polis, Brazil. Barbosa, C. (2005), Sensoriamento remoto da dinamica de circulac,;ao da agua do sistema planicie de Curuai/Rio Amazonas,
MELACK ET AL. Ph.D. thesis, 281 pp., Inst. Nac. de Pesqui. Espaciais, Sao Jose dos Campos, Brazil. Barbosa, C. C. F., E. M. L. M. Novo, J. M. Melack, R. M. Freitas, and W. P. Filho (2006), Metodologia de analise da dinamica de area e volume inundavel: 0 exemplo do varzea do Lago Grande de Curuai, Rev. Bras. Cartogr., 58, 201-210. Bates, P. D., and A. P. J. DeRoo (2000), A simple raster-based model of floodplain inundation, J. Hydrol., 236, 54-77. Belger, L. (2007), Fatores que influem na emissao de CO 2 e CH4 em areas alagaveis interfluviais do medio Rio Negro, Ph.D. thesis, Inst. Nac. de Pesqui. da Amazonia, Manaus, Brazil. Birkett, C. M., L. A. K. Mertes, T. Dunne, M. H. Costa, and M. J. Jasinski (2002), Surface water dynamics in the Amazon Basin: Application of satellite radar altimetry, J. Geophys. Res., I07(D20), 8059, doi:1O.1029/200IJD000609. Bonnet, M. P., et al. (2008), Floodplain hydrology in an Amazon floodplain lake (Lago Grande de Cmuai), J. Hydrol., 349, 18-30. Coe, M. T., M. H. Costa, A. Botta, and C. Birkett (2002), Longterm simulations of discharge and floods in the Amazon Basin, J. Geophys. Res., I07(D20), 8044, doi: I 0.1 029/200 IJD000740. Coe, M. T., M. H. Costa, and E. Howard (2007), Simulating the surface waters of the Amazon River basin: Impacts of new river geomorphic and flow parameterizations, Hydro!. Processes, 22, 2542-2553, doi: 10.1002/hyp.6850. Cole, J. J., et al. (2007), Plumbing the global carbon cycle: Integrating inland waters into the terrestrial carbon budget, Ecosystems, 10,171-184, doi:10. 1007/s10021-006-9013-8. Corredor, M. C. F. V. (2004), Influencia das variac,;oes temporais da disponibidade relativa de habitats sobre a communidade de peixes em um lago de varzea da Amazonia central, Masters thesis, Inst. Nac. de Pesqui. da Amazonia, Manaus, Brazil. Costa, M. (2005), Estimate of net primaty productivity of aquatic vegetation of the Amazon floodplain using Radarsat and JERSI, lilt. J. Remote SeilS., 26, 4527-4536. Costa, M. H., M. T. Coe, and J. L. Guyot (2009), Effects of climatic variability and deforestation on surface water regimes, Geophys. MOllogr. Ser., doi: I0.1 029/2008GM000738, this volume. Devol, A. H., T. M. Zaret, and B. R. Forsberg (1984), Sedimentary organic matter diagenesis and its relation to the carbon budget of tropical Amazon floodplain lakes, Verh. lilt. Vel'. Limllol., 22, 1299-1304. Devol, A. H., J. E. Richey, W. A. Clark, S. L. King, and L. A. Martinelli (1988), Methane emissions to the troposphere from the Amazon floodplain, J. Geophys. Res., 93, 1583-1592. Devol, A. H., J. E. Richey, B. R. Forsberg, and L. A. Martinelli (1990), Seasonal dynamics in methane emissions from the Amazon River floodplain to the troposphere, J. Geophys. Res., 95, 16,417-16,426. Doyle, R. D. (1991), Primaty production and nitrogen cycling within the periphyton community associated with emergent aquatic macrophytes in an Amazon floodplain lake, Ph.D. thesis, Univ. of Md., College Park. Dunne, T., L. A. K. Mertes, R. H. Meade, J. E. Richey, and B. R. Forsberg (1998), Exchanges of sediment between the flood plain and channel ofthe Amazon River in Brazil, Geol. Soc. Am. Bull., 110,450-467.
Engle, D., and J. M. Melack (1993), Consequences of riverine flooding for seston an.d the periphyton of floating meadows in an Amazon floodplain. lake, Limllo!. Oceallogr., 38,1500-1520. Engle, D., and J. J'vY/Melack (2000), Methane emissions from the Amazon f1oodpJllin: Enhanced release during episodic mixing of lakes, Bioge0611emistly, 51, 71-90. Engle, D. L., J. M. Melack, R. D. Doyle, and T. R. Fisher (2008), High rates of net primaly productivity and turnover for floating grasses on the Amazon floodplain: Implications for aquatic respiration and regional CO 2 flux, Global Change Bioi., 14, 369-381. Forsberg, B. R., C. A. R. M. Araujo-Lima, L. A. Martinelli, R. L. Victoria, and J. A. Bonassi (1993), Autotrophic carbon sources for fish of the central Amazon, Ecology, 74,643-652. Frappart, F., F. Seyler, J.-M. Martinez, J. G. Le6n, and A. Cazenave (2005), Floodplain water storage in the Negro River basin estimated from microwave remote sensing of inundation and water levels, Remote Sens. Environ., 99,387-399. Graciani, S. D., and E. M. L. M. Novo (2003), Determinac,;ao da cobertura de macr6fitas, in Anais XIII Simposio Brasilerio de Sensoriamento Remoto, pp. 2509-2516, Inst. Nac. de Pesqui Espaciais, Florian6polis, Brazil. Guyot, J. L., N. Filizola, and A. Laraque (2005), Regimes et bilan du flux sedimentaire a Obidos (Para, Bresil) de 1995 a 2003, in Sediment Budgets, edited by D. E. Walling and A. J. Horowitz, IAHS Publ., 291, 347-354. Hamilton, S. K. (1999), Potential effects of a major navigation project (the Paraguay-Parana Hidrovia) on inundation in the Pantanal floodplains, Reg. Rivers Res. Manage., 15, 289-299. Hamilton, S. K., S. J. Sippel, and J. M. Melack (1995), Oxygen depletion and carbon dioxide production in waters of the Pantanal wetland of Brazil, BiogeochemistlY, 30, 115-141. Hamilton, S. K., S. J. Sippel, and J. M. Melack (2002), Comparison ofinundation pattems among major South American floodplains, J. Geophys. Res., 107(D20), 8038, doi:l0.1029/2000JD000306. Hamilton, S. K., S. J. Sippel, and J. M. Melack (2004), Seasonal inundation patterns in two large savanna floodplains of South America: The Llanos de Moxos (Bolivia) and the Llanos del Orinoco (Venezuela and Colombia), Hydrol. Processes, 18, 2103-2116. Hamilton, S. K., J. Kellndorfer, B. Lehner, and M. Tobler (2007), Remote sensing of floodplain geomorphology as a surrogate for biodiversity in a tropical river system (Madre de Dios, Pem), GeomOlphology, 89, 23-38. Han, S.-C., C. K. Shum, C. Jekeli, and D. Alsdorf(2005), Improved estimation of terrestrial water storage changes fi'om GRACE, Geophys. Res. Lett., 32, L07302, doi:1O.1029/2005GL022382. Haugaasen, T., and C. A. Peres (2005), Tree phenology in adjacent Amazonian flooded and unflooded forests, Biotropica, 37, 620-630. Hess, L. L., J. M. Melack, S. Filoso, and Y. Wang (1995), Delineation of inundated area and vegetation along the Amazon floodplain with the SIR-C synthetic aperture radar, IEEE Trans. Geosci. Remote Sens., 33, 896-904. Hess, L. L., et al. (2002), Geocoded digital videography for validation of land cover mapping in the Amazon basin, Int. J. Remote Sens., 23, 1527-1555.
539
Hess, L. L., J. M. Melack, E. M. L. M. Novo, C. C. F. Barbosa, and M. Gastil (2003), Dual-season mapping of wetland inundation and vegetation for the central Amazon basin, Remote Sens. Environ., 87, 404-428. Instituto Brasileiro de Geografia e Estatistica (1980-1990), Estatlstica da Pesca-Brasil-Grandes Regioes-Unidades da Federac,;ao, report, vols. 1-11, Rio de Janeiro, Brazil. Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renovaveis (1993-1998), Estatistica da Pesca-Brasil-Grandes Regioes-Unidades da Federac,;ao, report, vols. 12-17, Brasilia. Junk, W. J. (Ed.) (1997), The Central Amazon Floodplain: Ecology ofa Pulsing System, Springer, Berlin. Junk, W. J., and M. T. F. Piedade (1993), Biomass and primatyproduction of herbaceous plant communities in the Amazon floodplain, Hydrobiologia, 263,155-162. Junk, W. J., and M. T. F. Piedade (1997), Plant life in the floodplain with special reference to herbaceous plants, in The Central Amazon Floodplain: Ecology ofa Pulsing System, edited by W. J. Junk, pp. 147-185, Springer, Berlin. Junk, W. J., and M. T. F. Piedade (Eds.) (2009), Amazonian Floodplain Forests: Ecophysiology, Ecology, Biodiversity and Sustainable Management, Springer, Berlin, in press. Junk, W. J., J. J. Ohly, M. T. F. Piedade, and M. G. M. Soares (Eds.) (2000), The Central Amazon Floodplain: Actual Use and Options for a Sustainable Management, Backhuys, Leiden, Netherlands. Kalliola, R., J. Salo, M. Puhakka, and M. Rajasilta (1991), New site formation and colonizing vegetation in primaty succession on the western Ama~on flood plains, J. Ecol., 79,877-901. Kemenes, A. (2006), Estimativa das emissoes de gases de effeito estufa (C0 2 e CH4) pela hidrel6trica de Balbina, Amazonia central, Brasil, Ph.D. thesis, Inst. Nac. de Pesqui. da Amazonia, Manaus, Brazil. Kemenes, A., B. R. Forsberg, and J. M. Melack (2007), Methane release below a tropical hydroelectric dam, Geophys. Res. Lett., 34, Ll2809, doi:l0.1029/2007GL029479. Kemenes, A., B. R. Forsberg, and J. M. Melack (2008), As hidreletricas e 0 aquecimento global, Genc. Hoje, 41, 44-49. Kvist, L. P., and G. Nebel (2001), A review ofPemvian flood plain forests: Ecosystems, inhabitants and resource use, For. Eco!. Manage., 150, 3-26. Laraque, A., N. Filizola, and J. L. Guyot (2005), Variations spatio-temporelles du bilan sedimentaire dans Ie bassin amazonien bresilien, a pattir d'un echantillonnage decadaire, in Sediment Budgets, edited by D. E. Walling and A. J. Horowitz, IAHS Publ., 291, 250-258. Lesack, L. F. W. (1993), Water balance and hydrologic characteristics of a rainforest catchment in the central Amazon basin, Water Resow'. Res., 29, 759-773. Lesack, L. F. W. (1995), Seepage exchange through the lakebed in an Amazon floodplain lake, Limnol. Oceanogr., 40,598-609. Lesack, L. F. W., and J. M. Melack (1995), Flooding hydrology and ,mixture dynamics of lake water derived from multiple sources in an Amazon floodplain lake, Water Resow', Res., 31, 329-345. Lloyd, J., et al. (2007), An airborne regional carbon balance for central Amazonia, Biogeosciences, 4, 759-768.
540
FLOODPLAIN ECOSYSTEM PROCESSES
Malhi, Y, 1. T. Roberts, R. A Betts, T. J. Killeen, W. Lee, and C. A. Nobre (2008), Climate change, deforestation, and the fate of the Amazon: Science 319, 169-172. Marshall, B. G.;tB. R. Forsberg, and M. J. F. Thome-Souza (2008), Autotrophic energy sources for Paracheirodon axelrodi (Osteichthyes, Characidae) in the middle Rio Negro, central Amazon, Brazil, Hydrabiologia, 596, 95-103. Martinez, 1.;M., and T. Le Toan (2007), Mapping of flood dynamics and spatial distl'ibution of vegetation in the Amazon floodplain using multitemporal SAR data, Remote Sens. Environ.,
108,209-223.
'
Maurice Bourgoin, L. M., M.-P. Bonnet, J.-M. Martinez, P. Kosuth, G. Cochonneau, P. M. Moreira-Turcq, 1.-L. Guyot, P. Vauchel, N. Filizola, and P. Seyler (2007), Temporal dynamics of water and sediment exchanges between the Curuai floodplain and the Amazon River main stream, Brazil, J. Hydrol., 335, 140-156. Mayorga, E., A. K. Aufdenkampe, C. A Masiello, A V. Krusche, J. I. Hedges, P. D. Quay, 1. E. Richey, and T. A Brown (2005a), Young organic matter as a source of carbon dioxide outgassing from Amazonian rivers, Nature, 436, 538-541. Mayorga, E., M. G. Logsdon, M. V. R. Ballester, and 1. E. Richey (2005b), Estimating cell-to-cell surface drainage paths from digital networks, with an application to the Amazon basin, J. Hydml., 315,167-182. McClain, M. E., and R. 1. Naiman (2008), Andean influences on the biogeochemistry and ecology of the Amazon River, BioScience, 58, 325-338. McGrath, D. G., O. T. Almeida, M. Crossa, A Cardoso, and M. Cunha (2005), Working towards community-based ecosystem management of the lower Amazon floodplain, PLEC News Views, 6,3-10. Melack, 1. M. (1996), Recent developments in tropical limnology, Verh. Int. Vel'. Limnol., 26, 211-217. Melack, 1. M. (2004), Remote sensing of tropical wetlands, in Manual of Remote Sensing, edited by S. Ustin, pp. 319-343, John Wiley, New York. Melack, 1. M. (2005), Floodplain lakes and reservoirs in tropical and subtropical South America: Limnology and human impacts, in Lakes Handbook, edited by P. O'Sullivan and C. Reynolds, pp. 241-257, Blackwell, Oxford, U. K. Melack, J. M., and D. Engle (2009), An organic carbon budget for an Amazon floodplain lake, Verh. Int. Vel'. Limnol., in press. Melack, J. M., and T. R. Fisher (1983), Diel oxygen variations and their ecological implication in Amazon floodplain lakes, Arch. Hydrobiol., 98, 422-442. Melack, 1. M., and B. Forsberg (2001), Biogeochemistry ofAmazon floodplain lakes and associated wetlands, in The BiogeochemistlY of the Amazon Basin and its Role in a Changing World, edited by M. E. McClain et aI., pp. 235-276, Oxford Univ. Press, Oxford, U. K. Melack, 1. M., and L. L. Hess (2004), Remote sensing of wetlands on a global scale, SIL News, 42,1-5. Melack, 1. M., and L. L. Hess (2009), Remote sensing of the distribution and extent of wetlands in the Amazon basin, in Amazonian Floodplain Forests: Ecophysiology, Ecology, Biodiversity
MELACK ET AL.
and Sustainable Management, edited by W. 1. Junk and M. Piedade, Springer, Berlin, in press. Melack, 1. M., L. L. Hess, M. Gastil, B. R. For$berg, S. K. Hamilton, I. B. T. Lima, and E. M. L. M. Novo (2004), Regionalization of methane emissions in the Amazon basin with microwave remote sensing, Global Change Bioi., 10, 530-544. Mertes, L. A K. (1994), Rates of floodplain sedimentation on the central Amazon River, Geology, 22, 171-174. Mertes, L. A K. (1997), Documentation and significance of the perirheic zone on inundated floodplains, Water Resow'. Res., 33, 1749-1762. Mertes, L. A K., M. O. Smith, and 1. B. Adams (1993), Estimating suspended sediment concentrations in surface waters of the Amazon River wetlands from Landsat images, Remote Sens. Environ., 43, 281-301. Mertes, L. A. K., D. L. Daniel, 1. M. Melack, B. Nelson, L. A Martinelli, and B. R. Forsberg (1995), Spatial patterns of hydrology, geomorphology, and vegetation on the floodplain of the Amazon River in Brazil from a remote sensing perspective, GeomOlphology, 13, 215-232. Moreira-Turcq, P., J. M. Jouanneau, B. Turcq, P. Seyler, O. Weber, and 1. L. Guyot (2004), Carbon sedimentation at Lago Grande de Curuai, a floodplain lake in the low Amazon region: Insights into sedimentation rates, Palaeogeogr: Palaeoclimatol. Palaeoecol., 214,27-40. Nebel, G., 1. Dragsted, and A Salazar Vega (2001), Litter fall, biomass and net primary production in flood plain forests in the Peruvian Amazon, For. Ecol. Manage., 150, 93-102. Novo, E. M. L. M., F. A Leite, 1. Avila, V. Ballester, and J. M. Melack (1997), Assessment of Amazon floodplain habitats using TMILandsat data, Cienc. Cult., 49, 280-284. Novo, E. M. L. M., C. C. F. Barbosa, R. M. Freitas, Y. E. Shimabukuro, J. M. Melack, and W. P. Filho (2006), Seasonal changes in chlorophyll distributions in Amazon floodplain lakes derived from MODIS images, Limnology, 7, 153-161, doi: 10.1007/s 10201-006-0179-8. Novo, E. M. L. M., A G. Affonso, and 1. M. Melack (2007), Multisensor approaches to access the relationship between wetland deforestation and Amazon floodplain lake eutrophication, in Anais XIII Simposio Brasilerio de Sensoriamento Remoto, pp. 34833490, Inst. Nac. de Pesqui. Espaciais, Florian6polis, Brazil. Putz, R. (1997), Periphyton communities in Amazonian black- and whitewater habitats: Community structure, biomass and productivity, Aquat. Sci., 59, 74-93. Quay, P. D., D. O. Wilbur, J. E. Richey, 1. I. Hedges, and A H. Devol (1992), Carbon cycling in the Amazon River: Implications from Ol3 C compositions of particles and solutes, Limnol. Oceanogr., 37, 857-871. Richey, 1. E., A H. Devol, S. C. Wofsy, R. Victoria, and M. N. G. Ribeiro (1988), Biogenic gases and the oxidation and reduction of carbon in Amazon River and floodplain waters, Limnol. Oceanogr., 33, 551-561. Richey, 1. E., L. A. K. Mertes, T. Dunne, R. L. Victoria, B. R. Forsberg, A C. N. S. Tancredi, and E. Oliveira (1989), Sources and routing of the Amazon River flood wave, Global Biogeochem. Cycles, 3,191-204.
Richey, J. E., J. I. Hedges, A. H. Devol, P. D. Quay, R. Victoria, L. Mmiinelli, and B. R. Forsberg (1990), Biogeochemistry of carbon in the Amazon River, Limnol. Oceanogr., 35, 352-371. Richey, 1. E., 1. M. I\1elack, A K. Aufdenkampe, V. M. Ballester, and L. L. Hess (~002), Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric CO 2, Nature, 416,617-620. Richey, 1. E., A. V. Krusche, M. S. Johnson, H. B. da Cunha, and M. V. Ballester (2009), The role of rivers in the regional carbon balance, Geophys. Monogr. Ser., doi:10.1029/2008GM000734, this volume. Rodrigues, R. (2007), Diversidade floristica, estrutura da comunidade arb6rea e suas relayoes com variaveis ambientais ao longo do lago Amana, ROSA, Amazonia Central, Masters thesis, Inst. Nac. de Pesqui. da Amazonia, Manaus, Brazil. Ronchail, 1., G. Cochonneau, M. Molinier, 1.-L. Guyot, A Chaves, V. Guimaraes, and E. Oliveira (2002), Interannual rainfall variability in the Amazon Basin and sea-surface temperatures in the equatorial Pacific and the tropical Atlantic oceans, Int. J. Climatol.,22, 1663-1686. Rosenqvist, A, M. Shimada, B. Chapman, A Freeman, G. De Grandi, S. Saatchi, and Y Rauste (2000), The Global Rain Forest Mapping Project-A review, Int. J. Remote Sens., 21, 1375-1387. Rosenqvist, A, B. R. Forsberg, T. Pimentel, Y. A Raullte, and 1. E. Richey (2002), The use of spaceborne radar data to model inundation patterns and tr'ace gas emissions in the central Amazon floodplain, Int. J. Remote Sens., 23, 1303-1328. Rudorff, C. M., E. M. L. M. Novo, L. S. Galvao, and W. Pereira Filho (2007), Analise derivativa de dados hiperespectrais medidos em nivel de campo e orbital para caracterizar a composiyao de aguas opticamente complexas na Amazonia, Acta Amazonica, 37,269-280. Schongart, 1., and W. 1. Junk (2007), Forecasting the flood-pulse in central Amazonia by ENSO-indices, J. Hydrol., 335, 124-132. Schongart, 1., W. 1. Junk, M. T. F. Piedade, 1. M. Ayres, A. Hlittermann, and M. Worbes (2004), Teleconnection between tr'ee growth in the Amazonian floodplains and the El Nino-Southern Oscillation effect, Global Change Bioi., 10,1-10. Schongart, 1., M. T. F. Piedade, F. Wittmann, W. 1. Junk, and M. Worbes (2005), Wood growth patterns of Macrolobium acaciifolium (Benth.) Benth. (Fabaceae) in Amazonian black-water and white-water floodplain forests, Oecologia, 145,454-461. Silva, R. M. (2005), Fisicoquimica e macr6fitas no lago Amana, Masters thesis, Inst. Nac. de Pesqui. da Amazonia, Manaus, Brazil. Sioli, H. (1984), The Amazon, Dr. W. Junk, Dordrecht, Netherlands. Sippel, S. 1., S. K. Hamilton, and J. M. Melack (1992), Inundation area and morphometry oflakes on the Amazon River floodplain, Brazil, Arch. Hydrobiol., 123, 385-400. Sippel, S. 1., S. K. Hamilton, 1. M. Melack, and B. 1. ChoudhUly (1994), Detelmination of inundation area in the Amazon River floodplain using the SMMR 37 GHz polarization difference, Remote Sens. Envimn., 48, 70-76. Sippel, S. 1., S. K. Hamilton, 1. M. Melack, and E. M. M. Novo (1998), Passive microwave observations of inundation area and
541
the area/stage relation in the Amazon River floodplain, Int. J. Remote Sens., 19, 3055-3074. Siqueira, P., S. Hensley, S. Shaffer, L. Hess, G. McGa11'agh, B. Chapman, and A Freeman (2000), A continental-scale mosaic of the Amazon basin using JERS-I SAR, IEEE Trans. Geosci. Remote Sens., 38, 2638-2644. Smith, L. K., J. M. Melack, and D. E. Hammond (2003), Carbon, nitr'ogen and phosphorus content and 21OPb-derived burial rates in sediments of an Amazon floodplain lake, Amazoniana, 17, 413-436. Tapley, B. D., S. Bettadpur, 1. Ries, P. Thompson, and M. Watkins (2004), GRACE measurements of mass variability in the Earth system, Science, 305, 503-505. Thieme, M., B. Lehner, R. Abell, S. K. Hamilton, 1. Kellndorfer, G. Powell, and 1. C. Riveros (2007), Freshwater conselvation planning in data-poor areas: An example from a remote Amazonian basin (Madre de Dios River, Peru and Bolivia), Bioi. Conserv., 135,484-501. Wang, Y, L. L. Hess, S. Filoso, and 1. M. Melack (1995), Understanding the radar backscattering fi'om flooded and nonflooded Amazonian forests: Results fi'om canopy backscatter modeling, Remote Sens. Envimn., 54, 324-332. Welcomme, R. L. (1979), Fisheries Ecology ofFloodplain Rivers, Longman, London. Wilson, M. D., P. D. Bates, D. Alsdorf, B. Forsberg, M. Horritt, 1. Melack, F. Frappart, and 1. S. Famglietti (2007), Modeling largescale inundation of Amazonian seasonally flooded wetlands, Geophys. Res. Lett., 34, Ll5404, doi: 10.1029/2007GL0301 56. Wittmann, F., and W. 1. Junk (2003), Sapling communities in Amazonian white-waterjrorests, J. Biogeogr., 30, 1533-1544. Wittmann, F., D. Anhuf, and W. 1. Junk (2002), Tree species distr'ibution and community structure of central Amazon varzea forests by remote sensing techniques, J. Trap. Ecol., 18, 805-820. Wittmann, F., W. J. Junk, and M. T. F. Piedade (2004), The varzea forests in Amazonia: Flooding and the highly dynamic geomorphology interact with natural forest succession, For. Ecol. Manage., 196, 199-212. Wittmann, F., 1. Schongart, 1. C. Montero, T. Motzer, W. J. Junk, M. T. F. Piedade, H. L. Queiroz, and M. Worbes (2006), Tree species composition and diversity gradients in white-water forests across the Amazon basin, J. Biogeogr., 33,1334-1347. Worbes, M. (1997), The forest ecosystem ofthe floodplains, in The Central Amazon Floodplain: Ecology of a Pulsing System, edited by W. Junk, pp. 223-265, Springer, Berlin. B. R. Forsberg and M. T. F. Piedade, Instituto Nacional de Pesquisas da Amazonia, 69011-970 Manaus, AM, Brazil. ([email protected]; [email protected]) L. Maurice, LMTG, Universite de Toulouse, 14 Avenue Edouard Belin, F-31400 Toulouse, France. ([email protected]) J. M. Melack, Department of Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, CA 93106, USA ([email protected]) E. M. L. M. Novo, Instituto Nacional de Pesquisas Espaciais, c.P. 515, 12227-010 Sao Jose dos Campos, SP, Brazil. (evlyn@ ltid.inpe.br)
Effects of Clinlatic Variability and Deforestation on Surface Water Regimes Marcos Heil Costa Departamento de Engenharia Agricola, Universidade Federal de ViI;osa, Vi<;osa, Brazil
Michael T. Coe Woods Hole Research Center, Falmouth, Massachusetts, USA
Jean Loup Guyot LlYJTG, Institut de Recherche pour Ie Developpement, Toulouse, France
The river flow regime of the Amazon basin exhibits considerable variability at the interannual and interdecadal scales. A major source of variation is the El Nino-Southern Oscillation (ENSO) events. El Nino events cause a decrease in rainfall, river flow, and inundation patterns throughout the entire region, with the strongest reductions happening in the northern part of the basin. On the other hand, La Nina events cause increased river flow for the northern tributaries and the main stream, but apparently do not cause a discernible pattern of climate variability in the southern part of the basin. ENSO events are different one from the other. While most El Nino events cause reductions in precipitation and river flow over the entire area of northern Amazonia, some El Nino events change precipitation only over northwestern Amazonia. The strength of the ENSO events through the decades is modulated by an interdecadal signal possibly associated with the Pacific Decadal Oscillation. La Nina (rainy) events are rainier during the 1940s-l950s and 1970s, while El Nino (dry) events are drier during the 1960s and since the 1980s. It is also apparent that the interannual variability was damped during the 1930s-l960s. In addition to these modes of variability caused by valying rainfall patterns, in some regions where changes in land cover are extensive, changes in evapotranspiration may drive increases in river flow, increasing the runoff coefficient. This has been clearly documented for the Tocantins basin, and there is evidence that the Obidos runoff coefficient is also increasing. I
1. INTRODUCTION Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union. 10.1029/2008GM000738
The Amazon basin surface water system presents noticeable interannual and interdecadal variability that are related to precipitation patterns, in addition to more recent variability related to changes in land cover. This chapter reviews the 543
544
EFFECTS OF CLIMATIC VARIABILITY AND DEFORESTATION
effects of climate variability on river flow and floodplain extension in the Amazon basin, including the analysis of some extreme river flow events. The effects of changes in land cover on tH'e Amazon basin river flow are also investigated. In most cases, papers are reviewed in chronological order, to indicate the advance of knowledge in these fields. An orientation map (Figure 1) shows the geographical features mentioned in the chapter. 2. EFFECTS OF CLIMATE'vARIABILITY ON RIVER FLOW IN THE AMAZON BASIN The availability of long-term precipitation data sets is a critical issue in the analysis of the effects of climate variability on the flow of the Amazon River and tributaries. The Amazon basin is a large remote region, with low density of rain gauges, and velY few gauges have a time series that span more than 50 years (Figure 2a). The period of highest availability of data was the 1980s, with available rainfall and discharge data decreasing after that. Additionally, the nature of the rainfall in the region is mainly convective and extremely variable spatially, which makes the interpolation from the few available rain gauges uncertain. Considering these limitations, there are only two data sets that are candidates to analyze multidecadal precipitation variability in this region. The first is the University of East Anglia CRU TS2.1 data set [New et aZ., 2000], which spans 1901 to 2002. Initial
COSTA ET AL.
tests with this database indicate that data (for the Amazon basin) before 1940 is umeliable, lacking basic features like seasonality. The second one has been assembled by the Hydro-geodynamique actuelle du bassin amazonien (HYBAM, or Hydrology of the Amazon basin) program [Callede et aZ., 2002] from 46 rain gauges for the period 1940-2003. Any other studies that evaluate interdecadal variability in Amazonian precipitation before 1940 are based on velY few stations and may lack representativeness. The longest fluvial record in Amazonia is the height of water at Manaus harbor, which started in 1903. Most discharge stations in the Amazon basin started operation in the 1960s (Figure 2b). Because of the integrative nature of fluvial records, they are sometimes used to make inferences about upstream precipitation patterns. The effects of climate variability on the Amazon basin river flow have been studied since the 1980s. Initial studies focused on detecting the interannual variability, followed by studies that attributed the causes of this variability to nearby oceanic phenomena, like the El Nino-Southern Oscillation and specific patterns of Tropical Atlantic sea surface temperature (SST). More recently, interdecadal modes of climate and river flow variability have been detected and attributed. Perhaps the first long time series analysis of river flow variability in the Amazon basin was performed by Richey et aZ. [1989a]. A reconstmction of discharge data for the Amazon main stream (Rio Solimoes at Manacapum) for the period
1200
545
1--;======;------------------1 ilIlColombia II!I Ecuador
1000
II!I Peru
(a) Number of Rainfall Stations 1--------------
o Bolivia II Brazil
800
600 - 1 - - - - - - - - - - - - - - - 400 - 1 - - - - - - - - - - - - - - - , - ; ; - 200 - 1 - - - - - - - - - - - - -
O+---r--r.......,.._~~.~ 1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
250,---------------------------, II!I Ecuador
(b) Number of Discharge Stations
ilIl Peru
200
o Bolivia II!I Brazil
150 + - - - - - - - - " - - - - - - - - - -
100 + - - - - - - - - - - - - - - - - - - - - - 0
50 + - - - - - - - - - - - - - - -
o-l----,----,-r-,--,---r---r-,-----,-----,----'ll
1910
O·
O·
10· S
10· S
20· S
20· S
80·W
Figure 1. Orientation map.
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
Figure 2. Availability of (a) rainfall and (b) discharge stations in the Amazon basin, for the period 1910-2005, used by the HYBAM data set.
1903 to 1985 from stage data at Manaus produced three important results. First, there is no statistically significant trend in discharge over the period; second, there is a strong 2- to 3year mode of variability in the discharge con-elated with the El Nino-Southern Oscillation phenomenon (ENSO), with El Nino years having reduced river flow and La Nina years having increased river flow; third, the variability predates the major human influences in the basin and, therefore, is not a resllit of human changes in the river basin at that point. Marengo [1995] and Marengo et al. [1998] used discharge data from a larger number of sites than Richey et aZ. [1989a] to provide information on the spatial distribution of river flow variability in addition to temporal variability. The ENSO-correlated river flow variability has a clear spatial component, with low river flow concentrated in the northwestern portions ofthe basin (e.g., the Rio Negro) in El Nino
years. Additionally, these studies describe an apparent decadal-scale variability of the river flow, with both the northern and southern regions showing increased river flow in the period 1945-1960 and decreased river flow in the period 1970-1990. Guyot et aZ. [1998] used discharge data from 17 large sub-basins in Brazil, showing a strong discharge decrease (increase) during El Nino (La Nina) events also in the northeastern part of the Amazon basin (Rio Branco and Rio lari). Zeng's [1999] analysis of an 8-year record of discharge for the Amazon at Obidos and the Xingu (a southern tributary) found a significant con-elation of river flow to ENSO with a 7-month time lag, corroborating Marengo's results from different tributaries. The author attributes the 7-month time lag to a 3-month lag in the precipitation correlation to ENSO and the influence of soil water drainage on discharge
546
EFFECTS OF CLIMATIC VARIABILITY AND DEFORESTATION
timing. Discharge is not significantly correlated to simulated evapotranspiration (ET) indicating that interannual discharge variability is most strongly influenced by interannual precipitation :Variability, not temperature and ET. A study of climate and discharge data by Foley et aI, [2002] clarified the spatial-temporal influence of ENSO on the climate and stream flow of the Amazon basin. Discharge data from nine locations on the main stream and its major tributaries not analyzed in previous studies show that the mean response to El Nino conditions since 1950 is decreased river flow in the southern and eastern portions ofthe Amazon basin (rios Tocantins, Xingu, and Tapaj6s) in addition to the northern portions of the basin (e,g., Rio Negro) described by Marengo et al. [1998]. The mean response to La Nina conditions is for increased discharge on the main stream and nOlihern tributaries (Amazon at Obidos, Solimoes at Manacapuru, and Jurua). The southern and eastern tributaries have mixed modest response to La Nina conditions, with reduced river flow in the Tapaj6s and modestly increased river flow in the Xingu. Ronchail et al. [2005b] analyze the temporal variability of discharge of 80 gauge stations located in the Amazon basin of Brazil, Bolivia, and French Guiana, for the 1981-2002 period. During El Nino, discharge is less than normal in the northwestern rivers (Solimoes, upper Negro, and Japura rivers) and in the southernmost regions (upper Madeira, Tapaj6s, and Xingu basins). During La Nina, discharge is greater than normal in the northeastern basin and, in contrast, is less than nOlmal in the Madeira basin. However, the ENSO signal is not stationary at Obidos during the twentieth century, when considering high river flow and mean discharge. During the low-flow season, higher than normal discharges rates in a vast central region of the Amazon basin are associated with cold SST over the Northern Tropical Atlantic. Furihermore, a common, though not permanent, near-decadal signal is observed in Obidos low river flow and Northern Tropical Atlantic SST. The discharge of the southernmost Amazonian rivers is related to the South Atlantic SST; however, these signals are space and time dependent. At about the same time, Botta et al. [2002] applied a singular spectral analysis to the 1939-1998 precipitation data from the CRU climate data set and identified a 28-year mode of precipitation variability (which explains 35% of the total variance in the time series), confilming and augmenting previous results of Marengo et aI, [1998]. Coe et al. [2002, 2007] used this precipitation data set to reconstruct, through modeling, monthly mean discharge for this same period (Figure 3), verifying that years of relatively high discharge are clustered in the 1940s-1950s and the 1970s, while low discharge years are clustered in the 1960s and 1980s-1990s. This pattern, although spatially variable in magnitude, is
COSTA ET AL. Annual mean discharge of the Amazon River at 6bidos
200,000 180,000
'.,
160,000
'"E 140,000 120,000 + - - - - - J l - - - - - - - - - - - - - - - - - - 1 100.000 +---+--j--+--+----j--t----+--+----1I---I------1 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 -
Callade et al. (2002) reconstructed
-
Callade et al. (2002) from obs
. ' Coe et al. (2007)
Figure 3. Amazon discharge at Obidos for 1903-2006, as reconstructed by Cae et aI, [2007] and Callede et aI, [2002]. Thick solid lines indicate period where there are water level observations at Obidos, while thin solid lines are reconstructions from the relationship with the Manaus harbor water level.
not limited to any geographic region, and individual ENSO events are clearly visible, embedded within the 28-year mode of variability. Another reconstruction of time series discharge has been done by Cal/Me et al. [2002, 2004] for the Obidos gauging station (see Figure 1) on the Amazon. This is the river gauging station measuring the world's largest drainage area (4,677,000 km2, mean discharge 163,000 m 3 S-I). Daily discharge data has been calculated from a good rating curve based on acoustic Doppler current profileI' (ADCP) measurements for the 1969-1999 period and extended to the 1928-1946 period using historical water level data. Monthly discharge data for the 1903-1927 and 1947-1968 periods have been calculated using correlation between Manaus water levels (Rio Negro) and Obidos discharge (the Amazon; Figure 3). The mean annual discharge at Obidos increases between 1903 and 1922, is more stable during the 1922-1969 period, and a break is observed in 1969/1970 with increased values aftelwards. During the 1903-1999 period, the mean discharge trend, measured by the Spearman's and the Kendal's coefficients, is positive, and there is a +9% increase between the two dates. Regarding annual maximum floods, the discharge value of 250,000 m3 S-1 was reached or exceeded 12 times from 1970 to 1999 and only five times from 1903 to 1969. Although this increase in the number oflarge floods since 1970 is consistent with the period of higher river flows, the reduced large flood events before 1970 may be an artifact ofthe reconstruction of the previous time series. The annual minimum discharge shows a downward trend since 1981, but is relatively stable for the whole period 1903-1999. No significant break is observed,
Another important feature that may be observed in Figure 3 is the decreased variability of the annual mean river flow in the 1927-1969 period (0 = 10,221 m3 s-I), when compared to the other perio\l~ (0 = 18,854 m 3 s-1 for 1904-1926; 0 = 14,958 m 3 s-1 fo1' 1970-2006]. This damped variability is also observed in the CRU precipitation records for the midtwentieth centmy period [Botta et al., 2002]. Labat et al. [2004] used Morlet continuous wavelet analysis to explore the time variability structure of rainfall (19501999) and discharge data of the Amazon basin at Obidos (1903-1999). On interannual scales, rainfall and high and low river flows highlighted quasi-biennial oscillations. As in previous analyses, other interannual processes (3.8 to 5.5 years) appear to be associated with ENSO. Rainfall and high river flow are characterized by energetic interannual oscillations at the beginning and the end of the centulY. This time variability is possibly associated with the interdecadal changes in the Southern Oscillation Index (SOl), with intervals of high variance (1875-1920 and 1960-1990) and others oflow variance (1920-1960). However, low flows are closely associated with these scales during nearly the whole centulY. The 1970's shift corresponds to interdecadal oscillation coinciding with major changes in SST patterns in the Atlantic and the Pacific [Enfield and Mestas-Nulies, 1999]. Labat et al. [2005] used a new wavelet analysis method on the reconstructed monthly discharge data at Obidos for the 1903-2000 period, and two well-known 10ng-telID climatologic indexes [SOl and the NOlih Atlantic Oscillation (NAO)]. This analysis also finds a strong and near-stationary annual cycle, as well as various weaker interannual and decadal to multidecadal oscillations. The interannual time scale (from 2.4 to 6.2 years) is well defined before 1940 and after 1980, i.e., when the variance of ENSO is the strongest. The coherence with SOl at a near biannual time scale is impOliant and nearly pelIDanent, and a near-decadaI coherence with SOl is also strong. As in previous studies, in all cases, the cOlTelation is positive indicating that discharge is low during EI Nino events. The Amazon discharge is characterized by little coherence with NAO, except on interdecada1 time scales, Important elements of the climate variability are the analysis of extreme events, both maximum and minimum. Ronchail et al. [2006] studied major flooding events of the Amazon at Obidos (defined here as discharge events over 250,000 m3 s-I) for the period 1984-2001. During this period, major flooding on the main stream was related to the following features: (1) greater than usual high water in the Rio Solimoes, the largest tributaly, (2) delayed discharge peaks in the southwestern tributaries and the Rio Amazonas (Peru, Bolivia) and/or advanced discharge peaks on the Rio Negro-Branco, (3) and unusual Februaty-April discharge
547
peaks in the western and northwestern tributaries, patiicularly in the Rio Negro. These two last features contribute to the simultaneous inflow of a great quantity of water from all the tributaries in April-May favoring major flooding of the Amazon at Obidos. The lowest annual river flow events of the Amazon, recorded at Manaus, happened in 1906,1916,1926,1958,1962, 1997, and 2005. Of these, two are remarkable because they have extended for a long period: 1926 and 2005 (Figure 4). The 1926 drought in Amazonia, the drought of the centuly, has been described by Williams et aI, [2005]. This drought is characterized by the lowest maximum river level and by a very low minimum river level (Figure 4). The year 1926 is widely recognized as an El Nino year, with reductions of 30-40% of rainfall upstream of Manaus, but the drought may have been concentrated in western Amazonia, with increased rainfall over eastern Amazonia and NOliheast Brazil. A recent analysis of the 2005 drought that severely affected the southwestern and central Amazon basin and comparison to previous droughts by Marengo et al. [2008] give interesting insights into the complex influences of climate variability on river discharge. This study shows that many of the severe droughts in the historical record (e.g" 1906, 1912, 1926, 1992, and 1997) are associated with strong El Nino events and are characterized by strongly decreased peak discharge on the main channel of the Amazon. Anomalously low river flow in these drought years is driven largely by decreased river flow fi'om the northern tributaries (e.g., Japura and Negro). However, the 2005 drought had velY different characteristics from these ENSO-driven droughts, with the decrease in river flow limited to the southern and western tributaries (e.g., Madeira, Tapaj6s), As a result, the low discharge anomaly is transferred downstream to the main channel but does not impact the river flow on the main channel until after the peak discharge season; the dly season river flow in the main channel is severely reduced in 2005 and 1962, not the peak wet season river flow as is the case in El Nino droughts. The authors attribute this difference to climatic factors driving anomalously warm sea surface temperatures in the tropical North Atlantic Ocean. 3. EFFECTS OF CLIMATE VARIABILITY ON AMAZON BASIN FLOODPLAIN EXTENSION The Amazon River system has a long and large flood season that is an impOliant part ofthe basin hydrology and ecology. In the lower reaches, the slow annual rise and fall of the Amazon has enabled floodplain flora and fauna to adapt to the flood pulse [Junk et al., 1989]. Many trees species fruit during the flood season and aquatic fauna have adapted to
COSTA ET AL.
548 EFFECTS OF CLIMATIC VARIABILITY AND DEFORESTATION 35
Max
30
I
=29.69 m
25
«> > .l!2
...,Qj
20
~
15 Min
=13.64 m
10
1910
1920
1930
1940
1950
1960
1970
1960
1990
2000
Figure 4. Annual maximum and minimum water levels of the Rio Negro at Manaus. Average maximum water height is 27.76 m; average minimum water height is 17.59 m. take advantage of flooded forest resources [Goulding, 1980]. limited by insufficient data. For example, the United States Most of the commercially important migratOly fish species, Operational Navigational Charts, which were deemed to be for example, feed on fiuits and nuts released at this time, and the best source of global infOlmation on flood extension and feeding and reproductive migrations are closely coupled to were used in most early attempts [e.g., Matthews and Fung, 1987], did not delineate any wetlands in the Central Amazon the annual flood pulse [Armijo-Lima and Goulding, 1997]. basin. Recent work has concentrated on using satellite data, The total flooded area each season is, of course, directly numerical models, and analysis of sediment cores to provide related to the magnitude of the discharge flood wave. In one of flood extension and variability. estimates ofthe first large-scale examinations ofthe Amazon discharge Sippel et al. [1998] used passive microwave observations flood wave, Richey et al. [1989b] used discharge data from [from the Scanning Multichannel Microwave Radiometer the main stream and major tributaries ofthe Amazon to show (SSMR), on Nimbus-7] of smface brightness temperature that the long (~9-month) discharge flood wave is a function combined with empirical mixing models of landscape units of the scale ofthe basin and its climate. The very large scale to calculate mean monthly flooded area within 12 reaches of the drainage basin and the tropical climate result in a 3of the Amazon main stream for the period 1979(segments) month phase lag in precipitation and discharge between the 1986. This analysis at 25 km horizontal resolution is limited north and south draining tributaries. This leads to a relatively 2 to about,176,000 km of the main stream and parts of several long time scale over which high water river flow reaches the large tributaries, but it provides the first multiyear spatial Central Amazon compared to other large river systems (e.g., analysis of flood extent on the Amazon. The mean flooded Mississippi and major Arctic rivers). The combined effects area on the main stream varies between 26,700 km2 in Noof the precipitation phase lag and the surface transport delay 2 makes the peak discharge happens in June. In addition, the vember and 67,300 km in June and has considerable intervery large volume of water stored on the floodplain itself annual variation, with the maximum wet season flooded area 2 2 results in a strong attenuation of the receding flood wave, varying between 91,000 km in 1982 to 55,000 km in 1983. Using a regression model of flooded area as a function of thereby extending the season. A more complete understanding of the extent and inter- river stage at Manaus harbor, the authors estimated a 94-year annual variability of flooded area in the Amazon basin has record (1903-1996) of flooding on the main stream of the been limited by the relative inaccessibility of the region and Amazon. This much longer proxy for flooded area suggests the vast area affected by floods each year. Early attempts at that the maximum wet season flooded area varies between 2 quantifying the seasonally flooded area of the Amazon were greater than 90,000 km (in several years of the record) and
a minimum of about 30,000 km2 in 1926. A further analysis of this long-term proxy record by Hamilton et al. [2002] showed that mean maximum flooded area over this region and for this periqo is about 73,250 km 2 with a coefficient of variation of 14~. Hess et al. [2003] used synthetic aperture radar imagely acquired by the JERS-1 to analyze the flooded area and ecosystems of a much larger area (1.77 million km2) of the Central Amazon basin. This analysis is limited to dual-season mapping (one during the 1995-1996 wet season, the other during the 1995-1996 dry season) but is at much higher spatial resolution (100 m versus 25 km for Sippel et al.) and is extensively validated against high-resolution digital videography. Hess et al. estimate that about 17% of the 1.77 million km2 regions studied are wetlands, about half of which are on the main stream and major tributaries of the Amazon. Of the wetland area, 96% (about 280,000 km2) is flooded at high water in 1996 and 26% at low water. Flooded forest constituted the largest single vegetation class (70% of the wetland area), with nonvegetated, woodland and herbaceous vegetation classes making up most of the remainder of the wetlands. Open water and extensive areas of aquatic macrophytes are more common in the middle and lower reaches of the Amazon because of the presence of large lakes along the main stream, while forested regimes are more common in the upper reaches and tributaries. Cae et al. [2002] presented the first spatially explicit hydrologic model of the 6.7 million km2 Amazon river and floodplain system. Simulations of the hydrology of the basin for the period 1939-1995 illuminated the factors causing the high variability in the basin hydrology. The mean simulated flooded area in that study has a coefficient of variation of 18% and is strongly influenced by the 28-year and 3- to 4-year modes of variability in the precipitation. Consistent with the results of Zeng [1999], the variability of ET has little influence on the variability of the flooded area. Deviation from the mean is strongly positive in the 1940s-1950s and 1970s and negative in the 1960s and 1980s-1990s. The strongest negative departure from the mean (~-50%) occurs in 1992 coincident with the dly phase of the 28-year mode of variability and the El Nino of 1992-1993. The strongest positive anomaly of the record occurs in 1949 at the peak of a positive phase of the 28-year mode of variability. An analysis of the influences of ENSO by Foley et al. [2002] on the same simulated flood extension as Cae et al. [2002] showed significant correlation to the SOl with a 4to 5-month time lag, consistent with a 2-month lag of the precipitation with SOl and the residence time of the water within the soils and streams in the large river basin. On average, El Nino years are associated with decreased discharge from the southern and eastern tributaries of the Amazon and
549
decreased flooding on the main stream and those tributaries, while La Nina years are associated with increased discharge in the western and northern portions of the basin and increased flooding on the main stream. These studies concluded that, although ENSO plays an important role in the flood variability, the 28-year mode of precipitation variability [Botta et al., 2002; Cae et al., 2002] explains most of the interannual variation of the flood extent. Numerical simulation is an important tool to reconstruct floodplain patterns when remote sensing products are not available. The advent of the Shuttle Radar Topography Mission (SRTM) topographic data [Kellndorfer et al., 2004] and morphometric data throughout the Amazon has led to considerable improvements in our capabilities to simulate flood extension (Figure 5) [Cae et al., 2007]. Continued expansion of satellite-derived products (L. L. Hess et a1., unpublished data, 2007) will provide much clearer quantification offloodedarea throughout the entire basin, at least for dual-season analysis. Coupled with improved climate data sets and climate analysis, one may expect continued improvements to our understanding of the variability and drivers of the Amazon flood extent. In addition to remote sensing and modeling, analysis of sediment cores is an important technique to study the spatial and interannual variability of the floodplain extension. Aalto
Figure 5. Simulated mean monthly Amazon flooded extent for the period 1939-1998 [Cae et al., 2007]. The fraction of each 5-min (~9 k111 x 9 km) grid cell flooded is shown from 100% flooded (black) to 0% flooded (white).
550
EFFECTS OF CLIMATIC VARIABILITY AND DEFORESTATION
et al. [2003] extracted a total of 276 sediment cores (65 to 160 cm deep) from the large floodplain of the Bolivian foreland basin (Mamore and Beni river basins), which receives sediment frOllt the rapidly eroding Andes Mountains. Cores were X-rayed to evaluate sedimentary structures, and clay normalized 210Pb profiles were measured using constant initial reach clay activity, unknown sedimentation procedure. More than 95% of the cores depict episodic silt sediment accumulation, as discrete packages of uniform age (typically 20-80 cm thick) across an observed depth range, indicating discrete large flood pulses. Any floodplain location receives sediment infi'equently: episodic sedimentation is the predominant mechanism for foreland floodplain accumulation. When accumulation dates from all cores are compiled, a wide pattern emerges for the Mamore basin: distinct sedimentation pulses are separated by years to decades (recurrence intervals of about 8 years) and correlates well with cold-phase ENSO events (La Nina), causing high rainfall and Rio Beni rapid-rise floods. More recently, two studies have investigated the relationship between recent flood events and ENSO, including its predictability. Ronchail et al. [2005a] investigated the relationships between inundations on the "Llanos de Mojos," drained by the Rio Mamore, rainfall and SSTs to determine whether inundations can be predicted. During the 1945-1999 period, 22 inundation events were recorded causing human and economic disasters. During this period, inundations have been associated with abundant rainfall in the Mamore basin, mainly in the Llanos and Yungas (Andean region). The role of rainfall in the inner dry Andes and downstream from Trinidad is more limited. When consecutive floods are observed, the groundwater storage contributes to the ocCUl1'ence of the second or third inundation events, as rainfall is generally weaker during the latter events. Rainfall in the Mamore basin is not well correlated with Sea Surface Temperature Anomalies (SSTA) in the Pacific and Atlantic Oceans during the 1953-1999 period. However, during the 1990s, the southern Atlantic SSTA account for 50% of rainfal1 variability. Inundations are also related to negative SSTA differences between the tropical and subtropical southern Atlantic. Two thirds ofthe 22 inundation events occurred in association with this oceanic anomaly that features a weak SSTA gradient in the southern Atlantic. Rainfall in the Mamore basin is more abundant, and inundations are more frequent after 1970, when the annual rainfall experienced a significant 15% increase. Despite being associated with major El Nino events (1982-1983, 1991-1992), inundations in the Mamore basin are not significantly related to the equatorial Pacific SSTA. Schongart and Junk [2007] examined the relationship between ENSO and the height and duration of the flood wave
at Manaus using the gauge data for the period 1903-2004. Their analysis corroborates the importance of ENSO events and water height but also quantifies its relationship with the length of the flood season. They find that the flood season is shotiened by as much as 44 days in El Nino years and lengthened by as much as 31 days in La Nina years compared to years that are not influenced by ENSO events; 4. EFFECTS OF CHANGES IN LAND COVER ON AMAZON BASIN FLOW Tropical forest conversion disrupts the hydrological cycle of a drainage basin, by altering not only the balance between rainfall, evaporation, and soil moisture dynamics, but also the runoff response of the area. The conventional approach to quantifying the hydrological impacts of changes in land cover has involved the use of experimental catchments, with direct manipulation of land cover. These approaches, although expensive and time-consuming, have been applied for more than a centUly, and extensive results have been documented in the literature [Costa, 2005]. Bosch and Hewlett [1982] reviewed the results of94 basin experiments throughout the world, including different time scales and rainfall regimes and concluded that removal of forest leads to higher stream flow, and reforestation of open lands generally leads to a decline in the overall stream flow. Bruijnzeel [1990, 1996] reviewed the effect of land cover transfotmation in the humid tropics and concluded that: (a) carefully executed light selective harvesting of trees (up to 20% removal of biomass) has little (if any) effect on stream flow; (b) removal of the natural forest cover may result in a considerable increase in water yield (up to 800 mm a-I), depending on the amount of rainfall received and the degree of surface disturbance; and (c) there is a decline in stream flow with time associated with any reforestation. Sahin and Hall [1996] used regression analysis and data from 145 experiments to study the effects of land use change on water yields. Their analysis of the tropical forest data (only five experiments were included) suggests that runoff increased 10 mm a-I after a 10% deforestation and 213 mm a-I after 100% deforestation. Due to the costs of a large-scale experiment and the enormous environmental impacts associated, it is not feasible to use direct experiments to determine the large-scale hydrological impacts of forest conversion. In this case, numerical modeling studies and analysis of climate records are the preferred tools. Costa and Foley [1997] used the Land Surface Scheme (LSX) coupled to a large-scale hydrological model to evaluate the changes in land cover (fi'om natural vegetation to pasture) in the hydrological budget of the Amazon basin, Their
COSTA ET AL.
simulation indicated a decrease in annual mean ET on the order of 0.5 mm d- I..!his reduction contributed to a higher production of runoff'that, when routed throughout thebasin, increased fro¢ 0 (for regions where natural vegetation is considered to/be grasslands) to a maximum of 47% (in regions with fOI'est vegetation cover). Although Richey et al. [1989a] concluded that the observed twentieth century flow variability is not a result ofhuman changes in the Amazon basin, they investigated a time series collected at a site that drained a region with relatively little change in land cover. Even today, agricultural land use upstream of Manacapuru is less than 10% of the land cover and is concentrated in the Andean border of the basin [Cardille et aI" 2002]. Other Amazonian regions have suffered strong changes in land cover throughout the last decades. One of these regions is the Tocantins basin, a region where more than 60% of the land is under some sort of agricultural use today. Costa et al. [2003] evaluated a 50-year long time series of discharge of the Rio Tocantins at POlio Nacional (175,360 2 km ), as well as precipitation over this drainage area, during a period where substantial changes in land cover OCCUlTed in the basin (1949-1998). Based on agricultural census data, they estimate that, in 1960, about 30% of the basin was used for agriculture, whereas Cardille et al. [2002] indicate that by 1995, agriculture had increased substantially, with about 49% of the basin area used as cropland and pastureland. A comparison of one period with little change in land cover (period 1, 1949-1968) with a period ofmore intense changes in land cover (period 2, 1979-1998), indicates that, while precipitation over the basin is not statistical1y different between period 1 and period 2, annual mean discharge in period 2 is 24% greater than in period 1, and the high-flow season discharge is greater by 28% (Figure 6). Costa et al. [2003] also estimated that the original land cover ET is 3.4 mm d- I, and the ET of the replacing pastureland is 2.7 mm d- I. Evidence of the effects of changes in land cover at a wider scale has been suggested by CallMe et al. [2004]. From their analyses, the mean annual rainfall for the whole basin (data from 43 rain gauges) for the 1944-1998 period correlates well with annual mean discharge at Obidos. From 1963 to 1973, rainfall and discharge normalized trends are similar and constant. Since 1974, the difference between these two trends (rainfall and discharge) is decreasing, suggesting that the increasing runoff is a possible consequence of the deforestation in the Brazilian part of the Amazon basin. In a follow up study, CallMe et al. [2008] used an extended period· of analysis (1903-2003 for the discharge) and mean annual rainfall over the whole basin recalculated on the basis of 163 rain gauges data for the 1940-2003 period. In the Obidos drainage basin, 341,000 km2 have been deforested between
551
7000
. , ,,
6000 5000
.
I ~~~ 1949-1968
••• , 1979-19981
..:; 4000
~ 3000
,
2000
"
1000
./
O+---,----,--,--,-----,---.---r---.--.--~__,
I
2
4
6 7 Months
8
9
10
11
12
Figure 6. Average river discharge (Q), for periods 1949-1968 and 1979-1998, for the Rio Tocantins at POlio Nacional. Modified from Costa et al. [2003].
1976 and 2003. The evolution of the runoff coefficient suggests a 5% increase of the mean annual discharge during 23 years (1981-2003). The number of strong floods has increased fivefold. No influence of deforestation has been detected in the low-water season. 5. CONCLUSIONS The river flow regime of the Amazon basin exhibits considerable variability at the interannual and interdecadal scales. Some of tHis variability is due to variability in climate, although the increasing deforestation in the basin is starting to become an impotiant source of long-term climate variability. A major source of variation of the river flow and floodplain extension in the Amazon basin is ENSO events. Warm events in the central Pacific (El Nino) cause a decrease in rainfall, river flow, and inundation patterns throughout the entire region, with the strongest reductions happening in the nOlihern part of the basin. On the other hand, cold events in the central Pacific (La Nina) cause increased river flow for the nOlihern tributaries and the main stream, but apparently do not cause a discernible pattern of climate variability in the southern pari of the basin. Although no conclusions have been reached yet, ENSO events are different one from the other [Trenberth, 1997]. While most El Nino events cause reductions in precipitation and river flow over the entire area of nOlihern Amazonia, some El Nino events change precipitation only over norihwestern Amazonia. The strength of the ENSO events through the decades is modulated by an interdecadal signal possibly associated with the Pacific Decadal Oscillation [Marengo et al., 1998; Botta et al., 2002; Marengo, 2004]. La Nina (rainy) events
552
EFFECTS OF CLIMATIC VARIABILITY AND DEFORESTATION
are rainier during the 1940s-1950s and 1970s, while El Callede, J., 1 L. Guyot, 1 Ronchail, Y. L'H6te, H. Niel, and E, de Oliveira (2004), Evolution du debit de l'Amazone a Obidos de Nino (dry) events are drier during the 1960s and since the 1902 a 1999, Hydrol. Sci. J., 49, 85-97. 1980s. It is also apparent that the interannual variability was Callede, J., J. Ronchail, J. L. Guyot, and E, de Oliveira (2008), damped dudhg the 1930s-1960s, with increased interannual Deboisement amazonien: Son influence sur Ie debit de l' Amazone variability before and after this period. These different ama 6bidos (Bresil), Rev, Sci, Eau, 21(1), 59-72. plitudes of variability are also tied to the interdecadal modes Cardille, J. A., 1. A. Foley, and M. H. Costa (2002), Characterizing ofthe Pacific. patterns of agricultural land use in Amazonia by merging satelIt has 'also been identified that some pattems of Atlantic lite classifications and census data, Global Biogeochem, Cycles, SST or the NAG may also influence precipitation, river flow, 16(3),1045, doi:1O.102912000GB001386. and inundation regimes in Amazonia, at both interannual Coe, M. T., M. H. Costa, A. Botta, and C. Birkett (2002), Longterm simulations of discharge and floods in the Amazon basin, J. and interdecadal scales, but these effects are secondary to Geophys, Res" 107(D20), 8044, doi:1O.1029/200lJD000740. the interannual and interdecadal Pacific modes. Coe, M. T., M. H. Costa, and E. Howard (2007), Simulating the The effects of climate variability on the river flow and surface waters of the Amazon River Basin: Impacts of new river inundation regimes are a consequence of the variability in geomorphic and dynamic flow parameterizations, Hydrol. Prorainfall. Interannual variations in temperature and ET have cesses, 22(14), 2542-2543, doi: 10.1 002/hyp.6850, not been documented to have a significant influence on the Costa, M. H. (2005), Large-scale hydrological impacts of tropical river flow and inundation variability [Zeng, 1999; Cae et al., forest conversion, in Forests, Water and People in the Humid 2002]. Tropics, edited by M. Bonell and L. A. Bruijnzeel, pp. 590-597. Despite this, in some regions where changes in land cover Cambridge Univ. Press, Cambridge, U.K. are extensive, changes in ET may drive increases in river Costa, M. H., and 1 A. Foley (1997), Water balance ofthe Amazon basin: Dependence on vegetation cover and canopy conductflow, increasing the runoff coefficient. This has been clearly ance, J. Geophys, Res., 102(D20), 23,973-23,989. documented for the Tocantins basin, and there is evidence Costa, M. H., A. Botta, and 1 A. Cardille (2003), Effects of largethat the Obidos runoff coefficient is also increasing. Acknowledgments, The authors wish to acknowledge Paul Lefebvre and Cleverson Alves for their efforts in producing graphics for this chapter.
REFERENCES Aalto, R., L. Maurice-Bourgoin, T, Dunne, D. R. Montgomery, C. Nittrouer, and 1 L. Guyot (2003), Episodic sediment accumulation on Amazonian floodplains influenced by El Nifio/Southern Oscillation, Nature, 425,493-497. Araujo-Lima, C., and M, Goulding (1997), So FI1Jitjitl a Fish: Ecology, Conservation, and Aquaculture of the Amazon's Tambaqui, 191 pp" Columbia Univ. Press, New York. Botta, A., N, Ramankutty, and J. A. Foley (2002), Long-term variations of climate and carbon fluxes over the Amazon basin, Geophys, Res, Lett" 29(9),1319, doi:10.102912001GL013607. Bruijnzeel, L. A. (1990), Hydrology of Moist Forests and the Effects of Conversion: A State ofKnowledge Review, UNESCO Int. Inst. for Aerospace Survey and Earth Science, Publication of the Humid Tropics Programme, 224 pp., Free Univ., Amsterdam. Bruijnzeel, L. A. (1996), Predicting the hydrological impacts of land cover transformation in the humid tropics: The need for integrated research, in Amazonian Deforestation and Climate, edited by J. H. C. Gash et al., pp. 15-56, John Wiley, Hoboken, N. 1 Callede, 1, J. L. Guyot, 1 Ronchail, M. Molinier, and E. de Oliveira (2002), L'Amazone a Obidos (Bresil). Etude statistique des debits et bilan hydrologique. Hydrol. Sci, J., 47, 321-333.
scale changes in land cover on the discharge of the Tocantins River, Southeastem Amazonia, J. Hydrol" 283, 206--217. Enfield, D. B" and A. M, Mestas-Nunez (1999), Multiscale variabilities in global sea surface temperatures and their relationships with tropospheric climate patterns, J. Clim, , 12, 2719-2733. Foley, J. A., A. Botta, M. T. Coe, and M. H. Costa (2002), El Nifio-Southern oscillation and the climate, ecosystems and rivers of Amazonia, Global Biogeochem. Cycles, 16(4), 1132, doi: 10.1029/2002GBOO 1872. Goulding, M. (1980), The Fishes and the Forest: Explorations in Amazonian Natural HistO/y, Univ. of Calif. Press, Los Angeles. Guyot, J. L., J. Callede, M. Molinier, V, Guimaraes, and E. de Oliveira (1998), La variabilite hydrologique actuelle dans Ie bassin de l'Amazone, Bull. Inst. Fr. Etudes Andines, 27(3), 779-788. Hamilton, S. K., S. J. Sippel, and 1 M, Melack (2002), Comparison ofinundation patterns among major South American floodplains, J. Geophys, Res., 107(D20), 8038, doi:10.1029/2000JD000306. Hess, L. L., 1 M. Melack, E. M. L. M. Novo, C. D. F. Barbosa, and M, Gastil (2003), Dual-season mapping of wetland inundation and vegetation for the central Amazon basin, Remote Sens. Environ., 87,404-428, doi:l0.l016/j.rse.2003.04.001. Junk, W., P. Bayley, and R, Sparks (1989), The pulse concept in river-floodplain systems, Proceedings of the Intemational Large River Symposium (LARS), Can. Spec. Publ, Fish. Aquat, Sci., 106, 110-127. Kellndorfer, J., W. Walker, L. Pierce, C, Dobson, 1 Fites, C. Hunsaker, J, Vona, and M. Clutter (2004), Vegetation height estimation from Shuttle Radar Topography Mission and national elevation datasets, Remote Sens, Environ., 93, 339-358. Labat, D., J. Ronchail, 1 Callede, 1 L. Guyot, E. De Oliveira, and W. Guimaraes (2004), Wavelet analysis of Amazon hydrological
COSTA ET AL. regime variability, Geophys. Res, Lett" 31, L02501, doi: 10.1 029/ 2003GL018741. Labat, D., 1 Ronch?,H, and 1 L. Guyot (2005), Recent advances in wavelet analyses: Part 2- Amazon, Orinoco and Congo discharges time s\?,llle variability, J. Hydrol., 314, 289-311. Marengo, J. A.·(I995), Variations and change in South American streamflow, Clim, Change, 21, 99-117, Marengo, 1 A. (2004), Interdecadal variability and trends ofrainfall across the Amazon basin, TheaI', App!. Climatol., 78,79-96. Marengo, J. A., J. Tomasella, and C, R. Uvo (1998), Trends in streamflow and rainfall in tropical South America: Amazonia, eastem Brazil, and northwestern Peru, J. Geophys. Res., 103(D2),1775-1783. Marengo, 1 A., C. A. Nobre, 1 Tomasella, M. D. Oyama, G. S. Oliveira, R, Oliveira, H. Camargo, L. M. Alves, and I. F. Brown (2008), The drought of Amazonia in 2005, J. Clim., 21, 495516. Matthews, E" and I, Fung (1987), Methane emission from natural wetlands: Global distribution, area, and environmental characteristics of sources, Global Biogeochem. Cycles, 1, 61-86. New, M., M, Hulme, and P. Jones (2000), Representing twentiethcentury space-time climate variability. Part II: Development of 1901-96 monthly grids of terrestrial surface climate, J. Clim., 13,2217-2238. Richey, 1 E., C. Nobre, and C. Deser (1989a), Aluazon River discharge and climate variability: 1903 to 1985, Science, 246, 101-103, Richey,l E., L. A, K. Mertes, T. Dunne, R. L. Victoria, B. R. Forsberg, A. C. N. S. Tancredi, and E.Oliveira (1989b), Sources and routing of the Amazon river flood wave, Global Biogeochem. Cycles, 3, 191-204. Ronchail, 1, L. BoutTel, G, Cochonneau, P. Vauchel, L. Phillips, A. Castro, 1 L. Guyot, and E. de Oliveira (2005a), Inundations in the Mamore basin (South-Westem Amazon - Bolivia) and sea-surface temperature in the Pacific and Atlantic oceans, J. Hydrol" 302,223-238, Ronchail, J., D. Labat, 1 Callede, G. Cochonneau, 1 L. Guyot, N. Filizola, and E. de Oliveira (2005b), Discharge variability
553
within the Amazon basin, in Regional Hydrological Impacts of Climatic Changes - Hydroclimatic Variability, IAHS Publ" vol. 296, edited by S, Franks, T, Wagener, E. B0gh, H. V, Gupta, L. Bastidas, C. Nobre, and C. 0, de Ga1vao, pp. 21-30, IAHS Press, Wallingford, u.K. Ronchail, 1, 1 L. Guyot, J. C. Espinoza, P, Fraizy, G. Cocholmeau, E. de Oliveira, N. Filizo1a, and J. 1 Ordonez (2006), Impact of the Amazon tributaries on major floods at 6bidos, in Climate Variability and Change Hydrological Impacts, IANS Pub!., vol. 308, edited by S. Demuth et al., pp. 1-6, IAHS Press, Wallingford, U.K. Sahin, V., and M. J. Hall (1996), The effects of afforestation and deforestation on water yields, J. Hydrol" 178, 293-309. Sch6ngart, 1, and W. 1 Junk (2007), Forecasting the flood-pulse in Central Amazonia by EN SO-indices, J. Hydro!., 335, 124-132, doi: 10.1 016/j.jhydrol.2006.11.005, Sippel, S, J., S. K. Hamilton, 1 M. Melack, and E, M. M, Novo (1998), Passive microwave observations of inundation area and the area/stage relation in the Amazon River floodplain, Int, J. Remote Sens., 19, 3055-3074. Trenberth, K. E. (1997), The definition ofEI Nifio, Bull, Am. Meteoral. Soc., 78,2771-2777. Williams, E., A. Dall'Antonia, V. Dall'Antonia, 1 M. de Almeida, F. Suarez, B, Liebmann, and A. C, M. Malhado (2005), The drought of the centmy in the Amazon basin: an analysis of the regional variation of rainfall in South America in 1926, Acta Amazonica, 35(2), 231-238, Zeng, N. (1999), Seasonal cycle and interannual variability in the Amazon hydrologic cycle, J. Geophys. Res" 104(D8), 90979106. M. T. Coe, Woods Hole Research Center, Falmouth, MA 02540, USA. M. H, Costa, Depatiamento de Engenharia Agricola, Universidade Federal de Viyosa, Viyosa, MG CEP 36570-000, Brazil. ([email protected]) 1 L. Guyot, LMTG, Institut de Recherche pour Ie Developpement, F-31400 Toulouse, France.
Results From LBA and a Vision for Future Amazonian Research M. Batistella, I P. Artaxo,2 C. Nobre,3 M. Bustamante,4 and F. Luiza0 5 This chapter summarizes selected results from the Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) and briefly describes a vision for future Amazonian research. The need for research on people and environment interactions is emphasized in the context of regional and global change. LBA developed institutional and scientific capacity in Amazonia, but its performance to promote sustainable development was restricted because the program has predominantly emphasized the advancement ofbasic knowledge, with less emphasis on integrative studies explicitly designed to influence public policies with consequences on land use and land cover. The challenge of transforming the natural goods ofAmazonia into human and economic benefits in an environmentally sustainable manner requires a new level of consciousness and collaborative work through the ability to move from simple diagnosis in the direction of actions at local, regional, and national levels. From the results of LBA, we may evolve to new experiences in which society will add value to the interactions between the biosphere and the atmosphere.
1. INTRODUCTION Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA) research has been steered by the recognition that Amazonia is undergoing a rapid and intense transformation associated with the processes of development and occupation. Changes in land use, land cover, and climate can affect the biological, chemical, and physical processes
lEmbrapa Satellite Monitoring, Campinas, Brazil. 2Institute of Physics, University of Sao Paulo, Sao Paulo, Brazil. 3Centro de Ciencias do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, Brazil. 4Department of Ecology, University of Brasilia, Brasilia, Brazil. 5Department of Ecology, INPA, Manaus, Brazil. Amazonia and Global Change Geophysical Monograph Series 186 Copyright 2009 by the American Geophysical Union 10.1029/2009GM000904
as well as the sustainable development of the region and its interactions with the regional and global climate. The journey through the chapters of this book confirms that LBA is a multidisciplinary program seeking to understand the functioning of Amazonian ecosystems and Amazonia as a regional entity of the Earth system. It also seeks to understand causes and consequences of the ongoing changes in the region and how to minimize impacts from its development. The original scientific agenda of LBA was designed around two overarching questions that are addressed through multidisciplinary research: (1) How does Amazonia function as a regional entity? (2) How will changes in land use and climate affect the biological, chemical, and physical functioning of Amazonia, including the sustainability of the region and the influence ofAmazonia on regional and global climate? Since the launching of the program in the 1990s, much has b~en accomplished to answer these questions, but much still has to be done through and beyond LBA science. This chapter summarizes lessons from LBA and briefly describes a vision for future Amazonian research. 555
556
BATISTELLA ET AL.
RESULTS FROM LBA AND VISION FOR FUTURE AMAZONIAN RESEARCH
2. SELECTED RESULTS FROM LBA
the early afternoon and less convection at night and in early morning over deforested areas. During the wet season, convective cloudiness is enhanced in the early night over deforested areas [Machado et al., 2004].
Since its beginning, LBA has produced countless publications, fucluding more than 2000 miicles in indexed journals, more than 200 book chapters, at least 13 books, and 10 special issues of scientific journals (Journal of Geophysical 2.2, Atmospheric Chemistly Research, Remote Sensing of Environment, Ecological ApThe effect of aerosols on the functioning of the Amazoplica{ions, Global Change Biology, Theoretical and Applied nian ecosystems was also extensively studied in LBA using Climatology, Acta Amazonica, Earth Interactions, Hydrologseveral approaches [Kesselmeier et al., this volume; Longo ical Processes, AtmospheriJ Chemistly and Physics, and the this volume; Artaxo et al., this volume; Davidson and et al., Brazilian Journal ofMeteorology). It is virtually impossible Artaxo, 2004; Artaxo et al., 2006]. The impact of trace gases to enumerate all the findings produced by LBA, but the foland aerosol emissions from biomass burnings in Amazonia lowing paragraphs highlight selected achievements, as sumwas also quantified [Artaxo et al., 2002], and its effects are marized in the LBA2 science plan [Batistella et al., 2007]. not limited to Amazonia, but reach a large portion of the Amazonia can be categorized as a region with high enSouth American continent. A strong influence of aerosol vironmental and social risks as far as climate change and particles was observed on mechanisms of formation and devariability are concerned. The risks are not only due to the velopment of clouds [Andreae et al., 2004] with impOliant predicted climate change, but also to the process of occupaimplications for the hydrological cycle. Cloud suppression tion of the region, including deforestation and changes in and a decrease in size of cloud droplets in areas with high land use. Models indicate the possibility of the occurrence, concentration of aerosols were also observed. The residence in the next decades, of an abrupt and irreversible replacetime of clouds is increased in the presence of large amounts ment of forested areas by vegetation with lower biomass of aerosols, which potentially reduces local precipitation and content, affecting biodiversity and the livelihoods of regional alters the radiation balance. New mechanisms of natural forpopulations. mation of clouds were observed with the production of cloud condensation nuclei from volatile organic compounds emit2.1, Physical Climate System ted by vegetation [Clayes et al., 2004]. The aerosols emitIn the context of climate change [Marengo et al., this ted through biomass burning can intercept large amounts of volume], projected scenarios until the end of this century solar radiation significantly altering the radiation balance. indicate possible reductions of up to 40% of rainfall and in- These particles also affect photosynthetic rates for forested creases of temperature of up to 8°C [Marengo et al" 2007]. areas far from the biomass-burning region, strongly influThe most severe impacts on the redistribution of species and encing carbon cycling in Amazonia. biomes would be felt in nOliheastern Amazonia, with less severe impacts in western Amazonia [Salazar et al., 2007]. Changes on land use have major effects on the meteorology of Amazonia [Nobre et al., this volume; Betts et al., this volume; Silva Dias et al., this volume; da Rocha et al., this volume]. In Rondonia, westerly and easterly intraseasonal wind oscillations were observed associated with different characteristics of rain, composition of aerosols, and cloud condensation nuclei. The extensive hydrological system of Amazonia affects the atmospheric circulation and the circulation pattern at the interface of large rivers with neighboring forest [Silva Dias et al., 2004]. Large areas of open water produce regional circulations that change the distribution patterns of clouds and the daily profile of winds at the forest-river interface, including alterations to the carbon flux in areas near large Amazonian rivers. Deforestation and changes in land use also influence cloud cover at seasonal scales and diurnal distributions [Durieux et al., 2003]. In the dry season, observations showed more low-level clouds in
2.3. BiogeochemistlY: Carbon Storage and Exchange, Other Trace Gases, and Nutrients
LBA has also conducted intensive studies into biogeochemical cycles [Davidson and Martinelli, this volume; Luiziio et al., this volume; Bustamante et al., this volume; Malhi et al., this volume; Phillips et al., this volume; Saleska et al., this volume; Houghton et al., this volume; Meir et al., this volume; Trumbore and de Camargo, this volume; Lloyd et al., this volume]. The cycling of nutrients such as phosphorus and nitrogen is critically important for ecosystems. The mechanisms and quantification of carbon cycling is one of the key scientific areas of LBA. The answer to the question if Amazonia is a source or a sink of carbon still remains open. But LBA has already produced important results about the mechanisms that regulate the carbon fluxes in natural ecosystems and in areas disturbed by land use changes. Research conducted throughout Amazonia showed a trend of
forest growth and biomass accumulation [Malhi et al., 2004]. This trend is especially stronger in western Amazonia, but its cause rem~}il.s unknown [Baker et al., 2004]. LBA has put in place a se,' of flux towers with semicontinuous operation over many>Years. The results of the flux towers also showed a predominant trend of carbon sink across several locations with some interesting examples of sites showing ecological disturbance as noticed in large quantities of aboveground dead biomass [Saleska et al., 2003]. These flux studies have also changed our conception about the seasonality of carbon fluxes in Amazonian forests. Several sites showed greater net carbon uptake during the dry season in comparison with the wet season, most likely due to the water availability in deep soils and larger photosynthetic radiation during the dry season [Silva and Avissar, 2006]. The knowledge of nuu'ient cycling is critical to the recovery of degraded areas in Amazonia. Davidson et al. [2007] described the complex mechanisms involved in phosphorus and nitrogen balance in chronosequences of vegetation recovery. It was observed that it takes at least 70 years to reestablish the nitrogen cycle after disturbance. 2.4. Land Surface Hydrology and Water Chemistly
Other studies demonstrated that land use and land cover changes, such as forest conversion to pastures, significantly alter the physical and chemical characteristics of rivers affecting their suucture and functioning [Richey et al., this volume; Tomasella et al., this volume; Melack et al., this volume; Costa et al., this volume]. When the forest is removed, the light availability favors the increase ofwater temperature and influences a series of chemical reactions such as the oxygen solubility [Neill et al., 2006]. The replacement of primary vegetation by pastures also enhances erosive processes [Thomas et al., 2004; Krusche et al., 2005]. Larger contributions of labile organic material cause an increase in the process of decomposition that is dependent on dissolved oxygen in water [Bernardes et al., 2004]. As a consequence, there is a decrease in the oxygen concentration with impacts on the biota. The nitrogen cycle is also affected since the available quantity of nitrogen decreases in pastures, thus decreasing its availability to small rivers, which become deficient for niu'ogen and not for phosphorous as they were originally [Carmo et al., 2005]. However, there are uncertainties as to whether these changes are extended to higher-order rivers, mainly those which drain mesoscale hydrological basins. 2.5, Land Use and Land Cover
ImpOliant advances were made through land use and land cover studies in LBA, as posed by the original research
557
questions [LBA, 1996; Alves et al., this volume; Asner et al., this volume; Schroeder et al., this volume]. In addition, new monitoring techniques, executed directly or indirectly within LBA, allowed the development of studies at different temporal and spatial scales, pariicularly concerning forested areas, During the entire period of LBA implementation, surveys on annual deforestation were conducted by the Brazilian National Institute for Space Research, which provided the scientific community with important information on rates and patterns of deforestation [Alves, 2002]. They also allowed LBA to identify new questions regarding local and regional patterns of land use, particularly regarding the dynamics of occupation in pioneering zones [Batistella and Moran, 2005], the dynamics of forest conversion to intensive agriculture [Morton et al., 2006], and the relationship between logging, forest degradation, and conversion rates [Asner et al., 2005a; Souza. et al., 2005]. Other studies indicated a variety of processes not only related to the expansion of agricultural frontiers but also to the intensification of land use and variations on production systems based on traditional and modern technologies. This work was conducted at several sites by means of robust remote sensing techniques and field work [Roberts et al., 2003; Lu et al., 2004]. The destiny of abandoned areas and their potential environmental services became extremely relevant, despite the evidence showing the r~duction of these areas in extremely degraded regions [Alves et al., 2003]. A significant advance was made in understanding the intensity and extension of logging in Amazonia as well as the possible environmental damage caused by this activity [Chambers et al., 2001]. Nepstad et al. [1999] found that areas under logging activities are equivalent in extension to areas deforested annually. Approximately 16% oflogged areas are converted to deforested area in the following year, and about 32% are deforested within 4 years. This means that logging does not occur prior to deforestation, but it is a way of disturbance in itself, facilitating the increase of impacted areas by human activities [Asner et al., 2005b]. Therefore, changes in land use can affect the environmental services, especially those related to the long-term functioning of ecosystems [DeFries et al., 2004]. The longtenn changes were assessed by Foley et al. [2007] through analysis of various LBA results. The authors indicated four examples of environmental services negatively affected by deforestation and degradation: carbon storage, hydrological flux, the influence on the regional climate, and disease vectors. Within plausible scenarios for changes in land use and land cover in Amazonia, analytical and theoretical tools were developed with the purpose of feeding regional climatic
558
RESULTS FROM LBA AND VISION FOR FUTURE AMAZONIAN RESEARCH
models including govemance altematives to current land use dynamics [Soares Filho et al., 2006]. It
2.6. Human Dimensions
Significant efforts were made in research on the human dimensions of environmental change in Amazonia within LBA [Walker et'al., this volume; Perz et al., this volume; Pfaff et al., this volume; Brondizio et al., this volume; Batistella et al., 2008]. Initially, ad hoc partn~rships were encouraged to discuss scientific questions identified' by the program community, in particular those aimed at reaching a better understanding of the processes of land use and land cover change. This action catalyzed the construction of more solid bridges with social scientists starting from their insertion into the LBA Science Steering Committee and, eventually, through systematic or programmatic initiatives. Among these initiatives, a survey of scientific production on Human Sciences, seminars and courses featuring the theme, and several publications can be considered the most significant results of the component on human dimensions ofLBA [da Costa et al., 2007]. 2.7. Training and Education
Besides the scientific production, the training and education activities are recognized as one of the major results and legacies of the program. More than 900 students were trained, including undergraduate, master, and doctorate students. In order to improve training of young scientists engaged in LBA, the Brazilian Council for Research (CNPq) created a special scholarship program. Three undergraduate, four master, and two doctorate courses were originated within LBA, all of them in Amazonian states. In addition to the scientific motivation, these human resources, now available in the region, represent a positive and cmcial factor toward a new vision for Amazonian research. 3. BEYOND SCIENTIFIC BARRIERS: PEOPLE, ENVIRONMENT, AND INTEGRATIVE RESEARCH IN AMAZONIA The first phase ofLBA tended to favor some research sites due to historical considerations and the experimental design based on ecological transects. Therefore, large areas and other important environments were undelTepresented. It is important now to review the geographic distribution of LBA efforts, seeking to balance the research planning, based on ecological and biogeographical aspects. This is vital for the integration of local, regional, and global studies. Basic environmental research must go on, but it is also cmcial to build new bridges for applications on sustain-
able development [Batistella and Luiziio, 2006]. The focus should continue to be the region as a whole, but with case studies in priority areas. For those integrated studies, a new research configuration is proposed, based on three disciplinary domains. One of these domains is being stmctured as "multiscale physical-chemical interactions on the biosphere-atmosphere interface." Its objective is to study the transpOliation and transformation of water, energy, trace gases, and aerosols in the Amazonian system and to identify the effects and impacts of human activities in the region. It is necessary to have a strong connectivity between the production of knowledge and its applications to the sustainable development, including technological innovations in cattle ranching, productive arrangements and chains, environmental predictions, and impacts on the human population. A second domain deals with the "social dimensions of the environmental changes and the dynamics of land use and land cover," with studies on the complex interactions between the environment and the society that characterizes the region. The third domain deals with "physical, chemical, and biological processes of aquatic and terrestrial systems and their interactions." Therefore, the issue on the environmental development strategies associated with sustainable practices of production becomes one of the important topics. Within those disciplinaly domains and after an extensive discussion with decision makers, stakeholders, and the scientific community, three research themes united the main questions to be addressed during the second phase of LBA: (I) the Amazonian environment in transformation; (2) the sustainability of environmental services and the systems of terrestrial and aquatic production; and (3) the climatic and hydrological variability and its dynamics: vulnerability, adaptation, and mitigation. An analysis of the results, to date, and the identification of remaining gaps reveal the need for regionalization through mesoscale integration. In particular, processes such as land degradation and land use intensification need to be studied and monitored using integrative approaches linking social and natural sciences. In this way, one can begin to constmct empirically based real-time approaches to policy making for Amazonia. We suggest the following actions as beneficial to the region, based on LBA science: (I) generation ofproducts at regional scale on land cover change based on multisenSOl' observations; (2) integration of biophysical data with the demands of social and economic agents, e.g., through zoning and territorial planning making use of science in management; and (3) modeling of the spatial distribution of deforestation, land degradation, and land use intensification and mapping vulnerabilities both biophysical and social.
To ensure the use of science in policy making and provide modeling with usable characteristics, it is important as we move fOlward·ill LBA: (l) to standardize data collec• rl hon whether at h4\:1usehold or less detailed scales of aggregation; (2) to chojJse models that permit proactive engagement of agents and other decision makers, not just scientists; and (3) to promote interaction among the scientists ofLBA, cutting across fields such as biogeochemistry, land change, social interactions, atmospheric chemistly, and regional climate models. Also, h"aining and education of young researchers as well as their engagement in local institutions will contribute significantly to enhance the capacity of Amazonians to lead and conduct studies about the functioning of ecosystems and its importance in the context of global change. There is a fundamental need to shift between regional analyses and case studies, as public policies based only on studies at regional scales may hide important sustainable practices. Moreover, the understanding of the characteristics, drivers, and consequences of land change in Amazonia should not depend only on land use and land cover studies per se but must include analyses ofthe human dimensions of environmental change. . The scientific community is engaged in a broad and participatory discussion on how to make the transition to the next decade of Amazonian research. For example, the Brazilian govemment, through the Ministly of Science and Technology, is coordinating the effOli to advance from the recognized success of the first phase of the program to the new phase of LBA, a challenging undertaking on integrative regional studies including complex linkages with global change issues. Much has to be done for LBA to progress from its multidisciplinary research to interdisciplinary and transdisciplinary integrative questions. In particular, it is vital to link local and regional processes with global change issues through mesoscale integration. Land use and land cover changes should still be at the core of the program because of their role in landscape transformation and consequent impacts on carbon dynamics, biogeochemical cycles, hydrological processes, atmospheric chemistly, and the physical climate. Recent processes in Amazonia, such as urbanization, agriculture intensification, and forest management should be of particular interest as they drive new changes to the region. An integrated perspective provides a rare opportunity to understand Amazonian ecosystems, landscapes, and region~ using a research strategy advocated by IGBP Phase II scientists [Moran et al., 2004]. This new science strategy views human-environment systems as integrated multiscaled systems and requires a new approach to scale up from case stud-
ies to the region, with particular att(~nti()ll nonlinearities. Some LBA imresltigllticlllS ha'{e(::orlttil:)lilt~d an integrative research agenda, thr'ough di~llo:gu(~s b'erureell natural and social sciences. A major challenge is regional differences as well as to understand 10c:al..sC,ale dynamics [Batistella and Brondizio, 2004; Batistella and Moran, 2007; Lahsen and Nobre, 2007]. A more complex framework for LBA research then arises, indicating the need for a continuous search for linkages between land use/land cover changes and other processes. LBA is initiating a new programmatic effort that will be essential to enhance its interface with other research initiatives. The recognition of continuous and creative reviews on the integration of social and natural sciences is among the incentives for a science of sustainability in Amazonia. 4. INSTITUTIONAL CONCERNS: A VISION FOR THE FUTURE Under the concept of institution building, we include a discussion about the social and environmental dimensions of change in Amazonia, as well as the articulation between the scientific and political domains. The role of communities in the management of natural resources and the challenges of institutional development has received special attention. The aliiculation between science and policy making is a necessalY step to respond to environmental change, taking advantage of concepts and results produced byLBA. The term "sustainability science" has been created to describe research aimed at facilitating the transition to sustainability. Focusing on the interaction between environmental sciences and science of development, science for sustainability is driven by the concem for the welfare of people and life on the planet, seeking to combine research activities and to understand how human and natural systems interact from local to global scales. Science for sustainability is also an initiative to include more social sciences, the priorities of least developed countries and subglobal foci, balanced against the general h'ends favoring the global environmental sciences and approaches that reflect priorities of most developed countries [Cash et al., 2003; Clark, 2007]. LBA indirectly promoted sustainability by developing the scientific and institutional capacity in the region, especially in Brazilian Amazonia, along with the fact that the knowledge and collected data can be freely accessed. The development of such capacity to face environmental change is necessary at the global level [Ambuj and VanDeveer, 2005], but it is especially so in less developed countries because they have fewer financial and human resources and are more
560
RESULTS FROM LBA AND VISION FOR FUTURE AMAZONIAN RESEARCH
vulnerable to multiple impacts resulting from rapid and simultaneous changes in social and environmental systems
[Kates et al., 2001]. However, thlcapacity ofLBA to promote sustainable development in Amazonia was restricted because the program has predominantly emphasized the natural sciences and the advancement of basic knowledge, with less emphasis in integrative studies explicitly designed to influence public policies and actions with consequences on land use and land cover in the region. As summarized by Batistella et al. [this volume], some notable LBA efforts have been undertaken successfully to promote the use of Science and Technology (S&T) for sustainable development in the region. The goals of sustainability could have been better achieved with more studies of this type. The difficulties in addressing sustainable development in the region reflect, in part, the difficulty of uniting the natural and social sciences [Philippi et al., 2003; Schor, 2005] and of understanding the relationship between scientific thinking and decision-making processes related to the design of public policies. Since the preliminmy discussions for the development of the first scientific plan ofLBA in the early 1990s, the perception about the role of Amazonia surpassed its geopolitical importance and its natural capital. Therefore, the challenge of sustainability remains. In response to worldwide growing concem about the risks of destructive environmental practices, govemments have launched environmental conservation plans for the region. Amazonia's biodiversity is now seen as the basis for cutting-edge sciences, especially biotechnology and geotechnology, capable of reconciling regional development and environmental conservation. However, this set of visions and good intentions is confi'onted with a well-established economic process based on the exploitation of natural resources as a mean to generate development. Such process is historically supported by the implementation of infrastructure projects and incentives for extensive land use change, with great potential to maintain the cycle of deforestation. The challenge of transforming the natural capital of Amazonia in human and economic benefits through an environmentally sustainable ma1l1ler requires a new level of consciousness and collaborative work. While S&T should play a central role to meet this challenge, there is little systematic knowledge about how to create institutions to effectively use S&T for sustainability. This challenge is significantly greater for Amazonia due to the absence of a model to be copied, that is, at least one tropical developed counhy or region with an economy strongly based on diversified natural resources, particularly forest resources, and intensive use of S&T through an educated and trained work-
BATlSTELLA ET AL.
force to promote development and environmental conservation [Nobre and Lahsen, 2008]. The solution to this dilemma depends on the ability to move from simple diagnosis in the direction of concrete, persistent, and integrated actions at local, regional, and nationallevels. While recognizing that much of this challenge is political, S&T have a central importance for the sustainable development of Amazonia, including the need for new knowledge about the environmental services provided by the region. The Amazonian countries should seek to implement a new development paradigm, minimizing deleterious environmental impacts. This is a huge challenge, particularly for developing countries with social liabilities. From the results ofLBA, we may evolve to new experiences in which society will add value to the interactions between the biosphere and the atmosphere. REFERENCES Alves, D. S. (2002), Space time dynamics of deforestation in Brazilian Amazonia, Int. J. Remote Sens., 23, 2903-2908. Alves, D. S., M. I. S. Escada, J. L G. Pereira, and C. A. Linhares (2003), Land use intensification and abandonment in Rondonia, Int. J. Remote Sens., 24, 899-903. Alves, D. S., D. C. MOlion, M. Batistella, D. A. Robelis, and C. Souza Jr. (2009), The changing rates and pattems of deforestation and land use in Brazilian Amazonia, Geophys. Monogr. Ser., doi: 10.1029/2008GM000722, this volume. Ambuj, D., and S. D. VanDeveer (2005), Capacity development for the environment: Broadening the scope, Global Environ. Politics, 5(3), 14-22. Andreae, M. 0., D. Rosenfeld, P. Artaxo, A. A. Costa, G. P. Frank, K. M. Longo, and M. A. F. Silva-Dias (2004), Smoking rain clouds over the Amazon, Science, 303,1337-1340. Artaxo, P., J. V. Martins, M. A. Yamasoe, A. S. Procopio, T. M. Pauliquevis, M. O. Andreae, P. Guyon, L V. Gatti, and A. M. C. Leal (2002), Physical and chemical properties of aerosols in the wet and dty seasons in Rondonia, Amazonia, J. Geophys. Res., 107(D20), 8081, doi: 10.1 029/200 1JD000666. Artaxo, P., P, H. Oliveira, L L Lara, T. M. Pauliquevis, L. V. Rizzo, C. Pires Jr., M. A. Paixao, K. M. Longo, S. de Freitas, and A. L. Correia (2006), Efeitos climaticos de particulas de aerossois biogenicos e emitidos em queimadas na Amazonia, Rev. Bras. Meteorol., 21(3), 1-22. Artaxo, P., et al. (2009), Aerosol particles in Amazonia: Their composition, role in the radiation balance, cloud formation, and nutrient cycles, Geophys. Monogr. Ser., doi: 10.1029/2008GM000778, this volume. Asner, G. P., D. E. Knapp, E. N. Broadbent, P. J. C. Oliveira, M. Keller, and J. N. Silva (2005a), Selective logging in the Brazilian Amazon, Science, 310, 480-482. Asner, G. P., D. E. Knapp, A. N. Cooper, M. M. C. Bustamante, and L. P. Olander (2005b), Ecosystem stmcture throughout the
561
Brazilian Amazon from Landsat observations and automated Cash, D. W., W. C. Clark, F. Alcock, N. M. Dickson, N. Eckley, spectral unmixing, Interact., 9(7), EIl34, doi: 10.1175/ D. H. Guston, J. Jager, and M. B. Ronald (2003), Knowledge EI134.1. systems for sustainable development, Proc. Natl. Acad. Sci. Asner, G. P., M. Keller, M. Lentini, F. Merry, and C. Souza Jr. U. S. A., 100(14), 8086-8091. (2009), Selective)ogging and its relation to deforestation, Geo- Chambers, .T. Q., J. Santos, R. 1. Ribeiro, and N. Higuchi (2001), phys. Monogr. Ser., doi:l0.1029/2008GM000723, this volume. Tree damage, allometric relationships, and above-ground net Baker, T., et al. (2004), Increasing biomass in Amazonian forest primary production in central Amazon forest, For. Ecol. Manplots, Phi/os. Trans. R. Soc. London, Ser. B, 359, 353-365. age., 152, 73-84. Batistella, M., and E. S. Brondizio (2004), Uma estrategia inte- Clark, W. C. (2007), Suslainability science: A room of its own, grada de analise e monitoramento do impacto ambiental de ashoc. Nat!. Acad. Sci. U. S. A., 104(6), 1737-1738. sentamentos mrais na Amazonia, in AvaliQl;iio e Contabilizac;iio Clayes, M., et al. (2004), Formation of secondmy organic aerosols de Impactos Ambientais, edited by A. R. Romeiro, pp. 74-86, through photooxidation of isoprene, Science, 303, 1173-1176. Editora Unicamp, Campinas. Costa, M. H., M. T. Coe, and J. L. Guyot (2009), Effects of climatic Batistella, M., and F. Luizao (2006), LBA and the future of Amavariability and deforestation on surface water regimes, Geophys. zonian research, Global Change Newsl., 67,4-5. Monogr. Ser., doi:1O.102912008GM000738, this volume. Batistella, M., and E. F. Moran (2005), Dimens6es humanas do uso da Costa, W. M., B. Becker, and D. Alves (Eds.) (2007), Dimene cobetiura das terras na Amazonia: Uma contribui<;ao do LBA, soes Humanas da Biosfera-Atmosfera na Amazonia, 178 pp., Acta Amazonica, 35(2), 239-247. EDUSP, Sao Paulo. Batistella, M., and E. F. Moran (2007), A heterogeneidade das mu- da Rocha, H. R., A. O. Manzi, and 1. Shuttleworth (2009), Evapodan<;as de uso e cobertura das terras na Amazonia: Em busca transpiration, Geophys. Monogr. Ser., doi:1O.1029/2008GM000744, de um mapa da estrada, in Dimensoes Humanas da Biosferathis volume. Atmosfera na Amazonia, edited by W. M. da Costa, B. Becker, Davidson, E. A., and P. Artaxo (2004), Globally significant and D. Alves, pp. 65-80, EDUSP, Sao Paulo. changes in biological processes of the Amazon Basin: Results Batistella, M., D. Alves, P. Artaxo, M. Bustamante, M. J(eller, F. of the Large-scale Biosphere-Atmosphere Experiment, Global Luizao, J. A. Marengo, L Martinelli, and C. A. Nobre (2007), Change BioI., 10,1-11. Plano Cient((ico LBA2-Programa de Pesquisas Sobre Inter- Davidson, E. A., and L. A. Matiinelli (2009), Nutrient limitations to ac;oes Biosfera-Atmosfera na Amazonia, LBA, Cachoeira Pausecondary forest regrowth, Geophys. Monogr. Ser., doi: 10.1029/ lista. (Available at http://lba.inpa.gov.br/lba/?p=plano_cientifico_ 2008GM000732, this volume. LBA2_vf2_1 &t= 1, accessed 10 August 2009). Davidson, E. A., et a1. (2007), Recuperation of nitrogen cycling in Batistella, M., E. F. Moran, and D. SAlves (Eds.) (2008), AmazoAmazonian forests following agricultural abandonment, Nature, nia: Natureza e Sociedade em TransformaC;Go, 304 pp., EDUSP, 447, 995-998. Sao Paulo. DeFries, R. S., J. A. Foley, and G. P. Asner (2004), Land-use Batistella, M., D. S. Alves, E. F. Moran, C. Souza Jr., R. Walker, choices: Balancing human needs and ecosystem function, Front. and S. J. Walsh (2009), People and environment in Amazonia: Ecol. Environ., 2, 249-257. The LBA experience and other perspectives, Geophys. Monogr. Durieux, L, L A. T. Machado, and H. Laurent (2003), The impact Ser., doi:10.1029/2009GM000902, this volume. of deforestation on cloud cover over the Amazon arc of deforBemardes, M. C., et al. (2004), Riverine organic matter compoestation, Remote Sens. Environ., 86,132-140. sition as a function of a land use changes, southwest Amazon, Foley, J. A., et al. (2007), Amazonia revealed: Forest degradation Ecol. Appl., 14,263-279. and loss of ecosystem goods and services in the Amazon Basin, Betts, A. K., G. Fisch, C. von Randow, M. A. F. Silva Dias, J. C. P. Front. Ecol. Environ., 5, 25-32. Cohen, R. da Silva, and D. R. Fitzjarrald (2009), The Amazonian Houghton, R. A., M. Gloor, 1. Lloyd, and C. Potter (2009), The boundaty layer and mesoscale circulations, Geophys. Monogr. regional carbon budget, Geophys. Monogr. Ser., doi:1O.1029/ Ser., doi: 10.1 029/2008GM000725, this volume. 2008GM000718, this volume. Brondizio, E. S., A. Cak, M. M. Caldas, C. Mena, R. BilsbolTow, Kates, R. W., et al. (2001), Environment and development: sustainC. T. Futemma, T. Ludewigs, E. F. Moran, and M. Batistella ability science, Science, 292(5517), 641-642. (2009), Small farmers and deforestation in Amazonia, Geophys. Kesselmeier, J., A. Guenther, T. Hoffinann, M. Piedade, and 1. Warnke Monogr. Ser., doi:1O.102912008GM000716, this volume. (2009), Natural volatile organic compound emissions from plants Bustamante, M. M. C., M. Keller, and D. A. da Silva (2009), and their roles in oxidant balance and particle formation, Geophys. Sources and sinks of trace gases in Amazonia and the cerrado, Monogr. Ser., doi: 1O.1029/2008GM000717, this volume. Geophys. Monogr. Ser., doi:1O.1029/2008GM000733, this vol- Kmsche, A. V., et al. (2005), Efeitos das mudan<;as do uso da terra ume. . na biogeoquimica dos corpos d'agua da bacia do rio Ji-Parana, Carmo, J. B., C. Neill, D. C. Garcia-Montiel, M. C. Piccolo, C. C. Rondonia, Acta Amazonica, 35, 197-205. CelTi, P. A. Steudler, C. A. Andrade, C. C. Passianoto, B. J. Lahsen, M., and C. A. Nobre (2007), Challenges of connecting Feigl, and J. M. Melillo (2005), Nitrogen dynamics during till international science and local level sustainability efforts: The and no-till pasture restoration sequences in Rondonia, Brazil, case of the Large-Scale Biosphere-Atmosphere Experiment in Nutr. Cycling Agroecosyst., 71,213-225. Amazonia, Environ. Sci. Policy, 10(1), 62-74.
562
RESULTS FROM LBA AND VISION FOR FUTURE AMAZONIAN RESEARCH
LBA (1996), Plano Cientifico Conciso, Programa de Grande Escala da Biosfera-Atmosfera na Amazonia, Cachoeira Paulista. Lloyd, 1., MvL. Goulden, J. P. Ometto, S. Patino, N. M. Fyllas, and C. A. Quesada (2009), Ecophysiology of forest and savanna vegetation, Geophys. Monogr. Ser., doi:IO.l029/2008GM000740, this volume. Longo, J(. M., S. R. Freitas, M. O. Andreae, R. Yokelson, and P. Artaxo (2009), Biomass burning in Amazonia: Emissions, long-range transport of smokl( and its regional and remote impacts, Geophys. Monogr. Ser., doi:IO.l029/2008GM000847, this volume. Lu, D., P. Mausel, M. Batistella, and E. F. Moran (2004), Comparison of land-cover classification methods in the Brazilian Amazon Basin, Photogramm. Eng. Remote Sens., 70(6),723-731. Luizao, F. 1., P. M. Fearnside, C. E. P. Cerri, and 1. Lehmann (2009), The maintenance of soil fertility in Amazonian managed systems, Geophys. Monogr. Ser., doi:lO.l029/2008GM000742, this volume. Machado, L. A., H. Laurent, N. Dessay, and 1. Miranda (2004), Seasonal and diurnal variability of convection over the Amazonia: A comparison of different vegetation types and large scale forcing, Theor. Appl. Climatol, 78,61-78. Malhi, Y., et al. (2004), The above-ground wood productivity and net primary productivity of 104 neotropical forests plots, Global Change Bioi., 10, 563-591. Malhi, Y., S. Saatchi, C. Girardin, and L. E. O. C. Aragao (2009), The production, storage, and flow of carbon in Amazonian forests, Geophys. Monogr. Ser., doi:lO.l029/2008GM000779, this volume. Marengo, 1., C. A. Nobre, R. A. Betts, P. M. Cox, G. Sampaio, and L. Salazar (2009), Global warming and climate change in Amazonia: Climate-vegetation feedback and impacts on water resources, Geophys. Monogr. Ser., doi:IO.l029/2008GM000743, this volume. Marengo, 1. A., C. A. Nobre, 1. Tomasella, M. D. Oyama, G. S. de Oliveira, R. de Oliveira, H. Camargo, and L. M. Alves (2007), The drought of Amazonia in 2005, J. Clim., 21(3), 495-516. Meir, P., et al. (2009), The effects of drought on Amazonian rain forests, Geophys. Monogr. Ser., doi: 10.1029/2008GM000882, this volume. Melack,1. M., E. M. L. M. Novo, B. R. Forsberg, M. T. F. Piedade, and L. Maurice (2009), Floodplain ecosystem processes, Geophys. Monogr. Ser., doi:1O.1029/2008GM00072I, this volume. Moran, E. F., D. L. Skole, and B. L. Turner (2004), The development ofthe international land use and land cover change (LUCC) research program and its links to NASA's land cover and land use change (LCLUC) initiative, in Land Change Science: Observing, Monitoring and Understanding Trajectories of Change on the Earth's SllIface, edited by G. Gutman et aI., pp. 1-17, Springer, Boston, Mass. Morton, D., R DeFries, Y. E. Shimabukuro, E. Arai, R. Freitas, L. O. Anderson, F. B. Espirito-Santo, and 1. Morisette (2006), Cropland expansion changes deforestation dynamics in the southern Brazilian Amazon, Proc. Nat!. Acad. Sci. U. S. A., 103,14,637-14,641. Neill, C., H. Elsenbeer, A. V. Krusche, J. Lehmann, D. Markewitz, and R O. Figueiredo (2006), Hydrological and biogeochemical
processes in a changing Amazon: Results from small watershed studies and the Large Scale Biosphere-Atmosphere Experiment, Hydrol. Processes, 20, 2467-2476. Nepstad, D., et al. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505-508. Nobre, C. A. and M. Lahsen (2008), Desenvolvimento sustentllvel na Amazonia: Desafios, potencial e 0 papel da ciencia e tecnologia, in Amazonia: Natureza e Sociedade em Transformaqiio, edited by M. Batistella, E. F. Moran, and D. S. Alves, pp. 291300, EDUSP, Sao Paulo. Nobre, C. A., G. O. Obregon, 1. A. Marengo, R. Fu, and G. Poveda (2009), Characteristics ofAmazonian climate: Main features, Geophys. Monogr. Sel'., doi: 10.1 029/2008GM000720, this volume. Perz, S., 1. P. Messina, E. Reis, R Walker, and S. 1. Walsh (2009), Scenarios of future Amazonian landscapes: Econometric and dynamic simulation models, Geophys. Monogr. Ser., doi:lO.l0291 2008GM000736, this volume. Pfaff, A., A. Barbieri, T. Ludewigs, F. Meny, S. Perz, and E. Reis (2009), Road impacts in Brazilian Amazonia, Geophys. Monogr. Ser., doi: 1O.l02912008GM000737, this volume. Philippi, A., Jr., F. R. de Aquino Neto, 1. Walker, F. R de Novais, and U. G. Cordani (2003), Revisiio de Meio Tel'll/o: Experimel/to de Grande Escala da Biofera-Atmosfera na Amazonia, LBA, Manaus, Santarem, Belem. Phillips, O. L., N. Higuchi, S. Vieira, T. R. Baker, K-1. Chao, and S. L. Lewis (2009), Changes in Amazonian forest biomass, dynamics, and composition, 1980-2002, Geophys. Monogr. Ser., doi:10.1029/2008GM000739, this volume. Richey, J. E., A. V. Krusche, M. S. Johnson, H. B. da Cunha, and M. V. Ballester (2009), The role of rivers in the regional carbon balance, Geophys. Monogr. Ser., doi:lO.l029/2008GM000734, this volume. Roberts, D. A., M. Keller, and 1. V. Soares (2003), Studies of land cover, land use, and biophysical propeliies of vegetation in the Large Scale Biosphere Atmosphere Experiment in Amazonia (LBA), Remote Sens. Environ., 87, 377-388. Salazar, L. F., C. A. Nobre, and M. D. Oyama (2007), Climate change consequences on the biome distribution in tropical South America, Geophys. Res. Lett., 34, L09708, doi:IO.l029/2007GL029695. Saleska, S. R., et al. (2003), Carbon in Amazon forests: Unexpected seasonal fluxes and disturbance-induced losses, Science, 302,li54-1157. Saleska, S., H. da Rocha, B. Kruijt, and A. Nobre (2009), Ecosystem carbon fluxes and Amazon forest metabolism, Geophys. Monogr. Sel'., doi:10.1029/2008GM000728, this volume. Schor, T. (2005), Ciencia e tecnologia: uma interpretayao da pesquisa na Amazonia-O caso do Experimento de Grande Escala da Biosfera-Atmosfera na Amazonia (LBA), Ph.D. thesis, Universidade de Sao Paulo, Sao Paulo. Schroeder, W., A. Alencar, E. Arima, and A. Setzer (2009), The spatial distribution and interannual variability of fire in Amazonia, Geophys. Monogr. Ser., doi:l0.l02912008GM000724, this volume. Silva, R. R., and R. Avissar (2006), The hydrometeorology of a deforested region of the Amazon Basin, J. Hydrometeorol., 7, 1028-1042.
BATISTELLA ET AL. Silva Dias, M. A. F.. iiP. L. Silva Dias, M. Longo, D. R. Fitzjarrald, and A. S. Dennil1g (2004), River breeze circulation in eastern Amazonia: O.~~ervations and modelling results, Theor. Appl. Climatol., 7{\1 111-121, doi:10.1 007/s00704-004-0047-6. Silva Dias, M: A., R. Avissar, and P. Silva Dias (2009), Modeling the regional and remote climatic impact of deforestation, Geophys. Monogl'. Ser., doi: 10. 1029/2008GM000817, this volume. Soares-Filho, B., et al. (2006), Amazon conservation scenarios, Nature, 440, 1-35. Souza, C. M., Jr., D. A. Roberts, and A. L. Monteiro (2005), Multitemporal analysis of degraded forests in the southern Brazilian Amazon, Earth Interact., 9(19), EI132, doi:l0.1175/EI132.I. Thomas, S. M., C. Neill, L. A. Deegan, A. V. Krusche, V. M. Ballester, and R L. Victoria (2004), Influences ofland use and stream size on particulate and dissolved materials in a small Amazonian stream network, Biogeochemistl)l, 68, 135-151. Tomasella, J., C. Neill, R. Figueiredo, and A. D. Nobre (2009), Water and chemical budgets at the catchment scale including nutrient exports from intact forests and disturbed landscapes, Geophys. Monogr. Ser., doi: 10.1 029/2008GM000727, this volume.
563
Trumbore, S., and P. B. de Camargo (2009), Soil carbon dynamICS, Geophys. Monogr. Ser., doi:l0.1029/2008GM000741, this volume. Walker, R, R. DeFries, M. del C. Vera-Diaz, Y. Shimabukuro, and A. Venturieri (2009), The expansion of intensive agriculture and ranching in Brazilian Amazonia, Geophys. MOl/ogr. Ser., doi: 10.1 029/2008GM000735, this volume.
P. Artaxo, Institute of Physics, University of Sao Paulo, Sao Paulo, SP 05508-050, Brazil. M. Batistella, Embrapa Satellite Monitoring, Avenida Soldado Passarinho 303, Fazenda Chapadao CEP 13070-115, Campinas SP, Brazil. M. Bustamante, Department of Ecology, University of Brasilia, CEP 70910-900, Brasilia, DF, Brazil. F. Luizao, Depmiment of Ecology, INPA, CEP 69060-001, Manaus, AM, Brazil. C. Nobre, Centro de Ciencias do Sistema Terrestre, Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, SP 12630, Brasil.
Index impacts of global change 273, 293, 555 instruments and techniques 145 interactions with particles and fields 183 isotopic composition and chemistly 451 land cover change 83, 293, 409, 451, 505, 555 land/atmosphere interactions 149, 251, 451 legislation and regulations 101, 555 limnology 525 middle atmosphere: composition and chemistry 183 middle atmosphere: energy deposition 389 modeling 83 newfields 43, 311 nutrients and nutrient cycling 299, 463 physical and biogeochemical interactions 505 plant ecology 1, 183 pollution: urban, regional and global I, 117 project evaluation 101 regional climate change 145, 555 regional modeling 251 remote sensing 5~5 restoration 117 ' science policy 1 soils/pedology 311, 451, 463 South America 83,101, 145,293 trace gases 489 tropical meteorology 251 water quality 261 water/energy interactions 261, 489, 543 watershed 485 wetlands 525
abrupt/rapid climate change 273 aerosols and particles 205, 231 agricultural systems 11,25,61,117,205,311,373 air/sea constituent fluxes 205 benefit-cost analysis 101 biodiversity 11,25,183,299,373,429 biogeochemical cycles, processes, and modeling 163,337, 355,429 biogeochemical kinetics and reaction modeling 409 biogeophysics 25, 43 biosphere/atmosphere interactions 25, 61,163,233,261,337, 373,389,463,489,543 carbon cycling 43,207,355,389,409,429,451,485,489 catchment 505 chemical kinetic and photochemical properties 389 climate change and variability 251,355 climate dynamics 11, 149,273,337,355,543 climate variability 149 cloud optics 233 cloud physics and chemistly 163,233 decision making under uncertainty 83 diel, seasonal, and annual cycles 261 eco-hydrology 485 ecosystems, structure and dynamics 1, 11, 43, 61, 117, 299, 337,373,409,429,451,463 floodplain dynamics 485, 525 general or miscellaneous 61, 299, 311, 543 geochemical cycles 163 global climate models 273 hydrological cycles and budgets 505
565
