Mediterranean Desertification
Mediterranean Desertification: A Mosaic of Processes and Responses Edited by N.A. Geeson
King’s College, University of London, UK C.J. Brandt
King’s College, University of London, UK and J.B. Thornes
King’s College, University of London, UK
Copyright 2002
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England Telephone (+44) 1243 779777
Email (for orders and customer service enquiries):
[email protected] Visit our Home Page on www.wileyeurope.com or www.wiley.com All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1T 4LP, UK, without the permission in writing of the Publisher. Requests to the Publisher should be addressed to the Permissions Department, John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England, or emailed to
[email protected], or faxed to (+44) 1243 770571. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the Publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought.
Other Wiley Editorial Offices John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany John Wiley & Sons Australia Ltd, 33 Park Road, Milton, Queensland 4064, Australia John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809 John Wiley & Sons Canada Ltd, 22 Worcester Road, Etobicoke, Ontario, Canada M9W 1L1 Library of Congress Cataloging-in-Publication Data Mediterranean desertification: a mosaic of processes and responses / edited by N.A. Geeson, C.J. Brandt, and J.B. Thornes. p. cm. Includes bibliographical references (p.). ISBN 0-470-84448-5 (alk. paper) 1. Desertification–Mediterranean Region. I. Geeson, Nichola. II. Brandt, C. Jane. III. Thornes, John B. GB618.68.M43 M4 2001 333.73′ 6′ 091822–dc21
2001046911
British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN 0-470-84448-5 Typeset in 9/11pt Times by Laserwords Private Limited, Chennai, India Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham, Wiltshire This book is printed on acid-free paper responsibly manufactured from sustainable forestry in which at least two trees are planted for each one used for paper production.
Contents
List of Contributors
ix
Preface
xv
Part 1
Thematic Issues
1
Section I
Introduction
3
Chapter 1
The Evolving Context of Mediterranean Desertification J.B. Thornes
5
Section II
Climate, Processes and Responses
13
Chapter 2
Extreme Climatic Events over the Mediterranean M. Conte, R. Sorani and E. Piervitali
15
Chapter 3
Potential Effects of Rising CO2 and Climatic Change on Mediterranean Vegetation C.P. Osborne and F.I. Woodward
33
Use of NOAA-AVHRR NDVI Data for Climatic Characterization of Mediterranean Areas Giovanni Cannizzaro, Fabio Maselli, Luciano Caroti and Lorenzo Bottai
47
Section III
Land Use, Processes and Responses
55
Chapter 5
The Effect of Land Use on Soil Erosion and Land Degradation under Mediterranean Conditions C. Kosmas, N.G. Danalatos, F. L´opez-Berm´udez and M.A. Romero D´ıaz
57
Chapter 4
Chapter 6
Agro-pastoral Activities and Land Degradation in Mediterranean Areas: Case Study of Sardinia G. Enne, G. Pulina, M. d’Angelo, F. Previtali, S. Madrau, S. Caredda and A.H.D. Francesconi
71
Chapter 7
Landscape Protection from Grazing and Fire N.S. Margaris and E. Koutsidou
83
Chapter 8
Bioengineering Principles and Desertification Mitigation J.N. Quinton, R.P.C. Morgan, N.A. Archer, G.M. Hall and A. Green
93
vi
Contents
Section IV
Physical Processes and Responses
Chapter 9
Differing Responses of Greek Mediterranean Plant Communities to Climate and the Combination of Grazing and Fire A. Dalaka, E. Papatheodorou, G. Iatrou, T. Mardiris, J. Pantis, S. Sgardelis, C. Lanara Cook, T. Lanaras, M. Argyropoulou, K.J. Diamantopoulos and G.P. Stamou
Chapter 10
Vegetation Cover Assessment in Mediterranean Semi-arid Landscapes F.J. Garc´ıa-Haro, J. Meli´a, M.A. Gilabert and M.T. Younis
Chapter 11
The Impact of Rock Fragments on Soil Degradation and Water Conservation B. van Wesemael, J. Poesen, C. Kosmas, N.G. Danalatos and J. Nachtergaele
107
109
119
131
Chapter 12
Aridification in a Region Neighbouring the Mediterranean ´ am Kert´esz, Tam´as Husz´ar, D´enes L´oczy, B´ela M´arkus, J´anos Mika, Ad´ Katalin Moln´ar, S´andor Papp, Antal S´antha, L´aszl´o Szalai, Istv´an T´ozsa and Gergely Jakab
147
Chapter 13
Soil Salinization in the Mediterranean: Soils, Processes and Implications L. Postiglione
163
Section V
Tools for Exploring Desertification
175
Chapter 14
Environmentally Sensitive Areas in the MEDALUS Target Area Study Sites A.C. Imeson and L.H. Cammeraat
177
Investigation on Environmental Characteristics to Underpin the Selection of Desertification Indicators in the Guadalent´ın Basin L.H. Cammeraat, A.C. Imeson and L. Hein
187
Chapter 15
Chapter 16
MEDRUSH: A Basin-scale Physically Based Model for Forecasting Runoff and Sediment Yield M.J. Kirkby, R.J. Abrahart, J.C. Bathurst, C.G. Kilsby, M.L. McMahon, C.P. Osborne, J.B. Thornes and F.I. Woodward
203
Part 2
Regional Studies
229
Section VI
The Guadalent´ın Basin, South-east Spain
231
Chapter 17
Natural Resources in the Guadalent´ın Basin (South-east Spain): Water as a Key Factor ´ and F. Belmonte F. L´opez-Berm´udez, G.G. Barber´a, F. Alonso-Sarria Serrato
Chapter 18
Local and Regional Responses to Global Climate Change in South-east Spain C.M. Goodess and J.P. Palutikof
233
247
Chapter 19
Chapter 20
Chapter 21
Chapter 22
Contents
vii
The Impact of Land Abandonment on Regeneration of Semi-natural Vegetation: A Case Study from the Guadalent´ın J.A. Obando
269
Lithology and Vegetation Cover Mapping in the Guadalent´ın Basin as Interpreted through Remote Sensing Data M.T. Younis, J. Mel´ıa, M.A. Gilabert, F.J. Garc´ıa-Haro and A.J. Bastida
277
Changing Social and Economic Conditions in a Region Undergoing Desertification in the Guadalent´ın Asunci´on Romero D´ıaz, Pedro Tobarra Ochoa, Franc´ısco L´opez-Berm´udez and Gonzalo Gonz´alez-Barber´a Management Plan to Combat Desertification in the Guadalent´ın River Basin L. Rojo Serrano, F. Garc´ıa Robredo, J.A. Mart´ınez Artero and A. Mart´ınez Ruiz
289
303
Section VII
The Agri Basin, Southern Italy
319
Chapter 23
General Description of the Agri Basin, Southern Italy F. Basso, E. Bove and M. del Prete
321
Chapter 24
The Agri Valley – Sustainable Agriculture in a Dry Environment: Crop Systems and Management F. Basso, M. Pisante and B. Basso
331
Chapter 25
Soil Erosion and Land Degradation F. Basso, M. Pisante and B. Basso
347
Chapter 26
Social and Economic Conditions of Development in the Agri Valley E. Bove and G. Quaranta
361
Chapter 27
Characterization of Soil Hydraulic Properties in a Desertification Context Alessandro Santini and Nunzio Romano
369
Chapter 28
Aspects of Forestry in the Agri Environment Agostino Ferrara, Vittorio Leone and Malcolm Taberner
385
Chapter 29
Modelling Large Basin Hydrology and Sediment Yield with Sparse Data: The Agri Basin, Southern Italy J.C. Bathurst, J. Sheffield, C. Vicente, S.M. White and N. Romano
397
Section VIII Conclusions
417
Chapter 30
419
Emerging Mosaics J.B. Thornes
Glossary
429
Index
433
List of Contributors
R.J. Abrahart
School of Geography, University of Nottingham, University Park, Nottingham NG7 2RD, UK
F. Alonso-Sarr´ıa
Laboratorio de Geomorfolog´ıa, Universidad de Murcia, Campus de “La Merced”, c/Santo Cristo 1, E-30001 Murcia, Spain
M. d’Angelo
Centro Interdipartimento di Ateneo NRD (Nucleo di Ricerca sulla Desertificazione), Dipartimento di Scienze Zootecniche, Universit`a degli Studi di Sassari, Facolt`a de Agraria, Via de Nicola, I-07100, Sassari, Italy
N.A. Archer
Division of Environmental and Applied Biology, Biological Sciences Institute, University of Dundee, Dundee DD1 4HN, UK
M. Argyropoulou
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
B. Basso
Dipartimento di Produzione Vegetale, Universit`a degli Studi della Basilicata, Via Nazario Sauro 85, 85100 Potenza, Italy
F. Basso
Dipartimento di Produzione Vegetale, Universit`a degli Studi della Basilicata, Via Nazario Sauro 85, 85100 Potenza, Italy
A.J. Bastida
Departamento de Geolog´ıa, Universtitat de Val`encia, Spain
J.C. Bathurst
Water Resource Systems Research Laboratory, School of Civil Engineering and Geosciences, University of Newcastle upon Tyne, Newcastle upon Tyne NE1 7RU, UK
F. Belmonte Serrato
Laboratorio de Geomorfolog´ıa, Universidad de Murcia, Campus de “La Merced”, c/Santo Cristo 1, E-30001 Murcia, Spain
L. Bottai
FMA, Via Einstein 36, 50023 Campi Bisenzio, Firenze, Italy
E. Bove
Dipartimento Tecnico-Economico perla Gestione del Territorio Agricolo-Foresstale, Universit`a degli Studi della Basilicata, Via Nazario Sauro 85, 85100 Potenza, Italy
L.H. Cammeraat
IBED-Fysische Geografie en Bodemkunde, Universiteit van Amsterdam, Nieuwe Achtergracht 166, NL 1018 WV Amsterdam, The Netherlands
G. Cannizzaro
TelespazioSpA, Via Tiburtina 965, 00156 Rome, Italy
S. Caredda
Centro Interdipartimento di Ateneo NRD (Nucleo di Ricerca sulla Desertificazione), Dipartimento di Scienze Zootecniche, Universit`a degli Studi di Sassari, Facolt`a de Agraria, Via de Nicola, I-07100, Sassari, Italy
L. Caroti
CeSIA-Accademia dei Georgofili, Logge Uffizi Corti 1, 50122 Firenze, Italy
x
List of Contributors
M. Conte (deceased)
Formerly at Istituto Fisica Atmosfera CNR, PZA L. Sturzo 31, 00144, Rome, Italy
A. Dalaka
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
N.G. Danalatos
Department of Agriculture, University of Thessaloniki, 38221 Volos, Greece
M. del Prete
Dipartimento di Produzione Vegetale, Universit`a degli Studi della Basilicata, Via Nazario Sauro 85, 85100 Potenza, Italy
K.J. Diamantopoulos
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
G. Enne
Centro Interdipartimento di Ateneo NRD (Nucleo di Ricerca sulla Desertificazione), Dipartimento di Scienze Zootecniche, Universit`a degli Studi di Sassari, Facolt`a de Agraria, Via Enrico de Nicola, 9-07100, Sassari, Italy
A. Ferrara
Dipartimento di Produzione Vegetale, Universit`a degli Studi della Basilicata, Via Nazario Sauro 85, 85100 Potenza, Italy
A.H.D. Francesconi
Centro Interdipartimento di Ateneo NRD (Nucleo di Ricerca sulla Desertificazione), Dipartimento di Scienze Zootecniche, Universit`a degli Studi di Sassari, Facolt`a de Agraria, Via Enrico de Nicola, 9-07100, Sassari, Italy
F. Garc´ıa Robredo
Fundaci´on Universidad Empresa de Murcia, Escuela de Negocios de la Regi´on de Murcia, Campus de Espinardo, 30100 Espinardo (Murcia), Spain
F.J. Garc´ıa-Haro
Remote Sensing Unit, Universitat de Val`encia, Dr Moliner 50, 46100-Burjassot, Val`encia, Spain
M.A. Gilabert
Remote Sensing Unit, Universitat de Val`encia, Dr Moliner 50, 46100-Burjassot, Val`encia, Spain
G. Gonz´alez-Barber´a
Departamento de Coservacion de Suelos y Agua, CEBAS-CSIC, Campus Universitario de Espinardo, Apartado 4195, 30080 Murcia, Spain
C.M. Goodess
Climatic Research Unit, University of East Anglia, Norwich, Norfolk NR4 7TJ, UK
A. Green
National Soil Resources Institute, Cranfield University, Silsoe, Bedford MK45 4DT, UK
G.M. Hall
National Soil Resources Institute, Cranfield University, Silsoe, Bedford MK45 4DT, UK
L. Hein
FSD, PO Box 570, NL 6700 AN Wageningen, The Netherlands
T. Husz´ar
Dept of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112 Budapest, Hungary
G. Iatrou
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
A.C. Imeson
IBED-Fysische Geografie en Bodemkunde, Universiteit van Amsterdam, Nieuwe Achtergracht 166, NL 1018 WV Amsterdam, The Netherlands
List of Contributors
xi
G. Jakab
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
´ Kert´esz A.
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
C.G. Kilsby
Water Resource Systems Research Laboratory, School of Civil Engineering and Geosciences, University of Newcastle upon Tyne, Newcastle upon Tyne NE1 7RU, UK
M.J. Kirkby
School of Geography, University of Leeds, Leeds LS2 9JT, UK
C. Kosmas
Laboratory of Soil Chemistry, Agricultural University of Athens, Iera Odos 75, 11855 Athens, Greece
E. Koutsidou
Department of Environmental Studies, University of the Aegean, “Xenia” Building, 81100 Mytilini, Lesvos, Greece
C. Lanara Cook
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
T. Lanaras
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
V. Leone
Dipartimento di Produzione Vegetale, Universit`a degli Studi della Basilicata, Via Nazario Sauro 85, 85100 Potenza, Italy
D. L´oczy
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
F. L´opez-Berm´udez
Department of Physical Geography, Laboratorio de Geomorfolog´ıa, Universidad de Murcia, Campus de “La Merced”, c/Santo Cristo 1, E-30001 Murcia, Spain
S. Madrau
Centro Interdipartimento di Ateneo NRD (Nucleo di Ricerca sulla Desertificazione), Universit`a degli Studi di Sassari, Facolt`a de Agraria, Via de Nicola, I-07100, Sassari, Italy
T. Mardiris
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
N.S. Margaris
Department of Environmental Studies, University of the Aegean, “Xenia” Building, 81100 Mytilini, Lesvos, Greece
B. M´arkus
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
J.A. Mart´ınez Artero
DGCONA, Ministerio de Medio Ambiente, Avda. Alfonso X El Sabio 6, 30008 Murcia, Spain
A. Mart´ınez Ruiz
Fundaci´on Universidad Empresa de Murcia, Escuela de Negocios de la Regi´on de Murcia, Campus de Espinardo, 30100 Espinardo (Murcia), Spain
F. Maselli
IATA-CNR, P. le delle Cascine 18, 50144 Firenze, Italy
M.L. McMahon
Infocom (UK) Ltd, York Science Park, York, UK
J. Meli´a
Remote Sensing Unit, Universitat de Val`encia, Dr Moliner 50, 46100-Burjassot, Val`encia, Spain
J. Mika
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
xii
List of Contributors
K. Moln´ar
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
R.P.C. Morgan
National Soil Resources Institute, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK
J. Nachtergaele
Laboratory for Experimental Geomorphology, Katholieke Universiteit Leuven, Belgium
J.A. Obando
Department of Geography, Kenyatta University, PO Box 43844, Nairobi, Kenya
C.P. Osborne
Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, UK
J.P. Palutikof
Climatic Research Unit, University of East Anglia, Norwich, Norfolk NR4 7TJ, UK
J. Pantis
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
E. Papatheodorou
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
S. Papp
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
E. Piervitali
CRATI s.c.r.l., Universit`a della Calabria, Rende (CS), Italy
M. Pisante
Dipartimento di Produzione Vegetale, Universit`a degli Studi della Basilicata, Via Nazario Sauro 85, 85100 Potenza, Italy
J. Poesen
Laboratory for Experimental Geomorphology, Katholieke Universiteit Leuven Redingenstraat 16, B-3000 Leuven, Belgium
L. Postiglione
Faculty of Agriculture, University of Naples Federico II, via Universit´a, 100, 80055 Portici (NA), Italy
F. Previtali
Dipartimento di Scienze dell’Ambiente e del Territorio, Universit`a di Milano–Biocca, Milano, Italy
G. Pulina
Centro Interdipartimento di Ateneo NRD (Nucleo di Ricerca sulla Desertificazione), Dipartimento di Scienze Zootecniche, Universit`a degli Studi di Sassari, Facolt`a de Agraria, Via de Nicola, I-07100, Sassari, Italy
G. Quaranta
University of Basilicata–DITEC, Via Macchia Romana, I-85100 Potenza, Italy
J.N. Quinton
National Soil Resources Institute, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK
L. Rojo Serrano
DGCONA, Ministerio de Medio Ambiente, Gran V´ıa de San Francisco 4, 28005 Madrid, Spain
N. Romano
Department of Agricultural Engineering, Division for Land and Water Resources Management, University of Naples “Federico II”, Via Universita’, 100, 80055 Portici (Naples), Italy
M.A. Romero D´ıaz
Department of Physical Geography, University of Murcia, Campus de “La Merced”, c/Santo Cristo 1, E-30001 Murcia, Spain
List of Contributors
xiii
A. S´antha
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
A. Santini
Department of Agricultural Engineering, Division for Land and Water Resources Management, University of Naples “Federico II”, Via Universita’, 100, 80055 Portici (Naples), Italy
S. Sgardelis
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
J. Sheffield
Department of Civil and Environmental Engineering, Princeton University, Princeton, New Jersey 08544, USA
R. Sorani
Servizio Meteorologico dell’Aeronautica, Rome, Italy
G.P. Stamou
Department of Biology, Aristotele University of Thessaloniki, GR 540 06 Thessaloniki, Greece
L. Szalai
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
M. Taberner
c/o Institute for Environment and Sustainability, Ispre, Italy
J.B. Thornes
Department of Geography, King’s College London, the Strand, London WC2R 2LS, UK
P. Tobarra Ochoa
Department of Fundamentals of Economical Analysis, University of Murcia, Spain
I. T´ozsa
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, H-1112, Budapest, Hungary
B. van Wesemael
D´epartement de G´eographie, Universit´e Catholique de Louvain, Place Louis Pasteur 3, B-1348 Louvain-la-Neuve, Belgium
C. Vicente
C/Cafetos #4, Col. Campestre, Cordoba, Veracruz 93653, Mexico
S.M. White
Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire, MK45 4DT, UK
F.I. Woodward
Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, UK
M.T. Younis
Remote Sensing Unit, Universitat de Val`encia, Dr Moliner 50, 46100-Burjassot, Val`encia, Spain.
Preface
Desertification has been recognized as one of the biggest problems facing the European Mediterranean countries. By desertification we mean land degradation resulting from various factors, including climatic variation and human impact, and it is the long history of human intervention, from Classical times onwards, that has particularly shaped the landscape here. Water resources have been exploited unsustainably, resulting in chemical pollution, salinization and exhaustion of aquifers. As economic activity has flourished in coastal areas so abandonment and degradation of land in the interior, previously sustained by traditional farming practices, have continued. Portugal, Spain, Italy and Greece are all now signatories to the United Nations Convention to Combat Desertification and implementation of the convention within national and regional action plans will require further organization of research and monitoring. The European Commission has funded a number of projects within the Environment Programme (DGXII), aimed at improving the understanding of the whole range of desertification issues. This book is based on the results of one of those projects, MEDALUS II, where 44 different universities and other institutions combined their expertise to clarify the processes of desertification operating in the Mediterranean environment, and the responses to those processes. Scientists of many disciplines, ranging from remote sensing to microbiology, researched climate, land use and the physical processes within soil and vegetation systems in order to design tools to describe and monitor desertification. Part 2 of this book describes how these processes and tools have been applied specifically. The regional studies illustrate how the application of remedial action cannot usually be uniform, but must respect the mosaic of physical environments and social and historical variations that interact within the geographical space of two of the target areas: the Guadalent´ın Basin of south-east Spain, and the Agri Valley of southern Italy. The editors feel privileged to have had the opportunity to work with the MEDALUS projects and to edit this book. All the authors should feel very proud of the unique spirit of co-operation that the projects have engendered. Each individual contribution makes up a part of the mosaic of our current knowledge, and the years of work behind this achievement are very much appreciated. Nichola Geeson Jane Brandt John B. Thornes Department of Geography, King’s College, University of London, UK November 2001
PART 1
THEMATIC ISSUES
Section I
Introduction
1
The Evolving Context of Mediterranean Desertification
J.B. THORNES
Department of Geography, King’s College London
1 INTRODUCTION In the last 10 years, the issue of desertification has not only become more widely recognized, both internationally and regionally, but the social and political framework has changed dramatically in a way that makes a change in the research approach crucial. It is the purpose of this chapter to outline these changes in order to set the context for further assessment of the problem. There have been a number of major syntheses that reflect the wider consciousness and appraisal of the problem. Despite these changes, the UNEP (United Nations Environment Programme) definition of desertification as “land degradation in arid, semi-arid and dry sub-humid areas resulting from various factors including climatic variations and human activities” remains as helpful today as it was in 1990 (UNEP 1990). Bearing in mind that “land” means the terrestrial bioproductive system that comprises soil, vegetation, other biota and the ecological and hydrological processes that operate within the system, the definition is particularly relevant. “Land degradation” means reduction and loss of the biological and economic productivity caused by land-use change, or by a physical process or a combination of the two. If anything, it would be useful to incorporate the rural depopulation implied in the French language usage, especially in a European context, where desertion of rural areas has been stressed as a pivotal problem in European Agricultural Reform. More light was spread on the problems of desertification in southern Europe by the conference held jointly by the Directorate General for Research of the European Commission and the Greek Government from 29 October to 1 November 1996. The proceedings have been published in two volumes (Balabanis et al. 1999, 2000). Another source is the documentation arising from the Concerted Action on Mediterranean Desertification, funded by the Research Directorate under Framework V and published in three volumes (Burke and Thornes 1998, in press a, b). A further important contribution, in addition to the publication of the two major books on the MEDALUS Project (Brandt and Thornes 1996; Mairota et al. 1998), is van der Leeuw’s brilliant synthesis of the Archaeomedes Project (van der Leeuw 1998).
2 AGENDA 21 AND SUSTAINABILITY At the international level, the UNCED Rio Conference of 1992 urged signatory nations to “reposition their economies, their societies and their collective purpose to maintain all life on earth, peacefully, equitably and with sufficient wealth to ensure that all are content in their survival” (O’Riordan and Voisey 1998, p. xiii). In Europe, this requirement was foreseen in the Fifth Environmental Plan, a precursor to the Rio Conference’s position on sustainability. Although progress has been relatively slow in some European countries and almost non-existent in others, the plan anticipates a level of public empowerment in environmental matters that will, in the longer term, enlighten environmental affairs. In Portugal, the establishment of Environmental Protection Associations at four different Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
6
Mediterranean Desertification
levels (Reibeiro and Rodrigues 1998) has strong affinities with the Land Care approach of Australia, in its emphasis on community and end-user involvement. This bodes well for the future contemplation of measures against desertification. Greece has been somewhat slower to act, according to Greek authors (Fousekis and Lekakis 1998), but the difficulties are the same: the lack of familiarity with, and acceptance of, consultation of the people; the late development of institutions of government, especially those concerned with environment; and the shortage of basic data that are required for decision making at the local level. Another major change since the start of the MEDALUS Project has been the shift in the Common Agricultural Policy (CAP) as a result of changing public awareness of the failures of the agricultural price support system and, specifically, the negotiations in 1992 of the General Agreement on Tariffs and Trade (GATT).
3
AGRI-ENVIRONMENTAL MEASURES AND AGENDA 2000
Perhaps the largest socio-economic change to occur in Europe that may be expected to have a bearing on the desertification problem is the Cork Declaration. In this, Commissioner Fischer stated his determination to reform the CAP into a more broadly based rural policy, integrating environmental issues. This was to bring to an end 40 years of price support and potentially affect an area of 141 million hectares, 44% of the total land surface of the 15 European Union states and probably change the landscape of Europe forever. There is a close link here to Agenda 21, because the need for sustainable agriculture is one of the key forces driving the reform of the CAP. This reform is called Agenda 2000. Here sustainable use means “The use of components of biological diversity in a way and at a rate that does not lead to the long-term decline of biological diversity, thereby maintaining its potential to meet the needs and aspirations of present and future generations” (according to the International Convention on Biodiversity, Rio 1992). In their assessment of the new CAP proposals, Birdlife International (1997) described the old CAP as “the engine of destruction in the countryside”. The productionist philosophy, with its link to regional and national development, led to the intensification of agriculture after the Second World War, through increased mechanization, fertilizer application and the promotion and extension of irrigation, that was so notable in the Mediterranean, leading to the conversion of dry farming to dense, fast-growing, heavily fertilized and pest-treated crops. It also led to a sharp increase in the demand for irrigation water and massive extraction of groundwater resources (see below). The original CAP (arising in the earliest days of the Community from the Treaty of Rome) was “an outdated, expensive, inefficient, inequitable and environmentally-damaging collection of policies that by 1992 was in need of further reform”. This was urgent for several reasons: • • • •
the proposed enlargement of the European Community; the requirement to meet the needs of the Uruguayan round of GATT and to prepare for the next round of World Trade negotiations with an emphasis on the reduction of trade-distorting subsidies; the commitment at the Rio Conference to promoting sustainable agriculture and protecting and enhancing the natural environment, as well as helping to meet the needs of rural communities; public demand for economic reform, relating to the budgetary costs and the economic inefficiency of the CAP.
The Commission decided to follow the McSharry approach to reforms by reducing support payments to world levels and replacing production incentives with direct payments. For example, the sharp cuts in beef and cereal prices are designed to allow food to be exported into world markets without subsidy enabling an enlarged EU to sell off its surpluses in these commodities. The East European countries that are joining will need to develop their agricultural systems in a sustainable way while meeting the needs of their rural communities. It is too early to see the outcomes of these policy shifts, which tend to be obscured by short-term fluctuations, such as the rise in grain prices that enabled export without subsidy in 1995–1996. The
The Evolving Context of Mediterranean Desertification
7
increased harvest in 1996 and the subsequent fall in world grain prices have reintroduced the need for export subsidies. The potential for significant increases in production brought some difficulties in meeting GATT limits on subsidized exports, requiring a significant increase in the set-aside rates. What is clear is that rural depopulation remains an important issue. A key objective of Agenda 2000 is to maintain the viability of rural communities by maintaining employment and incomes in rural areas through sustainable long-term use of resources. According to Birdlife International (1997), the number of people employed in agriculture in the 12 member states of the EU declined from 16.3 million in 1970 to 7.0 million in 1994, falling from 13.5% to 5.5% of total employment. At the same time, farm sizes and agricultural production have increased, resulting in increased levels of subsidies going to smaller numbers of farmers. As employment in agriculture continues to decline, the benefits of the (original) CAP are becoming less apparent. This valuable appraisal goes on to say that “Europe’s rural development problems cannot be addressed by support for agricultural production alone. They require a more integrated approach to rural policy, which places agriculture within the context of the whole rural economy” (Birdlife International 1997, p. 19). It is hard to disagree with this view and it must be added that the failure to address the most severe crisis in southern Europe, land degradation, highlights this lack of an integrated approach. The Agenda 2000 reforms are a great opportunity to couple economic regulations with environmental reforms. This has been done directly, to some extent through the “extensification measures” and indirectly through Environment Impact Directives. The ideal agri-environmental programme would, among other things, provide opportunities for all farmers to manage land for erosion mitigation rather than allow them to pass externalities (such as reservoir siltation) to the tax payers.
4 LAND ABANDONMENT It is often claimed that land abandonment invariably leads to land degradation and desertification, partially at least through the failure to maintain agricultural terraces. However, as Baudry (1991) points out, land abandonment is not a new phenomenon. It has been constantly occurring in Europe since 1950 and has been widespread in eastern North America since 1920. Rather than simply blame land abandonment on European Union policy, we need to know better what lessons can be learned from history. In the Mediterranean, there have been phases of strong outward migration. These have been both local (such as the impact of the Phylloxera plague on vines in the Spanish Alpujarra in the early years of the 20th century), and regional (as in the out-migrations for employment from southern Spain to northern Europe in the mid-20th century). Land abandonment does not necessarily mean that land is no longer used, either by agriculture or any other rural economy; it means a change in land use from the traditional or recent pattern to another, less intensive pattern. Nevertheless, we need to be able to identify how the landscape will change in relation to our knowledge of the erosion risk. Perhaps it is self-evident that the land at greatest risk is most likely to be abandoned. There are two sides to the coin: land abandonment occurs either because of external stresses and/or because of its inherently low productive capacity. Land abandonment occurs as a result of external driving forces, such as market changes, or internal changes that are “intrinsic”, for example if the system crosses some invisible threshold, such as the critical soil depth for plant growth. Once crossed, the tendency is for change to be negative, self-reinforcing and irreversible. Over the years, farming practice has brought the farming systems more stability, making them more resilient to changes. It is claimed that the mixed tree–grass–herb–grazing system of Extremadura, Spain (the dehesas) is highly stable to change because of its need for very low external inputs, its high biological diversity and the highly partitioned tree and herb layer (Bernaldez 1991). On the other hand, ecosystems are more unstable and susceptible to change when there is a strong competition between components. Thus Thornes (1990) was able to demonstrate the low stability in Mediterranean ecosystems where plants and soils compete for water, a situation that can lead to catastrophic changes as a result of small changes in the inputs and outputs (rainfall and grazing take-off, respectively). Progressive slow degradation
8
Mediterranean Desertification
can move the system towards an unstable state without the dangers being recognized. The trick is to identify the “position” of the threshold in state-space, so that trajectories towards instability can be recognized. The trajectory towards instability becomes apparent over time. After fire, it often takes 8–10 years before the pre-fire equilibrium between vegetation cover and sediment yield is reestablished. Unfortunately, abandonment and the associated neglect often bring the system rapidly to a threshold that, when crossed, may lead to irreversible erosion. Abandonment after ploughing results in a succession that requires about 20 years to reach equilibrium as a mature ecosystem under the prevailing grazing. Alpha diversity increases with succession and niche amplitude tends to diminish, the new plant species becoming specialists of increasingly narrower habitats (Pineda et al. 1981). Traditional sylvo-pastoral systems are subjected to either increases or decreases in grazing pressures. The former leads to destruction of natural pastures and the replacement of valuable grasses and legumes by unpalatable nitrophilous vegetation as has occurred at the MEDALUS field site in north-west Lesvos Island, Greece, observed by Kosmas et al. (1998). Replacement of nutritious herbs by rough pasture has also been described in Spain (De Miguel 1989). If this “matoralization” process proceeds unchecked, it eventually induces a decrease in biological diversity and a decrease in stability, as described by Naveh and Whittaker (1974), and an increase in fire risk.
5
WATER RESOURCES
Problems of water resources are inextricably bound to, but not synonymous with, desertification. As land degradation occurs, soil storage capacity is reduced, runoff increases and erosion thresholds are passed. The high inter-annual variability of rainfall moves Mediterranean soils inexorably towards the thresholds of land degradation as the pressure on vegetative cover increases through lack of soil moisture. The gathering pace of confidence in the observation of the existence of global warming and revised estimates by the ICCP indicate more difficult times ahead for hillslope hydrology as systems dry out. MEDALUS research suggests significant reductions in the biomass of grass and bushlands in areas having more than seven rain-free months per year in the Iberian Peninsula, as temperatures and atmospheric CO2 rise (Diamond and Woodward 1998), and estimates made by the Spanish Ministerio de Obras Publicas indicate important (17–20%) reductions in the flow of major Spanish rivers. Even accepting the scope for errors in these model estimates, the contemporary data already show that the supply of water for river flow replenishment and aquifer recharge is decreasing. In Mediterranean regions with average rainfalls of less than 300 mm per year, high inter-annual variability and high summer temperatures, there is a more or less continuous threat of water scarcity. In meteorological droughts this is caused by failure of precipitation, as has occurred in Italy, Greece and Spain in the last two decades of the last century. The whole of Italy was affected by severe drought during 1988–1999. A sequence of three years with low rainfall were accompanied by high temperatures; snow depths were also considerably reduced, with lower snowfalls than normal, combined with high temperatures. Large areas of Greece are susceptible to drought, notably eastern Greece and some Aegean islands. Catchments are often small and underlain by highly permeable karstic formations. There was an extended drought in the Athens area from 1987 to 1993, when rainfall was only 50% of normal, including two extremely dry years (1989/90 and 1991/2) that were the most severe over the last century. Most of Spain, except the north-west coast, was severely drought affected in the years 1990–1996. An analysis of seasonal rainfall (Institute of Hydrology and ISPRA 1999) indicates that the rainfall deficit was generally concentrated in winter and spring. Autumn rainfall was normal or above average and summer rainfall fairly regular. Mean percentage departure from normal rainfall exceeded −20% in the southern part of the country, which was worst affected. The drought reached its maximum coverage in September 1994 and August 1995 when rainfall reached −25%, and over two-thirds of Spain was affected. MEDALUS research by Goodess and Palutikof (Chapter 18) demonstrates the close coupling of the Atlantic Ocean pressure differences between the Azores High and the Iceland Low, on the one
The Evolving Context of Mediterranean Desertification
9
hand, and pressure fields over the Mediterranean that are linked to rainfall aberrations on the other. Earlier, Turkes (1996) showed, by the analysis of normalized rainfall patterns, that anticyclonic activity affected Turkey more frequently over the period 1973–1993. The abrupt decrease in rainfall since the early 1970s has been attributed to the northward shift of the Polar front, resulting from a more easterly extension of the drought-dominated subtropical anticyclone extending from the Azores to the eastern Mediterranean. According to the Institute of Hydrology/ISPRA report (1999), a study by Reynard et al. (1997) concluded, inter alia, that • there is a general tendency for an increase in annual average runoff in northern Europe and a decrease in southern Europe of over 30% in some areas; • the greatest sensitivity to change is in the drier parts of southern and eastern Europe; • before the 2050s there could be a substantial reduction in snowfall that would alter the current temporal distribution of river flows by reducing or eliminating the spring peak and substantially increasing winter flows in central and eastern Europe. In addition to the impacts of meteorological drought, the public perception of desertification has been heightened by water resource shortage arising from anthropologically induced water problems, including: • the huge and continuing rise in demand for water to meet the needs of tourism growth, which has locally caused salinization because aquifers have been drawn down, as in the case of Benidorm, Spain; • a number of major floods, whose magnitude and time-to-rise have almost certainly been affected by vegetation removal and soil erosion, but whose impact has resulted from the failure of planning measures to provide flood plain zoning; • the heavy reliance in Mediterranean countries on irrigation for agricultural production: in Greece, 80% of water is used for irrigation, in Italy 50%, in Spain 68% and in Portugal 52%; • the continued rise in the demand for irrigation water, which has led to a reversion to engineeringtype solutions. An example of the latter is the National Hydrological Plan of Spain, which foresees the transfer of water from the lower Ebro to both Catalonia (Barcelona) and Murcia. There has been a bitter debate by the people of Aragon who claim that the water needs for the poorer areas of Aragon are also exacerbating underdevelopment. There is a crisis of democracy because the central government has had to try to balance out the needs of the wet north and the dry south. In 1998, in the severe drought, the existing transfer canal taking water from the River Tajo in Castilla la Mancha to the River Segura in Murcia failed to stave off the impacts of drought in Murcia, where large numbers of fruit trees were lost. The Tajo–Segura Trasvase (transfer canal) has a capacity for transferring 6 × 108 m3 year−1 and the Spanish government ordered the diversion of a further 5.5 × 107 m3 to “save” Murcia. It is against this background that the current bitter row over transfer from the Ebro to Murcia is being waged. At a demonstration in the Aragon city of Zaragoza, two-thirds of the population of Aragon turned out to protest against the projected transfer, instead of letting the water flow to the Ebro delta and the irrigated lands around Tortosa. Meanwhile Barcelona is negotiating with France for water from the Rhˆone. Water quality deterioration is adding to the environmental crisis that has been confused and compounded with desertification and coupled to the issues of sustainability and the defence of rural areas. Again, the effects of productionist agriculture are evidently the major causes and any action taken to mitigate desertification through regulatory measures in an integrated catchment context will have to address the water quality problem (Foster 2000). The flux of fertilizer returns in water in the northern states is three times greater than in the southern states and contributes 73% of the total. Of the national amounts, the largest returns are of irrigation water (Egypt and Italy) and power station cooling water (France).
10
6
Mediterranean Desertification
A MOSAIC AND A PALIMPSEST
One of the major difficulties facing this planning operation is the fact that the Mediterranean landscape is one of the most complicated in the world. Over space, conditions rarely remain the same for more than a kilometre or two because of local variations in topography, soils, land use, climate and surface water conditions. Another source of variety is that almost every municipality bears the imprint of national, regional and local constraints throughout history. The challenge for those concerned with planning for environmental sustainability in a local Agenda 21, including desertification and land degradation, is threefold: • • •
to identify the local-scale causes of desertification and its manifestations, and develop suitable sensitive indicators to do this; to understand the historical development of the problem, also at different time-scales; to develop regulations that, far from being applicable to the whole of Europe, are sufficiently flexible to accommodate the local variations in history and conditions in the hope that this will facilitate implementation and contribute towards successful outcomes from the interventions.
Given the multiple pressures on national and regional governments from the International Convention, from the European Community and from national and local pressure groups, rural planning has shifted sharply into focus. With it has come the need for empowerment of local people in finding and negotiating optimal strategies to meet these legally binding requirements (Thornes 1998).
REFERENCES Balabanis P, Peter D, Ghazi A and Tsogas M (1999) Mediterranean Desertification. Research Results and Policy Implications, Volumes 1 and 2. Plenary Session Papers, European Commission, Directorate General for Science, Research and Development, EUR 19303, Brussels. Baudry J (1991) Ecological consequences of grazing, extensification and land abandonment. Role of interactions between environment, society and techniques. Options Mediterraneennes, Serie Seminaires 15, 13–19. Bernaldez FG (1991) Ecological consequences of the abandonment of traditional land use systems in central Spain. Options Mediterraneennes, Serie Seminaires 15, 23–29. Birdlife International (1997) A future for Europe’s Rural Environment: Reforming the Common Agricultural Policy. Birdlife International European Community Office, Brussels, p. 55. Brandt CJ and Thornes JB (1996) Mediterranean Desertification and Land Use. John Wiley, Chichester. Burke S and Thornes JB (1998) Volume 1, Actions taken by national governmental and non-governmental organisations to mitigate desertification in the Mediterranean; Volume 2, Thematic review (in press); Volume 3, Summary (in press). Concerted Action on Mediterranean Desertification. European Commission, Directorate General for Science, Research and Development, EUR 18490EN, Brussels, p. 349. De Miguel JM (1989) Estructura de un sistema silvopastoral de dehesa. PhD thesis, Universidad Complutense de Madrid, Facultat de Ciencias (in Spanish). Diamond S and Woodward I (1998) Vegetation modelling. In P Mairota, JB Thornes and N Geeson (eds) Atlas of Mediterranean Environments in Europe: The Desertification Context . John Wiley, Chichester, pp. 68–69. Foster S (2000) Sustainable groundwater exploitation for agriculture: current issues and recent initiatives in the developing world. Papers of the Groundwater Project, Madrid, Marcelin Botin Foundation, Series A, No. 6. Fousakis P and Lekakis J (1998) Adjusting to the changing reality: the Greek response. In T O’Riordan and H Voisey (eds) The Transition to Sustainability: The Politics of Agenda 21 in Europe. Earthscan, London, pp. 214–229. Institute of Hydrology (UK) and ISPRA 1999. Workshop on Drought and Drought Mitigation. Space Applications Institute, Ispra, Varese, Italy, February 1999. Kosmas C, Bakker M, Bergkamp G, Detsis V, Diamantopoulos J, Gerontidis St, Imeson A, Levelt O, Maranthianou M, Oortwijn R, Oustwoud Wijdnes D, Poesen J, Vandevkkerckhove L and Zaphirou Th (1998). Mairota P, Thornes JB and Geeson N (eds) (1998) Atlas of Mediterranean Environments in Europe: The Desertification Context. John Wiley, Chichester. MEDALUS III Meeting, Lesvos, 24–28 April 1998. MEDALUS Lesvos Field Guide. Laboratory of Soils and Agricultural Chemistry, Agricultural University of Athens.
The Evolving Context of Mediterranean Desertification
11
Naveh Z and Whittaker RH (1974) Structural and floristic diversity of shrublands and woodlands in northern Israel and other Mediterranean areas. Vegetation 41, 171–190. O’Riordan T and Voisey H (eds) (1998) The Transition to Sustainability: The Politics of Agenda 21 in Europe. Earthscan, London, pp. 214–229. Pineda FD, Nicolas JP, Ruiz M, Peco B and Bernaldez FG (1981) Succession, diversite at ampliyude de niche dans les paturages du centre de la Peninsula Iberique. Vegetation 47, 267–277 (in Spanish). Rebeiro T and Rodrigues V (1998) The evolution of sustainable development strategies in Portugal. In T O’Riordan and H Voisey (eds). The Transition to Sustainability: The Politics of Agenda 21 in Europe. Earthscan, London, pp. 202–214. Reynard NS, Hulme M, Conway D and Faulkner D (1997) In NW Arnell (ed.) The Impact of Climatic Change on Hydrological Regimes and Water Resources in Europe. Final Report to EC DCXII. Thornes JB (1990) The interaction of erosional and vegetational dynamics in land degradation: spatial outcomes. In JB Thornes (ed.) Vegetation and Erosion. John Wiley, Chichester, pp. 41–55. Thornes JB (1998) Mediterranean desertification and Di Castri’s fifth dimension. Mediterraneo 12/13, 149–166. Turkes M (1996) Meteorological drought in Turkey: an historical perspective 1930–1993. Drought Network News 8(3). UNEP (1990) Desertification revisited: proceedings of an ad hoc consultative meeting on the assessment of desertification. UNEP/DC/PAC, Nairobi, pp. 289–294. Van der Leeuw S (1998) The Archaeomedes Project – Understanding the Natural and Anthropogenic Causes of Land Degradation and Desertification in the Mediterranean Basin. European Commission, Directorate General for Science, Research and Development, EUR 18181EN, Brussels.
Section II
Climate, Processes and Responses
2
Extreme Climatic Events over the Mediterranean
M. CONTE,1 R. SORANI2 AND E. PIERVITALI3 1
Istituto Fisica Atmosfera CNR, Rome, Italy Servizio Meteorologico dell’Aeronautica, Rome, Italy 3 Universita` della Calabria, Rende (CS), Italy 2
1 INTRODUCTION Violent meteorological phenomena, including strong winds, heavy precipitation and intense thermal conditions, may lead to events such as floods and forest fires, with disastrous consequences to land cover and land use. The resulting damage, particularly to agricultural settlements, can lead to abandonment and degradation of once cultivated land. In the Mediterranean region the normal climate includes sparse rainfall and high temperatures, so that extreme meteorological events can have a big impact, destroying the fragile balance between climate, soils and vegetation. A small increase in aridity may be enough to prevent regeneration of vegetation, and cause soil erosion and salinization. In this way extreme climatic events are an agent of desertification, in a wider context. In this chapter attention is directed to some extreme meteorological events, covering large areas but having heavy consequences locally. “Meteorological bombs”, heat waves and precipitation patterns, particularly extreme rainfall episodes, have been studied.
2 THE METEOROLOGICAL BOMB IN THE MEDITERRANEAN 2.1 Introduction and Definition
Several studies have been devoted to meteorological “bombs” in the last few years because of the serious damage attributed to them. Strong winds, intense precipitation and resultant floods are generally associated with these “bombs”. T. Bergeron defined a very rapidly deepening extratropical low as “a depression in which the central sea-level pressure falls at a rate of 1 hPa h−1 or more for a period lasting at least 24 hours”. As Bergeron’s definition referred to the latitude of 60 ◦ N, a geostrophically equivalent rate can be obtained for a latitude ϕ by multiplying this rate by sin ϕ/sin 60◦ . The resulting critical rate, denoted as 1 bergeron by Sanders and Gyakum (1980), varies from 28 hPa 24h−1 at the pole to about 9 hPa 24h−1 at 20 ◦ N, which is the southern limit at which the phenomenon has been observed. In the Mediterranean, applying the geostrophic correction, the critical value of 1 bergeron is obtained with a deepening of 20 hPa 24h−1 at the extreme northern boundary, and of 14 hPa 24h−1 at the deep southern limit of the basin. An average value of 17 hPa 24h−1 is the critical value for an average latitude of 38 ◦ N. Sanders and Gyakum (1980) described this explosive extratropical cyclogenesis as a meteorological “bomb”. General case studies of “meteorological bombs” were examined by Mansfield (1974), Bosart (1981), Anthes Keiser (1979) and Mullen (1983). Specifically in the Mediterranean there have been studies by Bassani (1983), Capaldo et al. (1980) and Karakostas and Flocas (1983). In addition, a synoptic-dynamic climatology of the “bomb” was developed by Sanders and Gyakum Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
16
Mediterranean Desertification
(1980) for part of the northern hemisphere, but not including the Mediterranean. In this chapter a synoptic climatology of this extreme meteorological system in the Mediterranean Basin is presented for the 31-year period 1965–1995. Using various data sources, 101 “bomb” events from this time period have been examined. 2.2
Mechanism of Development of the ‘‘Bomb’’ in the Mediterranean
A synoptic but accurate analysis of the 101 events studied indicated that for the most part (93%) the “bombs” occurred, in broad outlines, following only two fundamental types of meteorological development. In the first type many elements pointed out by Karakostas and Flocas (1983) are recognizable. In broad terms the “bomb” develops from an interaction between a baroclinic, open long wave and an unstable short wave. In addition, the resulting cyclonic vorticity, the upper air temperature advection, and the sensible and latent heat exchange support the rapid and intense deepening of the cyclone. During the analyses carried out for this synoptic climatology over the Mediterranean region, it was observed that, in several situations, the effect of the Alpine barrier can be very important in initiating a small-scale cyclogenesis, known as “cyclogenesis in the lee of the Alps”. This can interact with a wave at a larger scale and degenerate to form a “bomb”. In other words, the function of the short wave of the Karakostas and Flocas mechanism may be enhanced by the Alps or assumed by cyclogenetic factors to be related to the Alps. This type of development of a “bomb” is hereafter denoted as KF (after Karakostas and Flocas 1983). The second type of development follows the dynamics reported in the work of Capaldo et al. (1980). In this case the “bomb” originates from the interaction between a middle-latitude depression at synoptic scale, deeply penetrated into the Mediterranean, and a depression of African origin, sometimes at sub-synoptic scale. Often the interaction can be an effective intrusion of a smallscale African depression into a larger scale low-pressure area drawn from middle latitudes. In this process the low-level jet-stream and the intense baroclinicity have very important roles, related to the strong thermal contrast between the two systems of quite different origins. Other relevant features, especially in the initial stage of the development, are the very intense upper air vorticity due to a marked closeness of a branch of the polar jet-stream to a branch of the subtropical jet-stream and the release of sensible and latent heat. This type of growth of a “bomb” is hereafter denoted as CC (after Capaldo et al. 1980). In both types of development the fact that the Mediterranean sea surface temperatures (SST) are higher than Atlantic sea surface temperatures seems to be of relevant importance. The “bomb” is essentially a meteo-marine phenomenon. 2.3
The Calendar and the Geographical Distribution of the ‘‘Bombs’’ in the Mediterranean for the Period 1965–1995
Annual calendars of the incidence of “bombs” occurring in the Mediterranean region in the period 1965–1995 were compiled, and an example for 1965 is shown in Table 2.1. The date of maximum intensity is reported, as well as a value for that intensity in bergeron, and the geographical location of the centre of the “bomb”. An indication of the mechanism of development (CC, KF or other) is also given. The complete set of calendars (1965–1995) is available from the authors. Figure 2.1 shows the geographical distribution of the “bomb” events between 1965 and 1995 within location quadrilaterals, (2◦ latitude) × (3◦ longitude). Three particular areas of incidence appear: the Corsican–Sardinian Sea and the central and southern Thyrrenian Sea; an area including the central and southern Adriatic Sea and the northern Ionian Sea; and an area including the Aegean Sea. This is in accordance with observations over sea areas by Sanders and Gyakum (1980), who stated that explosive developments of “bomb” type occur over a wide range of sea surface temperatures (SST), but frequently near and a little south of their strongest gradients. Actually the three preferential areas in the Mediterranean are situated a little south of moderate or strong gradients of SST.
17
Extreme Climatic Events over the Mediterranean Table 2.1 Calendar of ‘‘bomb’’ development in the Mediterranean region in 1965
Year
Date of maximum activity
1965
Location of “bomb” centre
21 Jan 4 Feb 9 Feb 20 Apr 12 Nov
1.29 1.25 1.25 1.10 1.25
40N-12E 36N-32E 41N-12E 43N-14E 41N-05E
11 Dec 31 Dec
1.29 1.06
40N-18E 39N-21E
long.N
Intensity (bergeron)
3 1 1
2
3 6
4
3 7
12
7
5 6
5
4
Mode of development
KF KF KF KF Frontal cyclogenesis in a western flux CC CC
40
1
1
1
6
1
7 1
1
9 1
1
1
30 0
lat.W
1
10
2
3
10
20
19
22
15
30
7
7
10
lat.E
1
3
Figure 2.1 Geographical distribution of ‘‘bomb’’ events between 1965 and 1995 within location quadrilaterals (2◦ latitude) × (3◦ longitude) 2.4 Simple Statistical Distribution of the Mediterranean ‘‘Bombs’’
The monthly distribution of all cases of Mediterranean “bomb” is reported in Figure 2.2, showing that the phenomenon is much more frequent in the cold winter season. Only two summer events were observed in 31 years. Most of the winter “bombs” were associated with KF or CC development. No “bomb” developed during summer associated with CC, since in this season the African depressions are very infrequent. Figure 2.3 indicates that the KF mechanism occurs more frequently than does CC development. Only 7 cases out of 101 are due to developments other than KF and CC. The influence of the Alps in initiating the process of development of a “bomb” has also been determined, by the examination of the cyclogenesis. Mediterranean low-pressure areas, which can degenerate into “bombs”, either arrive in the basin from external regions, generally from the Atlantic, or originate in the basin, in particular in association with the orographic effect of the Alps. Figure 2.3 shows that the Alps play a significant role. The cyclogenesis of about 42% of the “bombs” is influenced by the Alps. Our analysis has also shown that the frequency of “bomb” development is higher in months during which the air temperature is lower than normal. This is probably due to the fact that
18
Mediterranean Desertification 25
Number of "bombs"
20
15
10
5
1
2
3
4
5
6
7
8
9
10
11
12
Month
Figure 2.2 Monthly distribution of ‘‘bomb’’ cases identified over the Mediterranean during the period 1965–1995
80 70
Number of "bombs"
60 50 40 30 20 10 0 KF mechanism
CC mechanism
Other mechanism
cases influenced by the Alps cases not influenced by the Alps
Figure 2.3 Distribution of type of development of ‘‘bombs’’ identified over the Mediterranean (1965–1995): either KF (after Karakostas and Flocas 1983), CC (after Conte et al. 1986) or an alternative mechanism
19
Extreme Climatic Events over the Mediterranean
in cold periods the difference between the SST and the air temperature is higher than at other times. A high temperature difference favours a transfer of sensible heat and water vapour from the sea to the atmosphere, thus increasing the energy available for the development of intense cyclogeneses. 2.5 The ‘‘Bomb’’ in the Framework of Large-scale Mediterranean Atmospheric Circulation
The annual number of meteorological “bombs” is shown in Figure 2.4, with a clear negative trend over the whole period, and a shift in the mean since 1982. The difference between the mean of the period 1965–1981 and the mean of the period 1982–1995 is statistically significant at the 95% level (Student’s t-Test). The observed shift should not be attributed to inhomogeneity in the data series, since all data have been recorded by the Meteorological Service of the Italian Air Force, which also performs stringent data quality controls. Colacino and Conte (1993a) investigated the behaviour of the height of the 500 hPa level over the Mediterranean. It is well known that the evolution of this level represents very well the physical situation of the whole atmosphere in a large area, such as the Mediterranean Basin. The analysis concluded that a positive trend is present in the 500 hPa height, over most of the Mediterranean, during the past 45 years. The height of the 500 hPa level appears to have particularly increased during the 1980s. High values of the 500 hPa are related to high frequency and persistence of anticyclones, i.e. high pressure systems that should prevent or limit the cyclogenetic activity. As a consequence, we would expect that the annual frequency of the “bombs”, which are very severe cyclonic systems, should also be reduced by the increased anticyclonic patterns in the Mediterranean. This conclusion appears to be supported by the regression analysis between the annual numbers of “bombs” and the 500 hPa height: these two series are negatively correlated, with a correlation coefficient of −0.7. 8 7
Number of events
6 5 4 3 2 1 0 1960
1965
1970
1975
1980
1985
1990
1995
2000
Year
Figure 2.4 Number of Mediterranean ‘‘bombs’’ per year, showing how the mean annual number of events has changed since 1982
20
Mediterranean Desertification
2.6 The Meteorological ‘‘Bomb’’, Damage and Desertification The intense wind, torrential precipitation and floods associated with the “bombs” cause severe damage, and therefore research into their behaviour is important. One example of a particularly disruptive meteorological “bomb” occurred over the southern Tyrrhenian Sea on 25 October 1973. A deep depression, with intense precipitation, thunderstorms and winds, affected the northern coast of Sicily in particular, where the whole of Palermo Harbour was badly damaged. From an economical point of view the negative impact was evaluated in hundreds of billions of Italian lira (Lauteri et al. 1974). Impacts on a lesser scale frequently affect agricultural settlements. Damage to crops and soil erosion may be enough to ruin farmers, causing them to abandon their land and find alternative employment elsewhere. Land abandonment leading to land degradation and desertification in the relatively dry and hot environment of southern Europe is a widespread problem. 2.7 Conclusions on the Meteorological ‘‘Bomb’’ From the discussion above we can draw the following conclusions:
(i) Meteorological “bombs” are not unusual over the Mediterranean Basin during the winter season. (ii) “Bombs” can be produced by two different dynamic processes, and the process proposed by Karakostas and Flocas (1983) appears to occur more frequently than that proposed by Capaldo et al. (1980). (iii) Statistical analysis indicates that the greatest number of events occurs in the central Mediterranean, while a secondary maximum is found over the Aegean Sea. This distribution suggests a possible role of Alpine orography in triggering these events. (iv) In recent years an increase of the atmospheric pressure over the central and western basin has been recorded. This appears to be associated with a trend in which the annual number of “bombs” has been reduced. (v) Meteorological “bombs” can have serious impacts on agricultural settlements, damaging the terrain and reducing crop production.
3
HEAT WAVES IN THE CENTRAL MEDITERRANEAN BASIN
3.1 Introduction During the warm season (from June to September) over large areas of the Mediterranean Basin, the air temperature sometimes increases up to several degrees above the normal value. These hot spells can either be sudden and very intense, but of short duration (3–5 days), or more gradual and less intense, but of long duration (i.e. 10 days or more). Studies regarding such phenomena over Greece and surrounding regions of the eastern Mediterranean have been made by Karapiperis and Mariopoulos (1956), who defined these thermal events as “heat waves”, and more recently by Metaxas and Repapis (1978) and Metaxas and Kallos (1980). For the central Mediterranean Basin a synoptic study was carried out by Conte (1986), who examined some specific cases. This chapter presents research that analyses all heat waves that occurred over the central Mediterranean during the period 1950–1995. The mechanisms of their development are outlined, essentially from the point of view of synoptic meteorology. A simple statistical presentation of all events has also been carried out. The study was focused on the central Mediterranean area, but, since the patterns leading to the heat waves are of western origin, most of them also influence the Iberian Peninsula, southern France and the coastal areas of north Africa and other Mediterranean countries. 3.2 Definition of the Short- and Long-lasting Heat Waves We define a short-lasting heat wave as a sudden and disruptive increase of air temperature, which, in three separate reference stations located in southern Italy, reached temperatures from 7 ◦ C to 15 ◦ C above the normal monthly mean computed for the period 1951–1980. This event usually has a
Extreme Climatic Events over the Mediterranean
21
Table 2.2 Incidence of short-term heat waves over the central Mediterranean (1950–1995), showing that the highest frequency occurs in July and August
June 19–21/1972 24–27/1982
July 04–07/1952 19–21/1956 11–13/1962 23–27/1962 03–05/1965 20–23/1967 08–11/1968 01–04/1981 30–01 Aug/1982 03–07/1985 23–27/1985 24–28/1987 04–08/1988 26–29/1992 03–06/1993
August
September
12–15/1952 11–13/1960 25–28/1960 03–06/1963 13–15/1963 03–05/1967 07–09/1970 06–08/1971 03–05/1981 20–24/1985 14–17/1989 03–07/1992
07–11/1962 11–14/1970 03–05/1974 03–06/1988 21–25/1995
Table 2.3 Incidence of longer term heat waves over the central Mediterranean (1950–1995)
June 16–28/1950 10–27/1952 09–22/1966 14–26/1970 08–17/1981 02–13/1983 19–30/1990 01–10/1993
July 14–23/1964 16–29/1969 07–16/1974 01–20/1982 13–02 Aug 1983 07–16/1984 11–21/1993 02–15/1994 15–28/1995
August 16–31/1967 01–14/1969 12–23/1971 27 July–12/1980 08–18/1981 27 July–16/1986 13–28/1987 25 July–14/1988 17–03 Sep/1991 25 Jul/08 1994
September 08–17/1951 15–28/1961 14–27/1975 19–27/1983 08–24/1987 08–18/1992
duration of about 3–5 days, and encompasses all of Italy, also reaching Corsica, Malta, the Adriatic side of the former Yugoslavia, Albania, part of Greece and North Africa. A list of all the heat waves of this type that occurred in the period 1950–1995 is shown in Table 2.2. The total number of events was 34, with 131 days influenced by these heat waves, and a mean duration of the phenomenon of about four days. In contrast, the long-lasting heat waves give rise to a gradual air temperature increase, with temperatures that are about 5 ◦ C higher than the normal monthly mean over most of the central basin, and lasting for 10 days or more. Table 2.3 shows that in the study period there were 33 events that influenced, with a mean duration of about 14 days, 462 days of the warm summer season. 3.3 Mechanism of Development of Heat Waves
Short-term Heat Waves The mechanism of development of short-term heat waves is outlined using a simple composite analysis of 12 events lasting four days, which is the mean duration of this kind of phenomenon,
22
Mediterranean Desertification
and a synoptic analysis of the meteorological patterns of an intense event that occurred during 24–27 June 1982. Figure 2.5 shows the pattern of the tropopause/maximum windspeed on 25 June 1982, in which the Subtropical Jet Stream (STJ) appears to be largely north of its normal position, which is in the southern sector of the basin. Following the divergence–vorticity relationship (Palmen and Newton 1969), on the right-hand side of the Jet Streak (the band of the maximum windspeed) a strong upper air convergence gives rise to downward vertical motion, which, in turn, produces warming by adiabatic compression of the atmosphere and, thus, the heat wave. When the STJ returns to its normal position the temperatures return to normal values. Since the latitudinal oscillations of the STJ occur rapidly and are short term, the associated heat waves have the same characteristics.
(a)
275
275
PJ 250
80 0 10
250
225 60
STJ
225 200
175 200 (b)
0
0
0
0 D
0.3
0.2 0
0.1
0.1
0
Figure 2.5 (a) Heat wave development. Tropopause/maximum wind pattern on 00 UTC of 25 June 1982. STJ, Subtropical Jet Stream; PJ, Polar Jet Stream; heights of tropopause (- - - - ) in hPa; isotachs in knots ( ). (b) Vertical motion (Pa s−1 ) on 06 UTC of 25 June 1982. Data from the European Centre for Medium Range Forecasting (ECMWF). Reproduced by permission from Societe` Italiane di Fisice
23
Extreme Climatic Events over the Mediterranean
Longer term Heat Waves The mechanism of development of the long-lasting heat wave is briefly analysed, taking as an example a case that occurred in July 1983. In this type of event the atmospheric circulation in the Euro-Atlantic region, outlined using the pattern at the 500 hPa level, is always characterized by a socalled “omega” pattern (due to its resemblance to the last letter of the Greek alphabet) moving very slowly from west to east (Figure 2.6). The southerly winds blowing between the western depression of the “omega” and the central anticyclonic ridge progressively invade the western and central parts of the Mediterranean. With this southerly flux there is usually associated a horizontal advection of very warm air masses moving from north Africa, which invade the basin, producing the increase in air temperature.
Comparison of Short-term and Longer Term Heat Waves All cases in Tables 2.2 and 2.3 were identified using both the temperature increases recorded in the three reference stations (as well as information for the whole central Mediterranean) and synoptic analyses similar to those outlined above for the two typical cases (Conte 1986). The clear similarities between meteorological patterns associated with the different situations permitted us to group events into short-term and longer term cases. In summary, the main difference between the two heat wave mechanisms is as follows: the shortterm heat waves are essentially determined by downward motion, although the adiabatic compression is exerted on a dome of warm air of African type. The longer term events are produced essentially by horizontal motions, especially if temporary and brief incursions of the STJ can reinforce the phenomenon by adiabatic compression. 3.4 Some Statistical Considerations on Mediterranean Heat Waves
During the study period (1950–1995) the number of summer-time days influenced by a short or longer term heat wave was 586. Since the total number of days in June, July, August and September (summer season) in the whole 46 years is 5612, about 10% of the summer period in the central Mediterranean was influenced by warming due to heat waves. This is an appreciable percentage and indicates that the phenomenon is not infrequent or exceptional and it should probably be considered as a feature of the Mediterranean summer. Table 2.4 summarizes the monthly distribution of the events, the number of days influenced by heat waves and their relative percentages. Two-thirds of the total number of events occur in July and August. The number of days affected by heat waves shows a difference between July and August. Short-term events tend to be more frequent in July, with more long-term events in August. The number of heat wave days in each year of the study period is reported in Table 2.5. The linear trend gives an increase, with a value of 0.4 days year−1 and the polynomial smoothing of seventh order indicates a 20-year oscillatory pattern. Interesting patterns do occur, and recently Colacino and Conte (1993a) detected in the pressure field over the central and western Mediterranean an oscillation having the Hale period (i.e. about 22 years), together with a clear increasing trend. This Table 2.4 Summary of heat wave events (1950–1995)
Month
No. of short-term events
No. of longer term events
Total number of events
% per summer month
No. of days affected by events
% per total summer days affected
June July August September
2 15 12 5
8 9 10 6
10 24 22 11
15 36 33 16
109 183 198 96
19 31 34 16
Total
34
33
67
100
586
100
58
15 JULY 46
L
58
L
82 H
82
52
64
64 70
24
8 JULY L
64 70
70 76
70
82
64
76
H
L 58
82
76 70
L
L
82 76 88
L
88
H
H
94 22 JULY
94 29 JULY
L 52
64 H
52
52
L
70
64
58
64
58 70
64
82 76
88
94 H
82
H 94
H
L
82
70
76 L
76
L
76 88
88 88
H 94
Figure 2.6 Development of the ‘‘omega pattern’’ at the 500 hPa level, producing the long-term heat wave of July 1983. Reproduced by permission from Societe` Italiane di Fisice
25
Extreme Climatic Events over the Mediterranean Table 2.5 Number of heat wave days during the years of the study over the study area
Year
No. of heat wave days
Year
No. of heat wave days
Year
No. of heat wave days
Year
No. of heat wave days
Year
No. of heat wave days
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
13 10 26 0 0 0 3 0 0 0
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969
7 14 13 7 10 3 14 23 4 28
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
20 15 3 0 13 14 0 0 0 0
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
17 28 27 42 10 15 21 38 30 4
1990 1991 1992 1993 1994 1995
12 18 20 25 29 19
pattern of the pressure field was associated with an oscillatory, but progressive increase over the Mediterranean of the persistence of anticyclonic systems, particularly of the Azores anticyclone, over the last fifty years. Heat waves are strongly connected with anticyclonic patterns. A comparison of the number of days influenced by heat waves and the 500 hPa heights over Cagliari, Sardinia, which is in the core of the central basin, shows a general agreement between the two data sets, with low values during the 1950s and the 1970s and higher values in the 1960s and the 1980s. The correlation coefficient for the two data sets is 0.64, which is statistically significant at the 95% level. The heat waves produce very warm and dry environmental conditions, and prolonged drought can lead to desertification (Palutikof et al. 1996). In addition, dryness and high temperature can exacerbate ideal conditions for forest fires to rage out of control over large areas. Fires destroy crops, both forestry and agricultural, and can also destroy the ecological balance within the vegetation and fauna. Colacino and Conte (1993b) examined the pattern of forest fires in the Mediterranean region in connection with the number of heat waves. The number of heat waves recorded in the period 1980–1985 was about 70% higher than in the period 1970–1975, and a similar increase was recorded in the extension of forest burned in the regions of the Mediterranean Basin, for which data are available. Unfortunately more and more fires are started deliberately rather than naturally, but it is evident that the aridity of the soil and vegetation, and the warming associated with heat waves, play an important role in maintaining and extending the fires. Finally, it must be remembered that the heat waves can influence the health of the population, and mortality is often enhanced during these events. A study carried out for the heat wave of 13 July–2 August 1983 indicated that during and immediately after the warmest days, the number of deaths in Rome was 450 more than the normal average seasonal value (Todisco 1987). 3.5 Conclusions on Heat Waves
The following conclusions have been drawn: (i) (ii) (iii) (iv) (v)
There exist two different types of heat wave: the first very intense and of short duration, the second less intense but of longer duration. These two heat wave types are associated with different meteorological patterns. In the study period (1950–1995) the total number of heat waves was 67: 34 short-term events and 33 long-term episodes. Heat waves occur in summer, most frequently in August. Analysis of heat wave events during the study period suggested a 20-year oscillatory pattern, with a superimposed trend of increasing incidence.
26
Mediterranean Desertification
(vi)
(vii)
4
4.1
This behaviour is similar to that found for the same period for atmospheric pressure in the western and central Mediterranean and a connection with the anticyclones in the basin is probable. The impacts of heat waves are important because they may contribute to drought, desertification and forest fires, and may negatively influence the health of the population.
SPACE – TIME PRECIPITATION PATTERNS IN THE WESTERN AND CENTRAL MEDITERRANEAN BASIN AND ANALYSIS OF EXTREME CASES Introduction
Many examples of climatological research are at present focused on climate evolution in association with anthropogenic enhancement of the greenhouse effect. Studies concerning the trend of increased air temperatures and the impact of this increase include those of Jones et al. (1986) and Hansen and Lebedeff (1987). Particular attention is devoted to the hydrological cycle and to the precipitation regime because a reduction in precipitation limits water resources, while increased events of intense rain could cause more frequent floods. Papers published on this subject at a global scale (Bradley and Groismann 1989; Diaz et al. 1989; Vinnikov et al. 1990) indicate that an increase in precipitation has been recorded at latitudes higher than 50 ◦ N, while at lower latitudes an opposite pattern is found. However, the regional analyses do not confirm this general picture, and give contradictory results (Kutiel 1991; Ben-Gai et al. 1994; Beniston et al. 1994; Groismann and Easterling 1994; Lettenmaier et al. 1994; Norsallah and Balling 1996). Several studies have been carried out in the Mediterranean Basin, but some do not refer to the recent period (Maheras 1988) and only a few papers give quantitative results (Palutikof et al. 1994). Here analysis of the trends of yearly and seasonal precipitation in the central and western Mediterranean is given for the period 1951–1995. The study area is subdivided into three latitudinal belts: northern (>42 ◦ N), central (38 ◦ N–42 ◦ N) and southern (<38 ◦ N). For every belt, a regional time series using the Thiessen method (Ven Te Chow 1964) has been obtained and the Standardized Anomaly Index (Nicholson 1983) has been calculated. In every belt a significant decrease in the precipitation amount has been found. At the seasonal scale the strongest reduction has been shown in winter. A detailed analysis of the extreme events was carried out for the core area of the typical Mediterranean climate, using monthly values. The results showed a decreasing trend, similar to the previous analysis. 4.2
Data Set
Precipitation patterns in the central and western Mediterranean Basin, for the period 1951–1995, have been analysed using data from the stations reported in Figure 2.7. There are annual data (September to August) for all stations and additional seasonal and monthly rainfall totals for 14 stations. Italian data have been provided from the Meteorological Service of the Italian Air Force, data for Spain, Portugal, Algeria, Tunisia from the respective National Services, and data for the other countries from the World Climate Disc (1992). Quality controls have been performed by the data-gathering institutions. 4.3
Annual and Seasonal Analysis of Data
In order to evaluate precipitation trends over the Mediterranean Basin, a regional series using Thiessen’s method was constructed. The method is suitable for calculating a weighting for each station with respect to the location of the different observatories, since they are not uniformly spaced. For this data series a decreasing trend for the annual total precipitation has been observed (Figure 2.8). For the study period 1951–1995, the reduction in annual precipitation was about 142 mm (about 21%), but this is a relatively short time span, and the picture could be different over a longer time period.
Extreme Climatic Events over the Mediterranean
27
Northern belt
Central belt
Southern belt
Figure 2.7 Stations used for annual (•) or annual plus seasonal () precipitation data analysis across three latitudinal belts. The most southerly station has not been included in annual analyses. Reproduced by permission from Societe` Italiane di Fisice
850
Precipitation (mm)
800 750 700 650 600 550 500 450 1940
1950
1960
1970
1980
1990
2000
Year
Figure 2.8 Regional precipitation series of the central and western Mediterranean Basin, obtained by the Thiessen method. The trend of decreasing precipitation since 1951 is very clear. Reproduced by permission from Societe` Italiane di Fisice
In addition to this analysis, in order to obtain a regional normalized series, the Standardized Anomaly Index (SAI) has been computed. This index can be evaluated since the distribution of the yearly values is nearly Gaussian. The trend of the SAI for the three latitudinal belts in Figure 2.7 indicates that in all the cases the precipitation decreased significantly (statistically significant). The index ranges from values higher than 0.25 to values lower than −0.25 (Nicholson 1983). A possible interpretation is associated with the increase, recorded for the same period over the central and western basin, of the atmospheric pressure both at the surface and at the upper levels. This trend is known as the Mediterranean Oscillation and has been analysed by Conte et al. (1989), Colacino and Conte (1993a) and Douguedroit (1994).
28
Mediterranean Desertification 2.0 Autumn 1.5 1.0
SAI
0.5 0.0 −0.5 −1.0 −1.5 −2.0 1940
1950
1960
1970
1980
1990
2000
2.0 Winter 1.5 1.0
SAI
0.5 0.0 −0.5 −1.0 −1.5 −2.0 1940
1950
1960
1970 Year
1980
1990
2000
Figure 2.9 SAI trend for precipitation in autumn and winter, over the core area of the Mediterranean Basin since 1951. In winter the trend line exceeds the probability limits and the reduction in precipitation is statistically significant. Reproduced by permission from Societe` Italiane di Fisice
Precipitation analysis was then extended to look at seasonal behaviour in the core area of the Mediterranean climate (for stations marked by in Figure 2.7). The regional rainfall patterns over this part of the basin show trends of decreasing precipitation, but the reduction is not the same across all seasons, and tends to be greater in the cold season, in winter. In autumn the percentage reduction was 12.4%, in winter 29%, in spring 8.2% and in summer 6.7%. Figure 2.9 shows the trend for the SAI in autumn and winter. Only in winter was the reduction statistically significant. 4.4
‘‘Extreme’’ Monthly Events
Extreme precipitation events have also been investigated, using monthly data available for 14 stations in the core area of the typical Mediterranean climate. Mean monthly rainfall amounts for the 1951–1995 period have been computed and the most rainy month has been listed for each station. We have defined “extreme” as all the cases in which the monthly total precipitation in one station was higher than, or equal to, two times the mean precipitation value of the most rainy month,
Extreme Climatic Events over the Mediterranean
29
12
Number of events
10
8
6
4
2
0 1940
1950
1960
1970
1980
1990
2000
Years
Figure 2.10 The number of ‘‘extreme’’ precipitation events over the central Mediterranean Basin between 1951 and 1995. There is a trend of decreasing precipitation
Pi ≥ 2 Pim. Figure 2.10 shows the number of extreme episodes over the whole central Mediterranean Basin in the study period. The highest numbers were in 1959 (9 cases), in 1961 (7 cases), in 1966 (11 cases), in 1976 (8 cases), and in 1990 (7 cases). The linear trend shows a decreasing pattern, with a value of −0.06 cases year−1 (i.e. a reduction of 56.6% for the total 45 considered). 4.5 Conclusions on Extreme Events
The following conclusions have been drawn: (i)
The volume of annual precipitation received over the central and western Mediterranean Basin is decreasing. (ii) The SAI (Standardized Anomaly Index) trends for three latitudinal belts all show a statistically significant reduction in precipitation for the study period, 1951–1995. (iii) The seasonal analysis indicates that the rainfall decrease is mainly felt in winter. (iv) A decrease in the number of “extreme” precipitation events since 1951 has also been observed. These results indicate that there has been a decrease in the volume of precipitation over the central Mediterranean between 1951 and 1995. It is possible that this trend could be associated with human interventions in the environment, and the greenhouse effect (Frederick and Major 1997). However, for a true picture these trends would have to be analysed in the context of a much longer time-scale, as they may be merely part of a natural climatic variability. Unfortunately the long-term historical data for such an analysis is not available.
REFERENCES Anthes RA and Keiser D (1979) Test of a fine mesh model over Europe and United States. Monthly Weather Review 107, 963–984.
30
Mediterranean Desertification
Bassani M (1983) Brevi considerazioni su un ciclone a mesoscala nell’area mediterranea. Rivista di Meteorologia Aeronautica XLIII, 297–307. Ben-Gai T, Bitan A, Manes A and Alpert P (1994) Long term changes in annual rainfall patterns in Southern Israel. Theoretical and Applied Climatology 49, 59–67. Beniston M, Rebetez M, Giorgi F and Marinucci MR (1994) An analysis of regional climate change in Switzerland. Theoretical and Applied Climatology 49, 135–159. Bosart LF (1981) The President’s day snowstorms of 1819 February 1979. A subsynoptic scale event. Monthly Weather Review 109, 1004–1018. Bradley RS and Groismann PYa (1989) Continental scale precipitation variations in the 20th century. In Proceedings of the International Conference on “Precipitation Measurements”, WMO, Geneva, pp. 168–184. Capaldo M, Conte M, Finizio C and Todisco G (1980) A detailed analysis of a severe storm in the central Mediterranean: the case of the Trapani flood. Rivista di Meteorologia Aeronautica XL, 183–199. Colacino M and Conte M (1993a) Greenhouse effect and pressure patterns in the Mediterranean Basin. Nuovo Cimento C 16, 67–76. Colacino M and Conte M (1993b) Clima mediterraneo ed incendi boschivi, Consiglio Nazionale delle Ricerche, Istituto di Fisica dell’Atmosfera, Report 93-50. Conte M (1986) Short-lasting heat waves in the central Mediterranean. WMO Long-Range Forecasting Research Report, WMO/TD no. 87, pp. 146–152. Conte M, Giuffrida A and Tedesco S (1989) The Mediterranean Oscillation: impact on precipitation and hydrology in Italy. Conference on Climate and Water 1, Publications of the Academy of Finland 9/89, pp. 121–137. Diaz HF, Bradley RS and Eischeid JK (1989) Precipitation fluctuations over global land areas since the late 1800s. Journal of Geophysical Research 94, 1195–1210. Douguedroit A (1994) Regionalization of the precipitation in the Mediterranean Basin: Mediterranean Oscillation? Meeting on Atmospheric Physics and Dynamics in the Analysis and Prognosis of Precipitation Fields, Rome, 15–18 November. Frederick KD and Major DC (1997) Climate change and water resources. Climatic Change 37, 7–23. Groisman PY and Easterling DR (1994) Variability and trends of total precipitation and snowfall over the United States and Canada. Journal of Climatology 7, 184–205. Hansen J and Lebedeff S (1987) Global trends of measured surface air temperature. Journal of Geophysical Research 29(D11), 13 345–13 372. Jones PD, Wigley TML and Wright PB (1986) Global temperature variations between 1861 and 1984. Nature 322, 430–434. Karakostas TS and Flocas AA (1983) The development of the ‘bomb’ over the Mediterranean area. Proceedings of “Blue Water Green Water” meeting, Marseille. Karapiperis L and Mariolopoulos E (1956) On the annual march of the air temperatures in Athens and their anomalies. Publ Met, University of Athens. Kutiel H (1991) Recent spatial and temporal variations in mean sea level pressure over Europe and the Middle East and their influence on the rainfall regime in the Galilee, Israel. Theoretical and Applied Climatology 44, 151–166. Lauteri P, Conte M and Mancino L (1974) Some synoptic aspects of an advective storm over the southern Thyrrenian sea, 25 October 1973. Proceedings of the Regional Training Seminar “Meteorological Services to Marine and Coastal Activities”, WMO, Rome, 1–12 April 1974. Instituto di Fisica del’ Atmosfera–Consiglio Nazionale delle Ricerche, Rome, pp. 372–382. Lettenmaier DP, Wood EF and Wallis JR (1994) Hydro-climatological trends in the continental United States, 1948–88. Journal of Climatology 7, 586–607. Maheras P (1988) Changes in precipitation conditions in the Western Mediterranean over the last century. Journal of Climatology 8, 179–189. Mansfield DA (1974) Polar lows: the development of baroclinic disturbances in cold air outbreaks. Quarterly Journal of the Royal Meteorological Society 100, 312–328. Metaxas DA and Kallos G (1980) Heat waves from a synoptic point of view. Rivista di Meteorologia Aeronautica XL, 107–119. Metaxas DA and Repapis CC (1978) Large warm advection over Athens: a climatological and synoptic study. Archives f¨ur Meteorologie, Geophysik und Bioklimatologie B 26, 51–61. Mullen SL (1983) Explosive cyclogenesis associated with cyclones in polar air streams. Monthly Weather Review 111, 1537–1548. Nicholson SE (1983) Subsaharan rainfall and the years 1976–80: evidence of continued drought. Monthly Weather Review 111, 1646–1654.
Extreme Climatic Events over the Mediterranean
31
Norsallah HA and Balling RC (1996) Analysis of recent climatic changes in the Arabian Peninsula Region. Theoretical and Applied Climatology 53, 245–252. Palmen E and Newton CW (1969) Atmospheric Circulation Systems. Academic Press, New York. Palutikof JP, Goodess CM and Guo X (1994) Seasonal scenarios of the change in potential evapotranspiration due to the enhanced greenhouse effect in the Mediterranean Basin. International Journal of Climatology 14, 853–869. Palutikof J, Conte M, Casimiro Mendes J, Goodess CM and Espirito Santo F (1996) Climate and climatic change. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 43–86. Sanders F and Gyakum JR (1980) Synoptic dynamic climatology of the ‘bomb’. Monthly Weather Review 108, 1589–1606. Todisco G (1987) Indagine biometeorologica sui colpi di calore verificatisi a Roma nell’estate del 1983. Rivista di Meteorologia Aeronautica XLII, 189–197. Ven Te Chow PhD (1964) Handbook of Applied Hydrology (a compendium of water resources technology). McGraw Hill, New York. Vinnikov KYa, Groismann PYa and Lugina KM (1990) Empirical data of contemporary global climate changes (temperature and precipitation). Journal of Climatology 3, 662–677. World Climate Disc (1992) University of East Anglia. Chadwyck & Healey (global climatic change data on CD-ROM).
3
Potential Effects of Rising CO2 and Climatic Change on Mediterranean Vegetation
C.P. OSBORNE AND F.I. WOODWARD
Department of Animal and Plant Sciences, University of Sheffield, UK
1 INTRODUCTION The Mediterranean climate is characterized by seasonal variations in temperature and precipitation that result in a cool, wet winter and a hot, dry summer. In common with other dry climates, precipitation events are highly irregular and total precipitation varies considerably between years, resulting in a high frequency of drought (Ehleringer and Mooney 1983). Vegetation throughout the Mediterranean Basin is therefore adapted to periodic drought, being dominated by evergreen broadleaved species, which are typically deep-rooted and characterized by sclerophyllous leaves with adaptations for minimizing water loss and damage due to high temperatures (Archibold 1995). Water availability strongly limits net primary productivity in sclerophyllous vegetation, for example, net photosynthesis in Quercus suber declined by around 75% at a site in Portugal as a consequence of declining soil water potential (Tenhunen et al. 1987). The structure of Mediterranean sclerophyllous vegetation also varies significantly within the region, from sparse shrubland with a low stature, to forest which can reach 15 m in height, partly as a consequence of geographical variation in water availability (e.g. Lossaint 1973; Merino et al. 1990). Sclerophyllous shrubs and trees are gradually replaced by drought-deciduous shrubs and tussock grass steppe towards the warmer, drier parts of the region, where the summer drought period increases in duration and intensity, and by winter-deciduous trees in cooler, wetter areas, particularly at high altitudes (Tomaselli 1981; Woodward 1987). Water availability is therefore important in controlling both the productivity and structure of Mediterranean vegetation. Atmospheric CO2 is predicted to rise from its current concentration of 350 ppm to between 450 and 550 ppm by the middle of the 21st century (Schimel et al. 1996), and this increase is likely to have significant impacts on the structure and functioning of global vegetation (Betts et al. 1997). Climatic changes in the Mediterranean region are expected to accompany rising CO2 . Briefly, mean annual temperature is likely to rise by around 1.2 ◦ C by 2050 (Mitchell et al. 1995), and total precipitation is likely to decline, although the magnitude of this decrease is uncertain (Palutikof et al. 1994). However, increasing evaporation resulting from the rise in temperature is likely to cause a decline in soil water availability, irrespective of changes in precipitation (Palutikof et al. 1994). Understanding relationships between soil water availability and the effects of rising CO2 concentration will therefore be critical for predicting the future impact of climatic change on the productivity and structure of Mediterranean vegetation. This chapter explores some potential interactive effects of varying soil water and rising CO2 on Mediterranean sclerophyllous vegetation. First, results are reviewed from experiments that have investigated the effects of elevated CO2 on natural stands of sclerophyllous vegetation. Second, modelling approaches are used to examine interactions between rising CO2 and variation in soil water availability at a range of spatial and temporal scales. Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
34
2
Mediterranean Desertification
RISING CO2 AND WATER RELATIONS IN MEDITERRANEAN SCLEROPHYLLOUS SHRUBS
Experiments testing the effects of elevated CO2 on natural stands of Mediterranean sclerophyllous vegetation have been mainly conducted at sites in central Italy, and have used CO2 springs and open-top chambers to provide high CO2 environments during growth. CO2 springs are the vents for deep geothermal CO2 sources, and vegetation surrounding them has been exposed to elevated CO2 concentrations probably for many hundreds of years (Miglietta et al. 1993). Open-top chambers consist of transparent plastic cylinders that surround clumps of natural vegetation in situ, and atmospheric CO2 concentration is controlled using a vertical flow of air through the chamber. Results from both sets of studies suggest that the responses of Mediterranean vegetation to high CO2 may be similar to those of woody plants in general, and indicate interactions between CO2 concentration and plant water relations. Elevated CO2 concentration can indirectly affect plant water relations by causing stomatal closure, but decreased stomatal conductance does not necessarily result in reduced transpiration in stands of vegetation. Canopy boundary layer conductance may be low relative to stomatal conductance, resulting in feedbacks which tend to stabilize transpiration, and increased canopy leaf area may compensate for lower rates of water loss per unit area (Field et al. 1995). Reduced transpiration at elevated CO2 does result in slower decreases in soil water during drought periods in some experiments (e.g. Field et al. 1995). However, the effect of high CO2 on soil water availability for Mediterranean sclerophyllous vegetation varies between studies. Soil water availability can be assessed using measurement of pre-dawn plant water potential, which provides an indirect estimate of soil water potential, provided that the plant has reached equilibrium with the soil overnight (Richter 1997). High CO2 generally leads to decreases in leaf stomatal conductance, but responses are highly variable in woody species, varying between season and species in Mediterranean sclerophyllous shrubs (Morison 1985; Curtis and Wang 1998). Elevated CO2 led to a decline in stomatal conductance compared with controls during the spring in Arbutus unedo, but had no effect during the summer drought period, when conductance was low irrespective of CO2 concentration (Jones et al. 1995). However, the pre-dawn water potential of shoots was higher during the summer in elevated CO2 than controls, suggesting higher soil water potential (Jones et al. 1995). Stomatal conductance also decreased at elevated CO2 in Pistacea lentiscus during the spring, but was unchanged in Quercus ilex and Phyllirea angustifolia at the same site (Scarascia-Mugnozza et al. 1996a). The pre-dawn water potential in Q. ilex shoots was lower at elevated CO2 during the summer drought in this mixed stand of sclerophyllous shrubs (Scarascia-Mugnozza et al. 1996a). In contrast, there was no effect of elevated CO2 on pre-dawn water potential or stomatal conductance in Q. ilex a month earlier at a nearby site, and decreases in transpiration were inferred from sap flow measurements (Tognetti et al. 1996). Effects of CO2 concentration on soil water potential under stands of Mediterranean sclerophyllous vegetation therefore vary significantly between sites. Some of this variation can be attributed to differences between species in the sensitivity of stomatal conductance to CO2 , but other factors also seem to be involved. Photosynthesis in the growth environment of plants with the C3 photosynthetic pathway is usually higher in elevated CO2 than in ambient CO2 controls (Drake et al. 1997). Results from a range of Mediterranean sclerophyllous species fit this general pattern, showing a stimulation of leaf photosynthesis at elevated CO2 , irrespective of water availability, and with no evidence of acclimation of the photosynthetic system (Jones et al. 1995; Scarascia-Mugnozza et al. 1996b). A combination of increased photosynthesis without change or with a decrease in stomatal conductance results in higher instantaneous water-use efficiency (CO2 -uptake per unit water loss) at elevated CO2 , and can lead to higher growth rates in Mediterranean sclerophyllous shrubs. Branch volume in Q. ilex shrubs was significantly higher than in controls after three years’ growth in open-top chambers at elevated CO2 , but was lower than controls in Q. ilex trees growing in a CO2 spring (Scarascia-Mugnozza et al. 1996a; H¨attenschwiler et al. 1997a). However, trunk volume in the latter trees was greater than in controls (H¨attenschwiler et al. 1997a). The relative effect of elevated CO2 on trunk size was greatest
Effects of Rising CO2 and Climatic Change on Vegetation
35
in younger trees, and in years with a dry spring period, suggesting that increased water-use efficiency has an important effect on the growth of Q. ilex at high CO2 (H¨attenschwiler et al. 1997b). Results from the experimental studies reviewed here provide a basis for suggesting potential responses of vegetation in central Italy to future increases in CO2 concentration, and highlight mechanisms that could be involved in interactions between elevated CO2 and soil water availability. However, extrapolation of these findings to the whole Mediterranean region, and over longer time-scales, requires the use of dynamic simulation models. The remainder of this chapter uses a mechanistic simulation model of Mediterranean sclerophyllous vegetation to examine potential interactions between rising CO2 concentration and variation in water availability, at a range of spatial and temporal scales. Mechanistic models integrate the current understanding of vegetation responses to the environment, and allow the predictive power of this understanding to be assessed by comparison of model results with real observations. Given a satisfactory explanation of vegetation responses in terms of underlying mechanisms, models then provide a method for extrapolation to the future, and for examining the potential effects of rising CO2 and climatic change scenarios.
3 REGIONAL VARIATION IN RESPONSES OF MEDITERRANEAN SHRUBS TO RISING CO2 Experimental studies in Italy suggest interactions between high CO2 concentrations and the water relations of sclerophyllous shrubs. Stomatal conductance declines at elevated CO2 in some species, with important potential effects on transpiration and therefore soil water availability (see section 2 above). However, these effects on transpiration may be mediated at the canopy level by changes in leaf area which result from increased productivity and water-use efficiency at high CO2 . In addition, results from experimental studies suggest that stomatal conductance may be insensitive to CO2 in other species (section 2). This section explores potential interactions between elevated CO2 , stomatal conductance, leaf area, transpiration and soil water potential, and their implications for productivity and biomass, using the MEDRUSH vegetation model, which is a dynamic, mechanistic simulation model. The MEDRUSH vegetation model was developed for simulating primary production and water use in Mediterranean sclerophyllous shrub vegetation, using only standard meteorological data and CO2 concentration as inputs (see MEDRUSH in Chapter 18). The model successfully predicted 57% of biomass observations (n = 14), 80% of net primary productivity (NPP) observations (n = 5) and 77% of leaf area index (LAI) observations (n = 13) within an estimate of their 95% confidence interval, for sites throughout the Mediterranean region. The model was used to test the hypothesis that decreases in stomatal conductance at elevated CO2 will lead to a decline in transpiration and an increase in soil water potential in stands of Mediterranean sclerophyllous vegetation. The hypothesis was tested at three sites in the western Mediterranean Basin where model predictions of biomass, NPP and LAI were within the confidence interval for observations at each site. Water availability varied significantly between these sites because of regional variation in precipitation and potential evapotranspiration. The effects of an elevated CO2 concentration of 500 ppm were tested by running the model to equilibrium, which was defined as being when total biomass varied by less than 1% in consecutive years. Stomatal conductance (gs ) either declined by 18% at high CO2 (+CO2 , −gs ), or was insensitive to CO2 (+CO2 , = −gs ). Climate input data for the model consisted of a single year of long-term monthly means for Murcia (south-east Spain), Seville (south-west Spain) and Perpignan (southwest France), which were used repeatedly until equilibrium (M¨uller 1982). Annual precipitation and mean annual temperature were 304 mm and 18.0 ◦ C for Murcia, 535 mm and 18.8 ◦ C for Seville, and 639 mm and 12.5 ◦ C for Perpignan, and the sites therefore differed significantly with respect to climate, particularly precipitation. Biomass was always higher at elevated CO2 than in the control, irrespective of the site, or sensitivity of stomata to CO2 (Figure 3.1). When stomatal conductance declined at elevated CO2 , the increase in biomass was due to a rise in NPP, which varied from 32% for Murcia to 17% for
2
200
1
0
100
M
S Site
P
M
S Site
P
NPP (g m −2 year −1)
Mediterranean Desertification Biomass (kg m −2 )
36
0
LAI (m2 m−2)
300 200
1
100 0
M
S Site
P
Control + CO2, − gs + CO2, = gs
M
S Site
P
Transpiration (mm year −1)
2
0
M = Murcia S = Seville P = Perpignan
Figure 3.1 Potential effects of elevated CO2 on sclerophyllous vegetation at three sites in the western Mediterranean Basin: Murcia (south-east Spain), Seville (south-west Spain) and Perpignan (south-west France). Equilibrium model simulations of mean annual above-ground biomass, net primary productivity (NPP), mean annual leaf area index (LAI) and transpiration are shown for each. Model simulations tested the effects of an elevated CO2 concentration of 500 ppm with either an 18% decrease (+CO2 , −gs ) or no change (+CO2 , = gs ) in stomatal conductance compared with the control. CO2 in the control simulations remained at the current ambient concentration of 350 ppm
Seville, and 21% for Perpignan (Figure 3.1). The rise in NPP at elevated CO2 was accompanied by a decline in annual transpiration for Murcia (12%) and Seville (15%), but an increase in annual transpiration for Perpignan (8%) (Figure 3.1). Results therefore illustrate that a reduction in stomatal conductance does not necessarily result in decreased transpiration. In the case of Perpignan, this was because LAI was higher throughout the year at elevated CO2 than in the control (Figures 3.1 and 3.2). Model results also show that the decrease in total transpiration that can accompany a decline in stomatal conductance at elevated CO2 does not necessarily result in higher soil water potential. Reduced annual transpiration at elevated CO2 was associated with a higher soil water potential for Seville, but lower soil water potential for Murcia during the spring and summer (Figure 3.2). This discrepancy could be explained by seasonal patterns of transpiration. Transpiration followed similar seasonal patterns in elevated CO2 and the control for Seville, although it was lower throughout the year at elevated CO2 because of reduced stomatal conductance (Figure 3.2). However, transpiration was higher at elevated CO2 than in controls during the spring in Murcia, and soil water potential therefore declined more rapidly (Figure 3.2). This increase in transpiration occurred despite the decrease in stomatal conductance, and could be attributed to a higher LAI (Figure 3.2). During the summer, LAI declined markedly at elevated CO2 , and reached a value that was lower than the control (Figure 3.2). This change was accompanied by a decline in transpiration and a faster recovery of soil water potential from the summer drought at elevated CO2 than the control (Figure 3.2). Results therefore highlight the importance of considering seasonal variation in the responses of Mediterranean sclerophyllous vegetation to high CO2 .
37
Murcia
Seville
Perpignan
2 1
LAI (m2 m −2 )
0 2
Water potential (MPa)
Transpiration (mm day −1)
Effects of Rising CO2 and Climatic Change on Vegetation
0
1 0
−2 −4 Jan Apr Jul
Oct
Month Control;
Jan Apr Jul Month +CO2,− gs;
Oct
Jan Apr Jul
Oct
Month +CO2, = gs
Figure 3.2 Seasonal variation in the effects of elevated CO2 on water relations of sclerophyllous vegetation at three sites in the western Mediterranean Basin: Murcia (south-east Spain), Seville (south-west Spain) and Perpignan (south-west France). Equilibrium model simulations of transpiration rate, leaf area index (LAI) and soil water potential are shown for each. Model simulations tested the effects of an elevated CO2 concentration of 500 ppm with either an 18% decrease (+CO2 , −gs ) or no change (+CO2 , = gs ) in stomatal conductance compared with the control. CO2 in the control simulations remained at the current ambient concentration of 350 ppm
Annual transpiration for Murcia and Seville also decreased at elevated CO2 when stomatal conductance was insensitive to CO2 concentration (Figure 3.1). In both cases this was because LAI was lower at elevated CO2 than in the control, especially during the summer drought period (Figure 3.2). This resulted in soil water potential being greater during the summer at elevated CO2 than in controls, despite the insensitivity of stomatal conductance to CO2 concentration (Figure 3.2). A different pattern occurred for the Perpignan site, where high CO2 concentrations led to an increase in transpiration when stomatal conductance was unchanged (Figure 3.1). This resulted in a more rapid decline in soil water potential at elevated CO2 (Figure 3.2). Annual NPP was unchanged or higher at elevated CO2 for all sites when stomatal conductance was insensitive to CO2 (Figure 3.1). In addition, biomass was greater at elevated CO2 through higher NPP or reduced rates of litter production (Figure 3.1). However, the increase in biomass at elevated CO2 was always greater when stomatal conductance also decreased (Figure 3.1). The hypothesis that declining stomatal conductance will lead to higher soil water potential at elevated CO2 was tested. MEDRUSH simulations for three sites in the western Mediterranean Basin suggested that this hypothesis is not generally applicable. Indeed, at the Murcia site, soil water potential was lower at elevated CO2 during the summer when stomatal conductance declined, but higher when conductance remained unchanged (Figure 3.2). Model simulations suggested that the relationship between stomatal conductance, transpiration and soil water potential at elevated CO2 is complex, site-specific and mediated by seasonal changes in LAI (Figure 3.2). Simulation of LAI is dynamic in the MEDRUSH vegetation model and depends on a range of interacting factors. Leaf growth depends on flushes during the spring which utilize stored carbohydrate, and NPP, which depends on the balance between canopy photosynthesis and respiration. Leaf loss depends on the rate of litter production, which increases when carbohydrate stores decline to starvation levels during drought periods. Some potential interactions between these processes at elevated CO2
38
Mediterranean Desertification
are explored later in this chapter (section 5). However, the next section examines the potential for rising CO2 concentration to offset the adverse effects of long-term decreases in water availability. Model simulations for Murcia, Seville and Perpignan all indicated an increase in annual wateruse efficiency (NPP/transpiration) at elevated CO2 , irrespective of whether stomatal conductance decreased (Figure 3.1). Increased water-use efficiency could allow a similar NPP to be maintained at elevated CO2 using less water.
4
RISING CO2 AND LONG-TERM DECLINE IN WATER AVAILABILITY
The Mediterranean Basin is likely to become more arid in future because of a regional increase in evaporation, accompanied by declining precipitation in some areas (Palutikof et al. 1994; Segal et al. 1994). However, adverse effects of decreasing water availability on primary production could be offset by increasing water-use efficiency, resulting from the concurrent rise in CO2 concentration. Here, we use the MEDRUSH vegetation model to quantify the extent to which rising CO2 concentration could offset the adverse effects of decreasing precipitation. The study focuses on south-east Spain, an area where the MEDRUSH model has been tested, aridity has increased since pre-industrial times, and precipitation is likely to decline in future (Palutikof et al. 1994; Araus et al. 1997). Simulations test a range of CO2 and precipitation scenarios, and their interacting effects on simulated shrub biomass in south-east Spain. Future increases in CO2 concentration will depend on global changes in anthropogenic emissions, which are largely under political control. Uncertainty in predicting future changes in precipitation arises from unreliability in the predictive power of general circulation models (GCMs) of climate. The UN Intergovernmental Panel on Climate Change (IPCC) have used state-of-the-art global carbon cycle models to predict atmospheric CO2 in the 21st century. Future scenarios of CO2 emissions in these models depended on a range of political factors, and resulted in modelled CO2 concentrations for 2050 of between 450 and 550 ppm (Leggett et al. 1992; Schimel et al. 1996). This range was therefore used for model runs, in order to test the potential range of future vegetation responses. A temperature increase of 1.2 ◦ C in 2050 was used in all model runs (Mitchell et al. 1995); however, sensitivity of simulated biomass to increasing temperature is low compared with sensitivity to CO2 or precipitation (results not shown). A range of changes in precipitation were tested, from no change to a maximum decrease of 25% in 2050, encompassing projected decreases for south-east Spain from a range of studies (Palutikof et al. 1994; Rotmans et al. 1994; Jones et al. 1996). Modifications in climate and CO2 were all imposed as linear changes from the present day to 2050. In contrast to the previous section, the MEDRUSH model was run using a transient climate, and control climate data were a daily mean series from the Alcantarilla meteorological station. Simulations therefore accounted for inter-annual variability in climate. If CO2 remained constant, decreases in precipitation below about 5% in 2050 led to a decline in simulated biomass (‘+Temperature’, Figure 3.3). A 25% decline in precipitation in 2050 caused a 40% decline in simulated biomass (‘+Temperature’, Figure 3.3). Future effects of increasing aridity on sclerophyllous vegetation in south-east Spain could therefore be very significant in the absence of rising CO2 . However, rising CO2 concentrations partially offset these adverse effects of drought in the region. Biomass remained unchanged compared with the control simulation if precipitation declined by 14% and CO2 rose to 450 ppm in 2050 (lower limit of ‘Rising CO2 ’ range, Figure 3.3), the lowest concentration suggested by the IPCC. Similarly, a 19% reduction in precipitation in 2050 had no effect on biomass if it was accompanied by an increase in CO2 concentration to 550 ppm (upper limit of ‘Rising CO2 ’ range, Figure 3.3), the highest concentration suggested by the IPCC. Vegetation model simulations therefore indicate that a rising CO2 concentration has the potential to completely offset the adverse effects of increasing aridity suggested for the Mediterranean by regional climate models. However, the extent to which this occurs will depend critically on the magnitude of both increases in CO2 concentration and decreases in precipitation in the region. Experiments on Mediterranean sclerophyllous vegetation at high CO2 suggest that productivity could be greater (on a relative basis) when water availability is low, since elevated CO2 increases water-use efficiency, and can potentially decrease the rate of water use (see sections 2 and 3). A
39
Effects of Rising CO2 and Climatic Change on Vegetation 40 Rising CO2
Change in biomass (%)
20
0
−20
+ Temperature −40
−20
−10
0
Change in precipitation (%)
Figure 3.3 Potential change in the biomass of Mediterranean sclerophyllous vegetation caused by climatic change, relative to a control with no change in climate. Climatic change was imposed as linear changes in climate inputs with time. Above-ground biomass was simulated for Alcantarilla in south-east Spain using the MEDRUSH vegetation model, and the responses to climate shown were calculated using mean values for the period 2041–2050. The dashed line indicates the effects of increasing temperature by 1.2 ◦ C in 2050 for a range of decreases in precipitation. Dotted lines delimit the range of effects of increasing CO2 concentration to between 450 and 550 ppm in 2050 (the full range of IPCC scenarios), in conjunction with temperature and precipitation changes
review of the effects of elevated CO2 on natural herbaceous vegetation suggested that this could be a general pattern (Koch and Mooney 1996). However, in the long-term model simulations described in this section, elevated CO2 had a similar relative effect on biomass irrespective of water availability (Figure 3.3). The next section explores this potential interaction between water availability and rising CO2 on shorter time-scales, and suggests mechanisms that may mediate this response in Mediterranean sclerophyllous vegetation.
5 ELEVATED CO2 AND SHORT-TERM VARIATION IN WATER AVAILABILITY Precipitation varies considerably between years in dry parts of the Mediterranean Basin, and the positive effect of elevated CO2 on primary productivity might therefore also show significant interannual variation, being greater in dry years than in wet years (on a relative basis) due to increased water-use efficiency and reduced rates of water use. In addition, long-term decreases in water availability might be expected to increase the positive effect of rising CO2 for the same reasons. Here, we tested the hypothesis that the relative effects of CO2 on primary productivity are inversely related to water availability on a time-scale of months to years. We used simulation results from the MEDRUSH model to compare the responses of spring and annual growth to water availability and elevated CO2 .
40
Mediterranean Desertification
The MEDRUSH vegetation model was run until 2050 for a site in south-east Spain, with the CO2 concentration rising to 500 ppm (IS92a), and with no change in temperature or precipitation. The CO2 concentration was then held constant, and the model run for a further 30 years. Results were compared from two consecutive years within this latter period, where annual precipitation differed markedly, being 300 mm in the first, “wet” year and 113 mm in the second, “dry” year. Long-term mean annual precipitation for the site was 289 mm, close to that in the “wet” year. In agreement with the general trend suggested by experiments (H¨attenschwiler et al. 1997b), elevated CO2 had a larger relative effect on NPP during the spring when water availability was low, than when water availability was high. NPP was 35% greater at elevated CO2 than the control in the dry year, but only 22% greater in the wet year (Table 3.1 and Figure 3.4). This could be explained by differences between years in the relative effect of elevated CO2 on photosynthesis. Total canopy photosynthesis in the spring rose by 21% at elevated CO2 in the wet year; however, this stimulation was more than 30% in the dry year (Table 3.1 and Figure 3.4). Interaction between water availability and CO2 concentration arose through a greater stimulation of both photosynthesis per unit leaf area and canopy leaf area index at elevated CO2 in the dry year (Table 3.1). This interaction was associated with increased water-use efficiency for growth (higher NPP for a given Table 3.1 Seasonal variation in the effects of an elevated CO2 concentration of 500 ppm, compared with a control (350 ppm), on model simulations of the physiology and structure of Mediterranean sclerophyllous vegetation in south-east Spain. Results are shown for the spring and summer periods of two consecutive years where annual precipitation differed: 300 mm in the ‘‘wet’’ year and 113 mm in the ‘‘dry’’ year. Mean annual precipitation for the area was 289 mm
Wet Year
Spring (March–May) NPP (g m−2 ground area day−1 ) Total photosynthesis (g m−2 ground area day−1 ) Photosynthetic rate (g m−2 leaf area day−1 ) LAI (m2 m−2 ) Stomatal conductance (mmol m−2 leaf area s−1 ) Summer (June–August) Biomass in June (g m−2 ) Total respiration (g m−2 ground area day−1 ) Specific respiration rate (g CH2 O g−1 biomass) Wood biomass fraction (dimensionless) Biomass in August (g m−2 ) Whole Year NPP (g m−2 ground area day−1 )
Dry Year
Control
Elevated CO2
% Change
Control
Elevated CO2
% Change
0.80
0.97
22
0.32
0.43
35
6.4
7.7
21
2.6
3.5
32
17.6
20.4
16
4.8
5.7
20
0.56 77
0.62 58
0.35 123
0.36 93
3 −24
11 −24
468 5.9
596 7.1
27 21
426 3.1
563 4.2
0.007
0.006
−7
0.004
0.004
1
0.83
0.84
2
0.95
0.95
0
517
657
27
344
445
29
0.40
0.47
18
0.13
0.18
32
32 35
Effects of Rising CO2 and Climatic Change on Vegetation dry year
12 8 4 0 0 −4 −8
0 −1 −2 −3 −4
2 1
Biomass (g m −2)
0
Litter (g m −2 day −1)
NPP (g m −2 day −1)
Respiration (g m −2 day −1)
Photosynthesis (g m −2 day −1)
wet year
41
Aboveground
600 400 200
Storage
0
80
−2
60
−3
40
−4
20
−5
P (mm month−1) T air (°C)
Ψsoil (MPa)
0 −1
0 J FM A M J J A SO N D J FM AM J J A SO N D Month
Figure 3.4 Simulated effects of an elevated CO2 concentration of 500 ppm (solid symbols) compared with a control (350 ppm, open symbols) on canopy photosynthesis (g CH2 O m−2 day−1 ), maintenance respiration (g CH2 O m−2 day−1 ), above-ground net primary productivity (NPP, circles, g biomass m−2 day−1 ), above-ground litter production (triangles, g biomass m−2 day−1 ), above-ground biomass (circles, g m−2 ) and the mass of stored carbohydrate (squares, g m−2 ) for Mediterranean sclerophyllous vegetation in south-east Spain (all on a ground area basis). In all cases, fluxes that led to an increase in biomass are shown as positive values, and fluxes that led to a decrease in biomass are shown as negative values. Results are shown for two consecutive years with differing annual precipitation: 300 mm in the ‘‘wet’’ year and 113 mm in the ‘‘dry’’ year. Mean annual precipitation for the area was 289 mm. Seasonal changes in monthly precipitation (bars, mm), daily mean temperature (solid line, ◦ C) and soil water potential (lines with symbols, MPa) are also shown
42
Mediterranean Desertification
soil water potential) and reduced stomatal conductance during the spring at elevated CO2 (Table 3.1), but also with a slightly greater soil water potential during spring in the dry year (0.2 MPa; Figure 3.4). Underlying mechanisms for the interaction were therefore similar to those observed in experiments on Mediterranean sclerophyllous shrubs at elevated CO2 (section 2). Greater spring growth in Mediterranean sclerophyllous shrubs at elevated CO2 had important consequences for survival during the summer drought period. Total respiration during summer in the wet year was 21% greater at elevated CO2 , but 34% greater in the dry year (Table 3.1 and Figure 3.4). Increases in respiration were largely due to the greater respiring biomass at elevated CO2 compared with the control (Table 3.1). The respiration rate per unit mass was similar or lower in high CO2 , because of a higher proportion of biomass in woody organs, which have lower respiration rates than more metabolically active organs such as leaves and fine roots (Table 3.1). Stored carbohydrates are mobilized to provide the substrate for respiration during the summer when photosynthesis in Mediterranean sclerophyllous vegetation is limited by drought (Larcher 1995). However, in the dry year, carbohydrate stores were depleted before the end of the summer, and starvation of plant organs led to abscission of leaves and fine roots, and the progressive death of woody tissues, which formed litter (Figure 3.4). Since total respiration was higher at elevated CO 2 , litter production was also significantly greater (Table 3.1 and Figure 3.4). Although soil water potential during the summer was slightly lower at elevated CO2 than in the control, it was below the minimum water potential for photosynthesis in both (Figure 3.4; wilting point = −3.6 MPa), and would not have influenced the carbon balance. Greater rates of respiration and litter production during the summer partially offset the interactive effect of elevated CO2 and water availability during the spring. Biomass was 32% higher at elevated CO2 than in controls during June in the dry year, and 27% higher in the wet year (Table 3.1 and Figure 3.4). However, by the end of the summer this increase had fallen to 29% in the dry year, remaining unchanged in the wet year (Table 3.1). Seasonal changes in the effects of elevated CO2 therefore explain why elevated CO2 and annual precipitation do not have an interactive effect on simulated biomass at this site in south-east Spain. For the two-year example presented above, there was an interaction between water availability and NPP. Elevated CO2 stimulated NPP by 18% in the wet year, but by 32% in the dry year (Table 3.1). However, when 30 consecutive years from the model were considered, there was no interaction between CO2 and precipitation for annual NPP (Figure 3.5). A statistically significant correlation between stimulation of NPP in the spring and precipitation was observed, but the data showed considerable scatter (Figure 3.5). Why was the interaction between elevated CO2 and precipitation not apparent for annual NPP in the whole data set of 30 years? Two mechanisms can be suggested to explain the observed contrast between spring and annual NPP. First, as described above, the greater increase in respiration at elevated CO2 during dry years would have tended to offset the similar increase in photosynthesis. However, this mechanism cannot explain the whole discrepancy, since an interaction was still observed for the two years presented above as an example. Second, since elevated CO2 had only a minor effect on soil water potential, NPP fell to zero in response to water deficit at similar times in both elevated CO2 and the control (Figure 3.4). For extreme or long dry periods, therefore, elevated CO2 had no effect on NPP, and a similar effect would have occurred when photosynthesis was limited by low temperature (Long 1991). However, long wet periods would have allowed NPP to continue, leading to a positive effect of elevated CO2 on NPP for longer. In combination, these effects would have tended to reverse the expected interaction, leading to a greater effect of CO2 during wet than dry years, and seem to have been most pronounced during autumn (Figure 3.5). Trends for summer in the 30-year data set were intermediate to those in spring and autumn (Figure 3.5), and the CO2 effect on NPP during winter was limited by low temperature (not shown). In summary, model results for south-east Spain show that synergistic interaction between elevated CO2 and water availability does occur for photosynthesis and NPP when short periods are considered, especially during the spring. However, the interaction seems to vary significantly within and between seasons, and compensatory mechanisms during the dry season offset any interaction for annual NPP or biomass.
Effects of Rising CO2 and Climatic Change on Vegetation
43
Whole year 50
25
0
−25
0
200
400
600
Change in NPP due to elevated CO2 (%)
Precipitation (mm) Spring 100 75 50 25 0
−25 Autumn 100 75 50 25 0
−25
0
100
200 Precipitation (mm)
300
Figure 3.5 Inter-annual and seasonal variability in the effect of elevated CO2 on net primary productivity (NPP) for Mediterranean sclerophyllous vegetation in south-east Spain. The effect of elevated CO2 concentration (500 ppm) on simulated NPP is expressed relative to a control simulation (%). Variability is shown for annual NPP and NPP for the spring (March–May) and autumn (September–November), in relation to precipitation during the same periods (mm). Results are shown for 30 consecutive years. There was a statistically significant correlation between the relative change in NPP at elevated CO2 and precipitation for the spring (r 2 = −0.453, p = 0.014)
44
6
Mediterranean Desertification
CONCLUSIONS ON RISING CO2 AND WATER-USE
Experimental and model studies of Mediterranean sclerophyllous vegetation at high CO2 suggest the following: 1. The rate of water loss from soil under stands of Mediterranean sclerophyllous shrubs may decrease with rising CO2 because of reduced rates of transpiration, and lead to higher water availability for vegetation during the summer drought period. Experimental studies have suggested that this decrease in transpiration can result from stomatal closure at elevated CO2 (section 2). In addition, model studies suggest that transpiration can also potentially decline because of a reduction in leaf area at elevated CO2 , irrespective of stomatal responses (section 3). However, responses are highly site-specific, and both experimental and model studies demonstrate that higher soil water potential is not an inevitable consequence of rising CO2 . Stomata may be insensitive to CO2 in some Mediterranean sclerophyllous species, and leaf area could potentially increase at elevated CO2 (sections 2 and 3). As a result, transpiration may be higher and soil water potential lower under stands of vegetation in a high CO2 world (sections 2 and 3). 2. The efficiency of water use in primary production increases with rising CO2 in Mediterranean shrub vegetation. Experimental studies indicate an increased water-use efficiency for both photosynthesis and growth (section 2), and model studies support this finding for a range of sites in the Mediterranean region, irrespective of whether total water use decreases at elevated CO2 (section 3). The result is that rising CO2 has the potential to partially alleviate the adverse effects of drought on primary production, even in areas where soil water availability could decrease (sections 3 and 4). This finding could be of crucial importance in the future with increasing aridity in the Mediterranean Basin, especially in areas where precipitation is predicted to decline (section 4). 3. Responses of Mediterranean sclerophyllous vegetation to high CO2 vary on a seasonal and interannual basis in relation to soil water availability. Experimental studies indicate that stomata may be less sensitive to CO2 during the summer drought period than during the spring, when the majority of growth occurs in Mediterranean vegetation (section 2). Spring growth may be greater (on a relative basis) in dry years (section 2). In addition, model simulations suggest that the responses of transpiration, leaf area and soil water potential to high CO2 may also differ between spring and summer (section 3), and between years (section 5). Seasonal variation means that generalizations possible for other natural vegetation types may not be applicable to Mediterranean shrubs. In particular, the observation that effects of high CO2 on primary productivity (on a relative basis) may be greater when water availability is low, seems to be highly dependent on the time-scale under consideration. Model studies indicate that the interaction could potentially apply during the spring, but may be counteracted by vegetation processes occurring during the summer drought (section 5). In particular, relationships between respiration, storage and litter production seem to be important (section 5), and may underpin seasonal changes in water use (section 3). Understanding the effects of elevated CO2 on these processes may be critical to successful predictions of the structure and functioning of Mediterranean sclerophyllous vegetation in a high CO2 world.
ACKNOWLEDGEMENTS We thank Dr D.J. Beerling for helpful comments and discussion during the course of this work. Climate data for the Alcantarilla meteorological station were provided by C. Goodess of the Climate Research Unit, University of East Anglia, UK.
REFERENCES Araus JL, Febrebo A, Buxo R, Camalich MD, Martin D, Molina F, Rodriguez-Ariza MO and Romagosa I (1997) Changes in carbon isotope discrimination in grain cereals from different regions of the western Mediterranean Basin during the past seven millennia. Palaeoenvironmental evidence of a differential change in aridity during the late Holocene. Global Change Biology 3, 107–118.
Effects of Rising CO2 and Climatic Change on Vegetation
45
Archibold OW (1995) Ecology of World Vegetation. Chapman & Hall, London. Betts RA, Cox PM, Lee SE and Woodward FI (1997) Contrasting physiological and structural vegetation feedbacks in a climate change simulation. Nature 387, 796–799. Curtis PS and Wang X (1998) A meta-analysis of elevated CO2 effects on woody plant mass, form, and physiology. Oecologia 113, 299–313. Drake BG, Gonz´alez-Meler MA and Long SP (1997) More efficient plants: a consequence of rising atmospheric CO2 ? Annual Review of Plant Physiology and Plant Molecular Biology 48, 607–637. Ehleringer JR and Mooney HA (1983) Productivity of desert and Mediterranean-climate plants. In OL Lange, PS Nobel, CB Osmond and H Zeigler (eds) Physiological Plant Ecology IV. Ecosystem Processes: Mineral Cycling, Productivity and Man’s Influence. Springer-Verlag, Berlin, pp. 205–231. Field CB, Jackson RB and Mooney HA (1995) Stomatal responses to increased CO2 : implications from the plant to the global scale. Plant, Cell and Environment 18, 1214–1225. H¨attenschwiler S, Miglietta F, Raschi A and K¨orner Ch (1997a) Morphological adjustments of mature Quercus ilex trees to elevated CO2 . Acta Oecologica 18, 361–365. H¨attenschwiler S, Miglietta F, Raschi A and K¨orner Ch (1997b) Thirty years of in situ tree growth under elevated CO2 : a model for future forest responses? Global Change Biology 3, 463–471. Jones MB, Brown JC, Raschi A and Miglietta F (1995) The effects on Arbutus unedo L. of long-term exposure to elevated CO2 . Global Change Biology 1, 295–302. Jones PD, Hulme M, Briffa KR, Jones CG, Mitchell JFB and Murphy JM (1996) Summer moisture availability over Europe in the Hadley Centre general circulation model based on the Palmer drought severity index. International Journal of Climatology 16, 155–172. Koch GW and Mooney HA (1996) Response of terrestrial ecosystems to elevated CO2 : a synthesis and summary. In GW Koch and HA Mooney (eds) Carbon Dioxide and Terrestrial Ecosystems. Academic Press, San Diego, pp. 415–429. Larcher W (1995) Physiological Plant Ecology. Ecophysiology and Stress Physiology of Functional Groups. Springer, Berlin. Leggett J, Pepper WJ and Swart RJ (1992) Emissions scenarios for the IPCC: an update. In JT Houghton, BA Callander and SK Varney (eds) Climate Change 1992. The Supplementary Report to the IPCC Scientific Assessment. Cambridge University Press, Cambridge, pp. 69–95. Long SP (1991) Modification of the response of photosynthetic productivity to rising temperature by atmospheric CO2 concentrations. Has its importance been underestimated? Plant, Cell and Environment 14, 729–739. Lossaint P (1973) Soil–vegetation relationships in Mediterranean ecosystems of Southern France. In F di Castri and HA Mooney (eds) Mediterranean Type Ecosystems: Origin and Structure Chapman & Hall, London, pp. 199–210. Merino O, Martin MP, Martin A and Merino J (1990) Successional and temporal changes in primary productivity in two mediterranean scrub ecosystems. Acta Oecologia 11, 103–112. Miglietta F, Raschi A, Bettarini I, Resti R and Selvi F (1993) Natural CO2 springs in Italy: a resource for examining long-term response of vegetation to rising atmospheric CO2 concentrations. Plant, Cell and Environment 16, 873–878. Mitchell JFB, Johns TC, Gregory JM and Tett SFB (1995) Climate response to increasing levels of greenhouse gases and sulphate aerosols. Nature 376, 501–504. Morison JIL (1985) Sensitivity of stomata and water use efficiency to high CO2 . Plant, Cell and Environment 8, 467–474. M¨uller MJ (1982) Selected Climate Data for a Global Set of Standard Stations for Vegetation Science. Dr W Junk Publishers, The Hague. Palutikof JP, Goodess CM and Guo X (1994) Climate change, potential evapotranspiration and moisture availability in the Mediterranean Basin. International Journal of Climatology 14, 853–869. Richter H (1997) Water relations of plants in the field: some comments on the measurement of selected parameters. Journal of Experimental Botany 48, 1–7. Rotmans J, Hulme M and Downing TE (1994) Climate change implications for Europe – an application of the ESCAPE model. Global Environmental Change – Human and Policy Dimensions 4, 97–124. Scarascia-Mugnozza GE, De Angelis P, Matteucci G and Kuzminsky E (1996a) Carbon metabolism and plant growth under elevated CO2 in a natural Quercus ilex L. “Macchia” stand. In Ch K¨orner and FA Bazzaz (eds) Carbon Dioxide, Populations, and Communities. Academic Press, San Diego, pp. 209–230. Scarascia-Mugnozza GE, De Angelis P, Matteucci G and Valentini R (1996b) Long-term exposure to elevated [CO2 ] in a natural Quercus ilex L. community: net photosynthesis and photochemical efficiency of PS II at different levels of water stress. Plant, Cell and Environment 19, 643–654.
46
Mediterranean Desertification
Schimel D, Alves D, Enting I, Heimann M, Joos F, Raynaud D and Wigley T (1996) Radiative forcing of climate change. In JT Houghton, LG Meira Filho, BA Callander, N Harris, A Kattenberg and K Maskell (eds) Climate Change 1995: The Science of Climate Change. Cambridge University Press, Cambridge, pp. 65–131. Segal M, Alpert P, Stein U, Mandel M and Mitchell MJ (1994) Some assessments of the potential 2 × CO2 climatic effects on water balance components in the eastern Mediterranean. Climatic Change 27, 351–371. Tenhunen JD, Beyschlag W, Lange OL and Harley PC (1987) Changes during summer drought in leaf CO2 uptake rates of macchia shrubs growing in Portugal: limitations due to photosynthetic capacity, carboxylation capacity, and stomatal conductance. In JD Tenhunen, FM Catarino, OL Lange and WC Oechel (eds) Plant Response to Stress. Functional Analysis in Mediterranean Ecosystems. Springer-Verlag, Berlin, pp. 305–327. Tognetti R, Giovannelli A, Longobucco A, Miglietta F and Raschi A (1996) Water relations of oak species growing in the natural CO2 spring of Rapolano (central Italy). Annales des Sciences Foresti`eres 53, 475–485. Tomaselli R (1981) Relations with other ecosystems: temperate evergreen forests, mediterranean coniferous forests, savannahs, steppes and desert shrublands. In F di Castri, DW Goodall and RL Specht (eds) Ecosystems of the World 11. Mediterranean-type Shrublands. Elsevier Scientific, Amsterdam, pp. 123–130. Woodward FI (1987) Climate and Plant Distribution. Cambridge University Press, Cambridge.
4
Use of NOAA-AVHRR NDVI Data for Climatic Characterization of Mediterranean Areas
GIOVANNI CANNIZZARO,1 FABIO MASELLI,2 LUCIANO CAROTI1 AND LORENZO BOTTAI3 1
Telespazio SpA, Roma, Italy IATA-CNR, Firenze, Italy 3 FMA, Campi Bisenzio, Firenze, Italy 2
1 INTRODUCTION The current debate on global environmental change has stimulated great research efforts related to the monitoring of vegetation dynamics over large areas. Vegetation in fact plays a fundamental role as an interface between atmosphere and land surface, and it is also a direct indicator of environmental conditions (Sellers et al. 1995). Several studies have therefore resorted to remotely sensed data as the only effective means for monitoring vegetation at wide scales (Richards 1993; Hall et al. 1995). Satellite-derived vegetation indices, in particular, are an appropriate tool with which to characterize the status and changes of vegetation cover (Bannari et al. 1995). Of these indices, one of the most popular is the NDVI (Normalized Difference Vegetation Index), which is computed as (near-infrared reflectance − red reflectance)/(near-infrared reflectance + red reflectance) The NDVI is a direct expression of plant photosynthetic activity and, when derived from NOAAAVHRR data, offers the greatest potential for vegetation monitoring at a global scale (Prince 1991). In a number of studies NOAA-AVHRR NDVI data have been related to several environmental parameters, such as green biomass, plant productivity and meteorological driving factors (Baret and Guyot 1991; Benedetti et al. 1994; Bannari et al. 1995). Rainfall in particular has been demonstrated to be the major controlling factor for NDVI development in semi-arid regions, which is mainly due to the dependence of the local vegetation cycles on precipitation levels (Maselli et al. 1993). The situation is more complex in temperate and Mediterranean environments, where vegetation development is controlled contemporaneously by thermal and pluviometric factors as well as by human activities on the territory. As regards Mediterranean areas, recent studies of our research group have shown that inter-year NDVI variations of homogeneous zones are mainly dependent on the thermal factor at the beginning of the growing season (January–March) and on the pluviometric factor during the arid season (July–September), but a strong variability exists related to climatic and land-use features (Caroti et al. 1995). This behaviour, clearly related to the main vegetation cycles in Mediterranean environments, can be a good starting point for the derivation of aridity indices from NDVI data. As is well known, several aridity indices have been proposed by classical authors for characterizing the mean climate of a region as well as inter-year variations (De Martonne 1948; Thornthwaite 1948). These indices are generally derived from parameters that are measured from meteorological stations, mainly daily or monthly temperature and rainfall. The extension of the climatic estimates over the land surface therefore becomes a fundamental issue, especially in topographically complex regions Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
48
Mediterranean Desertification
(Trewartha and Horn 1980). The definition of sound and straightforward relationships between these indices and NDVI data would provide such extended estimates with a high spatial resolution (1.1 × 1.1 km) and could also serve for the prompt delineation of inter-year aridity variations (Maselli et al. 1998). This is the objective of the research described here, carried out in a region in central Italy with typical Mediterranean features (Abruzzo). In the case study NDVI images have been related to meteorological data taken at five ground stations over eight years. Some preliminary processing has allowed the definition of the station pixels most suitable for comparison. Correlation analysis and environmental considerations have then been employed to define an optimum NDVI combination that minimizes unimportant land-use variations and can therefore be taken as an aridity index. This has finally been extended over the region for the eight study years (1986–1993) to depict the variations in aridity levels during this period.
2
STUDY AREA AND DATA
The Abruzzo Region (central Italy) was a primary test site for the MEDALUS II Project (Figure 4.1). The topography of the region varies from the coastal plains to the inner mountains, with the climate ranging from Mediterranean to cold sub-humid. The flat and hilly areas near the coast are mainly used for agriculture, while the mountains inland are mainly covered by deciduous woods and pastures. The AVHRR data, already in the form of NDVI, were taken from the archives of Telespazio (Rome, Italy) within the framework of the MEDALUS II Project. The standard procedure for the production of these data comprised the geo-referencing and pre-processing of the original images by
Abruzzo
Italy
Figure 4.1 Geographical position of the Abruzzo Region
Climatic Characterization of Mediterranean Areas
49
a cubic convolution algorithm, the radiometric calibration of the first two bands to derive apparent reflectances following Rao and Chen (1994), and the computation of NDVIs to finally obtain maximum value composites, MVCs (Holben 1986), on a monthly basis. The final products were therefore 12 monthly NDVI MVCs covering all of 1990, and these were used for the current analysis. Standard meteorological measurements (daily minimum and maximum temperatures and rainfall) were taken from the available stations of the regional service. Only five stations had complete records for the study period (Tornimparte, Alanno, Teramo, Santa Eufemia a Maiella and Castel del Monte) and from them monthly values of mean temperature and total rainfall were derived as a basis for further processing. The five stations were located in different environmental and climatic areas (from the coastal area with a Mediterranean climate to the cold sub-humid zone of the inland mountains), so that, even with a small number of stations, a good representation of the entire region could be assumed.
3 DATA PROCESSING 3.1 Computation of Aridity Indices
Since only temperatures and rainfall were available for the meteorological stations studied, these data were used to compute simplified aridity indices as proposed by classical authors. The first index was that of De Martonne (1948), IDM , which is based on the following summation: IDM =
12 i=1
Pi /(Ti + 10)
(1)
where Pi is the total rainfall in month i (mm), and Ti is the mean temperature in month i ( ◦ C). As can be easily understood, the De Martonne index is actually a pluviometric one, since its increase corresponds to a decrease in aridity level. The same can be said for the second index considered, the humidity index of Thornthwaite (1948), IT , which relies on a more sophisticated computation of the water balance through the concept of potential evapotranspiration (ETP): IT =
12 i=1
(Pi − ETPi )/ETPi
(2)
where ETPi is the total potential evapotranspiration in month i (mm), computed from the mean temperature. Both indices were computed for each study year (1986–1993) with the meteorological data of the five stations. The period of summation was kept different from the solar year (January–December) since it had to approximate the effects of climate on vegetation development. Thus, both indices were calculated with monthly data from October of one year to September of the following year, when the growing cycle of Mediterranean vegetation is generally finishing. 3.2 Correlation Analysis
The annual aridity indices from the five study stations were first regressed against the available NDVI data. A linearity in the relationship between the two variables was assumed based on the fact that the response of NDVI to vegetation limiting factors can be considered approximately linear within restricted ranges of variation (Maselli et al. 1993). The analysis also assumed that the effect of land-use variability was minor with respect to that of the limiting factors. This assumption is valid in only a few cases, so further processing to reduce the consequent disturbances was carried out in a subsequent phase (see below). For the analysis, more or less homogeneous areas with respect to vegetation features were first selected around the five meteorological stations. This was done by means of a cluster analysis carried out on the NDVI profiles as fully described in Caroti et al. (1995). About 10 pixels were selected containing each station, and the relevant monthly NDVI data were extracted. Linear regression
50
Mediterranean Desertification
analysis was finally applied to the NDVI data from January to September (during the growing cycles of vegetation) with respect to the climatic indices.
4
RESULTS AND DISCUSSION
Both climatic indices considered showed interesting correlations with the monthly NDVI levels. As can be seen from Table 4.1, negative correlations were found for the first months of the year, with a peak in February. The period April–May showed lower correlations, while high positive correlations were noted for the summer months, with a peak in July. Slightly higher correlations were found with the De Martonne index, and most of the correlation coefficients were significant at least at the 95% confidence level. These results can be interpreted on the basis of the previous study carried out on the same data set (Caroti et al. 1995). In that case it was found that temperature was positively correlated with NDVI at the beginning of the year and negatively correlated after the April–May period. The opposite trend was found for rainfall. On this basis, it can be hypothesized that the thermal factor contained in the climatic indices is most influential on the NDVI levels at the beginning of the season, while the pluviometric factor is dominant during the arid months. Since the effects of the two factors on the values of the indices are contrasting, the negative correlations at the beginning of the season and the positive correlations in summer can be understood. In effect, it is well known that in Mediterranean environments vegetation development is mainly limited by temperature in winter and by rainfall in summer (Pinna 1977). The scheme of Figure 4.2 shows typical NDVI profiles in Mediterranean and temperate environments of the northern hemisphere and can help in interpreting the results obtained. Typical NDVI profiles in temperate–humid areas show a marked increase in spring and a long plateau in summer when actual transpiration is at a maximum due to high temperatures and water availability. As the climate becomes more Mediterranean–arid, the vegetation growing cycle changes due to the increased actual transpiration in late winter–early spring, while a summer minimum appears due to increasing aridity. In practice, the transition from temperate–humid to Mediterranean–arid climates is accompanied by a primary decrease in vegetation activity in summer and a secondary increase in late winter. The trend is more confused in spring due to the interaction of different contrasting factors. This is obviously only a general scheme which can be strongly modified by several factors, particularly land-use differences. Such modifications can also help to explain why the correlation coefficients never reached very high values. Table 4.1 Correlation coefficients found between the two study aridity indices and the monthly NDVI data from January to September (1–9 = sequential months), with relevant significance levels (∗ = significant correlation, P < 0.05; ∗∗ = highly significant correlation, P < 0.01)
Month 1 De Martonne aridity index Thornthwaite aridity index
2
3
4
5
6
7
8
9
−0.59∗∗ −0.65∗∗
−0.49∗∗ −0.23
−0.39∗
0.72∗∗ 0.81∗∗ 0.83∗∗ 0.81∗∗
−0.54∗∗ −0.65∗∗
−0.52∗∗ −0.28
0.30
0.63∗∗ 0.76∗∗ 0.73∗∗ 0.73∗∗
Calculations are based on eight-year data from five weather stations. A unique De Martonne or Thornthwaite value is computed for each year and weather station and all the resulting values are regressed against the monthly values of NDVI. For example, the correlation coefficients in column 9 derive from a regression analysis between the NDVI September values of each of the eight years and the De Martonne or Thornthwaite values computed for each of the same eight years
51
Climatic Characterization of Mediterranean Areas 0.7 0.6
Mediterranean − arid 0.5
Temperate − humid
NDVI
∗∗ 0.4
Intermediate
0.3 Increasing aridity
∗ 0.2 0.1
∗
Thermal factor
∗∗
Pluviometric factor
0 1
2
3
4
5
6
7
8
9
10
11
12
Month
Figure 4.2 General scheme for the interpretation of vegetation development in relation to the prevalent climatic regime
These considerations hold both for spatial variations and for inter-year differences. They can therefore be a good starting point for the definition of an NDVI-derived index which can optimally approximate a climatic one. Such an index should possess the following desirable features: 1. 2. 3. 4.
easy derivability from monthly NDVI data; sound conceptual basis according to the previous discussion; reduction of the effects of land-use differences; maximization of the correlation with climatic indices.
On the basis of these points and of the results of the previous analysis, various combinations of winter (January–February) and summer (July–August) NDVI values were considered mainly in analogy with difference, ratio and normalized difference indices (e.g. a ratio-like index was computed as NDVIwinter /NDVIsummer ). Among these combinations, the highest correlations with the climatic indices were obtained using the following formula: INDVI = (NDVI7 + NDVI8 ) − (NDVI1 + NDVI2 )
(3)
where the subscripts refer to the sequential months. This new index responded to all the requirements mentioned and gave a high correlation coefficient with both traditional climatic indices (Figure 4.3). In practice, these correlations were very close to the global correlation coefficients obtained by multivariate regression analysis with all monthly NDVI data from January to September. As for the two traditional climatic indices, here also increases in INDVI correspond to decreases in aridity levels. The distribution of this “optimum” index in the eight study years (1986–1993) is shown in Figure 4.4. The high spatial variability of the index in the region can be appreciated, mainly related to its topographic features. Also, strong inter-year differences in the index can be noted, which are mostly due to different meteorological seasons.
5 CONCLUSIONS The current debate on the possible causes and effects of climate changes has enhanced the utility of aridity indices that were proposed by classical authors several decades ago. Conventional methods
52
Mediterranean Desertification (a)
100 90
r = 0.882
80 De Martonne Index
70 60 50 40 30 20 10 0 0
−0.25
0.25
0.5
0.75
1
1.25
0.75
1
1.25
NDVI Index 3.0
(b)
r = 0.858
2.5
Thornthwaite Index
2.0 1.5 1.0 0.5 0.0 0
−0.25
0.25
0.5
−0.5 −1.0 −1.5 NDVI Index
Figure 4.3 Linear regressions between the (a) De Martonne and (b) Thornthwaite climatic indices and the new index derived from NDVI data measured in the five study stations for the eight study years (both correlations are highly significant, P < 0.01)
of obtaining these indices are generally based on point measurements from meteorological stations. The collection of the input data, however, is not always easy and, in particular, the extension of the estimates on the land surface is usually problematic, especially in complex terrain. In this research a simple and straightforward methodology is presented to derive aridity estimates from NOAA-AVHRR NDVI data in temperate and Mediterranean areas. These data are available for most regions in the world at least on a monthly basis, and with a relatively high spatial resolution. Based on theoretical considerations and the correlations experimentally found for a region in central Italy, it has been hypothesized that a difference between NDVI levels measured in summer (when the effect of the pluviometric factor on vegetation prevails) and in winter (when the thermal factor mostly determines vegetation development) can be taken as an estimator of usual aridity indices.
Climatic Characterization of Mediterranean Areas
−1
1986
1987
1988
1989
1990
1991
1992
1993
0
53
1
Figure 4.4 Aridity maps of the Abruzzo Region derived from monthly NDVI data (NDVI7 + NDVI8 − NDVI1 − NDVI2 ) for the eight study years, 1986–1993
This hypothesis has been validated according to the De Martonne and Thornthwaite indices. More generally, it can be supposed that such an approach will work anywhere where the thermal factor is limiting for vegetation development in winter and the pluviometric factor is limiting in summer. In practice, the methodology could be applied to most areas with climate types ranging from Mediterranean–arid to temperate–cold.
ACKNOWLEDGEMENTS The present work was partly funded by the EU MEDALUS II Project. The authors would like to thank the Coordinator of the MEDALUS research group in the Abruzzo Region, Dr Claudio Conese, and the Nuova Telespazio team and Dr Claudio Serafini, for their collaboration within the framework of the Project.
REFERENCES Bannari A, Morin D, Bonn F and Huete AR (1995) A review of vegetation indices. Remote Sensing Reviews 13, 95–120. Baret F and Guyot G (1991) Potentials and limits of vegetation indices for LAI and APAR assessment. Remote Sensing of Environment 46, 213–222. Benedetti B, Rossini P and Taddei R (1994) Vegetation classification in the Middle Mediterranean area by satellite data. International Journal of Remote Sensing 15, 583–596.
54
Mediterranean Desertification
Caroti L, Maselli F and Serafini C (1995) Valutazione dell’informazione agroecologica contenuta in profili NOAA-NDVI. In A Zaghi and M Gomarasca (eds) Proceedings of the VII National Congress of AIT (Associazione Italiana di Telerilevamento). CSEA, Torino, pp. 487–492. De Martonne E (1948) Trait´e de G´eographie Physique, 7th edition. Colin, Paris. Hall FG, Townshend JR and Engman ET (1995) Status of remote sensing algorithms for estimation of land surface state parameters. Remote Sensing of Environment 51(1), 138–156. Holben BN (1986) Characteristics of maximum-value composite images from temporal AVHRR data. International Journal of Remote Sensing 7, 1417–1434. Maselli F, Conese C, Petkov L and Gilabert MA (1993) Environmental monitoring and crop forecasting in the Sahel through the use of NOAA NDVI data. A case study: Niger 1986–1989. International Journal of Remote Sensing 14, 3471–3487. Maselli F, Petkov L and Maracchi G (1998) Extension of climate parameters over the land surface by the use of NOAA-AVHRR and ancillary data. Photogrammetric Engineering and Remote Sensing 64, 199–206. Pinna M (1977) Climatologia. Ed UTET, Torino, Italy. Prince SD (1991) Satellite remote sensing of primary production for use with coarse resolution satellite data. International Journal of Remote Sensing 12, 1313–1330. Rao CRN and Chen J (1994) Post-launch calibration of the visible and near infrared channels of the Advanced Very High Resolution Radiometer on NOAA-7, -9, and 11 spacecraft. NOAA Technical Report NESDIS 78, US Department of Commerce, Washington, DC, August 1994. Richards JA (1993) Remote Sensing Digital Image Analysis: An Introduction, 2nd edition. Springer-Verlag, Heidelberg. Sellers PJ, Meeson BW, Hall FG, Asrar G, Murphy RE, Schiffer RA, Bretherton FP, Dickinson RE, Ellingson RG, Field CB, Huemmrich KF, Justice CO, Melack JM, Roulet NT, Schimel DS and Try PD (1995) Remote sensing of the land surface for studies of global change: models–algorithms–experiments. Remote Sensing of Environment 51(1), 3–26. Thornthwaite CW (1948) An approach toward a rational classification of climate. Geographical Review 38, 55–94. Trewartha GT and Horn LH (1980) An Introduction to Climate, International Student Edition, Tokyo.
Section III
Land Use, Processes and Responses
5
The Effect of Land Use on Soil Erosion and Land Degradation under Mediterranean Conditions
3 ´ ´ C. KOSMAS,1 N.G. DANALATOS,2 F. LOPEZ-BERM UDEZ AND M.A. ROMERO D´ıAZ3
1
Laboratory of Soil Chemistry, Agricultural University of Athens, Greece Department of Agriculture, University of Thessaloniki, Volos, Thessaloniki, Greece 3 Department of Physical Geography, University of Murcia, Spain 2
1 INTRODUCTION Soil erosion is the most serious land degradation hazard in the Mediterranean uplands. It brings about siltation of water courses, infilling valleys and reservoirs, and forming deltas in coastal areas. It drastically reduces soil productivity, due to soil structural deterioration, nutrient wash out, and reduction of the water-holding capacity, limits vegetation growth and eventually leads to extensive desertification. Soil erosion is a natural phenomenon, a process by which all protruding surfaces of land are worn down gradually by the elements, only to be uplifted and re-formed time and again by geological forces. Soil erosion problems begin when human activity intervenes to accelerate the natural process far beyond the rate at which nature can redress it (Hillel 1991). Though it is clear that land degradation in the Mediterranean is in many ways a product of human activity, nevertheless it is significantly amplified by physical factors. Thus, the high erosion rates occurring in Mediterranean type areas are attributed to the climatic regime (Langbein and Schumm 1958), to the existing generally poor vegetation cover, and to land-use management (Douglas 1969; Reed 1971; Williams and Reed 1972; Patton and Schumm 1975; L´opez-Berm´udez et al. 1984; Newson 1985; Bryan and Campbell 1986). Mediterranean environments are characterized by strong seasonal contrasts in climate. When this is coupled with low annual rainfall and parent materials of high erosional susceptibility, the potential for erosion is very high (Romero D´ıaz et al. 1988). The extreme intensity and irregularity of annual precipitation events is a primary cause of soil erosion in Mediterranean Europe. During such rainfall events, the energetic impact of rainfall at the soil surface (rainsplash) modifies the soil physical properties; as a consequence, the soil particles are destabilized, detached and subsequently transported downslope by the running water. The pressure exerted by the raindrops subjects the soil to strong compaction and consolidation of its surface, resulting in an impermeable crust that inhibits infiltration and increases runoff. Romero D´ıaz et al. (1995) found that, as expected, the mean rainfall intensity was the factor best correlated with runoff coefficient and sediment production. Total rainfall correlated modestly, whereas vegetation cover influenced the runoff coefficient but no effects on sediment yields were detected. Studies conducted during MEDALUS projects I and II point to the existence of a relationship between annual rainfall and runoff, and sediment loss for different kinds of land use, and therefore between annual precipitation and soil erosion in the Mediterranean region. The following land-use types can be graded in order of decreasing effect on soil erosion: vines > eucalyptus > wheat > shrub land > olives. In the Mediterranean the topography of a region exerts a powerful influence on settlement patterns and land-use practices, as well as being a contributory factor in soil erosion. The physiography Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
58
Mediterranean Desertification
of the Mediterranean area includes a diverse array of land-forms, a large proportion of which are dominated by sparsely vegetated upland zones (Perez-Trejo 1992). Such a physiographic relief presents ideal conditions for the generation of water erosion on slopes, resulting in loss of soil productivity and desertification. The slope aspect greatly influences the temperature of the local environment, which in turn affects evaporation and subsequently vegetation growth and resilience. A greater slope inclination also influences infiltration rates and accelerates runoff and sediment loss. The steepest slopes generate mass movements, such as landslides and mudflows. The vegetation patterns that cover a landscape affect the soil in all its dynamics, including water redistribution over and within the soil, and microbiological activity. Biotic interactions occur which generate and maintain soil structure in the upper soil through the process of aggregation. Aggregation is a strong determinant of the soil’s hydrological and biological characteristics (Imeson 1984), and affects erosional response. Extensive Mediterranean areas cultivated with rainfed crops such as cereals, vines, almonds and olives are mainly confined to hilly lands with shallow soils very sensitive to erosion. These areas become vulnerable to erosion and desertification because the reduced vegetation cover means less protection from raindrop impact during heavy rains (Faulkner 1990), the reduction of infiltration rates due to compaction from heavy machinery (Fullen 1985), and the formation of a surface crust (Morin and Benyamini 1977; Casenave and Valentin 1992; Romero D´ıaz et al. 1998). Many authors have demonstrated that in a wide range of environments, both runoff and sediment loss decrease exponentially with increasing percentage of vegetation cover (Elwell and Stocking 1976; Lee and Skogerboe 1985; Francis and Thornes 1990). Without vegetation, all the runoff energy is directed to soil erosion and the removal of the detached material over various distances. Thus, vegetation and land use are in that respect of paramount importance, controlling the intensity and the frequency of overland flow and surface wash erosion (Bryan and Campbell 1986; Mitchell 1990). Large-scale deforestation of semi-arid areas accompanied by intensive cultivation and overgrazing has resulted in accelerated erosion and the formation of badlands with very shallow soils. Erosion rates measured in Mediterranean badlands vary widely from 0.4 to 1.7 mm year−1 (Yair et al. 1982; Benito et al. 1992). However, even greater erosion rates have been reported elsewhere, as in the Trevelez river basin in Spain in which an average soil loss value of 2.4 mm year−1 has been measured (Martin-Vivaldi and Jimenez Olivencia 1992). In critical semi-arid areas of Spain, soil loss of 12 mm year−1 or 200 t ha−1 year−1 has been reported (L´opez-Berm´udez 1990). These values can easily be surpassed during heavy rainfall events occurring occasionally over the Mediterranean (Alias-Perez and Ortiz-Silla 1986; L´opez-Berm´udez et al. 1991; Romero D´ıaz et al. 1995).
2
HISTORICAL EVOLUTION OF LAND USE IN THE MEDITERRANEAN REGION
The Mediterranean must be the region of the world most badly affected by human-induced degradation over thousands of years. The evidence of degradation is very clear, with only relict patches of the indigenous forest cover remaining and entire landscapes no longer able to sustain any cultivation. Accelerated soil erosion is as old as farming. Two early leaders of the US Conservation Service, Hugh Bennett and Clay Lowdermilk, wrote in the 1938 Yearbook of Agriculture: “soil erosion began when the first heavy rain struck the first furrow turned by a crude implement of tillage in the hands of prehistoric man. It has been going on ever since, wherever man’s culture of the earth has bared the soil to rain and wind” (USDA 1938). Soil erosion was first reported by Homer in The Iliad. Greek hillsides were originally forested and covered by a fertile soil mantle, which, however, was rather shallow and vulnerable to erosion. Upland grazing and farming probably began around the middle of the second millennium and began the initial damage to forests. Several thousands of years of exploitative agriculture have greatly contributed to a dramatic reduction of agricultural productivity in the region, something that had already been mentioned by Plato, who, speaking for Attica in the 4th century BC (Critias III), noted the occurrence of massive floods and landslides, the disappearance of forests and the denudation of cattle pasture. This description provides us with one of the earliest recorded examples of degradation and desertification, but also implicates climatic as well as anthropogenic causes. Two centuries earlier, Solon had already advocated discontinuing grain
Effect of Land Use on Soil Erosion and Land Degradation
59
cultivation on the sloping lands of Attica, and recommended planting olives and grapes instead. His advice was echoed in the 4th century BC by Theophrastus in his “Cause of Plants”. Considering the effects of land use on erosion and particularly the positive effects of olive groves (see below), one realizes how suitable this early land-use change plan of Solon was. However, neither man’s advice addressed the root cause of the problem, which was not the choice of the crop as such but the process of erosion and the failure of the ancient Greeks to control it. Additional historical evidence relating to the effects of degradation on vegetation can be traced to Roman times when land degradation resulted in the creation of large pastoral estates. Wherever Romans established their dominion, they repeated the same pattern of extensive forest clearing, over-cultivation and overgrazing of land to satisfy the avaricious demands of their centre of power (Hillel 1991). Land-use changes in the Mediterranean during recent history are mainly due to physical and technical factors as well as socio-economic reasons. Particular land uses have been related to specific population behaviours, spatial distribution changes, and pressure over natural resources. The region has suffered important transformations since the middle of the 19th century, when the agricultural development really began. Land mismanagement stimulated by demographic dynamics resulted in shifting of the agricultural population (and activities) to marginal areas unsuitable for agriculture. Human impact on the landscape was increasingly negative through conventional large-scale extensive agriculture, negatively affecting soil properties and enhancing the erosion processes. The extension of cultivated areas at the expense of forest land implies high ecological alterations due to deforestation and the break-up of the original equilibrium between cultivation, grazing and forestry. Short-term capital investment and intensive cultivation have often resulted in land degradation. Land profits are usually not invested for land conservation measures, but are simply reinvested for cultivating another area. The most significant change in the current land-use distribution in Mediterranean Europe is the increasing intensification of agricultural land in terms of mechanization, extensive use of agro-chemicals, and irrigation. The Guadalent´ın Basin in south-eastern Spain may serve as an example for demonstrating the impacts of land transformation changes and population evolution on land degradation (Barbera et al. 1997). The basin is characterized by the greatest hydrological deficit in the Iberian Peninsula and also in Europe. Historically the lack of water resources and the pressure for land-use change have been constant factors. Land-use changes have been related to specific population behaviour and spatial distribution, and pressure on natural resources, often as a response to economic demands. The Guadalent´ın has suffered significant transformation since the latter half of the 19th century, when agricultural development began. Since then agricultural activities and some mining have seriously affected the rural landscape and the whole environment in general. Population evolution in the Guadalent´ın Basin has been analysed from the middle of the 18th century and indicates substantial interaction of population dynamics with land-use changes (L´opez-Berm´udez et al. 1995). On the other hand, extreme climatic events typical for the region have also exerted an important socio-economic influence. Soil erosion cannot be considered as a human-induced disaster of only recent times (Wise 1982). Archaeological and geomorphological evidence from the badlands in southern Spain shows that the basic physical properties such as drainage patterns and degree of slope have been in place for some 4000 years. In the hilly Guadalent´ın Basin, human-induced land degradation has been particularly due to intensive cereal cropping, grazing and exploitation of Quercus forest resources (Figure 5.1). Inappropriate agricultural practice and management in relation to soil properties, topography and climate have stimulated economically based political decisions that have resulted in the migration of people and their agricultural activities to marginal areas with poor soils not necessarily suitable for agriculture. Another negative human impact on the landscape has been through conventional largescale extensive agriculture using mechanization, weakening soil properties in relation to weathering and erosive processes. Due to economic reasons and also as a response to soil degradation, large areas then had to be abandoned or used only for grazing. The following discussion focuses on the impacts of precipitation and land use on erosion rates based on an extensive database collected in various northern Mediterranean sites, located in Portugal, Spain, France, Italy and Greece. These sites represent a variety of landscapes under a variety of land
60
Mediterranean Desertification Dry land
Irrigated land
Forested land
1850
1981
60
Area (%)
50 40 30 20 10 0 1755
Year
Figure 5.1 Changes in land-use types in the Guadalent´ın Basin (Spain) since 1755 ´ (Lopez-Berm udez et al. 1995) ´
uses typical for the Mediterranean region, such as agricultural land cultivated with rainfed cereals, vines, olives, eucalyptus plantation or under natural vegetation (shrub land).
3 3.1
LAND USE AND EROSION RATES The Impact of Vegetation and Surface Soil Conditions
The effects of soil surface conditions and percentage vegetation cover are of paramount importance to rainwater runoff and sediment loss. These effects were clearly demonstrated across a hillslope catena, where land use was the only dependent parameter and where all other factors, e.g. weather, soils and topography, remained almost standard. The hillslope (gradient 14.5–16.2%) is formed on a sandstone formation near Athens (southern Greece). The climate of the area is Thermo-Mediterranean with an average air temperature of 17.8 ◦ C. The average annual precipitation is 495 mm with more than two-thirds (71%) falling between November and April. The following soil-surface conditions/land uses were studied, all being typical for Mediterranean environments: • • • • •
Olive grove under semi-natural conditions, with winter-annual understorey vegetation. The soil surface was sufficiently protected from raindrop impact by the ground cover (including the plant residues). No ploughing of the soils took place for more than 20 years. Vine cultivation with moderate inputs involving sufficient weed control. The soil was ploughed parallel to the contours. Surface roughness was estimated at 14 cm, and clod/furrow angle at 30◦ . Bare land abandoned for 2.5 years, without any vegetation (kept bare by controlling weeds) and with an average soil surface roughness of 4 cm. Land abandoned for 2.5 years. The soil contained large rock fragments (15 cm average diameter) partially embedded in the soil surface, and covering 17.8% of it. The average soil-surface roughness was 4 cm. Land with annual vegetation and abandoned for 2.5 years. The soils were under natural vegetation (no weeding) with an average soil-surface roughness of 4 cm.
Four rainfall events (27.5, 24.9, 28.5 and 18.2 mm) inducing incipient ponding fell between the end of November and late January 1994, and runoff volumes were measured from the different plots. There was considerable variation in the total runoff, reflecting the enormous importance of surface conditions on runoff generation and land degradation (Figure 5.2). The presence of annual vegetation and the plant residues covering about 90% of the soil surface in the olive grove prevented
61
Effect of Land Use on Soil Erosion and Land Degradation 12
(a)
Runoff (mm)
9
6
3
0
27.5
24.3
28.5
18.2
Rainfall events (mm) (b) 64.10
Sediment loss (t km−2)
olives 48.07 vines bare
32.05
rock fragments annual vegetation
16.02
0.00
27.5
24.3
28.5
18.2
Rainfall events (mm)
Figure 5.2 (a) Rainfall runoff and (b) sediment loss measured during four rainfall events under different land uses and surface soil conditions
the formation of surface sealing and minimized the velocity and volume of runoff water. A total runoff of only 1.0 mm was measured from four rainfall events (Figure 5.2(a)). An intermediate water runoff (16.3 mm) was measured on the plots of the abandoned land where annual vegetation had been allowed to grow. In contrast, the lack of vegetation cover in the plots kept bare, or in the vineyard, favoured much greater volumes of runoff, with values of 22.6 mm and 21.0 mm, respectively, from the four rainfall events. The greatest runoff (30.3 mm) was generated from the bare soil containing rock fragments (cobbles), at a rate even higher than the bare, stonefree soil. Figure 5.2(b) shows how the total sediment loss varied according to land use after each of the four ponding rainfall events. The sediment loss was at a maximum (203.3 t km−2 ) on the soil containing rock fragments at the soil surface. In contrast, the abandonment of the olive grove for a long time, and thus the presence of annual vegetation and plant residues on the soil surface, was responsible for
62
Mediterranean Desertification
the drastic reduction of soil loss to negligible values (0.1 t km−2 ). Therefore, under olives grown like this, further degradation of the land is very restricted. Rock fragments on the soil surface appeared to play the most important role against erosion, especially during particularly heavy showers (i.e. heavier than those mentioned previously). A plant cover of 48% growing after abandonment of the land for 2.5 years reduced the total soil loss due to erosion by 35% as compared to the stone-free bare soil. Contour ploughing of the soils under vines also significantly decreased runoff and sediment loss. The absence of vegetation and the low aggregate stability (mean aggregate size equal to 0.6 mm) favoured surface sealing and increased runoff and sediment loss. 3.2
The Combined Effects of Land Use and Climate
The considerable variation in total runoff and sediment loss measured at various field sites along the northern Mediterranean reflect the great importance of total rainfall as well as land use on runoff generation and sediment loss, and therefore soil erosion. A number of runoff plots were installed at eight different sites (Figure 5.3), mainly formed on sedimentary rocks, i.e. schist, slates and phyllites, limestones, marls, sandstone-marls or unconsolidated alluvial deposits with slightly gravelly to gravelly, coarse to moderately fine-textured soils. In the following paragraphs, the effect of total rainfall on runoff and sediment loss is compared for the six study sites kept under the same land use.
Cereals Rainfed cereals, particularly wheat and barley, are widespread on the Mediterranean uplands. However, in some years the prevailing weather conditions during the growing period of these crops may be so adverse that the soils remain bare, creating favourable conditions for overland flow and erosion. Any loss of soil volume from these marginal lands greatly reduces the potential for biomass production, ultimately leading to desertification. Desertification at present threatens only the shallow and severely eroded soils. This threat, however, may expand to the majority of soils due to the adverse effects of global climatic change. Figure 5.4 indicates that the total annual runoff from the fields under rainfed cereals is positively related to the annual rainfall. It appears that runoff is a very small portion of the total rainfall (less than 1.5%) if the latter does not exceed 380 mm. However, amounts of rainfall greater than
Var Roussillon Petralona
Vale Formoso EI Ardal
Spata Santa Lucia
Rambla Honda
Figure 5.3 Location of the eight experimental field sites where erosion rates under various land uses were studied 1, Vale Formoso (Portugal); 2, El Ardal (Murcia, Spain); 3, Rambla Honda (Almeria, Spain); 4, Roussillon (Pyrenees, France); 5, Var (Pyrenees, France); 6, Santa Lucia (Sardinia, Italy); 7, Spata (Athens, Greece); 8, Petralona (Thessaloniki, Greece)
Effect of Land Use on Soil Erosion and Land Degradation Vale Formoso
(a) 300
El Ardal
63
Petralona
Y = −3.83 − 0.12∗X + 0.00056∗X 2
Runoff (mm year −1)
R = 0.82, n = 65 200
100
0 100
300
500
700
900
Rainfall (mm year −1) (b)
Portugal
Sediment loss (g m−2 year −1)
120
Spain
Greece
Y = −12.7 + 0.046∗X + 0.000083∗X 2 R = 0.60, n = 65
90
60
30
0 100
300
500
700
900
Rainfall (mm year −1)
Figure 5.4 (a) Rainwater runoff and (b) sediment loss versus total annual rainfall measured at three Mediterranean sites under rainfed wheat (Kosmas et al. 1997)
700 mm generated runoff volumes of up to 24% of the total precipitation. Most runoff events under Mediterranean conditions occur in the period from early October to late February. The rains falling during this period are of high intensity and long duration, and the soils cultivated with rainfed cereals are not sufficiently covered and protected from raindrop impact. As Figure 5.4 shows, in Mediterranean areas with a total precipitation of less than 280 mm, sediment loss is really not a threat. Sediment loss increases with increasing rainfall and may fluctuate between about 15 and 90 t km−2 year−1 for the range of 280–700 mm rain per year (Kosmas et al. 1997). Inbar (1992) reported a value of 20 t km−2 year−1 for the Catalunya area of Spain with an annual precipitation of 600–700 mm which is less than the values measured in wheat fields in wet years. Despite the wide variability existing, the obtained data show an increasing trend of sediment loss with increasing annual precipitation. The area cultivated with cereals around the northern Mediterranean is currently diminishing following a decline in the market prices for cereals, the rising cost of fertilizers and fuel, and the increased frequency of dry years. Most uplands with shallow soils have already been abandoned, and this abandonment seems likely to continue in the future.
64
Mediterranean Desertification
Vines The available data suggest that vine cultivation creates conditions for increased water runoff and sediment loss. This is because the soils cultivated with vines remain almost bare during autumn, winter and early spring due to the removal of annual vegetation by ploughing or the application of pesticides (to control weeds). Very high runoff rates have been measured in Spata (Greece) and Roussillon (France) with values of up to 31.8% of the annual precipitation, which reached 850 mm. The greatest runoff occurred typically in winter when the soil was wet and characterized by a low sorptivity and infiltration rate (Danalatos 1993). Infiltration was further diminished by the high compaction of the plough layer. It should be noted that the soils under vines are usually ploughed twice in mid-spring and are treated once or twice a year with herbicides. As the soils are very susceptible to dispersion, and rainfall intensities can be extremely high in the area (e.g. rainfall of 700 mm in one day was recorded at the Roussillon site in 1947, and 185 mm at Spata in 1994 in one day with a maximum intensity of 335 mm h−1 ), soil crusting occurs very often after ploughing, creating favourable conditions for overland flow and erosion. As with rainwater runoff, the greatest volumes of sediment loss were measured under vines, ranging from 67 to 460 t km−2 year−1 . These values greatly exceed those measured in fields cultivated with wheat. Data for sediment yields are not available from the Roussillon site. Further experiments and data are required in order to establish a clear trend of runoff and sediment loss in relation to annual rainfall along the whole northern Mediterranean.
Eucalyptus Eucalyptus cultivation, especially for pulp production, is very important, covering more than 500 000 ha in both Spain and Portugal and about 70 000 ha in Italy. Available data on soil erosion under eucalyptus plantations, which have been collected at Rio Santa Lucia (Sardinia), suggest that, as with vines, eucalyptus creates conditions for increased overland flow and erosion. Eucalyptus plantations are dense and dark and create adverse conditions for the growth of understorey annual or perennial vegetation so that the soil remains almost bare during the whole year. The total annual runoff under eucalyptus measured over a period of four years ranged from 0.6% to 8.2% of the annual precipitation, which varied from 171 mm to 564 mm. Figure 5.5 illustrates that the average sediment loss ranged from 1.4 to 65.6 t km−2 year−1 (SD = 1.2–46.8 t km−2 year−1 ) for the same precipitation range, demonstrating a serious erosion hazard for any soil reforested with eucalyptus as compared to the soils left under natural vegetation
Runoff (mm year −1)
Sediment loss (g m−2 year −1)
Runoff and sediment loss
70 60 50 40 30 20 10 0 171
453
473
564
Rainfall (mm year −1)
Figure 5.5 Runoff and sediment loss measured in hilly areas cultivated with eucalyptus at Rio Santa Lucia, Sardinia, over a four-year period (Kosmas et al. 1997)
Effect of Land Use on Soil Erosion and Land Degradation
65
(Aru and Barrocu 1993). These erosion rates are generally lower than those measured from soils under vines and generally higher than those measured under wheat. Runoff and sediment loss may be expected on any cultivated Mediterranean upland area, but especially where the soil is left bare for large parts of the year. If, in addition, heavy cultivation machinery is used, soil aggregate stability and organic matter content are decreased. This further increases the likelihood of soil erosion.
Olives Olive groves cover an appreciable part of the Mediterranean hilly areas. Where they grow as seminatural vegetation, annual vegetation and accumulating plant residues provide a high soil-surface cover, occasionally up to 90%, so preventing surface sealing and minimizing the velocity of the runoff water. Figure 5.6 shows that runoff in excess of 5% of the total rain, and sediment loss greater than 5.3 t km−2 year−1 never occurred under olive groves monitored for four years in southern Greece (Spata, Athens). Thus, the presence of annual vegetation and plant residues on the soil surface allows negligible soil loss and olives can play a big part in protecting Mediterranean uplands from further degradation and desertification. In fact, large areas around the Mediterranean region have been covered with olive trees since ancient times but many now grow untended as the prices obtained for olive oil have declined and made harvesting uneconomic in some places. Land-use planning should recognize the benefit of growing olives for the conservation of the soil before advocating alternative crops, such as eucalyptus, for only a short-term profit. Olives show a particularly high adaptation and resistance to long-term droughts and support a remarkable diversity of flora and fauna, greater than in some natural ecosystems (Margaris et al. 1995). The olive groves can be considered as a natural forest highly adapted to dry Mediterranean conditions, with lower vulnerability to fire than pine or eucalyptus forests, and protecting hilly areas from desertification in many ways. Much Mediterranean upland has been terraced for cultivating cereals, vines, olives and other crops (Figure 5.7). In many cases the stonewalled terraces are hundreds or even thousands of years old. Sometimes individual crescent-shaped terraces have been carefully constructed for individual trees. Soil was removed from other places to fill these terraces. This type of conservation management requires high labour costs to maintain the terraces in good repair. In the last few decades the value of such terraces for an agricultural return has markedly declined because of difficulties associated with poor accessibility and the limited use of labour-saving machinery. Many of these
Sediment loss (g m−2 year −1)
Runoff (mm year −1)
Runoff and sediment loss
10 8 6 4 2 0 349
453
508
575
Rainfall (mm year −1)
Figure 5.6 Runoff and sediment loss versus total annual precipitation measured between 1991 and 1994 at an olive grove under semi-natural conditions in Greece (Spata, Athens)
66
Mediterranean Desertification
Figure 5.7 Terraced olive grove well protected from erosion on the island of Lesvos
areas have been abandoned and if the stonewalling is allowed to collapse, removal of the retained soil by the runoff water can happen very quickly. Unfortunately maintaining such abandoned terraces appears a very expensive practice in labour terms compared to most other alternatives for soil erosion control.
Shrub Land Through the First and Second World Wars much upland was cleared of natural vegetation and cultivated mainly for cereal production to ensure sufficiency of cereal supply to the local populations. In the first years of cultivation, the production was fairly good, but very soon soil degradation reached serious levels, productivity started to decline dramatically and so did the local population and agriculture in those uplands. Land abandonment has continued until recent times, so that more areas are left under semi-natural vegetation. At least this is not accelerating soil erosion. Vegetation cover in the abandoned areas is variable and depends on the amount and distribution of rainfall. Figure 5.8 demonstrates the tendency of increasing overland flow with decreasing annual rainfall. Of course there is an inevitable variation between the different experimental sites, which is attributed to soil-surface properties, slope grade and length, and rainfall intensity and duration. Vegetation cover is crucial to runoff generation and alters throughout the Mediterranean uplands depending on climatic conditions and the period of the year. In areas such as southern Spain (Almeria, Murcia), with an annual precipitation lower than 280 mm year−1 and high evapotranspiration rates, the soil water available to plants is drastically reduced and therefore the soil remains relatively bare, favouring overland flow. Runoff reached values up to 10% of the total rainfall at the rather dry Almeria site. The available data show a peak of runoff with an annual precipitation total of 280–300 mm. The relationship between annual sediment loss and precipitation shows a trend of increasing loss with decreasing precipitation as long as the latter exceeds 280–300 mm year−1 . If annual precipitation falls below this range, then erosion decreases with increasing aridity. Inbar (1992) reported similar trends for different watersheds in the coastal area of Israel. Data for sediment loss from the wetter experimental sites of Petralona (northern Greece) and Rio Santa Lucia (Sardinia), having an average annual precipitation of 464 and 448 mm, respectively, showed the lowest values, ranging from 13.8 to 0.5 t km−2 year−1 . Sediment loss increases if one moves from areas of higher precipitation to areas of lower precipitation (such as Murcia and Almeria, southern Spain). The maximum value of sediment loss was recorded in Almeria (21.5 t km−2 year−1 ), associated with an annual precipitation of 282 mm. Under drier climatic conditions, sediment loss is greatly reduced to values similar to those measured in the relatively wetter sites.
Effect of Land Use on Soil Erosion and Land Degradation
67
20
Runoff (mm year −1)
n = 18
15
10
n = 21
5
n =8
n = 12
0
n = 17
100−200 200−300 300−400 400−500 500−600 Rainfall (mm year −1)
Sediment loss (g m−2 year −1)
20
15
10
5
0
100−200 200−300 300−400 400−500 500−600 Rainfall (mm year −1)
Figure 5.8 Annual runoff and sediment loss versus rainfall measured in shrub lands at various experimental sites across the northern Mediterranean region (Kosmas et al. 1997)
3.3 Land Abandonment
There is a growing interest in the evolution of abandoned dry lands (Gordon et al. 1981; RuizFlano et al. 1992), which are now marginal from an environmental and socio-economic point of view (Esteve et al. 1993). The abandoned fields may show quite different evolutions depending on various environmental and land-use features. Some of these, especially soil type, water availability and the type of previous and post-abandonment land use, could play a more important role in some places than others. A wide variety of situations are generated, so that it is difficult to predict future evolution. The evolution of vegetation types depending on age of abandonment shows clear tendencies: the predominance of annuals in the fallow land and the field abandoned for five years, and a progressive decrease of annuals until they barely appear by the time scrub lands have developed. Shrubs and herbaceous perennials show the opposite behaviour, though in a less pronounced way and with
68
Mediterranean Desertification
different time-scales. It can take more than 20 years for shrub lands with a high percentage of ground cover to develop. Martinez-Fernandez et al. (1996) studied the effects of land abandonment in eight abandoned fields in Murcia (Spain) with 1–30 years since abandonment, N–S exposure, similar present-day conditions and limestone substratum. The pedological property most related to vegetation dynamics was found to be the organic matter content. This factor also has important implications in soil degradation processes. Organic matter content shows a clear relationship with age of abandonment, which confirms a tendency already pointed out in previous studies (Martinez-Fernandez et al. 1994). The agricultural use of these fields led to a loss of organic matter content, being usually less than 1% under actual crops close to these abandoned fields. The results show that, after the abandonment of agricultural practices, an evident recovery of this factor may be detected, even in the early stages. The organic matter content gradually reverts to the situation before agricultural use. Recovery is helped if the abandoned field has a northern exposure, which loses soil moisture less readily than a field with a southern exposure. The post-abandonment uses of the abandoned fields have major importance in their evolution. Moderate grazing has a minor effect on the partial rejuvenation of the communities, visible through the maintaining of intermediate successional stages and high diversity index. These results may be of interest for all Mediterranean semi-arid areas showing similar environmental conditions, in which the abandonment practices are relevant. This is also true of sediment production. Generally, the greatest soil losses occur during the months with the highest rainfall, especially those with high hourly intensities, such as happen frequently in the month of October. However, in other months with significant rainfall, such as April and November 1984, November and December 1985, and January and April 1986, there was hardly any sediment production. Such observations confirm that the degree of correlation between rainfall, runoff and sediment production is low in semi-arid south-east Spain (Romero D´ıaz et al. 1998). Other controls are still only partly understood (Fisher et al. 1985), such as annual variations in the content and retention of soil moisture, biomass production, the incorporation of organic material, the quantities of soluble anions and cations, conductivity, etc. All these factors play an important role in the complex relations between rainfall, runoff and sediment production.
REFERENCES Alias-Perez LJ and Ortiz-Silla R (1986) Proceedings of XIV National Soils Meeting. Spanish Sciences Society of the Soil (CSIC), University of Murcia, Spain. Aru A and Barrocu G (1993) The Rio Santa Lucia catchment area. In Mediterranean Desertification and Land Use, MEDALUS Final Report. Commission of the European Communities. Contract number EPOC-CT900014-(SMA), pp. 533–559. Barbera GG, L´opez-Berm´udez F and Romero D´ıaz MA (1997) Cambios de uso del suelo y desertificac en el Mediterraneo: El caso del Sureste Iberico. In JM Garcia-Ruiz and P Lopez Garcia (eds) Accion humana y desertification en ambientes semiaridos. Instituto Pirenaico de Ecologia, Zaragoza, pp. 9–39. Benito G, Gutierrez M and Sancho C (1992) Erosion rates in Badland areas of the Central Ebro Basin (NESpain). Catena 19, 269–286. Bryan RB and Campbell IA (1986) Runoff and sediment discharge in a semi-arid drainage basin. Zeitschrift f¨ur Geomorphologic 58, 121–143. Casenave A and Valentin C (1992) A runoff capability classification system based on surface features criteria in semi-arid areas of West Africa. Journal of Hydrology 130, 231–249. Danalatos NG (1993) Quantified analysis of selected land use systems in the Larissa region, Greece. PhD thesis, Agricultural University of Wageningen, Wageningen. Douglas I (1969) Sediment yields from forested and agricultural lands. Proceedings of the Symposium on The Role of Water in Agriculture. University of Wales (Aberystwyth) Memorandum No. 12, E1–E22. Elwell HA and Stocking MA (1976) Vegetal cover to estimate soil erosion hazard in Rhodesia. Geoderma 15, 61–70. Esteve MA, Calvo F, Ibernon M, Gimenez A, Palazon JA and Ramirez-Diaz L (1993) Tierras marginales en ecosistemas semiaridos del Sureste Iberico: descriptores, relacion con los factores fisicos y aplicaciones a la gestion ambiental. Problematica Geoambiental y Desarrollo. V Reunion Nacional de Geologia Ambiental y Ordenacion del Territorio, Murcia, pp. 777–786.
Effect of Land Use on Soil Erosion and Land Degradation
69
Faulkner H (1990) Vegetation cover density variations and infiltration patterns on piped alkali sodic soils: implications for the modelling of overland flow in semi-arid areas. In JB Thornes (ed.) Vegetation and Erosion, Processes and Environments. John Wiley, Chichester, pp. 317–346. Fisher GC, Romero D´ıaz A, L´opez-Berm´udez F, Thornes JB and Francis F (1985) Vegetation litter production and effects in an eroding Mediterranean ecosystem, Mula, SE Spain. IX Coloquio de Geografos Espanoles, Murcia, Poniense, V2. Francis CF and Thornes JB (1990) Runoff hydrographs from three Mediterranean vegetation cover types. In JB Thornes (ed.) Vegetation and Erosion, Processes and Environments. John Wiley, Chichester, pp. 363–384. Fullen MA (1985) Soil compaction, hydrological processes and soil erosion on loamy sands in East Shropshire, England. Soil and Tillage Research 29(6), 17–29. Gordon M, Guillerm JL, Poissonet J, Poissonet M, Thiault M and Trabaud L (1981) Dynamics and management of vegetation. In FDi Castri, DW Goodal and R Specht (eds) Mediterranean-type Scrublands. Elsevier, Amsterdam, pp. 317–344. Hillel D (1991) Deforesting the earth. In D Hillel (ed.) Out of the Earth, Civilization and the Life of the Soil . University of California Press, Berkeley and Los Angeles, pp. 175–185. Imeson A (1984) An eco-geomorphological approach to the soil degradation and erosion problem. In R Fantechi and NS Margaris (eds) Desertification in Europe. Proceedings of the Information Symposium in the EEC Programme on Climatology. Reidel, Dordrecht, pp. 153–168. Inbar M (1992) Rates of fluvial erosion in basins with a Mediterranean type climate. Catena 19, 393–409. Kosmas C, Danalatos N, Cammeraat LH, Chabart M, Diamantopoulos J et al. (1997) The effect of land use on runoff and soil erosion rates under Mediterranean conditions. Catena 29, 45–59. Langbein WB and Schumm SA (1958) Yield of sediment in relation to mean annual precipitation. American Geophysical Union Transactions 39, 1076–1084. Lee CR and Skogerboe JG (1985) Quantification of erosion control by vegetation on problem soils. In Al Swaify, WC Moldenhauer and A Lo (eds) Soil Erosion and Conservation. Soil Conservation Society of America, Ankeny, IA, pp. 437–444. L´opez-Berm´udez F (1990) Soil erosion by water on the desertification of a semi-arid Mediterranean fluvial basin: the Segura basin, Spain. Agriculture, Ecosystems and Environment , 33(2), 129–145. L´opez-Berm´udez F, Thornes JB, Fisher G and Francis C (1984) Erosion y Ecologia en la Espana semiarida (Cuenca de Mula, Murcia). Cuadernos de Investigacion Geografica 10(1–2), 113–126. L´opez-Berm´udez F, Romero D´ıaz MA and Martinez-Fernandez J (1991) Soil erosion in semi-arid Mediterranean environment. El Ardal experimental field (Murcia, Spain). In M Sala, JL Rubio and JM Garcia-Ruiz (eds) Soil Erosion Studies in Spain. Geoforma Ediciones, Logrono, pp. 137–152. L´opez-Berm´udez F, Sancez-Fuster MC and Romero D´ıaz A (1995) Incidencia de los modelos de gestion socioeconomica (siglos XIX y XX) en la degradacion del suelo en el Campo de Lorca (Cuenca del Guadalentin, Murcia). Papeles de Geografia 22, 5–18. Universidad de Murcia. Margaris N, Koukoutsidou E, Giourga Ch, Loumou A, Theodorakis M and Hatzitheodoridis P (1995) Managing desertification. In MEDALUS II Project 3, Managing Desertification, EV5V-CT92-0165, pp. 83–110. Martinez-Fernandez J, Romero D´ıaz MA, L´opez-Berm´udez F and Martinez-Fernandez J (1994) Parametros estructurales y funcionales de Rosmarinus officinalis en ecosistemas mediterraneos semiaridos. Studia Oecologica, 10–11, 309–316. Martinez-Fernandez J, Romero D´ıaz MA and Belmonde-Serrato F (1996) Evolution of vegetation and pedological characteristics in fields with different age of abandonment: a case study in Murcia (Spain). In JL Rubio and A Calvo (eds) Soil Degradation and Desertification in Mediterranean Environments. Geoforma Ediciones, Logrono, pp. 279–290. Martin-Vivaldi MC and Jimenez Olivencia Y (1992) Estudio de la erosion en la cuenca del Rio Trevelez (Granada). In F L´opez-Berm´udez C Conesa-Garcia and A Romero D´ıaz (eds) Estudios de Geomorfologia en Espana. Sociedad Espanola de Geomorfologia, Murcia, pp. 93–103. Mitchell DJ (1990) The use of vegetation and land use parameters in modelling catchment sediment yields. In JB Thornes (ed.) Vegetation and Erosion, Processes and Environments. John Wiley, Chichester, pp. 289–314. Morin J and Benyamini Y (1977) Rainfall infiltration into bare soils. Water Resources Research 13, 813–817. Newson MD (1985) Forestry and water on the uplands of Britain – the background of hydrological research and options for harmonious land use. Journal of Forestry 79, 113–120. Patton PC and Schumm SA (1975) Gully erosion, North-western Colorado: a threshold phenomenon. Geology 3, 83–90. Perez-Trejo F (1992) Desertification and Land Degradation in the European Mediterranean. European Commission, EPOCH programme, Directorate General XII, Science, Research and Development, EUR 14 850.
70
Mediterranean Desertification
Reed LA (1971) Hydrological and sedimentation of Corey Creek and Elk Run Basins, North-Central Pennsylvania. US Geological Survey Water Supply Paper. Rodriguez-Aizpeolea J, Perez-Badia R and Cerda-Bolinches A (1994) Colonizacion vegetal y produccion de escorrentia en bancales abandonatos: Vall de Gallinera. Alacant. Cuaternario y Geomorfologia. Romero D´ıaz MA, L´opez-Berm´udez F, Thornes JB, Francis CF and Fisher GC (1988) Variability of overland flow erosion rates in a semi-arid Mediterranean environment under matorral cover in Murcia Spain. Catena, supplement 13, 1–11. Romero D´ıaz A, Barbera GG and L´opez-Berm´udez F (1995) Relaciones entre erosion del suelo, precipitacion y cubierta vegetal en un medio semiarido del sureste de la peninsula iberica. Lurralde 18, 229–243. Romero D´ıaz A, L´opez-Berm´udez F and Belmonte Serrato F (1998) Erosion y escorrentia en el Campo Experimental de “El Ardal” (Murcia). Nueve anos de experencias. Papeles de Geografia 26, 129–144 Ruiz-Flano P, Garcia-Ruiz JM and Ortigosa L (1992) Geomorphological evolution of abandoned fields. A case study in the Central Pyrenees. Catena 19, 301–308. USDA (1938) Soils and Men, 1938 Yearbook of Agriculture. USDA, Washington, DC. Williams KF and Reed LA (1972) Appraisal of stream sedimentation in the Susquehanna River basin. US Geological Survey Water Supply Paper. Wise SM (1982) How old are the badlands? A case study from south-east Spain. In R Bryan and A Yair (eds) Badland Geomorphology and Piping. GeoBooks, Norwich, pp. 259–277. Yair A, Goldberg P and Brimer R (1982) Long term denutation rates in the Zin-Havarim badlands of northern Negev, Israel. In R Bryan and A Yair (eds) Badland Geomorphology and Piping. GeoBooks, Norwich, pp. 279–291.
6
Agro-pastoral Activities and Land Degradation in Mediterranean Areas: Case Study of Sardinia
G. ENNE,1 G. PULINA,1 M. D’ANGELO,1 F. PREVITALI,2 S. MADRAU,1 S. CAREDDA1 AND A.H.D. FRANCESCONI1 1
Dipartimento di Scienze Zootecniche, University of Sassari, Sassari, Italy Dipartimento di Scienze dell’Ambiente e del Territorio, Universita` di Milano–Biocca, Milano, Italy 2
1 INTRODUCTION Livestock farming is one of the main agricultural activities in the Mediterranean Basin, both in terms of the numbers of people employed and in its distribution throughout the region. There are about 100 million livestock units (LSU) of herbivores in the countries of the Mediterranean Basin, 53.8% of which are in Europe, 23.2% in Africa and 23% in Asia (FAO 1995). Most LSU are ruminants (cattle, sheep and goats) and their main feeding source is natural or cultivated pastures grazed directly. The international scientific community has recognized that agro-pastoral activities are clearly one of the main causes of land degradation in the Mediterranean. In southern Spain, Greece and Portugal, wide areas intensively exploited by small ruminants have already reached a severe level of land degradation. The spread of agro-pastoral activities in most Mediterranean countries and the increased grazing pressure are also related to past European Union (EU) policies, which favoured the uncontrolled development of modern agricultural practices. Those policies provided a system of guaranteed prices and subsidies to farmers for the production of meat and wheat which resulted in the cultivation of marginal areas. In addition, the Common Agricultural Policy (CAP) led to a steep rise in productivity by encouraging mechanization. As has been shown by several studies, this production-orientated model protected farmers against the economic consequences of environmental degradation and also removed their responsibility toward environment management (Buller 1992). Fortunately, since the beginning of 1992, all State Members of the EU have focused their attention on a new concept of environmental sustainability, as laid out by the Treaty of Maastricht. The EU concerns about the rural environment, land abandonment and its consequent degradation have led to it supporting environmentally friendly practices. For example, Council Regulation Number 2078/92 concerns aid to encourage agricultural production methods with a low impact on the environment, while Number 2080/92 concerns aid to encourage afforestation. Furthermore, in order to give scientific assistance to the political decisions on land degradation issues, the EU has promoted several research programmes, including MEDALUS, aimed at understanding the major causes of land degradation, and developing schemes to mitigate and prevent land degradation.
2 SARDINIA: AN ISLAND THREATENED BY LAND DEGRADATION Sardinia is one of the Italian regions most threatened by land degradation. Although it is a relatively low-lying region (the highest peak being Punta la Marmora, 1834 m a.s.l.), which does not exceed Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
72
Mediterranean Desertification
the upper limit of vegetation, the unproductive lands, excluding urban and coastal areas and inland waters, represent about 12% of the total area. These are distributed over the whole island as a result of the land use through the centuries, which has always been agro-pastoral to a great extent. Today, about 85% of Sardinian land is used for agriculture (ISTAT 1976, 1982, 1992), and livestock farming is one of the main economic activities. There are about 622 835 LSU of herbivores, of which 50% are dairy sheep (Table 6.1). This fact has greatly influenced land use in Sardinia: meadows and pastures are intensively grazed, and both wooded areas and arable land are cultivated to provide forage and other animal feeding sources. Agro-pastoral activities are a major cause of fires. A detailed analysis of the causes of forest fires shows that more than 90% of the total number of fires are started deliberately, and are historically and traditionally related to human activities (Figure 6.1). Fire has been considered an important practical and economical tool for clearing lands for grazing. Land fragmentation and the heterogeneity of land cover, typical of Mediterranean environments, have in many cases favoured fire propagation from grasslands to shrublands and wooded areas, thus compromising forested ecosystems (Figure 6.2). Although in the Mediterranean Basin fire has always been present in the ecosystem, promoted by hot dry periods common in the Mediterranean climate and the particularly inflammable characteristics of typical Mediterranean vegetation (Molina 1996), in the last 50 years its occurrence has dramatically increased, and is now a major factor of desertification. The destruction of the vegetation cover and the effects on the underlying soil (Chandler et al. 1983) result in an increased erosion risk. In Table 6.1 Grassland, livestock and stocking rate evolution (1971–1991)
Year
1971 1981 199l a
Agricultural land (ha)
2 159 245 2 047 811 2 050 731
Grassland (ha)
1 613 279 1 497 503 1 539 224
Livestock number (LSU)a Cattle
Sheep
Goats
Total
273 050 287 798 286 840
215 323 226 714 313 129
25 070 22 463 22 867
513 443 536 875 622 835
LSU = 450 kg live weight (1 cattle; 10 sheep; 10 goats).
Figure 6.1 Typical aspect of an over-exploited pasture in Sardinia. Overgrazing and the frequent use of fires are among the main causes of land degradation
73
Agro-pastoral Activities and Land Degradation in Sardinia
Figure 6.2
Erosive processes in an area recently affected by fire
Table 6.2 Number of fires and areas affected by fires in Sardinia (1982–1996)
Year No. of fires Wooded area (ha) Pasture (ha) Other areas (ha) Total area (ha)
1982–1986
1987–1991
1992–1996
Annual average (1982–1996)
14 406 64 720 217 621 17 299 299 640
16 111 35 694 150 246 12 562 198 502
16 438 48 537 128 324 14 429 191 290
3 130 9 930 33 079 2 953 45 962
Source: RAS (1996). Sardinia, during the period 1982–1996, an area of 689 432 ha of land was swept by fire (Table 6.2). The mean annual area affected by fire amounts to about 46 000 ha, about 72% of which is pasture. With particular reference to woodlands, during the period 1989–1993 about 1.6% of the total area was annually swept by fire; when compared to the European Mediterranean average (1%) (EEC 1996), this datum shows the significant incidence of this phenomenon in Sardinia. These are the main reasons why Sardinia has been considered a representative study area for the impact of grazing systems on desertification processes. On the island there are two main forage systems: agro-silvo-pastoral activities in hilly and mountainous areas and cereal–dairy sheep farming on the plains and low hills. Between these two systems there are various intermediate conditions. The traditional agro-pastoral system, based on pasture with or without fertilization and with shortterm forage crops, is very common in hilly and mountainous areas of Sardinia. This system makes it very difficult for forage availability and animal feed requirements to coincide (Caredda et al. 1992). The green forage production is mainly concentrated in spring, while the maximum dairy sheep feed requirements are concentrated in autumn–winter, at the end of ewe pregnancies and at the beginning of lactation. Generally, this problem is solved either by vertical transhumance or, more often, by feeding sheep with hay and concentrates. During the 1950s and 1960s, a reduction of the traditional rotation of cereal-grazed fallow brought about the abandonment of arable lands and different land-use management. As a consequence, the
74
Mediterranean Desertification
Figure 6.3 Maquis clearing in favour of cultivated pastures can result in severe surface erosion when carried out on unsuitable lands
balance between crop and pasture was altered, leading to an increased abundance of low-growing palatable species on the abandoned land and a reduction in forage availability. Unplanned and irrational grazing patterns (autumn overgrazing and/or spring undergrazing) caused a progressive reduction of palatable species and an increase in bare soil areas, which generated soil losses, particularly on slopes. During the 1970s and 1980s an increase in sheep milk prices led to a large increase in sheep numbers in lowlands and hilly areas. In order to increase forage availability on natural pastures, farmers utilized different agronomic techniques which, particularly on slopes, tended to lead to erosion. For example, inadequate machines (e.g. scrapers) were used to clear the ground of stony and shrubby vegetative cover and trees, and removed the topsoil irreversibly. Generally, forage intensification is based on cultivation of short-term forage crops that are grazed in winter and harvested as hay or grain in late spring. However, because of the lack of sufficient flat surfaces, forage crops are also cultivated on slopes, often by ploughing across rather than parallel to the contours, and without adopting soil conservation practices. As a consequence, gullying due to runoff after autumn rainfalls is frequently observed in hilly regions (Figure 6.3). Indeed, the risk of soil erosion on hillslope areas is concentrated at the end of the summer and autumn, when frequent and intense rains coincide with the soil being recently ploughed or bare, and very vulnerable to the strength of rainwater (Rivoira et al. 1989; Roggero et al. 1995). Even though agro-pastoralism, especially overgrazing, has often been considered as the main cause of land degradation, Seligman (1996) maintains that “among all the factors that contribute to landscape degradation in the Mediterranean Basin, high stocking rates must be placed low on the list”. In order to contribute to the clarification of this matter, this chapter mainly deals with the general effects of agro-pastoral activities on vegetation and soil degradation, and presents some experimental results of the MEDALUS II research carried out in Sardinia on this topic. Finally, practical implications of irrational grazing practices on soil fertility are examined and some guidelines on proper land management in an agro-pastoral context are proposed.
3
EFFECTS OF GRAZING ON VEGETATION AND SOIL
In agro-pastoral systems, animals, plants and environment interact with each other in a complex manner, and are directly affected by human activities. The common belief that grazing is always
Agro-pastoral Activities and Land Degradation in Sardinia
75
detrimental to vegetation is a result of the confusion between grazing and overgrazing, with only the latter being destructive to plants and soil. Mathematical models have been devised to aid the understanding of the complex interactions of the diverse components that make up the soil–animal–plant interaction system (Doucet and Sloep 1992). Rational management should encourage certain positive influences of grazing on vegetation and soil resources. Indeed, grazing delays maturation of the vegetation (Vallentine 1990), keeping plants in a vegetative, forage-producing state. Grazing also stimulates growth and regrowth by its pruning effect, maintaining optimum leaf area index (LAI = total leaf area per unit ground area), improving the nutritive value of available forage, and reducing excessive accumulations of standing dead vegetation and mulch. This reduces the vegetation biomass which, if allowed to build up, provides the undergrowth that favours outbreaks of fires and their propagation during the hot dry periods of summer. On the other hand, other common agro-pastoral activities can have negative effects. Soil is easily compacted due to trampling by livestock, and the soil can become bare due to overgrazing or ploughing. If fire is used to destroy vegetation to clear land for pasture, valuable organic matter is also lost from the soil. An understanding of the effects of grazing on vegetation and soil is fundamental for the development of a rational grazing management strategy within a more sustainable agro-pastoral context. 3.1 Grazing Effects on Vegetation
Grazing involves biting, pulling and breaking off plant parts, which causes defoliation, or even pulling entire plants out of the ground, if they are not well rooted. Furthermore the trampling and treading of the vegetation may damage the stand. Seed dispersal, internally through the animal digestive system, or externally by temporary attachment to animal hair, fleece or hooves, is an ecological factor affecting a perennial forage stand, but the impact will range from favourable to unfavourable, depending upon the plant species and site being affected (Vallentine 1990). Covering some parts of the vegetation with faeces and urine is another effect of grazing. Manure spots are generally avoided by animals visiting later, even though the nutritive quality of the affected forage (particularly rich in nitrogen) may be better than that on adjacent ground. Rejection is presumably on grounds of palatability based on smell or taste, and perhaps designed to avoid recycling internal parasites (Van Soest 1994). Measurement of the leaf area index (LAI) is a very useful indicator of how forage responds to grazing. Undergrazing allows overgrowth and shading by senescent foliage, which reduces photosynthesis and increases respiration. Optimum grazing pressure improves the effective LAI, whereas higher pressures, which result in excessive defoliation and a related decrease in forage yield, diminish it (Van Soest 1994). In addition to the reduction in leaf area per plant by grazing, the thinning out of grass species, which is a consequence of the selective feeding action of the animals or of the exposure of roots to the cutting edge of the hooves, can have a strong impact on final LAI and quality of forage. Plant recovery from defoliation depends not only on the available carbohydrate reserves but also on the quantity of the remaining foliage and its photosynthetic capacity. The rate of development of new foliage and photosynthetic capacity of new leaves is also important (Caldwell 1984). Generally, the lower the level of reserve carbohydrates, the more important the remaining leaf area is in promoting regrowth. Also, while perennial forage plants are influenced by the conditions in current and preceding years, which affect their root reserves and spring regrowth, annual forage plants are not. The supply of nutrients needed for the regrowth of annual forage plants after defoliation is primarily dependent on the remaining leaf area rather than storage compounds (Vallentine 1990). The optimum rational grazing management strategy should allow the maximum level of defoliation that will still maintain sustainable forage production and animal response. The definition of the optimum moment for grazing, the optimum frequency and duration of grazing, and the intensity of defoliation, by using a proper stocking rate, are of great importance (Brandano and Rossi 1975).
76 3.2
Mediterranean Desertification Grazing Effects on Soil
The negative effects of grazing on the soil are felt directly, through trampling, and indirectly, through the reduction of vegetation cover and removal of organic matter from the soil (Pulina et al. 1995a). The treading of soil by grazing animals may be detrimental, causing soil compaction, surface horizon disruption, reduction of infiltration, creation of terracing on steep slopes, development of animal trails, and thus erosion (Vallentine 1990). The degree of impact the animal treading has on a specific site depends on the interaction between vegetation, soil, weather and animals. Soil compaction by hooves causes a reduction in soil porosity which reduces water infiltration and percolation in the soil, leading to increased water runoff and erosion on sloping terrain, and a tendency to hydromorphism or to stagnation on flat terrain. Soil compaction depends both on the characteristics of the animals’ behaviour, such as their tendency to walk, run or jump, or to graze in groups, and on agro-pastoral interventions, such as the presence of concentration areas (shade of trees, areas protected from predominant winds, drinking places and artificial feeding places, etc.). Compaction depends not only on the stocking rate but also on the specific pressure of the hooves per square centimetre. For instance, a calculation based on hoof area and body weight of various animals has estimated an average pressure per unit area of 0.47 kg cm−2 for sheep, 0.98 kg cm−2 for cattle and 1.01 kg cm−2 for donkeys (Pulina et al. 1995a). The destruction of the soil surface by penetration of hooves is more likely to occur when soils are wet, where there are clay-textured soils and where there is poor vegetation cover. Other factors that may accentuate damage to soil properties include allowing grazing in the wet winter months, high stocking rates at any time, or a preponderance of cattle rather than lighter animals. The creation of terracing on sloping terrain, and of trails on flatter terrain, is a result of the routes chosen by animals while grazing and being transferred from one pasture to another. Trails, which become areas of bare soil, are created in direct proportion to stocking rates and in inverse proportion to the availability of forage. At times these trails may become a high proportion of the total pasture area, especially at waiting points near gates and in areas with a high movement of animals, such as near drinking places. These areas may suffer from significant wind erosion during the dry season. Overgrazing may remove part of the vegetation cover which, in its turn, brings about an increase in raindrop impact and surface soil crusting, and a decrease in organic soil matter, aggregate stability, and water infiltration rates (Blackburn 1983, 1984). All these effects may cause increased water runoff, reduced soil water content, and increased erosion. Organic matter in the soil is an important component of soil fertility and essential for the maintenance of good soil structure, which can counteract the erosive action of water and wind. The removal of organic matter by animals is due to an imbalance between the amount of dry matter they consume and the dry matter that returns to the soil in the form of faeces and urine (Pulina et al. 1995b). It is not easy to estimate the quantity of organic matter actually returned and incorporated into the soil by animals. Organic matter restitution is efficient only during wet seasons, when faeces are easily incorporated into the soil. At this time it is soft as a consequence of the animals’ intake of fresh grass with a high moisture content. On the other hand, in dry seasons, the faeces are much drier as a result of feeding on dry stubble and some internal body defence mechanisms that protect animals from wasting water. Dry faeces can remain on the soil surface for months and are likely to be completely oxidized.
4
THE CASE STUDY OF SARDINIA
In order to evaluate the effect of agro-pastoral activities on land degradation in Sardinia, a series of laboratory and field experiments were conducted under the aegis of the MEDALUS II project. A two-year experiment (1994–1995) was carried out on the main factors influencing cattle grazing behaviour. Initially the experimental site in north-western Sardinia at the Astimini-Fiume Santo basin (Figure 6.4), a region characterized by semi-arid climate (mean annual rainfall is 544 mm over a 39-year period), was overstocked (20 heifers of Limousine × Bruno Sarda cross occupying 10 ha with an initial stocking rate of 450 kg ha−1 ) (Enne et al. 1996). The average slope in this hilly area
77
Agro-pastoral Activities and Land Degradation in Sardinia Experimental area Stintino E A
Porto torres
Sardinia (Italy)
H
B
G I
Astimini-fiume santo basin
km 0
F
C
D
Experimental area 5
10 Legend Enclosure limits Sub-catchment and sub-areas limits
Subarea Surface (ha) A B C D E F G H I
Figure 6.4
0.40 1.87 1.02 1.49 0.35 0.84 1.46 0.69 1.08
Variables Aspect N NE N N W W NW N NW
Slope
Land cover
Soil
<30 Maquis Shallow soil 30−35 Sparse shrubs Shallow soil 30−35 Natural sward Shallow soil but eroded 35−40 Natural sward Shallow soil <30 Natural sward Shallow soil <30 Sparse shrubs Shallow soil >40 Natural sward Shallow soil <30 Sparse shrubs Shallow soil >40 Maquis − rock outcrop Rock outcrops
Location and main characteristics of the study area (Sardinia, Italy)
was greater than 35% and height above sea level ranged from 235 to 379 m. Soils can be classified as dystric, lithic, lithic ruptic–xerorthentic xerochrepts and lithic xerorthents (Soil Survey Staff 1996) over Paleozoic metamorphic substrates. Grazing behaviour was assumed not to be influenced by forage quality and availability because the vegetation was composed of a natural sward with mainly homogeneous species and ground cover, except for a small part of the area that was covered with maquis and scattered shrubs. Animals were given 4.7 kg DM hay daily (65% of their voluntary intake) throughout the experimental period, to compensate for the high stocking rate. Following a photo-interpretation method of analysis, the experimental area was theoretically divided into nine sub-areas homogeneous for topographical features, land-cover typologies and pedological characteristics. Site visitation by animals was influenced by the location of an artificial feeding site and of a resting area, time of day and season, local meteorological conditions, and topographical characteristics. Very early in the morning animals moved directly from the resting area, which was located at the upper part of the enclosure, downhill to the artificial feeding site, where they were afterwards found grazing. Then they moved around in regular patterns, shifting gradually to higher sub-areas in the afternoon and evening (Figure 6.5). During the experimental period, the animals tended to be distributed more homogeneously over the area, probably due to the effects of overstocking, there being too many of them for them all to find food in one place. In dry seasons animals tended to visit the sub-areas closest to the foraging site more frequently, while in wet seasons the distribution pattern was more homogeneous, and animals also tended to move and graze at higher elevations. On windy days, especially when winds blew from
78
Mediterranean Desertification 2
2
Legend Morning Afternoon Evening
250
250
1 2
300
300 1
Resting area Feeding area
1
350
350 (a) Wet season
0
(b) Dry season
100
200 m
Figure 6.5 Topographic map showing seasonal movement patterns of cattle in the experimental area Table 6.3 Average cumulative trampling pressure (kg cm−2 ) for each sub-area
Sub-area
A B C D E F G H
Period
Total
J–F
M–A
M–J
N–D
J–F
M–A
0.00 0.00 5.52 3.03 21.06 2.46 2.71 0.90
1.05 4.71 1.87 1.22 34.38 3.45 0.00 0.06
1.32 8.21 0.00 1.40 39.20 3.40 0.00 0.06
2.31 1.77 5.09 0.21 50.26 4.43 3.05 2.37
0.16 3.70 4.38 4.43 34.97 4.86 3.22 2.91
0.00 5.52 4.35 2.69 31.73 10.74 5.48 4.90
4.85 23.91 21.21 12.97 211.61 29.34 14.46 11.20
the north-west, animals preferred the most protected and leeward areas. The most visited sub-area in all seasons was the one nearest to the foraging area (E), followed by a bordering area (B) that was protected against strong winds. The least visited sub-areas were one mainly covered with maquis (A) and another that was relatively inaccessible and more exposed to strong winds (H). Even under high stocking rates, animals preferred some sites, which had a cumulative trampling pressure as high as 211.61 kg cm−2 , while other sites were visited much less frequently, with a cumulative trampling pressure as low as 4.85 kg cm−2 (Table 6.3). This part of the experiment showed that the traditional stocking rate (LSU ha−1 ) cannot be considered as a grazing pressure index without taking animal behaviour into account. In order to determine the effects of stock trampling on soils, samples of soil were taken from the grazed area of the above-described experiment and from an ungrazed control area (Previtali 1996). The surface horizons of the ungrazed soils showed an average bulk density of 1.0 g cm−3 (wellstructured medium-textured soil, with a medium to high organic matter content) and an average penetration resistance (soil strength) of 2.5 kg cm−2 (low penetration resistance). On the other hand, the surface horizons of the grazed soils showed some structural modification as shown by higher average bulk density (1.6 g cm−3 ) and higher average penetration resistance (4 kg cm−2 , moderate penetration resistance). Some micromorphological soil alterations were found after a one-year grazing period (Table 6.4), especially in the structural porosity (SP) of the soil which varied from 28% to 12% in the surface A horizon, and from 20% to 10% in the subsurface B horizon, in the
79
Agro-pastoral Activities and Land Degradation in Sardinia Table 6.4 Micromorphological characteristics in the grazed and ungrazed areas
Horizon
A horizon B horizon
Ungrazed area
Grazed area
Structural porosity (SP) %
Textural porosity (TP) %
Structural porosity (SP) %
Textural porosity (TP) %
28.4 19.7
27.9 18.2
11.7 10.0
33.1 29.9
ungrazed and grazed areas, respectively. These values show that cattle trampling influenced not only the surface horizons, but also the structural characteristics of the subsurface horizons. Analysis of the hydrological state of the soil in the basin studied highlighted that restoring normal soil water storage did not take place before November (Madrau et al. 1995). From November until March, the period in which rainfall in this semi-arid climate is more concentrated, the average soil moisture content on some days crossed the modified threshold of plasticity limit of 31%. From springtime onwards, the soil moisture dropped below such a critical value. Thus, winter is the season during which the risk of soil degradation by animal trampling is highest in this climatic and pedological context. As part of the large-scale experiment described above, the effect of continuous high stocking grazing was compared with the situation in ungrazed areas. Pastoral value and forage production were studied in this natural pasture region of north-western Sardinia for two years (1993/94 and 1994/95) (Caredda et al. 1996). The natural pasture of the experimental basin, which was representative of many other Sardinian marginal areas, was characterized by low productive potential, because of the steep slopes, superficial rockiness and limited soil depth. The use of a high and continuous stocking rate caused an increase in the area of bare soil surface particularly in the second year (10% higher than in the ungrazed area), thus increasing the risk of soil erosion. Grazing influenced the floristic composition of the vegetation by reducing the presence of some weeds, such as Galactites tomentosa, and grasses, probably due to animal selection; grazing also increased the presence of subclover (Trifolium subterraneum), thus improving the pastoral value. The botanical composition of the fresh grass available in both treatments was dominated mainly by species from the Compositae family. In both treatments, dry matter production was always below 2 t ha−1 year−1 , with a maximum value of forage availability and growth rate in spring. Seasonal rainfall distribution and temperature determine the length of growing period in Sardinia, leading to a huge variation of forage availability during the year. As a consequence, pasture tends to be overgrazed in dry autumn and/or cold winter, and undergrazed during favourable conditions of spring. For grazed and ungrazed treatments, the amount of residual organic matter was very high, particularly in autumn, because of the great quantity of non-utilized biomass, unpalatable and thorny species from the previous year. An increase in the percentage of residual organic matter, on the basis of dry matter content, occurred at the end of spring. In the ungrazed basin, total residual dry matter in summer was high, increasing the risk of fire. In another trial carried out in northern Sardinia, comparisons were made between a ploughed– cereal cultivated area, and an area with natural Mediterranean maquis. Based on this study and on some other trials that have been conducted for 10 years (Caredda et al. 1996), it was found that a good permanent pasture sward or a dense Mediterranean maquis was adequate vegetation cover to maintain soil losses far below critical levels in the autumn months. The intensification of cropping on hillslopes using annual short-term forage and winter crops increased erosion risk, particularly if crop establishment was slow and ploughing was done across the contours, rather than parallel to them. Figure 6.6 compares the effect of land use on percentage vegetation cover in the winter months, and illustrates the difference in potential erosion risk between maintaining pasture and cultivating arable crops, which leave the soil surface almost bare for weeks or months. Therefore, the incorrect methods of ploughing on steep slopes, and the lack of vegetation between ploughing and sowing time can be considered as dangerous agricultural practice as far as desertification is concerned.
80
Mediterranean Desertification 100
Vegetation cover (%)
80
60
40
20
0 Oct
Nov
Dec
natural pasture short-term forage crop
Jan
Feb
Mar
improved pasture cereal
Figure 6.6 Comparison of the change in % vegetation cover for pasture and for arable crops between October 1993 and March 1994, at Ottava-Sassari
Furthermore, cereal crops cultivated in those marginal areas not only increase erosion risk but are uneconomic as they also show a low yield.
5
CONCLUSIONS
Agro-pastoralism is one of the most important economic activities in the whole of the Mediterranean Basin. In southern Mediterranean countries, the increased demographic pressure, urbanization, the improved standard of living and the needs of the human diet, based on proteins from milk and meat of goats and sheep, will lead to an increased food demand in the near future. Therefore, one might expect an increase in over-exploitation of agricultural lands in these areas due to an increase in stocking rates of small ruminants. On the other hand, particularly during the last decade, the north Mediterranean countries have experienced land abandonment in rural areas which has led to the appearance of land degradation phenomena. The studies carried out in Sardinia, an area characterized by both over-exploitation and land abandonment, showed that practices related to agro-pastoral activities, such as overgrazing, badly planned cultivation and the use of fire to clear pastures, can be considered the main causes of desertification. Based on those studies, the following general guidelines for preventing and mitigating desertification can be suggested: 1. Evaluation of land suitability for grazing should be done by standardized methods that must be adapted to different scales (regional, local and point scale). 2. Estimation of pasture productivity using mathematical models based on meteorological and vegetation parameters may allow the determination of sustainable stocking rates according to the season and year. 3. Determination of low-impact animal production systems should consider animal species and varieties adapted to local environmental conditions, rational grazing management, and lowimpact pasture improvement practices. 4. Development of integrated models for the management of agro-silvo-pastoral resources in the area with particular reference to common lands. Farm development plans should be linked to planning and management tools at a larger scale.
Agro-pastoral Activities and Land Degradation in Sardinia
81
REFERENCES Blackburn WH (1983) Livestock grazing impacts on watersheds. Rangelands 5(3), 123–125. Blackburn WH (1984) Impacts of grazing intensity and specialized grazing systems on watershed characteristics and responses. In Developing Strategies for Rangeland Management , National Research Council and National Academy of Science. Westview Press, Boulder, Colorado, pp. 927–983. Brandano P and Rossi G (1975) La razionale utilizzazione dei pascoli ed i problemi inerenti la determinazione del carico. L’Informatore Agrario 32, 20 139–20 148. Buller HJ (1992) Agricultural Change and the Environment in Western Europe. Mansell Press, London. Caldwell MM (1984) Plant requirements for prudent grazing. In Developing Strategies for Rangeland Management, National Research Council and National Academy of Science. Westview Press, Boulder, Colorado, pp. 117–152. Caredda S, Porqueddu C, Roggero PP, Sanna A and Casu S (1992) Feed resources and feed requirements in the sheep agro-pastoral system of Sardinia. Proceedings of the IV International Rangeland Congress, Montpellier, 22–26 April 1991, pp. 734–737. Caredda S, Ledda L, Porqueddu C and Sulas L (1996) Pastoral value and forage production of grazed and ungrazed natural swards in a semi-arid Mediterranean area of Sardinia. In G Parente, J Frame and S Orsi (eds) Grassland and Land Use Systems. Arti Grafiche Friulane, Tavagnacco, pp. 395–399. Chandler C, Cheney P, Thomas P, Trabaud L and Williams D (1983) Fire in Forestry. Forest Fires Behaviour and Effects. John Wiley, New York. Doucet P and Sloep PB (1992) Mathematical Modelling in Life Sciences. Ellis Horwood, Chichester. EEC (1996) Forest Fires in the South Mediterranean Union. European Commission, Directorate General for Agriculture, Brussels. Enne G, Pulina G, d’Angelo M and Masala G (1996) The role of animal grazing behaviour on land degradation in Mediterranean environments. In International Conference on Land Degradation, Abstract. University of Cukurova, ¸ Adana, Turkey. FAO (1995) FAO Yearbook Production 1994 . FAO Statistics Series, FAO, Rome. ISTAT (1976) II Censimento Generale dell’Agricoltura. Istituto Centrale di Statistica, Rome. ISTAT (1982) III Censimento Generale dell’Agricoltura. Istituto Centrale di Statistica, Rome. ISTAT (1992) IV Censimento Generale dell’Agricoltura. Istituto Centrale di Statistica, Rome. Madrau S, Deroma MA, Dess`ı G and Goussikpe Y (1995) Soil properties and trafficability of the Rio AstiminiFiume Santo experimental plots (North-western Sardinia). In A Aru, G Enne and G Pulina (eds) Land Use and Soil Degradation. MEDALUS in Sardinia. La Celere Editrice, Alghero, pp. 211–220. Molina MJ (1996) Impact of forest fires on desertification processes: a review in relation to soil erodibility. In JL Rubio and A Calvo (eds) Soil Degradation and Desertification in Mediterranean Environments. Geoforma Ediciones, Logrono, Spain. Previtali F (1996) Soil degradation processes and land use in north-western Sardinia (Italy). Proceedings of the International Conference on Land Degradation, Abstract. University of Cukurova, ¸ Adana, Turkey. Pulina G, Zanda A and Enne G (1995a) The impact of animal husbandry on the degradation of the soil. In A Aru, G Enne and G Pulina (eds) Land Use and Soil Degradation. MEDALUS in Sardinia. La Celere Editrice, Alghero, pp. 231–240. Pulina G, Masala G, Zanda A and Enne G (1995b) Analysis of biomass balance and stocking rate in cattle and sheep production systems in Mediterranean areas. MEDIT 1, 27–30. RAS (1996) Piano per la difesa contro gli incendi nei boschi e nelle campagne. Regione Autonoma della Sardegna–Assessorato della Difesa dell’Ambiente, Cagliari. Rivoira G, Roggero PP and Bullitta S (1989) Influenza delle tecniche di miglioramento dei pascoli sui fenomeni erosivi dei terreni in pendio. Rivista di Agronomia 4, 372–377. Roggero PP, Porqueddu C, Sulas L and Caredda S (1995) Relationships between land use and water run-off and soil erosion in a Mediterranean environment. In NE West (ed.) Proceedings of the V International Grassland Congress, Society of Range Management, Denver, Colorado, pp. 481–482. Seligman NG (1996) Management of Mediterranean grasslands. In J Hodgson and AW Illius (eds) The Ecology and Management of Grazing Systems. CAB International, Oxon, pp. 359–391. Soil Survey Staff (1996) Keys to Soil Taxonomy. United States Department of Agriculture, National Resource Conservation Service, Washington, DC. Vallentine JF (1990) Grazing Management. Academic Press, San Diego. Van Soest P (1994) Nutritional Ecology of Ruminants, 2nd edition, Cornell University Press, Ithaca, New York, pp. 93–107.
7
Landscape Protection from Grazing and Fire
N.S. MARGARIS AND E. KOUTSIDOU
Department of Environmental Studies, University of the Aegean, Lesvos, Greece
1 INTRODUCTION During recent decades, the number of wildfires and the surface area burned have increased in most countries of the Mediterranean region. Forest fires have become a major concern for politicians and the general public. Although fires have affected the Mediterranean region for millions of years, it is likely that changes in land use brought about by recent socio-economic changes may have affected when, where, how big, how frequent and how intense these fires are. Until now, in most countries fire suppression has been the main policy, and large and increasing sums are spent on fighting fires every year. To introduce new policies that would reduce fire risk and its impacts, it is necessary to have a good understanding of how fire affects the structure and functioning of ecosystems. The purpose of this chapter is to evaluate the main ways in which fire (combined with overgrazing) affects ecosystem properties in Mediterranean-type areas and what ecological principles management policies should take into account to reduce fire-related hazards (Figure 7.1). A major part of the Mediterranean landscape is made up of the semi-natural (non-cultivated but managed) system. The plants dominating in this system in the “Mediterranean-type” climate have developed adaptations to water stress during the dry period. They may be evergreen sclerophyllous (called maquis in the Mediterranean Basin, choresh in Israel, matorral in Chile, and chaparral in California) at the wet end of the precipitation gradient (975 mm annual precipitation), or, at the dry end of the precipitation gradient (around 275 mm annual precipitation), they may show seasonal dimorphism, which considerably reduces the transpiring surface area of the plant in the hot season. Such plants (known as phrygana in Greece, tomillares in Spain, batha in Israel, and coastal sage in California) are mainly thorny with a low cushion structure. Evergreen sclerophyllous plants of the maquis are characterized by small thick leathery evergreen leaves. These leaves often show a combination of anatomical modifications including thick cuticles, well-developed palisade mesophyll and the enclosure of stomata within small pits or grooves which are often partially occluded by hair or wax tubules. In general, evergreen species have leaves with a relatively high mass per unit area. This greater density of the evergreen leaf may be important in promoting water conservation and predator protection. Over the Greek Peninsula and the Aegean Islands these plant types are widely distributed, absent only in mountainous areas, and becoming sparse on the northern coast of the Aegean Sea. It has been estimated that the phrygana occupy between 1 and 1.6 million hectares while the maquis area covers about 775 000 ha (Papanastasis 1977; Diamantopoulos 1983). The total area of phryganatype vegetation in the Mediterranean Basin has been estimated at about 25 million hectares (Le Houerou 1981).
2 ADAPTATION OF THE MEDITERRANEAN ECOSYSTEMS TO SURVIVE FIRE Fire is a very common event in the areas covered by Mediterranean-type ecosystems, due to the combination of high air temperatures and water deficiency during the summer. During the long Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
84
Mediterranean Desertification
Figure 7.1 Annual fire pre-suppression costs are not related to the area of burned land or the financial loss caused by fires at national level
dry season, maximum temperatures may exceed 40 ◦ C, dry continental winds sweep coastward and air relative humidity may drop below 5%. The vegetation becomes extremely dry (less than 10% moisture on a dry weight basis). In this desiccated state the Mediterranean-type ecosystems are very susceptible to fire, a fact long recognized by scientists (Griesebach 1872). Typically, major fires in any particular area occur with a frequency of once every 20 to 30 years. It is clear that fire is an important element in the actual evolution of maquis and other sclerophyllous shrub vegetation in all regions with a Mediterranean-type climate. The positive influence of fire in the evolution of these ecosystems has been indicated by many authors (Shantz 1947; Biswell 1974; Margaris 1981; Trabaud 1982). Surprisingly, there seem to be evolutionary tendencies towards increased flammability in these plants, for example by the production of volatile oils in many phryganic species. It has been suggested
Landscape Protection from Grazing and Fire
85
that the plants are adapted to survive fairly regular burning, where fire temperatures are controlled, and the plants are only superficially damaged. However, if the volume of flammable plant material on the surface has been allowed to build up, then any subsequent fire would be intense and at very high temperatures, damaging plant structures above and below ground, and destroying seed banks in the surface soil (Moreno and Oechel 1994). Plants dominating Mediterranean-type ecosystems are equipped with adaptive strategies to help their post-fire recovery (Margaris 1981; Arianoutsou and Margaris 1982). Thus fire and adaptations to fire are part of the properties of the ecosystem. Under normal fire frequencies (every 80–100 years for pine forest, every 20–30 years for shrubs; Trabaud 1982), fire acts as a selective force in these systems so that the results of fire can be considered as fire-induced or fire-adapted. Post-fire survival of the woody community can be accomplished by re-sprouting and by increased seed germination from the soil seed bank. Re-sprouting is carried out from dormant buds located at the crown where they are not destroyed by fire. This can be observed quickly after fire in the maquis and after the first autumn rains in the phrygana. Vegetative regeneration is very fast, and in a period of less than 10 years full reconstitution takes place because the rate of photosynthesis in the young stems is particularly high (Naveh 1973). In the phryganic species this fast reconstitution can be due to the new leaves which are both larger and richer in chlorophyll than normal. For example, the increase in the chlorophyll content in leaves after fire was 150% compared to that before fire in Sarcopoterium spinosum, allowing photosynthesis to be significantly increased (Arianoutsou and Margaris 1982). Survival of the dominant woody species by increased germination of its own seed is an adaptation that occurs mainly in Pinus species and in the phrygana, while in the maquis the only method of survival seems to be the ability to re-sprout. This re-sprouting soon results in a stand without great age dissimilarities between the individuals. Under normal fire frequencies herbaceous plants appear in large numbers during the first few post-fire years, which is the result of the activated germination of seeds of those herbaceous species lying in the soil seed bank. This short-lived herbaceous interlude has two major advantages. First, the vegetation cover protects against soil erosion, and second, through the nitrogen-fixing ability of leguminous species, atmospheric nitrogen is brought back to the soil, since approximately 95% of the nitrogen contained in the above-ground biomass is removed from the ecosystem in smoke from fires. Leguminous species return nitrogen to the soil via the nodules on their roots (Arianoutsou and Margaris 1981). The proportion of the herb biomass is greater during the first two years after fire, then declines rapidly, with shrubby species dominating from the third year onwards. The life span of the dominant woody species of the phrygana, such as Phlomis fruticosa, Euphorbia acanthothamnos, Sarcopoterium spinosum and Cistus sp., may be a few decades, but this period might be less than the time interval between two successive fires. Cistus sp. live for up to 15 years and the same is true for Phlomis fruticosa. Euphorbia acanthothamnos, Sarcopoterium spinosum and Thymus capitatus have longer life spans, reaching 50 years (Arianoutsou and Margaris 1982). Consequently these species must base their recovery mechanisms on germination of dormant seeds after fire, and this is reflected in their age distribution. This is normal for the woody plants of the phrygana but not for the maquis. We have shown that there is an optimum frequency of fires to ensure the survival of the ecosystem species, but the intervention of human activities may result in a fire frequency that is either higher or lower than the optimum. In both cases the equilibrium is affected, and systems may become degraded. This will be our point of focus, since only a higher fire frequency is considered to cause damage. About 200 000 ha of forests and shrublands are burned annually in the Mediterranean Basin (Le Houerou 1973). The numerous adaptations to fire shown by plants dominating these ecosystems suggest that fire has been a selective force for a long time. The important point is that the plant ecosystems can recover from the regular burning regime that they have become adapted to, but interference from humans, and other associated factors such as grazing or trampling by livestock, can interrupt that process and cause long-term damage to the plant communities. Humans have been burning the Mediterranean forest for the past 6000 years in order to obtain better pasture and crop lands, and this practice still remains an important cause of forest fire in Corsica, certain parts of Italy, Greece, Turkey, Spain and Algeria. When fires occur too frequently
86
Mediterranean Desertification
(every 10 years or less) the plant community structure loses its ability to regenerate and becomes very simple, with only low shrubs and herbs (Le Houerou 1981).
3
GRAZING IN MEDITERRANEAN ECOSYSTEMS AND THE RELATIONSHIP WITH FIRE
The breeding of sheep and goats began in the eastern Mediterranean between 12 000 and 8000 BP. Humans have partly co-evolved with these ecosystems. The breeding of ruminants is still the principal agricultural activity in the Mediterranean countries (Le Houerou 1981). Rangelands, as surveyed by the National Statistical Service of Greece, include all non-productive forests as well as maquis and phryganic ecosystems. In fact, all semi-natural vegetation, apart from a few timber-producing forests, is used by domestic herbivores and can be considered as rangeland (Paraskevopoulos 1991). As with fire, the Mediterranean vegetation is well adapted to withstand complete devastation by grazing livestock by having thorns and aromatic substances that make it unpalatable (Le Houerou 1981). These are the essential oils and resins that also make the vegetation highly flammable during periods of drought. Like fire, grazing only causes permanent damage to the plant communities if it is uncontrolled. If it is carried out using rational criteria (i.e. choosing the optimum time of the year for grazing, not allowing grazing in one place all year round, and recognizing an optimum period for which land should be left fallow) and with the correct stocking rate, grazing will improve plant productivity and biodiversity, and will reduce the accumulation of flammable dry biomass. In general, one year of grazing should be followed by two years left ungrazed (Paraskevopoulos 1991). The plant ecosystems have adapted to withstand fire and grazing, except where the pressure is too great. Shepherds know very well that herbaceous plants are abundant after fire and that leguminous species in particular are an excellent source of food for sheep. For this reason fires may be started on purpose. If fires are followed by intense grazing, and the juicy herbaceous plants and re-sprouting are either eaten or trampled during grazing, the whole system will be out of equilibrium and cannot recover because the otherwise efficient adaptive mechanisms cannot continue. The only plants capable of survival are those particularly resistant to fire and grazing. Since most of these are unpalatable, shepherds wishing to get rid of them use the same practice, i.e. setting fires more frequently. This is especially true for Sarcopoterium spinosum, a Mediterranean dwarf shrub, because this shrub is not suitable for animal fodder. Another reason why the overgrazing that follows wildfires has a destructive impact on the land productivity is that the supply of nitrogen to the system is reduced. Also, because the low productivity of the Mediterranean rangelands may not be sufficient to cover the dietary needs of livestock, shepherds are often forced to supplement the diet with fodder. Keeping livestock is not an easy living, and in order to increase productivity and to reduce the production costs of dairy products on a short-term basis the cycle of fires and overgrazing continues every year. This approach completely undermines the equilibrium between vegetation, fire and grazing, and leads to widespread degradation. Maintaining an equilibrium between vegetation, fire and grazing, within the limits of the adaptations described above, is the only method available of sustaining the ecosystem in the long term. An example of where the increase in grazing pressure (even now there is a trend to increase the number of grazing animals) has resulted in ecosystem degradation and desertification, is on the Aegean Islands. In some desertified areas the dominant species is Asphodelus microcarpus, a plant well adapted to dry conditions, fire and overgrazing since its deep-rooted underground part avoids permanent damage to the plant (Pantis 1987). In general, these asphodel deserts can be considered to be the last stage of degradation of the maquis systems. Under normal conditions Asphodelus microcarpus is a geophyte that grows in the maquis as well as in the phrygana. It has a flower stalk 1 m high, which is greater than the height of most shrubs, so it can be distinguished easily by pollinators and therefore reproduces effectively. Asphodel deserts are systems of extremely low productivity. The above-ground biomass does not surpass 150 g m−2 , compared to non-degraded maquis with an above-ground biomass of at least 4 kg m−2 .
Landscape Protection from Grazing and Fire
87
Since historical times the inhabitants and farmers of the Mediterranean landscape have been acquainted with the potential problems and have taken measures against desertification. On the Aegean Islands, isolation and scarcity of resources forced islanders to develop special systems of management in which agriculture and animal husbandry were key parts. Landscapes that lacked flat plains and a reliable water supply forced the development of terrace agriculture and the cultivation of either annual (cereals, legumes, etc.) or perennial (olives, almonds, figs, etc.) plant species. This agro-pastoral system included grazing and was usually organized on a rotation system so that 50% of the land was under cultivation and the other 50% under grazing, reversing the land use the following year. Grazing was permitted only in certain areas and not all year round. For hundreds of years this form of management allowed a continuous but sustainable exploitation of environmental resources right across the Mediterranean Basin (Margaris 1992). As long as equilibrium is maintained, rates of desertification remain very low. This optimal system has collapsed dramatically during recent decades. On the Aegean Islands this has mainly been due to the increased mechanization of agriculture on the plains of the Greek mainland, which has meant that the islanders could no longer produce cereal, etc., at competitive prices and so were forced to abandon their usual agriculture activities. With the abandonment of terrace cultivation, animal husbandry has begun to dominate on all the islands (Table 7.1). Although the area given over to pasture has increased, the grazing pressure has not been reduced because in recent years the number of animals has also increased, mainly as a result of subsidies provided by the EC. The result has been an increase in desertification processes on the islands, a phenomenon that requires immediate action wherever possible. The relatively high stocking rate and the lack of planning for development and exploitation of pastures, together with the frequency of fires (almost annual) on the Aegean islands, has led to a general impoverishment of the pastoral value of the natural vegetation, followed by a progressive reduction and then elimination of the most favoured species (legumes and grasses) and their replacement by species that are unpalatable to livestock (Le Houerou 1977; Clark 1996). The most serious damage is done when grazing is permitted on soil still bare during the first year after a burn, when the woody or herbaceous young shoots are still tender. Herbaceous plants and re-sprouting shoots are eaten or trampled by grazing animals and the vegetation has no further means to regenerate. Where overgrazing removes seedlings before new seed is set, the seed bank for regeneration with annual herbaceous species after fire is progressively depleted. While a natural phryganic ecosystem supports approximately 20 woody species, where degradation is occurring the number of species declines so that eventually only two or three species, particularly Sarcopoterium spinosum, Thymus capitatus or Cistus sp., remain. Only plants that can regenerate from deep-seated underground parts, like the asphodel, can survive the most intensive fire and grazing. On the Aegean Islands it is clear that the productive capacity of the pastures is not enough to support the current number of animals on a long-term basis. Since the low productivity of the rangeland is not sufficient to cover the dietary needs of the livestock, shepherds are forced to supplement the diet with imported fodder. The total animal food products imported to the islands is increasing year by year. On the island of Lesvos, for example, natural pasture biomass is only providing 30% of animal requirements. Shepherds wishing to get rid of unpalatable thorny plant species and promote the short-term growth of palatable herbaceous and re-sprouting species are Table 7.1 Changes in the number of sheep and goats (in flocks and nomadic) between 1965 and 1995 on selected Aegean islands
Island Dodecanisa Cyclades Lesvos Samos Chios
1965
1970
1975
1980
1985
1990
1995
119 540 175 605 183 501 29 647 13 542
121 119 172 999 188 620 30 629 24 472
78 788 132 763 174 896 13 194 15 220
117 300 239 047 293 172 22 042 31 830
146 320 242 247 336 533 25 039 23 778
145 802 281 237 372 856 28 232 37 736
150 063 248 623 405 129 32 081 42 613
88
Mediterranean Desertification
still using fire on a frequent basis. As many as 40% of the total number of fires on the islands may be caused intentionally by shepherds (Koutsidou 1995). The growth of the livestock population combined with the low natural productivity of the pastures explains very clearly the rapid and intense degradation of Mediterranean vegetation seen in recent decades. Without intervention, the final result will be irreversible desertification.
4
THE MANAGEMENT OF MEDITERRANEAN-TYPE ECOSYSTEMS
Ecosystem degradation caused by the combination of frequent fires and subsequent overgrazing has led to the consideration of fire as a catastrophic event usually attributed to criminal actions. The healthy ecosystem’s natural requirement for regular burning tends to be forgotten. Due to this misunderstanding a policy of complete fire exclusion was initiated, and this brought more problems. The direct consequence of a fire exclusion policy is that biomass accumulates in the ecosystem. Therefore, when fire does break out, the burns reach high temperatures and the vegetation and the seed bank may be too damaged to allow recovery afterwards. This fact was realized about 25 years ago in California, and today carefully controlled burning techniques are a common practice in California (Conrad and Oechel 1982). Also, from the ecological point of view, in the systems where fire is excluded the diversity is reduced in terms of both plant and animal species. So fire exclusion policies are not protecting the maquis at all. Fires are needed, prescribed regularly, and controlled during the moister part of the year to reduce the temperatures reached. This technique is also being used as a modern and efficient tool for the management of certain other vegetation types commonly reputed to be very sensitive, such as the Pinus halepensis forests of Greece. The result is real protection not only of these ecosystems but also of human settlements. The shortcomings of this technique are that (a) enormous quantities of energy and organic matter are lost and (b) there is always the danger of the fire running out of control. Although experience from California is very positive, countries around the Mediterranean are proving unwilling to abandon fire exclusion policies, which are both inefficient and very costly. In Greece, even political stability is threatened because of annual fires. There, an area of approximately 15 000 ha is burnt every year. Failure of these measures to solve the problem has never led to any evaluation of their validity, as common logic might suggest. On the contrary, every Mediterranean government seems willing, even after years of widespread destruction, to further increase the budget available for so-called “fire protection policies” and for reforestation afterwards. It is ironic that today Mediterranean countries are importing 80% of their liquid fuels, while at the same time mismanaging the potential of natural vegetation as fuel. Huge amounts of money are spent on fire fighting and fire protection, and every year about 200 000 ha are burned out in the Mediterranean Basin. Margaris et al. (1985) have shown that if maquis ecosystems are harvested every 10 years, this practice can provide fuel and also reduce the risk of uncontrolled high-temperature fires which cause so much damage. After 10 years, biomass accumulation in the maquis is moderate, but not as great as at the mature stage, which takes 20–30 years. Education of farmers and landowners is required, to avoid further unnecessary exploitation and damage.
5
ECONOMIC ANALYSIS OF DESERTIFICATION: THE CASE OF THE ISLAND OF CHIOS
The northern territory of the island of Chios is a typical example of an area where there have been annual fires throughout a whole decade (1981–1991). Productivity of the rangelands was not sufficient to feed the animals, and fodder imports increased every year. At the same time, to minimize the costs of imported fodder, the natural ecosystem has been completely overgrazed. The combination of overgrazing and fires is the main cause of desertification in this area. The objective of this analysis is to evaluate and compare the total costs of overgrazing and fire with a more sustainable approach. Our aims would be to eliminate the deliberate starting of fires by shepherds to bring about a short-term increase in shoots palatable to livestock, and to alter the form
89
Landscape Protection from Grazing and Fire
of stock-farming from extensive to intensive, by subsidizing stock farmers so that they replace the plant biomass that is removed from the pasture by the livestock. The environmental benefits from the achievement of this goal are apparent both in terms of the reduction in the number of fires on the national and local scale, and in relation to the regeneration of vegetation that takes place when grazing is prohibited (Koutsidou 1995). Table 7.2 shows official classifications of the causes of fires on the northern part of Chios, and also for all of Greece for the period 1981–1991. It is likely that many of those fires with an unknown cause were in fact arson, but this is always difficult to prove. So we shall consider that 30% of fires of unknown causes and 20% of those caused due to negligence were unproved acts of arson. With this assumption, 44.1% of the fires in northern Chios were probably started deliberately, specifically to increase the area suitable for pasture (in the decade 1981–1991, 41.6 ha of agricultural land and 17.7 ha of forests became pasture). An economic analysis of the costs involved includes pre-suppression costs (PC), suppression costs (SC), partial-rectification costs (RC), added to suppression costs, and fire damage (FD). Pre-suppression costs are regarded as the purchase of machinery (aircraft, fire engines, wireless, communication systems, etc.), the construction work involved in creating anti-fire protection facilities (fire control watch towers, anti-fire zones, etc.) and the salaries of personnel (forest guards, seasonal fire-fighters, etc.). Suppression costs are regarded as all the costs that arise during firefighting operations (labour hours, operational costs of machinery, consumable equipment, etc.). An attempt has been made to estimate the forest-fire prevention costs in northern Chios by means of a graduated approach. Of the total annual pre-suppression costs, 50% are apportioned to northern Chios, while the total annual suppression costs are apportioned between northern and southern Chios according to the location where the fire-fighting operations took place. Damages caused by the fire (FD) were estimated according to the percentage of burnt land area in the either part of the island, while partial-rectification costs (RC) in northern Chios are considered to be 50% of the total costs. Table 7.3 shows an analysis for northern Chios, for the period 1981–1991, taking 1991 as the base year, with costs expressed in American dollars. This analysis includes the commercial value of the wood destroyed, the cost of fire prevention and fire fighting, and the cost of regeneration, but does not take into account associated damage such as flash floods causing sheet erosion, leading to sedimentation and silting up of dams and canals, and the reduction of fertility and productivity of damaged soils and vegetation (Turner et al. 1994). The annual fluctuation of pre-suppression costs in order to fight fires in northern Chios does not necessarily cause a change on the same scale as the burnt land area or the financial loss caused by the fires. This lack of correspondence between the scale of the annual pre-suppression costs and the burnt land surface is also the case at the national scale (Skourtos and Dimitrakopoulos 1991). The average (1981–1991) expenditure per year, including direct and indirect costs, amounts to $9 884 170 per 915.97 ha (average burnt area per year) of burnt land surface on northern Chios. Table 7.2 Classification of causes of fires in Greece and northern Chios, 1981–1991
Cause of fire
Deliberate and arson Negligence Unknown cause Thunderbolt Ammunition used during military exercises Miscellaneous
All of Greece (%)
Northern Chios (%) Official classification
Alternative classification
27.66 24.31 40.38 2.92 0.84
27.5 24.5 40 3.44
44.1 19.6 28.3 3.44
3.89
4.56
4.56
Source: Chlikas, Ministry of Agriculture, Greece, personal communication, 1993.
Costs
1981
Aircraft 36 612 Vehicles 18 092 Buildings 408 Functional 377 Vehicle maintenance 336 Salaries of permanent 11 099 personnel 4 372 Salaries of seasonal and contract workers (SC)a Prevention strategiesb – Total PC for Chios 71 296 35 648 50% of PC for N. Chiosc PC for N. Chios (fixed 194 796 on 1991 rates) Burnt land on N. Chios 6147.5 (ha), and as % of (100%) total area burned on Chios Suppression and 47 631 rectification costs (based on 1991 rates) of N. Chios FD of N. Chios 61 881 648 (based on 1991 rates) Total PC + SC + RC + FD a
62 124 080
1982
1983
1984
20 901 17 188 511 359 295 15 322
26 066 16 283 561 480 553 19 483
94 707 15 378 651 613 1022 23 974
182 072 57 237 916 615 1855 30 622
7 530
10 079
33 280
32 484
– 62 105 31 053 139 875
3809 77 316 38 658 144 786
329 169 954 84 977 268 914
69.4 (47.8%)
370.7 (100%)
27 335
78 269
204 (49.3%)
28 437
1985
1988
1989
1990
526 316 102 727 2086 1089 2234 56 122
– 96 632 1579 1356 1365 67 859
29 210 – 90 536 166 678 1904 2309 903 1165 2958 8158 79 609 96 118
50 526
71 191
66 868
134 270 146 424
3332 309 132 154 566 409 986
5523 18 056 213 207 1 578 799 106 604 789 400 229 747 1 461 849
22 079 783 842 391 921 638 306
21 375 257 033 128 516 184 118
36 885 37 602 376 273 458 454 188 137 229 227 223 700 229 227
678 (70%)
126.1 (100%)
1083.8 (100%)
98 (65.5%)
413.2 (43.8%)
3289
227 782
45 365
15 710
78 602
1986
1987
– 1 350 329 115 247 108 822 1086 727 914 994 3934 1411 44 372 47 934 42 132
1060 (63.2%)
145 604
1991
45 (64.8%)
23 075
2 055 474
709 122 3 752 589 6 875 428 1 292 763 10 966 740 10 704 605 1 012 766 4 167 763 460 526
2 223 790
881 243 4 099 770 7 364 020 1 525 800 12 656 370 11 488 520 1 242 250 4 407 170 712 828
The cost covers salaries of contract workers and drivers working on a seasonal basis, fire-fighters and fire-guards. The cost covers construction work for the construction of anti-fire zones, fire watch-towers, etc. c Northern Chios comprises 42.61% of the total area of the islands. However, it was considered rational to allocate the costs equally in both parts of Chios due to the more frequent occurrence of fires in northern Chios. b
90
Table 7.3 Pre-suppression (PC), suppression (SC), reforestation–rectification costs (RD) and financial damage (FD) caused by fires on northern Chios between 1981 and 1991 (in American dollars, at current rates with 1991 as the base year)
Landscape Protection from Grazing and Fire
91
Table 7.4 Estimated cost of substantial cereal diet during the gestation period (in American dollars, at current rates with 1991 as the base year)
A cereal diet Vetch Straw Maize Roughage Cotton cake Total fodder for seven months per animal Number of sheep and goats
Animal needs per day (kg) 0.3125 0.13 0.781 0.25 0.0625 312.5 37 900
Cost 0.15 0.11 0.16 0.14 0.13 48.75 1 847 548
Source: Union of Agricultural co-operatives of Chios island, personal communication, 1994. Of this expenditure, 40% is due to fires caused by shepherds and amounts to $3 953 668 for the corresponding year. The costs presented in Table 7.4 show clearly the benefits to be had from using land-use practices that minimize the incidence of fire damage. The productivity of pastures in northern Chios was estimated as 10 732 612 kg dry matter (average productivity of 1989 and 1990; Koutsidou 1995). The grazing lands are used chiefly during the period from March to September. According to the statistics provided by the National Statistics Service of Greece for 1991, the extensive stock-raising comprises 37 900 animals: 7777 sheep and 30 123 goats. The seven months’ dietary needs of the animals that graze in the northern part of Chios, measured in dry matter, are 11 258 425 kg. It is clear that the productivity of the phryganic rangeland is not sufficient to cover the dietary needs of the livestock capital that grazes in the northern parts of Chios and additional fodder has to be imported. In the summer period shepherds are forced to supplement the diet with a total of 525 813 kg fodder (Koutsidou 1995). Therefore, to reduce the costs associated with imported fodder, shepherds cannot resist the temptation to set fires to grazing lands in order to increase, just for a short period of course, the pasture productivity, and subsequent livestock productivity, in terms of dairy products. A substantial cereal diet for all livestock during the gestation period, which equals the quantity of dry matter that would be taken from the ecosystem by the animals for seven months, costs $1 847 548. It is obvious that the subsidy of fodder costs less than the cost of fire-fighting, so we can avoid these fires by paying much less money directly to shepherds. It must be noted that the above figure is even smaller if the amount of money paid by shepherds as rental costs for the grazing land is taken into account. Finally, if we take into account that the best management for pasture rangelands is one year of grazing followed by one year of land being set aside, the total cost of subsidies may be reduced by half.
6 CONCLUSION We can say that Mediterranean-type ecosystems are very resilient and sustainable plant communities, following a long evolutionary history of frequent disturbances (fires and grazing). Fire is beneficial if low-intensity fires occur every 10–20 years and grazing is controlled. The problems of degradation and desertification only begin if the system is brought out of balance. Problems occur if (1) fire is used annually to stimulate new palatable growth for livestock, but a ground cover of vegetation is not maintained; (2) overgrazing also depletes the ground cover; and (3) highintensity fires occur because of the accumulation of dry biomass in mature Mediterranean ecosystems. The Mediterranean ecosystem can, however, be managed to allow all the required land uses, but in moderation.
92
Mediterranean Desertification
REFERENCES Arianoutsou M and Margaris NS (1981) Fire induced nutrient losses in a phryganic ecosystem. International Journal of Biometeorology 25, 341–347. Arianoutsou M and Margaris NS (1982) Phryganic (East Mediterranean) ecosystem and fire. Ecologia Mediterranea 8, 473–480. Biswell H (1974) Effect of fire on chaparral. In T Kozlowski and CE Ahlegren (eds) Fire and Ecosystems. Academic Press, New York, pp. 321–364. Clark SC (1996) Mediterranean ecology and an ecological synthesis of the field sites. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 271–299. Conrad CE and Oechel WC (1981) Dynamics and Management of Mediterranean-Type Ecosystems. USDA/Forest Service, Pacific Southwest Forest and Range Experimental Station, Berkeley, California. Diamantopoulos J (1983) Structure and distribution of phryganic ecosystems in Greece. PhD thesis, Aristotelian Thessaloniki University, Thessaloniki. Griesebach A (1872) Die Vegetation der Erde nach ihrer Klimatischen Leipzing, Anordnung. Koutsidou E (1995) Regeneration of degraded Mediterranean ecosystem after protection from overgrazing: the case of Chios island. PhD thesis, Department of Environmental Studies, University of the Aegean, Mytilini (in Greek with English summary). Le Houerou HN (1973) Fire and vegetation in the Mediterranean Basin. Proceedings of the 13th Annual Tall Timbers Fire Ecology Conference, Tallahassee, Florida, pp. 237–277. Le Houerou HN (1977) Biological recovery versus desertification. Economic Geography 53(4), 413–420. Le Houerou HN (1981) Impact of man and his animals on Mediterranean vegetation. In F di Castri, DW Goodall and RL Specht (eds) Mediterranean-type Shrublands. Elsevier, Amsterdam, pp. 479–521. Margaris NS (1981) Adaptive strategies in plants dominating mediterranean ecosystems. In F di Castri, DW Goodall and RL Specht (eds) Mediterranean-type Shrublands. Elsevier, Amsterdam, pp. 309–315. Margaris NS (1992) Primary sector and the environment in the Aegean Islands, Greece. Environmental Management 16(5), 569–574. Margaris NS, Arianoutsou M, Paraskevopoulos S, Diamantipoulos J and Vokou D (1985) Harvesting before the Fire for Energy Mediterranean-type Ecosystems: Costs & Benefits. Commission of the European Communities, DG XII, EUR 9968 EN, Luxemburg. Moreno JM and Oechel WC (1994) Fire intensity as a determinant factor of postfire plant recovery in Southern California Chaparral. In JM Moreno and WC Oechel (eds) The Role of Fire in Mediterranean-type Ecosystems. Springer-Verlag, New York, pp. 26–45. Naveh Z (1973) The Ecology of Fire in Israel. Proceedings of the 13th Annual Tall Timbers Fire Ecology Conference, Tallahassee, Florida, pp. 131–170. Pantis J (1987) Structure, dynamics and management of Asphodelus deserts in Thessaly. PhD thesis, Aristotelian Thessaloniki University, Thessaloniki. Papanastasis V (1977) Fire ecology and management of phrygana communities in Greece. Proceedings of the Symposium on Environmental Consequences of Fire and Fuel Management in Mediterranean Ecosystems. USDA Forest Service, General Technical Report Wo-3, Washington, DC, pp. 476–482. Paraskevopoulos S (1991) Study of regeneration of Mediterranean type ecosystems after harvesting for energy and biomass. PhD thesis, University of Aegean, Mytilini. Shantz HL (1947) The Use of Fire as a Tool in Management of the Bruch Ranges of California. California State Board of Forestry, Sacramento. Skourtos M and Dimitrakopoulos A (1991) Economic analysis of fighting forest fires in Greece. Proceedings of the 2nd Conference of Environmental Science and Technology, Mytilini, pp. 299–307 (in Greek). Trabaud L (1982) Effects of past and present fire on the vegetation of the French Mediterranean Region. In LE Conrad and W Oechal (eds) Dynamics and Management of Mediterranean Type Ecosystems. USDA/Forest Service, Pacific Southwest Forest and Range Experimental Station, Berkeley, pp. 450–457. Turner RK, Pearce D and Batema I (1994) Environmental Economics. TJ Press, Cornwall, UK.
8
Bioengineering Principles and Desertification Mitigation
J.N. QUINTON,1 R.P.C. MORGAN,1 N.A. ARCHER,2 G.M. HALL1 AND A. GREEN1 1 2
National Soil Resources Institute, Cranfield University, Silsoe, UK Biological Sciences Institute, University of Dundee, UK
1 INTRODUCTION In the semi-arid regions of the Mediterranean the conservation of the soil and water resource is of particular importance; soils are often poor and soil and water are both in short supply. Since ancient times technologies have been adopted to conserve soils and water. Bench terraces with risers protected with drystone walls are found throughout much of the region. However, mechanical methods, such as terracing, are expensive to construct and maintain, making them an unattractive proposition for the protection of land with low value. Currently social, economic and biophysical pressures are forcing many farmers to abandon large areas of previously cultivated land, often leaving it devoid of vegetation and vulnerable to erosion. To protect this land, a cost-effective means of conserving soil and water is required. Vegetation is an important component of soil and water conservation systems and its role in protecting the soil from erosion has long been recognized (Hudson 1996; Morgan 1996). Vegetation modifies the soil and changes the energy characteristics of rainfall and runoff. Aggregate stability and soil fertility can be improved by the addition of organic matter. Raindrops may be shattered and have their energy reduced by the foliage. Plant roots can increase soil strength and may promote infiltration. The use of natural vegetation has been suggested as one way of managing such land within the Mediterranean region (Francis and Thornes 1990; Morgan 1991; Quinton et al. 1997). The use of vegetation in engineering applications such as this has been termed “bioengineering” (Coppin and Richards 1990; Morgan and Rickson 1995). However, there are few investigations of the bioengineering properties of semi-natural Mediterranean vegetation, and in particular the role of plant properties on soil hydrology. Such information is important if vegetation with suitable bioengineering properties is to be chosen for soil and water conservation works. This chapter describes work carried out in south-east Spain that examines the effects of plant properties, both above and below the ground, on infiltration, runoff generation and soil erosion. It then uses this information and the available literature to suggest a number of species that might prove useful in a revegetation programme.
2 ABOVE-GROUND PROPERTIES 2.1 Field Sites
Two study sites, Barranco de la Cantina (site 1) and Barranco del Charco de Castro (site 2), were chosen as representative of abandoned land in the semi-arid area of south-east Spain. Both sites are within the Guadalent´ın Basin, and are located close to the village of Zarzadilla de Totana (Figure 8.1). The Barranco de la Cantina site (37◦ 52′ N, 1◦ 40′ W) consists of a series of abandoned terraces on the pediment between the Sierra de Espu˜na and Burras. Close to the site are terraces Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
94
Mediterranean Desertification
Figure 8.1 Location of study sites (from Quinton et al. 1997). Reproduced by permission from British Society of Soil Science
cropped for grapes and almonds. There are also signs of erosion, including severe rilling, on other abandoned fields less than 1 km away. The site is botanically diverse with a mixture of shrubs and annual species (Table 8.1). The range of species and size of many of the shrubs suggests that the terraces have been abandoned for at least 10 years, and possibly longer. Local people were unable to confirm this date. The Barranco del Charco de Castro site (37◦ 51′ 20′′ N, 1◦ 45′ W) is in an area of abandoned terraces to the south of the Sierra de Madro˜no (Figure 8.1). Here, according to local farmers who were interviewed, abandonment took place 20 years ago when olive trees were planted which died shortly afterwards due to drought. At this site the vegetation is dominated by annuals. Both sites have soils formed over marls which are classified as calcic xerosols according to the FAO–UNESCO (1974) legend. The closest meteorological station at Zarzadilla de Totana (less than 5 km from both sites) has a mean annual rainfall of 355 mm (Navarro 1991). The distribution of rainfall through the year is uneven, with dry periods during the summer months and during January and February. Temperatures are high during the summer dry period, with a mean July maximum of 26.5 ◦ C. The site can be classified as semi-arid and within the upper thermomediterranean bioclimatic belt according to Peinado et al. (1992). The region experiences extreme rainfall events, with Totana (25 km south-west of the field sites) recording a rainfall of 350 mm in 10 h on 18 and 19 October 1973, from which severe flooding resulted (Thornes 1976). While estimates put the return period of such an event in excess of 500 years (Thornes 1976), 11 large flood events have occurred on the Rio Guadalent´ın in the last 200 years, with the discharge of the April 1802 event estimated at four times that of the 1973 flood (L´opez Berm´udez 1993). 2.2
Methodology for Field Studies To characterize the hydrological and erosional response of the dominant vegetation types on each site, rainfall simulation experiments were carried out at sites 1 and 2 in October 1993 and repeated in May 1994. These represented the periods of minimum and maximum vegetative growth (see
Bioengineering Principles and Desertification Mitigation
95
Table 8.1 List of species present on the Barranco de la Cantina and Barranco del Charco de Castro field sites (+ denotes presence)
Species
Artemisia herba-alba Genista cinera Anthyllis cytisoides Biscutella didymax Helichrysum stoechas Teucrium sp. Adenocarpus sp. Artemisia campestris Pistacia lentisco Eryngium compestre Stipa tenacissima Onopordon nervosum Plantago albicans Digitalis (mullain type) Bupleurum ancifolium Erica australi Helianthemum sp. Picnomon acarn Sedum sediforme Sedum sp. Sideritis sp. Foeniculum sp. Genista equisetiformis Osyris alba Other grasses
Barranco del Charco de Castro
Barranco de la Cantina
+ − − − + + + − − − + + + + + + − + + + + − − − +
+ + + + + + + + + + − − − − − − + − + + + + + + +
Table 8.2 Treatments used in rainfall simulation experiments (m = mature; s = semi-mature, i = immature)
Barranco de la Cantina
Barranco del Charco de Castro
Bare soil Artemisia herba-alba (m, s, i) Anthyllis cytisoides (m, s) Mixed community
Bare soil Plantago albicans Plantago albicans–Poa mix Stipa tenacissima
Orshan 1989). Eleven treatments were identified, representing five plant species and one bare soil treatment at each site, with some species of perennial vegetation being investigated at different stages of maturity which was assumed to be proportional to shrub size. The treatments are given in Table 8.2. Each treatment had four replicates. These were demarcated as single plots of variable size, between 1 and 1.75 m2 depending on the vegetation type. The need to experiment on single species necessitated the use of variable-sized plots. These were bounded on all sides by plastic rainwater guttering, which channelled the runoff and sediment to the collecting point.
96
Mediterranean Desertification
The following characteristics of each plot were described: species composition; per cent canopy cover; per cent ground cover; plant height, long axis and short axis for each plant on the plot; plant position on plot; litter biomass; shear strength at saturation measured using a shear-vane and torvane (Soil Test Inc.); per cent stone cover, and moisture content at beginning of simulation. A single nozzle rainfall simulator was used for the experiments. Water was supplied, at 100 KPa, to a nozzle (Delavan 1/2BQM32) on an adjustable mounting bracket, which was attached to a frame constructed from plastic drainpipe. The simulator was able to deliver approximately 120 mm h−1 to plots of approximately 1.5 m2 . Uniformity of the rainfall was reasonable, with a Christiansen (1942) coefficient of 81%. The median drop size, determined by the flour pellet method (Hudson 1964), was 1.4 mm. In windy conditions, the simulator was surrounded by a 10 m × 3 m wind shield made from “shade netting” to prevent spray drift. Each simulation was run for 15 min from the start of runoff. The discharge and sediment concentrations were determined at 3 min intervals throughout the simulation. 2.3
Effect of Plant Species and Season on Runoff and Soil Erosion
From Figure 8.2(a) it is apparent that all of the vegetation treatments reduced soil loss compared to the bare soil plots on both occasions. These differences are all significant at the P = 0.05 level. Figure 8.2(b) shows that, with the exception of the immature Artemisia herba-alba, the mean runoff is reduced by the vegetation treatments. On the Barranco de la Cantina site, only the mature Artemisia herba-alba and the semi-mature Anthyllis cytisoides gave significant decreases in runoff in October, whereas in May 1994 it was the semi-mature and mature Anthyllis cytisoides and the mixed community treatment that significantly decreased runoff. However, on the Barranco del Charco de Castro site, runoff was significantly reduced on both occasions by all the treatments. Although all treatments decreased soil loss and runoff compared with the bare soil, there were few differences between vegetation treatments. The reason for the higher runoff and soil loss from the immature Artemisia herba-alba plots, when compared to the other vegetation treatments, was the poor protection afforded to the soil surface. This species treatment had a significantly smaller canopy cover than the other vegetation treatments. It is apparent from Figure 8.2(a) that mean soil losses were lower for all treatments, including the bare soil, for the May 1994 simulations. There were no significant differences between the same vegetation treatments assessed during the two different seasons. No such differences occur for the runoff (Figure 8.2(b)). We cannot be certain why the soil loss from the May 1994 simulations is less. It may be a function of increased biological activity in the soil after the wet season, the effect of the previous simulations on the soil surface, i.e. the removal of erodible material leaving an erosion pavement, or perhaps changes in the canopy cover afforded by the vegetation. This latter reason is partly corroborated for all the perennial treatments where the canopy cover had increased over the winter. The lower canopy cover in October may have been due to drought avoidance strategies in the summer which decrease the surface area of the plants during times of water stress: Anthyllis cytisoides is summer deciduous; Artemisia herba-alba decreases the size of its transpiring parts during periods of drought; and Stipa tenacissima decreases the transpiring area by curling its leaves. However, Plantago also reduces its surface area by body reduction, yet the mean canopy cover was less in May 1994 than in October 1993. From observations in the field, it appeared that much of the Plantago had been killed by the combination of the summer drought and the dry winter. 2.4
Effect of Plant Properties on Soil Loss and Erosion Of the plant properties measured, only plant cover showed significant correlations, at P < 0.05, with soil loss (g/m−2 ) and runoff (mm) for both the October 1993 and May 1994 data. Broadly, as the cover increased, the soil loss and runoff fell (Figure 8.3). The data show that the decrease in soil loss was particularly marked as cover increased from 0 to 30%, which is in broad agreement with studies worldwide (Elwell and Stocking 1976; Lang and McCaffrey 1984; Brown et al. 1989) and with results from similar areas in southern Spain (Francis and Thornes 1990). The data also show that the greatest variation in the values of soil loss occurred when the ground was bare. As
97
Bioengineering Principles and Desertification Mitigation (a)
700.00 600.00
Soil loss (g m−2)
500.00
Oct 93 May 94
400.00 300.00 200.00 100.00
(b)
Stipa
Bare soil
Plantago /grass
Plantago
Mixed
Anthyllis (s)
Anthyllis (m)
Artemisia (m)
Artemisia (s)
Artemisia (i)
Bare soil
0.00
0.80 0.70
Runoff coefficient
0.60
Oct 93 May 94
0.50 0.40 0.30 0.20 0.10
Stipa
Bare soil
Plantago/grass
Plantago
Mixed
Anthyllis (s)
Anthyllis (m)
Artemisia (m)
Artemisia (s)
Artemisia (i)
Bare soil
0.00
Figure 8.2 (a) Soil loss and (b) runoff coefficients from the rainfall simulation experiments carried out on 11 vegetation treatments during October 1993 and May 1994 (from Quinton et al. 1997). Reproduced by permission from British Society of Soil Science
the vegetation cover increased, the variation diminished, indicating that there was little difference between the vegetation types when their cover exceeded 70%. If soil loss is assumed to have declined exponentially as the percentage cover increased, the value for the exponent was within the rather large range of 0.01 to 0.15 obtained by other workers (Brown et al. 1989). It is close to the value of 0.045 proposed for rangelands in the USA (Simanton and
98
Mediterranean Desertification 1000 900 800
Soil loss (g m−2)
700 600 500 400 300 200 100 0 0
10
20
30
40
50
60
70
80
90
100
Percentage canopy cover
Figure 8.3 Relationship between plant canopy cover and soil loss for the combined October 1993 and May 1994 data (from Quinton et al. 1997). Reproduced by permission from British Society of Soil Science
Renard 1992), but larger than the value of 0.026 obtained for the effects of random roughness and crop residue on agricultural land (Cogo et al. 1984; Worku Bekele and Thomas 1992) and the value of 0.025 on rangeland in Swaziland obtained with a rainfall simulator (Morgan et al. 1997). Within this general relationship, however, there were important differences between vegetation types. Values were much higher for Anthyllis cytisoides than for Artemisia herba-alba while rather low values were found for Stipa tenacissima. There were insufficient data to obtain relationships for the other plant species for covers in the 10–40% range.
3
BELOW-GROUND IMPACTS
The hydrological effects of the above-ground biomass or plant canopy have been well studied and documented (Coppin and Richards 1990; Styczen and Morgan 1995; Quinton et al. 1997). In contrast, the hydrological effects of plant roots have been neglected. This part of the study concentrates on how and to what extent individual plant species and different root morphologies affect soil permeability. 3.1 Methodology for Field Studies For each site (see section 2.1 in Chapter 14) the two dominant plant species were chosen for investigation along with a bare soil control (Table 8.3). Each treatment consisted of four individual plants chosen randomly and was replicated four times. A steel ring infiltrometer (diameter 53 cm, depth 25 cm) was placed around the plant and pressed into the soil to a depth of 2 cm to prevent leakage. The time taken for the water level to drop by Table 8.3 Treatments used in infiltrometer experiments
Barranco de la Cantina
Barranco del Charco de Castro
Artemisia herba-alba Anthyllis cytisoides Bare soil
Plantago albicans Stipa tenacissima Bare soil
Bioengineering Principles and Desertification Mitigation
99
Root density (g mm−3)
0.035 0.03
Bare soil a
0.025
Bare soil b
Anthyllis Artemisia Plantago Stipa
0.02 0.015 0.01 0.005 0 0
50
100
150
200
250
Infiltration rate (mm day −1)
Figure 8.4 The relationship between root density and infiltration rate determined at the Barranco del Charco de Castro and the Barranco de la Cantina field sites
1 cm was recorded; this was repeated three times to obtain the infiltration rate. The infiltrometer and the above-ground vegetation were removed and soil cores were taken inside the infiltrometer using an auger corer. In total, seven cores were taken: five cores at 0–9 cm depth and two cores at 9–18 cm depth. The cores were then washed and all the roots present removed and dried. The five 0–9 cm cores were combined; the two 9–18 cm cores were also combined. For each sample the roots were separated into size fractions depending on their diameter as follows: coarse (>4 mm), medium (4–1 mm), and fine (1–0.1 mm). The total root length of each fraction was estimated separately using the Root Length Intersection Method (Newman 1966). The roots were positioned randomly in a glass dish over a grid so that they did not overlap. Counts were made of the intersections of the roots with the vertical and horizontal gridlines. The intersection counts were converted to centimetres using the following equation (Newman 1966): root length (R) = 11/14 × number of intersections (N) × grid unit
(1)
After counting, the roots were dried and weighed. The root density was calculated as the dry weight of roots per unit volume of soil. 3.2 Results of Root Studies, Spain
Statistical analysis of the data shows significant correlations between infiltration rate (mm day−1 ) and root density (g cm−3 ) (r = 0.402, P < 0.05) (Figure 8.4). Within individual species groups, different responses are found between the two sites. For Stipa, a sharp increase in infiltration rate with root density is recorded. This is in line with investigations carried out by the University of Amsterdam (Imeson et al. 1994) which showed that deep infiltration occurred around active roots connected to the living standing stems in the Stipa tussock, and that this was due to preferential flow along vertical root systems. For Plantago, a general increase in infiltration rate is observed, but the relationship is not as pronounced because Plantago has a much lower rooting density and rooting depth. For Anthyllis and Artemisia, the relationship between root density and infiltration rate is unclear; this may be linked to the rooting morphology. Anthyllis possesses a deep tap root and Artemisia has a dendritic root system. Both show a low root density (0.002–0.015 g cm−3 ) in the top 18 cm of the soil. Again this is in broad agreement with the findings of the University of Amsterdam (Imeson et al. 1994) in which no preferential flow of water along roots under Artemisia bushes was found and therefore infiltration rates were similar to those of bare soil.
4 DISCUSSION 4.1 Discussion of Rainfall Simulation Studies The rainfall simulation studies have shown that there are few differences in the ability of the vegetation types studied to control runoff or soil erosion. Of the plant properties considered, only
100
Mediterranean Desertification
plant canopy cover shows a significant relationship with soil loss and runoff. Differences in the rate of exponential decay in soil loss with increasing percentage cover appear to relate more to the differences in erosion rates, both spatially and temporally, on the bare soils than to the behaviour of different vegetation types at high percentage covers. Nevertheless, the rate of decline in erosion with cover is greater for shrub and bush covers than for grasses and is higher in autumn than in late spring. These findings have implications for research into the management of abandoned sites prone to erosion. They suggest that future research should be concentrated on developing ecological successions and revegetation methods that promote a sustainable, high value of canopy or ground cover. Although plant height showed no significant effect on erosion in this study, it should be noted that the range of plant heights considered was low, and a stronger effect might be observed under taller canopies. 4.2
Selection of Vegetation for Revegetation
In order to select vegetation species for revegetation purposes the following strategy was adopted. From the rainfall simulation studies, it was possible to identify how the individual components of each plant species affected the hydrological and mechanical behaviour of the soil. This information combined with that from a review of previous studies on the ecology of semi-arid lands allowed criteria to be drawn up to aid identification and selection of potentially suitable species for the revegetation of abandoned land. The criteria were then used to compile a list of potentially suitable species for erosion control based on detailed descriptions from the available literature on their morphology and ecology. The species were selected on the basis of their ecological and engineering properties, as for soil erosion mitigation it is important for the chosen species to be both ecologically acceptable and able to perform the required engineering function. The results of the rainfall simulation experiments and root studies suggest that any revegetation programme should involve the design of an ecological succession to sustain a dense ground cover and a root system that promotes macroporosity of the soil. Any attempt to revegetate an arid or semi-arid area will have its difficulties. The propagation and growth of the natural vegetation is severely restricted by low rainfall, frequent droughts and poor soil conditions. Therefore, when selecting individual plant species, several points need to be taken into consideration. These are whether the plant possesses the necessary characteristics to control runoff and erosion, whether it is adapted to thrive under the hostile conditions of a semi-arid environment, and lastly how the selected plants can be combined into a suitable ecological succession. A list of 47 potentially useful species for erosion control has been compiled (Table 8.4) with a description of their habitat, morphology and ecological and/or bioengineering properties. Some of these species not only protect the soil surface from erosion but will also improve the nutrient status, structure, organic matter content and biological activity. Using information obtained from this study and from the literature, a catalogue of species, indigenous to the Mediterranean, which lists their bioengineering properties, ecological behaviour and Table 8.4 Potentially useful plant species for the revegetation of abandoned lands
Species GRASSES Agrostis tenuis
Habitat
All soil types but especially on dry acidic soils
Arrhenatherum elates
Dry regions
Bromus molliformis
Mediterranean soils
Description
Ecological/bioengineering properties
Tufted perennial grass Good on poor dry soils, spreading by good ground cover rhizomes forming a dense turf Tall perennial grass Dense fibrous root system, 50–150 cm excellent soil binder Loosely tufted Good on poor dry soils, biennial grass good nurse crop
Bioengineering Principles and Desertification Mitigation
101
Table 8.4 (continued)
Species Bromus tectorum Cynadon dactylon
Poa pratensis
Habitat Poor soils and waste land Mediterranean basin, dry regions
Dry and well-drained soils
LEGUMES Lotus corniculatus
Most soil types
Lotus tenuis
Dry grassy places
Medicago arborea
Meillotus alba
S Mediterranean regions, on sandy rocky soils Open habitats, native to Mediterranean Open habitats
Trifolium dubium
Dry grassy places
Trifolium fragiferum
Dry areas
Trifolium subterraneum
Dry grassy places
Medicago sativa
HERBS Plantago albicans Plantago notata SHRUBS Acacia farnessiane
Anabasis articulata
Dry areas, Mediterranean Dry places, SE Spain
Description
Ecological/bioengineering properties
Annual grass
Good on poor soils
Vigorous perennial grass, extensively creeping with both rhizomes and stolons Dense grass, spreads by creeping rhizomes
Good ground cover, soil binding
Apomictic, good soil binder
Decumbent perennial Good ground cover, herb 10–40 cm nitrogen-fixing high Densely branched low Tolerant of dry areas, perennial herb N-fixing, good cover Sericeous shrub N-fixing, soil improver, 1–4 m in height good cover Deep rooting erect perennial Erect branched annual herb Procumbent annual up to 25 cm high Creeping perennial herb
N-fixing, good cover, soil binding N-fixing, soil binding
Low perennial herb with one rosette Low perennial herb with one rosette
Excellent ground cover
N-fixing, good ground cover N-fixing, good ground cover, tolerant of saline soils Prostrate annual up to Good on sandy and 20 cm high gravelly soils
Deciduous densely Exotic naturalized branched shrub, on sandy soils can reach a height min. rainfall of 4 m 250 mm year−1 S Spain and N Africa Fleshy small on marls and saline persistent shrub, soils, min rainfall 10–30 cm in 250 mm height, densely branched
Excellent ground cover, drought resistant Good cover and litter production, N-fixing
Good cover, surface protection, infiltration
(continued overleaf )
102
Mediterranean Desertification
Table 8.4 (continued)
Species Anthyllis cytisoides
Anthyllis terniflora
Artemisia campestris
Artemisia herba-alba
Artemisia reptans
Atriplex canescens
Atriplex glauca
Atriplex halimus
Atriplex nummularia
Ceratonia siliqua
Haloxylon tamariscifolia
Kochia prostrata
Habitat
Description
W Mediterranean on marls, calcareous and siliceous soils, min. rainfall 200 mm S and SE Spain on marls, calcareous and siliceous soils, min. rainfall 200 mm Semi-arid areas
Erect branched woody shrub up to 1 m in height
Good cover and litter production
Similar to A. cytisoides
Good cover and litter production
Erect densely branched perennial up to 1.5 m in height Low densely branched evergreen aromatic perennial shrub Strongly aromatic small shrub 12–30 cm in height Shrub up to 50 cm in height, densely branched from the base
Good cover, litter production and surface protection
Dwarf shrub up to 50 cm in height
Good cover, rainfall interception
Erect stout shrubby silvery-white perennial up to 2.5 m in height Erect shrub, much branched, up to 3.5 m in height Evergreen tree or shrub up to 10 m in height. Small shrub 10–60 cm in height with articulate stems Small shrub up to 80 cm in height, with procumbent stems
Good cover, rainfall interception
Semi-arid/arid areas of Spain, France and N Africa, min. rainfall 200 mm S and SE Spain, min. rainfall 200 mm SE Spain and N Africa on saline and sandy soils, min. rainfall 200 mm Central, E and S Spain on saline and sandy soils, min. rainfall 250 mm Mediterranean Basin on marls and saline soils, min. rainfall 250 mm Exotic on clay, saline, loamy soils, min. rainfall 200 mm Mediterranean Basin, min. rainfall 250 mm SE Spain and N Africa on marl slopes, min. rainfall 250 mm Mediterranean Basin mainly on marl slopes, min. rainfall 300 mm
Ecological/bioengineering properties
Good ground cover, drought tolerant, unpalatable to grazing animals Good ground cover, unpalatable to grazing animals Good cover, rainfall interception
Good cover, shade provider Good year-round cover, rainfall interception Soil binder, good cover
Good cover, litter production
Bioengineering Principles and Desertification Mitigation
103
Table 8.4 (continued)
Species
Habitat
Ononis fruticosa
Spain and France on marls, min. rainfall 350 mm
Ononis natrix
Mediterranean Basin on dry calcareous soils, min. rainfall 350 mm
Ononis rotundifolia
SE Spain on calcareous soils, min rainfall 400 mm Spain and N Africa on marls and gypsum soils, min. rainfall 250 mm Exotic naturalized, min. rainfall 150 mm
Ononis tridenta
Opuntia ficus-barbarica
Mediterranean Basin, min. rainfall 250 mm Quercus coccifera Mediterranean Basin, min. rainfall 250 mm Quercus rotundifolia Native to Spain, Portugal and France, min. rainfall 350 mm Retama monosperma Spain on sandy soils, min. rainfall 200 mm Retama sphaerocarpa Spain and Portugal, on most soil types but prefers acid soils, min. rainfall 200 mm Rhus coriaria Mediterranean Basin, min. rainfall 400 mm Rosmarinus officinalis Mediterranean Basin
Pistacia lentiscus
Description
Ecological/bioengineering properties
Erect dwarf shrub Good cover, surface 25–100 cm in protection height, young stout pubescent stems Small shrub Good ground cover, 20–60 cm in surface protection height, erect stems very densely branched Erect branched dwarf Good cover, surface shrub 30–35 cm in protection height Erect or procumbent Good cover, surface dwarf shrub protection 15–40 cm in height Large shrub or small tree, segmented stem, member of cactus family Small evergreen tree or shrub 1–8 m in height Evergreen bushy shrub up to 2 m in height Evergreen tree or shrub
Good soil binder, litter production
Good shade provider, interception, litter production Good cover all year round, shade provider Good cover all year round, shade provider
Shrub or small tree Good shade provider, up to 3 m in height, interception, soil much branched binding Deciduous shrub up Soil binding, provides to 3 m in height, shade for summer with thin whippy annuals branches Shrub
Erosion control
Woody, much Erosion control branched evergreen shrub, either small and spreading or erect and over 2 m in height (continued overleaf )
104
Mediterranean Desertification
Table 8.4 (continued)
Species Salsola opositifolia
Salsola vermiculata
Senecio bicolor
Spartium junceum
Tetraclinis articulata
Habitat
Description
Spain and S Italy on Shrub marls, saline and nitrificated soils, min rainfall 250 mm Native to Spain, Shrub Portugal and Italy, on marl slopes, min. rainfall 300 mm year−1 Mediterranean Basin Perennial erect shrub up to 60 cm in on rocky and sandy height, much soils, min. rainfall branched with large 350 mm year−1 variegated leaves crowded at base Much branched stiff Native to rush-like bush Mediterranean Basin, min. rainfall 350 mm year−1 Native to Large shrub with Mediterranean densely arranged Basin, especially branches in SE Spain, min. rainfall 250 mm year−1
Ecological/bioengineering properties Erosion control
Erosion control
Good cover and surface protection
Erosion control
Good cover, rainfall interception
Sources: Tutin et al. (1964–1979), Vedel (1968), Goodin and Nottingham (1985), Correal et al. (1987). means of propagation has been developed. This provides the basis for the selection of plants to fulfil specific purposes within a revegetation programme, e.g. providing good ground cover, high levels of litter production or filling gaps in a proposed ecological succession.
5
CONCLUSIONS
The findings described in this chapter suggest that future research should concentrate on developing ecological successions and revegetation methods that promote a sustainable, high value of canopy or ground cover. This will require further knowledge of how different combinations of plant species compete in their search for water and nutrition in the often harsh environments of the Mediterranean.
ACKNOWLEDGEMENTS The authors would like to thank the staff of CEBAS-CSIC and the Geography Department of the University of Murcia, Murcia, Spain for their help with the fieldwork. Work on this topic was undertaken in the MEDALUS III project supported by the European Commission. The funding of the European Commission (project EV5V-CT92-0165) is gratefully acknowledged.
Bioengineering Principles and Desertification Mitigation
105
REFERENCES Brown LC, Foster GR and Beasley DB (1989) Rill erosion as affected by incorporated crop residue and seasonal consolidation. Transactions of the American Society of Agricultural Engineers 32, 1967–1978. Christiansen JE (1942) Irrigation by Sprinkling. Agricultural Experiment Station, University of California, Bulletin No. 670. Cogo NP, Moldenhauer WC and Foster GR (1984) Soil loss reduction from conservation tillage practices. Soil Science Society of America Journal 48, 368–373. Coppin NJ and Richards IG (1990) Use of Vegetation in Civil Engineering. CIRIA, Butterworths, London. Correal E, Sanchez P and Alcareaz F (1987) Woody species (trees and shrubs) of multiple value for the arid and semi-arid zones of North Mediterranean EEC countries. Seminar on ‘Les Especes Ligneuses A Usages Multiples Des Zones Arides Mediterraneenes’, Instituto Agronomico Mediterraneo de Zaragoza, September 1987. Elwell HA and Stocking MA (1976) Vegetal cover to estimate soil erosion hazard in Rhodesia. Geoderma 15, 61–70. FAO–UNESCO (1974) Soil Map of the World , 1:5 000 000, Volume 1, Legend. UNESCO, Paris. Francis CF and Thornes JB (1990) Mattoral: erosion and reclamation. In J Albaladejo, M Stocking and F Diaz (eds) Soil Degradation and Rehabilitation in Mediterranean Environmental Conditions. CSIC, Murcia, Spain, pp. 87–116. Goodin JR and Nottingham DK (1985) Plant resources of arid and semi-arid lands. Academic Press, Sidcup, Kent. Hudson NW (1964) The Flour Pellet Method for Measuring the Size of Raindrops. Department of Conservation and Extension, Research Bulletin No. 4, Salisbury, Rhodesia. Hudson NW (1996) Soil Conservation. Batsford, London. Imeson A, Cammeraat LH, Cerd`a Bolinches A, Prinsen H and Garcia Alvarez A (1994) Field site integration, data base management, desertification response units. Fourth Interim Report for MEDALUS II, pp. 1–6. Lang RD and McCaffrey LAH (1984) Ground cover: its effect on soil loss from grazed runoff plots, Gunnedah. Journal of Soil Conservation of New South Wales 40, 56–61. L´opez-Berm´udez F (1993) Presentation of research area: Guadalentin, Spain. In MEDALUS II, start-up document, European Union project EV5V CT92. Morgan RPC (1991) Technical and policy options for dealing with desertification in the European community. In JL Rubio (ed.) Desertification and Water Resources in the European Community, European Parliament, pp. 339–360. Morgan RPC (1996) Soil Erosion and Conservation. Longman Scientific and Technical, Harlow. Morgan RPC and Rickson RJ (1995) Slope Stabilisation and Erosion Control: A Bioengineering Approach. E & FN Spon, London. Morgan RPC, McIntyre K, Vickers AWV, Quinton JN and Rickson RJ (1997) A rainfall simulation study of soil erosion on rangeland in Swaziland. Soil Technology 11, 291–299. Navarro F (1991) El Sistema Hidrogafico del Guadalentin. Comunidad Aut´onoma de la Region de Murcia, Murcia, Spain. Newman EI (1966) A method of estimating the total length of root in a sample. Journal of Applied Ecology 3, 139–145. Orshan G (1989) Plant Pheno-morphological Studies in Mediterranean Type Ecosystems. Geobotany 12, Kluwer Academic, Dordrecht. Peinado M, Alkaraz F and Martinez Parras JM (1992) Vegetation of South-eastern Spain. J Cramer, Berlin. Quinton JN, Edwards GM and Morgan RPC (1997) The influence of vegetation species and plant properties on runoff and soil erosion: results from a rainfall simulation study in SE Spain. Soil Use and Management 13(3), 143–148. Simanton JR and Renard KG (1992) Upland erosion research on rangeland. In AJ Parsons and AD Abrahams (eds) Overland Flow: Hydraulics and Erosion Mechanics. UCL, London, pp. 335–375. Styczen ME and Morgan RPC (1995) Engineering properties of vegetation. In RPC Morgan and RJ Rickson (eds) Slope Stabilisation and Erosion Control: A Bioengineering Approach. E & FN Spon, London, pp. 5–58. Thornes JB (1976) Semi Arid Erosional Systems: Case Studies in Spain. London School of Economics and Political Sciences, London. Tutin TG, Heywood VH, Burges NA, Moore DM, Valentine DH, Walters SM and Webb DA (1964–79) Flora Europaea, Vols 1–5, Cambridge University Press, Cambridge. Vedel H (1968) Trees and Shrubs of the Mediterranean. Harmondsworth, Penguin. Worku Bekele M and Thomas DB (1992) The influence of surface residue on soil loss and runoff. In H Hurni and K Tato (eds) Erosion, Conservation and Small Scale Farming. Geographica Bernesia, Bern, pp. 439–452.
Section IV
Physical Processes and Responses
9
Differing Responses of Greek Mediterranean Plant Communities to Climate and the Combination of Grazing and Fire
A. DALAKA, E. PAPATHEODOROU, G. IATROU, T. MARDIRIS, J. PANTIS, S. SGARDELIS, C. LANARA COOK, T. LANARAS, M. ARGYROPOULOU, K.J. DIAMANTOPOULOS AND G.P. STAMOU
Aristotele University, Thessaloniki, Greece
1 INTRODUCTION Patterns of distribution of world vegetation types have been studied since the beginning of the 19th century, particularly in relation to climate. Gradually the mechanisms through which climatic control is exercised have been determined, through the physiological responses of plants to climate factors (Woodward 1987). Mediterranean-type ecosystems have received particular attention due to their pronounced seasonality, and possibly also due to their proximity to centres of population, now and through history. Such areas were considered ideal to study the notion of convergent evolution, not only studying the components (species) but extending the idea to the structure and function of the whole community (Mooney 1977; Conacher and Sala 1998). The concept is of course also applicable to other biomes that have evolved under other sets of similar macroclimatic conditions, such as the temperate forests or the tropical forests (both rainy and seasonal). It is expected, therefore, that under similar climatic conditions communities develop an analogous composition in terms of life forms. Today in the Mediterranean region human intervention shapes the life-form composition of the present Mediterranean vegetation much more than natural forces do. This has been primarily through allowing grazing by livestock (and associated firing to enhance grasslands), and through agriculture, reforestation and any other mitigation schemes for the purpose of exploitation. Only very few plant species now share a common existence in areas having a Mediterranean-type climate on a world scale, so a direct species-by-species comparison is not possible. Nevertheless a suitable life-form classification scheme can offer a sound basis for inter-comparisons and the study of the match between communities and their environments (Begon et al. 1990). The aim of the work described in this chapter was to compare the life-form composition of the vegetation in the MEDALUS I and II project sites with available data from two other Mediterranean-type areas of the world (i.e. California and Chile) with different land-use patterns and histories. The sites are listed in Table 9.1. It was expected that, under the same conditions, the growth-form composition of a community would be similar in sites with similar environmental conditions. The convergent evolution concept predicted that under similar environmental conditions plant species would develop similar adaptive traits (e.g. sclerophyll tissue). (In this chapter, ‘life-form’ and ‘growth-form’ are used synonymously.) Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
110
Mediterranean Desertification
Table 9.1 Landscape characteristics of MEDALUS I and II sites as well as some sites from California and Chile (Miller 1981) Sites
California, USA Echo Valley Torrey Pines Park
Latitude
32◦ 55′ N 32◦ 35′ N
Chile Fundo Santa Laura 33◦ 04′ S Greece Hortiatis 40◦ 36′ 34′′ N Petralona 40◦ 21′ 38′′ Larissa 39◦ 38′
Spain Guadalent´ın Basin Almeria (Rambla Honda) Murcia (El Ardal, Mula) Sardinia (Olias, Santa Lucia)
39◦ 00′ 37◦ 07′ 55′′
Elevation Annual Mean (m a.s.l.) precipitation annual (mm) temperature (◦ C)
Dominant plant growth formsa
1000 110
476 200
13.4 15.8
Chaparral Sev2Coastal sage Ssc3- and Sscev2 scrub–chaparral– coastal pine
1000
593
12.4
Matorral
400 220 73
440 407 466
16.2 17.7 16.2
Sev2 Sev2 Sev2 and desertified slopes
300 660
270 219.3
17.7 17.75
Tev2ndl, and S-2 S-1, DS dc3 nal and GP
266.4
17.15
440.5
16.36
38◦ 04′ 34′′ 39◦ 11′ 18′′
Dominant vegetation type (local name)
130
Sev2 and Ssc3
Sev2
a T,
tree; S, shrub; DS, dwarf shrub; GA, grass annual; GP, grass perennial; HA, herb annual; HP, herb perennial; ev, evergreen; di, dimorphic; dc, deciduous; sc, semideciduous; 1, photosynthetic stem; 2, sclerophyllous; 3, malacophyllous; bdl, broad-leaved; nal, narrow-leaved; ndl, needle-leaved. For example, pine corresponds to Tev2ndl (tree, evergreen, sclerophyllous, needle-leaved).
2
METHODOLOGY
In the analysis, the Bray–Curtis ordination method was followed. It is an early ordination method but it remains efficient (Kent and Coker 1996). Each point on the resulting graph corresponds to a site, and distances between points on the graph reflect their similarities in terms of the factors controlling them (here it was the environmental parameters used and growth-form composition at each particular site). The Sorensen index was used as a similarity measure.
3
COMMUNITY STRUCTURE
The landscape characteristics most often used and assumed to have ecological implications or which have been useful in comparisons are presented in Table 9.1. A Bray–Curtis ordination diagram summarizes their similarities and differences (Figure 9.1). All the MEDALUS sites are grouped fairly closely together since they have similar properties. Almeria (southern Spain) is further apart, due to its more southerly position and climatic indexes. The American sites are also apart because of their lower latitudinal position and higher elevation (except Torrey Pines). In a Mediterranean region analogy these sites (Echo Valley and Fundo Santa Laura) are equivalent to sites somewhere in the Atlas mountains. In Table 9.2 the life-form composition in each area is presented following the scheme of Stewart and Webber (1981). Two additional categories were used: the narrow-leaved sclerophyllous tree to accommodate the Pinus life forms and the drought-deciduous scrubs to accommodate the phrygana (e.g. coastal sage) formations.
111
Responses to Climate and Combined Grazing and Fire
Torrey Pines
Axis 2
Larissa Guadalentín Sardinia Murcia
Echo Valley Almería
Petralona Fundo St Laura Hortiatis Axis 1
Figure 9.1
Two-dimensional plot of the study sites according to their land characteristics
30
30
Tev2ndl S(ev)2nal
20 Juox
10 Juox
S-3bdl S-3nal Succulent S-1
70 Baac, Ciin
1 Quco
50 Qudu
25 Piha 20 Roof, Ancy
50
25
60
5
5
45
15
20 Arca 10
25 Ancy 40 Resp
20
Fundo Santa Laura
Murcia
50 Quco, Oleu, Phla
Torrey Pines
Sardinia
45 Quco, Pile, Phla
Echo Valley
Larissa
65 Quco
Almeria
Petralona
Sev2bdl
Guadalent´ın
Hortiatis
Table 9.2 Cover per cent allocation between growth forms in seven MEDALUS and three American study sites. Categories are as described by Stewart and Webber (1981). Species named are the dominant species in these categoriesa
5 20 15 15
a Quco,
Quercus coccifera; Pile, Pistacia lentiscus; Phla, Phillyrea latifolia; Oleu, Olea europaea; Qudu, Quercus dumosa; Piha, Pinus halepensis; Juox, Juniperus oxycedrus; Roof, Rosmarinus officinalis; Baac, Ballota acetabulosa; Ciin, Cistus incanus; Ancy, Anthyllis cytisoides; Arca, Artemissia californica; Resp, Retama sphaerocarpa. Index of similarity according to Sørensen 2c Is = A+B
where c is the number of species common to two releves or communities, A is the total number of species in releve (or community) A, and B is the total number of species in releve (or community) B (Mueller-Dombois and Hellenberg 1974).
112
Mediterranean Desertification
Figure 9.2 is a Bray–Curtis ordination diagram summarizing the data of Table 9.2, where the sites are plotted according to the life-form composition of their vegetation. Visual comparison of Figures 9.1 and 9.2 shows that the use of the two different factors does not result in the same distribution of points on the graphs. This means that the two factors affect point distribution differently, and hence no strict relation exists between them (Dargie 1984). In Figure 9.1 MEDALUS sites are grouped fairly closely together. Here the difference in land uses and their effects on vegetation disperse the group. Almeria, with its extreme conditions which have a big impact on its vegetation, stays further apart. Details of the alterations due to grazing have been studied at the species level for the association Quercus coccifera–Stipa bromoides and at the community level for the site at Larissa, in Greece. In evergreen sclerophyll formations between shrub patches the soil is covered by a large number of annual species and a few perennials, among which Stipa bromoides dominates. It forms rather large tussocks of tillers which grow up to 1 m in height and seem to occupy mainly the margins of the shrub patches. The maximum density of tussocks is found in the marginal zone and the minimum density under tall shrubs and in the uncovered soil. The spatial distribution of the tussocks is almost uniform in the marginal zone and aggregated in the uncovered areas as well as under the canopies of tall shrubs where a few micro-sites are available for S. bromoides to grow. The establishment and growth of the grass is facilitated at the margins where plants appear more robust (considering the height of the tillers and their number per tussock) and produce a large number of reproductive tillers. Sharp gradients of light intensity, water and nutrient availability created across the transect from shrub clusters to the uncovered areas possibly explain the spatial arrangement of tussocks. For example, soil water potential at the surface soil layers is higher in the uncovered areas during midsummer (Table 9.3). Plants growing in such areas appear to be able to extract water and remain
Axis 2
Guadalentín
Hortiatis
Almería
Petralona Echoval Murcia
Sardinia Larissa
Torrey Pines Fundo St Laura
Axis 1
Figure 9.2 Two-dimensional plot of the study sites according to their life-form composition
Responses to Climate and Combined Grazing and Fire
113
Table 9.3 Values of water potential (in Mpa) for the soil and parts of Stipa bromoides inside and outside Quercus coccifera shrubs during August
Soil Roots Stem Old leaves New leaves
Inside shrub
Outside shrub
−4.58 −17.8 −24.6 −25.3 −25.6
−15.1 −16.9 −21.0 −21.1 −24.7
active during summer but the cost of water extraction is greater than where plants are growing under the shrub canopies. So, inside shrubs of 30 cm height we can count 2.31 tiller m−2 . This number decreases to 1.68 inside higher shrubs (60–70 cm) but reaches a maximum at the shrub periphery (2.76), while being only 0.72 in open spaces. Greater levels of nitrogen in the marginal zone may also facilitate the growth of the plants at this specific site. By growing in the marginal zone and under the canopies of low shrubs, S. bromoides avoids competition from the fast-growing annuals for nutrients, and competition from Q. coccifera for light. It also avoids, to a certain degree, the effects of trampling during the establishment phase. Under grazing pressure and depending on its intensity, which varies in time and space, the shrub boundaries and height are in a dynamic state of expansion and shrinkage. Grazing, at a certain moderate level, facilitates the establishment and growth of S. bromoides. In the absence of grazing, the growth of Q. coccifera canopies and grasses is expected. In overgrazed areas the abundance of different species is expected to diminish since soil conditions will cease to be favourable.
4 FUNCTIONAL TRENDS 4.1 Response to Nutrients
Availability of nutrients might affect the presence or absence of particular life forms. For this reason levels of nitrogen and phosphorus were studied at different sites, as well as their interaction with particular life forms. At the Hortiatis (Greece) site, the pattern of nitrogen (N) and phosphorus (P) circulation between soil and Quercus coccifera shrubs was studied. The N content in the tissues was greater during the growing period and there was no direct correlation of N with temperature and humidity. The study of the N distribution in different plant parts (leaves, wood, fine and large roots) and in three categories of plants, i.e. heavily, moderately and non-grazed, as well in the soil, showed that N concentration increases simultaneously in the soil and the grazed shrubs while showing a time lag in the non-grazed shrubs. The reason for this delay might be attributed to the larger leaf mass in the non-grazed shrubs. Unlike nitrogen, the phosphorus distribution in the Q. coccifera tissues is controlled by a combination of temperature and humidity, and shows significant differences between autumn–winter and spring–summer samples.
5 LAND USES 5.1 Disturbance Regime
Extensive grazing and fires are the main disturbance factors in all Mediterranean areas of the world. Although some authors (Stewart and Webber 1981) consider that in California urbanization is the biggest threat to the chaparral vegetation, the same cannot be said for the Mediterranean region
114
Mediterranean Desertification
itself. The combined effect of grazing and fire leads to communities where as the grazing pressure increases along with the frequency of fire, total biomass is reduced and the evergreen broad-leaved sclerophyllous species are replaced by thistles and annual grasses. The results of this study are summarized in Table 9.4, according to the species participating in each group. Taking the growth forms percentage in each group, it was observed that the effect of grazing shifted the character of the community from a shrubland to a prairie. Grouping of specimens from each sampling site gave the community types presented in Table 9.4 in terms of species composition. The Sørensen similarity index is given in Table 9.5. Figure 9.3 shows the pathways of degradation. Community 1 is more or less the climax stage. Light grazing opens up the closed structure and leads to Community 2. From this stage combinations of fire frequency and fire intensity follow either the path from Communities 5 and 6 to Communities 7 and 9 or through Communities 4 and 3 to Community 8. As can be seen from Table 9.4, Communities 8 and 9 are dominated by annuals or perennial herbaceous species and grasses. The effects of grazing on shrub morphology, seed ecology and herb layer have been studied also by Puigdefabregas (1999) for the dwarf shrub Anthyllis cytisoides (classified as narrow-leaved sclerophyllous scrub). The herb layer under and between Retama has been studied and results showed that the responses to grazing level were significant only if grazing was held constant throughout a bush’s life. Also, grazing had no impact on seed production, or the seedling development of the shrub. The effects of grazing on the herb layer led to the dominance of species associated with open semi-arid environments, as expected.
Community 1 57.1% Low grazing 45.5% Overgrazing
Community 2 57.1% Overgrazing + low frequency of fires
50% Overgrazing + high frequency Overgrazing of fires 53% Community 6
Overgrazing + high frequency of fires
Community 5
46.7%
Community 4
50% Overgrazing + high frequency of fires Community 7
75%
Community 3
Overgrazing + low frequency 57% of fires Community 8
Overgrazing + 62% high frequency of fires
Community 9
Figure 9.3 Diagrammatic representation of the degradation paths followed, according to the intensity of grazing and fire frequency
115
Responses to Climate and Combined Grazing and Fire
Table 9.4 The plant species occurring in each of the nine communities that can be distinguished along a fire-grazing intensity gradient
Plant species
Quercus coccifera Olea europaea Cistus incanus Thymus capitatus Helianthemum sp. Asphodelus aestivus Asparagus acutifolius Ballota acetabulosa Pterocephalus papposus Crataegus sp. Dactylis glomerata Phillyrea latifolia Poa sp. Chrysopogon gryllus Eryngium campestre Euphorbia cf. myrsinites Carlina vulgaris Bupleurum glumaceum Trifolium sp. Aegilops sp. Eryngium creticum Teucrium polium Scolymus hispanicus Pyrus amygdaliformis Thymus striatus Petrorhagia sp. Anthemis sp. Avena sp. Tordyllium sp. Allium sphaerocephalon Hordeum sp. Phalaris sp. Paliurus australis Melica sp. Taeniatherum caput-medusae Sideritis curvidens Nigella sp. Cynodon dactylon Lagurus ovatus Centaurea sp. Bromus sp. Sonchus sp. Carlina sp. Euphorbia sp. Vulpia sp.
Growth form
EBLSS EBLSS BLMDS NLSDS NLSDS PH PThistle BLMDS H DTree G EBLSS H G AT ENLDS AT HNL H AG HThistle ENLMDS PScrub DTree Scrub HNL H AG ABL PBL AG PG Scrub AG G H H AG AG H AG H AT NLSDS H
Communities 1
2
+ + + + + + + + + + + + + +
+
3
5
6
7
8
9
+ +
+ + +
4
+ +
+
+ +
+ + +
+ +
+
+ + + + + + + + +
+ +
+ +
+ +
+ +
+ + + +
+
+
+
+
+
+ + +
+
+
+
+
+ + + + + + + + + +
+ + +
+ + +
+ + +
+ + + + + +
+
+ +
+ + + + + + + + + + +
+ + +
(continued overleaf )
116
Mediterranean Desertification
Table 9.4 (continued)
Plant species
Growth form
Communities 1
Silybum merianum Marrubium peregrinum Verbascum sp. Alyssum sp.
2
3
4
5
6
7
8
9 + + + +
AT Scrub, hairy leaves H H
EBLSS, evergreen broad-leaved sclerophyllous shrub; BLMDS, broad-leaved malacophyllus deciduous scrub; NLSDS, narrow-leaved summer deciduous shrub; PH, perennial herb; AG, annual grass; ABL, annual broad-leaved; PBL, perennial broad-leaved; PG, perennial grass; AT, annual thistle; ENLDS, evergreen narrow-leaved dwarf scrub; ENLMDS, evergreen narrow-leaved malacophyllus dwarf scrub; A, annual; P, perennial; H, herbaceous; G, grass; D, deciduous Table 9.5 The values of the Sørensen similarity index for the communities described in Table 9.4 based on species composition
Com.1 Com.2 Com.3 Com.4 Com.5 Com.6 Com.7 Com.8 Com.9
Com. 1
Com. 2
Com. 3
Com. 4
Com. 5
Com. 6
Com. 7
Com. 8
57.1% 57.1% 75% 30% 38.5% 46.7% 25% 37.5%
11% 53% 31.6% 10.5% 50% 40% 62%
22% 50% 23.5% 7.1% 21.1% 42%
21.4% 45.5% 24% 22% 20.1%
33.3% 32.3% 57.1% 21%
5.5% 26.7% 35.3%
0% 43.5%
7.1%
Com. 9
5.2 Reforestation As grazing and fire (especially fire) represent powerful means of changes in the plant species and accordingly the growth-form composition in any given community, it was judged useful to study the impact of mitigation schemes, specifically reforestation. Reforestation represents another way in which the life-form composition of an area can be changed, though this time with good intentions. Results from a census of 12 reforested areas in mainland Greece are presented in Table 9.6. They show that species number, species diversity and evenness in natural ecosystems of broad-leaved sclerophylls are higher than in reforested areas, and suggest that these communities in continental Greece are well established, with remarkable homogeneity in their structural characteristics, independent of the management practices that were applied at each site. However, in the reforested areas, the values of the community characteristics are significantly lower than those in the maquis ecosystems, revealing the presence of immense differences in the community structure of these two types of ecosystems. Factors interacting include the species used in reforestation and the year of plantation. In addition, species number, species diversity and evenness in the reforested areas decrease as time from the year of plantation passes, since strong dominance of the pine species leads to the extinction, under their canopy, of most of the other woody species. Thus, minimum values of the above-mentioned community structure parameters are observed in the oldest plantations. This indicates that with reforestation, as a restoration practice, simplified ecosystems are developed, strongly differing in species composition and structure from the initial natural vegetation. Even if in the first years of reforestation there exist similarities between the two types of ecosystems (maquis and reforestation), as time passes their divergence gradually increases, resulting in two different ecosystems after 70–80 years.
117
Responses to Climate and Combined Grazing and Fire
Table 9.6 Results of the census in 12 reforested areas in mainland Greece and the neighbouring community of evergreen broad-leaved sclerophylls. The year is the date of reforestation. Indexes used are species number (S), Shannon-Wiener (H’) and Evenness or Pielou (J)
Sites
Nafpaktos Nafpaktos Atalante Atalante Ermione Tristinika Taygetos Rapsani Polygyros N Marmaras Stylida Stylida Kavala Basilica Basilica Mt Pelion
Year
1914 1948 1928 1936 1965–1966 1979 1951 1957 1970 1974 1937 1953 1945–1947 1958 1966 1970
Evergreen broad-leaved S
H’
J
8 ” 8
1.67 ” 1.69
0.81 ” 0.81
12 8 9 8 10 12 8
1.53 1.84 2 1.47 1.66 1.78 1.63
0.62 0.89 0.91 0.71 0.72 0.72 0.78
11 12
2.08 1.83
12
1.78
0.87 0.74 2 0.72
Reforestation S 3 4 3 3 8 7 3 4 5 4 5 6 6 3 0.2 8
H’
J
0.29 0.80 0.31 0.33 1.07 1.48 0.27 0.25 0.62 0.87 0.64 1.04 0.94 0.15 1 1.24
0.26 0.31 0.29 0.30 0.52 0.76 0.24 0.18 0.39 0.63 0.4 0.58 0.52 0.13 0.30 0.60
6 CONCLUSIONS The concept of convergent evolution of species under similar environmental conditions was used in order to evaluate the evolution of Mediterranean plant communities, especially those in Greece. In the case of communities, the growth-form composition was used, instead of species characteristics. The Bray–Curtis polar ordination method was used and the per cent participation of growth forms in its composition was used to show community characteristics. Ordination of sites according to their environmental characteristics produced different results from when the growth-form composition basis was used. This finding persisted when examining sites at a global, regional or local level. So, while specific characteristics at the level of species seem to converge, at the community level the growth-form composition may vary considerably. The impacts of grazing and fire on soil nutrients and community composition were investigated. Results showed that while grazing affected mainly individual species (such as the shrub Quercus coccifera) it was fire or “corrective actions” such as reforestation that changed species composition and altered the community status. Also, the duration of that action rather than specific environmental parameters seemed to be the cause of divergence between communities. This indicates how carefully land use has to be planned when acting to reverse desertification (Thornes and Brandt 1997). Further work on the life-form classification and participation in the structure of the vegetation in an area can help us gain a better understanding of the relationship between vegetation and environmental conditions, in the inter-comparisons between areas of the Earth geographically apart but having the same macroclimate (Oechel et al. 1981). Such inter-comparisons are useful, especially for the Mediterranean region itself since the intensity of the land use over the centuries has profoundly changed the natural vegetation and its supporting landscape. Life-form structure may be the common link for a better design of the corrective measures and the exploitation practices that are needed.
118
Mediterranean Desertification
REFERENCES Begon M, Harper JL and Townsend CR (1990) Ecology, Individuals, Populations and Communities. Blackwell, London. Conacher AJ and Sala M (eds) (1998) Land Degradation in Mediterranean Environments of the World . John Wiley, Chichester. Dargie TCD (1984) On the integrated interpretation of indirect site ordinations: a case study using semi-arid vegetation in south-eastern Spain. Vegetatio 55, 37–55. Kent M and Coker P (1996) Vegetation Description and Analysis. John Wiley, Chichester. Miller PC (ed.) (1981) Resource Use by Chaparral and Matorral . Springer-Verlag, New York. Mooney H (ed.) (1977) Convergent Evolution in Chile and California. Dowden, Hutchinson & Ross, Pennsylvania. Mueller-Dombois D and Hellenberg H (1974) Aims and Methods of Vegetation Ecology. John Wiley, New York. Oechel WC, Lawrence W, Mustafa J and Martinez J (1981) Energy and carbon acquisition. In PC Miller (ed.) Resource Use by Chaparral and Matorral . Springer-Verlag, New York, pp. 151–183. Puigdefabregas J (1999) The Rambla Honda Field Site. MEDALUS III Core Project Final Report, London, pp. 13–20. Stewart D and Webber PJ (1981) The plant communities and their environments. In PC Miller (ed.) Resource Use by Chaparral and Matorral . Springer-Verlag, New York, pp. 43–67. Thornes JB and Brandt JC (1997) Final Report for MEDALUS II , vols 1–4. MEDALUS II, London. Woodward FI (1987) Climate and Plant Distribution. Cambridge University Press, Cambridge.
10
Vegetation Cover Assessment in Mediterranean Semi-arid Landscapes
´ ´ M.A. GILABERT AND M.T. YOUNIS F.J. GARCIA-HARO, J. MELIA,
` Remote Sensing Unit, Universitat de Valencia, Spain
1 INTRODUCTION In semi-arid areas, routine techniques available to estimate parameters of plant canopies from reflectance data of broadband sensors are limited (Elvidge et al. 1993). For example, vegetation indices such as the normalized difference vegetation index (NDVI), which combines the red and near-infrared parts of the spectrum, have been commonly used in remote sensing to observe vegetation patterns and dynamics. The NDVI has a high correlation, at the local scale, with various plant parameters such as LAI (leaf area index), biomass, production and percentage cover (Justice 1986). However, the NDVI is affected by soil spectral properties (Huete et al. 1985; Baret and Guyot 1991), particularly in areas with sparse vegetation where the observations are strongly influenced by the signal from the soil background (rock, soil, and litter materials). Alternative classification techniques which assign a unique cover class to each pixel according to Boolean algorithms are insensitive to the continuous variation of the ground vegetation cover of mixed pixels. A different technique, namely linear spectral mixture modelling (LSMM), has been developed in recent years to extract land-cover information at a sub-pixel level. This procedure divides each ground resolution element into its constituent materials using endmembers that represent the spectral characteristics of the cover types. This modelling approach has been successfully used in geological applications (Adams et al. 1986; Thomson and Salisbury 1993), climatological studies (Rambal et al. 1990) and vegetation studies (Smith et al. 1990). Certainly, in semi-arid landscapes the vegetation cover varies at scales characteristically smaller than the image pixel size. Since satellite observations integrate the radiance of all elements within the pixel, techniques to invert fractional coverage of components provide a reliable representation of ground surfaces. The basic physical assumption is that there is not a significant amount of photon multiple scattering between the macroscopic materials, in such a way that the flux received by the sensor can be expressed as a summation of the fluxes from the cover types (macroscopic materials) and the fraction of each one is proportional to its covered area. When applied to multispectral satellite data, the result is a series of images, each depicting the abundance of a cover type. LSMM has been shown to be a valuable remote sensing tool to assess the compositional distribution of vegetation and its change at a regional scale and also to accurately estimate its areal extent for yield or inventory purposes (Garc´ıa-Haro 1997). In the frame of the MEDALUS project we have been interested in producing a remotely sensed derived map, sensitive to vegetation activity and quite independent of the soil background, in a pilot area, the Guadalent´ın Basin (south-east Spain). The semi-arid features of this landscape and its high lithological variability are suitable for the application of the LSMM procedure, to obtain vegetation maps more accurately than with traditional remote sensing techniques such as those based on traditional vegetation indices. These maps will be incorporated into models developed to study the physical processes involved in desertification at a basin scale. The unmixed vegetation fraction is not, however, a straightforward estimation of the vegetation amount; careful analysis is required. In fact, the correlation between the derived vegetation fraction Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
120
Mediterranean Desertification
and the vegetation cover is not yet well understood and seems to depend on the type of vegetation canopy. Moreover, the derived vegetation fraction is sensitive to a great number of factors, the most important being as follows: 1.
the location and number of wavebands of the spectral data (non-linear effects caused by the multiple scattering of radiation are wavelength dependent); 2. the radiometric accuracy and precision of the spectral data; 3. the presence of shadows, which should be interpreted to measure the fractional amount of vegetation; 4. the selection of the set of endmembers. The difficulty of unambiguously characterizing the spectral and biophysical properties of the endmembers could introduce an additional bias in resulting fractions. Moreover, the number and spectral properties of endmembers are critical (Sabol et al. 1992). The performance of the LSMM to extract vegetation-related parameters has been tested and is described in section 2. In this case, high spectral resolution reflectance measurements of soil–plant mixtures with different soil brightness and plant densities were carried out in a laboratory radiometry experiment under controlled conditions. In section 3, the feasibility of the model inversion to retrieve canopy parameters (basically the vegetation fraction) is discussed. This was validated using simulated multispectral data derived from a canopy reflectance model (Garc´ıa-Haro et al. 1997a). These artificial data have enabled us to examine and control the influence of dominant canopy parameters and facilitate the interpretation of the results. Sections 2 and 3 show the importance of carefully selecting the endmembers used in the inversion procedure. In fact, identifying and estimating the spectral signature of components that form the scene is a crucial point in the application of this methodology, since they vary depending upon the scale and purpose of the study. In order to solve this problem, three new methods are proposed in section 4 that can be used to estimate the spectra of pure components from a set of unknown mixture spectra. The different procedures have been tested for the case of three endmembers using simulated and real data corresponding to mixtures of vegetation and soil. Factors that limit the accuracy of the results, such as the number of channels and the level of data noise, have been analysed. Finally, these procedures have been applied to a Landsat-5 TM scene of the study area. The results from this application have highlighted the suitability of this approach for the analysis of the biophysical and compositional character of ground surfaces.
2
DEVELOPMENT AND TESTING OF THE LINEAR SPECTRAL MIXTURE MODELLING APPROACH: LABORATORY EXPERIMENT
A laboratory experiment under controlled conditions was designed in order to 1. verify how well the LSMM performs; 2. test the feasibility of the spectral unmixing procedure; 3. interpret the reflectance values in Landsat-5 TM wavebands. High-resolution reflectance spectra (from 0.4 to 2.5 µm) were acquired from plots with different amounts of vegetation. The classical estimator constrained to normalization condition (i.e. the analytical least-squares method using Lagrange multipliers) was applied to estimate the fractions of components (Garc´ıa-Haro et al. 1996), since it is fast and provides an unbiased solution (Settle and Drake 1993; Garc´ıa-Haro 1997). Its required inputs are the measured reflectance spectra of the sample, its covariance matrix (which is likely to be diagonal) and the signature of the hypothesized components. A set of 21 plots was designed, consisting of seven different vegetation densities and three different soil backgrounds. Only one vegetation species was considered, Quercus ilex “rotundifolia”, which is common in the Mediterranean Basin. Plots with different vegetation amounts were obtained by planting Holm oak plants of about 25 cm height in closely
Vegetation Cover Assessment in Semi-arid Landscapes
121
spaced holes uniformly distributed over boxes of dimensions 29 cm × 42 cm, providing canopies with a quasi-planophile leaf distribution. The original soil background was mainly composed of red clay conglomerates. The effect of the soil spectral properties was modelled by covering the soil with two different densities of coal (16 and 40 g m−2 ) in order to account for the brightness–colour effect. Coal has a low reflectance in all wavelengths and therefore it spectrally minimizes the soil contribution. Reflectance spectra measured include those for the 21 designed plots along with those corresponding to the pure endmember selected. Three replicates of each observation were made and the average spectra were used. Finally, filters reproducing the TM bands were applied. This sensor was used since it is well suited for vegetation monitoring (Franklin et al. 1986). According to the LSMM principles, spectral measurements of vegetation canopies were modelled through linear combinations of spectra of vegetation, background (considered as a macroscopic mixture of soil and coal) and shadow. Figure 10.1 shows an example of the LSMM performance corresponding to a plot with LAI = 0.56 and 16 g m−2 coal. It reveals the close agreement between the measured and modelled spectra. In fact, the residuals (i.e. the differences between the measured reflectance and the reflectance predicted by the model) were very low in all cases, presenting values generally similar to the precision of the measurements. This result was confirmed by the chi-square test, concluding that the measured spectra were reproduced by the modelled ones with a confidence level of 95%. It was found that residuals are not randomly distributed as would be expected if they were attributable to experimental noise. There are some regions in which they seem to present slightly higher values than the experimental accuracy. In fact, the main sources are systematic errors attributed to the unmixing procedure, e.g. the presence of components not considered; mixing non-linearity in some regions of the spectra; and inaccuracies in the definition of endmembers, particularly those of vegetation and shadow. Figure 10.2 shows the NDVI values as well as the unmixed values of vegetation fraction versus LAI. We can observe that the vegetation fraction obtained using LSMM is well correlated with LAI, and less sensitive to soil optical properties than the NDVI. Also it does not appear to saturate for high values of LAI as has occurred with NDVI. The most outstanding feature, however, is that the effect of the soil optical properties has been reduced. Despite this, the vegetation fraction appears to diminish when the soil is blackened. The effect of darkening the soil causes an increase in the unmixed fraction of coal and a similar diminution of the soil endmember, without altering significantly the fraction of vegetation. However, estimation of the vegetation fraction shows a systematic error, roughly 0.03, basically due to
20
Reflectance (%)
17 14 11 8 5 2 −1 0.4
0.7
1
1.3
1.6
1.9
2.2
2.5
l (µm)
Figure 10.1 Measured (solid line), modelled (dotted line) and residual (dashed line) spectra for the plot with LAI = 0.56 and 16 g m−2 coal. The straight grey line represents the zero base line
122
Mediterranean Desertification (a)
1
NDVI
0.8 0.6 0 g m−2
0.4
16 g m−2 40 g m−2
0.2 0 0
0.5
1
1.5
2
2.5
(b)
1
Vegetation fraction
LAI
0.8 0.6 0 g m−2 0.4
16 g m−2 40 g m−2
0.2 0 0
0.5
1
1.5
2
2.5
LAI
Figure 10.2 (a) NDVI values versus LAI for the three soil backgrounds. (b) Vegetation fraction, as derived from the LSMM, versus LAI for the three soil backgrounds. Symbols refer to plots with 0, 16 and 40 g m−2 coal. Reproduced by permission of Taylor & Francis Ltd. (International Journal of Remote Sensing, 1996, 17: 3373–3400)
inaccuracies associated with experimental noise. Similar low errors were obtained using the measurements of reflectance in TM bands. This fact confirms the validity of the modelling approach in vegetation-related studies, even for the application of data from satellites with a limited number of spectral bands. In conclusion, the demonstrated capacity of LSMM to separate out the soil contribution makes it suitable for mapping the vegetation in semi-arid areas quantitatively. The accuracy attainable by this modelling approach can be considered sufficient for many practical purposes, being operational in the monitoring of vegetation from satellite data. In the case of Landsat-5 TM data, accurate estimates of the three dominant endmembers seem to be possible in environments with vegetation.
3
MODELLING PHYSICAL SCENES TO ASSESS THE SENSITIVITY OF SPECTRAL INDICATORS
The retrieval of parameters from remotely sensed data is constrained by the complexity of the physical processes involved. Variations in the radiance received at the sensor associated with variations in the illumination geometry are difficult to disentangle and quantify. In addition, the estimate of canopy structural parameters is limited when dealing with real data, particularly in studies at a regional scale. In fact, most canopy parameters are highly correlated and thus variations of one parameter may be erroneously attributed to variations of some others. In order to examine and quantify the main sources of error in remote sensing data, artificial images have been generated using the principles of geometric models, e.g. where the sensor integrates the
Vegetation Cover Assessment in Semi-arid Landscapes
123
radiance from composing materials (plant, shaded soil and illuminated soil). Multispectral images were generated using a process based on simulated, three-dimensional trees and a light–canopy interaction model. This modelling approach is linked with the interpretation of the LSMM and seems to be a suitable remote sensing tool to study heterogeneous canopies. It also served to gain an insight into the scaling processes involved in the relationship between remotely sensed data (particularly, the LSMM-derived fractions) and land surface variables. Radiative transfer modelling in discontinuous canopies requires a geometrical representation and a spatial distribution of the vegetation elements. In this work solid symmetrical shapes (ellipsoids and cylinders) were used to represent crowns and trunks of trees, which were distributed randomly in space on a contrasting flat background. Two different shapes were used to replicate pine trees and bushes. Plant dimensions followed a Poisson distribution within the scene. Several scenarios were generated using various vegetation densities, with average vegetation cover from 4 to 50% (Figure 10.3). Shadows were computed by means of a Lambertian reflectance model, which was used due to its simplicity and generality (St-Onge and Cavayas 1995). To generate the shadows a geometrical method, based on the intersection of crowns and trunks with the direction of both the sun and the sensor, was developed. This method was demonstrated to be more efficient than the ray tracing or Monte Carlo procedures (Gerard and North 1997). The output is a bi-dimensional grid showing the
Image Pixel
aggregation of pixels aggregation of cells
ZOOM
qs
qv
Pixel size = 30 m
Figure 10.3 Representation of a simulated physical based scene. Tree dimensions have been amplified for viewing purposes
124
Mediterranean Desertification
proportions of four components: shaded ground, illuminated ground, shaded soil and illuminated soil. Different sun positions at satellite overpass time were considered. We considered different levels of aggregation which allowed us to analyse the spatial heterogeneity effect in the inter-pixel variance. The reflectance of each pixel was calculated using the spectra of the individual components as inputs. We mixed them according to the estimated proportions in order to derive noise-free satellite imagery. Data quality was then diminished to simulate the instrumental noise and the systematic errors associated with atmospheric alteration and sensor degradation. The level of noise was modelled as a Gaussian variable. In order to evaluate the influence of soil background optical properties, different soils (quaternary, marl, phyllite and red clay soil) were considered. Then, LSMM was applied to un-mix the fractions of vegetation, soil and shadow, paying special attention to vegetation. It was observed that the instrumental noise limits the spectral contrast between components, producing, for LSMM estimates of each fraction, a distribution of values around the “real” value. The variance of the distribution is proportional to the instrumental noise. The scale and pattern of shadows affect results indirectly since they determine the spectral response and composition of the scene. In general, the greater the number of endmembers, the lower the accuracy of the estimations and the greater the time needed for computations. In fact, when applying LSMM from data with a small number of spectral bands, such as Landsat-5 TM data, the accurate estimation of a number of endmember proportions larger than three is limited when the spectral separability of the materials is small, regardless of the method used to extract the endmembers (Garc´ıa-Haro 1997). Furthermore, the attainable accuracy (which is proportional to the hyper-volume occupied by the endmembers’ positions in the features space) considerably diminishes when one endmember is collinear in some spectral regions with the rest of the endmembers (Sabol et al. 1992; Garc´ıa-Haro 1997). For example, the low spectral contrast between brown herbage and soil makes it very difficult to separate them. As a result, a TM scenario cannot support the computation of four endmembers with the desired accuracy. Overall, the classical estimator, constrained by the normalization condition, proved to be the most accurate analytical un-mixing method. In fact, when the normalization condition is not applied, statistical and specially systematic errors of estimated fractions are unacceptably high. Furthermore, this analytical method provides outcomes for the fractions similar to more sophisticated numerical procedures such as MIGRAD or SIMPLEX algorithms (Minuit 1992), but it is faster and, therefore, particularly appropriate to be applied on large satellite areas monitored at different time-scales. The study has also shown that an important source of systematics is the radiometric accuracy of the original spectral data, mainly attributable to inadequate knowledge of calibration constants (Markham and Barker 1985) and radiometric alterations (atmospheric and topographic). Nevertheless, the most limiting factor that determines the accuracy of LSMM is the estimation of the endmembers. For example, error in a component spectral signature will influence the estimated vegetation fraction in direct relationship to the proportion of the component incorrectly modelled. Moreover, the error of the vegetation fraction is significantly higher when the erroneous endmember is that of vegetation. Due to its importance, the problem of endmember estimation is addressed in the next section.
4
EXTRACTION OF ENDMEMBERS FROM SPECTRAL MIXTURES
Three original methods have been developed to identify the number and the spectral signature of the major components from a set of mixtures spectra. A detailed description of the methods can be found in Garc´ıa-Haro et al. (1997b, 1999a). The large dependence between reflectance at different regions implies that the same amount of variability can be carried by a smaller number of new variables, each of which would be mutually independent. These abstract variables (representing a new coordinate reference axis) can be defined and related to the original ones via linear equations, operations that are termed factor analyses. The first method entails deducing the best possible rotation of the factor abstract axes to “align” them upon the directions of the real component spectra. The second method
125
Vegetation Cover Assessment in Semi-arid Landscapes
is based on the position of all the objects in the reduced “factor space”. Both methods employ a minimization procedure of an objective function appropriately defined. The third method consists of the design of a neural network whose architecture implements the LSMM principles. Endmembers are selected from the data space itself, and therefore, the methods are less problematic than library based methods which can hardly take into account all processes and factors influencing the data, which is especially critical when the number of spectral channels is reduced. A detailed formulation of the three methods can be found in Garc´ıa-Haro et al. (1999a). The three methods have been validated for the case of three endmembers. In fact, although it is theoretically possible to apply each method to a greater set of endmembers, the accuracy of the estimates is conditioned by the real dimensionality of the scene and the contrast of the materials (it is proportional to the hyper-volume of the space defined by the data). High spectral resolution spectra from 0.4 to 2.5 µm corresponding to three components – vegetation, soil and coal (which was introduced to address the influence of soil brightness) – were mixed in appropriate proportions, obtaining a set of 21 spectral mixtures. We subsequently combined them with experimental noise to simulate a real situation. Logically, the deviation between original and estimated spectra was directly proportional to the level of noise (0.5, 1.0, 1.5 and 2.0). This residual also increased with decreasing number of channels (as shown in Figure 10.4) since the Gaussian noise becomes more apparent when the data dimensionality decreases (Sabol et al. 1992). However, what is important is not only the number of wavebands but also the width and location of the spectral channels. For example, the performance attained from TM wavebands was better than from spectra sampled at 200 nm, despite the fact that the dimensionality of the TM data is smaller. These results are in agreement with results obtained by other authors (Sabol et al. 1992). Results demonstrated the validity of the three methods, reproducing the original data for a 95% confidence level. In general, the three component spectra are well regenerated in all regions, although the performance seems to be less accurate for the coal component in the near-infrared region and for the soil component in the middle-infrared region. We have also analysed the performance of the method on the basis of the deviation (residual) between the actual and the estimated values of matrices. Figure 10.4 shows the values of the root mean square (RMS) residual of the endmembers matrix, R, as obtained using methods 2 and 3. The analysis of errors showed that inaccuracies of the pure component spectra are below the introduced noise and that the errors of the fractions are relatively small in all cases. In general, the RMS residual of estimated spectra increases with increasing level of noise and decreasing number of wavebands (mainly in methods 1 and 3). Typical values of the RMS residual were below 0.2 for endmembers reflectance and 0.04 for fractions. The error of estimated parameters supplied by the minimization algorithm was similar. Results indicated that although method 3 seems to be more (a) 5
(b) 5
RMS (R)
4 3 2 1
0.5 1.0 1.5 2.0
4 RMS (R)
0.5 1.0 1.5 2.0
3 2 1
0
0 0
50
100
150
Wavebands
200
250
0
50
100 150 200 Wavebands
250
Figure 10.4 Values of the RMS residual of the endmembers matrix, R (expressed in per cent reflectance units), versus the number of bands for different levels of noise (0.5, 1.0, 1.5 and 2.0, also expressed in reflectance units), as obtained using (a) method 2 and (b) method 3
126
Mediterranean Desertification (b) 35
(a) 35 Vegetation
30 Soil
25 20 15 10 Coal
5
Reflectance (%)
Reflectance (%)
30
Vegetation Soil
25 20 15 10 Coal
5 0
0 0.4 0.7
1
1.3 1.6 1.9 2.2 2.5 l (µm)
0.4 0.7
1
1.3 1.6 1.9 2.2 2.5 l (µm)
Figure 10.5 Original (solid lines) and estimated (dashed lines) spectral signatures in TM bands of the three pure components. Case (a) corresponds to external endmembers, i.e. with the absence of the three pure data points, while case (b) corresponds to the absence of mixtures on the soil–coal edge
accurate in general, method 2 is more accurate when the number of wavebands is reduced. For example, the error does not diverge when the number of wavebands is very small, which indicates the suitability of this method when applied to data with a reduced dimensionality, such as TM (as confirmed subsequently). It should be noted that the accuracy of the outcome depends on the quality of the sample. Figure 10.5(a) shows the results obtained by method 2, applying it from a set of 18 mixtures with the absence of the three pure data points. It reveals that endmembers are not particularly affected by taking off the three samples of pure endmembers, which indicates that the method does not require the presence of vertex-extreme samples. For example, estimated outcomes represent purer spectra than could be found in any of the mixtures. Figure 10.5(b) corresponds to the case of 16 mixtures with the absence of five of the six data points on the soil–coal edge. It can be observed that the method addressed the spectra of the three components reasonably well, particularly those of vegetation and soil. In conclusion, though the methods may identify external endmembers, samples should express the variability of the system of interest and present a significant variation in the mixture proportions. We then extended results to a more realistic case using data from a set of 21 mixtures of vegetation, soil and coal measured in the laboratory (Garc´ıa-Haro et al. 1996), as described in section 2. Designed plots represented seven varying amounts of vegetation characterized by leaf area index (LAI) values from 0 (bare soil) to 2.4 (close to 100% coverage). The spectra estimated by the three methods resembled the spectral signature of vegetation, soil and coal. Assuming residuals to be normally distributed, the minimum of the chi-square of the model fit is a chi-square variable. This feature let us apply the chi-square test of the fit performance of the model to disprove the null hypothesis that for p = 0.05 (95% confidence level) the null hypothesis cannot be invalidated for any of the spectra except for the case of LAI maximum without coal (this case could present non-accountable experimental inadequacies), even from data with a reduced dimensionality. This result confirms the validity of the method. Results of the application of the proposed methods to a TM scene are shown in the next section. Concluding, the three methods provide accurate estimates of the spectral endmembers, especially the third method, which is based on a neural network. They provide a deep insight into the nature of spectral endmembers, offering new possibilities in imaging spectroscopy. Moreover, the second method, which is based on the exploration of the mixture positions in the factor space, has been demonstrated to be the most appropriate when the dimensionality of the data is reduced, as in the case of a TM data set. Further research dealing with TM images confirmed these results and showed the capability of method 2 to derive suitable endmembers of soil and different types of semi-natural vegetation.
Vegetation Cover Assessment in Semi-arid Landscapes
127
5 A CASE STUDY: APPLICATION OF THE LSMM TO THE MEDALUS TARGET AREA IN SPAIN The LSMM was applied to Landsat Thematic Mapper (TM) images over the Guadalent´ın Basin, a MEDALUS target area in Spain. This area shows high variability of lithological exposure, dominated by carbonate, marl, quaternary deposits, phyllite, greywackes and gypsum. Quaternary deposits include pebbles of various sizes and compositions of sedimentary and metamorphic origin (Younis et al. 1997). The procedure previously described in this chapter was applied to a Landsat-TM scene to un-mix the spectra of soil and different types of semi-natural vegetation. TM data were obtained from a Landsat-5 overpass taken on 7 April 1993 (a floating window from the scene 199/34 of 80 × 60 km in size, and limited by the UTM coordinates: Xmin = 565 000; Xmax = 645 000; Ymin = 415 300; Ymax = 421 300) which comprises the major part of the Guadalent´ın Basin. The image was atmospherically corrected using an invertible atmospheric radiative transfer model (Gilabert et al. 1994). As mentioned before, the extraction of endmembers was carried out following method 2, which was selected due to the optimal performance offered on data with a reduced dimensionality such as those corresponding to TM images. In order to control soil influence, we analysed the zones presenting uniform soil background separately. This was done by utilizing a lithological classification of the area (Meli´a et al. 1995). Once the major components were identified, the LSMM was applied to estimate the fractions of each component in each pixel image. As an example, Figure 10.6 shows the estimated spectra of the three endmembers for areas with a quaternary background, where semi-natural vegetation frequently grows. The application of this methodology to the areas with different lithologies produced similar values: a vegetation spectrum, a soil spectrum and a third endmember corresponding to shadow. Results demonstrated the validity of the method. For example, the soil endmember is well related to the ground spectral signature of soil. Similarly, the vegetation endmember resembles the typical spectral signature of green vegetation and is quite independent of soil background. Furthermore, estimated endmembers guaranteed the acceptability of the fractions. The un-mixing procedure was applied to obtain different images depicting the abundance of the components in the scene. Plate 1 (in colour plate section) shows a fraction vegetation image derived from applying the procedure described to the TM image of the Guadalent´ın Basin. The mean RMS residual (average of all the image pixels) was around 0.02 (comparable to data error), which is indicative of a good model fit. In addition, the derived images have demonstrated the potentiality of the un-mixing procedures to address the fractional coverage of vegetation at a pixel scale. In fact, the abundance of vegetation at a sub-pixel level has been shown to be more appropriate to monitoring vegetation in semi-arid landscapes than traditional techniques, such as those using NDVI and classification procedures. For example, although the vegetation fraction has 50 end.1
Reflectance (%)
40
end.2
30
end.3
20 10 0 TM1
TM2
TM3
TM4
TM5
TM7
TM Bands
Figure 10.6 Estimated spectra of the three pure components (shadow, vegetation and soil), for pixels representing quaternary background. (Shadow = end.1, vegetation = end.2, soil = end.3)
128
Mediterranean Desertification
a relatively high correlation with NDVI, it has shown a higher range of variability, probably due to the reduction of soil influences. Furthermore, there is a large number of pixels with non-zero but very low vegetation fraction values, which means that this quantity magnitude is quite sensitive to variations in vegetation, even in areas classified as bare soil. The main drawback of the LSMM method is the non-linear mixing behaviour of natural ground surfaces due to the multiple scattering and transmission of canopy elements, as has been observed by a range of authors (Roberts et al. 1993; Ray and Murray 1996; Garc´ıa-Haro et al. 1999b). In fact, energy leaving the canopy surface is a non-linear function of the optical properties of leaves and soil. The future development of satellite technology will provide increased spatial and spectral resolution of remote sensing data, allowing for the adequate interpretation of data by means of reflectance models describing the complex process of the radiative transfer within the canopy. An additional question is to solve the inverse problem of determining vegetation structural and optical parameters with a minimum computer time. For operational purposes, the simplicity of LSMM ensures the estimation of vegetation parameters using a reduced number of spectral data. In fact, LSMM provides an interpretation of the vegetation ground coverage similar to more accurate models at this scale. Nevertheless, regardless of the method used to select the endmembers, calibration based on ground measurements is needed to convert derived fractions to absolute abundances. As a general conclusion, LSMM is recommended for semi-arid environments containing a large number of mixed pixels (e.g. the Guadalent´ın Basin), since this technique provides accurate estimates of sub-pixel land-cover composition. In these areas, where reflectance is usually dominated by the soil background, the LSMM-estimated vegetation fraction is less biased and shows a higher capability than NDVI to detect vegetation levels. In particular, this magnitude has been shown to be appropriate for monitoring the spatial distribution of vegetation patterns and to assess the temporal changes of shrub communities. A crucial point of our future work will be the analysis of the temporal variations of this magnitude for different natural vegetation species, in order to provide inputs to be incorporated into models that predict desertification at a regional scale within the frame of MEDALUS.
REFERENCES Adams JB, Smith MO and Johnson PE (1986) Spectral mixture modelling: a new analysis of rock and soil types at the Viking Lander 1 Site. Journal of Geophysical Research 91, 8098–8112. Baret F and Guyot G (1991) Potentials and limits of vegetation indices for LAI and APAR assessment. Remote Sensing of the Environment 35, 161–173. Elvidge CD, Chem Z and Groeneveld DP (1993) Detection of trace quantities of green vegetation in 1990 AVIRIS data. Remote Sensing of the Environment 44, 271–279. Franklin J, Logan T, Woodcock C and Strahler A (1986) Coniferous forest classification and inventory using Landsat Thematic Mapper imagery. IEEE Transactions on Geoscience and Remote Sensing GE-24, 139–149. Garc´ıa-Haro FJ (1997) Modelizaci´on y estimaci´on de par´ametros relacionados con la cubierta vegetal en teledetecci´on. PhD thesis, Universitat de Val`encia, Spain. Garc´ıa-Haro FJ, Gilabert MA and Meli´a J (1996) Linear spectral mixture modelling to estimate vegetation amount from optical spectral data. International Journal of Remote Sensing 17, 3373–3400. Garc´ıa-Haro FJ, Gilabert MA and Meli´a J (1997a) Sensitivity study of vegetation related parameters by simulating reflectance scenes. Proceedings of the 7th International Symposium on Physical Measurements and Signatures in Remote Sensing, 6–11 April, Courchevel, France, pp. 251–257. Garc´ıa-Haro FJ, Gilabert MA, Younis MT and Meli´a J (1997b) Estimation of pure components spectra from spectral mixtures. Proceedings of the 7th International Symposium on Physical Measurements and Signatures in Remote Sensing, 6–11 April, Courchevel, France, pp. 509–516. Garc´ıa-Haro FJ, Gilabert MA and Meli´a J (1999a) Estimation of endmembers from spectral mixtures. Remote Sensing of the Environment 68, 237–253. Garc´ıa-Haro FJ, Gilabert MA and Meli´a J (1999b) A radiosity model of heterogeneous canopies in remote sensing. Journal of Geophysical Research 72, 328–345. Gerard FF and North PRJ (1997) Analysing the effect of structural variability and canopy gaps on forest BRDF using a geometric–optical model. Remote Sensing of the Environment 62, 46–62.
Vegetation Cover Assessment in Semi-arid Landscapes
129
Gilabert MA, Conese C and Maselli F (1994) An atmospheric correction method for the automatic retrieval of surface reflectances from TM images. International Journal of Remote Sensing 15, 2065–2086. Huete AR, Jackson RD and Post DF (1985) Spectral response of a plant canopy with different soil backgrounds. Remote Sensing of the Environment 17, 37–53. Justice CO (guest ed.) (1986) Monitoring the grasslands of semi-arid Africa using NOAA AVHRR data. International Journal of Remote Sensing (special issue) 7, 1383–1622. Markham BL and Barker JL (1985) Spectral characterization of the Landsat Thematic Mapper sensors. International Journal of Remote Sensing 4, 697–716. Meli´a J, Bastida J, Gand´ıa S, Gilabert MA, Younis MT, Lopez-Buend´ıa A and Garc´ıa-Haro FJ (1995) MEDALUS II Project 2, Final Report. MEDALUS, London. Minuit (1992) Minuit Minimization Package, Function Minimization and Error Analysis, Reference Manual. CERN Program Library, Computing and Network Division, Geneva. Rambal B, Lacaze B and Winkel T (1990) Testing an area-weighted model for albedo or surface temperature of mixed pixels in Mediterranean woodlands. International Journal of Remote Sensing 11, 1495–1499. Ray TW and Murray BC (1996) Nonlinear spectral mixing in desert vegetation. Remote Sensing of the Environment 55, 59–64. Roberts DA, Smith MO and Adams JB (1993) Green vegetation, non-photosynthetic vegetation, and soils in AVIRIS data. Remote Sensing of the Environment 44, 255–269. Sabol DE, Adams JB and Smith MO (1992) Quantitative sub-pixel spectral detection of targets in multispectral images. Journal of Geophysical Research 25, 2659–2672. Settle JJ and Drake NA (1993) Linear mixing and the estimation of ground cover proportions. International Journal of Remote Sensing 14, 1159–1177. Smith MO, Susan LU, Adams JB and Gillespie AR (1990) Vegetation in deserts: I. A regional measure of abundance from multispectral images. Remote Sensing of the Environment 31, 1–26. St-Onge BA and Cavayas F (1995) Estimating forest stand structure from high resolution imagery using the directional variogram. Remote Sensing of the Environment 16, 1999–2021. Thomson LJ and Salisbury JW (1993) The mid-infrared reflectance of mineral mixtures (7–14 mm). Remote Sensing of the Environment 17, 37–53. Younis MT, Gilabert MA, Meli´a J and Bastida J (1997) Weathering process effects on spectral reflectance of rocks in a semi-arid environment. International Journal of Remote Sensing 18, 3361–3377.
11
The Impact of Rock Fragments on Soil Degradation and Water Conservation
B. VAN WESEMAEL,1 J. POESEN,2 C. KOSMAS,3 N.G. DANALATOS4 AND J. NACHTERGAELE2 1
´ ´ Departement de Geographie, Universite´ Catholique de Louvain, Belgium Laboratory for Experimental Geomorphology, Katholieke Universiteit Leuven, Belgium 3 Laboratory of Soil Chemistry, Agricultural University of Athens, Greece 4 Department of Agriculture, University of Thessaloniki, Greece
2
1 INTRODUCTION The increasing water demands for domestic and industrial use are restricting the amount of water available for agriculture in the Mediterranean region. It is foreseen that water scarcity will increase as a result of global climate change (Parry and Carter 1991). High summer evapotranspiration rates, greatly restricting water availability, characterize the climate of the Mediterranean, so that biomass production on hilly areas in the semi-arid zone of this region is largely dependent on the amount and distribution of precipitation. In this context, water conservation strategies, such as the use of mulches, become increasingly important in restricting upward water movement from the root zone to the soil surface. During a normal growing season, evaporation from the soil surface may reach up to 50% of evapotranspiration (Peters 1960). Unlike transpiration losses, evaporation losses are not unavoidable and they can be modified. Thus, in arid and semi-arid regions, farming is directed towards reducing evaporation from the soil surface using techniques such as shallow tillage and mulching. Various natural mulch materials have been reported in the literature including inorganic materials such as gravel or stones (Meyer et al. 1972; Jennings and Jarrett 1985; Gilley et al. 1986; Lopez et al. 1987; Abrahim and Rickson 1989). A large proportion of Mediterranean soils, especially those on rolling and hilly topography, contain substantial amounts of rock fragments. According to Poesen (1990), and Poesen and Lavee (1994), stony soils cover more than 60% of the land area in this region (Figure 11.1). Rock fragments are considered the main factor in preserving a favourable soil structure with a high macroporosity (Magier and Ravina 1984). The literature review by Poesen and Lavee (1994) illustrates how rock fragments at the surface or in the soil profile modify the intensity of the most important desertification processes in the European part of the Mediterranean region. Still, relatively little experimental evidence exists on the effect of rock fragments on soil moisture conservation and protection against soil erosion. Soil loss as a result of water erosion is a major component of environmental degradation and a worldwide threat to sustainable agriculture. In many Mediterranean areas, soil erosion is an increasing problem because of the increased intensity of land use over recent decades. Rock fragments may have positive or negative effects on soil properties affecting runoff and erosion. An example of the contrasting influence of rock fragments on hydrological processes was given by Magier and Ravina (1984). These authors demonstrated that increasing rock fragment content decreases the water-conducting area and thus decreases the hydraulic conductivity of uncompacted soils, whereas the opposite occurs in compacted soils. Depending on their size and placement, a surface cover Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
132
Mediterranean Desertification
Figure 11.1 Soil rich in rock fragments in a typical Mediterranean hilly landscape (photo by C. Kosmas, March 1996)
of rock fragments may be important in decreasing soil erosion by dissipating raindrop impact, decreasing the erodible surface, and decreasing the velocity of runoff water. Poesen and IngelmoSanchez (1992) concluded that rock fragments (a) protect the underlying topsoil structure against degradation by raindrop impact, (b) increase infiltration rates and therefore decrease runoff under certain conditions, (c) increase hydraulic roughness, slowing down overland flow, and (d) decrease transported sediment detached by raindrop impact. Other studies have shown that rock fragments can protect substantial amounts of water from evaporation, thus increasing water availability for vegetation (Hanson and Blevins 1979). Ravina and Magier (1984) concluded that coarse fragments contribute to improved soil physical conditions by acting as a “skeleton” that resists soil compaction. Magier and Ravina (1984) reported that orchard trees planted on soils with high amounts of rock fragments were larger, had a better developed root system and produced higher yields than on soils without rock fragments. Kemper et al. (1994) demonstrated that a 5 cm thick gravel mulch significantly decreased evaporation, allowing 80–85% of the annual precipitation to percolate through the soils to the aquifers. In the following paragraphs, the results of detailed laboratory and field studies on the effect of rock fragments on soil degradation and water conservation conducted within the framework of MEDALUS II are summarized and discussed.
2
PHYSICAL SOIL DEGRADATION
Freshly tilled soils are characterized by a rough surface, a low bulk density and a high macroporosity. Over time, topsoils become smooth and compacted due to rainfall and subsequent drainage. The increase in vulnerability to water erosion is mainly due to a decrease in water detention and an increase in velocity of runoff at the smooth soil surface as well as a decrease in infiltration rate in the compacted topsoil. Such changes in topsoil characteristics are often referred to as physical degradation and are largely controlled by aggregate stability (De Ploey and Poesen 1985). The factors determining the aggregate stability of the fine earth fraction have been extensively described (e.g. Le Bissonnais 1996). Although several authors mention the role of the coarse fraction in providing a skeleton that preserves the soil structure (e.g. Ravina and Magier 1984; Childs and Flint 1990), little is known of the effect of rock fragment content on the physical degradation of topsoils. The effects of rock fragments on physical degradation were assessed, therefore, by measuring changes in soil surface roughness and in bulk density during and after simulated rainfall in the laboratory. At a mean intensity of 71 mm h−1 , a median drop diameter of 2.2 mm and a fall height of
133
Impact of Rock Fragments on Soil Degradation
3.2 m, kinetic energy at the soil surface amounted to 15.8 J m−2 mm−1 . Total rainfall volume equalled 192.5 mm for each experiment. A homogeneous mixture of a well-structured silt loam and wellrounded river gravel (17–27 mm) was used, simulating a plough layer. Along four fixed transects the soil surface elevations were determined at 2 mm intervals using a Selcom laser microreliefmeter driven by a computer-controlled motor (R¨omkens et al. 1988). Soil surface roughness was expressed as random roughness (RR, cm), i.e. the mean standard deviation of surface elevations of de-trended transects (Bertuzzi et al. 1990). The mean of the surface elevations was used to calculate fine earth bulk density (BDfe, kg m−3 ) as well as total bulk density (BDt, kg m−3 ; van Wesemael et al. 1995a,b). The random roughness of the topsoil without rock fragments is significantly lower after 192.5 mm rainfall, whereas for the soil containing rock fragments (rock fragment content by mass: Rm = 0.52 kg kg−1 ) the opposite appears to be true (Figure 11.2). The behaviour of the non-stony soil is in agreement with the exponential decrease of RR with cumulative rainfall described by Onstad et al. (1984). In more detail, the RR of the stony soils decreases during the first 17.5 mm of rainfall due to soil aggregate breakdown and washing of fine particles in the depressions between the rock fragments. After this stage, most surface aggregates are destroyed and roughness is controlled by protruding rock fragments, while at the same time rock fragment cover increases by emergence of rock fragments from the subsurface. This process contributes to the formation of a rock fragment pavement. It should be noted that the effect of rock fragments on soil surface roughness is most pronounced for relatively low rock fragment contents (Rm < 0.3 kg kg−1 ; Figure 11.3(a)). A regression analysis by van Wesemael et al. (1995b) using the equations proposed by Onstad et al. (1984) demonstrates that BDfe increases non-linearly with cumulative rainfall for both stony and non-stony soils. A significantly lower rate of increase was only observed for soils with very high rock fragment contents (Rm > 0.7 kg kg−1 ). The effect of rock fragment content on the bulk density of the fine earth is shown by plotting BDfe after 192.5 mm of cumulative rainfall versus rock fragment content (Figure 11.3(b)). BDfe is only affected by the presence of rock fragments for Rm (a)
20
15
10 0
5
10
15
20
25
30
35
40
45
50
55
60
65
0
5
10
15
20
25
30 cm
35
40
45
50
55
60
65
(b) 24 22
cm
20 18 16 14 12 10
Figure 11.2 Evolution of soil surface roughness and compaction of freshly tilled topsoils. Measured cross-sections before (solid line) and after (gray line) 192.5 mm rainfall for (a) soils without rock fragments and (b) soils with gravels (rock fragment content by mass (Rm) = 0.52 kg kg−1 ; after van Wesemael et al. 1995a). Reproduced by permission of John Wiley and Sons Ltd
134
Mediterranean Desertification (a)
0.8
Random roughness (cm)
0.7 0.6 0.5 0.4 0.3 0.2 0
0.1
0.2
0.3
0.4 0.5 Rm (kg kg−1)
0.6
0.7
0.8
(b) 2000 1750
BD (kg m−3)
1500 1250 1000
BDfe BDt
750 500 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Rm (kg kg−1)
Figure 11.3 The role of rock fragments in reducing physical degradation of topsoils. (a) Surface roughness and (b) bulk density of topsoils containing various rock fragment contents. Random roughness, fine earth bulk density (BDfe) and total bulk density (BDt) of bare soils exposed to 192 mm of cumulative rainfall are plotted (after van Wesemael et al. 1995a,b)
values in excess of 0.3. Poesen et al. (1994) demonstrated that soil erosion decreases exponentially with surface cover. Hence, the effectiveness of a rock fragment cover is highest at low rock fragment contents. Therefore, the role of rock fragments at the soil surface is more important than that of rock fragments in the soil profile, since the latter mainly protect the soil structure at high rock fragment contents (i.e. from Rm > 0.3 upwards; cf. Figures 11.3(a) and (b); Poesen and van Wesemael 1995).
3
SOIL EROSION BY WATER
Rock fragments play an important role in controlling overland flow and sediment loss (Poesen et al. 1994). The effects of different sizes, amounts and positions of rock fragments on water erosion were studied under field conditions in experimental plots. Thirty experimental plots, 2 m × 5 m each, were
Impact of Rock Fragments on Soil Degradation
135
Figure 11.4 Experimental plots for studying the effect of rock fragments on soil erosion and water conservation (Photo by C. Kosmas, May 1994)
installed on a hillslope (slope gradient of 17%) located 30 km east of Athens (Figure 11.4). The site is characterized by a thermo-mediterranean climate (UNESCO-FAO 1963), with an average annual air temperature of 17.8 ◦ C and an annual precipitation of 496 mm. The soil is a well-structured, dark, stony calcareous clay loam, formed on marl interbedded with sandstones. It is classified as a typic calcixeroll according to the Soil Taxonomy (Soil Survey Staff 1975). Rock fragments (>1.5 cm) were completely removed from the upper 40 cm of the soil. Then rock fragments of two sizes, classified according to Miller and Guthrie (1984) as coarse gravel with 4.4 cm average diameter (range 1.9–6.7 cm), and cobbles with 14.6 cm average diameter (range 9.4–18.9 cm), were either incorporated into the upper 15 cm of soil or partially embedded in the soil surface. Most plots were kept bare, and in the rest natural vegetation was allowed to grow, dominated by winter annuals and perennials of the following species: Avena fatua, Aegilops ovata, Sinapis arvensis, Echium sp. and Thymus capitatus. Each plot was enclosed by trenches to divert runoff originating upslope. Runoff from the plots was drained into covered metal containers along the lower side of each plot. The containers were cleared of sediment after each runoff event. Additional equipment was installed in seven plots, representing various treatments, to automatically measure and sample runoff with tipping buckets. The data were recorded on a data logger at 5-min intervals. The volumetric soil moisture content was measured at weekly intervals and after every rainfall event at four depths (5–10 cm, 15–20 cm, 20–30 cm, and 30–40 cm) using a neutron and gamma probe. 3.1 Runoff Generation
Soils containing rock fragments exhibit various effects on runoff generation. Generally, large rock fragments (cobbles) cause greater runoff than smaller fragments (coarse gravel). Over five rainfall events with maximum intensity ranging from 21 to 50 mm h−1 , the largest amounts of runoff were generated from bare soils containing abundant large rock fragments, either partially embedded in the surface or incorporated in the upper soil part (Figure 11.5). This is in line with results from laboratory experiments reported by Poesen and Lavee (1991). Soils containing abundant gravel on the surface exhibited a variable effect depending on the rainfall characteristics during individual events: they generated small amounts of runoff and soil loss under rainfalls of high intensity and low duration, but generated more runoff and soil loss under low-intensity rainfalls. Thus, the runoff collected from such soils is even lower than the runoff collected from bare soils in the events with the highest intensities (31 mm h−1 in Figure 11.5). There is a protective effect of coarse gravel at the
136
Mediterranean Desertification 12 surface cobbles
Runoff (mm)
9
incorporated cobbles surface gravel
6
incorporated gravel stone-free
3 vegetation
0
21
31
100 90 Sediment loss (g m−2)
80 70 60 50 40 30 20 10 0
31 21 Rainfall intensity (mm h−1)
Figure 11.5 Runoff and soil loss from erosion plots with various soil surface conditions (after Moustakas et al. 1995)
surface; it probably prevents surface sealing, and this counteracts and eventually exceeds the runoff generated on the impervious rock fragments themselves. As expected, soils protected by a vegetation cover generate less runoff than bare soils. Among the vegetative treatments, the stony soil had the highest biomass production and vegetation cover, and thus generated the least runoff. The higher biomass production in stony soils has been attributed to the generally higher water conservation in such soils (Danalatos et al. 1995). A natural vegetation cover of 48% restricted runoff rates even more; in some cases, the runoff volume was even less than half of the volume measured on a bare soil, depending on the duration and the intensity of the rain. 3.2
Soil Loss
Rock fragments on the soil surface appear to play an important role in the protection against erosion, especially during the heaviest showers. Soil loss is greater from soils containing cobbles than from soils with coarse gravel (Figure 11.5), which is also in line with results reported by Poesen and Lavee (1991). A soil rich in rock fragments with a vegetative cover of 48% reduced the total soil loss by 74% during the rainfall with the highest intensity (Moustakas et al. 1995).
Impact of Rock Fragments on Soil Degradation
137
Coarse gravel on top of the (bare) soil surface has a variable effect on soil loss and erosion. Soil loss is greatly reduced as compared to the stone-free soil during rainfall events with high intensity and short duration, but generates appreciable sediment loss during rainfall with low intensity and long duration. The soil loss varied according to the characteristics of the rainfall. In the stone-free, bare soil it ranged from 2.7 to 38.7 kg m−2 , whereas in the soil with 23% coarse gravel on the surface the soil loss varied from 13.2 to only 19.2 kg m−2 because of the protective effect of the rock fragments against raindrop impact. The ratio of sediment mass to runoff varied from 2.4 to 8.1 kg m−2 mm−1 when averaged for five events, depending on the amount, size and position of rock fragments. In particular, this ratio was 2.4–2.9 for the vegetative plots and 3.0 for the bare soil with coarse gravel on the surface. Among the rest of the bare soils, the ratio was 3.7 for the stone-free bare soil and fluctuated between 4.8 and 8.1 for the other soils containing rock fragments. The ratio in the plots with coarse gravel on the surface was the lowest among the bare soils and slightly higher than in the soils with a vegetation cover.
4 WATER CONSERVATION 4.1 Evaporation Under Laboratory Conditions
Rock fragments influence evaporation from bare soils by changing the soil–atmosphere interface. Numerous efforts have been undertaken to modify the topsoil characteristics (mulching, tillage) in order to create a thin dry topsoil that reduces evaporation. However, little attention has been paid to the role of topsoil stoniness with respect to evaporation. Soil columns with various rock fragment contents simulating a stony plough layer were left to evaporate at a constant evaporative demand. The conditions during the experiments are summarized in Table 11.1. Two evaporation runs were carried out with initial water contents typical for the end of the wet season and the dry season in a Mediterranean environment: • Soils at field capacity: moist silt loam soils (gravimetric moisture content: 0.2 g g−1 ) with different river gravel contents were subjected to 24 mm rain. This condition simulates the beginning of the growing season when excess rainfall has infiltrated to greater depth. • Air-dry soils with different gravel contents received 20 mm of rain. This condition simulates a dry period in which a limited amount of rain falls. For soils at field capacity, the initial total soil ( it ) and fine earth ( ife ) water contents decrease with increasing rock fragment content (Table 11.1). This decrease can be explained by the drainage of excessive moisture and the limited retention of moisture in stony soils (Childs and Flint 1990; Poesen and Lavee 1994; van Wesemael et al. 1995b). In the case of air-dry soils, it decreases slightly with rock fragment content, but ife increases with rock fragment content due to the concentration of an equal amount of rainfall in a smaller volume of fine earth (Table 11.1). After 10 days, clear differences in cumulative evaporation between soil columns with different rock fragment contents could be observed. These differences correspond to the differences in initial fine earth water contents (Figure 11.6 and Table 11.1). For the soils at field capacity, cumulative evaporation decreases with rock fragment content (Figure 11.6(a)), whereas for the air-dry soils cumulative evaporation increases with rock fragment content (Figure 11.6(b)). A rock fragment mulch reduces evaporation compared to a non-stony soil in both experiments (Figure 11.6). These experiments illustrate the ambivalent impact of rock fragments with respect to evaporation rates. During the wet period (winter), when soils are at field capacity, excess precipitation can penetrate below 25 cm (the depth to which evaporation losses are largely restricted; Hanks and Ashcroft 1980). Soils containing rock fragments have a lower fine earth water content in their top layer due to the small water retention capacity of stony soils. Therefore, evaporation rates are smaller in soils containing rock fragments compared to stone-free soils. The high efficiency of a rock fragment mulch under wet conditions in reducing evaporation losses has already been reported by Bond and Willis (1969), Hillel (1980) and Kamar (1994). During dry periods, an equal amount of rain
138
Mediterranean Desertification Table 11.1 Set-up of the three sets of laboratory experiments with mean air temperature, relative humidity and evaporative demand. Figures in parentheses are minimum and maximum values
P = 24 mm, soil at field capacity Temperature: 20.1 ◦ C (16–28 ◦ C) Relative humidity: 77.8% (58–95%) 7.71 mm day−1 (5.6–10.1 mm day−1 ) Evaporative demand (Eo ): Rv (m3 m−3 )
it (m3 m−3 )
ife (m3 m−3 )
0 0.19 0.35 0.53 mulch
0.38 0.30 0.23 0.10 0.37
0.38 0.37 0.35 0.21 0.37
P = 20 mm, air-dry soil Temperature: Relative humidity: Evaporative demand (Eo ):
18.5 ◦ C (15–23 ◦ C) 77.6% (57–94%) 9.24 mm day−1 (6.3–11.7 mm day−1 )
Rv (m3 m−3 )
it (m3 m−3 )
ife (m3 m−3 )
0 0.16 0.30 0.46 mulch
0.14 0.12 0.12 0.11 0.16
0.14 0.14 0.17 0.20 0.16
P , total rainfall amount; Rv , rock fragment content by volume; it , total soil water content at the start of evaporation; ife , fine earth water content of the fine earth at the start of evaporation. Mulch is a 5-cm-thick continuous gravel layer. All figures represent mean of duplicates. that falls on a dry soil is concentrated in a smaller volume of fine earth with increasing rock fragment content, thus leading to a higher fine earth water content. Hence, evaporation rates increase with fine earth water content and rock fragment content. It should be kept in mind that evaporation during the laboratory experiments was caused by convection rather than by a combination of radiation and convection, and the soil columns were rather short. The following section will discuss the pertinence of these laboratory experiments to water conservation under field conditions. 4.2
Evaporation Under Field Conditions
Soil moisture storage was measured in the upper part (5–15 cm) of a stone-free soil and the same soil covered by coarse gravel (28%) and by cobbles (18%) under field conditions for one year (Figure 11.7). It can be seen that soil water content was generally higher in the cobbly soil, pointing to greater water conservation by the cobbles for most of the study period. Only after the end of July did the water storage in the cobbly soil decrease sharply and show values lower than both the control
139
Impact of Rock Fragments on Soil Degradation
Cumulative evaporation (mm)
(a) 60 field capacity; P = 24 mm
50 40 30 20 10 0 0
Cumulative evaporation (mm)
(b) 24 22 20 18 16 14
5
10
15 Days
20
25
30
air-dry soil; P = 20 mm
12 10 8 6 4 2 0
Rv = 0 0
5
Rv = 0.16 10
Rv = 0.30 15
Rv = 0.46 20
25
mulch 30
Days
Figure 11.6 Cumulative evaporation depth from soil columns containing different contents of rock fragments. Rv is rock fragment content by volume. Experiments were carried out in the laboratory with (a) soils at field capacity subjected to 24 mm rainfall and (b) air-dry soils subjected to 20 mm rainfall (after van Wesemael et al. 1996). Reprinted from Journal of Hydrology 182, B. van Wesemael, J. Poesen, C.S. Kosmas, N.G. Danalatos and J. Nachtergaele, Evaporation from cultivated soils containing rock fragments, 65–82. Copyright 1996, with permission from Elsevier Science
and the gravelly soils for the rest of the dry and hot period (Figure 11.7). This is apparently due to a much greater heating of the rock fragments at that period (Danalatos et al. 1995). Conversely, the soil containing coarse gravel had the lowest water storage and therefore the highest evaporation losses throughout the wet period and the period of moderate drought. Only during the dry and hot summer were values higher than those of the soil with cobbles embedded in the surface (Figure 11.7). Data on soil moisture loss obtained from the weighing lysimeters demonstrated that the presence of cobbles on the soil surface is extremely important, especially the first day after a rainfall or irrigation event. As Figure 11.8(a) illustrates, heating of the cobbly soil during daytime in summer resulted in a great loss of water as compared to the loss from the stone-free soil. In the following days, the rate of water loss remained almost the same in both lysimeters due to the formation of a desiccated layer, drastically reducing the evaporation loss. Conversely, the presence of cobbles
140
Mediterranean Desertification cobbles
25
free of RF Soil water (mm)
20
gravel
15 10 heavy rainfall
5 0 120
M
J
J
A 220
S
O
N
D
J
320 Time (days)
F
M
420
A
Month 520
Figure 11.7 Evolution of the soil moisture stored in the 5–15 cm soil layer for soils with rock fragments of different sizes on the soil surface (after Kosmas et al. 1995) (a) Dry period irrigation 21 mm
Soil weight (kg)
243 241 239 237 235 233 231
232
233
234
235
236
249
250
(b) Wet period
Soil weight (kg)
242 irrigation 21 mm
240 238 236 234 245
246 cobbles 28%
247 248 Time (days) stone-free
Figure 11.8 Changes in soil weight with time measured in lysimeters with a stone-free soil and a soil with rock fragments, during a period of (a) high and (b) moderate evaporative demands (after Kosmas et al. 1995)
141
Impact of Rock Fragments on Soil Degradation
reduced the evaporation during the wet period the first day after irrigation as compared to the stonefree soil (Figure 11.8(b)). These field experiments are in agreement with the ambivalent role of rock fragments illustrated in the laboratory. Increasing rock fragment cover is associated with decreasing evaporative water loss during periods of no to moderate drought, such as from late fall to early summer, but with an increased evaporation during the dry and hot months. Stony soils are generally warmer during daytime and cooler at night than soils free of rock fragments. In the warmest month (July) the diurnal amplitude reached 14.3 ◦ C in the stone-free soil and 24.1 ◦ C in the stony soils under climatic conditions prevailing in the region of Attica. Considering that maturation of rainfed crops occurs in late spring, rock fragments appear to be very important in conserving appreciable amounts of soil moisture for growing plants in late spring, when rain can be scarce in the Mediterranean region. Water conservation in stony soils supports considerable biomass production, and protects extensive hilly lands from desertification. 4.3 Water Vapour Adsorption
Rock fragments present on the soil surface restrict evaporation as well as water vapour adsorption by the soil by reducing the soil–atmosphere interface. Daily fluctuations in soil moisture tension measured in the upper 3–5 cm soil layer and in patches that were free of rock fragments (a transect between two rock fragments), were greater than those measured under the cobbles of the same soil (Figure 11.9). In soil patches under rock fragments, the maximum and minimum values of soil moisture tension occurred one or two hours later than in soil patches free of rock fragments. During the day, fluctuations in soil moisture tension in stone-free soils were about twice those of stony soils, while during the night soil moisture tensions reached almost the same value in both cases (Figure 11.9). This points to the importance of rock fragments in conserving soil moisture from evaporation under Mediterranean conditions. Rock fragments restrict evaporation losses during the day, whereas during the night, water adsorbed as water vapour by the stone-free soil–air interface is transmitted and protected from evaporation under the rock fragments. This water storage can be of great importance for rainfed crops throughout the Mediterranean. 4.4 Biomass Production
The presence of rock fragments on the soil surface (i.e. stone mulches) is extremely important in dry years in order to conserve appreciable amounts of soil water and prevent large areas from
Soil moisture tension (kPa)
30 28 26 24 22 20 351
353
357
355
359
361
Time (days) Under RF
Free of RF
Figure 11.9 Changes in soil moisture tension with time measured in soil patches free of rock fragments (solid line) and patches under rock fragments (dashed line) in the same soil (after Kosmas et al. 1995)
142
Mediterranean Desertification
desertification. Despite their normally low productivity, stony soils formed on conglomerates and shale–sandstones may supply appreciable amounts of previously stored water to the stressed plants and, therefore, secure a substantial biomass production even during extremely dry years (Kosmas et al. 1993). Soils formed on parent materials free of rock fragments such as marl are, despite their considerable depth and high productivity during normal and wet years, very susceptible to desertification. Such soils are unable to support any vegetation during particularly dry years due to adverse soil physical properties and the absence of gravel and stone mulching. Investigations into the relationship between biomass production and evapotranspiration rate, taking into account the rock fragment cover, were conducted along catenas of central and northern Greece. Rock fragments were removed from a number of plots in distinct landscape positions after the sowing of wheat, and the total aboveground biomass production of wheat was measured at the end of the growing period of cereals and compared with that where cobbles remained on the soil. The presence of cobbles on the soil surface conserves appreciable amounts of soil water by surface mulching, which results in increased biomass production, particularly in dry years, by preventing desiccation of the soil. After removing all rock fragments from the surface of 32 plots, biomass production of rainfed cereals decreased by 2–30%. In an attempt to relate measured biomass production to the actual evapotranspiration (ETa ) and to the percentage of rock fragment cover (Rc), the ETa was calculated from its maximum value (ETm ) and the momentary soil moisture content in the root zone using a simple water balance model based on Doorenbos and Pruitt’s (1977) methodology. The maximum crop evapotranspiration rate (ETm ) was calculated from the potential evapotranspiration rate (ETp ) and the crop (leaf area) coefficient of wheat according to Doorenbos and Pruitt (1977). Finally, the potential evapotranspiration rate was calculated from daily values of maximum and minimum air temperature, sunshine duration, air humidity and wind speed, according to Penman (1948; modified by Frere 1979). It was found that the relative biomass production (RBP) of rainfed wheat could be related to the relative evapotranspiration rate (ETa /ETm ) and the percentage of rock fragment cover, according to the following empirical relation (Kosmas et al. 1995): RBP = 0.97 + 0.54∗ ln(ETa /ETm ) + 0.035∗ ln(Rc)
R = 0.90
n = 52
(1)
The relative biomass was determined from the measured biomass production divided by the maximum value estimated for each landscape position and parent material under conditions of no water deficit (Kosmas et al. 1993). The equation above is valid only when a soil water deficit occurs (ETa /ETm < 1) during the growing period, which is normally the case under Mediterranean conditions. In the case that there is no water deficit, rock fragments negatively affect biomass production due to the combined effect of a reduction in effective rooting depth and a decreased soil volume available for adsorption of nutrients.
5
A PRACTICAL EXAMPLE OF GRAVEL MULCHING
Few studies have investigated the effect and behaviour of a gravel mulch in the field. Gale et al. (1993) describe the application of gravel mulches in the loess belt of north-west China, and Caldas and Salguero (1988) report mulching with lapilli on the Canary Islands. These studies remain descriptive and lack experimental data. Fieldwork was carried out in the vineyards of an alluvial fan in the upper Rhˆone valley in Switzerland (Nachtergaele et al. 1988). An artificial gravel mulch of 20 cm was applied to most vineyards totalling approximately 10 km2 . Although mean annual precipitation in the Upper Rhˆone Valley is amongst the lowest in Switzerland (597 mm year−1 ), irrigation water is readily available from mountain streams. An inquiry amongst the wine-growers revealed that the thermal properties are higher ranked than the hydraulic characteristics of the mulch. Since the role of rock fragments on water conservation and soil erosion has already been discussed in previous sections, we will concentrate here on the thermal properties of a rock fragment mulch. Soil temperature at 3 cm below the soil surface is constantly higher for the topsoil with a mulch compared to the topsoil without a mulch (mean difference: 0.7 ◦ C; Figure 11.10). The difference in
143
Impact of Rock Fragments on Soil Degradation 27
25
26 20
24 15
23 22
10
21 20
Rainfall depth (mm)
Temperature (°C)
25
5
19 18
0 Time (Julian days)
ˆ Figure 11.10 Soil temperature measured in July 1994 at 3 cm depth in vineyards of the Rhone Valley (Switzerland). The solid line represents values under a gravel mulch and the dashed line represents the control situation without a gravel mulch. Vertical bars indicate rainfall depth (after Nachtergaele et al. 1998). Reprinted from Soil and Tillage Research 46, J. Nachtergaele, J. Poesen and B. van Wesemael, Application and efficiency of gravel mulches in southern Switzerland, 51–59. Copyright 1998, with permission from Elsevier Science
soil temperature between the treatments is affected by soil moisture status. During and immediately after a rainy period, the temperature difference is much less than in drier conditions (Figure 11.10). These results are in agreement with the findings of Childs and Flint (1990) and Gras (1994), indicating that the presence of non-porous rock fragments in a dry soil profile increases thermal diffusivity (i.e. the ratio of thermal conductivity to heat capacity). Measurements of the soil surface temperature also indicate that there is a systematic temperature increase due to the mulch cover (mean increase 2.2 ◦ C; Nachtergaele et al. 1998). The implications of the higher soil and soil surface temperatures for the mulched vineyards are: (i) plant roots are protected from low temperatures at night or during spring; (ii) plants and fruits receive an extra amount of radiation.
6 CONCLUSIONS Rock fragments appear to have a profound impact on physical soil degradation, soil erosion, soil moisture conservation, plant growth and thermal properties under Mediterranean conditions. Rock fragments at the soil surface preserve the surface roughness of freshly tilled soils during rainfall even at low rock fragment contents. At high rock fragment contents a skeleton structure will develop in the topsoil which prevents soil compaction. Maximum runoff and erosion are expected in soils containing large amounts of cobbles partially embedded or incorporated in the soil. However, soils containing abundant gravel on the surface may show various effects. They generate small amounts of runoff and cause little soil loss under rainfalls of high intensity and short duration, but greater runoff and soil loss under low-intensity rainfalls. Cobbles have a beneficial effect on soil moisture conservation under conditions of moderate water stress such as those prevailing in spring and early summer, which is the most crucial period for winter crops. Later in the summer, their effect becomes negative because they cause greater heating of the soils. This is not harmful, however, since in that period (late summer to early autumn) only irrigated crops may survive or give reliable yields. The presence of rock fragments can be very valuable, particularly in dry years, by conserving appreciable amounts of water stored in previous rainy periods or adsorbed at night, thus protecting large areas from degradation and eventual desertification. It
144
Mediterranean Desertification
is believed that both coarse gravel and cobbles in soils are important for their combined positive effects on erosion rates (coarse gravel) and soil moisture conservation (cobbles).
REFERENCES Abrahim YB and Rickson RJ (1989) The effectiveness of stubble mulching in soil erosion control. Soil Technology 1, 115–126. Bertuzzi P, Rauws G and Courault D (1990) Testing roughness indices to estimate soil surface roughness changes due to simulated rainfall. Soil and Tillage Research 17, 87–99. Bond JJ and Willis WO (1969) Soil water evaporation: surface residue rate and placement effects. Soil Science Society of America Proceedings 33, 445–448. Caldas FE and Salguero TMK (1988) Mulch farming in the Canary Islands. In N Wichiro (ed) Agro-geology in Africa. Commonwealth Science Council Technical Publication 226, pp. 242–256. Childs SW and Flint AL (1990) Physical properties of forest soils containing rock fragments. In SP Gessel, DS Lacate, GF Weetman and RF Powers (eds) Sustained Productivity of Forest Soils. Proceedings of the 7th North American Forest Soils Conference, University of British Columbia, Faculty of Forestry Publication, Vancouver, pp. 95–121. Danalatos NG, Kosmas C, Moustakas N and Yassoglou N (1995) Rock fragments: II. Their effect on soil properties and biomass production. Soil Use and Management 11, 121–126. De Ploey J and Poesen JW (1985) Aggregate stability, runoff generation and interrill erosion. In KS Richards, RR Arnett and S Ellis (eds) Geomorphology and Soils. George Allen and Unwin, London, pp. 99–120. Doorenbos J and Pruitt WO (1977) Guidelines for predicting crop water requirements. Irrigation and Drainage Paper 24, FAO, Rome. Frere M (1979) A method for the practical application of the Penman formula for the estimation of potential evapotranspiration and evaporation from a free water surface. FAO, AGP: Ecol./1979/1. FAO, Rome. Gale WJ, McColl RW and Xie Fang (1993) Sandy fields traditional farming for water conservation in China. Journal of Soil Water Conservation 48, 474–477. Gilley JE, Finker SC and Varvel GE (1986) Runoff and erosion as affected by sorghum and soybean residue. Transactions of the American Society of Agricultural Engineers 29, 1605–1610. Gras R (1994) Sols Caillouteux et Production V´eg´etale. Institut National de la Recherche Agronomique, Paris. Hanks RJ and Ashcroft GL (1980) Applied Soil Physics. Springer-Verlag, New York. Hanson CT and Blevins RL (1979) Soil water in coarse fragments. Soil Science Society of America Journal 43, 819–820. Hillel D (1980) Introduction to Soil Physics. Academic Press, London. Jennings GD and Jarrett AR (1985) Laboratory evaluation of mulches in reducing erosion. Transactions of the American Society of Agricultural Engineers 28, 1466–1470. Kamar MJ (1994) Natural use of stone and organic mulches for water conservation and enhancement of crop yield in semi-arid areas. Advances in GeoEcology 27, 163–179. Kemper WD, Nick AD and Corey AT (1994) Accumulation of water in soils under gravel and sand mulches. Soil Science Society of America Journal 58, 56–63. Kosmas C, Danalatos NG, Moustakas N, Tsatiris B, Kallianou Ch and Yassoglou N (1993) The impacts of parent material and landscape position on drought and biomass production of wheat under semi-arid conditions. Soil Technology 6, 337–349. Kosmas C, Yasoglou N, Moustakas N and Danalatos N (1995) Field site: Spata. In Mediterranean Desertification and Land Use, Basic Field Programme, Phase 2 . Final report of MEDALUS II-Project 1, contract EV5VCT92-0128, MEDALUS Office, Thatcham, UK. Le Bissonnais Y (1996) Aggregate stability and assessment of soil crustability and erodibility: 1. Theory and methodology. European Journal of Soil Science 47, 425–438. Lopez PR, Cogo NP and Levien R (1987) Erosion reduction effectiveness of types and amounts of surfaceapplied crop residues. Revista Brasileira de Ciencia do Solo 11, 71–75. Magier J and Ravina I (1984) Rock fragments and soil depth as factors in land evolution. In JD Nichols, PL Brown and WJ Grant (eds) Erosion and Productivity of Soils Containing Rock Fragments. Special Publication no. 13, Soil Science Society of America, Madison, Wisconsin, pp. 13–30. Meyer LD, Johnson CB and Foster GR (1972) Stone and woodchip mulches for erosion control on construction sites. Journal of Soil Water Conservation 27, 264–269. Miller FT and Guthrie RL (1984) Classification and distribution of soils containing rock fragments in the United States. In JD Nichols, PL Brown and WJ Grant (eds) Erosion and Productivity of Soils Containing Rock Fragments. Soil Science Society of America, Madison, Wisconsin, pp. 1–6.
Impact of Rock Fragments on Soil Degradation
145
Moustakas N, Kosmas C, Danalatos NG and Yassoglou N (1995) Rock fragments. I: Their effect on runoff, erosion and soil properties under field conditions. Soil Use and Management 11, 115–120. Nachtergaele J, Poesen J and van Wesemael B (1998) Application and efficiency of gravel mulches in southern Switzerland. Soil and Tillage Research 46, 51–59. Onstad CA, Wolf ML, Larson CL and Slack DC (1984) Tilled soil subsidence during repeated wetting. Transactions of the American Society of Agricultural Engineers 27, 733–736. Parry WL and Carter TR (1991) Climatic changes and future land use potential in Europe. In R Fantechi, G Maracchi and ME Almeida-Teixeira (eds) Climatic Change and Impacts: A General Introduction. Commission of the European Communities, Directorate-General Science, Research and Development, Report EUR 11 943 EN. Penman HL (1948) Natural evaporation from open water, bare soils and grass. Proceedings of the Royal Society 193, 120–145. Peters DB (1960) Relative magnitude of evaporation and transpiration. Agronomy 52, 536–538. Poesen J (1990) Conditions for the evacuation of rock fragments from cultivated upland areas during rainstorms. In DE Walling, A Yair and S Berkowicz (eds) Proceedings of the Jerusalem Workshop on Erosion, Transport and Deposition Processes. IAHS Publication 189, IAHS Press, Wallingford, pp. 145–160. Poesen J and Ingelmo-Sanchez F (1992) Runoff and sediment yield from topsoils with different porosity as affected by rock fragment cover and position. Catena 19, 451–474. Poesen J and Lavee H (1991) Effects of size and incorporation of synthetic mulch on runoff and sediment yield from interrills in a laboratory study with simulated rainfall. Soil and Tillage Research 21, 209–233. Poesen J and Lavee H (1994) Rock fragments in topsoils: significance and processes. Catena 23, 1–28. Poesen J and van Wesemael B (1995) Effects of rock fragments on the structural collapse of tilled topsoils during rain. In E Derbyshire, T Dijkstra and IJ Smalley (eds) Genesis and Properties of Collapsible Soils. NATO Advanced Science Institute Series Vol. 468, Kluwer Academic, Dordrecht, pp. 333–343. Poesen JW, Torri D and Bunte K (1994) Effects of rock fragments on soil erosion by water at different spatial scales: a review. Catena 23, 141–166. Ravina I and Magier J (1984) Hydraulic conductivity and water retention of clay soils containing rock fragments. Soil Science Society of America Journal 48, 736–740. R¨omkens MJ, Wang JY and Darden RW (1988) A laser microreliefmeter. Transactions of the American Society of Agricultural Engineers 31, 408–413. Soil Survey Staff (1975) Soil Taxonomy: A Basic System of Soil Classification for Making and Interpreting Soil Surveys. USDA-SCS Agric Handbook 436. US Government Print Office, Washington, DC. UNESCO-FAO (1963) Bioclimatic map of the Mediterranean zone. Explanatory notes. Arid Zone Research XXI. FAO, Rome. Van Wesemael B, Poesen J, de Figueiredo T and Govers G (1995a) Effects of rock fragments on soil surface roughness evolution during rainfall. Earth Surface Processes and Landforms 21, 399–441. Van Wesemael B, Poesen J and de Figuiredo T (1995b) Effects of rock fragments on physical degradation of cultivated soils by rainfall. Soil and Tillage Research 33; 229–250. Van Wesemael B, Poesen J, Kosmas CS, Danalatos NG and Nachtergaele J (1996) Evaporation from cultivated soils containing rock fragments. Journal of Hydrology 182, 65–82.
12
Aridification in a Region Neighbouring the Mediterranean
´ ´ AM ´ ´ ´ HUSZAR, ´ ´ ´ ´ ´ AD KERTESZ, TAMAS DENES LOCZY, BELA MARKUS, JANOS MIKA, ´ ´ ´ ´ ´ ´ ´ TOZSA KATALIN MOLNAR, SANDOR PAPP, ANTAL SANTHA, LASZLO SZALAI, ISTVAN AND GERGELY JAKAB
Department of Physical Geography, Geographical Research Institute, Hungarian Academy of Sciences, Budapest, Hungary
1 INTRODUCTION AND OBJECTIVES Interpretation of temperature, precipitation and potential evaporation anomaly patterns, and the scenarios of regional climate change based on the General Circulation Model, generally suggest that climate modification may be predicted for the northern part of the Mediterranean region as well as the south. Hungary, lying in the heart of the Carpathian Basin, among the countries of central and eastern Europe, is a flat, low-lying country, and faces some severe problems. The effects of three main climatic influences felt in Hungary (Mediterranean, continental and Atlantic) may become modified and result in a changed, more difficult, climate. The main desertification problems in Hungary have always been connected with drought. Dry periods, ranging from several years up to 25 years, have, in the past, led to serious water deficits and water imbalances affecting natural systems and land resource production systems. The term aridification was introduced to characterize the increasing dryness (aridity) of the climate as a result of global climate change and its environmental consequences. To consider these possible consequences of global climate change, an aridification research programme was launched within the framework of the MEDALUS II project. Objectives included climatological investigations to explore the impact of global climate change on the climate of Hungary. Some test areas were studied in detail, and the physical processes of aridification were examined and tested. It has yet to be shown whether medium- and short-term oscillations do indicate a tendency towards a warmer and dryer climate. Changes in soil properties, water reserves and vegetation were studied in areas considered most environmentally sensitive. Special attention was paid to water budget parameters. Recent groundwater level changes have been monitored, and future trends predicted. Soil moisture dynamics in soil profiles, and the impacts of groundwater level changes on soil processes were studied. The species composition of the natural vegetation of the central Great Hungarian Plain was evaluated to provide further climatic change parameters. Remote sensing and GIS techniques were used to map land-use changes between 1975 and 1991, and the trends that emerged were considered from environmental, agricultural and economic viewpoints.
2 CLIMATE CHANGE IN HUNGARY As in other parts of Europe, the meteorological record for recent years in Hungary (Bussay et al. 1995) shows major deviations from long-term mean values, from data available since 1881. For Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
148 Temperature (°C)
Mediterranean Desertification
12 10.2 9 800
Precipitation (mm)
700 600 500 400 300 200 100 0 1881
1891
1901
1911
1921
1931 1941 Year
1951
1961
1971
1981
1991
Figure 12.1 Trends of annual mean temperature and annual precipitation for Budapest, 1881–1991 (Matyasovszky 1995)
example, the precipitation deficits observed in the summer of 1994 (when rainfall was only 43% of the long-term average) were followed in the next year by the almost unprecedented low October precipitation of 3 mm and then record rainfall amounts in December 1995. Figure 12.1 shows the long-term trends of annual mean temperature and annual total precipitation for Budapest. From this, and with data from other meteorological stations, changes in other climatic variables can be deduced, and scenarios of climate change can be suggested. To test a hypothesis of increasing aridification since 1900, time series of monthly mean temperatures and precipitation between 1900 and 1990 were analysed using records for 16 stations across Hungary (17 for precipitation). The locations of the stations are shown in Figure 12.2. Although they are not geographically evenly spaced, the stations do seem to represent the full range of climate experienced over Hungary (Matyasovszky 1995; Moln´ar and Mika 1997). The inhomogeneity of temperature data series can lead to some uncertainty when establishing trends. In the present survey, the data series was homogenized applying Szentimrey’s (1994) data correction procedure, which relies on an undistorted reference (the temperature series for Kremsm¨unster, Austria). The inhomogeneities are due to the changes of the measurement frequency of meteorological data in Hungary in the middle of the 1960s. The station of Kremsm¨unster, with an undisturbed data series, is located near to the Hungarian border. The time series for 1900–1990 was divided into two intervals: 1900 to 1949, and 1950 to 1989, and these intervals were also analysed for trends of climatic change. 2.1
Temperature Trends
There are definite suggestions of climatic warming over Hungary since 1900. In areas of colder climate, annual mean temperatures were 0.2–0.3 ◦ C higher than the long-term average during the interval 1950–1989. For stations with the warmest climate the increase was even more remarkable (+0.3–0.5 ◦ C). Monthly mean temperatures also showed an increasing trend. The change in January (coldest month) temperatures can be illustrated with the examples of Ny´ıregyh´aza where there was a
149
Aridification in a Region Neighbouring the Mediterranean 16
16
18 1
2 18
16 3
4 14
7
18 5
16
6 8 9
10
12
11 18
N
14
18 13
20
W
E S
18
16
17
15 20
Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
0
40
80 km
Location Mosonmagyaróvár Sopron Eger Nyíregyháza Szombathely Pápa Budapest Debrecen Zalaegerszeg Keszthely Kecskemét Túrkeve Kalocsa Szarvas Pécs Baja Szeged Kiskunfélegyháza groundwater well
Figure 12.2 Locations of meteorological stations (and a groundwater well) across Hungary. Precipitation data are available for all stations since 1881, and temperature data for all stations except Eger. July isotherms (◦ C) are also shown
rise from −2.7 to −2.4 ◦ C, and of Debrecen where there was a rise from −2.2 to −1.8 ◦ C. Warmer areas experiencing warming in January included P´ecs (from −0.4 to −0.1 ◦ C) and Kecskem´et (from −0.5 to −0.1 ◦ C). July (warmest month) mean temperatures do support the warming tendency, but not so clearly. At the station with the coldest climate (Zalaegerszeg) July mean temperature was 20.0 ◦ C in the study period up to 1949, but 19.7 ◦ C after 1949. Regarding the two warmest areas of Hungary, similar trends were observed. The mean July temperature in the centre of the Great Hungarian Plain (Kecskem´et) between 1900 and 1949 was 23.1 ◦ C and between 1950 and 1989 it
150
Mediterranean Desertification ◦
was 22.9 C. From studying the means of other months, it became clear that most of the warming has been associated with rising temperatures in the winter half of the year. Annual temperature range has decreased at every meteorological station, although not at the same rate. The biggest changes were observed along the Danube (Kalocsa: −1.0 ◦ C), and at Zalaegerszeg (−0.9 ◦ C). The nationwide mean change in annual temperature range was found to be −0.56 ◦ C. This may indicate reduced continentality of the Hungarian climate. In conclusion, we suggest that the 1950–1989 temperatures are several tenths of degrees higher than those measured between 1900 and 1949. The annual mean temperatures of the 16 stations increased, on average, by 0.3 ◦ C during the period 1950–1989. Milder winters and the reduced annual range experienced in this region may be associated with global warming. 2.2
Precipitation Trends
At stations with the lowest annual precipitation, there has been a decrease in precipitation since 1949. The degree of change, however, varies greatly, e.g. for Szarvas (from 585 mm to 495 mm), and T´urkeve (from 544 mm to 534 m). Both these stations are on the Great Hungarian Plain. Data from the stations with the highest annual precipitation (western Hungary) also show a clear reduction in rainfall of 40–65 mm, e.g. Zalaegerszeg (from 740 mm to 701 mm) and Sopron (from 715 mm to 657 mm). Mean monthly precipitation figures confirm the tendency towards decreasing precipitation since 1949. Figure 12.3 shows that the pattern of rainfall distribution throughout the year has changed considerably. In March there has been a decrease, except for two stations (Szombathely and T´urkeve), where no change was detected. There was a significant drop in precipitation for all the 17 stations studied in April (mean change of −7 mm) and also May in most regions (about −4 mm). June was the only month when an increase in precipitation was found for all 17 stations. The average for the first half of the century was 68.8 mm in this month, becoming 78.1 mm in recent decades. In July and August no change was observed. In September all stations recorded a mean monthly decrease 90.00
Monthly average precipitation (mm)
80.00 70.00 60.00 50.00 40.00 30.00 20.00
1900−1949 1950−1989
10.00 0.00 Jan
Feb
Mar
Apr
May
Jun Jul Month
Aug
Sep
Oct
Nov
Dec
Figure 12.3 A comparison of monthly precipitation between the intervals 1900–1949 and 1950–1989, average values for 17 stations
151
Aridification in a Region Neighbouring the Mediterranean 80 1900−1949 1950−1989 70
Precipitation (mm)
60 50 40 30 20 10 0 1
2
3
4
5
6
7
8 9 10 11 Station number
12
13
14
15
16
17
Figure 12.4 Differences in October mean precipitation between the intervals 1900–1949 and 1950–1989; averages for 17 stations across Hungary
in precipitation. The mean fall was from 53.4 mm to 42.8 mm. The decrease was even greater in October, from 55.3 mm to 38.1 mm (Figure 12.4). In September the effect was mainly uniform across Hungary but in October, three particular areas with a climate of Mediterranean character have shown a reduction in autumn rainfall. The distribution of annual and monthly precipitation has more anomalies than the corresponding distributions of temperature. It is common to find any month without rain, or alternatively, there may be high rainfall in any month of the summer (200–300 mm) associated with intense storms. 2.3 Statistical Significance of Precipitation and Temperature Trends A statistical T-test was applied to the results of linear regression analysis, to establish which stations have shown statistically significant trends for warming and precipitation distribution changes, at the 95% level. Table 12.1 lists the temperature changes for 16 stations using the basic data. Table 12.2 is for the same stations but the data have been adjusted and corrected to conform with the Kremsm¨unster series. In Table 12.1, if the trends are averaged, an increase of only 0.003 ◦ C year−1 is found. However, where the corrected data are shown, in Table 12.2, all 16 stations showed a moderate warming tendency of at least 0.010 ◦ C year−1 . For each station studied, a trend of decreasing annual precipitation was found – trends that were statistically significant for 12 stations out of the total 17 at the 95% significance level. An average precipitation decrease of −0.917 mm year−1 (from Table 12.3) is similar to the findings of Koflanovits-Adamy and Szentimrey (1986). Ambr´ozy et al. (1990) studied change over a 84-year period (1901–1984) and claim that over the Great Hungarian Plain the first decades of the 20th century were characterized by increased humidity, followed by a long period of little change, and then evidence of drought. The range amounts to almost 10% of annual precipitation, meaning that the oscillations of the mean annual precipitation in a dry period can be 10% of the long-term average. The main concern is that reduced annual precipitation and rising temperatures in Hungary are leading to increased aridity.
152
Mediterranean Desertification
Table 12.1 Temperature trends for 16 meteorological stations in Hungary (1900–1990) (non-corrected database)
Number
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Location of meteorological station in Hungary
Temperature trend (◦ C year−1 )
Statistical T value
Statistically significant at 95% level
Mosonmagyar´ov´ar Sopron Ny´ıregyh´aza Szombathely P´apa Budapest Debrecen Zalaegerszeg Keszthely Kecskem´et T´urkeve Kalocsa Szarvas P´ecs Baja Szeged Mean for 16 stations
+0.004 +0.003 +0.003 +0.003 −0.003 +0.010 +0.002 −0.003 +0.002 −0.003 +0.002 −0.000 −0.001 −0.006 +0.003 −0.010 +0.003
1.740 1.724 1.301 1.436 1.448 5.599 0.836 1.337 0.836 1.291 0.992 0.212 0.622 2.632 1.335 4.446
no no no no no yes no no no no no no no yes no yes
Table 12.2 Temperature trends 1900–1990 (corrected database)
Number
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Location of meteorological station in Hungary
Temperature trend (◦ C year−1 )
Statistical T value
Statistically significant at 95% level
Mosonmagyar´ov´ar Sopron Ny´ıregyh´aza Szombathely P´apa Budapest Debrecen Zalaegerszeg Keszthely Kecskem´et T´urkeve Kalocsa Szarvas P´ecs Baja Szeged Mean for 16 stations
0.010 0.010 0.011 0.010 0.011 0.011 0.011 0.011 0.010 0.011 0.010 0.011 0.010 0.011 0.011 0.010 0.0105
5.307 5.174 5.532 5.372 5.330 5.806 5.707 5.307 5.490 5.332 4.795 5.727 4.635 5.545 5.718 4.519
yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes
153
Aridification in a Region Neighbouring the Mediterranean
Table 12.3 Precipitation trends for 17 meteorological stations across Hungary (1900–1990)
Number
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Location of meteorological station in Hungary Mosonmagyar´ov´ar Sopron Eger Ny´ıregyh´aza Szombathely P´apa Budapest Debrecen Zalaegerszeg Keszthely Kecskem´et T´urkeve Kalocsa Szarvas P´ecs Baja Szeged Average
Precipitation trend (mm year−1 )
Statistical T value
Statistically significant at 95% level
−0.61 −1.22 −0.47 −1.02 −1.41 −0.27 −1.09 −0.47 −0.56 −0.39 −0.86 −0.98 −0.43 −0.79 −2.30 −0.71 −0.91 −0.918
1.953 3.518 1.618 3.086 4.432 0.745 3.309 1.358 1.526 0.994 2.892 2.980 1.436 2.754 5.765 2.070 3.002
no yes no yes yes no yes no no no yes yes no yes yes yes yes
For calculating the frequency of warm and dry years (using the method of Tar 1992), a sample meteorological station was selected and the time series was divided into 10-year intervals. The database included corrected annual and monthly mean temperatures (Szentimrey 1994) and annual and monthly precipitation sums for a total of 110 years (1881–1989). Each year with a mean annual temperature higher than or equal to the 110-year average was defined as a warm year, and each year lower than average, defined as a cold year. The same classification was made for precipitation (as wet and dry years). Categories of humid and dry years were defined.
3 ARIDIFICATION PROCESSES 3.1 Groundwater Level Changes
A major impact of changes toward a drier climate is the depletion of groundwater reserves. This has been studied in one of the most severely affected regions of Hungary, on the Danube–Tisza interfluve (Husz´ar et al. 1996). The database analysed derives from the observation well network operated by the Research Centre for Water Resources Development (VITUKI). A most serious aspect of the aridification trend here is extremely reduced infiltration into the soils and reduced recharge of groundwater. In the 1980s, significantly diminished autumn and winter precipitation only allowed infiltration (of insignificant amount) on two occasions. According to hydrologists (P´alfai 1991), a combined effect of several factors is responsible for falling groundwater levels: lower precipitation and increased evaporation explain about 50% of the drop, but the extraction of confined groundwater for drinking water supply (25%), afforestation and other land-use changes (10%), drainage regulation (7%), direct extraction of free groundwater as well as reduced recharge from the neighbouring hills and from the Danube (6%) are also significant factors. In the 1980s and early 1990s the deficit in autumn and winter precipitation and diminished infiltration led to a reduction of peaks on the annual groundwater graph (P´alfai 1995). The changes in
154
Mediterranean Desertification 102.00
101.00
m a.s.l.
100.00
99.00
98.00
97.00
96.00 1950 1954 1958 1962 1966 1970 1974 1978 1982 1986 1990 1952 1958 1980 1984 1968 1972 1978 1980 1984 1988 1992 Year
Figure 12.5 Fall in groundwater levels on the Danube–Tisza interfluve illustrated by the ´ ´ long-term (1950–1993) curve of monthly average groundwater levels in the Kiskunfelegyh aza observation well (no. 883)
the well shown in Figure 12.5 clearly show both the gradual fall of the annual average groundwater table level and the reduced seasonal range which is a consequence of inadequate winter rain infiltration. Three representative SSE–NNE-aligned cross-sections of the Danube–Tisza interfluve (perpendicular to the strike of the interfluve ridge) were analysed. Each included 12 to 15 observation wells, at around 5 to 7 km intervals along the cross-section. Groundwater levels in the early 1990s were compared to the average of the 1960s, when infiltration was above the long-term average level, and therefore this decade most certainly preceded the beginning of aridification. Through the geographical interpolation of groundwater well observation records, a map of changes in the annual mean groundwater level was constructed (Figure 12.6). It shows that in some of the most susceptible, rapidly drained areas (loess-mantled as well as sand regions), 2–4 m falls in the water table are common. With falling groundwater levels, soil moisture contents also reduced considerably during the 1990s. For instance, in spring 1990 in some sections of the Danube–Tisza interfluve the uppermost 1 m of soil had only 60–70% soil-moisture reserves, as opposed to the long-term average of 100% field capacity. In 1992, in the same area, the 0–0.5 m topsoil contained less than 15% moisture, which is below the wilting-point of most agricultural crops (P´alfai 1996). Before the wet winter of 1995, the winter precipitation deficit had maintained a decreasing trend of relative moisture content in the topsoil for 12–15 years. The drought also involves water level falls in ponds traditionally used to irrigate crops. Then confined groundwater reserves suffer from increased water use for irrigation. The levels have recently sunk more than 20 m at some locations (Ber´enyi and Erd´elyi 1990). The area affected is virtually the same as in the case of free groundwater. After the depletion of the Quaternary aquifer of the alluvial fan, the Pliocene aquifers come into use and their pressure conditions are also now being affected.
155
Aridification in a Region Neighbouring the Mediterranean
No change <1 m 1− 2 m
Danube
2−3 m 3−4 m Budapest
N
0
15
30
45 km
´ Figure 12.6 Changes of groundwater levels in the counties Pest and Bacs-Kiskun (on the ´ Danube–Tisza interfluve) between 1967 and 1993 (Loczy and Szalai 1995)
156 3.2
Mediterranean Desertification Impact of Climate Change on Land Capability
For the survey of the agro-ecological impacts of drought, the same region, the Danube–Tisza interfluve, was chosen as a test area. In the land capability survey, topographic, climatic and soil parameters served to represent the physical environment (Table 12.4) and they were contrasted with the ecological requirements of arable crops. Particularly in sandy areas, land capability has been most severely restricted by the reduced availability of groundwater. Closely connected factors that control land suitability for arable crops were jointly assessed. Monthly precipitation was combined with soil texture and soil depth to groundwater table, viewed together with the grain-size composition of the parent material (Figure 12.7). A scoring scheme was applied and five suitability classes were identified. Table 12.4 Parameters of the land capability GIS for lowland areas of the Danube–Tisza interfluve
1 2–9 10–16 17 18 19 20 21 22 23
Horizontal dissection of the surface (drainage density + dirt roads + patches of forest, m km−2 ) Monthly mean temperatures for the growing season (◦ C) Monthly average precipitation for the growing season (mm) Total winter precipitation (mm) Genetic soil type (Hungarian classification system) Depth to parent material (cm) Soil texture Parent material Depth to groundwater table (m) Soil reaction and CaCO3 content
N Dabas
Danube
Cegléd
Kecskemét
Settlements Suitability classes: Danube
unsuitable
intermediate
poor
good
km
very good 0
5
10
Figure 12.7 Land capability of a test area in the central part of the Danube–Tisza interfluve, ´ based on soil parameters only (Loczy and Szalai 1995)
157
Aridification in a Region Neighbouring the Mediterranean
N Cegléd
Danube
Dabas
Kecskemét
Settlements Danube
Suitability classes:
unsuitable poor
intermediate good
km 0
5
10
Figure 12.8 Predicted land capability under reduced water availability, i.e. precipitation ´ reduced by 10 mm during the months of the growing season (Loczy and Szalai 1995)
The best land capability (shown as “very good” in Figure 12.7) indicates that on a given tract of land a wider range of crops can be grown with success. The most valuable agricultural areas of Hungary are those suitable for winter wheat, maize and sunflower growing at suitability levels of good and very good. The land capability GIS is able to simulate changes of agro-ecological potential under the conditions of reduced precipitation and lowered groundwater table. The assessment technique remains the same but in the database long-term average values are replaced by a set of data representing the conditions in the 1980s and early 1990s in more detail. The effects of reducing the average monthly precipitation by 10 mm were mapped, with regard to soil texture, to provide a land capability map for a drier scenario (Figure 12.8). 3.3 Natural Vegetation
The hypothesis that even small changes in climate could affect the distribution of vegetation was put forward. Microhabitats in a sandy area of the Danube–Tisza interfluve were identified and classified. Three complex spatial vegetation types were differentiated: those on sand dune hollows, sand dune summits and north-facing slopes. Each site type has a different microclimate, more or less extreme, and makes up a mosaic landscape pattern. The natural vegetation of these sandy areas is the best indicator of changing ecological conditions. The mosaic of habitats is occupied by plant associations with similar ecological demands (manifested by species composition and physical appearance). In the most adverse habitats, open sandy grasslands with low percentage cover develop into closed grasslands and occasionally even to arboreal associations (open or closed associations of Convallario–Quercetum; Szujk´o-Lacza and Kov´ats 1993).
158
Mediterranean Desertification
In the study area, species with a continental range predominate, and a little over one-third belong to the Mediterranean group. The ratio of Pannonian (endemic) species is relatively high, while adventives and cosmopolitans are negligible. As Pontian–Mediterranean elements are classed with the continental group, the Mediterranean character is even more pronounced. According to ecological indicators developed in Hungary, temperature and water demands of plants have been evaluated. In the cenological surveys, 10 test squares 5 × 5 m2 were identified for each microhabitat. The interpretation of the index of relative heat demand (TB) differentiated between the following types: • • • • •
mesophilous broad-leaved forest species (TB: 5) – 5.3% of flora submontane broad-leaved (TB: 6) – 17.8% of flora thermophilous (TB: 7) – 38.9% of flora submediterranean woodland and grassland species (TB value: 8) – 34.5% of flora plants of eu-mediterranean evergreen belt (TB value: 9) – 3.5% of flora
The distribution of flora types according to the index of relative groundwater or soil moisture (WB) showed a dry character: the cumulative ratio of categories WB 1–3 amounted to 80%. The most frequently observed WB 2 category indicated a particular Mediterranean influence. The types are as follows: • • • • • • • •
plants of extremely dry habitats of bare rock (WB: 1) – 22.9% of flora xero-indicators on habitats with a long dry period (WB: 2) – 45.9% of flora xero-tolerant species also occurring on moist soils (WB: 3) – 16.5% of flora plants of semi-arid habitats (WB: 4) – 9.3% of flora plants of intermediate semi-humid habitats (WB: 5) – 1.8% of flora plants of moist soils (WB: 6) – 1.8% of flora plants of wet soils, not drying out, and well aerated (WB: 7) – 0.9% of flora plants of wet soils tolerating short periods of waterlogging (WB: 8) – 0.9% of flora.
The index of continentality (CB) shows the degree of tolerance to extreme climate and indicates the continental character of the region. At the same time, the transitional nature is indicated by the high proportion of species in the CB 5 category. The types are distributed as follows: • • • • • • •
(sub)oceanic species, whole of Central Europe (CB: 3) – 2.8% of flora suboceanic, mainly in Central Europe but expanding to east (CB: 4) – 10.8% of flora intermediate with slight suboceanic–subcontinental character (CB: 5) – 31.5% of flora subcontinental species, eastern Central Europe (CB: 6) – 9.9% of flora (sub)continental species, main area in Eastern Europe (CB: 7) – 12.6% of flora continental species only reaching eastern Central Europe (CB: 8) – 12.6% of flora eu-continental, main area in Siberia and Eastern Europe (CB: 9) – 19.8% of flora
Investigations have suggested that in the test area (and in the broader environment) the typical continentality is accompanied by a remarkable (sub)mediterranean character. 3.4
Soil Dynamics
On the Danube–Tisza interfluve the shift of wind-blown sand was only stopped by the expansion of an open grassland vegetation surface. The decomposition of organic matter gradually produced humus in sufficient quantities to stabilize the quartz sand grains and hinder further movement. This is the origin of the humic sand, a characteristic soil type of dune summit levels in the Kiskuns´ag National Park. Some dune summits are still covered by blown sand or blown sand veneer. Most of the soil humus content is concentrated in the upper soil horizon. The open grassland (20–25% cover) is a pioneer association over surfaces of extremely adverse conditions for plants. They are strongly exposed to radiation, and the sandy soil shows very rapid infiltration and therefore poor
Aridification in a Region Neighbouring the Mediterranean
159
water retention capacity, and only contains traces of mineral clay colloids, important for both soil structure and plant nutrition. In hollows between dunes and flat areas the water table may be close to the surface, forming hydromorphic, often alkaline, soils. The most common genetic soil type here (V´arallyay et al. 1981) is double-layered humic sand under hydromorphic influence both from above and below (although it has to be noted that the influence of capillary water rise is certainly greater). Decreasing rainfall in the area has contributed to the gradual lowering of the water table. Alkaline ponds have decreased in area or dried up altogether (e.g. Szappan-sz´ek pond near F¨ul¨oph´aza). Direct contact between the groundwater and the salt-affected soils is interrupted, the dynamics of the solonchak soils cease, and leaching may even ensue. The sodium salts previously accumulated in the soil profile are transported by rainwater (available in a sparse but sufficient amount) into the subsoil or into the groundwater now present at a depth of some metres. Disregarding salt-affected horizons, hollows, on the whole, have soils with a higher water retention capacity and thus have more favourable ecological conditions. This allows development of a more closed vegetation, richer in species, and therefore the tolerance of this vegetation to aridification is much higher.
4 LAND USE The purpose of land-use studies was to detect and map spatial and temporal changes of land use between 1975 and 1991 and to find relationships between the changes and their possible causes. The test area, a 56 km × 56 km square, was again in the Danube–Tisza interfluve. This area is a mosaic of arable fields, pastures and discontinuous protected areas of the Kiskuns´ag National Park. Three LANDSAT images were classified and compared: 15 April 1975 (MSS), 3 August 1985 (TM) and 10 October 1991 (TM), and interpreted for land use using the ARCVIEW program. The methodology developed within the CORINE Land Cover Project was employed. ARC-INFO was used for comparing land-use changes. The database included mean annual precipitation (mm); mean date of the first frost day; mean annual minimum temperature; number of frosty days per year; roads and settlements; relief based on a digital elevation model with a 10 m contour line interval; genetic soil types; depth to groundwater (m) in 1967 and 1993; water surfaces (lakes); and land-use classification for the years of the images. The following outputs were derived from the database: – – – – – – – – – –
changes in forest area, 1975–1985 changes in forest area, 1985–1991 intersection of the changes, 1975–1985 and 1985–1991 union of the changes, 1975–1985 and 1985–1991 groundwater level changes, 1967–1993 composite of water surfaces and swamps with groundwater level changes between 1967 and 1993 composite of ground water level with elevation a.s.l. composite of land-use changes 1975–1993 with groundwater level changes union of water surfaces with soils union of changes 1975–1985 with changes 1975–1991
Analysing the changes over the entire monitoring period (1975–1991; Table 12.5, Figure 12.9) reveals the following trends: an increase in the area covered by forest (an increase of 7.8%), meadows (increased by 6.4%), orchards (increased by 26.1%), open water (increased by 29.4%), and a decrease in the area covered by vineyards (decreased by 20.3%). These changes are not great enough to be directly relatable to climatic trends, but there are some interesting features. The desirability of land-use changes can be evaluated from several different points of view, depending on whether the interest is economic or ecological. The growth in the area of forest is seen as a positive change, ecologically because forests help to conserve both soil and water, and economically because forests also provide a commercial return. The area covered by vineyards has
160
Mediterranean Desertification
Table 12.5 Land-use changes, 1975–1991 (hectares)
Settlements
Forests
Meadows Vineyards
Orchards
Settlements Forests
– –
Meadows
–
Vineyards
–
Orchards
–
Swamps
–
Water
–
–
–
Arable
–
3648
2766
1375
704
62 700
39 993
4479
8338 63 623
140000
–
–
Water
– –
– –
720
–
–
524
–
–
6262
–
–
27
–
245
−1973 (37.6%) – – 3193
–
+269 – (29.4%) 269 −1099 (1%) 914 130 359
160
1975 1991 %
120000
Arable
140
100000
Area (ha)
120 80000 100 60000 80 40000
Changes in % (1975 = 100%)
1991 total
– – – – +4940 146 58 – (7.8%) 138 +4022 53 6 (6.4%) 2023 942 −8926 495 (20.3%) 2 10 – +1167 (26.1%) – 831 88 2
Swamps
60
20000
40
0 Settlement Forest
Meadow Vineyard Orchard
Swamp
Lake
Arable
Land use
Figure 12.9 Land-use changes on the Danube–Tisza interfluve between 1975 and 1991
decreased considerably. The change has been mainly to arable land (64% of the changed area). It is interesting that only relatively small vineyards changed their land use, while larger vineyards usually survived. The decrease in the area covered by arable land was very small, and the land that changed became forest, meadow or, in one case, a reservoir. With dropping groundwater tables the area covered by
Aridification in a Region Neighbouring the Mediterranean
161
swamps has diminished, tending to become meadow. Some small areas that had been swamp became arable land or even vineyard, but only in those areas where groundwater subsidence was relatively rapid and groundwater reached great depths. Small areas of wetland lying near roads disappeared in great numbers and there is an interesting economic reason for this. These areas were easier to access than others lying further away from roads, and have therefore been most easily brought into alternative use. Table 12.5 indicates that there has been a surprising increase in the area covered by open water. This is not what would be expected in a climate of increasing aridity. The reason for this is that only lakes with considerable depth could be identified when classifying the space images. Although there has been a widespread decrease in the area of shallow water, there has been an increase in the area covered by deep reservoirs and fish ponds.
5 CONCLUSIONS There are three main conclusions: • There is a very well defined trend towards aridity in terms of temperature increases and precipitation decreases since 1900. (The differentiation of the rate of change of the mean annual temperature values between 1950 and 1989 is given below.) • Aridification processes have led to detectable changes of environmental factors. • Land-use changes in the last two decades have not yet shown a direct relationship with aridification. Annual mean temperatures during 1950–1989 were frequently higher than those during 1900–1949. The annual mean temperatures of the stations with the warmest climate increased more remarkably (+0.3–0.5 ◦ C). Monthly mean temperatures (particularly in the autumn–winter) also showed an increasing trend. The average fall in the annual temperature range for the 16 stations was found to be −0.56 ◦ C, suggesting a moderate reduction of continentality across Hungary. At stations with the lowest annual precipitation, an annual precipitation decrease was observed. Data from the stations with the highest annual precipitation (western Hungary) also showed a remarkable fall (a decrease of 40–65 mm). On average, a decreasing precipitation trend of −0.917 mm year−1 has been found. There has been a marked rearrangement in the temporal pattern (monthly distribution) of precipitation. This may have an important effect on traditional agriculture, if crops or pasture no longer grow well in an altered climate. The physical processes of aridification were studied mainly on the Great Hungarian Plain (the Danube–Tisza interfluve), where groundwater levels have fallen 2–4 m since the 1960s. A serious aspect of the aridification trend is the extremely reduced infiltration into the soils, and the poor recharge of groundwater. Natural vegetation tends to adjust to changed environmental conditions. The original clearlydistinguishable plant associations of the rolling sand dune areas are merging together (e.g. the open Junipero–Populetum association of dry dune summits and slopes). When the groundwater levels were higher, the dry dune summits had a very different microclimate to the wetter interdune hollows and supported different vegetation associations. Now that the hollows are drying out, the micro-environments are not so different. The survey of land-use change indicates no direct relationship between land-use changes and the growing aridity. Comparing those areas that underwent changes during 1975–1991 with the distribution of the partial areas of the Kiskuns´ag National Park, it is obvious that in many cases there is a direct relation between them. Those areas where the land use changed during 1975–1991 are at or near the border of the Park and the changing land use involves changing environmental conditions just at the border of the Park. Changes may endanger the natural ecosystems of the Kiskuns´ag National Park. Future research aims to define and delineate environmentally sensitive areas in Hungary, and to apply models designed to predict the impacts of aridification and land degradation processes at the catchment scale.
162
Mediterranean Desertification
REFERENCES Ambr´ozy P, Koflanovits E and K¨ov´er B (1990) A csapad´ekeloszl´as id¨obeli a´ trendez d´ese Magyarorsz´agon (Changes of the temporal distribution of precipitation in Hungary). Id¨oj´ar´as 94(2–3), 156–167. Ber´enyi P and Erd´elyi M (1990) A r´etegv´ız szintj´enek s¨ullyed´ese a Duna–Tisza k¨oz´en (Sinking confined groundwater under the Danube–Tisza interfluve). V´ız¨ugyi K¨ozlem´enyek 72(4), 377–397. Bussay A, Dunay S and Szeleczky M (1995) Hot dry weather diminishes 1994 crop yields in Hungary. Drought Network News, University of Nebraska 7, 2. ´ L´oczy D, Moln´ar K and Mika J (1996) Simulation of possible climate change effects on Husz´ar T, Kert´esz A, soil water content. Proceedings of the Climate Change Conference, Visegr´ad, Budapest, pp. 171–176. Koflanovits-Adamy E and Szentimrey T (1986) The variations of the precipitation amounts in the Carpathian Basin during the present century. Id¨oj´ar´as, 90(2–3), 206–216. L´oczy, D and Szalai, L (1995) Korszer s´ıtett term helymin s´ıt´es e´ s agro¨okol´ogiai k¨orzetes´ıt´es f¨oldrajzi inform´aci´os rendszer felhaszn´al´as´aval (Land capability survey for agroecological zoning using GIS in Hungary). F¨oldrajzi ´ Ertesit 44(1–2) 23–37. Matyasovszky I (1995) Temperature and precipitation trends in the Hungarian Great Plain during the present century. MEDALUS Working Paper No. 58, London. Moln´ar K and Mika J (1997) Climate as a changing component of landscape: recent evidence and projections for Hungary. Zeitschrift f¨ur Geomorphologie, Supplementband 110, 185–195. P´alfai I (1991) Az 1990 e´ vi asz´aly Magyarorsz´agon (Drought in Hungary in 1990). V´ız¨ugyi K¨ozlem´enyek 73(2), 117–133. P´alfai I (1995) A Duna–Tisza k¨ozi h´ats´ag v´ızgazd´alkod´asi probl´em´ai e´ s megold´asuk lehets´eges u´ tjai (Water management problems on the Danube–Tisza interfluve and possible solutions). V´ız¨ugyi K¨ozlem´enyek 76(1–2), 144–164. P´alfai I (1996) A talajnedvess´eg e´ s a talajv´ız´all´as v´altoz´asai az Alf¨old¨on (Changes in soil moisture and groundwater level in the Great Hungarian Plain). V´ız¨ugyi K¨ozlem´enyek 78(2), 207–218. Somogyi S and Marosi S (eds) (1990) Magyarorsz´ag Kist´ajainak Katasztere I (Microregions of Hungary). GRI Hungarian Academy Science, Budapest. Szentimrey T (1994) Magyarorsz´agih m´ers´ekleti adatsorok inhomogenit´as´anak becsl´ese (Estimation of inhomogeneity for the temperature series of Hungary). Climatological and Agrometeorological Papers, No. 2 Hungarian Meteorological Service, Budapest. Szujk´o-Lacza J and Kov´ats D (eds) (1993) The Flora of the Kiskuns´ag National Park in the Danube–Tisza Mid-region of Hungary, Vol. 1. The Flowering Plants. Hungarian Museum of Natural History, Budapest. Tar K (1992) T´urkeve e´ ghajlat´anak megv´altoz´asa (Changing climate of T´urkeve). In Papers of Conference on the 145th anniversary of the birth of Kabos Hegyfoky, Debrecen, T´urkeve, pp. 156–164. V´arallyay Gy, Szcs L, Mur´anyi A, Rajkai K and Zilahy P (1981) Magyarorsz´ag agro¨okol´ogiai potenci´alj´at meghat´aroz´o talajtani t´enyezk 1:100 000-es m´eretar´any´u t´erk´epe (Map of soil properties controlling the agroe´ cological potential of Hungary on 1 to 100 000 scale). F¨oldrajzi Ertesit 30, 235–250.
13
Soil Salinization in the Mediterranean: Soils, Processes and Implications
L. POSTIGLIONE
Faculty of Agriculture, University of Naples Federico II, Portici, Italy
1 INTRODUCTION In all Mediterranean countries there are areas where the soils have become saline. This can pose a great problem for soil use and conservation, especially as their salinity is continually increasing due to the characteristic climate of dry, hot summers and mild, wetter winters. The excessive presence of salts, especially sodium salts, alters or destroys the soil structure as well as increasing the swelling and dispersion of clay aggregates. These smaller soil particles, which are almost always richer in nutritive elements and in organic matter, become more liable to be transported away by water or wind. The soil thus loses its fertility and the environment can become desertified. During the summer, high temperatures cause a considerable increase in evaporation and this, together with the lack of summer rainfall, favours salt accumulation. In the autumn–winter period, rain is often of short duration but of considerable volume and intensity, causing clay and organic colloids to swell and deflocculate, altering the soil structure. Intense rainfall events produce torrents that wash the easily eroded soil away, sometimes causing landslides and rockfalls. In soils affected by salinization, cultivation is rendered difficult in the short term. The majority of crop species suffer due to the change in the water/oxygen ratio. Plants are subject to water and salinity stress, and to consequent nutritional deficits, reducing yields considerably. In the medium term many plants cannot survive and the symptoms of land degradation emerge, while in the long term desertification occurs.
2 HISTORICAL PERSPECTIVE Agriculture has been practised in the Mediterranean since Neolithic times (c. 10 000 BC). Continual cultivation, especially as populations have expanded and technology has allowed it, has in many places over-exploited the soil and impoverished it. Intensive cropping and excessive use of mineral fertilizers continue to make the situation worse. In addition, in this arid or semi-arid climate, water for agriculture is an increasingly limited resource as more is needed for urban and industrial use. Therefore economics dictate that water known to be saline and bad for soil fertility is now frequently used for irrigation of crops. Sometimes projects designed to aid crop production fail due to widespread salt accumulation. For example, in Egypt, following the construction of the Aswan Dam on the River Nile, 60% of farmland was affected by salinity. In particular, the highly clayey soils of the Nile Delta are affected by salinity and alkalinity (sodic-saline), due to continuous irrigation with saline groundwater (Chanduv`ı 1977). It is worth remembering that the decline of ancient civilizations, such as those that flourished in Mesopotamia, was as much due to the salt accumulation in their irrigated soils as to warfare or any other reason. The problem of salinization in soils is assuming greater prominence today. The greenhouse effect caused by the increasing levels of carbon dioxide and other gases in the atmosphere appears to be leading to a slow but continuous rise in mean annual temperatures in many places. As a result of Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
164
Mediterranean Desertification
this, scientists foresee two possible scenarios for the future (Postiglione 1991). According to the first scenario, sea-water evaporation will increase and consequently cloudiness will increase too, causing more rainfall in dry, pre-desert and desert areas. Both agricultural and forestry production will increase, induced by the effects of a CO2 -rich atmosphere on the photosynthesis and physiology of plant production (Idso 1989). The second scenario is not so optimistic and suggests that there will be a decrease in rainfall during the summer months, an increase in evapotranspiration and a consequent increase in soil salinization. The extent of water courses will be reduced and areas that were formerly temperate will become arid (Adams et al. 1990). Water available for irrigation is likely to have an even higher salt content, due to concentration following evaporation, again promoting soil salinization and desertification (Bultot et al. 1988).
3
DEFINITION
Salinity is a general term including various soluble salts that can affect soil and water in different ways. A preliminary classification distinguishes saline soils, saline–sodic soils and sodic soils (Table 13.1). Electrical conductivity is used as a measure of saline concentration in water (ECw ) and saturated soil paste (ECe ) and is expressed as dS m−1 . Sodium ions are particularly important as they drastically affect soil characteristics and also the absorptive capacity of the plant root system. The ESP (exchangeable sodium percentage) index is the ratio between sodium ions and the total cation exchange capacity. Saline soils, characterized by the predominance of Ca and Mg ions, are white, with a slightly alkaline reaction and are generally well structured. However, the presence of salts leads to an increase in the osmotic pressure of the circulating solution, making it difficult for the plant roots to absorb water and nutritive elements, which in extreme cases may cause the plasmolysis of the root cells. The overall situation is aggravated by the frequent droughts during the summer in Mediterranean environments. The excessive presence of certain anions or cations may affect the absorption and development of some plant species, while other ions (Cl, B and Na) may even be toxic. Sodic soils have an acceptable salinity (ECe < 4), but they are found in areas of low rainfall and are characterized by the excessive presence of sodium ions, which causes deflocculation of the clay colloids and hence the loss of structure. This in turn leads to a reduction in permeability, with the ponding of water during the rains or irrigation and a marked lack of soil oxygen and nutritive elements. In times of drought a surface crust or deep cracks are formed. All these circumstances have an adverse effect on the root system, which in some cases suffers irreversible damage, and on the whole underground biomass, exacerbated both by the high pH and by the fact that sodium ions in excess are toxic for many plant and micro-organism species. Sodic–saline soils occur more frequently in Mediterranean environments, and have the same drawbacks as those found in saline and sodic soils, but to a greater degree of severity, i.e. chiefly the dispersion of colloids and concomitant problems, as well as high osmotic pressure in the circulating solution. Their pH is slightly lower, especially when there are some anions present, although it increases if the anions or some soluble salts are leached. Soils characterized by high salinity, and especially sodic soils, have different problems according to their location, soil texture and soil profile. If they are situated on a slope, they support few Table 13.1 Properties of saline, sodic–saline and sodic soils (Riverside US Salinity Laboratory)
Soils Saline Sodic–saline Sodic
ECe
ESP
pH
>4 dS m−1 >4 dS m−1 <4 dS m−1
<15% >15% >15%
<8.5 <8.5 >8.5
Soil Salinization in the Mediterranean
165
Mediterranean maquis species and are easily eroded by the impact of rainfall until bare rock is exposed, which is one of the frequent causes of desertification of Mediterranean hill soils. In fact, the situation of the Mediterranean hill soils is itself very fragile, even when salinity problems are not pressing, though it may be possible to help the situation with suitable cultivation techniques and suitable crops or plants (Postiglione 1988; Postiglione et al. 1993). With regard to this issue there was a tragic event in May 1998 in Sarno (near Salerno, southern Italy), when a rainfall event, not even of great intensity, moved all the soil, which was mainly volcanic deposits above calcareous rocks, down the valley. As a consequence of this event, in addition to the destruction of the woodland vegetation and bedrock exposition, the mud flooded an entire village and a large number of deaths were recorded. On the plains, there are sometimes marshes that permit only sparse vegetation of palustrine plants. In better situations, when abundant rainfall occurs, surface ponding may occur that creates well-known problems concerning soil structure and plant root asphyxia. Subsequently, such surface waters disperse as runoff, taking with them clay which is already suspended due to colloid dispersion. Thus the finer particles are removed, i.e. those that are richer in nutrients, organic matter and microorganisms, essential for soil fertility. At times, besides the removal of fertile material, there are cases in which the lower soil horizons have highly toxic concentrations, or limestone or large quantities of sand accumulate, and degradation becomes irreversible (Schertz 1983).
4 CAUSES OF SALINITY IN THE MEDITERRANEAN Soil salinization may be of a primary nature, when salt accumulation arises through pedogenetic processes, or of secondary origin, due either to abiotic factors such as excessive evaporation or sea-water infiltration, or resulting from human intervention, chiefly saline water irrigation. Primary salinization occurs during the pedogenesis of certain soils. In the weathering of some rock types in extremely dry environments the normal leaching of salts and especially of cations does not take place. A great quantity of salts and cations accumulate over time in the soils. Excessive evapotranspiration, typical in dry environments (Figure 13.1), causes salinization since only the solvent (water) evaporates. As a result, the surface layers continuously accumulate salts found in the circulating solution, both in the upper and underlying layers, and the circulating solution present in the latter rises by capillarity consequent to the evaporation. This fact is very important in Mediterranean regions in which evaporation reaches even 8–10 mm day−1 . Naturally, the phenomenon is accentuated if there is saline groundwater close to the soil surface. Sea-water infiltration frequently occurs along coastal plains, being particularly marked in those areas which in different geological eras, or even in historical times, experienced the phenomenon of bradyseism, which occurs quite commonly in Mediterranean environments. Clearly, in this case the salts in question almost entirely consist of sodium chloride. Soil salinization from irrigation depends on the quality and salt concentration of the water used and the nature of the soils. In fact, the damage caused by using saline water increases in particular in high clay-content soils. By contrast, the rainfall pattern in the rainy season is very important because it may be conducive to the leaching of the salts from irrigation water. Other anthropogenic causes of salinization may include overgrazing and deforestation in semiarid environments, the excessive use of chemical products, and the contribution (via the air or water) of pollutants emitted by industry. In particular, overgrazing in semi-arid environments leads directly to desertification when even the poor-grade pasture diminishes and no other fodder resources are available (Szabolcs 1994). In reality, soil salinity almost always stems from the concurrence of two or more of the above factors. However, constant or increasing salinity is chiefly caused by the use of highly saline irrigation water, compounded by excessive evapotranspiration in dry areas. Nevertheless, in the soils of the Mediterranean Basin, given the hot, dry climate in the spring–summer period, only by resorting to irrigation can high crop yields be achieved. As long as there is water with a low salt content available, irrigation is also a means of leaching salts and thus improving the condition of such soils. In reality, water with a salinity of ECw < 3 dS m−1 (similar to about 2‰ of the total concentration)
166
Mediterranean Desertification
50
mm per decade
40
30
20
10
2 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 13.1 Example of potential evapotranspiration trend in relation to rainfall: Scafati, Campania Region, 25 years average (Postiglione 1972)
is considered useful for irrigation. The problem becomes more complicated when the water available for irrigation has a high salt content. As a vital resource for agriculture in Mediterranean environments, irrigation water may be extracted from surface water (springs, rivers, streams) or groundwater (phreatic boreholes, artesian wells). In the case of surface water resources, in the Mediterranean area springs are sometimes salt-rich since such water passes through rock layers and saline or sodic soils where there is an excess of sodium which has stayed in situ during the pedogenetic process, or because there is sea-water infiltration. Such infiltration frequently occurs in some aquifers when well supplies are over-abstracted or when the groundwater fails and is not recharged due to a shortage of rain during the winter. Both types of process may leave a void to be filled by sea-water infiltration. Of course, the situation becomes even more serious when excessive abstraction means that wells have to be sunk deeper, with the consequent risk of reaching saline groundwater. A case of mixed-origin salinization has occurred in the Sele River Plain (southern Italy), in the area called the Paestum Plain, where there are six springs with a considerable salt content not associated with high temperatures, and a total flow of 3 m3 s−1 . Mean electrical conductivity varies from 11.9 dS m−1 in the most saline spring with the greatest flow, to 8–9 dS m−1 for two other springs and 3.9 for the other three. A study was conducted (Celico et al. 1982) on the origin of such salinity. That study examined the following aspects: the geological and chemical nature of the catchment basin; the pattern of the piezometric surfaces in the surrounding mountains; the various cations and anions present in the water and the relationships between them; the effects of using isotope 18 O; soil and subsoil properties in the plain; and the circumstances whereby, during eustatic movements in the Quaternary, the current springs were actually below sea level. The study in question showed that
Soil Salinization in the Mediterranean
167
the mineralization of such waters is due to various degrees of mixing between low salinity calcium bicarbonate water and sea-water. On the basis of the findings, the following causes were ruled out: derivation from connate water or the circulating solution in evaporitic soils, and the contribution of deep fluids and sea-water infiltration due to groundwater abstraction. “A model is proposed in which the hydrodynamics of the groundwater in the carbonatic massif are such as to remove the sea water trapped below the source level during the last eustatic movements in the Quaternary” (Celico et al. 1982). In a coastal area of Apulia, again in southern Italy, it was observed (Caliandro et al. 1997a) that some wells used for irrigation supplied water with conductivity varying from 1.10 to 4.20 dS m−1 and that water quality is not affected by the distance of the wells from the sea, but by the heterogeneity of the karst aquifer of the area in question.
5 EXTENT OF THE PROBLEM IN THE MEDITERRANEAN AND IMPACT ON PRODUCTION AND ENVIRONMENT 5.1 Soils
Saline soils are a problem for all the countries in the Mediterranean area, and this problem becomes more and more serious each year because of the climate characteristics and human intervention, which is not always careful to safeguard natural resources. Consequently, in the future, if appropriate measures are not adopted, agricultural yield will decrease consistently, while environmental problems will increase towards desertification, which already affects thousands more hectares each year. Following trials carried out in different countries, the serious damage to soil structure and reduction of fertility caused by salinity has emerged. So, in the Sele River Plain (southern Italy), on a clayey-silty soil treated for six years with saline water (with 1% NaCl added), the sodium was observed to cause deflocculation of the clay particles, thereby altering soil porosity, i.e. reduced macroporosity and increased microporosity (a typical blockage of the macropores with the formation of microporosity). The increase in microporosity meant that more water was retained, although this was retained with greater force and the plants were not always able to draw such water with their suction capacity (Tedeschi et al. 1996). In central Sicily, in saline-water irrigation trials conducted for several years on two different soils, salt accumulation was observed, in so far as the winter rainfall was too low to ensure complete leaching. Furthermore, in a vertic soil, an increase in magnesium and sodium soluble ions was observed with the shift from calcic solonchak to magnesium solonchak, and an increase in exchangeable sodium percentage (ESP), with a consequent worsening of the soil structure (Fierotti et al. 1982). In Greece, saline soils are present and diffused over soils on flat areas due to natural–ecological factors, both abiotic (climate, geomorphology, hydrogeology) and biotic (vegetation, soil fauna). Human activities have been very important, especially with the extension of irrigation and undisciplined use of saline water at the beginning of the 20th century, which has caused over-pumping, and the consequent sea-water infiltration into the groundwater layer. Sustainable management of groundwater resources together with the restoration of drained freshwater wetlands is carried out whenever it is possible (Zalidis 1998). In Cyprus, sodic and saline soils are present due to irrigation with salt-rich water, which is used because of the scarcity of freshwater. The use of fertilizers and municipal wastewater also contributes to the salinity. In order to limit secondary salinization damage, modern irrigation technologies and cropping systems are now recommended (Papadopoulos 1998). In Turkey, two groups of saline soils are present: hydromorphic saline alluvial soils and solonchak soils. Both groups may have saline, sodic and saline–sodic classes (S¨onmez 1997). In Israel, there are four major areas that are affected by salinity: Galilee (north), Western and Central Negev (centre) and Arava valley (south). In northern Israel, salinity causes different degrees of damage and it is connected with high evaporation and irrigation water quality, particularly when
168
Mediterranean Desertification
treated sewage water is used. In coastal areas sea-water infiltration is widespread. In central Israel, soils are damaged mainly by increasing sodicity. For cropping of these soils it is recommended that the optimal land use is chosen according to site characteristic parameters such as water quality, duration of use, soil type, crop type, presence of drainage tiles, etc. In Israel, salinity causes a big decrease in cotton, tomato and corn seedling survival, reduction of citrus tree yield and reduction of the crop development rate in deciduous trees and all perennials, i.e. pears, peaches and plums (Nadler 1998). In Spain, salt-affected soils are widespread. Saline soils are present in coastal areas, where the main salt affecting the soil is sodium chloride. In the Guadalquivir river valley and the Marismas, located at the estuary of the river, alkalinity is the prevailing factor. Alkalinity is also present in the Ebro river valley (north-east Spain) and also in some southern areas of Spain. Along the Mediterranean coast the problem of soil salinity is increasing due to scarcity of precipitation and irrigation with low quality water (Moreno 1998). In Mediterranean regions of the African continent, water and soil salinity problems are particularly severe, principally in areas where rainfall is less than 500 mm per year. Such areas are found in Egypt, Libya and Tunisia (Chanduv`ı 1997). 5.2
Vegetation
Vegetation is affected by salts in the soil to different degrees, with some plants being more susceptible to certain ions than others. Generally salinity causes a decrease in soil hydraulic conductivity and root aeration, and an increase in resistance to root penetration. Moreover, roots meet greater difficulty in suction of water and absorbtion of nutritional elements. Under these conditions, in natural ecosystems only Mediterranean maquis vegetation will grow, especially those plants which have adapted themselves to salinity and drought conditions. In extreme cases excessive salinity can cause the disappearance of some plant species, causing a reduction in biodiversity. In agro-ecosystems it is worth stressing that when crops are grown on saline land or land irrigated with saline waters, even if salt-resistant crops are used, a large number of alterations occur in plants (e.g. osmotic adjustments and various reactions) which affect the various organs from the roots to the stems and leaves, and which may actually prevent plant growth. These alterations depend on the salt concentration and the species, variety and age of the plants. The most widely studied and evident variations occur in the leaves, and concern both the morphology (leaf surface, surface/weight ratio) and physiology, in particular the gas exchanges and photosynthesis. Such alterations naturally affect yield quantities, which decrease to a different extent according to the circumstances, and product quality, which generally becomes less desirable. In fact, there are some tables, regarding some main crops, in which it has been shown that potential yield reduction is a function of soil ECe (Ayers and Westcot, 1985). Regarding this issue there is much literature about certain crops. For example, in southern Italy there have been studies on tomatoes (Barbieri et al. 1990; Caruso and Postiglione 1993), eggplant (Ruggiero and Perniola 1992; Barbieri et al. 1994; Ruggiero et al. 1994; Sifola et al. 1995), peas (De Pascale and Barbieri 1996), snap beans (De Pascale et al. 1996), broad beans (De Pascale and Barbieri 1997), watermelons (De Pascale et al. 1998), peppers (De Pascale et al. 2000) and sunflowers (d’Andria et al. 1997; Tedeschi et al. 1997).
6
MANAGEMENT OF SODICITY AND SALINITY
There are several criteria and techniques for conserving soil and limiting salinity damage. Besides irrigation with fresh water, mention should also be made of the benefits of adding organic matter. This, in fact, besides supplying nutritional elements, develops a stabilizing action on pH, and above all it develops a protective action on mineral colloids. In the presence of organic matter, clays can resist the dispersing action of sodic cations, preserving the crumb structure for longer. Unfortunately, in the Mediterranean environment organic matter is in short supply and, where it is available, it is subject to rapid oxidation due to the high temperatures. Hence its effects are not long-lasting.
Soil Salinization in the Mediterranean
169
Of great use, particularly in sodic soils, is the addition of appropriate correctives, such as chalk (CaSO4 ), which reduces alkalinity but obviously increases salinity. Its application must therefore be combined with rapid salt-leaching conditions (rain or irrigation and, fundamentally, good drainage). With such soils great attention should also be paid to using mineral fertilizers, both in terms of quantity and composition. With regard to fertilizer composition, in sodic soils it is preferable to use “physiologically acid” fertilizers, e.g. fertilizer in which cations are absorbed by plant roots and anions remain in the circulating solution. In arid and semi-arid climates, freshwater resources are not infinite, and more and more water is needed for civil or industrial purposes, so the use of saline water for agriculture will continue to increase. In these circumstances it is essential to monitor the situation as regards the use of vegetation or crops and land management techniques, to limit soil degradation and stop the advance of the desert. The situation improves when land improvement schemes are conducted either through appropriate channelling or, better still, through drainage with underground pipes, which allow the leaching of salts into deep layers. Under these conditions in areas with sufficient rainfall during the autumn–winter period, such leaching occurs naturally. In some cases, if low salt-content water is available, leaching may be carried out by irrigation. As regards irrigation water, attention is nowadays focused on the nature of the salts, in so far as the presence of calcium may mitigate the adverse effects of sodium. Hence if the ratio of sodium to calcium and magnesium, the SAR (sodium absorption ratio), is less than 10, the water can be used without causing damage. When the SAR increases, some limitations occur, and when the SAR exceeds 26, water is generally not utilizable. In many Mediterranean regions such as Israel, Egypt, Libya and Tunisia and in several areas of Spain, Italy, Albania, Greece and Turkey, where water is chiefly saline, “it nonetheless constitutes a resource” (Chanduv`ı 1997), and if well-managed, may contribute to limiting damage from soil salinity. Thus in the Nile Delta in Egypt, groundwater is abstracted, while in the same area and in the Nile valley, drainage water from settlements on higher ground is used: in all cases, such waters are saline. In Libya, in the Turga area east of Tripoli, 3000 ha are irrigated with water with a conductivity starting from 2.5 dS m−1 , and in some coastal areas well water is used which, due to continuous pumping, has led to sea-water infiltration. In Fezzan, water is used with an electrical conductivity that ranges from 9.0 to 1.79 dS m−1 (Chanduv`ı 1997). In southern Tunisia, in the Tatouine area, the climate is pre-Saharan, with irregular winter precipitation (120 mm annual average) and hot dry summers. The soils have developed over sedimentary materials, are sandy, rich in carbonates and sulphates, are salt-affected, and the natural vegetation is very scarce, dominated by perennial shrubs. This area has been the subject of research, using remote sensing, to study the water infiltration rate for an eventual irrigation scheme (Escadafal et al. 1993). In the Carpathian Basin, irrigation water as a solvent reactant and transporting agent plays a decisive role in the development of salt-affected soils, but above all it is connected with drainage (Varallyay 1998). In Apulia, southern Italy, a commonly used technique to prevent excessive soil salinization is to irrigate very frequently with amounts of water exceeding crop evapotranspiration so as to keep the soil sufficiently wet throughout the season and, at the same time, ensure a little continuous leaching of salts (Cavazza et al. 1984). For the above reasons, all the studies and research being conducted on the possibility of using saline water in various countries are of great interest for the future of agriculture and in reducing the risks of desertification. Of particular importance is the relationship between “irrigation = salt in” and “leaching = salt out” (Papadopoulos 1998). In other words, in a dry environment, when saline-water irrigation is carried out to avoid progressive salt accumulation in the soil, a quantity of water must be applied in addition to that normally calculated on the basis of the climate, soil and crops, so as to promote the leaching of the applied salts and their removal, by means of drainage, from the root zone (known as the leaching requirement). As long ago as 1929, E. de Cillis, a scholar in the field of dry-farming in Mediterranean environments (southern Italy and Libya), wrote: “Saline water will have to be used (in irrigation) in great abundance and on well-drained soils, so that continuous washing takes place and salts do not
170
Mediterranean Desertification
accumulate.” Nowadays the amount of water needed to satisfy the leaching requirement (LR) is calculated according to the electrical conductivity of the irrigation water (ECw ) and the electrical conductivity of the soil saturation extract (ECe ) for a given crop appropriate to the tolerable degree of yield reduction, according to the formula LR = ECw /(5ECe − ECw ). Moreover, there are cases in which the water cannot wet all the soil uniformly (e.g. due to the presence of cracking, or poor irrigation systems for distributing water). Therefore, appropriate consideration must also be given to the leaching efficiency. This is 100% in sandy soils using the sprinkler distribution system, but may fall to 30% in poorly structured or cracked, clayey soils. Thus the percentage obtained with the above formula is divided by the leaching efficiency percentage, and the amount of water required increases further. In order to be certain of actual salt leaching, the nature of the soil must be duly considered. In loose grounds, leaching usually occurs without great difficulty; in medium soil, two-layer tillage appears useful (ploughing to ∼40 cm and ripping to ∼60 cm). In compact clayey soils, as are the majority in many Mediterranean areas, a good drainage network first needs to be built, which involves a considerable technical and economic commitment. In Egypt, for example, rudimentary forms of drainage along the course of the Nile date back to ancient times. Drainage programmes have been operating along modern lines since 1909, and by 1997, 1.9 million hectares of farmland had been drained, with much of the drainage water being re-used, as stated above, to irrigate other land (Ramadan 1998). In pluriennial trials conducted in the Sele River plain, southern Italy, on a clayey-silty soil (fluentic xerochrepts) with three irrigation frequencies (every 2, 5 and 10 days), using freshwater (ECw = 0.54 dS m−1 ) and four levels of saline water obtained by adding 0.125, 0.25, 0.5 and 1% commercial NaCl (ECw = 2.30, 4.43, 8.46, 15.73 dS m−1 ), the highest soil salinization values (ECe = 20 dS m−1 ) were found at the end of each irrigation season in the most saline treatment (Figure 13.2). The latter fell to only 7 dS m−1 after the rainy season, while in the freshwater treatment the ECe varied little and returned to normal after the rainy season. After six years, despite seasonal oscillations, the ECe values showed an increasing trend in the treatments irrigated with various salinity levels, which was more accentuated in the treatments irrigated every two days compared with those irrigated every 10 days. With the latter treatment a larger expansion of the sheet of water occurs in the underlying
Figure 13.2 Piana del Sele experimental field after six years of irrigation by waters with different salt amounts. From the left are plots watered by fresh water and by water with 0.125, 0.25, 0.5 and 1% NaCl respectively. Beginning with the 0.25 treatment, the presence of saline efflorescence increasing with salt concentration is evident
Soil Salinization in the Mediterranean
171
layers. Moreover, the ESP values also increased considerably, with the pH shifting from 7 to 8. The structure proved degraded due to the deflocculation of the clay caused by the sodium, with very low soil permeability (Postiglione et al. 1995). Hence from these trials it emerges that by spacing out watering events, less salt accumulation occurs in the upper layer. Nowadays it is recommended that watering events (excluding drip irrigation) are spaced out, in so far as if the surface is kept continuously wet, evaporation causes salt concentration in the surface soil layers. In fact, when the surface is constantly wet, roots absorb water first from the upper layers, unlike when the soil surface is dry due to spaced-out irrigation events when roots absorb water from the lower layers. In other words, evaporation and root absorption processes together cause surface saline concentration when frequent irrigation occurs (Ragab 1996). It is worth stressing that among irrigation systems, when saline water is used, drip irrigation causes less stress to plants, since it maintains a continuous supply of water in the root zone. There is therefore less probability of a higher salt concentration occurring. In fact, since water is released over small areas, leaching processes prevail on evaporation and root absorption shares. Finally, when only a small amount of freshwater is available, saline water can be used for saltresistant crops and freshwater for salt-intolerant crops, or it is possible to alternate saline-water irrigation with freshwater irrigation, so that the latter can accelerate salts leaching to deeper layers. Another strategy, which appears preferable, is to use freshwater for the young stages of the crops, and to use saline water subsequently (Shalhevet 1994). In Apulia, on the basis of initial results from irrigation trials with saline water on clayey-silty or clayey-sandy soils, it emerged (Caliandro et al. 1997b) that the winter rains are sufficient to leach the salts added by irrigation and that in environments with an autumn rainfall of about 400 mm it does not appear necessary to add the quota of leaching water. However, where irrigation is abundant, the soils appear to be becoming sodic. In other areas, on soils rich in iron and aluminium sesquioxides, which have good structure, permeability and good natural drainage, saline-water irrigation is traditionally applied to vegetable crops in alternate years. The salt is thus leached by the rainfall occurring during two winters. Finally, it must be recalled that one of the main means of maintaining soil vitality is the presence of vegetation, whether trees, shrubs or crops. When deficient soils are completely abandoned, they undergo further degradation and decline towards desertification. Also harmful for soil conservation is the adoption of unsuitable cultivation systems (i.e. single-crop farming, very intensive cultivation) and unsustainable vegetation exploitation (e.g. excessive cattle densities causing overgrazing, destruction of forests). In the last few years there has been more research on species and varieties resistant to different levels of salinity and on their tolerance to some ions (Cl, B, etc.). The aim of such studies is to be able to advise on crops, cropping systems and technologies in order to make agriculture possible on saline soils or by using salt-rich water to irrigate. There have also been attempts to grow halophytes (salt-tolerant species) on saline soils in order to produce biomass for conversion into bio-fuels. All these interventions to restore degraded soils due to excessive salinity and their use according to a modern concept of agriculture (”sustainable agriculture”) are undoubtedly fundamental to providing an efficient defence against the slow but progressive advance of the desert.
REFERENCES Adams RM, Rosenzweig C, Peart RM, Ritchie JT, McCarl BA, Glyer JD, Curry RB, Jones JW, Boot KJ and Allen LH Jr. (1990) Global climate change and US agriculture. Nature 345(6272), 219–224. Ayers RS and Westcot DW (1985) Water quality for agriculture. Irrigation and Drainage Paper 29, FAO, Rome. Barbieri G, Caruso G and Postiglione L (1990) Response of processing tomato to irrigation with saline water. Proceedings of the 1st Congress of the European Society of Agronomy, 5–7 December, Paris, pp. 5–6. Barbieri G, De Pascale S and Sifola MI (1994) Effetti della frequenza di irrigazione sulle funzioni di risposta produttiva della melanzana (Solanum melongena L.) alla salinit`a. Riv di Agronomia 28, 241–252. Bultot F, Dupriez GL and Gellens D (1988) Estimated annual regime of energy-balance components, evapotranspiration and soil moisture for a drainage basin in the case of a CO2 doubling. Climate Change 12, 39–56. Citato da Idso, 1989.
172
Mediterranean Desertification
Caliandro A, Rubino P and Lonigro A (1997a) Monitoraggio di acque di falda a diverso contenuto in sale. Agricoltura e Ricerca, XIX, 171, 13–16. Caliandro A, Rubino P and Lonigro A (1997b) Evoluzione dei terreni irrigati con acque salmastre. Agricoltura e Ricerca, XIX, 171, 5–12. Caruso G and Postiglione L (1993) Effetti dell’irrigazione con acqua a diversa concentrazione salina sul suolo e su cultivar di pomodoro da industria (Lycopersicon lycopersicum L.). Riv di Agronomia 27, 211–219. Cavazza L, Chisci GC, Fierotti G and Lauciani E (1984) Aspect of irrigation and drainage in problem soils in Italy. Proceedings of the 12th Congress, International Commission on Irrigation and Drainage, 28 May–2 June, 1984, Fort Collins, Colorado, pp. 333–343. Celico P, de Gennaro M, Ferreri M, Ghiara MR and Stanzione D (1982) Geochimica delle sorgenti mineralizzate della Piana di Paestum (Campania, Italia). Periodico di Mineralogia 51, 249–274. Chanduv`ı F (1997) Salinity problems in some countries in the Mediterranean Basin. Proceedings of 1st Transnational Meeting on Salinity as Limiting Factor for Agricultural Productivity in the Mediterranean Basin, 24–25 March 1997, Leone and Steduto, Napoli, pp. 77–86. D’Andria R, Lavini A, De Lorenzi F, Alvino A, Martorella A, Calandrelli D and Tedeschi P (1997) Growth analysis of field-grown sunflower (Helianthus annuus L.) under different salt concentrations of irrigation waters. In Water Management Salinity and Pollution Control toward Sustainable Irrigation in the Mediterranean Region, Vol. IV, 22–26 September 1997, Valenzano, Bari, pp. 381–395. De Cillis E (1929) Trattato delle coltivazioni: Agronomia. Della Torre, Portici. De Pascale S and Barbieri G (1996) Effetti della salinit`a del terreno sull’accrescimento e sulla produzione di una coltura di pisello da consumo fresco (Pisum sativum L.). Italus Hortus 2, 9–17. De Pascale S and Barbieri G (1997) Effects of soil salinity and top removal on growth and yield of broad bean as green vegetable. Scientia Horticulturae 71, 147–165. De Pascale S, Ruggiero C and Barbieri G (1996) Consumptive water use and plant growth and water relations of snap bean. Acta Horticulturae 449(2), 649–655. De Pascale S, Ruggiero C and Barbieri G (1998) Consumptive water use and plant growth of watermelon as affected by irrigation and N fertilization. Acta Horticulturae 458, 49–56. De Pascale S, Ruggiero C and Barbieri G (2000) Growth, yield and quality of pepper (Capsicum annuum L.) as affected by irrigation water salinity. Acta Horticulturae 537, 687–695. Escadafal R, Pontanier R, Belghith A and Mtimet A (1993) Remote sensing of potential infiltration rate of arid soils from Tunisia. Proceedings of the IV International Conference on Desert Development, Mexico City, 25–30 July 1993. Fierotti G, Lombardo V, Sarno R and Stringi L (1982) L’action des eaux saumatres sur la fertilit`e des sols argileux. In F Lanza (ed.) Evolution du Niveau de Fertilit`e des Sols dans Differents Systems de Culture: Criteres pour Mesurer cette Fertilit`e . Istituto Sperimentale Agronomico, Bari, 1982, pp. 367–380. Idso SB (1989) Carbon Dioxide and Global Change: Earth in Transition. IBR Press, Temple, Arizona. Moreno F (1998) Soil salinity in Spain and its impact on agriculture: research activities carried out by the group of IRNAS. Proceedings of 1st Trans-national Meeting on Salinity as Limiting Factor for Agricultural Productivity in the Mediterranean Basin, 24–25 March 1997, Leone and Steduto, Napoli, pp. 43–54. Nadler A (1998) Salinity problems in Israel and suggestions for solutions. Proceedings of 1st Trans-national Meeting on Salinity as Limiting Factor for Agricultural Productivity in the Mediterranean Basin, 24–25 March 1997, Leone and Steduto, Napoli, pp. 55–66. Papadopoulos I (1998) Use of saline waters for irrigation in Cyprus. Proceedings of 1st Trans-national Meeting on Salinity as Limiting Factor for Agricultural Productivity in the Mediterranean Basin, 24–25 March 1997, Leone and Steduto, Napoli, pp. 97–105. Postiglione L (1972) L’evapotraspirazione potenziale a Scafati nel periodo 1949–1971. Il Tabacco 742, 2–8. Postiglione L (1988) Esperienze di sistemazione nella collina meridionale. Sistemare la collina per difendere il suolo e tutelare l’ambiente. Assoc. Naz Bonifiche (ed.) Il Mulino, Bologna, pp. 281–285. Postiglione L (1991) Variazioni climatiche, vegetazione naturale e colture agrarie. Agricoltura e Ricerca 124, 2–10. Postiglione L, Basso F and De Franchi AF (1993) Soil conservation in the Mediterranean environment: results of twenty-two years of research in the hill areas of Southern Italy. Proceedings of the IV International Conference on Desert Development, Mexico City, 25–30 July 1993, pp. 338–394. Postiglione L, Barbieri G and Tedeschi A (1995) Long-term effects of irrigation with saline water on some characteristics of a clay loam soil. Riv di Agronomia 29(1), 24–30. Ragab R (1996) Constraints and applicability of irrigation scheduling under limited water resources, variable rainfall and saline conditions. In Irrigation Scheduling: From Theory and Practice. Proceedings of the ICID/FAO Workshop on Irrigation Scheduling, Rome, Italy, 12–13 September 1995, pp. 149–165.
Soil Salinization in the Mediterranean
173
Ramadan MF (1998) Drainage and salinity control for agricultural land in Egypt. Proceedings of 1st Transnational Meeting on Salinity as Limiting Factor for Agricultural Productivity in the Mediterranean Basin, 24–25 March 1997, Leone and Steduto, Napoli, pp. 13–22. Ruggiero C and Perniola M (1992) Sviluppo e distribuzione dell’apparato radicale della melanzana (Solanum melongena L.) sottoposta a tre diversi regimi irrigui. Irrigazione e Drenaggio 2, 31–34. Ruggiero C, De Pascale S and Barbieri G (1994) Effetti dell’irrigazione con acque a diverso contenuto salino sullo strato idrico, sull’accrescimento e sulla produzione della melanzana (Solanum melongena L.). Riv di Agronomia 28, 222–234. Schertz DL (1983) The basic for soil loss tolerances. Journal of Soil and Water Conservation 38, 10–14. Shalhevet J (1994) Using water of marginal quality for crop production: major issues. Agricultural Water Management 25, 233–269. Sifola MI, De Pascale S and Romano R (1995) Analysis of some quality parameters of yield in eggplant grown under irrigation with saline water. Proceedings of the 1st International Symposium on Solanacea for Fresh Market, Malaga, Spain, 28–31 March 1995, p. 67. S¨onmez B (1997) Experiences in the management of salt affected soils in Turkey. International Conference on Water management, salinity and pollution control towards sustainable irrigation in the Mediterranean region, Vol IV. Istituto Agronomico Mediterraneo, Valenzano, Bari, Italy, 22–26 September, 37–49. Szabolcs I (1994) Soils and salinizations. In M Pessarakli (ed.) Handbook of Plant and Crop Stresses. Marcel Dekker, New York, pp. 3–11. Tedeschi A, Hamminga W, Postiglione L and Menenti M (1996) Sustainable irrigation scheduling: effects of saline water on soil physical properties. In Irrigation Scheduling: from Theory and Practice. Proceedings of the ICID/FAO Workshop on Irrigation Scheduling, Rome, 12–13 September 1995, pp. 195–204. Tedeschi P, d’Andria R, Lavini A, Giorio P, Sorrentino G and Romano G (1997) Grain yield, oil, leaf water status and photosynthesis as affected by saline irrigation water in a field-grown sunflower crop (Helianthus annuus L.). Proceedings of conference on water management salinity and pollution control toward sustainable irrigation in the Mediterranean region, Vol. IV, 22–26 September 1997, Valenzano, Bari, pp. 69–83. Varallyay G (1998) Salinity/alkalinity as limiting factors of soil productivity in the Carpathian basin. Proceedings of 1st Trans-national Meeting on Salinity as Limiting Factor for Agricultural Productivity in the Mediterranean Basin, 24–25 March 1997, Leone and Steduto, Napoli, pp. 23–41. Zalidis G (1998) The impact of water and soil resources management on salt budget in Greece. Proceedings of 1st Trans-national Meeting on Salinity as Limiting Factor for Agricultural Productivity in the Mediterranean Basin, 24–25 March 1997, Leone and Steduto, Napoli, pp. 87–96.
Section V
Tools for Exploring Desertification
14
Environmentally Sensitive Areas in the MEDALUS Target Area Study Sites
A.C. IMESON AND L.H. CAMMERAAT
IBED-Fysische Geografie en Bodemkunde, Universiteit van Amsterdam, The Netherlands
1 INTRODUCTION The MEDALUS target area studies were selected so as to be located in areas that were extremely sensitive to desertification. In the recent past they had all undergone extensive changes in land use and at the same time contained areas that had become severely degraded. All of the target area studies shared similar objectives. It was expected that they would lead to a better understanding of the sensitivity of different environments to desertification while also contributing towards a methodology for identifying desertification sensitive areas. It was reasoned that it should be possible to interpret information about the distribution of desertification-affected areas within the target areas, in terms of the factors and processes that were responsible. The objective of this chapter is to explain how the authors, in the MEDALUS target area studies in which they worked in Portugal, Spain and Greece, approached this challenge. 1.1 Environmentally Sensitive Areas
The term “environmentally sensitive area” has a long history and many connotations. The EU and many of the constituent countries have implemented policies aimed at protecting environmentally sensitive areas (ESAs). These policies were usually developed to save endangered and vulnerable ecosystems perceived as being threatened by environmental change. The policy was applied to physically and functionally distinct land units or biotopes, such as riverine meadows or freshwater coastal lagoons, and the main concern was the loss of biodiversity and habitat. Generally, ESAs were small areas that provided critical environmental functions within the larger landscape. Protecting areas performing vital ecological functions is considered essential for maintaining ecosystem integrity and resilience. Sensitivity is present at two levels: there is the sensitivity of the land unit itself and the sensitivity at the level of the landscape in which it is found, should the unit be compromised. This application of the sensitivity concept is based on ecological considerations. Burke and Thornes (1998) pointed out that, particularly in Mediterranean Europe, the notion of environmentally sensitive areas is also appropriate to desertification. In this case the main concerns are land degradation and water availability. It concerns areas that are specifically sensitive to erosion, land degradation, and to declines in soil and ecosystem quality. Both the biological and desertification viewpoints emphasize the importance of resilience. From the ecological point of view the contribution of small, particularly sensitive areas to landscape-scale resilience is usually emphasized. From the land degradation perspective the resilience of the site is important in relation to the time needed (after a disturbance) for the soil and vegetation to recover its protective role against the forces that are causing degeneration or erosion. From the above it might be concluded that an area can be diagnosed as sensitive, first, if it has an inherently low resilience; second, if it is performing threatened critical functions that are needed Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
178
Mediterranean Desertification
at a coarser scale; and third, if the degradation may lead to off-site problems that impair the normal functioning of other areas.
1.2
Sensitivity to Land Degradation
Sensitivity can be quantitatively evaluated from the relative rates of “loss”, “recovery” and “persistence”. In the case of the loss of soil by erosion, this requires information about the rate of soil erosion (E), the resilience of the soil (R), and the depth or volume of erodible soil (D), all of which can be theoretically measured and quantified at specified temporal and spatial scales. Sensitivity to erosion can be calculated as follows: S = (E − R)/D
(1)
where E and R are in rates of erosion (m year−1 ) and D is the soil depth (m). Similarly, the sensitivity of the soil water retention capacity, following a change in land use, can be quantified if equivalent information is available. In both cases, resilience reflects the relative rates of degenerative and regenerative processes. The effects of this “competition” between, for example, erosion and vegetation development are easy to measure but the processes themselves may reflect highly complex dynamics. The usefulness of this approach to sensitivity depends upon the objectives. At the landscape scale the top-down influences of climate, parent material and land use are so dominant that regional differences in sensitivity can be obtained from maps, remote sensing images, climatological data and readily available information. Taking erosion as an example, according to the above criteria, limestone areas are sensitive to erosion because of the thin soils and slow rates of weathering. Marl areas are sensitive because of the high rates of erosion. When degraded, the resilience of both lithologies is high. On marls, when the soil is shaded by plants, stable soil aggregates develop rapidly to increase the water storage capacity of the soil. On limestone, there is frequently a very high availability of water in joints and fissures as the soil is protected by stones. At scales relevant for management and mitigation, there is usually too little information and too much spatial and temporal variability in data to go much further in this direction.
2 2.1
COMPARING THE SENSITIVITIES OF DIFFERENT TARGET AREAS Introduction
An initial research task was to develop the conceptual and practical tools for identifying critical areas, establishing threshold conditions and making comparisons between areas. The different disciplinary backgrounds and experiences of the various MEDALUS research groups meant that a variety of concepts and procedures were used to identify ESAs in the target areas (in Kosmas et al. 1999). Gradually, however, a common methodology evolved, which, as a first step, required establishing the distribution and pattern of desertification-affected areas. This pattern was deduced from studies of key indicators of soil, climate and management qualities to which critical values could be given. The approach was both simple and flexible in order to cope with different physical and socio-economic conditions and with different levels of data availability, and knowledge. Large geographical differences between the target areas, as well as the sensitivity of desertification to initial conditions, did not legitimize the simple transfer of indicators between target areas. Nevertheless the indicators can be used and combined with weightings to classify areas as being either critical, potential, or at no risk (Kosmas et al. 1999). Furthermore, in many cases the target areas contain areas that are superficially similar. The photographs from the Guadalent´ın (Figure 14.1), the Alentejo (Figure 14.2) and Lesvos (Figure 14.3) are characterized by large areas that have vegetation colonizing abandoned land, and by areas of silty dispersive soils (Figure 14.4) that are
Environmentally Sensitive Areas in the MEDALUS Target Area
179
Figure 14.1 This area is characteristic of the environmentally sensitive area in Spain. Note the mosaic of land use. Only 40% of the area has the same land use as in 1956. Erosion is extensive, especially on the marl soils
Figure 14.2 Area of abandoned agricultural land near Mertola, Portugal. The thin soils were abandoned about 12 years ago and the grazed area is being recolonized by Cistus. Adjacent ungrazed areas have complete vegetation cover. The south-facing slopes are especially sensitive to erosion. Runoff from these areas of thin soils can cause gully erosion downslope
highly erodible. In all areas there has been a response of the soil and vegetation to very heavy grazing pressure. To be relevant for management, the target area studies had a focal scale ranging from 1:20 000 to 1:200 000. If the general trends occurring within the target areas are considered at the scale of the regional investigations considered elsewhere (Kirkby 1999), there is a useful overlap which allows the target area studies to be embedded in the regional ones. Furthermore, the differences
180
Mediterranean Desertification
Figure 14.3 Area of thin soils in central Lesvos. Although the western part of the island is extremely desertified, the central part of the island has soils and vegetation that are the result of centuries of heavy grazing pressure
Figure 14.4 With very poor behavioural characteristics, silty dispersive soils such as these are extremely sensitive to erosion. They represent some of the most environmentally sensitive soils that are to be found. With the addition of organic matter, they can be improved dramatically
occurring within the target areas, established from regional-scale indicators, enable global differences in sensitivity resulting from underlying factors of geology, geomorphology and climate to be used as a basis for stratifying the selection of the field sites, where studies are being made to underpin indicators. Conducting a regional-scale investigation can be considered to be the first top-down preliminary phase of an ESA analysis.
Environmentally Sensitive Areas in the MEDALUS Target Area
181
2.2 The Pressures Responsible for Desertification
Desertification is often described as being “driven” by a combination of physical and socio-economic “pressures”. For a full discussion of landscape responses to combined climatic and anthropogenic stresses in the Mediterranean, reference is made to Puigdefabregas and Mendizabel (1998). Each target area has had its own unique history of pressures, which in terms of desertification are often remarkably similar. So many areas have had to persist under desertification pressures for so long that they are in comparable degraded states. The location of an area in a high- or low-pressure regime is probably only observable with respect to impacts that occur during the initial phases of land degradation. This is because the rate of soil formation is relatively low. The correlation between current “pressure” and “impact” indicators derived from soil and vegetation data is for this reason often found to be weak. This makes the comparison between different areas even more tenuous. Nevertheless, many biophysical and socio-economic pressures are correlated with topography. For example, soil moisture limitations and conditions are related to aspect, altitude and slope position; most soil erosion pressures are related to slope angle, position and curvature; and gully erosion is correlated with topographic indices obtained from slope and drainage area measurements. Many direct human and socio-economic pressures are also correlated with the topography, through the effect that they have on access and the exploitability of resources. For a variety of different reasons, therefore, topographic information is a valuable indicator of both exposure and impact. With respect to present impacts in the target areas studied, it was found that most cases of observable degradation were linked to pressures resulting from either land-use practices or from land-use change. It is under such circumstances that sensitivity is greatly influenced by the soil and other system properties (Kosmas et al. 1999). 2.3 Functions Resisting Desertification in the Target Areas
In order to understand the processes that are responsible for sensitivity, it is interesting to consider how and why the soil and land cover is resisting erosion or degradation. Which processes and mechanisms enable the soil–vegetation system to provide what could be called a successful “desertification control” function? The idea of a “soil erosion control” function for comparing seminatural and agricultural systems has been described by Imeson (1998, 2000). This was applied to the Alentejo target area and is presented in more detail below. How well a landscape unit (response unit, slope or landscape) is performing its “desertification control” function can be deduced from “sensitivity indicators”. It is useful to consider how both functions and pressures work during desertification. For example, should erosion reduce the production or ecological functions of a “response unit” (see below) and make it unfit for cereal production or nature conservation, then the losses of these functions can be exactly quantified (see Chapter 15). The losses can also be related to the critical values of indicators, such as soil depth, which can be used to establish threshold conditions (e.g. soil depth) and be related to sensitivity. A functional approach enables areas that are physically or socio-economically dissimilar to be compared. 2.4 Function Performance Indicators
An indicator is required to summarize specific aspects of the effects of complex processes and it should be easy to measure and relate to critical conditions. A major problem is that simple universal indicators are impossible to find because properties of the soil and vegetation, and processes that are dependent on chemical or physical threshold conditions (e.g. frost action or salt accumulation), usually limit the validity of an indicator to a specific geo-ecological domain. This is exemplified by soil depth. Soil depth has often been proposed as a good indicator of desertification (Reining 1984). Research by Kosmas et al. (1999) illustrates how the effects of soil depth are conditioned by lithology. Consequently the actual values that are critical are dependent upon the same factors that influence sensitivity as defined in equation (1).
182
Mediterranean Desertification
Soil depth itself can never be used as more than a global indicator of degradation (Reining 1984). This is because the real point of interest is whether or not the soil retains its productive capacity on a slope. In this context, the pattern of soil depth is more important than an average soil depth. The type of parent material, and the resilience of the disturbed soil must also be known to establish critical values of soil depth and pattern. Therefore, to calculate and compare the sensitivity of different areas, soil depth and pattern need to be evaluated with respect to their roles as indicators of a function in a clearly defined context. 2.5
The Effect of Desertification on Critical Soil Functions Sensitive to Desertification Desertification is a process bringing about changes in soil properties that can be measured in time. Examples of properties include the organic matter content, soil bulk density, the albedo and the electrical conductivity. When changes in such properties are monitored or mapped (Kosmas et al. 1999), it becomes apparent that soils are very variable in terms of their sensitivity to desertification drivers. Critical threshold values of indicators may be deduced from the damage to or disappearance of the functions that they are being used to monitor. As an illustration, functions will be considered at three scales: “fine scale”, referring to processes in the soil; “focal slope scale” and “coarse scale”, referring to catchment or landscape-scale processes.
The Water and Nutrient Regulation Functions of the Soil The loss of water and nutrient regulating functions can be caused by the deregulation of several processes. Water regulation functions require a soil medium that is able to store and retain water. This depends on the maintenance of the soil porosity and permeability. This is in turn favoured by a soil in which the primary particles are aggregated into larger water-stable agglomerations. The processes of agglomeration are often a consequence of biological activity and reflect the dynamics associated with the input and mineralization of organic matter. Since biological activity depends on favourable temperature and moisture conditions and regimes, as well as on chemical or toxic conditions, less favourable microclimatological conditions are soon felt. In practice, since other ecological and physio-chemical processes also come into play, cause and effect can often only be unravelled by detailed research. Nevertheless, in general the water-stable soil aggregation is indicative of the success or failure of biological activity in creating and maintaining the water and nutrient regulation function, which in turn favours retention of available soil moisture and good water and nutrient transmission characteristics. In turn, this depends on there being both a sufficient amount of food (suitable organic matter) and periods of time during which soil moisture and temperature do not critically limit biological activity. A reduction or diminution in soil aggregation will only lead to a loss of function in soils when it is a key process. When this is the case, soil aggregation indices can provide good sensitivity indicators. When it is not the case (e.g. in clay soils), a slightly different approach is appropriate. In more arid soils, there is a tendency for clay minerals to disperse. Clay dispersion is a key climate-sensitive process affecting soil functions, in particular the water regulation functions. Dispersive conditions are frequently found in soils that contain low amounts of salt but where a large proportion of this consists of sodium. In southern Europe, there is a threshold above or below which climatological conditions lead to the soil being either flocculated or dispersed. Where the rainfall is below about 400 mm year−1 , annual precipitation dispersion is often the main process regulating infiltration (Lavee et al. 1996). Above 600 mm year−1 , salts are flushed from the soil. Anywhere salt accumulates in the soil (e.g. at slope-foot positions or because of irrigation), sensitivity is increased by its effect on permeability and swelling. The areas of soils affected by dispersion vary both seasonally and from year to year according to the amount of rainfall. The water regulating function is also influenced by at least two other pressures. First, chemicals can poison the soil biota (e.g. dissolved organic compounds from eucalyptus, heavy metals from mines, and agro-chemicals), either through the effects they have on clay dispersion, or on the soil biota. They inhibit the biota from restoring the soil structure and quality. They therefore dramatically
Environmentally Sensitive Areas in the MEDALUS Target Area
183
reduce resilience. Second, mechanical pressure and compaction (e.g. tillage and grazing) also affect the soil’s water-regulating performance in complicated ways.
Soil and Water Conservation Function at the Slope Scale The soil on a slope is a store of minerals, water and nutrients. Together with the vegetation it prevents erosion from occurring. If the fine-scale processes that regulate the fine-scale functions described above are affected by desertification, erosion and runoff at the response unit or slope scale can be the result. This is reflected in three responses emergent at the slope scale: 1.
Changes in the distribution of sediment and runoff source and sink areas. There are changes in the size, behaviour and location of sinks and sources of water and sediments on slopes. 2. A general loss of soil depth at source sites and a potential increase at accumulation sites. 3. An increase in the connectivity of slope runoff processes during extreme events. Changes in soil depth on impermeable rocks sometimes result in positive feedback, whereby the more shallow soils become, the more likely erosion is to occur. On the other hand, the desertification regulation functions can also sometimes improve. For example, this can occur on abandoned land or during a succession of relatively wet years, if there is an increase in the water storage capacity of the soil–vegetation system. However, overgrazing and fire can also damage this function. It has also been demonstrated that one of the negative effects of forest fires is to reduce the amount of water that can be stored in the soil. Under semi-natural vegetation, the soil and water conservation functions of the soil depend upon both weathering and on organisms interacting with mineral material. This is why the weatherability of the parent material and the soil depth are important. On cultivated land, a “soil and water conservation capacity” is provided by tillage and by the construction of terraces. Additions of manure or chemical fertilizers can have a positive effect, promoting higher storage and stability. However, the effects can also be negative. The long-term effect of terracing is to accumulate soil and to create a potential time bomb should abandonment occur and maintenance no longer be forthcoming. Also on cultivated land, ploughing and tillage erosion affect the long-term degradation of the soil by means of gravitational effects. Increasing stoniness can also indicate the loss of fine soil particles that are harvested with root crops.
Catchment-scale Functions At the catchment scale the river channels and slopes may also store colluvium, alluvium and water. The performance of this function can be assessed by catchment-scale assessments of erosion that describe sinks and sources and the amount of erosion that is produced by extreme events. For example, near Mertola in the Alentejo target area, the channel incisions and ephemeral gullies resulting from heavy winter rainfall were used as an indicator of this catchment-scale sensitivity. The ephemeral gully incision summarized the failure of the catchment to perform its soil and water conservation function at this scale (Imeson 2000). When surface runoff contains dispersed clay, which is transported into rivers, this does not settle so the water remains turbid almost indefinitely. Sharma (1998) therefore proposed using the area covered and the turbidity of water as hydrological indicators of desertification in the Jodhpur District of India.
3 DESERTIFICATION RESPONSE UNITS AND SENSITIVITY To apply the ESA method requires data and information that can be related to geographically explicit areas. Field measurements required to obtain parameter values for equations or to underpin indicators are often point measurements that require extrapolation over larger areas. The tools available for this purpose include remote sensing, models and the desertification response unit (DRU) methodology. The advantage of the DRU method is that the response units can be identified from air photos;
184
Mediterranean Desertification
they provide convenient units for considering and upscaling functions and they can be used for identifying and investigating site (response unit)-specific indicators. If land-use or erosion models are available the DRUs also can be used for defining geographically explicit inputs. The application of this approach is illustrated in Chapter 15. 3.1
Desertification Response Units
The desertification response unit method was used to scale up from point measurements to large areas and to explain the contrasted response of different areas to desertification at different scales. Detailed descriptions are found in Imeson et al. (1996, 1998). It considers the changes that are taking place in the soil–water–vegetation patterns in the landscape system at different scales, resulting from the movement of water, sediment and nutrients. A response unit is defined as a slope or land unit with relatively homogeneous soil and vegetation characteristics that reflect the history, processes and characteristics of the site under the influence of climate and land use. The units are characterized and identified in such a way that a specific unit will give a typical response with respect to its hydrological behaviour, land degradation sensitivity or any other environmentally important aspect. By identifying the different response units for larger areas and by studying their spatial relationships, an integrated assessment with respect to environmental sensitivity for a catchment or region can be carried out. At the smallest sub-patch scale, relatively wet and moist micro-environments occur as the indirect result of many processes (Imeson and Cammeraat 2000; see Chapter 15). These include sorting, by rain-wash, the soil meso-fauna and the effects of the irregular distribution of roots. The irregularities in water and nutrient distributions associated with these differences are incorporated in the larger scale processes so that some units retain and others produce runoff and sediment. The spatial and temporal scales are of course different from those at the patch scale. The effects of the coarsescale processes feed back to and influence the dynamics of the finer scale units. These transfers and feedbacks produce patterns in the soil and vegetation that are characteristic of either specific landscape positions or disturbances. If there is degradation of the vegetation cover, subsurface expressions of the vegetation pattern remaining in the soil may enhance the ability of the vegetation to recover because they concentrate soil moisture above the critical levels required for sprouting, seed germination or root development. In other words, they exert a major influence on resilience and hence sensitivity. Different vegetation mosaics may be characteristic for different units, reflecting the internal redistribution of water, sediments and nutrients, but also the contributions or losses of water from adjacent units, when there is active spatial linkage between the units. External factors, such as changing rainfall regimes, grazing or land-use change will affect these units in different ways, depending on their internal stability and resilience and their spatial connections.
4
CONCLUSIONS
The authors of this and subsequent chapters have had the advantage of being able to learn from comparisons made in the different target areas. As mentioned in the introduction, desertification is from an ecological and economic perspective experienced as the loss of critical ecological and economic functions that when present enable the system to function in a way that is experienced as non-desertified. On the one hand it is evident that the experience of desertification both by scientists and by the target area is highly subjective, reflecting values, historical myths and psychological attitudes. On the other hand, the impact of non-sustainable land use is apparent in all of the target areas. From the target area comparisons, the following three conclusions are put forward: 1.
Desertification has left large areas in a degraded state and these have lost viable productive functions. Recognizing desertification control “functions” for these areas would be useful.
Environmentally Sensitive Areas in the MEDALUS Target Area
185
2.
Even the most resilient and intrinsically insensitive area can become desertified when policies ignore the desertification pressures they generate. 3. Even the most intrinsically sensitive areas have a high resilience and can recover lost functions under good management.
ACKNOWLEDGEMENTS This work was supported by the MEDALUS II and MEDALUS III (Mediterranean, Desertification and Land Use) research projects, under the EU Environment and Climate Research Programme, with contract numbers EV5V-CT92-0128 and ENV4-CT95-0119. Part of the work was also granted by the Netherlands Geosciences Foundation (GOA) with financial aid from the Netherlands Organization for Scientific Research (NWO) within the framework of the ‘Hierarchy in Land Degradation Processes in a Mediterranean Environment’ programme, under contract number 1003.750.294.03. The authors gratefully acknowledge this support. Jeroen Schoorl, Anja de Wit, Gonzalo Gonz´alez-Barber´a and Hein Prinsen are thanked for their help in the field.
REFERENCES Brandt CJ and Thornes JB (1996) Mediterranean Desertification and Land Use. John Wiley, Chichester. Burke S and Thornes JB (1998) Actions taken by national governmental and non-governmental organisations to mitigate desertification in the Mediterranean. Office for Official Publications of the European Communities, Luxembourg. Imeson AC (1998) Functional indicators for evaluating the effect of desertification on soils. Mediterraneo 12/13, 19–41. Imeson AC (2000) Indicators of land degradation in the Mediterranean Basin. In G Enne, Ch Zanolla and D Peter (eds) Desertification in Europe: Mitigation Strategies, Land use Planning. EUR 19 390 pp. 47–58. Imeson AC and Cammeraat LH (2000) Scaling up from field measurements to large areas using the Desertification Response Unit and Indicator Approaches. In O Arnalds and S Archer (eds) Rangeland Desertification. Advances in Vegetation Science 19, Kluwer Academic, Dordrecht, pp. 99–114. Imeson AC, Perez-Trejo F and Cammeraat LH (1996) The response of landscape units to desertification. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 447–469. Imeson AC, Cammeraat LH and Prinsen H (1998) A conceptual approach for evaluating the storage and release of contaminants derived from process based land degradation studies: an example from the Guadalent´ın basin, Southeast Spain. Agriculture, Ecosystems and Environment 67, 223–237. Kirkby MJ (1999) Coordinators’ summary report. In: Mediterranean Desertification and Land Use MEDALUS III (1996–1999) ENV4-CT95-0121, EU, Luxembourg, pp. 1–12. Kosmas C, Kirkby M and Geeson N (eds) (1999) The MEDALUS Project, Mediterranean Desertification and Land Use: Manual on Key Indicators of Desertification and Mapping Environmentally Sensitive Areas to Desertification. Office for Official Publications of the European Communities, Luxembourg. Lavee H, Pariente S and Imeson AC (1996) Aggregate stability dynamics as affected by soil temperature and moisture regimes. Geografiske Annaler 78A, 73–82. Puigdefabregas J and Mendizabel T (1998) Perspectives on desertification: the western Mediterranean. Journal of Arid Environments 39, 209–224. Sharma KD (1998) The hydrological indicators of desertification. Journal of Arid Environments 39, 121–132. Thornes JB (ed.) (1990) Vegetation and Erosion. John Wiley, Chichester.
15
Investigation on Environmental Characteristics to Underpin the Selection of Desertification Indicators in the Guadalent´ın Basin
L.H. CAMMERAAT,1 A.C. IMESON1 AND L. HEIN2 1 2
IBED-Fysische Geografie en Bodemkunde, Amsterdam, The Netherlands FSD, Wageningen, The Netherlands
1 INTRODUCTION This study describes the application of desertification indicators for a test area in semi-arid environments in Spain. Indicators are used to assess the degradation vulnerability for this area and the results are combined with the socio-economic function analysis of the area. Studies to prevent, assess and mitigate desertification are necessary as changes in land use due to human influence and changes or variations in climate and their possible adverse influence on the environment are perceived as major sources of problems within the Mediterranean Basin (Jeftic et al. 1993; Brandt and Thornes 1996; Burke and Thornes 1998). In this study an attempt is made to use key indicators, reflected in different properties of the soil and vegetation, to define the degradation vulnerability for the test area based on the hypothesis that finer scale processes or properties (e.g. soil aggregation, crusting and vegetation development) may influence processes at a broader scale (e.g. vegetation banding, adsorption of water in soil) and that these properties, patterns and changes can be detected and categorized for homogeneous landscape units.
2 KEY INDICATORS FOR (SEMI-)ARID REGIONS An indicator is required to summarize specific aspects of the effects of complex processes and also to be easy to measure and relate to critical conditions. A major problem is that simple universal indicators are impossible to find because properties of the soil and vegetation, and processes that are dependent on chemical or physical threshold conditions (e.g. frost action or salt accumulation) usually limit the validity of an indicator to a specific geo-ecological domain. Function performance indicators are required that not only tell us how well an area is performing its function but also give early warning of a risk. 2.1 Physical and Biotic Indicators
Rapport (1989), Hunsaker and Carpenter (1990) and Mouat et al. (1992) defined key indicators as follows: “An environmental indicator is defined as an environmental attribute that, when measured, quantifies the magnitude of stress, habitat characteristic, degree of exposure to the stressor, or degree of ecological response to the exposure” to the stressor. Cammeraat and Imeson (1998) stated that a key indicator should also reflect linkages to other biotic and abiotic processes at both the same and Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
188
Mediterranean Desertification
higher hierarchical scale levels. In this way key indicators can be used for upscaling. Desertification indicators were developed and proposed in the 1980s (Reining 1984) and have since been adapted and improved (Hunsaker and Carpenter 1990; Enne et al. 1998; Kosmas et al. 1999). There is also a strong resemblance between concepts and indicators of desertification and concepts of soil health or quality (Doran and Jones 1996; Doran and Parkin 1996; Harris et al. 1996; Karlen et al. 1997). In fact, the indicators of soil quality summarized by Karlen et al. (1997) and the concepts they propose are all appropriate to desertification. For a full discussion of the use of ecological indicators and of the problems and potentials, see Landres (1992) and Harris et al. (1996). Ecological indicators are usually indicators of ecosystem structure or function. An important aspect of applying such indicators is the requirement of indicators for reference areas where the potential conditions can be studied. The conversion of point indicators to larger areas and the application of upscaling/downscaling techniques is an important issue. Shoshamy et al. (1995) and Hill et al. (1998) applied remote sensing imagery combined with detailed point indicators related to degradation and climate gradients in the Mediterranean Basin. Another approach is the use of indicators combined with the response unit methodology (Imeson et al. 1996; see Chapter 14) which aims at the definition of land units with defined characteristics and that are functioning in a coherent way (with respect to hydrology, land degradation or overgrazing). By defining and quantifying key indicators for each response unit, an alternative way of upscaling is applied. This approach will be extended in the following text. 2.2
Socio-economic Function Analysis and Indicators
In addition to the physical and biotic indicators, the socio-economic effects of land degradation should be assessed. These effects are estimated by assessing the socio-economic functions of the environment. Environmental sensitivity analysis is used to assess the chances of erosion leading to a loss of the socio-economic functions of the environment. Socio-economic function analysis involves the evaluation of the socio-economic value of the different functions of the environment, as described by Constanza et al. (1997) and De Groot (1992). In order to describe and quantify the functions, the land use of the area has to be mapped. Next, the functions of the different landscape units are assessed, and, where possible, an economic value is attached to these functions.
3
´ BASIN THE STUDY AREA OF THE GUADALENTIN
The analysis of sensitivity was carried out within the MEDALUS target area of the Guadalent´ın catchment in the Province of Murcia, south-east Spain (Figure 15.1). The sample area is approximately 100 km2 in size and has a relief of about 300 m. The geology is dominated by limestone and marls from the Upper Cretaceous, Eocene and Miocene with more recent calcareous Holocene, Pleistocene and Late Tertiary deposits in depressions (Mapa Geologico de Espa˜na 1977, 1981). The area is drained by the perennial Rio Luchena and intermittent Rio de Turilla and Barranco de la Casa, and by many other smaller ramblas. The area has a semi-arid climate with an average annual rainfall of only about 300 mm year−1 , falling mainly in the autumn and spring. The average minimum and maximum temperatures are 9.3 and 26.0 ◦ C respectively (Navarro-Herv´as 1991). An air photo interpretation for 1958 and 1995 showed that land-use changes in the region have been considerable. In the Ca˜nada de Cazorla, for example, only 30% of the area in 1995 has the same land use as in 1958 (Imeson et al. 1998). In 1958, the area consisted of small fields in the valleys, and areas that are now abandoned were then used for grazing. These areas, taken into cultivation during the 1960s when they were used for dry land farming, were the first to be abandoned or set aside in the 1980s and 1990s. Within the region, areas were selected for more detailed studies. These include the Ca˜nada de Cazorla, the Ca˜nada Hermosa and the Alquer´ıa area, collectively referred to below as the Alquer´ıa area. Typical sections across these areas can be seen in Figures 15.2 and 15.3. Digital aerial photographs taken in April 1997 illustrate the vegetation cover and reflectance of the different cover
189
Selection of Desertification Indicators
Es
pu
na
Sierra de Cambron
rra
de
La Paca
orre alvi lla
Valdeinfierno Reservoir
Sie
Zarcilla de Ramos 1
R. T
Aledo
ra d e
l Gi
gan
te
2
Sier Rio
ra er Si
Parroquia
V
Alama de Murcia
z ele Lorca
de
ia rc Te
R
Totana
ua io G
dale
ntin
N
Sierra de Torrecilla
eN
oga
lte
de
Puerto Lumbreras
rra
Spain
Sie
10
0
< 500 > 1000 altitude (metres)
Alm en
R. d
ara
Fieldsite
kilometres
Figure 15.1 Location map of the study area. 1 and 2 indicate the locations of Figures 15.2 and 15.3 respectively
730
SSW
3A/B
N
Altitude (m)
720 Limestone Marls Response unit boundary
710
1B 1C
700
1A
690 680
2
2
1B
1A 0
200
400
600
800
1000
1200
Distance (m)
˜ Figure 15.2 Cross-section of the Canada de Cazorla area, indicating the lithology and response unit boundaries
190
Mediterranean Desertification N 9
7
8
6
S
1A
2
5 4
750
Altitude (m)
a b
700
c 650
d
600 lim
m
ar l
300
m
um ps gy
ne
200
l+
to
100
ar
es
ne
0
ne
lim
500
m
to es
l+ to es
ne
lim
to
ly
es
ar
ar
m
lim
550
400
500
600
800
700
900
Distance (m)
Figure 15.3 Cross-section of the Alquer´ıa area, indicating the lithology and response unit boundaries: a, response unit number; b, response unit boundary; c, barrier for water with high threshold value; d, direction of surface water flow
1A 1B 1B
1C
2
1C
1B
3B 3A
3A 1B 1A
˜ Figure 15.4 High-resolution digital aerial photograph mosaic of the Canada de Cazorla plateau and its surrounding active pediment surfaces. A label indicates the individual response units. Broken lines are response unit boundaries. The position of the cross-section shown in Figure 15.2 is indicated (thick white line). The photo is orientated to the north and the distance over the lower side of the photograph is approximately 500 m
types, as shown for the Ca˜nada de Cazorla area in Figure 15.4. Detailed information on the test sites can be found in Cammeraat and Imeson (1998, 1999) and in Imeson and Cammeraat (2000).
4
´ REGION FUNCTIONS IN THE ALQUERIA
Environmental sensitivity to desertification involves the loss of soil and nutrients, but also changes in biodiversity or landscape fragmentation. Sensitivity is influenced by land use, which is on the one
191
Selection of Desertification Indicators
hand dominated by the properties and resources of the area and on the other hand by the driving socio-economic forces. Therefore an analysis is essential if we are to understand impacts on the environment, to assess gains or losses of natural and socio-economic functions, and how indicators can be used to monitor changes in this. An analysis of the goods and services provided by the environment (see e.g. De Groot 1992; Constanza et al. 1997) in the Alquer´ıa region is based on the following functions: agricultural production, animal husbandry, supply of water, hunting, nature conservation, and the supply of wood and other materials. 4.1 Agriculture
Wheat is the most important crop in the area, of which over 80% is grown without irrigation. The extent of this crop is decreasing as a result of increases in other crops. Almonds are an important crop and their area of cultivation is rapidly expanding in the area with stony soils. Olives are also a traditional rain-fed crop, but the area with irrigated olives is increasing. A fast expansion of irrigated crops such as melons, broccoli, peppers, lettuce and tomatoes can be found on the more marly soils. In Table 15.1, the total gross yields in tonnes per hectare and pesetas per hectare are presented for the major crops. 4.2 Animal Husbandry and Hunting
Sheep and goat production is an important economic activity in the Guadalent´ın Basin. Although the number of herds has greatly decreased over the last 20 years, the number of animals is thought to have remained constant (120 animals km−2 ). An increasing number of large pig farms are present in the research area, both bio-industry farms and free-range pig farms. This activity is not dependent on a particular environment and is not affected by land degradation. However, free-range pig farms are built on both former cultivated and cleared semi-natural fields. Soil degradation, both by wind and water erosion, is affecting these areas, as they are kept free of vegetation. Hunting is an important function of the area. Game includes wild boar, partridge and rabbits. Most of the study area is classified as hunting area, and a hunting licence costs about 1000 pts ha−1 year−1 . Because the forests provide more shelter and food for animals, the forest is expected to have a higher value for hunting than esparto grass (Stipa tenacissima). The valleys and ramblas are also important for hunting because the animals depend on them for drinking water. Table 15.1 Yields of the major crops in the Alquer´ıa region (1997)
Crop Wheat (not irrigated) Wheat (irrigated) Almonds (irrigated) Almonds (not irrigated) Olives (irrigated) Tomatoes (irrigated) Melons (irrigated) Peppers (irrigated)
Yield (t ha−1 year−1 )
Yield (pts kg−1 )
Yield (103 pts ha−1 year−1 )
EU subsidy (103 pts ha−1 year−1 )
0.8–1.5 3–5 2–4 0.7–0.9
30 30 225a 225a
24–45 90–150 450–900 158–203
5 5 40 40
2–4 30 20 10
80 25 30 75 (wet)
160–320 750 600 750
? – – –
a (Shell + nut). The yield expressed as pts ha−1 year−1 is a gross yield. It does not include labour costs and investments, e.g. in irrigation equipment. Subsidies are not included in the figures. (100 pts = 0.601 euros)
192 4.3
Mediterranean Desertification Water Supply
Water is the most important limiting factor both for agricultural activities and in the natural environment. Interviews with farmers showed that water is considered by far the most important limiting factor for agriculture in the area. Annual rainfall is highly variable, and in years with low rainfall, non-irrigated crops cannot be harvested. The irrigated crops have a high water demand (Table 15.2). For example, in the case of almonds, the water supplied by irrigation is approximately 400 mm year−1 . Drinking water for the nearby villages of Zarcilla de Ramos and La Parroquia/Fuensanta comes from karst springs in the nearby mountains. The various farms depend on natural private wells, which usually have an output of only a few litres of water per minute (e.g. Cortigo de Alquer´ıa: 3 dm3 min−1 baseflow at the end of dry season, but has permanently fallen dry since the autumn of 1998). For irrigation, water comes from irrigation canals, groundwater sources (deep drilled wells) and reservoirs such as the Embalse de Puentes. The area has a high diversity in flora, influenced by variations in altitude and humidity, and endangered animals are also present. The greatest diversity of plants can be found at locations with a somewhat wetter microclimate, e.g. in the ramblas, on concave slopes, or in areas that receive runoff from uphill. The economic value and the nature conservation value of the landscape units are presented in Table 15.3. The figures are average values for the units; in some cases a range is given.
Table 15.2 Water demand of irrigated crops
Crop
Water supplied by irrigation (m3 ha−1 year−1 )
Wheat Almonds Olives Melons Tomatoes
3500–4000 4000 4500 5500 7000
Source: Comarcal (1997).
Table 15.3 Socio-economic functions of the landscape units. The natural value is presented as a relative value
Landscape unit
Irrigated agriculture Non-irrigated agriculture Abandoned meadow Esparto Forest Reforested Valleys a
Huntingc Total gross Agriculturea Grazingb Value for nature (×1000 pts) (×1000 pts) (×1000 pts) economic value conservation (×1000 pts) (relative scale) 100–750 24–45
– 2–4
– –
– – – – –
2–4 2–4 – – 2–4
– 0.5–1 1–2 0.5–1 1–2
100–750 25–50 2–4 2.5–5 1–2 0.5–1 3–6
? ? ? + +++ + +++
Gross value excluding labour costs and investment costs (irrigation equipment, machinery, etc.). Excluding labour costs. c The distinction between the landscape units is based on estimation of the relative value of the units for foraging of the hunted animals. b
Selection of Desertification Indicators
193
5 SENSITIVITY TO SOIL DEGRADATION 5.1 Indicators and Response Units
Prior to the degradation assessment the area was subdivided into several response units. This was done by applying the methodology tested in two representative training sites (Ca˜nada de Cazorla and Alquer´ıa; Figures 15.2 and 15.3). In these training sites many indicators have been used to characterize the individual response units. For an extended discussion of the indicators applied to the training sites, see Imeson and Cammeraat (2000). Examples are worked out for the Alquer´ıa area, concentrating on the fine- and intermediate-scale indicators, and are given in Table 15.4. The next step was to evaluate these indicators for each of the response units. The stability and resilience of the response units was evaluated from the scores in Table 15.4. The results are shown in Table 15.5 where the scores for the three classes are summed. Indicators with a good score received a weight of 3, intermediate scores received a weight of 2 and poor scores received a weight of 1. The final scores were summed and converted to a relative scale of 1 to 100. They are shown in the last column of Table 15.5 and are also given a ranking number in descending order of vulnerability per response unit. This enables a characterization for larger areas, as the whole area of study can be characterized in response units. In Figure 15.5, the southern slope of the Alquer´ıa hill is visible, showing different response units, corresponding with the profile of Figure 15.3. From the top of the valley to the bottom, the units are 5 (bare limestone), 4 (narrow band of shrubs), 2 (marl slopes with esparto cover) and 1A (cultivated area). 5.2 Soil Erosion
Erosion measurements have been carried out at many different sites in the Mediterranean, including in the Guadalent´ın and neighbouring areas. It is very hard if not impossible to directly translate these literature data to the Guadalent´ın as the actual values depend very much on local differences in slope, soil and vegetation cover, erodibility and erosivity. Furthermore, many of these data come from bounded plot experiments which are especially limited in value for semi-natural areas (Romero D´ıaz et al. 1999) and which have been maintained for periods that were too short to cover the high temporal variation in Mediterranean precipitation. Also, the application of the Universal Soil Loss
Figure 15.5 Alquer´ıa hill showing different response units, and different typical indicator characteristics
194
Mediterranean Desertification
Table 15.4 Physical and biological indicators of ecosystem function and structure relevant for soil and water conservation applied to the response units (adapted and extended from Imeson and Cammeraat 2000)
Indicator
Good
Intermediate
Poor
Well-defined small flow lines and associated deposits Some displacement also of larger organic debris Few micro-terraces, stones moved
Numerous flow paths and associated deposits Extreme movement during each event
Physical indicator Flow paths
Litter
Little evidence of water movement from unit In place
Rainwash
No evidence
Crusting and sealing Exposure of tree and shrub roots Surface cover Rills Gullies
None or very limited
Soil conservation dams
Crusting obvious, reducing infiltration Some
Significant movement of large stones and exposure of roots Hard crusts strongly reducing infiltration Abundant exposure
>0.5 dams ha−1
Incomplete protection Occasionally present Few but not very active <0.5 dams ha−1
Little protection Very common Numerous active on 20% of length None
Good representation of life forms and number of species Good diversity of height size and distribution of plants and roots
One or two life forms; only 30% expected species Moderate diversity of plants and root sizes height and distribution
Most plants productive and alive Dense spotted or banded structure Mechanisms adequate for plant maintenance Very little
Signs of mortality, production declining Intermediate spotted structure Mechanism marginally adequate Obvious but recovery possible
Many ant nests and other soil fauna
Less than 5 nests per 2500 m2 , no other soil fauna
One life form class dominant; 50% expected species One or two life form classes dominant with poor distribution of species and roots Evident dead and dying plants, poor production Open spotted structure Mechanisms inadequate to support life forms Serious damage, plants unable to recover Less than 1 ant nest per 10 000 m2 , no other fauna observed
None Soil protected Uncommon None
Biological indicator Diversity
Plant diversity
Plant status
Spatial vegetation structure Nutrient cycle
Grazing pressure
Soil fauna
195
Selection of Desertification Indicators
Table 15.5 Number of indicators (nine physical and seven biological) with good, poor or intermediate status for the different response units at Alquer´ıa (adapted and extended from Imeson and Cammeraat 2000)
Response unit
1A. 1B. 1C. 2. 3A. 3B. 4. 5. 6. 7. 8. 9. 10. 11.
Colluvium + soil cons. Footslope cultivated Footslope, abandoned Sloping esparto mattoralb Plateau with calcrete cultivated Plateau with calcrete mattoral Slope, limestone shrubs Bare limestone slope Slope, pine forest Degraded forest slope Slope, degraded mattoral Slope, esparto mattoralc Irrigated agriculture Reforestation
Biological indicators
Physical indicators
Combined
Ranking
G
I
P
G
I
P
G
I
P
No.
(%)a
1 1 0 4
1 1 4 2
5 5 3 1
3 0 1 2
0 3 5 4
6 6 3 3
4 1 1 6
1 4 9 6
11 11 6 4
9 11 8 4
(71.9) (81.3) (65.6) (43.7)
2
1
4
3
3
3
5
4
7
7
(56.3)
2
3
2
3
3
3
5
6
5
5
(50.0)
5 0 4 0 0
2 6 3 3 3
0 1 0 4 4
5 2 7 0 1
3 5 1 1 2
1 2 1 7 6
10 2 11 0 1
5 11 4 5 5
1 3 1 11 10
2 6 1 12 10
(21.9) (53.1) (18.7) (84.4) (78.1)
3 2 1
2 1 1
2 4 5
3 2 0
5 5 2
1 2 7
6 4 1
7 6 3
3 6 12
3 8 12
(40.6) (65.6) (84.4)
Abbreviations: G, good; I, intermediate; P, poor. Percentage of max. possible score: 100 = maximum vulnerability; 0 = minimal vulnerability. b On marl. c On limestone. a
Equation (Wischmeier and Smith 1978) is questionable, as this equation is developed for other climate regimes, limited slope angles and arable land. From the literature it can be seen that the overall erosion rate in the Mediterranean appears to be low, with the exception of badlands (Benito et al. 1992; Kosmas et al. 1997; Romero D´ıaz et al. 1999). However, erosion shows a strong spatial heterogeneity. Local (and temporal) variability is very high and scale-dependent. The problem of strong spatial heterogeneity may be overcome by the approach to define relative homogeneous response units, with a distinctive response to erosion. The connectivity between these units and the different thresholds determining these links was found to be crucial in the area of study (Cammeraat 2002). During a 46.4 mm rainstorm, with 21.4 mm in one hour in September 1997, soil erosion from shrublands remained very low (<0.1 ton ha−1 ), whereas in the terraced valley bottoms erosion rates of up to 30 ton ha−1 occurred, due to the failure of terrace rims (Cammeraat 2002). Much of the soil eroded from semi-natural hillslopes is accumulating behind these terraces in the agricultural areas in the valley bottom and only during intense rainfall is material removed from these terraces. The rainfall intensity and duration related to this failure and degradation threshold is >20 mm h−1 for at least one hour (Cammeraat 2002). Erosion and degradation rates therefore show high spatial heterogeneities, related to soil and water conservation measures and to connectivity between different response units. This must be considered when characterizing an area with respect to degradation sensitivity. Figure 15.6 illustrates the El Fraile hill, showing many rills on the non-terraced footslopes originating from the esparto grass-covered upper slopes. These rills were formed after two days with more than 50 mm of rainfall (October 2000), with a return period of about seven years.
196
Mediterranean Desertification
Figure 15.6 El Fraile hill with rills originating from runoff generating slopes covered with esparto grass
The assessment of degradation is further based on field observations of erosion. These data are supplemented with data from Pulleman and Coppus (1996) for the erosion rates in the Ca˜nada de Cazorla, Hein (1997) for the total area, and K¨onigel (1997) for the Ca˜nada Hermosa and data from the Alquer´ıa catchment (Cammeraat 2002). Extrapolations of the field observations were made on the basis of substrate, slope angle (based on the elevation lines from the 1:25 000 topographic map) and vegetation cover (based on field observations and aerial photographs). 5.3
The Resilience to Erosion
The effects of erosion in marly areas are compensated for by weathering and colluviation. Weathering rates are difficult to measure and literature values are scarce. For limestone, weathering rates of 9–20 mm per 1000 years were found for a number of studies in humid temperate conditions (Ollier 1984). In semi-arid regions, weathering rates will be much lower (a factor of 5 is mentioned by Young (1972) in Ollier (1984)). This means that the weathering rates for limestones are low compared to the erosion rate for sites where erosion is apparent. For the marl substratum, weathering rates are probably higher than soil erosion rates, but this does not mean that erosion is not a problem. Downhill, the accumulation of soil material does take place during high-intensity rainfall events, and is temporarily stored in natural and artificial depressions (vegetation structures, soil and water conservation terraces and in the alluvial domain of the valley bottoms). Another indication of sediment transport and sedimentation is the silting up of barrier lakes such as the Embalse de Puentes, which became filled with sediments over a period of 30–40 years. Although there is also erosion of soil by the larger ephemeral rivers, the accumulation is larger and the resilience of the river valleys is high enough to counteract the erosion. Bull et al. (1999) stated that with an event with a recurrence interval of seven years (1997), only minor changes were found in the channel of the nearby Rio de Torrealvilla, but they suppose that this event was close to the threshold where changes could occur. 5.4
The Persistence of Erosion
The persistence of functions of the environmental system under current rates of erosion depends on the depth of the soil relative to the minimum depth necessary to sustain the socio-economic functions of the soil. The A horizon is the most relevant part of the soil for vegetation rooting and is
Selection of Desertification Indicators
197
taken as an indicator of the depth of the topsoil. Although some plants (e.g. esparto) are able to root in the underlying C horizon in cracks and faults, the depth of the A horizon is used as an indicator of the resistance of the environmental system to erosion. This can, however, show important spatial heterogeneities, especially in limestone areas, although findings of Boer et al. (1996) show that there is a broad-scale spatial pattern in soil depth related to lithology. It is concluded that for the agricultural landscape units, the persistence of functions is high. For the other landscape units, a large variability occurs within the units. It is assumed that for the largest parts of the non-agricultural site of the study area, the depth of the topsoil is limited and erosion of topsoil is not buffered by a thick topsoil. At these sites, the persistence of functions under soil erosion is low.
6 THE SOCIO-ECONOMIC RISK OF EROSION: AN APPLICATION OF INDICATORS 6.1 Introduction
After mapping the landscape units of the Alquer´ıa region, defining the response units, determining indicator values for each response unit to express degradation vulnerability, and carrying out a final analysis of its socio-economic functions, the socio-economic values of the different landscape units can be determined. The first ones have been described in section 4. In section 5, the vulnerability of the Alquer´ıa region to erosion was summarized and related to the indicators and response units. In this section these findings will be combined and evaluated, including vulnerability, risk and socio-economic values for the area. 6.2 The Vulnerability of the Alquer´ıa Region to Erosion
It will now be clear that by using literature studies with regard to erosion and weathering rates, no conclusions can be drawn about erosion vulnerability. However, by applying a set of easily recognizable indicators within different landscape units (response units), a classification can be made to express relative vulnerability to erosion. This alone is not sufficient as the spatial heterogeneity and spatial connection between landscape elements is crucial for the transport of water and sediment. Only by incorporating those thresholds that determine the scale level in the landscape where runoff and overland flow are generated, will it be possible to indicate the risk of erosion. Both the irrigated and the non-irrigated agricultural areas have a high vulnerability to erosion, but the risk of erosion is low, especially over short time-scales, and is dependent on the magnitude of the runoff-generating storm. Events reaching the threshold of continuous runoff generation in arable fields have a return period of approximately 5–7 years. The arable fields generally have a very low slope (< 5%). Rainfall infiltrates almost completely after cultivation and there is no overland flow on recently ploughed areas except for during low-frequency storms. In such low-frequency storms, water is also coming from surrounding semi-natural areas, feeding the lower valley bottoms. Cereal fields are normally cropped for one year, and left fallow the following year. Fallow fields are crusted and so have low infiltration rates, which increases the risk of erosion. Some irrigated agricultural fields are partly covered by plastic to reduce evapotranspiration, but during intense rainfall this leads to strong concentrated erosion of the non-covered parts of the fields and sediment accumulation on the plastic covered area (Figure 15.7). The vulnerability to erosion of the abandoned fields is also high. This is because they are usually located on areas with a steeper slope, ploughing no longer counteracts crust formation and vegetation remains sparse for decades after abandonment. Abandoned fields are often located between natural and cultivated fields, and transmit water and sediment towards the arable zone. They generate runoff even during low-magnitude storms and the risk of erosion is high. The vulnerability of the semi-natural areas dominated by esparto cover is dependent on the substrate. Vulnerability is higher on marls than on limestone. The highest sensitivities are found on marls with dispersive clays, or on pediment surfaces where the massive top of the calcrete is not present. Some badlands are related to active stream rejuvenation, where gully back-wall retreat occurs
198
Mediterranean Desertification
Figure 15.7 Sediment accumulation on irrigated and plastic-covered area, with serious pipe erosion under the non-covered areas on marls with dispersive clays
related to steep gradients caused by neo-tectonic activity and also by the presence of dispersive clays in the marls. The vulnerability and risk to erosion of the natural forest is very low. The substrate under the natural forest is mostly limestone. The erosion rate is controlled by a combination of a persistent substrate, high infiltration rates into joints and deep soil pockets, and a dense, sometimes banded vegetation cover. The vulnerability of the reforested areas is much higher than that of natural forests. The vegetation cover is low and the reforestation terraces are not maintained, which leads to the accumulation of overland flow and thus rilling and gullying. Occasionally, rills break through the terraces, causing high rates of erosion. Runoff is often generated and these areas are prone to erosion. The vulnerability of the valleys is high, but the risk to erosion is low for most of the time. During low-frequency–high-magnitude events, erosion and channel modification are important factors. Erosion is an ongoing process, but sedimentation prevents soil loss and causes a high resilience of the landscape unit. As erosion and sedimentation are dynamic processes, patches with net erosion and net sedimentation occur. However, the perpendicular walls of the valleys, eroded into the alluvial deposits in the west of the Alquer´ıa region, are highly sensitive to erosion. Erosion causes continuous retreat of the valley walls. 6.3
The Environmental Risk of Erosion
Table 15.6 shows the combined socio-economic values and vulnerability data for the Alquer´ıa region. It shows that vulnerability to erosion is low to medium for the large majority of the area, and that the areas with the highest socio-economic values are the least sensitive to erosion, in normal years. Abandoned fields and reforested areas show the highest sensitivity but have very little socio-economic value. For the natural areas the erosion risks are variable depending on the vegetation cover and degradation status. Although the current socio-economic risk to erosion of most of the Alquer´ıa region is low, erosion is still a factor that must be considered in the management of the area. The esparto and forest ecosystems generally have a thin A horizon (1–15 cm) and their persistence is low. This means that if the factors that control the erosion rate are disturbed (e.g. by cutting of forests, overgrazing) and erosion increases, the functions of the environmental system will be affected immediately. The
199
Selection of Desertification Indicators Table 15.6 The socio-economic risk of erosion of the landscape unit
Land use
Socio-economic Nature value conservation (103 pts ha−1 year−1 ) value
Unit 10: irrigated agriculture Unit 1A: non-irrigated agriculture Unit 1B: non-irrigated agriculture Unit 3A: non-irrigated agriculture Unit 1C: abandoned fields Unit 9 esparto/shrubs limestone Unit 2: esparto/shrubs marls Unit 3B: esparto/shrubs calcrete Unit 8: degraded esparto/shrubs Unit 4 + 6: natural healthy forest Unit 7: natural degraded-forest Unit 11: reforested Unit 12: valley a b
Vulnerability (according to indicators)
Risk of erosion
90–750a
–
High
(65.6)
Very lowb
25–50a
–
High
(71.9)
Very lowb
25–50a
–
High
(81.3)
High
25–50a
–
High
(56.3)
Medium
2–4 2.5–5
– +
High (65.6) Medium (40.6)
High Lowb
2.5–5
+
Medium (43.7)
Mediumb
2.5–5
+
Medium (50.0)
Medium
2.5–5
+
High
(78.1)
Mediumb
1–2
+++
Low
(20.3) Very low
1–2
+++
High
(84.4)
Medium to high
1–2 3–6
+ +++
High High
(84.4) (n.d.)
High Lowb
Gross yields, i.e. excluding investment and labour costs. Risk is (very) high in the case of low-frequency–high-magnitude event.
resilience of these ecosystems is also very low, which means that recovery of the environmental system from erosion is a very slow process.
ACKNOWLEDGEMENTS This work was supported by the MEDALUS II and MEDALUS III (Mediterranean, Desertification and Land Use) research projects, under the EU Environment and Climate Research Programme, with contract numbers EV5V-CT92-0128 and ENV4-CT95-0119. Part of the work was also granted by the Netherlands Geosciences Foundation (GOA) with financial aid from the Netherlands Organization for Scientific Research (NWO) within the framework of the ‘Hierarchy in Land Degradation Processes in a Mediterranean Environment’ programme, under contract number 1003.750.294.03. The authors gratefully acknowledge this support. Jeroen Schoorl, Anja de Wit, Gonzalo Gonz´alez-Barber´a and Hein Prinsen are thanked for their help in the field.
REFERENCES Benito G, Gutierrez M and Sancho C (1992) Erosion rates in Badland areas of the Central Ebro Basin (NE Spain). Catena 19, 269–286. Boer M, Del Barrio G and Puigdefabregas J (1996) Mapping soil depth classes in dry Mediterranean areas using terrain attributes derived from a digital elevation model. Geoderma 72, 99–118.
200
Mediterranean Desertification
Brandt CJ and Thornes JB (1996) Mediterranean Desertification and Land Use. John Wiley, Chichester. Bull LJ, Kirkby MJ, Shannon J and Hooke JM (1999) The impact of rainstorms on floods in ephemeral channels in southeast Spain. Catena 38, 191–209. Burke S and Thornes JB (1998) Actions taken by national governmental and non-governmental organisations to mitigate desertification in the Mediterranean. Report, Office for Official Publications of the European Communities, Luxembourg. Cammeraat LH (2002) Scale dependent thresholds in hydrological and erosion response of a semi-arid catchment. Agriculture, Ecosystems and Environment : submitted. Cammeraat LH and Imeson AC (1998) Deriving indicators of soil degradation from soil aggregation studies in SE Spain and S France. Geomorphology 23, 307–321. Cammeraat LH and Imeson AC (1999) The evolution and significance of soil–vegetation patterns following land abandonment and fire in Spain. Catena 37, 107–127. Comarcal (1997) Officina Comarcal Agricultre Alto Guadalent´ın, Lorca. Statistic regions Parroqu´ıa and Zarcilla de Ramos. Constanza R, d’Arge R, de Groot RS, Fraber S, Grasso M, Hannon B and Limburg K (1997) The value of the world’s ecosystem services and natural capital. Nature 387, 253–260. De Groot RS (1992) Functions of Nature. Wolters-Noordhoff, Groningen. Doran JW and Jones AJ (1996) Methods for Assessing Soil Quality. SSSA Special Publications, 49. Soil Science Society of America, Madison, Wisconsin. Doran JW and Parkin PW (1996) Quantitative indicators of soil quality: a minimum data set. In JW Doran and AJ Jones (eds) Methods for Assessing Soil Quality. Soil Science Society of America, Madison, WI, pp. 25–38. Enne G, D’Angelo M and Zanolla C (eds) (1998) Indicators for Assessing Desertification in the Mediterranean. Nucleo Recerca Desertificazione, Sassari. Harris RF, Karlen DL and Mulla DJ (1996) A conceptual framework for assessment and management of soil quality and health. In JW Doran and AJ Jones (eds) Methods for Assessing Soil Quality. Soil Science Society of America, Madison, Wisconsin, pp. 61–82. Hein L (1997) Socio-economic risk assessment of erosion in the Alquer´ıa region, South East Spain Report. Foundation for Sustainable Development, Wageningen. Hill J, Sch¨utt B and Jarmer T (1998) Mapping complex landscape systems along climatological gradients by optical remote sensing. In ERMES II, Final Report. FGBL, Amsterdam, pp. 227–263. Hunsaker CT and Carpenter DE (1990) Ecological indicators for the environmental monitoring and assessment programme. EPA 600/3-90/060, US Environmental Protection Agency, Office of Research and Development, Research Triangle Park, North Carolina. Imeson AC and Cammeraat LH (2000) Scaling up from field measurements to large areas using the Desertification Response Unit and Indicator approaches. In O Arnalds and S Archer (eds) Rangeland Desertification. Advances in Vegetation Science 19, Kluwer Academic, Dordrecht, pp. 99–114. Imeson AC, Perez-Trejo F and Cammeraat LH (1996) The response of landscape units to desertification. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 447–469. Imeson AC, Cammeraat LH and Prinsen H (1998) A conceptual approach for evaluating the storage and release of contaminants derived from process based land degradation studies: an example from the Guadalent´ın basin, Southeast Spain. Agriculture, Ecosystems and Environment 67, 223–237. Jeftic L, Milliman JD and Sestini G (eds) (1993) Climatic Change and the Mediterranean. Edward Arnold, London. Karlen DL, Mausbach MJ, Doran JW, Cline RG, Harris RF and Schuman GE (1997) Soil quality: a concept, definition, and framework for evaluation. Soil Science Society of America Journal , 61, 4–10. K¨onigel C (1997) Response unit mapping in a semi-arid environment. Thesis, University of Amsterdam. Kosmas C, Danalatos N, Cammeraat LH, Chabart M, Diamantopuolos J, Farand R, Gutierrez L, Jacob A, Marques H, Martinez-Fernandez J, Mizara A, Moustakas N, Nicolau JM, Oliveros C, Pinna G, Puddu R, Puigdefabregas J, Roxo M, Simoa A, Stamou G, Tomasi D, Usai D and Vacca A (1997) The effect of land use on soil erosion rates and land degradation under Mediterranean conditions. Catena 29, 45–59. Kosmas C, Kirkby M and Geeson N (eds) (1999) The MEDALUS Project, Mediterranean Desertification and Land Use: Manual on Key Indicators of Desertification and Mapping Environmentally Sensitive Areas to Desertification. Office for Official Publications of the European Communities, Luxembourg. Landres PB (1992) Ecological indicators: panacea or liability. In DH McKenzie, DE Hyatt and VJ McDonald (eds) Ecological Indicators, Vol I. Elsevier Applied Science, London, pp. 1295–1318. Mapa Geologico de Espa˜na (1977) 1:50 000, Sheet Velez-Blanco (no. 952). Servicio de Publicationes Ministerio de Industria, Madrid.
Selection of Desertification Indicators
201
Mapa Geologico de Espa˜na (1981) 1:50 000, Sheet Lorca (no. 953). Servicio de Publicationes Ministerio de Industria, Madrid. Mouat DA, Fox CA and Rose MR (1992) Ecological indicator strategy for monitoring arid ecosystems. In DH McKenzie, DE Hyatt and VJ McDonald (eds) Ecological Indicators, Vol I. Elsevier Applied Science, London, pp. 717–737. Navarro-Herv´as F (1991) El Sistema Hidrogr´afico del Guadalent´ın. Consejeria de Politica Territorial, Obras Publicas y Medio Ambiente, Murcia. Ollier C (1984) Weathering. Longman, Harlow. Pulleman M and Coppus R (1996) Inventarisatie van bodemerosie en erosiegevaar en veranderingen in landgebruik in de Ca˜nada de Cazorla, Z-O Spanje. Report, University of Amsterdam. Rapport DJ (1989) What constitutes ecosystem health. Perspectives in Biology and Medicine 33, 120–132. Reining P (1984) Handbook on Desertification Indicators. AAAS Publication 78-7. Romero D´ıaz A, Cammeraat LH, Vacca A and Kosmas C (1999) Soil erosion at three experimental sites in the Mediterranean. Earth Surface Processes and Landforms 24, 1243–1256. Shoshamy M, Lavee H and Kutiel P (1995) Seasonal vegetation cover change as indicators of soil types along a climatological gradient: a mutual study of environmental patterns and controls using remote sensing. International Journal of Remote Sensing 16, 2137–2151. Wischmeier WH and Smith DD (1978) Predicting Rainfall Erosion Losses. Agriculture Handbook 57, US Department of Agriculture, Washington, DC.
16
MEDRUSH: A Basin-scale Physically Based Model for Forecasting Runoff and Sediment Yield
M.J. KIRKBY,1 R.J. ABRAHART,2 J.C. BATHURST,3 C.G. KILSBY,3 M.L. MCMAHON,4 C.P. OSBORNE,5 J.B. THORNES6 AND F.I. WOODWARD5 1
School of Geography, University of Leeds, UK School of Geography, University of Nottingham, UK 3 Water Resource Systems Research Laboratory, School of Civil Engineering and Geosciences, University of Newcastle upon Tyne, UK 4 Infocom (UK) Ltd, York Science Park, York, UK 5 Department of Animal and Plant Sciences, University of Sheffield, UK 6 Department of Geography, King’s College London, UK 2
1 INTRODUCTION: CONCEPTUAL DEVELOPMENT OF THE MEDRUSH MODEL A major objective within the MEDALUS project has been the creation, development and testing of MEDRUSH, a substantially new model that is designed to forecast runoff and sediment yield by water erosion, for catchments of up to 2000 km2 , over periods relevant to global change (c. 100 years). This model was coded in C++ and integrated within the GRASS (Geographic Resources Analysis Support System) GIS, for application within desktop work-station environments. The finished model has been applied primarily to the MEDALUS Agri target catchment, in the Basilicata region of southern Italy, but there is still insufficient validation data at the catchment scale, and confidence in the model is largely based on the behaviour of individual components. The spatial and temporal scales have been chosen to address the needs of regional planning, both in dealing with an area large enough to be of relevance to planners, and over a period in which climate and land-use change is expected to be both rapid and cumulative. The limitation to periods of 100 years, however, means that changes in landscape morphology are likely to be negligible, even where there is substantial soil loss. Thus an exceptionally high average annual soil loss of 5 mm will lead to a net lowering of only 0.5 m over a century, which may severely damage soil productivity, but is very minor in terms of the landscape as a whole. Nevertheless, the change of conditions involved in a 100-year scenario may produce profound changes in both vegetation and soil properties, and it is essential that the period of interest is simulated continuously to demonstrate the accumulation of changes over time. Examples of the types of interaction that are thought to be significant over a 100-year time-span, and that are not included in current soil erosion models, include the following: 1. 2.
Changes in vegetation cover and type influence the formation and decay of organic matter in the soil, which in turn influence soil hydrological parameters. Surface erosion increases the microtopography of the soil surface, modifying the patterns and intensity of soil erosion.
Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
204 3.
Mediterranean Desertification
Erosion and deposition modify surface stone size through the development of an armour layer, and ultimately change the stoniness of the whole soil profile, influencing moisture retention and subsequent erosion rates.
The model has been developed for the Mediterranean region, and is therefore biased towards representing semi-arid conditions, under which most overland flow is generated by rainfall intensities that exceed the infiltration capacity (Hortonian overland flow), although some allowance is made for subsurface saturation and return flow. The perceived widespread vulnerability of abandoned land to soil erosion, and the widespread use of tree crops in Mediterranean steeplands, has also led to a focus on uncultivated and non-arable land in the model. This provides a major point of departure in comparison with most existing erosion models, which focus on arable land. Arable crop routines have now been added to MEDRUSH, but are not described here. MEDRUSH addresses the scale issue by an explicit nesting strategy, which can be broken down into stages: 1. Subdivision of the catchment into up to 200 sub-basins, each of 5–50 km2 . Some sub-basins represent headwater areas, and others straddle larger channels. 2. Identification from the DEM (Digital Elevation Model) of further permanent channels within each sub-basin, where possible, to provide elementary DEM catchment units. 3. Breaking elementary DEM catchment units into a spatial family of representative flow strips, each forming a path from the local divide to the elementary DEM catchment unit outlet. 4. Simplified catena modelling for a single representative flow strip, using methods compatible with the MEDALUS I catena model (Kirkby et al. 1993, 1996). 5. Full coupling within the catena of a vegetation growth model, which distinguishes major functional types. 6. Transfer of changes forecast for the representative flow strip to all other corresponding parts of the whole sub-basin – for vegetation growth and soil loss. 7. Providing water and sediment yields by linearized routing through the network at the wholecatchment level. This provides linkages to flood-plain changes and groundwater storage. The primary areas where new physical model concepts have been generated are the representation of medium-term (decadal) interactions, through a dynamic formulation of surface microtopography, surface armouring, and changes in soil properties due to erosion or deposition. These processes themselves interact with the vegetation growth model, primarily through transpiration and the water balance, but also through changes in erodibility. Wherever possible, integration of analytical functions over frequency distributions has been used to provide an explicit scaling up from scales of relevance to the scales represented within the model. Scales of relevance may, for example, be very short intervals of intense rainfall, or the local distribution of microtopography. Explicit integration is used to scale up these effects to the hourly intervals of rainfall data and the computational time step; and to spatial steps (5–20 m) along the flow strips. This allows the explicit inclusion and parametrization of relevant fine-scale processes within the coarser time and space increments needed to allow rapid computation. After considering alternatives based on regular grid squares, sub-basin units were adopted as the basis for partitioning the catchment, each sub-basin being connected to the network of main channels. The high specific water and sediment yields from upland areas require a greater density of sub-basins in steep headwater areas. For a 2000 km2 catchment, sub-basin sizes therefore range from approximately 5 km2 in upland areas, increasing to 50 km2 at the downstream end. This gradation takes account of the fact that upstream areas are generally of greater importance in connection with the generation of floods and large erosional events. Thus the smaller sub-basins in steep headwater regions will be modelled with a higher density of sub-basins, to reflect their greater importance in the overall sediment budget. Over the time-spans of interest, it is assumed that there are no changes in the network linking the sub-basins, and that this network is represented by permanent
205
MEDRUSH: A Basin-Scale Physically Based Model GRASS GIS Digital contour data
Vegetation and parameter field measurement
DEM
Remote sensing images and products
Thematic maps
Output Visualization of data and forecasts
Neural net or other classification scheme Updates
Sub-basin definition
Vegetation and parameter distribution
Flow strip topography
Distributed hydrology and sediment yield
Rates of change Flow Strip Models Vegetation
Atmosphere
Surface
SubBasins Integration to areal frequency distributions
Catchment Routing of water and sediment
Soil
Figure 16.1
Conceptual structure of the MEDRUSH model for land degradation
channels (though not necessarily by perennial flow). Figure 16.1 gives an overview of the proposed MEDRUSH model structure.
2 THE SUB-BASIN COMPONENT 2.1 Sub-basin Response
Each sub-basin is envisaged as consisting of a defined family of flow-strip profiles. Within the subbasin, the spatial set of profile and flow strip geometries can therefore be represented in the model by a single representative flow strip, containing the range of possibilities, from hollows through straight slopes to convex spur-heads, and classified by major land-use and/or lithological units. We know that hollows, although they occupy only a small percentage of the area, are disproportionately important in their hydrological and sediment contribution, whereas spurs tend to be much less significant, and the representative flow strip is chosen to be similarly weighted. At the scales of interest, it is clear that sub-basins contain many channels. Some of these are permanent while others may be dynamically extending or infilling in response to changes in climate or land use/vegetation. Channel flow generated within the sub-basin can be characterized in a number of ways, but for a coarse-scale model, we use linear routing algorithms taking a Geomorphological Unit Hydrograph (GUH) approach. Each sub-basin is represented by a flow strip which includes a permanent channel segment within it, together with a time-varying component of dynamic channel extension. The model is designed to allow this channel extension to generate appropriate discharges and sediment yields, so that it is able to represent the behaviour of both unchannelled hillslopes and headwater channels.
206
Mediterranean Desertification
Hillslope profile flow can perhaps best be represented using a modified version of TOPMODEL (Beven and Kirkby 1979), which relates lateral subsurface flow to net rainfalls and local topography/soils. This establishes the current distribution of saturation deficits, based on the distribution of the topographic wetness index ratio a/(tan β) (i.e. area drained per unit width ÷ local gradient). This is established from seasonal net rainfalls, and storm rainfalls then top up this deficit to produce patterns of saturation overland flow. In addition, a fixed near-surface store related to current vegetation and organic soil is used to represent infiltration (Hortonian) overland flow. Leakage (i.e. infiltration) occurs from this store at a rate that increases with the saturation deficit defined by TOPMODEL. Overland flow thus occurs in two ways: •
•
Hortonian overland flow (HOF) occurs as soon as the surface store is filled. A proportion of subsequent storm rainfall provides the HOF, while the infiltration goes to reduce the saturation deficit. Even when saturation deficits are very high, not all the rainfall infiltrates, so that there can always be some HOF. The pattern of saturation deficits defines the proportion of the catchment yielding saturation overland flow (SOF). Once the saturation deficit is less than the surface store depth, there is no more infiltration.
The HOF component is likely to be dominant in semi-arid conditions, and the SOF component in more humid conditions. For generality both have been retained. Overland flow and gradient drive sediment transport, with an erosion threshold related to soil properties, turf cover and possibly armouring, although it may be hard to estimate this directly at the coarse resolution available. The downslope build-up of overland flow is influenced by the duration of storms, and therefore the lengths of slope over which overland slope generally accumulates. In general terms, transport increases smoothly with discharge, without explicit distinction between rainsplash, rainflow and rill wash, although they are implicit in the effectively rising exponent of discharge, following a law such as S ≈ a + bq + c(q − qT )2 where qT is the local erosion threshold. Perhaps the best current sediment formulation distinguishes travel distance and detachment processes, allowing effective selective grain-size sorting down the length of a flow strip. Detachment is by raindrop impact (rainsplash and rainflow) or by flow traction (rillwash); travel is aerial (rainsplash) or in the flow (rainflow and rillwash) (Kirkby 1992). Criteria for determining whether channels are enlarging or infilling for single storms have been discussed elsewhere (Kirkby 1994). Summed over a series of storms, the increase in sediment transport provides a Smith and Bretherton (1972) erosion stability criterion, which establishes the dynamically changing drainage density. For Mediterranean areas with sparse vegetation, we expect drainage density to respond dynamically in storms with a return period of a few years. Under these conditions, soil erosion is generally the major contributor of sediment to the channel network, with significant removal from many hillslope areas. For the more densely vegetated areas, the total contribution of soil erosion is less important except perhaps in storms with a return period of more than 100 years, which have a significant long-term impact on the landscape and on sediment supply through the formation of new gully systems, thereby providing a persistent feature of the landscape. By running the same storm characteristics through a series of realizations, we can build up a distribution of outcomes, which helps to define the reliability of the forecasts. The Smith and Bretherton criterion can also be applied on a storm-by-storm basis, giving rill extension, etc., in major events, with infilling in between, and these criteria can be applied to the single profile, even though it cannot represent the detailed rill–gully morphology at our scale of interest. 2.2
Distribution of Overland and Subsurface Flow Routing At each point down the length of the flow-strip catena, infiltration rate is estimated from current saturated deficit, unsaturated storage capacity and unsaturated storage. For each potential delivery
MEDRUSH: A Basin-Scale Physically Based Model
207
point downslope, the routing travel time is computed to give the proportion of Hortonian overland flow traversing the intervening distance. At long distances and long travel times, mean intensity rarely exceeds infiltration capacity. Long-distance overland flow runoff is therefore small except where mean intensity exceeds the infiltration capacity. Brief bursts of intense rain, however, quite commonly exceed the infiltration capacity and generate surface runoff. Overland flow discharge is therefore able to accumulate, but only over the short distances the flow travels overland during the intense burst. A similar computation is made for the integrated sum of the overland flows squared, which is used to estimate sediment transport. The integrated sum is estimated from a fractal distribution of short-term intensities and their durations. The variations in intensity must also be integrated across the microtopographic roughness elements which concentrate flow and erosion into rills and proto-rills. The proposed approach is to examine, separately and then together, first the distribution of flow and sediment discharge across microtopographic roughness elements, and second the time distribution of flows, and the way in which this varies downslope. The following assumptions have been demonstrated to provide a reasonable approximation to flow on a more general rough surface: 1. The routing velocity for overland flow is constant (c). Note that this is not the same as assuming that overland flow velocity is constant. 2. Sediment transporting capacity per unit width is proportional to discharge per unit width squared. Lateral effects, as from rill walls, are neglected. There is scope to disaggregate transport into detachment and travel distance, giving the grain-size dependence needed for armouring, but this is not illustrated here. 3. The standard deviation in mean rainfall intensity for a period t is proportional to a negative power α of t. If rainfall intensity values were not autocorrelated (i.e. into storms), then this exponent would be 0.5. We propose that the exponent is about α = 0.3, allowing a Hurst persistence of 0.2, and that the standard deviation for a period of one day is approximately equal to the rainfall total, since daily rainfalls fit fairly closely to an exponential distribution, which has this property. 4. Microtopography is roughly normally distributed, following distributions observed in the field, and is characterized by a standard deviation height of h0 . For a flow over the microtopography, stage h may be defined relative to the mean elevation. Points on the surface at elevation z occur with probability density 2 1 −z p(z) = √ exp (1) 2h20 h0 2π where h0 is the standard deviation of microtopography. Using this probability density as a weighting, we have, for flow at stage h, 2 h 1 −z (h − z) exp dz z0 = √ 2h20 h0 2π −∞ 2 h −z c (h − z) exp dz q= √ 2h20 h0 2π −∞ 2 h −z Kc2 dz (h − z)2 exp S= √ 2h20 h0 2π −∞
(2)
where z0 is the mean water depth, q is the discharge per unit width, K is an erodibility constant and S is the sediment transporting capacity per unit width. At high flow depths, when most of the surface is submerged, mean flow depth is almost identical to stage, and sediment discharge is almost directly proportional to discharge squared. At low depths, when flow is confined to topographic lows, the concentration of flow in a few threads produces a
208
Mediterranean Desertification
sediment transport that is disproportionately high relative to q 2 (though absolutely low). Empirically, the expression above may be approximated as S = 0.85(q 2 + q q0 ) K
(3)
where q0 is the unit discharge c.h0 . Comparing this estimate with the full calculation using equation (1) above, it may be seen that there is excellent agreement, except at very shallow depths, where sediment transport and mean flow depth are, in any case, negligible. To estimate water and sediment discharge down the length of the slope, we note that the flood-wave travel time from a distance x upslope is x/c, so that the standard error estimate for the mean intensity should be scaled up to s(x) = (x/x0 )−α i0 , where i0 is the mean daily rainfall intensity, x0 is the critical length c.1 and the time unit is taken as one day. This makes use of widely available daily rainfall data, although shorter time units may also be used. At present we plan to use one-hour time periods as a standard in the model wherever possible. Summing over the frequency distribution of rainfall intensities, which is assumed normal, and ignoring microtopography initially (i.e. h0 = 0), ∞ x −(i − i0 )2 √ di (i − f ) exp q= 2s 2 (x) s(x) 2π i=f ∞ −(i − i0 )2 Kx 2 (i − f )2 exp di (4) √ S= 2s 2 (x) s(x) 2π i=f where f is the daily rate of infiltration. These expressions may be integrated in the forms: exp(−u2 /2) q − uf (u) √ = y (1−a) i0 x0 2π u exp(−u2 /2) S (2−2a) 2 √ = y (1 + u )f (u) − K(i0 x0 )2 2π
(5)
where √ f (u) = 21 erfc(u 2) u = −(i0 − f )/s(x) = −1(1 − β)y α β = f/i0 y = x/x0 S0 indicates the sediment discharge for zero roughness. For mean rainfall less than infiltration rate (β ≤ 1), equation (5) can be taken as it stands. Significant contributions come from flows generated over the entire distance upslope. However, where infiltration exceeds mean rainfall rate, equation (6) implicitly contains an additional integral. Only small and infrequent flows are derived from the full slope length. Additional flows are derived more locally, so that contributions must be added for values of distance x less than the full distance from the divide, weighted for the extension of the frequency distribution for lower values of u. Thus for i0 < f (i.e. β > 1 and u > 0), we have u (1−a)/a exp(−u2 /2) v q = y (1−a) − uf (u) + f (v) dv √ i0 x0 b−1 2π 0 u exp(−u2 /2) S (2−2a) 2 √ = y (1 + u )f (u) − K(i0 x0 )2 2π (2−2a)/a u 2 exp(−v /2) v + √ dv (6) − vf (v) b−1 2π 0
209
MEDRUSH: A Basin-Scale Physically Based Model
Equations (5) and (6) show that, at low infiltration rates, there is net overland flow produced everywhere, so that discharge increases linearly downslope, and sediment discharge as the square of the distance (for a given gradient). As infiltration exceeds mean rainfall, only occasional bursts of rain produce overland flow, and accumulate only over short stretches of the slope, so that water and sediment discharge flatten off with distance progressively closer to the divide. This is in accord with observed behaviour on semi-arid to arid slopes, as has been noted by Dunne and Aubry (1986) and Yair and Lavee (1982) among others. For a rough surface, the discharge squared term, q 2 = [(i − f )x]2 in equation (4) for sediment discharge must be replaced by the modification of equation (3). The forecast sediment discharge is thus replaced by a linear combination of the two terms shown in equations (5) and (6), which provides an estimate of the roughness dependence. The equations for water discharge are unchanged, provided that our initial assumption of constant overland flow routing velocity is observed. Combining the equations gives S0 h0 q Sh = + (7) K(i0 x0 )2 K(i0 x0 )2 i0 i0 x0 where Sh indicates the sediment discharge for roughness h0 . Equation (7) shows that the concentration of flow within roughness elements produces increases in sediment discharge everywhere. Increases are greatest at rainfalls that are low relative to infiltration, and greatest near the top of the slope. Increases in infiltration lead to a reduction in sediment transport that is greatest downslope. Thus days when the soil is dry may lead to deposition downslope, while smaller rainfalls after wetter antecedent conditions may be associated with similar rates of sediment transport overall, but with downslope incision. Increases in microtopographic roughness give increases in sediment transport everywhere, but these increases are least downslope. The effect of roughness decreases, however, as storm size increases, and the roughness elements are drowned out. Subsurface flow is calculated using TOPMODEL (Beven and Kirkby 1979), but with allowance for downslope differences in the amount of water percolating into the saturated zone (Kirkby 1986). The hourly time step is broken down into variable increments to maintain computational stability during intense rainfall events. Where necessary, exfiltrating saturation overland flow (return flow) is also added to the Hortonian overland flow described above. This is calculated from the intersection of the rough (microtopography) surface with the mean saturated deficit level within the soil. The relationships between the flow components are shown schematically in Figure 16.2. 2.3 Grain-size Effects
The effect of changes in grain size with travel downslope may similarly be integrated over the frequency distribution, at least to a first, linear approximation. Many grain-size distributions are approximately log-normal in form, so that the source distribution at a point, before transport, may be expressed in the following form: ... 2 1 d − 1 d (8) p(d, 0) = √ exp − 2 σ σ 2π ...
where d is the grain size in phi units, d = − log2 (grain size in millimetres), and d and σ are the mean and standard deviation of the distribution. Assuming that travel distance is inversely proportional to grain size to the mth power, then the mean travel distance for size d is ... ... x = x0 2(d− d )m (9) ...
where x0 is the mean travel distance for the mean diameter d . For moderate travel distances, at which the source material is not exhausted, an inverse exponential (or Ŵ (1)) distribution of travel distances is appropriate (Kirkby 1991), giving a rate of deposition
210
Mediterranean Desertification Fractal distribution of intra-hour intensities
Hourly rainfall Infiltration rate
Infiltration capacity
Unsaturated infiltration store
Hortonian overland flow
Key Water flow Causality Flow
Percolation rate Surface roughness
Saturated subsurface store
Store
Subsurface flow referred to mean surface
Intersections with depressions in microtopographic surface
Return flow (exfiltration)
Reduced subsurface flow
Figure 16.2 Schematic relationships between overland and subsurface flow components in the MEDRUSH model
at distance x, for grain size d, of
x ... 1 x ... 1 exp − ... = 2(d− d )m exp − 2( d −d)m x x x0 x0 ...
(10)
Combining these expressions, the mean grain size of the transported material in phi units is given as ... ... x −1 (11) d (x) = d (0) + mσ 2 ln(2) x0 ...
where d (x) indicates the mean after travel distance x. This approximation shows that the mean grain size is coarsened close to the source, unchanged at the mean travel distance and finer at greater distances. Summing and weighting across the distribution of deposition given in equation (9), the mean grain size of the transported material as a whole is properly conserved. It is assumed here that the phi standard deviation (σ ) is preserved during transport. 2.4
Sediment Transport in General
Sediment transport of all kinds is modelled as an erosion-limited process. This is similar in principle to its inclusion in the MEDALUS I catena model, and has been described in greater detail in Kirkby (1992), although set there in the context of integration of storm impacts over longer periods. This approach allows fine sediment transport to be effectively limited by supply, while coarse sediment transport is limited by a limited travel distance, and is essentially flux-limited. In this approach, sediment transport is governed by the continuity equation, and constrained by a sedimentation balance. For each individual process, the rate is determined by two quantities, the rate of detachment D and the travel distance h, both of which generally vary with rainfall, flow and/or surface conditions. ∂S S dw S ∂z = − =D− (12) − ∂t ∂x w dx h
MEDRUSH: A Basin-Scale Physically Based Model
211
where S is the actual sediment transport per unit width, w is the local flow strip width, z is elevation, and x is the distance measured down the flow strip. The first part of this equation represents the continuity equation, taking due allowance for flowstrip convergence or divergence. The second part of the equation is the sedimentation balance, in which the rate of detachment, D, is balanced against the rate of sedimentation, S/ h. Clearly the capacity transport rate C = Dh. Where the travel distance, h, is small, then the sediment transport is close to its capacity, and where h is very large relative to the flow-strip length, removal is essentially supply-limited, following the equation ∂z =D (13) − ∂t 2.5 Sediment Transport by Wash Processes
Wash processes are considered to include rainsplash, rainflow and rillwash. In the first two of these processes, detachment is by raindrop impact, and in the third by flow traction. In rainsplash, travel is by aerial saltation, and in the last two processes travel is within the overland flow. We therefore need to specify two processes of detachment and two processes of travel. Raindrop detachment is modelled as independent of gradient and grain size, varying as the square of rainfall intensity. Detachment is limited by flow depth beyond about 6 mm. The attenuation is modelled as D = (1 + y) exp(−y) (14) D0 where D0 is the detachment on a bare surface and y is the ratio of flow depth, z, to attenuation depth, z0 . A given average flow depth, z¯ , may be converted to an actual flow depth relative to the roughness elements, by solving the following equation for z:
z 1 z′ 2 1 ′ dz′ (15) (z − z ) exp − z¯ = √ 2 h0 h0 2π −∞ Summing for equation (14) over this distribution, the overall efficiency of detachment is
∞ z ′ D z − z′ 1z2 1 1 ′ 1 + exp − √ dz + = √ D0 2 h20 z0 h0 2π z h0 2π −∞ 1 z′ 2 z − z′ exp − dz′ × exp − z0 2 h
(16)
Using this relationship, it may be shown that the attenuation depth controls the decay with depth for smooth surfaces. For surfaces with a roughness greater than the attenuation depth, the dominant effect is the exposure of significant unsubmerged areas due to the concentration of flow in the depressions. Movement of splashed material takes place both downslope and laterally. For erosion of the hillslope as a whole, the downslope direction is relevant. The travel distance may be calculated if all or part of the momentum of a raindrop is transferred to an underlying particle, and this impulse is used to project particles equally at all vertical and horizontal projection angles. On a gradient this leads to a net mean downslope travel distance, x, of n dR 4 vT2 2 (17) x= π g d where vT is the raindrop terminal velocity, dR is its diameter, d is the grain diameter, 2 is the slope gradient, and n is an exponent that takes the value of 2 for d < dR , and 6 otherwise.
212
Mediterranean Desertification
This expression contains rainfall-related terms that have already been included in the detachment process, leaving the gradient and grain-size terms as the independent components of travel distance. For momentum transfer to an average detached layer, the detachment component, D0 , is D0 =
vT2 2g
The exponent, n, is controlled by the way in which momentum is transferred. If the raindrop is smaller than the grain, then the whole of its momentum is transferred. If larger than the grain, then only a cylindrical cross-section impacts the grain, the remainder of the drop falling around it. Clearly for a range of raindrop sizes and velocities, and partial grain overlaps, there is a cross-over zone between these behaviours, and travel distance can be approximated, ignoring constants, as follows: 2 (18) x= 2 t (1 + t 4 ) where t is the ratio d/dR . This expression gives a smooth transition in the region of the raindrop diameter, dR , and a maximum grain transport (defined as xd 3 ) for d = 0.77 dR . In the cross-slope direction, splash is a critical process in softening microtopographic roughness. At low roughness, the driving gradients increase linearly with roughness, h0 , but beyond a critical point, hc , lateral gradients encourage rill-wall collapse, which has been identified as an important process (de Ploey 1983). Summed over relevant grain sizes, the contribution of rainsplash to roughness reduction may be expressed in the approximate form h0 dh = −µh 1 + (19) dt hc where µ has the depth dependence characterized by equation (16). We will return to the issue of roughness generation below. Flow detachment, Dc , is modelled through a threshold power, in the following form: Dc ∝ q2 −
(20)
where q is the overland flow discharge per unit width, and is the detachment threshold. Thresholds may be set by turf strength or grain characteristics according to surface conditions. For Mediterranean conditions, grain thresholds are widely relevant, with components for grain friction, cohesion and corrections for steep gradients. An appropriate form (Kirkby et al. 1993) is d2 tan φ d+ c (21) = 0.06c4 tan φ − 2 d where 4 is the ratio of submerged grain to water density (≈1.65), φ is the angle of grain friction (≈35◦ ), c is the overland flow routing velocity, and dc is the grain size for minimum traction threshold (≈0.1 mm). Travel distance in rill and inter-rill flows is taken as proportional to flow discharge per unit width, q. Thus the total transporting capacity in rill flow, for example, is given as q(q2 − ) = 2 q2 − q (22) C = Dc h c ∝ The summation takes place, as above, by integrating over the full range of flow depths within the distribution of microtopography. Rillwash is also selective in modifying microtopography. In an erosive or depositional event, the deepest flow strands are most strongly affected, while high areas above the flow level are unaltered. The overall effect can be summarized by the change in the standard deviation of microtopographic elevation, which is our parameter for describing surface roughness.
MEDRUSH: A Basin-Scale Physically Based Model
213
2.6 Changes in Surface Roughness Over Time
In a rillwash event, sediment transport is dominated by the q 2 term in equation (22). Over a short distance downslope, discharge increases more or less linearly with distance, so that the rate of erosion is proportional to d(q 2 )/dx ∼ q. In other words, local erosion within the microtopography is proportional to flow depth. Summing across the microtopography, we obtain z 1 z′ 2 λ ′ (z − z ) exp − dz′ = λ¯z E= √ 2 h h 2π −∞ ∞ 1 1 z′ 2 ′2 ′ 2 h = √ (z + E) exp − dz′ 2 h h 2π z z 1 z′ 2 1 ′ ′ 2 dz′ (23) [z + E − λ(z − z )] exp − + √ 2 h h 2π −∞ where E is the local integrated erosion rate, z¯ is the mean overland flow depth, h′ is the modified roughness, calculated from the definition of variance, and λ is a measure of the local erosional intensity (<0 for deposition). The first term in the modified roughness is the contribution from unsubmerged areas, and the second is the contribution from the flow-covered areas, allowing in both cases for the change in mean by the erosion, E. Over an incremental time period, we may ignore terms in E 2 , and derive the rate of change in roughness as follows: z z E 1 1 z′ 2 1 z′ 2 1 dh ′ exp − exp − dz = λh √ dz′ (24) = h √ dt z¯ h 2π −∞ 2 h 2 h h 2π −∞ Empirical examination of this relationship shows that the following expression is a working fit to equation (24) for all significant erosional events: dh = λh[1 − exp(−1.8¯z/ h)] dt
(25)
We have identified two influences on surface roughness, infilling and roughness reduction by rainsplash, and erosion or deposition by rillwash. A third significant factor is the biological activity directly and indirectly associated with vegetation. Low mounds surrounding long-lived perennial shrubs are thought to form in a number of ways: 1. through protection from rainsplash; 2. through reduction in overland flow and rill erosion (both by enhanced infiltration and by flow diversion around the mound topography); 3. through the preferential accumulation of plant litter; and 4. through preferential faunal burrowing beneath the shade of vegetation. In the absence of clear process guidelines, these processes are lumped together as a single roughnessenhancing constant rate, V . The overall expression for the rate of change in roughness is then dh = V − µh(1 + h/ hc ) + λh dt
(26)
where λ, µ are measures of the erosional environment, and are taken to incorporate depth dependent terms. µ, the intensity of splash, is expected to be strictly positive, while λ reflects the change in rillwash transport downslope, and may take positive or negative values. For any steadily applied erosional environment, equation (26) has a stable equilibrium solution: h∞ =
hc (λ − µ) + [(λ − µ)2 + 4V µ/ hc ]0.5 2µ
(27)
214
Mediterranean Desertification
This solution includes the special case of no rillwash (λ = 0). Typically the critical roughness hc for bank collapse will not be reached in this case, so that for unrilled and depositional environments (λ ≤ 0), the equilibrium roughness is approximately V /(µ − λ). The magnitude of the coefficient µ may be estimated from splash diffusivity (∼3 × 10−3 m2 year−1 ) and the size of the characteristic topography (∼1 m) as 3 × 10−3 year−1 , indicating survival times for abandoned mounds and rills of about 300 years. If an unrilled topography has a roughness of 100 mm, then the vegetation term, V , in equation (26) is about 3 × 10−4 m year−1 , or 0.3 mm year−1 net growth in roughness due to vegetation mounding. To be consistent, we would expect the critical roughness, hc , to generate bank collapse to be about 200–500 mm. Where rill erosion is active enough for λ to exceed µ, then rill channels tend to enlarge, and would grow in an unstable positive feedback if bank collapse did not limit them. The condition λ > µ may therefore be taken as a primary condition for the presence of active rilling. The balance of factors in this roughness equilibrium is equivalent to the conditions for stability described by Smith and Bretherton (1972) in theory, and by Dunne and Aubry (1986) for Kenya in the field. In a changing erosional environment, the roughness is constantly adjusting to current conditions. For much of the time roughness is very slowly decreasing, with bursts of regeneration in storms. Rates of change of roughness vary with roughness and erosional intensity, including the equilibrium relationship with zero rate of change. Down the length of a slope catena, we expect the erosional intensity to be low near the divide, to increase downslope, and perhaps to change to deposition (negative intensity) near the slope base. The equilibrium line suggests that there should be a corresponding increase in roughness with intensity, falling to lowest roughnesses in the depositional area. Using these order of magnitude values, it can be shown that responses to storm events may be rapid, but that in general equilibration to average conditions takes several hundred years, and is completed sooner where the change is towards lower roughness (i.e. in depositional environments) than where roughness is increased. 2.7
Construction of Sub-basins and Representative Flow Strips
Following the construction of suitable digital elevation models for the Agri catchment, an automated procedure for surface water routing and subdivision of the catchments was undertaken using GRASS modules written for the purpose. These modules calculate the accumulated upslope drainage area and principal flow direction at each point to create sub-basin raster maps and a river network vector map for flows within and between sub-basins. Two different types of sub-basin are produced, some covering headwater areas (leaf-type sub-basins) and others containing one or more through-flowing streams (stem-type sub-basins). Sub-basins are also selected with a threshold size that increases with stream order, to provide greater detail in the catchment headwater areas. Flow paths and networks are accumulated by sorting all cells in the DEM in altitude order, and applying a multiple flow direction algorithm to share the outflow between all lower neighbours. The merits of alternative weighting schemes, and ways of ensuring that streams follow their thalwegs are discussed by Quinn et al. (1991). Here a cubic weighting of gradients to the eight neighbouring points has been used, and this has been found as a satisfactory compromise which gives strong dominance to thalweg flow paths, but still allows some distribution of flow on fan and other divergent flow areas. The flow direction vector is then drawn in the direction of the neighbouring cell with the highest accumulated upslope drainage area value. For large flat areas, this procedure has been implemented by creating a tree of drainage directions, working upstream from the lowest exit point. Finally, the catchment outlet point is located by following the flow direction map to the edge of the catchment, and then recursively ascending against the principal flow direction to accumulate drainage area. Where the total accumulated area first exceeds the threshold associated with main-stem stream order, it defines the position as the mouth of a sub-basin, and each sub-basin is tagged to avoid re-use and link it to the catchment network structure, using the “segment ordering” system of Shreve (1967, cited in Gregory and Walling 1973). This scheme, using a principal flow direction, is a compromise between the single- and multi-path analyses, and has the advantage of providing unambiguous subbasin definition. This scheme was used to generate 208 sub-basins for the Agri Basin, ranging in
MEDRUSH: A Basin-Scale Physically Based Model
215
1) Section 1 (top) 2) Section 2 3) Section 3 4) Section 4 5) Section 5 (foot)
Figure 16.3 (top to foot)
Division of the Agri catchment into sub-basins and representative catena sections
size from 4.70 to 28.83 km2 , with the majority in each instance being of a similar proportion, as shown in Figure 16.3. The principal flow strips for each sub-basin were also extracted automatically within GRASS. At the same time, a relative strip width figure is calculated for each cell in that particular track according to local surface curvature, based on accumulated upslope drainage area and downslope distance from the start of the track. This figure therefore provides a quantitative estimate of both convergence and divergence on the track. The representative width, w, is approximated as 0 da − 1 dx (28) 1/a w = exp dx z where a is the accumulated upslope drainage area, and x is the distance from the top of the slope. These paths, which as stated can start at both sub-basin boundaries and at the occasional internal within-area peaks, perforce will be of different lengths. Thus the next step is to normalize the data set for each sub-basin such that the overall lengths and total drop match that of the longest principal strip. Normalized values are then averaged, taking the median value in each case, to produce a representative hillslope profile, which in combination with the mean width figure at each point produces the representative flow strip. The dimensions of the representative flow strips are scaled up to match the real length and drop of the longest strip in each sub-basin. The reason for using the longest flow strip is that, given an idealized basin, it is the centre strip that is most representative of the sub-basin as a whole. The final product from this exercise is therefore a set of representative flow strips on which the MEDRUSH hillslope model can be run. A number of strategies were tried to create a suitable representative flow strip. None is fully satisfactory, and the best are only considered adequate for estimating short-term, and therefore relatively minor, changes in the sub-basin. To transfer data to the representative flow strip, mean values are calculated for each required input value, thus all equivalent parts of the slope are modelled using equivalent input statistics. Mean values were used to ease the computational load, since median values are more time-consuming to compute, and mean values are considered sufficient to provide acceptable input data at this level of spatial generalization. To update the spatial database from the hillslope models, as they evolve, three strategies have been tried, each with some positive features, although none is fully satisfactory (Kirkby 1999). One proposal is to transfer changes in each variable on the basis of common values of area drained per
216
Mediterranean Desertification
unit contour length (referred to as unit area below); a second is to use the wetness index (the ratio of unit area to local gradient); and a third is to use elevation. There are theoretical reasons for preferring each of these in certain ideal circumstances, but all tend to change steadily down-catchment, so the differences between them are not large in relation to the errors except over periods longer than those for which MEDRUSH has been designed.
3
THE VEGETATION GROWTH MODEL
The model simulates processes of primary productivity and evapotranspiration in stands of vegetation, and focuses on vegetation functions that are likely to be involved in mediating responses to atmospheric and climate change (see Osborne et al. 2000). In particular, responses of canopy gas exchange to rising CO2 are considered important, and are explored elsewhere in this volume (Chapter 3). A generic vegetation model is applied to groups of plant species by using a different set of key model input variables for each. 3.1
Plant Functional Types
The use of plant functional types in landscape-scale models is preferable to modelling at the species level, because of the reduced complexity required and the scarcity of data available for most individual species (Smith et al. 1993). Four functional types are currently simulated: evergreen sclerophyllous shrubs, drought-deciduous shrubs, perennial tussock grasses and winter annual grasses. Functional types are defined according to their strategies for surviving summer drought. Sclerophyllous shrubs have tough, evergreen leaves, which have adaptations for minimizing water loss and damage due to high temperatures, and remain physiologically active throughout the summer drought period (Table 16.1). In contrast with the other functional types, they tend to be deep-rooted and many reach depths of over 5 m, often allowing access to water throughout the summer (Specht 1988; Table 16.1). Despite this, they tend to conserve water through stomatal closure, and can remain physiologically active at low soil water potentials (Archibold 1995; Table 16.1). In contrast, drought-deciduous shrubs “avoid” the summer drought, becoming dormant after leaf abscission at Table 16.1 Comparison between Mediterranean plant functional types that are simulated by the MEDRUSH vegetation model
Life history Leaf phenology Plant life span Drought adaptations Rooting depth Wilting point Growth during drought? Primary productivity Photosynthetic rate Respiration rate Storage capacity
Evergreen sclerophyllous shrub (e.g. Pistacea lentiscus)
Droughtdeciduous shrub (e.g. Anthyllis cytisoides)
Perennial tussock grass (e.g. Stipa tenacissima)
Winter annual grass (e.g. Vulpia ciliata)
Evergreen
Deciduous
Deciduous
Perennial
Perennial
Facultatively deciduous Perennial
Deep Low Yes
Shallow High No
Shallow Low Opportunistic
Shallow High No
Low Low High
High Low High
Low Low Moderate
High High Low
Annual
217
MEDRUSH: A Basin-Scale Physically Based Model
the beginning of the summer, which may be triggered by low soil water potential or phenological cues such as day length (Margaris 1975; Smith et al. 1997; Table 16.1). Leaves tend to be intolerant of drought and physiological activity ceases at a relatively high water potential (Clark et al. 1998; Table 16.1). The leaves of perennial tussock grasses persist and growth continues throughout the summer period, provided that sufficient water is available (Pugnaire and Haase 1996; Pugnaire et al. 1996; Table 16.1). In common with sclerophyllous shrubs, stomatal closure and morphological adaptations of leaves tend to minimize water loss in transpiration and damage due to high irradiance, and tussock grasses can remain physiologically active at low soil water potentials (Pugnaire and Haase 1996; Pugnaire et al. 1996; Table 16.1). Winter annual grasses avoid drought by completing their life-cycle before or at the start of the summer dry season (Clark et al. 1998; Table 16.1). While their productivity during the wet season is high, drought resistance is low, and physiological activity ceases at a relatively high soil water potential (Table 16.1). 3.2 Model Functions
Model functions are summarized below and in Figure 16.4, and are described in detail by Osborne et al. (2000). Further applications of this model are presented elsewhere (Woodward and Osborne 2000; Osborne and Woodward 2001). The vegetation model requires only climate data, CO2 concentration and soil water potential as inputs, and predicts biomass, net primary productivity (NPP), leaf area index (LAI), evapotranspiration and litter production. Canopy photosynthesis provides carbohydrate for growth, and is calculated as a linear function of absorbed solar radiation (Figure 16.4; Monteith 1972; Haxeltine and Prentice 1996). Photosynthetic rate varies in response to atmospheric CO2 concentration, air temperature and soil water potential. Increases in photosynthesis which occur at high CO2 concentration interact with solar radiation and temperature, according to functions that were developed using a biochemical model of canopy photosynthesis (Wilks et al. 1995). Respiration consumes carbohydrate, and is partitioned between maintenance and growth processes, the former being dependent on temperature (Figure 16.4;
CO2
Photosynthesis a
CH2O
Maintenance b respiration
CO2
Storage
Biomass
Growth NPP
Rain H2O
Soil H2O
Evapotranspiration c
Litter + Fruit
H2O Vapour
Figure 16.4 Overview of vegetation model processes. Flow of matter is shown by the solid arrows; pools of matter are highlighted in bold boxes (CH2 O = simple carbohydrates); and model processes are enclosed by normal boxes. Atmospheric CO2 , climate and soil influence model processes through (a) the response of canopy photosynthesis to atmospheric CO2 concentration, temperature and soil water potential; (b) the temperature-dependence of maintenance respiration; (c) changes in evapotranspiration via effects of air temperature, vapour pressure deficit (VPD) and soil water availability, and the response of canopy stomatal conductance to CO2 concentration, temperature, VPD and soil water potential. Productivity and evapotranspiration models interact via soil water potential (dashed arrow)
218
Mediterranean Desertification
Thornley 1970). Carbohydrate is partitioned between storage and new growth (NPP; Figure 16.4) of leaves, woody stems, woody roots, fine roots and reproductive tissues in fixed proportions, which vary according to phenology. Winter annuals have no storage capacity, and neither grass functional type has woody tissues. Three phenological stages are distinguished for each functional type: a period of vegetative growth, in the winter or spring, when a large proportion of canopy and root growth occurs; reproductive growth, during the late spring or summer, when secondary thickening of wood also occurs in shrub functional types; and a period of dormancy, in the autumn and winter for sclerophyllous shrubs, and the summer for drought-deciduous shrubs and winter annuals. There is no dormancy in tussock grasses because of the opportunistic nature of growth. Rates of abscission and the death of organs are calculated using coefficients that relate to longevity and vary with phenological stage. However, litter production increases when storage is low and starvation of tissues occurs. Evapotranspiration is calculated using a modified version of the Penman–Monteith equation, which was developed for sparse canopies, and partitions water loss between canopy and soil surfaces according to net radiation at each (Shuttleworth and Wallace 1985; Shuttleworth and Gurney 1989). Canopy stomatal conductance is estimated following the approach of Jarvis (1976), and varies in response to atmospheric CO2 concentration, air temperature, soil water potential and vapour pressure deficit. Stomatal closure in response to CO2 concentration is assumed to be linear, following the general response of C3 plants (Morison 1985; although see Osborne and Woodward, Chapter 3 in this volume). The evapotranspiration model interacts with the vegetation productivity model via its effect on soil water potential (Figure 16.4), and influences the distribution of soil water, modifying infiltration, subsurface flow and other physical processes within MEDRUSH. 3.3
Vegetation Model Testing
The model has been tested for sites throughout the Mediterranean Basin, using the method of Mitchell (1997) and Mitchell and Sheehy (1997), where the deviation of model predictions from observations is compared with a standard that is set using independent criteria, and in advance of the comparison. Simulations of biomass, NPP and LAI were tested using published observations for Mediterranean sclerophyllous shrub vegetation at 18 sites (Figure 16.5). Observations were summarized as a mean for each site, and simulations were carried out for each using mean climate data from a nearby meteorological station (M¨uller 1982). The model was run to equilibrium using a daily timestep for primary productivity and an hourly timestep for evapotranspiration. The precision of model predictions was assessed by comparison with an estimate of the 95% confidence interval for observations. In ten observations of biomass made at two different sites in the Mediterranean, the 95% confidence limits were, on average, ±43% of the mean, varying between ±25% and ±75% (Trabaud 1991; Puigdef´abregas et al. 1996). The confidence interval tended to increase on a relative basis when biomass was less than 200 g m−2 . We therefore estimated confidence limits to be approximately ±50% of the mean value for biomass, NPP and LAI. Confidence limits for values of biomass and NPP less than 200 g m−2 were estimated as ±100 g m−2 , and for LAI less than 1.0, estimated as 0.5 m2 m−2 . Eight out of 14 predictions of biomass (57%), 4 out of 5 predictions of NPP (80%) and 10 out of 13 predictions of LAI (77%) were within our estimate of the 95% confidence interval for observations (Figure 16.6). Model predictions tended to be closest to observations for sites in the western Mediterranean Basin, in Portugal, Spain and France, and furthest from observations for sites in the eastern Mediterranean Basin, in Italy and Greece (Figure 16.5). Results also suggested a negative bias in model predictions at the most productive sites, many of which were located in the eastern Mediterranean (Figure 16.6). However, model predictions tended to be too high for the Rambla Honda site in south-east Spain, where productivity was very low (Figure 16.6). Success of the vegetation model in predicting biomass, NPP and LAI of sclerophyllous shrubs therefore varied between sites, but was generally good, especially for the western Mediterranean, where nearly 90% of observations (n = 9 for biomass and LAI) were predicted within the estimate of their 95% confidence limits.
MEDRUSH: A Basin-Scale Physically Based Model
219
(a)
Error (kg m−2)
(b)
2
Biomass
0 −2 −4 −6
0
2
4
6 −2)
(c)
Error (g m−2 year −1)
Observation (kg m 300 NPP 100 −100 −300
0
100
200
300 −2
Observation (g m (d)
400
500
year−1)
Error (m2 m−2)
4 LAI
2 0 −2 −4
0
1 2 3 Observation (m2 m−2)
4
Figure 16.5 (a) Locations of sites within the Mediterranean Basin used to test the vegetation model. Simulated values of (b) above-ground biomass, (c) above-ground net primary productivity (NPP) and (d) leaf area index (LAI) for Mediterranean sclerophyllous vegetation at each site were compared with published observations (Sources: Specht 1969; Lossaint 1973; Rapp and Lossaint 1981; Catarino et al. 1981; Rambal and Leterme 1987; Malanson and Trabaud 1988; Specht 1988; Tsiouvaras 1988; Merino et al. 1990; Trabaud 1991; Valentini et al. 1991; Arianoutsou and Paraskevopoulos et al. 1992; Pitacco et al. 1992; Diamantopoulos et al. 1993; ´ Harrison et al. 1993; Rambal 1993; Paraskevopoulos et al. 1994; Puigdefabregas et al. 1996; ´ Lopez-Berm udez et al. 1996; Rambal et al. 1996; Scarascia-Mugnozza et al. 1996) ´
4 THE CHANNEL ROUTING COMPONENT In the model the basin is divided into sub-basins of varying size and shape but typically larger than 5 km2 . Connections between sub-basins are provided by the river links or reaches and each reach accepts lateral inputs from the hillslope component (i.e. overland, subsurface baseflow and tributary
Mediterranean Desertification Biomass
2000 0 −2000 −4000 −6000
0
2000 4000 6000 Observed Biomass (g m−2)
300 NPP
200 100 0 −100 −200 −300
Error (Observed-Simulated)
Error (Observed-Simulated)
Error (Observed-Simulated)
220
0
100 200 300 400 Observed NPP (g m−2 y−1)
500
4 LAI 2 0 −2 −4
0
1
2 3 4 Observed LAI (m2 m−2)
5
Figure 16.6 Error in model simulations, calculated as the difference between model predictions and observations, for: above-ground biomass (kg m−2 ); above-ground NPP (g m−2 year−1 ); LAI (m2 m−2 ). Positive errors indicate an overestimation, and negative errors an underestimation, compared with observations. Dotted lines delimit an estimate of the 95% confidence limits for observations (see text for explanation). Symbols distinguish sites located in different countries (see map above)
flows) and inputs from upstream. The channel component was required to route water and sediment (by size fraction) along the channel network from the sub-basins to the basin outlet on the main river network and to simulate discharge at the outlet and at any point along the network. 4.1
Water Flow Routing
The routing scheme is required to be fast, to be computationally simple and to deliver a distributed output. The first two requirements are satisfied by using linear transfer functions derived from analytical solutions of the convection–diffusion equation. In connection with the third, two routing modes
MEDRUSH: A Basin-Scale Physically Based Model
Cascade mode
221
Direct mode
Figure 16.7 Cascade and direct routing modes for the MEDRUSH channel flow component, illustrated for a system of channel reaches
have been developed (Figure 16.7): a cascade system, routing from reach to reach and providing a spatially distributed output; and a direct, superposition scheme in which the discharge at each link is routed directly to the outlet. The direct scheme is faster and simpler than the cascade scheme but provides discharge at the outlet only and does not therefore allow sediment transport modelling.
Derivation of Transfer Functions The transfer functions describe the characteristic time of water flow through each reach. A parcel or impulse of water enters the head of the reach and the function provides the percentages of the total parcel which arrive at the reach outlet in given time intervals, e.g. 0% in hour 1, 10% in hour 2, 45% in hour 3. A range of functions, different for each reach, are required to allow for spatial and temporal variation in the routing time at the reach scale. A linear solution of the convection–diffusion approximation to the Saint Venant equations ∂Q ∂ 2Q ∂Q +C − D 2 = Cq ∂t ∂x ∂x
(29)
is used, where Q is channel discharge; C is a convection or celerity coefficient; D is a diffusion coefficient; q is lateral discharge per unit distance; t is time; and x is distance along the channel. Analytical solutions to impulse (parcel) inputs to a river reach can be found for two cases: upstream point input and uniformly distributed lateral input. Integration of these impulse responses provides pulse responses, equivalent to transfer functions. The discharge out of the reach is then given by Qout (t) =
M 2 j =1 i=1
H (j, i)Qin (j, t − i + 1)
(30)
where Qout is discharge out of the reach; Qin is discharge into the reach; H is the transfer function; i is the time index of the transfer function; j is input type (lateral or point); and M is memory (length) of the transfer function.
Parametrization Celerity, C, and diffusivity, D, are related to the discharge and channel characteristics by C=
1 dQ B dy
(31)
D=
Q 2BSf
(32)
222
Mediterranean Desertification
where Q is discharge; B is channel width; Sf is friction slope; and y is depth. Using the Manning or Ch´ezy resistance relationship, it can be shown that the celerity varies as αU , where U is flow velocity and 1.5 < α < 1.666 depending on channel geometry and the resistance relationship used. To linearize equation (29), C must be constant. Velocity–discharge (U –Q) curves are therefore discretized for each reach to define ranges through which C may be considered constant. A transfer function is then calculated for each value of C (corresponding to a U –Q range). The transfer function to be used in the simulation changes when the discharge into the reach changes to a different U –Q range. Ten U –Q ranges are discretized at each reach to ensure smooth transitions between functions.
Calculations of C and D for Cascade Routing To calculate celerity and diffusivity at each reach, a number of parameters are required, determined as follows: 1. characteristic discharges, including the mean annual flood, estimated by interpolation and drainage area weighting; 2. channel bankfull width, calculated using the between-site equation of hydraulic geometry for width (Leopold and Maddock 1953); 3. friction slope, Sf , approximated by the mean channel bed slope over the reach, calculated from a digital elevation model (DEM) of the basin; 4. the Manning resistance coefficient, n, determined from formulae or tables. The mean flow velocity, U , is calculated for the 10 discharge ranges using the Manning relationship, with the simplifying assumption of a rectangular channel. C and D are then calculated from equations (31) and (32). Parametrization of the channel routing component requires a digitized river network, normally to be obtained using network node information (position coordinates and elevation) supplied from the automated subcatchment division program operating on the catchment DEM (see section 2.7). The channel transfer functions are then calculated from values of C and D derived from the network node spacings, mean channel slopes and estimated channel dimensions. 4.2
Channel–Aquifer Interaction Groundwater effects are represented by a “bank storage element” which exchanges water with the channel according to the relative head difference. The bank element has the area L × W and contains groundwater in a matrix of porosity θ with head ha (see Figure 16.8 for an explanation of terms). Flow between the channel and the bank element is specified by the relative heads and a user-defined transmissivity. The exchange discharge Qb is calculated as
Qb = k(hc − ha )
(positive out of the channel)
Qb ≤ 0 if d = 0
(limited by available water)
(33)
and ha changes through each timestep as ha1 = ha2 + Qb θ dt/Lw
(34)
where ha1 and ha2 are the values at the end and start of the timestep respectively. The discharge Qb is then supplied to the channel routing component at the next timestep as a lateral input (which may be positive or negative) and routed downstream. 4.3
Sediment Routing
The sediment routing scheme runs in step with the flow routing scheme, to solve the equation for conservation of sediment mass. An upwind difference scheme was developed, operating on the
223
MEDRUSH: A Basin-Scale Physically Based Model
Qb d
zb
hc
w w = width of bank element (m) (user defined) L = length of channel element (m) (user defined) Qc = channel discharge (m3s−1) (routing model) zb = bed elevation (m) (user defined) ha = aquifer head (m) (calculated) hc = channel head (m) (calculated = d + zb) d = channel water depth (m) (calculated from Qc by a stage-discharge relationship) k = transmissivity (m2s−1) (user defined) Qb = exchange discharge
Figure 16.8
L
ha
Qc
Schematic diagram of the MEDRUSH channel–aquifer interaction model
channel reaches in descending order of elevation. An adaptive time weighting has been used, dependent on the Courant number at each reach and timestep. This ensures stability, whilst minimizing numerical dispersion, which is already present owing to the water flow transfer function method. For each sediment size fraction and each reach, a semi-implicit finite-difference mass balance gives An+1 Lcn+1 − An Lcn = Qnup + Qnl − An Vsn cn+σ 4t
(35)
where A is flow cross-sectional area; L is reach length; c is volumetric concentration of sediment transport; Qup is volumetric rate of upstream input of sediment; Ql is net volumetric rate of input from sources (such as bank erosion, overland flow, infiltration into the bed); Vs is sediment particle velocity; 4t is timestep; σ is a time weighting factor (σ = 0 gives an explicit scheme, σ = 1 gives an implicit scheme); and n indicates the time level. The last group of terms on the right-hand side represents the rate at which sediment leaves the bottom end of the reach. The scheme allows for two particle sizes, fine and coarse. The fine fraction moves at the water velocity, while the coarse fraction moves more slowly. Transport is limited by a capacity rate; excess sediment falls to the bed, and may be re-suspended if discharge increases. The transport capacities and coarse sediment velocity are pre-computed in order to reduce program running time, and are held in a look-up table referenced by the channel discharge. 4.4 Verification of Routing Schemes
Channel Routing The channel routing component is designed to accept lateral inputs from each MEDRUSH subbasin and to route these inputs along the channel network. A full test of the component therefore requires data on the inputs as well as the corresponding discharges along the channel. However, measurements of lateral inputs along an entire network are not generally available for large basins. The component (not including the channel–aquifer interaction model) was therefore tested using artificial data and data generated by other models.
224
Mediterranean Desertification
Good agreement was obtained in comparisons of the analytical channel routing solution with the numerical solution of the MIKE11 hydraulic software package (DHI 1989) for an idealized single reach case. Good agreement was also obtained between MEDRUSH and SHETRAN discharge simulations for the 300-km2 upper South Tyne catchment in England, using lateral input data generated by SHETRAN. Application of the routing model to the channel network of the Agri target basin in southern Italy, using artificial input data and a range of input conditions, likewise produced satisfactory discharge simulations, with a physically reasonably level of dispersion. (For a description of SHETRAN and the Agri Basin, see Chapter 29.) The tests with the Tyne and Agri channel networks showed the model to behave as expected as the values of the routing parameters C and D vary. Crucially, reasonable values of C may be derived from estimates of the water flow velocity and measurable channel properties. However, there is a need for field data to provide a basis for checking such parametrization. Values of D are more difficult to derive but the sensitivity of the scheme to D is low. The transfer function approach has also proved to be robust and very fast. For example, a one-year simulation for the full Agri channel network was completed in less than one minute on a SUNSparc20 workstation.
Sediment Routing The sediment routing scheme was tested on the Agri channel network using artificial inputs designed to invoke the full range of possible conditions in the model. Sediment pulses were satisfactorily routed and mass conservation was observed exactly. In general, the sediment routing scheme is of similar robustness and speed to the flow routing scheme. Its main limitation is the dispersion introduced by considering each channel reach as a computational element for the finite-difference scheme. The scheme assumes immediate mixing of inputs at a reach, which is more realistic for short rather than long reaches. Channel reaches should therefore be less than about 5 km in length for the false dispersion introduced to become a second-order error.
5
IMPLEMENTATION
MEDRUSH has been implemented in C++, within the GRASS GIS, using Borland Turbo C++ compilers on Windows and UNIX platforms. Some features are only available within UNIX. A detailed manual, describing parameter requirements, set-up procedures and program modules, is available through the MEDALUS website, at http://www.medalus.demon.co.uk.
6 1.
CONCLUSIONS
The MEDRUSH model contains many innovative features, which have been integrated through collaboration between modelling groups specializing on hillslope runoff processes, eco-physiology and channel hydrology. 2. The sub-basin model contains a flow-strip model for water and sediment routing which provides strong interactions with soil and vegetation as they change over time, both seasonally and in the longer term. This is a critical component to allow forecasting in a global change context. The representative flow-strip concept is acceptable for distributing short-term changes, but would not be satisfactory for longer term (>100 year) forecasts. 3. Simulations of vegetation productivity and canopy size showed good agreement with observations made throughout the Mediterranean region, giving confidence in these key model variables. The vegetation model is grounded firmly in plant eco-physiology, and therefore provides a mechanistic basis for plant sensitivity to climate, soil properties and atmospheric CO2 . However, it remains computationally straightforward, thereby allowing the relatively rapid simulation of many vegetation patches for long time intervals. 4. The channel water flow and sediment transport routing component was subjected to a thorough verification programme. Comparison of test simulations with alternative model solutions
MEDRUSH: A Basin-Scale Physically Based Model
5.
225
and with expected mathematical performance showed excellent agreement. Through its analytical solution the component is fast, robust and flexible, it retains a firm physical basis and it incorporates an innovative approach to flow routing. Verification of model performance to date has primarily been at the level of individual components, through a clear physical understanding of each component process. Attempts to validate the model for large catchments have proved impracticable, and further development of the model is likely to concentrate on smaller catchments, and consequently with timesteps shorter than the 1-hour increments currently used. These methods are being applied to 10–150 km2 subcatchments, in both the Agri and Guadalent´ın (south-east Spain) catchments.
ACKNOWLEDGEMENTS The following contributed significantly to the development of the MEDRUSH channel component: Dr John Ewen (University of Newcastle upon Tyne), Dr Pascal Lardet and Douglas Clark (both former members of WRSRL, University of Newcastle upon Tyne). Most of the work reported here was funded within the MEDALUS II project, by the European Commission under its Environment Programme, contract numbers EV5V-CT92-0128/0164/0165 and 0166, and this support is gratefully acknowledged.
REFERENCES Archibold OW (1995) Ecology of World Vegetation. Chapman & Hall, London. Arianoutsou M and Paraskevopoulos S (1992) Some aspects of the mineral cycling in a maquis (evergreen sclerophyllous) ecosystem of northeastern Greece. Israel Journal of Botany 41, 135–144. Beven KJ and Kirkby MJ (1979) A physically based variable contributing area model for basin hydrology. Hydrological Sciences Bulletin 24, 43–69. Catarino FM, Correia OCA and Correia AIVD (1981) Structure and dynamics of Serra da Arr´abida mediterranean vegetation. Ecologia Mediterranea 8, 203–222. Clark SC, Puigdef´abregas J and Woodward FI (1998) Aspects of the ecology of the shrub-winter annual communities of the Mediterranean Basin. In P Mairota, JB Thornes and N Geeson (eds), Atlas of Mediterranean Environments in Europe. The Desertification Context. John Wiley, Chichester, pp. 44–47. De Ploey J (1983) Runoff and rill generation on sandy and loamy topsoils. Zeitschrift fur Geomorphologie, Supp Bd 46, 15–23. DHI (1989) MIKE11: A Microcomputer Based Modelling System for Rivers and Channels. Documentation and User’s Guide. Danish Hydraulic Institute, Hørsholm, Denmark. Diamantopoulos J, Stamou GP, Pantis J and Sgardelis S (1993) Petralona, Thessaloniki, Greece. In JB Thornes and CJ Brandt (eds) MEDALUS I Final Report. European Community, pp. 560–580. Dunne T and Aubry BF (1986) Evaluation of Horton’s theory of sheetwash and rill erosion on the basis of field experiments. In AD Abrahams (ed.) Hillslope Processes. Allen & Unwin, London, pp. 31–53. Gregory KJ and Walling DE (1973) Drainage Basin Form and Process: A Geomorphological Approach. Edward Arnold, London. Harrison A, Taberner M and Hurcom S (1993) Site-based remote sensing of vegetation and land cover. In JB Thornes and CJ Brandt (eds) MEDALUS I Final Report. European Community, pp. 225–263. Haxeltine A and Prentice IC (1996) A general model for the light-use efficiency of primary production. Functional Ecology 10, 551–561. Jarvis PG (1976) The interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the field. Philosophical Transactions of the Royal Society, London, Series B 273, 593–610. Kirkby MJ (1986) A runoff simulation model based on hillslope topography. In VJ Gupta (ed.) Scale Problems in Hydrology. Reidel, Amsterdam, pp. 39–56. Kirkby MJ (1991) Sediment travel distance as an experimental and model variable in particulate movement. Catena Supplement 19, 111–128. Kirkby MJ (1992) An erosion-limited hillslope erosion model. Catena Supplement 23, 157–187. Kirkby MJ (1994) Thresholds, instability and frequency distributions. In MJ Kirkby (ed.) Process Models and Theoretical Geomorphology. John Wiley, Chichester, pp. 295–314.
226
Mediterranean Desertification
Kirkby MJ (1999) Translating models from hillslope (1 ha) to catchment (1000 km2 ) scales. In B Diekkr¨uger, MJ Kirkby and U Schr¨oder (eds) Regionalization in Hydrology. IAHS Publication 254, pp. 1–12. Kirkby MJ, Baird AJ, Lockwood JG, McMahon MD, Mitchell PJ, Shao J, Sheehy JE, Thornes JB and Woodward FI (1993) MEDALUS Final report. Kirkby MJ, Baird AJ, Diamond SM, Lockwood JG, McMahon ML, Mitchell PJ, Shao J, Sheehy JE, Thornes JB and Woodward FI (1996) The MEDALUS slope catena model: a physically based process model for hydrology, ecology and land degradation interactions. In JB Thornes and J Brandt (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 303–354. Leopold LB and Maddock T, Jr. (1953) The hydraulic geometry of stream channels and some physiographic implications. US Geological Survey Professional Paper 252, US Government Printing Office, Washington, DC. L´opez-Berm´udez F, Romero D´ıaz A and Mart´ınez-Fern´andez J (1996) The El Ardal field site: soil and vegetation cover. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 169–188. Lossaint P (1973) Soil–vegetation relationships in Mediterranean ecosystems of Southern France. In F di Castri and HA Mooney (eds) Mediterranean Type Ecosystems Origin and Structure, Chapman & Hall, London, pp. 199–210. Malanson GP and Trabaud L (1988) Vigour of post-fire resprouting by Quercus coccifera L. Journal of Ecology 76, 351–365. Margaris NS (1975) Effect of photoperiod on seasonal dimorphism of some Mediterranean plants. Berichte der Schweizerenischen Botanischen Gessellschaft 85, 96–102. Merino O, Martin MP, Martin A and Merino J (1990) Successional and temporal changes in primary productivity in two mediterranean scrub ecosystems. Acta Oecologia 11, 103–112. Mitchell PL (1997) Misuse of regression for empirical validation of models. Agricultural Systems 54, 313–326. Mitchell PL and Sheehy JE (1997) Comparison of predictions and observations to assess model performance: a method of empirical validation. Applications of Systems Approaches at the Field Level. Volume 2. Proceedings of the Second Annual Symposium on Systems Approaches for Agricultural Development, held at IRRI, Los Ba˜nos, Philippines, 6–8 December 1995 (eds Kropff MJ, Teng PS, Aggarwal PK, Bouma J, Bouman BAM, Jones JW, Van Laar HH), Kluwer Academic, Dordrecht, pp. 437–451. Monteith JL (1972) Solar radiation and productivity in tropical ecosystems. Journal of Applied Ecology 9, 747–766. Morison JIL (1985) Sensitivity of stomata and water use efficiency to high CO2 . Plant, Cell and Environment 8, 467–474. M¨uller MJ (1982) Selected Climatic Data for a Global Set of Standard Stations for Vegetation Science. Dr W. Junk Publishers, The Hague. Osborne CP and Woodward FI (2001) Biological mechanisms underlying recent increases in the NDVI of Mediterranean shrublands. International Journal of Remote Sensing 22, 1895–1907. Osborne CP, Mitchell PL, Sheehy JE and Woodward FI (2000) Modelling the recent historical impacts of atmospheric CO2 and climate change on Mediterranean vegetation. Global Change Biology 6, 445–458. Paraskevopoulos SP, Iatrou GD and Pantis JD (1994) Plant growth strategies in evergreen-sclerophyllous shrublands (Maquis) in central Greece. Vegetatio 115, 109–114. Pitacco A, Gallinaro N and Giulivo C (1992) Evaluation of the actual evapotranspiration of a Quercus ilex L. stand by the Bowen ratio–Energy Budget method. Vegetatio 99–100, 163–168. Pugnaire FI and Haase P (1996) Comparative phenology and growth of two perennial tussock grass species in a semi-arid environment. Annals of Botany 77, 81–86. Pugnaire FI, Haase P, Incoll LD and Clark SC (1996) Response of the tussock grass Stipa tenacissima to watering in a semi-arid environment. Functional Ecology 10, 265–274. Puigdef´abregas J, Alonso JM, Delgado L, Domingo F, Cueto M, Guti´errez L, L´azaro R, Nicolau JM, S´anchez G, Sol´e A, Videl S, Aguilera C, Brenner A, Clark SC and Incoll LD (1996) The Rambla Honda field site: interactions of soil and vegetation along a catena in semi-arid southeast Spain. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 137–168. Quinn P, Beven KJ, Chevallier P and Planchon O (1991) The prediction of hillslope flow paths for distributed hydrological modelling using digital terrain modes. Hydrological Processes 5, 59–79. Rambal S (1993) The differential role of mechanisms for drought resistance in a Mediterranean evergreen shrub – a simulation approach. Plant, Cell and Environment 16, 35. Rambal S and Leterme J (1987) Changes in aboveground structure and resistances to water uptake in Quercus coccifera along a rainfall gradient. In JD Tenhunen, FM Catarino, OL Lange and WC Oechel (eds) Plant Response to Stress. Functional Analysis in Mediterranean Ecosystems. Springer-Verlag, Berlin and Heidelberg, pp. 191–200.
MEDRUSH: A Basin-Scale Physically Based Model
227
Rambal S, Damesin C, Joffre R, M´ethy M and Lo Seen D (1996) Optimization of carbon gain in canopies of Mediterranean evergreen oaks. Annales des Sciences Foresti`eres 53, 547–560. Rapp M and Lossaint P (1981) Some aspects of mineral cycling in the garrigue of southern France. In F di Castri, DW Goodall and RL Specht (eds) Ecosystems of the World 11. Mediterranean-type Shrublands. Elsevier Scientific, Amsterdam, pp. 289–301. Scarascia-Mugnozza GE, De Angelis P, Matteucci G and Valentini R (1996) Long-term exposure to elevated [CO2 ] in a natural Quercus ilex L. community: net photosynthesis and photochemical efficiency of PS II at different levels of water stress. Plant, Cell and Environment 19, 643–654. Shuttleworth WJ and Gurney RJ (1989) The theoretical relationship between foliage temperature and canopy resistance in sparse crops. Quarterly Journal of the Royal Meteorological Society 116, 497–519. Shuttleworth WJ and Wallace JS (1985) Evaporation from sparse crops – an energy combination theory. Quarterly Journal of the Royal Meteorological Society 111, 839–855. Smith SD, Monson RK and Anderson JE (1997) Physiological Ecology of North American Desert Plants. Springer-Verlag, Berlin and Heidelberg. Smith TM, Shugart HH, Woodward FI and Burton PJ (1993) Plant functional types. In AM Solomon and HH Shugart (eds) Vegetation Dynamics and Global Change. Chapman & Hall, London, pp. 272–292. Smith TR and Bretherton FP (1972) Stability and the conservation of mass in drainage basin evolution. Water Resources Research 8(6), 1506–1529. Specht RL (1969) A comparison of the sclerophyllous vegetation characteristic of mediterranean type climates in France, California, and southern Australia. II. Dry matter, energy, and nutrient accumulation. Australian Journal of Botany 17, 293–308. Specht RL (1988) Mediterranean-Type Ecosystems: A Data Source Book. Kluwer Academic, Dordrecht. Thornley JHM (1970) Respiration, growth and maintenance in plants. Nature 227, 304–305. Trabaud L (1991) Fire regimes and phytomass growth dynamics in a Quercus coccifera garrigue. Journal of Vegetation Science 2, 307–314. Tsiouvaras CN (1988) Long-term effects of clipping on production and vigor of Kermes Oak (Quercus coccifera). Forest Ecology and Management 24, 159–166. Valentini R, Scarascia-Mugnozza GE, De Angelis P and Bimbi R (1991) An experimental test of the eddy correlation technique over a Mediterranean macchia canopy. Plant, Cell and Environment 14, 987–994. Wilks DS, Wolfe DW and Riha SJ (1995) Simple carbon assimilation response functions from atmospheric CO2 , and daily temperature and shortwave radiation. Global Change Biology 1, 337–346. Woodward FI and Osborne CP (2000) The representation of root processes in models addressing the responses of vegetation to global change. New Phytologist 147, 223–232. Yair A and Lavee H (1982) Factors affecting the spatial variability of runoff generation over arid hillslopes, southern Israel. Israel Journal of Earth Sciences 31, 133–143.
PART 2
REGIONAL STUDIES
Section VI
The Guadalent´ın Basin, South-east Spain
17
Natural Resources in the Guadalent´ın Basin (South-east Spain): Water as a Key Factor
1 ´ ´ ´ 2 F. ALONSO-SARRIA ´ 1 AND F. LOPEZ-BERM UDEZ, G.G. BARBERA, 1 F. BELMONTE SERRATO
1 2
Laboratorio de Geomorfolog´ıa, Universidad de Murcia, Spain CEBAS-CSIC, Campus Universitario de Espinardo, Murcia, Spain
1 INTRODUCTION The Guadalent´ın Basin is located in south-eastern Spain, and covers an area of 3300 km2 (see Plate 2 in the colour plate section). The climate is semi-arid, this being one of the driest areas of Europe, with high inter-annual variability in rainfall. The Guadalent´ın River has an extremely irregular flow, which can change within hours from a dry channel to catastrophic floods. The relief is variable: there are two wide plains surrounded by mountains reaching 500–2100 m. Natural vegetation is severely limited by climate, and most of the semi-natural ecosystems are shrublands of diverse types, although in the mountains there are Pinus halepensis forests. Desertification is a complex set of processes that results in degradation of the land, with a loss of productive value. Much attention has been paid to the local causes and effects of these processes, such as deforestation, overgrazing and soil erosion. There are also off-site effects of these primary processes, such as changes in the hydrological dynamics of channels, floods and sedimentation. However, less attention has been paid to the global relationship between the development of socioeconomic systems and the progress of desertification. In this chapter such a relationship is studied in the Guadalent´ın Basin, by attempting to synthesize some relevant aspects of environmental conditions, the constrictions imposed on the historic development of socio-economic systems, and how both aspects affect the natural resources of the basin. The relationship between the human society and its environment is reflected in the land use system. Here we introduce the ways in which this system has been modelled through time, and the key factors in these processes. The objective has been to isolate the most relevant aspects in order to characterize the main problems associated with desertification in the basin. Socio-economic systems and natural systems have strong links that interact with and influence each other. The way in which socio-economic systems have evolved has been closely related to environmental factors, but also according to changing politics and technology. Historical changes in land use in this basin are discussed in relation to climate and the exploitation of water resources.
2 THE ADVERSE CLIMATIC CONDITIONS OF A SEMI-ARID BASIN In the Guadalent´ın Basin aridity is constant, with less than 300 mm of annual rainfall and 900–1000 mm of potential evapotranspiration (PET) throughout most of the territory. The extreme variability in rainfall is characterized by long periods of drought and sudden extreme torrential precipitation events causing soil erosion. Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
234
Mediterranean Desertification
In spite of this generally semi-arid environment, mostly with an annual deficit of more than 300 mm, the Guadalent´ın Basin, like many of the larger Mediterranean semi-arid basins, has a sub-humid sector at higher altitude which provides most of the water supply for the basin. With regard to this variability, the Guadalent´ın Basin can be divided into three sectors (L´opez Berm´udez et al. 1998), which are outlined in Figure 17.1. • • •
The western sub-humid sector in the highest areas of Sierra de Mar´ıa (2000 m a.s.l.) has a positive water balance (negative hydraulic deficit). This area covers 0.4% of the territory. The higher relief of the northern sector (Sierra Espu˜na and Sierra del Cambr´on, 1400–1500 m a.s.l.) and the upper part of the basin upstream of Puentes Reservoir support a semi-arid sector with hydraulic deficits lower than 300 mm year−1 . These areas cover 16.8% of the territory. The largest sector is arid, with annual hydraulic deficits exceeding 300 mm. This sector includes the eastern and south-eastern sectors and most of the valley floor, covering 82.8% of the territory.
The temporal variability of rainfall totals is extreme. This variability produces a characteristic alternation between humid and dry periods that is best studied through a long time series, such as the Murcia 1862–1997 series (Figure 17.2). Drought periods can be considered as climatic hazards
4200
Deficit > 300 mm year −1
4190 Deficit < 300 mm year −1
4180 4170
Positive balance
4160 560 570 580 590 600 610 620 630 640 650 660
Figure 17.1 Water balance in the Guadalent´ın Basin. The scales on the axes are UTM (Universe Transverse Mercator) coordinates
800
Rainfall (mm)
600
400
200
0 1860 1870 1880
1890
1900
1910
1920
1930 1940
1950
1960 1970
1980
1990
Year Annual rainfall
Moving average (5 yr)
Trend
Figure 17.2 Annual precipitation for Murcia (1862–1997), showing the decreasing trend
235
Natural Resources in the Guadalent´ın Basin Table 17.1 Basic climatic characteristics of the Guadalent´ın Basin
Station Alcantarilla Aledo Alhama de Murcia Do˜na In´es Librilla Lorca Puerto Lumbreras Puentes Totana Zarzilla de Ramos Zarzadilla de Totana Mar´ıa Topares V´elez Rubio
H
YP
D
Tav
Tmax
Tmin
PET
72 620 760 786 168 335 465 450 225 652 861 1200 1192 838
321 528 448 329 343 261 295 265 259 550 359 391 408 391
69 41 45 26 50 38 46 36 67 31 23 67 30 43
17.3 14.5 15.0 14.1 18.5 18.1 17.2 17.8 17.3 16.1 17.7 11.4 12.5 13.5
23.9 19.7 19.7 21.0 25.0 24.5 24.4 24.7 22.7 24.2 22.9 17.1 16.8 19.3
10.7 10.2 10.2 7.2 12.0 11.8 10.0 10.9 11.9 8.0 12.5 5.8 8.2 7.7
904 795 795 774 981 837 899 942 894 843 932 669 713 728
H , height (m a.s.l.); Y P , annual precipitation (mm); D, precipitation days; Tav , average temperature (◦ C); Tmax , average maximum temperature (◦ C); Tmin , average minimum temperature (◦ C); PET, potential evapotranspiration (mm). After Garc´ıa de Pedraza and Reija Garrido (1994). with very fuzzy spatio-temporal limits, because of (a) their inherent variability and (b) the subjectivity of their evaluation. In this century, a large number of drought periods have occurred. The most remarkable were those of 1911–1913, 1925–1928, 1934–1941, 1944–1945, 1952–1953, 1955–1956, 1963–1964, 1978–1979, 1981–1984 and 1993–1995. There has been a clear trend of decreasing annual rainfall during the period monitored. Table 17.1 provides temperature and rainfall information for the main weather stations in the Guadalent´ın. Data were available to estimate average precipitation volume per day. However, daily precipitation has a far from Gaussian distribution, and it is better fitted to a General Extreme Values distribution (Alonso-Sarr´ıa 1995). This means that a high percentage of the yearly rainfall falls during a few very intense events. These high-intensity rainfall events can occur in either humid or dry periods. The main cause of a drought period is a low circulation index for the middle troposphere, causing the eruption of low pressure cells in the Mediterranean that result in high-intensity rainfall events. Alternatively, local breeze and valley winds, originating due to land–sea pressure gradients, favour the formation of warm and humid air masses that can rise as they reach higher relief, and release precipitation (Mill´an and Estrela 1994). The combination of these processes produces high-intensity rainfall events. With these climatic conditions, and taking into account the relief characteristics of the Guadalent´ın Basin, the hydrology is dominated by channels that are usually dry, but experiencing episodic flash floods. The specific drainage network of the Guadalent´ın Basin has a considerable influence on the origin and paths of extreme flash floods due to its morpho-structural configuration, with high relief and steep slopes, in the tributary sub-basins and a flat and subsident valley floor. A high number of flash flood events have been recorded at the Puentes Reservoir (located in the centre of the Guadalent´ın River headwaters). The most severe incidents in the last 200 years were those of 1802, 1830, 1831, 1838, 1846, 1860, 1943, 1948, 1973 and 1982. In the 20th century, 23 significant flood events were recorded in the Guadalent´ın Basin.
3 IRRIGATION AS AN EARLY RESOURCE TO OVERCOME ARIDITY The combination of low precipitation, high potential evapotranspiration, and the infrequency of rainfall events, seriously constrains the ecosystem productivity in the Guadalent´ın Basin. It is clear
236
Mediterranean Desertification
that even under undisturbed conditions, primary production is naturally low, limiting the development of the most structurally complex terrestrial ecosystems such as forests. Aridity is a major factor for these ecosystems and has strongly influenced the way in which human societies have exploited the land. In turn, the environment has influenced the human population evolution. The spatial distribution of the present land uses (Figure 17.3(a)–(d)) is a result of the interaction between environmental conditions, human evolution and technological capabilities and advances. It can be observed from Figure 17.3(c) that the main area covered by irrigated crops is located in a belt in the centre of the basin, following an E–W direction. This corresponds to the valley bottom along the Guadalent´ın River. The main land uses at present include huerta (a Spanish word to name a complex of herbaceous crops of different species including little orchards of irrigated trees), citrus trees, almond trees, and greenhouses. Dry crops are grown near the borders of irrigated land and on the plains surrounding the valley bottom, although in the southern part of the basin they can also be grown on mountains composed of soft siliceous metamorphic rocks. The main crops are barley and almond trees. Finally, seminatural ecosystems are mainly located in the mountains and their margins. Shrublands are more extensive than forests. The former are complex but most are dominated by Stipa tenacissima (a perennial tussock herb), Rosmarinus officinalis and Anthyllis cytisoides. The distribution of Stipa is deeply influenced by humans as it was used for fibre production for centuries. Forests are dominated by Pinus halepensis, in part as a result of afforestation policies over the last 150 years. The Guadalent´ın Basin has been exploited by humans for many years, particularly with the early development of agriculture and pastoralism in Neolithic times, about 5000 BP (Walker 1979; Camel-Avila 1998; L´opez-Berm´udez and Mariscal 1996). In the context of the naturally low ecosystem productivity, the use of water as a basic production factor appeared very early in the human exploitation of the basin. For example, in the Argar culture (about 3500 BP), an autochthonous culture of south-eastern Spain in the Bronze Age, two types of villages are found in the basin: lowland and highland villages. In lowland villages irrigation schemes were a major characteristic, although alongside other activities. In highland villages the people seemed to rely more on hunting and stockbreeding for their livelihood (Ayala Juan 1991). The extensive use of irrigation in the Argar period is even more remarkable if one takes into account that there is evidence that the climate was once more humid than at present (Ayala Juan 1991). Irrigation is a characteristic that has been maintained throughout the history of the basin, but continuously increased. It is well known that in Roman and Arabic times irrigation schemes were widely improved and enlarged. After the Christian Reconquest, in the 13th century, most of the agriculture was concentrated close to the town of Lorca, mainly on areas irrigated by the Guadalent´ın River. The total ploughed area (irrigated and non-irrigated) is estimated to have been about 10 000 ha in the first third of the 14th century (Torres Fontes 1994). During the 13th and 14th centuries, the territory of Lorca was virtually free of agriculture and nearly uninhabited, except in a small area surrounding the town. Stock-breeding was the most important occupation. During the 16th and 17th centuries, once the last Muslim kingdom of Granada had been reconquered in 1492, there was a slow expansion of agriculture, mainly in the valley close to Lorca town and the nearby hillslopes, where people took advantage of the irregular irrigation from boqueras. In 1550, the population of the central part of the Guadalent´ın Basin was estimated at 8000 inhabitants (Figure 17.4), about 4 people km−2 , 92% of whom were employed in agriculture (Lemeunier 1980). Agricultural production in 1550 and 1750, as an indicator of change in land use during that period, is shown in Table 17.2. Land use was changing to a subsistence economy based on cereal and on sheep products. The distribution of land turned over to dryland cereal production affected the land available for grazing, and semi-natural ecosystems became important for pasture. By 1635, irrigated lands were estimated to cover about 9150 ha, and these sustained most of the agricultural production. By 1694, the irrigated area was estimated to be 10 000 ha (Gil Olcina 1980), so the increase in the
Natural Resources in the Guadalent´ın Basin
237
(a)
Forests
0− 5% 5− 20% 20− 40% 40− 60% 60− 80% 80−100%
(b)
Shrublands
0 −5% 5− 20% 20− 40% 40−60% 60−80% 80−100%
Figure 17.3 Distribution of the main land use types over the Guadalent´ın Basin, 1996: (a) forests; (b) shrublands; (c) irrigated crops; (d) dryland crops
area of irrigated lands was slow but continuous. However, without further technology, only basic channel building for irregular, or even ephemeral, flow diversion towards the fields was possible, and this kind of development of food production and subsequent human population increase was severely limited (Figure 17.4).
238
Mediterranean Desertification
(c)
Irrigated crops
(d)
Dryland crops
0 −5% 5 −20% 20−40% 40−60% 60−80% 80−100%
0 −5% 5 − 20% 20−40% 40−60% 60−80% 80−100%
Figure 17.3 (Continued)
4
EXPANSION OF AGRICULTURE
In the previous section it was shown that irrigation has been an important factor since prehistoric times in determining land use in the Guadalent´ın Basin. Lack of technological knowledge limited the extent of irrigation, but the low population density was also due to political factors. Then came the chance to expand irrigation into areas not previously used for agriculture. The 18th century marked a huge change in land use around Lorca. Crown lands were sold to the town, and then they were progressively privatized and ploughed. Between 1700 and 1775 the increase in the area of agricultural land was relatively slow, but between 1775 and 1790 the process was accelerated. Between 1771 and 1807, it is estimated that 33 000 ha were transformed into agricultural
239
Natural Resources in the Guadalent´ın Basin 80
Population (× 1000)
70 60 50 40 30 20 10
0 1500
1550
1600
1650
1700
1750
1800
1850
1900
Year
Figure 17.4 Evolution of the human population in the central sector of the Guadalent´ın Basin (1500–1900). Sharp increases occurred, first when Crown lands were sold off, and later, when cereal production became dominant
Table 17.2 Relative value of agricultural production in the central sector of the Guadalent´ın Basin in the 16th, 18th and 20th centuries
Product
1550
1750
1997
Cereal Legumes Vegetables Fruits Wine Olive oil Barrillaa Wool and meat Other
67% – <1% <1% 5% 4% – 13% 11%
82% 1% <1% <1% <1% 2% 3% 9% 4%
1% <1% 35% 23% 3% <1% – 28% 10%
a
Barrilla are halophitic plants (mainly Halogeton sativus). The burned ashes were used as raw material for soap production. It was an industrial crop of great importance in the 17th century right across the Guadalent´ın Basin. 1997 figures are based on estimates for the whole region of Murcia. After Lemeunier (1980) and Centro de Estad´ıstica Regional de Murcia (1999).
land (Gil Olcina 1971, 1980), a figure representing 15% of the Lorca territory in those times. Most of these new agricultural lands were drylands planted with cereal. This new direction of the economy is easily observed from figures for agricultural production in Table 17.2. Cereal production became more dominant as new lands were ploughed, and sheep breeding was in turn reduced as the area used for pasture decreased. Unfortunately, forest and esparto production at that time is unknown, as taxes were not collected on these products, thus preventing a more accurate estimation of the regression of forests and other semi-natural ecosystems. In the 18th century there was a sudden increase in population (Figure 17.4). This increase cannot be proved to be a direct result of agricultural expansion, but there was no doubt a steady increase in population as agriculture grew. For example, over 80% of the population worked in agriculture
240
Mediterranean Desertification
in 1750, and 84% of the agricultural production was locally consumed at the onset of 19th century (Lemeunier 1980). This indicates a subsistence economy. This accelerated land-use change from semi-natural ecosystems to dryland agriculture implied the ploughing of land of marginal quality. A clear indicator of this process was the decrease of cereal productivity in the drylands. Classifying them in three quality classes, between 1771 and 1819 the average productivity of first-class land decreased by 60%, second-class land by 62.5% and thirdclass land by 80% (P´erez Picazo 1980). There may be other socio-economic factors involved, and more detailed research on the topic is needed, but it seems clear that the relatively small change in human population seen in the first half of the 19th century (Figure 17.4) could have been strongly influenced by the change in land use which resulted in a decrease in productivity. At the end of the 18th century the distribution of land use was 5% irrigated land, 50% dryland, and 45% pasture, shrubland and forest (Lemeunier 1980). It is difficult to say how accurate these figures are as the actual area of public land (covered by semi-natural ecosystems) was not clearly known by local authorities in 1755 (Gil Olcina 1990). However, as the relative proportion of irrigated and dryland crops seems quite accurate, it is possible to state that the increase of land exploitation in the 18th century was based in the drylands. It is interesting to look at the distribution of crops in irrigated and drylands (Table 17.3). Cereal production was completely dominant in both farm systems, pointing to the lack of a clear specialization of production. At the end of the 18th century the Enlightenment Reforming Policy and the development of new technology led to the building of the Valdeinfierno and Puentes dams on the Guadalent´ın River. The dimensions of these dams were far greater than those of any other dam in Europe at the time. Both dams were almost finished in 1795 (Mula G´omez et al. 1986), but there was little opportunity to expand the area of irrigated land before the Puentes Dam burst in 1802, causing 600 deaths in Lorca town. This catastrophic event put a stop to dam building in the Mediterranean for several decades, until both dams were rebuilt in 1884 and 1897 (Bautista Mart´ın and Mu˜noz Bravo 1986). In spite of this early failure in the expansion of irrigated land, the ploughing of drylands continued. In the second half of the 19th century more municipality property was privatized according to new desamortizaci´on laws. In this way, 28 560 ha of the town lands were sold in 1855. Again, most of the land was used for cereal crops (Table 17.4). The important change here was from subsistence farming to growing fruit and vegetables for the commercial market. After this period, it seems that most land suitable for agriculture was in use. The reduction in the area of the semi-natural ecosystems used for Table 17.3 Relative area covered by crops in irrigated and drylands in the mid-18th century in the central sector of the Guadalent´ın Basin
Land use
Irrigated land
Dryland
87.6% 4.0% 3.8% 3.0% 1.4% 0.2%
98.7% 0.3% 0.1% 0.7% 0.2% 0%
Cereals Olives Mulberries Vineyards Fruit trees Vegetables
Table 17.4 Changes in ploughed area by crop in the 19th century in the central sector of the Guadalent´ın Basin (ha)
Year
Cereals
Olives
Vineyards
Fruit trees
Vegetables
1822 1861
109 982 162 617
1098 ?
1027 7560
1890 7953
218
241
Natural Resources in the Guadalent´ın Basin
pasture is clearly indicated by the reduction in the number of sheep from 56 018 to 16 514 between 1865 and 1914 (Mart´ınez Carri´on 1991). This time period was also marked by a big increase in the human population (Figure 17.4). This can be attributed mostly to new mining enterprises, but perhaps also to the new wave of dryland farming development.
5 EXHAUSTIVE USE OF WATER RESOURCES Changes in land use in the 20th century in the Guadalent´ın Basin are examined in Chapter 5. The most important change has been the exhaustive use of water resources, associated with three main processes: (i) definitive regulation of surface resources; (ii) exploitation of groundwater; (iii) the use of resources from far afield. Each of these processes is associated with problems for the sustainable and reliable long-term use of water resources. 5.1 Reservoirs
The rebuilding of the Puentes and Valdeinfierno dams at the end of the 19th century allowed complete regulation of Guadalent´ın headwater, and therefore made irrigation of the traditional irrigated lands of the valley more reliable, resulting in some expansion of permanently irrigated lands. This was supposed to allow farmers to reach the upper limit in the possible use of local surface water resources. However, the reservoir capacity is not constant but has been progressively reduced by rapid sediment accumulation. Over the last 100 years, the storage capacity of both reservoirs has been severely reduced by sedimentation (Table 17.5 and Figure 17.5), caused by flash floods. Within 90 years the capacity of the reservoirs decreased by more than 50%. The building of a new dam at the Puentes reservoir has allowed the accumulation of sediment to be studied in detail. One method used to estimate the erosion in a basin is the determination of the basin-specific degradation (Fournier 1960) from the silts accumulated in a reservoir during a set period of time (ICOLD 1989). This is done by means of bathymetric and sedimentological techniques, and the calculation of parameters including grain size, density of silts, retention capacity of the reservoir, exploitation of the reservoir, and photogrammetry (Avenda˜no et al. 1995). The volume of silt transported by overland flow depends on the intensity and duration of rainfall, slopes, lithology, previous mobilization of silt associated with cultivation techniques, and general land use. Accumulated silts in the Puentes reservoir, since 1889, are now thicker than 30 m, particularly at the end of the reservoir furthest from the dam. Dating of the silts using radionucleid Cs137 (Cobo Ray´an et al. 1996) indicates that the sedimentation rates have varied with time as a function of the Table 17.5 Soil erosion and annual and total reservoir capacity losses in the Guadalent´ın Basin
Reservoir
Puentes
Valdeinfierno
Date of construction Basin area (km2 ) Estimateda gross erosion (t ha−1 year−1 ) Sediment production rate (t ha−1 year−1 ) Sediment delivery ratio (t ha−1 year−1 ) Loss of capacity total (Hm3 ) % per year (Hm3 ) %
1884 1042 30.2 2.65 0.088
1897 427 30.62 2.79 0.091
18.7 59.3 0.19 0.59
11.2 44.6 0.07 0.29
a
Average gross erosion of each basin was estimated by USLE. After Cabezas et al. (1992) and Sanz Montero et al. (1998).
242
Mediterranean Desertification
Figure 17.5 The accumulation of sediments in the reservoirs of the Guadalent´ın Basin has reduced their capacity by 50% in 90 years
rainfall regime, land-use changes and changes in reservoir exploitation. The Cs137 cannot be detected below a depth of approximately 3 m. This is roughly the thickness of silts that accumulated between 1954 and 1994. From the analysis of recent sedimentary series, three distinct periods were detected. Between 1954 and 1972, around 120 cm of silt was deposited (an average sedimentation rate of 6.7 cm year−1 ). Between 1972 and 1977, another 120 cm of silt was deposited (an increase to 24 cm year−1 ). The period between 1977 and 1994 showed an average silt thickness of 30 cm (a rate of 1.8 cm year−1 ) (Cobo Ray´an et al. 1996). Changes in sedimentation rates, as well as the sediment lithology, are due to variations in the speed of soil erosion in the basin and to the water retention capacity in the reservoir, which in turn depends on the degree of exploitation of the reservoir. Not all the silts that arrive at the reservoir are retained, however. A high percentage of the silt is carried away through drainage pipes for irrigation. Therefore real sedimentation rates are greater than those that have been measured. A new dam has recently been built at Puentes in order to improve water storage capacity, but, of course, sedimentation processes never stop. 5.2 Groundwater With the complete regulation of the Guadalent´ın headwaters, the next possibility for providing new water resources was from groundwater. In the 1950s and 1960s the availability of new submersible pumps and a reliable electric power supply, as well as a general improvement in the economy, led to an increase in irrigated land. In the Guadalent´ın Basin there are five hydrogeological units (Senent Alonso and Arag´on Rueda 1995). These are, in order of importance, the Guadalent´ın Valley, Valdeinfierno, Carrascoy, Y´echar and Orce-Mar´ıa. Their aggregated area is about 1200 km2 (35% of the basin area). These units are in different hydrogeological domains, having different characteristics related to the stratigraphy and tectonics of their geological formations. In the Guadalent´ın Valley unit there are two aquifer formations. The Upper (235 km2 ) and Lower Guadalent´ın (562 km2 ) have different permeabilities and piezometric levels, which are frequently independent. At present, the depth to groundwater ranges from 20 m to more than 200 m. Another unit is the Valdeinfierno, about 250 km2 , where the depth to water is between 0 and 100 m, while the Orce-Mar´ıa (shared with Guadalquivir Basin) is only 20 km2 . These units are in the headwaters regulated by the Valdeinfierno–Puentes reservoir system. For the marginal units of Y´echar (112 km2 ) and Carrascoy (27 km2 ), depths to groundwater are around 62–234 m and 30–150 m respectively.
243
Natural Resources in the Guadalent´ın Basin
Table 17.6 Some parameters of the main hydrogeological units of the Guadalent´ın Basin
Hydrogeological unit Guadalent´ın Valley Yechar Carrascoy Valdeinfierno
Input (Hm3 )
Natural output (Hm3 )
Pumping (Hm3 )
23.3 0.4 3.6 8.7
0 0 0 7.1
92 3.5 5.6 1.6
Total output (Hm3 ) 92 3.5 5.6 8.7
Deficit (Hm3 )
Reserves (Hm3 )
Theoretical horizon (years)
68.7 3.1 2 0
6000 50 300 1230
87 16 150 –
After Consejo Econ´omico y Social de Murcia (1995). The Guadalent´ın Valley, Y´echar and Carrascoy are, at present, severely over-exploited units (Table 17.6). The theoretical horizon or exhaustion limit may be calculated as the reserve/deficit ratio. However, the actual time period to exhaustion is much less, because of associated problems of salinization and the economic cost of pumping water from greater depths. Over-exploitation has brought many problems, mainly salinization, but also pollution due to increases in industrial waste, municipal waste, and intensive pig breeding in the valley. Groundwater over-exploitation is, historically, very recent but the ecological effects have been felt very quickly. The upper valley aquifer system discharges near the towns of Alhama and Totana where the natural ecosystem includes halophytic vegetation on saline soils (saladares) around several saline ponds. Recent sharp falls in piezometric levels have reduced the extent of these unique ecosystems. The present depth of piezometric levels in the main aquifer in the Guadalent´ın Valley suggests that recharge hardly exceeds 0, and that sufficient water from the surface no longer infiltrates back to replenish the groundwater. 5.3 The Tagus–Segura Canal
Since all locally available water resources were in use or over-exploited, in 1979 the Tagus–Segura Canal was opened (Nicol´as Mart´ınez 1995). This is an infrastructure devoted to diverting water from the Tagus headwaters (in central Spain) towards the south-east. Part of these resources came to the Guadalent´ın Basin, promoting the development of new irrigated lands in the valley. The canal is legally authorized to divert up to 600 Hm3 from the Tagus to the Segura. However, the recurrent droughts of the 1980s and 1990s only allowed the transfer of half that volume. In the severe drought of the early 1990s, with a critical situation in both basins, there were inter-regional disputes about the transfer of an emergency flow to save irrigated lands in the south-east. At present, the situation is being discussed in the framework of the National Hydrology Plan, approved by the Spanish government in 2002, with suggestions for new canals from the Ebro river in northern Spain. However, inter-regional tensions still have to be resolved, and the final result is not clear.
6 DISCUSSION: WATER AS THE KEY RESOURCE In the semi-arid Guadalent´ın Basin, water has always been an important factor in socio-economic development. However, in the 20th century the situation became more critical. Detailed studies of recent land-use changes in the Guadalent´ın have shown that land-use systems have changed in two ways (Barber´a et al. 1997). The area of irrigated land has spread as new water resources have been exploited so that its share in the value of agricultural production has risen to 90%. At the same time drylands and semi-natural ecosystems have become more and more marginal from an economic point of view. Crops in drylands are increasingly maintained by subsidies, even if production has become almost unprofitable. In the 18th and 19th centuries the most important feature of land-use change was that semi-natural ecosystems were being brought into production for dryland agriculture. In the 20th century a limit to this was reached because no more suitable land was available, and because
244
Mediterranean Desertification
transport costs and other economic reasons no longer allowed cereal crops to be produced here at a competitive price. Now the exploitation pressure on the semi-natural ecosystems is decreasing as rural depopulation proceeds and alternative sources of energy and materials are provided by modern technology. In summary, from the point of view of land degradation, the present land-use model is one where semi-natural ecosystems are mostly stable or even slowly recovering from former exploitation. The area of existing drylands, and also some semi-natural vegetation, has been encroached on by the increased area of irrigated lands (Barber´a et al. 1997), provoking site degradation, by massive clearing and levelling, and over-exploitation of the basic resource, water. The present land-use system (Figure 17.3(a)–(d)) has stabilized, after the rapid changes during the last 300 years. Dryland crops are expected to be further reduced, and the area of irrigated land, used for high-value crops, cannot increase where water resources are already over-exploited. If no new water sources are brought to the basin from outside, a progressive abandonment of many of the irrigated farms is expected, as the seriously depleted groundwater resources are non-renewable in the present climate. In terms of sustainable agriculture, a solution must be found to conserve water in terms of quantity and quality. Natural rainfall cannot be relied upon as it is scarce and irregular. Therefore, water can be only managed for agriculture by building dams and diverting the flows toward the crops. However, if the flow is diverted the ecosystem of the river and surrounding land which once depended on that water is seriously affected. From the perspective of the scientist, the only reasonable way to exploit groundwater and at the same time maintain its quality and quantity is to use the aquifer as a “buffer” for drought periods. Thus, during drought periods more water is pumped to support a stable level of water supply, and in the humid years pumping is reduced to allow recharge. Also, with this method of managed exploitation the conservation of ecosystems associated with discharge areas can be satisfactorily achieved. Unfortunately, in the semi-arid Mediterranean climate, where farmers are trying to make a living, water conservation is at odds with market forces. In conclusion, water is the key factor required to sustain both the human society and the natural environment in the basin. Lack of water is the main environmental problem to be solved, but the dynamics of the economy work against finding a long-term solution. External political regulation is necessary but unpopular, as it interferes with the short-term economy. However, unless steps are taken now, the present rate of water use will continue the trend towards degradation and complete water resource exhaustion. Water supply is the main factor governing the future of both the environment and human society in the Guadalent´ın Basin. The choice of management practices for the drylands and the semi-natural ecosystems today may be critical when considering long-term consequences. Much of the land degradation in the area occurred in the 18th and 19th centuries, and although such land could be recovered, it would not be economical to do so at present. The scale of water consumption on the irrigated lands cannot be sustained, as the climate does not provide sufficient rainfall and the exploitation of the aquifers has reached unacceptable depths. Plans to divert water from wetter parts of Spain are both expensive and difficult, causing tension between regions. Irrigated horticulture to provide high-value produce has been profitable for farmers on a short-term basis, but it is clear that there is a price to be paid, and without restrictions there will not be sufficient water available for farmers to make a living at all.
REFERENCES Alonso-Sarria F (1995) Estudio del Alto Guadalent´ın desde la Perspectiva Econ´omica de la Gesti´on del Agua Subterr´anea. Caja de Ahorros del Mediterr´aneo, Murcia. Avenda˜no C, Cobo R, G´omez Monta˜na JL and Sanz Montero ME (1995) Procedimiento para evaluar la degradaci´on espec´ıfica (erosi´on) de cuenca de embalses a partir de los sedimentos acumulados en los mismos. Aplicci´on al estudio de embalases espa˜noles. Ingenieria Civil 99, 51–58.
Natural Resources in the Guadalent´ın Basin
245
Ayala Juan MM (1991) El poblamiento arg´arico en Lorca: estado de la cuesti´on. Real Academia Alfonso X El Sabio, Murcia. Barber´a GG, L´opez-Berm´udez F and Romero D´ıaz MA (1997) Cambios de uso del suelo y desertificaci´on en el Mediterr´aneo: El caso del Sureste Ib´erico. In JM Garc´ıa Ruiz and P L´opez Garc´ıa (eds) Acci´on Humana y Desertificaci´on en Ambientes Mediterr´aneos. Instituto Pirenaico de Ecolog´ıa, Zaragoza, pp. 9–39. Bautista Mart´ın J and Mu˜noz Barvo J (1986) Las Presas del Estrecho de Puentes. Confederaci´on Hidrogr´afica del Segura, Murcia. Cabezas F, Romero D´ıaz MA and L´opez-Berm´udez F (1992) Erosion and fluvial sedimentation in the river Segura Basin (Spain). Catena 19, 379–392. Camel-Avila M (1998) G´eomorphogen`ese Holoc`ene dans le Bas-Guadalent´ın. Th´ese, Universit´e de Pau et des Pays de L’Adour, UFR de Sciences Humaines. Centro de Estad´ıstica Regional de Murcia (1999) Anuario Estad´ıstico de la Regi´on de Murcia. Consejer´ıa de Econom´ıa y Hacienda, Murcia. Cobo Ray´an R, G´omez Monta˜na JL, Plata Bedmar A, Sanz Montero ME and Avenda˜no C (1996) Influencia del r´egimen de explotaci´on del embalse de Puentes en las tasas de sedimentaci´on durante el per´ıodo 1954–1994. V Jornadas Espa˜nolas de Presas. Correcci´on hidrol´ogica de Cuencas y Aterramiento de Embalses: 155–169. Consejo Econ´omico y Social de Murcia (1995) Recursos H´ıdricos y su Importancia en el Desarrollo de la Regi´on de Murcia. Consejo Econ´omico y Social de La Regi´on de Murcia (Ed), Murcia. Fournier F (1960) Climat et erosion. La relation entre l’erosion du sol par l’eau et les pr´ecipitations atmosph´eriques. Press Universitaires de France, Paris. Garc´ıa de Pedraza L and Reija Garrido A (1994) Tiempo y clima en Espa˜na. Meteorolog´ıa de las Autonom´ıas. Dossat, Madrid. Gil Olcina A (1971) El Campo de Lorca. Estudio de Geograf´ıa Agraria. Instituto Juan Sebasti´an Elcano, Valencia. Gil Olcina A (1980) Configuraci´on de la estructura de la propiedad del suelo en el Campo de Lorca. In M Mu˜noz Barber´an (ed.) Ciclo de Temas Lorquinos. Caja de Ahorros de Alicante y Murcia, Murcia, pp. 21–42. Gil Olcina A (1990) Lorca 1755: seg´un las Respuestas Generales del Catastro de Ensenada. Ministerio de Econom´ıa y Hacienda, Madrid. ICOLD (International Commission on Large Dams) (1989) Sedimentation Control of Reservoirs. Bulletin 67. Lemeunier G (1980) Lorca del XVI al XVIII. Introducci´on a una ciudad del Sureste. In M Mu˜noz Barber´an (ed.) Ciclo de Temas Lorquinos. Caja de Ahorros de Alicante y Murcia, Murcia, pp. 136–155. L´opez-Berm´udez F and Mariscal B (1996) Deserticaci´on en el Sureste Espa˜nol. Secuencias Paleoambientales Holocenas. Notes de Geograf´ıa F´ısica 25, 67–83. L´opez-Berm´udez F, Romero D´ıaz A, Cabezas F, Rojo-Serrano L, Mart´ınez-Fern´andez J, Boer M and Del Barrio G (1998) The Guadalent´ın Basin, Murcia, Spain. In P Mairota, J Thornes and N Geeson (eds) Atlas of Mediterranean Environments in Europe. John Wiley, Chichester, pp. 130–142. Mart´ınez Carri´on JM (1990) Las transformaciones agrarias en Lorca durante el S.XIX y comienzos del XX. In Lorca Pasado y Presente. Aportaciones a la Historia de Murcia, vol. II. Caja de Ahorros del Mediterr´aneo, Lorca, 129–148. Mart´ınez Carri´on JM (1991) La Ganader´ıa en la Econom´ıa Murciana Contempor´anea. Consejer´ıa de Agricultura, Ganader´ıa y Pesca, Murcia. Mill´an M and Estrela MJ (1994) Manual pr´actico de introducci´on a la Meteorolog´ıa. Centro de Estudios Ambientales del Mediterr´aneo, Valencia. Mula G´omez AJ, Hern´andez Franco J and Gris Mart´ınez J (1986) Las Obras Hidr´aulicas en el Reino de Murcia durante el Reformismo Borb´onico. Los Reales Pantanos de Lorca. Colegio de Ingenieros de Caminos, Canales y Puertos. Nicol´as Mart´ınez JL (1995) Trasvase Tajo-Segura. In M Senent and F Cabezas (eds) Agua y Futuro en la Regi´on de Murcia. Asamblea Regional de Murcia, Cartagena, pp. 129–141. P´erez Picazo MT (1980) Aspectos socio-econ´omicos del S.XIX en Lorca. In M Mu˜noz Barber´an (ed). Ciclo de Temas Lorquinos. Caja de Ahorros de Alicante y Murcia, Murcia, pp. 156–176. Sanz Montero ME, Avenda˜no C, Cobo R and G´omez JL (1998) Determinaci´on de la erosi´on en la Cuenca del Segura a partir de los sedimentos acumulados en sus embalses. Geogaceta 23, and 135–138. Senent Alonso M and Arag´on Rueda R (1995) Descripci´on hidrogeol´ogica de la Cuenca del Segura. In Agua y futuro en la Regi´on de Murcia. M Senent Alonso and F Cabezas Calvo-Rubio (eds) Asamblea Regional de Murcia, Murcia, pp. 110–127. Torres Fontes J (1994) Repartimiento de Lorca, 2nd edn. Real Academia Alfonso X el Sabio, Murcia. Walker MJ (1979) From hunter-gatherer to pastoralist: rock painting and Neolithic origins in Southeastern Spain. National Geography Society, Research Report, Washington, pp. 511–545.
18
Local and Regional Responses to Global Climate Change in South-east Spain
C.M. GOODESS AND J.P. PALUTIKOF
Climatic Research Unit, University of East Anglia, Norwich, UK
1 INTRODUCTION Climate change is one element in the mosaic of processes and responses that must be considered in any effective study of Mediterranean desertification. The focus of much ongoing research is the enhanced greenhouse effect, which is considered to be the most likely cause of climate change in the short-term future (Palutikof et al. 1996). Following the second scientific assessment by the Intergovernmental Panel on Climate Change it was concluded that “the balance of evidence suggests that there is a discernible human influence on global climate” (Houghton et al. 1996). Regional scenarios of climate change due to the enhanced greenhouse effect are needed in order to investigate the impacts on Mediterranean desertification processes over the next 50 years. Climate scenarios are not definite predictions but “internally-consistent pictures of a plausible future climate” (Wigley et al. 1986). The construction of regional scenarios for the Guadalent´ın Basin target area in south-east Spain is described in this chapter. First, however, in order to provide a context for these future climate scenarios, the present-day climatology of the Guadalent´ın Basin and variations in rainfall over the period of instrumental record are outlined.
2 CLIMATOLOGY OF THE GUADALENTI´N BASIN The climatic regime of the Guadalent´ın Basin is typically Mediterranean. Winters are relatively mild. The mean temperature of the coldest month (either January or December) varies from 7 ◦ C at high altitudes (Alhama de Murcia; Table 18.1) to about 13 ◦ C on the coast (Aguilas; Table 18.1). As expected in this typical Mediterranean regime, summers are hot. The mean temperature of the warmest month (either July or August) for the six stations listed in Table 18.1 ranges from 23 ◦ C to 27 ◦ C. The rainfall regime of the Guadalent´ın Basin is characteristic of Mediterranean-type climates in that the winter rainfall amount is at least three times the summer rainfall (Ko¨ ppen 1936). Month by month, however, the seasonal rainfall cycle differs from that across much of the Mediterranean Basin in having two peaks: a major peak in October and a slightly lower peak in April. Mean annual rainfall varies widely across the region, from less than 200 mm near the coast to over 400 mm at the highest altitude stations (Figure 18.1 and Table 18.1). The mean annual number of rain days also varies widely, from fewer than 30 days to just under 50 days (Table 18.1). The varying precipitation regime across the Basin reflects the complex topography (see Chapter 17). Changes in annual and seasonal rainfall in the Guadalent´ın Basin over the period 1958–1987 have been investigated using standardized anomaly indices. By compositing data from a number of stations, it is often possible to obtain a clear signal of trend or pattern in the data, which would not be apparent by examining time series from individual stations. Because the means and standard Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
248
Mediterranean Desertification
Table 18.1 Details of the six Guadalent´ın stations. Means are for the period 1958–1987
Latitude
Longitude
Altitude (metres)
Mean annual rainfall (mm)
Annual rainfall: standard deviation
Mean number of rain days
Rain days: standard deviation
37.4 38.0 37.9 37.7 37.7 37.8
−1.6 −1.2 −1.5 −1.2 −1.7 −1.5
5 75 760 200 335 200
178 289 418 272 234 293
74 113 145 107 83 121
29 48 44 34 38 28
11 12 13 8 11 7
Aguilas Alcantarilla Alhama de Murcia Fuente Alamo Lorca Totana
38.25°N 250 350 38.00°
AL 350 400
37.75°
TO
LO
FA
300
250 250
37.50°
200 37.25° 2.00°
250
AM
1.75°
AG
1.50°
1.25°
1.00°
0.75°
0.50°W
Figure 18.1 Mean annual rainfall (mm) for 20 stations in the Guadalent´ın Basin, 1958–1987, and the location of the six stations used here: AG, Aguilas; AL, Alcantarilla; AM, Alhama de Murcia; FA, Fuente Alamo; LO, Lorca; TO, Totana. Reproduced by permission of John Wiley and Sons Ltd
deviations of individual records differ, it is necessary to standardize prior to compositing. The formula for the calculation of standardized anomaly indices (SAIs) is i=n
SAI =
1 [(xij − x¯i )/σi ] n i=1
(1)
where n is the number of stations, x is the precipitation at station i for year j , x¯ is the station mean, and σ is the standard deviation. Annual and seasonal SAIs (Figure 18.2) have been constructed using the six stations listed in Table 18.1. Linear trends were calculated for each annual and seasonal SAI but are not shown because none had an r 2 value greater than 0.4 or a significance level greater than 75%. Although significant linear trends cannot be identified, a number of distinctive subperiods are apparent. The relatively dry periods of, on the one hand, the early and mid-1960s and, on the other, the late 1970s and 1980s, are separated by the relatively wet early 1970s. The seasonal SAIs indicate that this pattern of change is seen throughout the year except in the winter months. The shape of the seasonal rainfall cycle in the Guadalent´ın Basin appears to have changed over time. This is demonstrated in Table 18.2 for one representative station, Alhama de Murcia, by
Local and Regional Responses to Global Climate Change Spring
3.0
3.0
2.5
2.5
2.0
2.0
1.5
1.5
Rainfall amount
Rainfall amount
Winter
1.0 0.5 0.0
1.0 0.5 0.0
−0.5
−0.5
−1.0
−1.0 −1.5
−1.5 1958 1962 1966 1970 1974 1978 1982 1986 1960 1964 1968 1972 1976 1980 1984
1958 1962 1966 1970 1974 1978 1982 1986 1960 1964 1968 1972 1976 1980 1984 Autumn
Summer 3.0
3.0
2.5
2.5
2.0
2.0
1.5
1.5
Rainfall amount
Rainfall amount
249
1.0 0.5 0.0
1.0 0.5 0.0
−0.5
−0.5
−1.0
−1.0
−1.5
−1.5 1958 1962 1966 1970 1974 1978 1982 1986 1960 1964 1968 1972 1976 1980 1984
1958 1962 1966 1970 1974 1978 1982 1986 1960 1964 1968 1972 1976 1980 1984 Annual 2.0 1.5
Rainfall amount
1.0 0.5 0.0
−0.5 −1.0 −1.5 1958 1962 1966 1970 1974 1978 1982 1986 1960 1964 1968 1972 1976 1980 1984 1988
Figure 18.2 Basin
Annual and seasonal standardized anomaly indices of rainfall for the Guadalent´ın
comparing mean seasonal and annual rainfall totals for two different decades. The more recent decade (1977–1986) is considerably drier than the earlier decade (1967–1976). The percentage difference in mean annual rainfall between the two decades is 30%. The greatest seasonal difference occurs in spring (41%) and the smallest difference in winter (18%). The rainfall changes in the Guadalent´ın Basin reflect the general trend towards decreasing rainfall which is evident in SAIs constructed for four Mediterranean regions including the Iberian Peninsula. The SAIs for the Iberian Peninsula (not shown) are constructed using data for 1951–1992 from 17
250
Mediterranean Desertification Table 18.2 Mean annual and seasonal rainfall at Alhama de Murcia calculated for two consecutive decades
Annual Winter Spring Summer Autumn
Mean rainfall (mm) 1967–1976
Mean rainfall (mm) 1977–1986
Difference
509 107 197 55 150
353 88 117 38 110
30% 18% 41% 30% 27%
Figure 18.3 Observed seasonal and annual rainfall time series for 1901–1996 for a 0.5◦ grid box centred over the Guadalent´ın Basin
Local and Regional Responses to Global Climate Change
251
stations. The annual, winter and spring SAIs all show a clear downward trend. The SAI for autumn indicates two approximately 10-year periods of above-average rainfall (1958–1969 and 1982–1992) interrupted by a period of below-average rainfall (1970–1981). This broadly reflects the pattern of change in the Guadalent´ın Basin (Figure 18.2). The results obtained in MEDALUS II and outlined above are based on rainfall records up to 1987 for the Guadalent´ın Basin and up to 1992 for the Iberian Peninsula and Mediterranean Basin. With very few exceptions they indicate a drying climate. To show more recent trends, seasonal and annual time series for 1901–1996 for station data interpolated to a 0.5◦ grid box centred over the Guadalent´ın Basin (New et al. 1998a,b) are shown in Figure 18.3. There is some indication of increased inter-annual variability with more frequent extremes since the mid-1980s, particularly in winter and summer. For example, the two very wet winters of 1989 and 1996 are separated by a period of drier conditions, including the very dry winter of 1992. December 1996 was a particularly extreme month in Spain with flooding and some loss of life, and the destruction of US$4 million worth of crops (McKinlay 1996). Despite these events, it is not yet possible to say that the drying trend has been reversed or terminated.
3 MODELLING FUTURE CLIMATE The MEDALUS scenarios of future climate change in response to the enhanced greenhouse effect are all based on General Circulation Models, or GCMs. These are widely regarded as the best source of information for this purpose. GCMs are complex, three-dimensional computer-based models of the atmospheric circulation developed from numerical weather forecasting models (McGuffie and Henderson-Sellers 1996). Results are output for a low-resolution global three-dimensional grid. In the UK Hadley Centre high-resolution GCM (UKTR) (Murphy 1995; Murphy and Mitchell 1995), for example, the spatial resolution of the model grid is 2.5◦ latitude by 3.75◦ longitude at Mediterranean Basin latitudes. The UK Hadley Centre has performed a 75-year transient experiment with this coupled atmosphere–ocean GCM (Murphy 1995; Murphy and Mitchell 1995). Atmospheric CO2 concentrations were increased by 1% per year compounded, doubling in year 70. Some model results for the grid square closest to the Guadalent´ın Basin are shown in Figure 18.4. They illustrate the changes in winter (December to February) temperature and in rainfall during the main rainy season (October to March). The changes are presented for each decade of the perturbed experiment. Although the results for the summer months are not shown, the trends are in the same direction as in winter. Figure 18.4 shows that the changes, particularly for rainfall, are not necessarily linear with time. This is to be expected with a fully coupled atmosphere–ocean GCM which should be able to incorporate, in part, the effects of natural variability, as well as greenhouse warming. Even for temperature, there is cooling between the fifth and sixth model decades, although overall there is a clear trend towards warmer and drier conditions for this particular grid square. The changes shown in Figure 18.4 represent one possible scenario for south-east Spain based on the UKTR experiment. Experiments performed with other GCMs provide alternative, equally plausible, scenarios. The 1995 assessment by the Intergovernmental Panel on Climate Change, for example, reviewed the precipitation changes indicated by seven models for southern Europe in winter and summer (Kattenberg et al. 1996). Six models showed increased precipitation in winter (whereas UKTR indicates a decrease), and five indicated a decrease in summer (in agreement with UKTR). Different scenarios can be obtained from the same underlying climate model by specifying a different forcing pattern. In MEDALUS II, we have evaluated precipitation output for Northern Mediterranean grid boxes from transient simulations performed with the state-of-the-art fully coupled Hadley Centre GCM (HADCM2) (Mitchell et al. 1995). The experiment in which only the concentration of atmospheric greenhouse gases is varied (HADCM2GHG), shows a long-term decline in rainfall, beginning in the 1940s, of around 0.5 mm day−1 over the experimental period (1860–2099). However, when atmospheric sulphates are added to the experiment (HADCM2SUL), there is no observable trend over the period 1860–2040. After that, there is some indication of a reduction in rainfall which may be because sulphate emissions decline in the later years of the model run.
252
Mediterranean Desertification 2.0
Spain
Temperature (°C)
1.5 1.0 0.5 0.0 −0.5 5−14
15−24
25−34
35−44
45−54 55−64
65−74
45−54 55−64
65−74
Decade Spain 0.2
Rainfall (mm day−1)
0.1 0.0 −0.1 −0.2 −0.3 −0.4 −0.5 −0.6 5−14
15−24
25−34
35−44 Decade
Figure 18.4 Changes in winter (December–February) temperature and in rainfall during the main rainy season (October–March) over the perturbed run of the UKTR model for the grid square closest to the Guadalent´ın Basin
Although GCMs are considered to provide the greatest potential for scenario construction there are two problems for the generation of regional scenarios: • •
even in the most recent generation of GCMs, the model grid is coarse in comparison to the spatial resolution at which information is required; and, the extent to which we can have confidence in the results from GCMs is scale-dependent. As we move from the global to the regional scale, and from the annual to the monthly and even daily scale, confidence diminishes (Hewitson and Crane 1992a,b; von Storch et al. 1993). The accuracy of GCMs can be tested through the procedure of validation, i.e. comparison of output from the control or 1 × CO2 simulation with observations. At high spatial and temporal resolutions, the results from validation are generally poor when raw GCM output is compared with observations (Palutikof et al. 1997).
To overcome these problems, techniques of downscaling have been developed. Downscaling may be defined as “sensibly projecting the large-scale information on the regional scale” (von Storch et al. 1993). A number of different methods have been proposed and can be divided into two general categories: model-based and empirical (Cubasch et al. 1996; Hewitson and Crane 1996;
Local and Regional Responses to Global Climate Change
253
Schubert and Henderson-Sellers 1997). Model-based approaches include the use of high-resolution limited area models nested within a GCM (Giorgi et al. 1990, 1992; Jones et al. 1995). This approach is considered to offer good long-term potential (Hewitson and Crane 1996) but is very computerintensive (Schubert and Henderson-Sellers 1997). Empirical downscaling is less computer-intensive. It involves the identification of relationships between the observed large-scale and regional climates. These relationships are then applied to GCM output. In MEDALUS II, we developed and tested the potential of two different approaches to empirical downscaling in the Guadalent´ın Basin. First, empirical transfer functions were developed and applied to output from the Canadian Climate Centre GCM to produce daily scenarios of maximum and minimum temperature (Palutikof et al. 1997; Winkler et al. 1997). Second, atmospheric circulation classifications were used to construct daily rainfall scenarios using output from the UK Hadley Centre high-resolution transient GCM (Goodess and Palutikof 1998). Both downscaling methods rely on the assumption that the observed relationships between the large-scale climate and the daily temperature or rainfall regime will be unchanged in a future warmer world. This assumption cannot be fully tested and a similar assumption must be made when using any of the empirical approaches to downscaling. The construction of these scenarios is described in the following two sections and their implications for the Guadalent´ın Basin are considered in the final section of this chapter.
4 CONSTRUCTION OF DAILY MAXIMUM AND MINIMUM TEMPERATURE SCENARIOS The Canadian Climate Centre second-generation GCM (Boer et al. 1992; McFarlane et al. 1992) forms the basis of the method to downscale maximum and minimum temperatures in the Guadalent´ın. Output from this equilibrium-response simulation is available as 10-year time series of twice-daily values (daily in the case of rainfall and temperature) for both 1 × CO2 (control) and 2 × CO2 (perturbed) conditions and for 21 climate variables. Amongst these 21 variables are the free-atmosphere variables sea-level pressure and 500 hPa geopotential height. This raises the possibility of using transfer functions to relate these free-atmosphere variables to the required surface variables (here maximum and minimum temperature) in what is essentially a “Perfect Prog” type approach (Klein 1982; Glahn 1985). Such an approach to downscaling was first employed by Karl et al. (1990) and the method described here is heavily influenced by these authors. The method is based on the hypothesis that the free-atmosphere variables are better simulated by GCMs than the surface variables. Thus, before proceeding to develop downscaled scenarios it is necessary to demonstrate, first, that the required surface variables are poorly simulated by the GCM such that downscaling is required and, second, that the free atmosphere variables are sufficiently well simulated by the GCM to form the basis of a downscaling methodology. Throughout this section, we use the example of Alcantarilla (see Table 18.1, and Figure 18.1) to illustrate the application of the downscaling methodology for maximum and minimum temperatures in the Guadalent´ın. Thirty years of daily temperature observations are available for this site. 4.1 Comparison of Control-Run Output with Observations
Table 18.3 compares summary statistics for seasonal and annual maximum (TMAX) and minimum (TMIN) temperatures calculated from observations for 1975–1984 and from 1 × CO2 model data. A 16-point Bessel interpolation scheme was used to interpolate model output from the GCM grid to the site of Alcantarilla. Modelled maximum temperatures are too low, minimum temperatures too high. In consequence, daily mean temperatures are well-simulated: the annual mean temperature is 17.6 ◦ C for 1975–1984, and 17.3 ◦ C for 1 × CO2 . Standard deviations in winter are far too small, although in the other three seasons they are well modelled. This is shown more clearly in Figure 18.5, where the frequency distributions for the winter and summer seasons are plotted. Although the frequency distributions in summer are reasonably well simulated by the model, in winter the kurtosis is far
254
Mediterranean Desertification Table 18.3 Summary temperature statistics for 1975–1984 observations and model control run
Maximum (◦ C)
Annual 1975–1984 obs. GCM control Winter 1975–1984 obs. GCM control Spring 1975–1984 obs. GCM control Summer 1975–1984 obs. GCM control Autumn 1975–1984 obs. GCM control ∗
Mean
SD
Mean
SD
24.3 20.9∗
6.8 7.3
11.2 13.6∗
6.0 5.9
17.3 12.5∗
3.6 1.7∗
5.1 7.7∗
3.8 1.3∗
22.4 18.2∗
4.1 3.7
9.3 10.8∗
3.5 3.2
32.1 29.7∗
3.5 3.1
18.0 21.0∗
2.7 2.9
25.2 23.0∗
5.3 5.3
12.3 14.7∗
4.7 4.4
Significant differences, control against observed. Control run 40
30
30
20 10 15
0 −25 −15
25
30
30
20
%
30
20 10
15
25
35
0 −5
45
40
40
30
30
20 10
5
15
25
35
7
−3
7
−3
%
30
%
30 20 10 5
15
25
35
0 −5
25
35
45
−3
7
0 −33 −23 −13
17
30
−5
15
10
40
0
5
20
40
10
25
30
20
40
20
15
40
49.8
0 −33 −23 −13
17
5
20
0 −5
45
10
0 −33 −23 −13
−5
10
%
5
%
%
15
40
−5
%
5
−5
40
0
Summer
20 10
40
10
MINIMUM TEMPERATURES Winter
20
0 −25 −15
25
%
%
5
−5
49.9
30
10
0 −25 −15 Summer
Perturbed run 40
43.2 %
40
%
%
1975−1984 MAXIMUM TEMPERATURES Winter
Minimum (◦ C)
17
20 10
5
15
25
35
0 −5
5
15
25
35
Temperature (°C)
Figure 18.5 Frequency distributions of winter and summer daily TMAX and TMIN (◦ C) for Alcantarilla
255
Local and Regional Responses to Global Climate Change ◦
too high, with over 40% of values occurring in the 11–13 C bin for TMAX and in the 7–9 ◦ C bin for TMIN. The cause of the low standard deviations in winter is clearly shown by Figure 18.6. For each day, the highest and lowest of the 10 values (in this case, for TMIN) are plotted. There is a threshold, at around 5 ◦ C, below which temperatures do not fall in winter, and which is responsible for the low winter standard deviation in the modelled data. Further investigation of grid-point data has shown that the true threshold is at 0 ◦ C, and that the higher threshold at Alcantarilla is an artefact of the interpolation from the 16 model grid points, some of which are over land and some of which are over sea (see Palutikof et al. (1997) for a further discussion of this phenomenon). Clearly, the simulation of daily TMAX and TMIN by the GCM is unsatisfactory. The next step is to test whether the model simulates the free atmosphere adequately. Table 18.4 shows summary information for the two free-atmosphere variables retained from the Canadian Climate Centre model simulation: sea-level pressure (SLP) and 500 hPa geopotential height (Z500). The observed data are taken from the gridded meteorological data sets from the National Meteorological Centre (NMC) operational analyses (available on CD-ROM), for the period 1975–1984. Again, a 16-point Bessel scheme was used to interpolate the observed and modelled data to the site of Alcantarilla. From Table 18.4, the largest difference between modelled and observed mean SLP is 6.3 hPa, in winter. For Z500, the largest mean difference is in summer (42 m). Standard deviations are generally too high in the model for SLP, and too low for Z500, and the differences are generally statistically significant. Figure 18.7 shows the evolution through the year of the daily mean for both variables. Particularly for Z500, the observed seasonal cycle of the two variables is well simulated by the model. The cycle of SLP is exaggerated, with too many high pressure events in winter and too many low pressure events in summer. In summary, there are statistically significant differences between the observed and modelled free-atmosphere data, although numerically these differences are small. Importantly, there are no systematic errors of the type caused by the threshold in the temperature data. It was 30
25
Temperature (°C)
20
15
10
5
0
−5
−10
0
50
100
150 200 Julian day
250
300
350
Figure 18.6 Highest and lowest daily TMIN (◦ C) from 10 years of observations for 1975–1984 (solid line) and control run output from the Canadian Climate Centre GCM (dashed line) for Alcantarilla
256
Mediterranean Desertification Table 18.4 Summary free-atmosphere variable statistics for 1975–1984 observations and model control run
Annual 1975–1984 obs. GCM control Winter 1975–1984 obs. GCM control Spring 1975–1984 obs. GCM control Summer 1975–1984 obs. GCM control Autumn 1975–1984 obs. GCM control
Sea level pressure (hPa)
∗
500 hPa height (m)
Mean
SD
Mean
SD
1018.1 1019.3∗
6.25 8.56∗
5732.1 5722.4
127.4 94.2∗
1020.5 1026.8∗
8.08 7.45
5652.3 5649.1
114.8 70.3∗
1016.4 1019.1∗
5.69 6.87∗
5657.1 5673.0∗
96.2 71.6∗
1016.5 1010.9∗
3.17 4.54∗
5854.5 5812.6∗
65.5 54.4∗
1018.9 1020.6∗
6.07 6.50
5766.6 5753.8
100.0 74.6∗
Significant differences, control against observed.
1040 1030 1020 1010 1000
500 hPa geopotential height (m)
Sea level pressure (hPa)
0
50
100
150 200 Julian day
250
300
350
0
50
100
150 200 Julian day
250
300
350
6000 5900 5800 5700 5600 5500 5400
Figure 18.7 Mean daily sea-level pressure (hPa) and 500 hPa geopotential height (m) calculated from data at 0000z for Alcantarilla. The solid line indicates observations (1975–1984) and the dotted line the Canadian Climate Centre GCM control run
Local and Regional Responses to Global Climate Change
257
considered that the simulation of the free-atmosphere variables by the GCM is adequate as a basis for downscaling, although the errors in the GCM simulations will propagate through to the downscaled scenarios and should be borne in mind in their interpretation. 4.2 Transfer Function Development
The transfer functions for downscaling use the free-atmosphere variables in multiple regression equations to predict TMAX or TMIN (separate equations were developed for each). The 10-year period 1975–1984 was used for construction, and data from the period 1965–1974 were retained for validation. The predicted variables in the regression equations are either TMAX or TMIN. We chose not to use raw values of these variables; if we did, there is a strong possibility that, particularly in the case of the 2 × CO2 model run, the regression equations would be applied in the prediction stage to cases that are beyond the range of the data used in the construction stage. For the same reason, all meteorological predictor variables are expressed as standardized anomalies from the monthly mean. However, TMAX and TMIN cannot be expressed in this way, because the eventual aim is to have a time series of actual temperature values derived from the GCM data. If we predicted straightforward standardized anomalies of TMAX and TMIN, we would be left with the problem of how to recreate these actual values, in the absence of information regarding the true mean and standard deviation of the series. These considerations are discussed further by Winkler et al. (1997). There must always be tension between the desire to utilize all the information in the predictors (which would favour using non-standardized variables) and ensuring that the assumptions of regression analysis are not invalidated (which here requires standardized variables). We explore here the results from the latter approach. Winkler et al. (1997) examine the implications of both assumptions. We therefore based the predicted variables on our finding that the mean temperature ((TMAX + TMIN)/2) is well simulated in the control run, and we assume that this is also the case in the perturbed simulation. TMAX and TMIN are standardized not by their own mean and standard deviation, but by the mean and standard deviation of the daily mean temperature. There remains the problem that the variance, even of daily mean temperatures, is too low in the model, particularly in winter. This can be overcome by assuming that the model variance : observed variance ratio remains constant between the control and perturbed simulations. Thus, when the transfer functions are initialized with model data, the predicted values of TMAX and TMIN can be de-standardized using the means and variances of the modelled daily mean temperature, after first adjusting the variances by the ratio of observed-to-control variances. This approach generates time series of absolute TMAX and TMIN from the modelled free-atmosphere variables. We used stepwise multiple regression analyses to construct the transfer functions. The candidate independent variables were derived from the values of SLP and Z500, available at 0000 UTC and 1200 UTC. First, the two point variables were included. Second, north–south and east–west gradients (calculated over 4◦ of latitude and longitude) were used to approximate the strength and orientation of the low- and upper-level airflow. Third, persistence was incorporated through 24-hour backward and forward changes in both SLP and Z500, calculated using 0000 UTC values. Finally, to capture the effects of seasonality, the sine and cosine of Julian Day were added (as one full cycle over the year). Four regression equations were calculated, one for each season. The decision to work at the seasonal scale was determined, first, by the need to preserve a sufficient number of degrees of freedom and, second, to limit the number of discontinuities between the predicted series. 4.3 Validation
The regression equations were validated on observations from the period 1965–1974. The performance is summarized in Table 18.5. In this table, the values of TMAX and TMIN have been reconstructed from the dependent variable, prior to the calculation of the test statistics. It can be seen that the transfer functions perform well on the independent data set. In terms of the correlation
258
Mediterranean Desertification Table 18.5 Summary temperature statistics for observations for 1965–1974 and transfer-function-generated series (TFG) based on the GCM control simulation
Maximum (◦ C)
Winter Observed TFG Spring Observed TFG Summer Observed TFG Autumn Observed TFG ∗
Minimum (◦ C)
Mean
SD
R2
Mean
SD
R2
16.7 16.4
3.4 3.4
0.74
4.5 4.2
3.6 3.5
0.74
22.5 22.4
4.8 4.1∗
0.83
9.0 9.2
3.8 3.9
0.76
31.7 31.7
3.6 3.4
0.85
18.0 18.0
2.7 3.0
0.77
24.4 25.1*
5.3 5.6
0.90
12.5 11.8*
4.7 4.3
0.86
Significant differences, TFG against observed. Table 18.6 Summary statistics for TFG TMAX and TMIN using control- and perturbed-run output for initialization
Winter Control Perturbed Spring Control Perturbed Summer Control Perturbed Autumn Control Perturbed
Maximum (◦ C)
Minimum (◦ C)
Mean
SD
Mean
SD
16.2 20.2
2.9 3.2
4.1 5.9
4.7 5.5
21.1 26.3
3.8 4.9
7.9 11.1
4.2 4.6
32.3 37.2
3.4 3.1
18.4 23.5
3.5 3.6
25.3 28.9
5.4 5.3
12.4 17.2
5.1 5.4
between the observed and predicted series, the results for TMAX are slightly better than those for TMIN, although both are acceptable. 4.4
Using the Transfer Functions for Downscaling
First, the regression equations were applied to data from the control run of the GCM. The results are shown in Table 18.6, and can be compared with Table 18.3 to demonstrate the improvements achieved. Two problems were noted with respect to the raw control-run TMAX and TMIN data in Table 18.3. First, the winter standard deviations were far too low in the modelled data (1.7 ◦ C for TMAX and 1.3 ◦ C for TMIN). In Table 18.6, they have increased to 2.9 ◦ C and 4.7 ◦ C respectively, compared with 3.6 ◦ C for observed TMAX and 3.8 ◦ C for observed TMIN. Second, mean modelled
Local and Regional Responses to Global Climate Change
259
TMAX was found to be too low, and mean modelled TMIN was found to be too high, compared to observations. The differences between control-run and observed (1975–1984) temperatures are found to be much lower for the transfer-function-generated series than for the raw model data, as shown in Table 18.7. The improvement is particularly good in autumn and winter. The overall improvement can be judged by comparing the winter and summer frequency distributions from the transfer-function generated (TFG) model series (not shown) with those from the observed and raw control-run series (Figure 18.5). In particular, with respect to winter, the shape of the TFG series is much closer to the observed than is the shape of the raw control-run series. Table 18.6 also shows the results from initialization of the regression equations with 2 × CO2 data. The differences, 2 × CO2 minus 1 × CO2 , for the raw model data and for the TFG data, are shown in Table 18.8. For TMAX, the TFG mean change is greater in three seasons (winter, spring and autumn), and lower in summer, than the change indicated by the raw model data. The annual change is only slightly different: 4.4 ◦ C for TFG data, 4.1 ◦ C for the raw model data. For TMIN, the mean change for the TFG data is smaller in winter and spring, and larger in summer and autumn, than the raw model change. The annual TMIN warming is 3.6 ◦ C for the TFG data, and 3.7 ◦ C for the raw model data. There are some surprisingly large differences between the changes in variance indicated by the two data sets. For example, the raw data shows a very large change in the variance of autumn TMAX, whereas a small decrease is suggested by the TFG data. The large increase in spring TMAX variance in the TFG data is not repeated in the raw data.
Table 18.7 Differences between control-run (raw model and TFG) and observed temperatures for 1975–1984
Winter Spring Summer Autumn
Maximum temperatures
Minimum temperatures
Raw
TFG
Raw
TFG
−4.8 −4.2 −2.4 −2.2
−1.1 −1.3 0.2 0.1
2.6 1.5 3.0 2.4
−1.0 −1.4 0.4 0.1
Table 18.8 Changes (perturbed − control) for TFGgenerated and raw TMAX and TMIN
Maximum (◦ C)
Winter Raw TFG Spring Raw TFG Summer Raw TFG Autumn Raw TFG
Minimum (◦ C)
Mean
SD
Mean
SD
3.7 4.0
−0.2 0.3
2.2 1.8
0.4 0.8
4.7 5.2
0.4 1.1
3.6 3.2
0.7 0.2
5.2 4.9
0.2 −0.3
4.8 5.1
−0.3 0.1
2.8 3.6
2.3 −0.1
4.1 4.8
0.4 0.3
260
Mediterranean Desertification Table 18.9 Measures of temperature extremes, calculated from 10 years of data∗
NoHD
HDG
Observations 1965–1974
17
29
15
12
Raw control Raw perturbed Change
1 58 57
0 186 186
0 0 0
0 0 0
TFG control TFG perturbed Change
22 93 70
28 287 259
42 32 −10
17 13 −4
∗
NoFD
FDG
To obtain these measures “per year”, it is necessary to divide by 10.
4.5 Changes in Extreme Events A principal reason for downscaling in the Guadalent´ın was to generate meaningful time series which could be analysed with respect to the occurrence of extreme events. We therefore examined the following measures of extreme temperatures using TMIN:
• •
The number of frost days, i.e. with TMIN ≤ 0 ◦ C (NoFD). The number of freezing degree days (FDG). These are found by summing the number of degrees Celsius below zero which occur over the period of interest. For example, over a five-day period with TMIN values of +5 ◦ C, −5 ◦ C, −2 ◦ C, −3 ◦ C and +1 ◦ C, the number of degree days would be 0 + 5 + 2 + 3 + 0 = 10.
Using TMAX: • •
The number of excessively hot days, i.e. with TMAX ≥ 35 ◦ C (NoHD). The number of degree days per year above the 35 ◦ C threshold (HDG).
The number of very cold days and their severity have implications for agriculture, particularly with respect to timing. Severe cold in mid-winter can be beneficial for disease and pest control. Towards the end of winter, however, it may be responsible for the destruction of new crop growth. The occurrence of very high temperature events has implications for human health and morbidity, as well as for the viability of crops. High temperatures imply high rates of evapotranspiration, with corresponding reductions in soil moisture and increased frequency of drought, other things being equal. Table 18.9 shows the comparison between the extremes in the observed temperature time series for 1965–1974, and their occurrence in the GCM control simulation, both raw and TFG. Looking first at the raw control data, the model is not able to simulate the extremes of the distribution–there are too few very hot days and no frost days. Although the number of hot days increases substantially in the perturbed experiment, the poor performance at the validation (control) stage must reduce confidence in this result. The control-run TFG extremes accurately simulate the occurrence of very hot days, and the number of freezing degree days, although the number of days with temperatures at or below 0 ◦ C is too high. However, the improvement over the raw model data is considerable. The perturbed-run TFG results indicate the expected reduction in frost frequency and severity. Less expected is the very large increase in the occurrence of high temperature events. This would lead to a substantial increase in high temperature stress for both plants and human beings.
5
CONSTRUCTION OF DAILY PRECIPITATION SCENARIOS
The daily temperature scenarios for the Guadalent´ın Basin described above are constructed using empirical transfer functions to describe the relationship between the large-scale and local climate.
261
Local and Regional Responses to Global Climate Change
Here, downscaling is performed on the basis of a classification of pressure patterns into atmospheric circulation types, which are then related to local rainfall amounts at the daily scale (Goodess and Palutikof 1998). If strong relationships exist between the circulation types and daily rainfall in the observed data sets, and similar relationships exist in a future warmer climate regime, then GCM output can be used to investigate changes in the frequency and intensity of rain storms due to global warming. For example, if the frequency of circulation types associated with high-intensity rainfall events is predicted to increase, it can be assumed that the frequency of high-intensity rainfall events will also increase. 5.1 The Automated Circulation-Type Scheme
The automated circulation-type scheme used to classify daily gridded sea-level pressure (SLP) data for the Iberian Peninsula by direction and type of flow has been developed using the National Meteorological Center (NMC) CD-ROM data set. It is based on Lamb weather types (Lamb 1972) and a method first applied to UK data (Jenkinson and Collinson 1977; Hulme et al. 1993; Jones et al. 1993). In order to match the grid spacing of the UK Hadley Centre high-resolution transient GCM experiment (UKTR), the data have been interpolated to a 2.5◦ latitude by 3.75◦ longitude grid over the area 36.25 ◦ N to 46.25 ◦ N and 16.88 ◦ W to 9.38 ◦ E. Flow and vorticity parameters, including resultant flow (F ) and total shear vorticity (Z), have been calculated for this grid and used to define 14 basic circulation types (Table 18.10). Over the Guadalent´ın Basin, many of the 14 circulation types are relatively infrequent in at least some seasons. This makes it difficult to establish reliable Table 18.10
The 14 basic circulation types
Values of Z and F Z
2F Z>0 Z<0 Z < 6 and F < 6 Z>0 Z<0 F < Z < 2F Z>0 Z<0
Table 18.11
Type C HYC UC A/HYA UA W/NW/SW/N E/NE S/SE
Description
Type ◦
Directional (resolution of 45 )
N, NE, E, SE, S, SW, W, NW
Cyclonic Anticyclonic
C A
Unclassified cyclonic Unclassified anticyclonic
UC UA
Hybrid cyclonic Hybrid anticyclonic
HYC HYA
The eight circulation-type groups
Description Cyclonic Hybrid cyclonic Unclassified/light flow cyclonic Anticyclonic/hybrid anticyclonic Unclassified/light flow anticyclonic Westerly/northwesterly/southwesterly/ northerly directional types (202.5◦ –22.5◦ ) Easterly/northeasterly directional types (22.5◦ –112.5◦ ) Southerly/southeasterly directional types (112.5◦ –202.5◦ )
Mean annual frequency (days) 19.1 19.4 90.8 41.9 79.0 68.6 33.1 12.3
262
Mediterranean Desertification
statistics for the rainfall regime associated with each type, particularly given the low number of rain days overall for this region of Spain. The 14 basic types were, therefore, regrouped into eight types, including three directional groups (Table 18.11). 5.2
Observed Circulation-type Frequencies
The mean seasonal cycles of circulation-type frequency calculated from observed SLP for the period 1956–1989 are shown in Figure 18.8 (plotted relative to the hydrological rather than calendar year). Over the whole year, the most frequent circulation types are the two unclassified or indeterminate flow types (UC and UA) (Table 18.11). The UC type has a very strong seasonal cycle, with a minimum in winter and a maximum in summer. The UA type is most frequent in autumn, but does not have a strong seasonal cycle. In contrast, the A/HYA group has a very strong seasonal cycle, with a pronounced winter maximum. The C and HYC types have a late spring/summer maximum and a less pronounced seasonal cycle. The W/NW/SW/N group has a strong seasonal cycle which peaks in late autumn/winter. The E/NE group is one of the least frequent types and does not have a seasonal cycle. The least frequent category is the S/SE group which is particularly infrequent from May to September. 5.3
Observed Relationships between the Circulation Types and the Rainfall Regime
For each season, the relationships between the circulation types and the daily rainfall regime have been investigated for six stations in the Guadalent´ın Basin (Table 18.1). It is possible to identify those of the eight circulation types for which the amount of rainfall occurring is significantly higher (or lower) than average (Figure 18.9). The percentage contribution to annual rainfall from the UC type is the same as the percentage of days that are of this type (about 25%), indicating that the amount of rainfall occurring on UC type days is close to the mean annual average daily rainfall. In contrast, the E/NE types contribute about 25% of annual rainfall although they occur on less than 10% of days. The percentage contributions to annual rainfall of the C, HYC and S/SE types are also somewhat greater than the percentage of days on which these circulation types occur. Conversely, the rainfall contribution from the A/HYA, UA and W/NW/SW/N types is less than average. The relatively high percentage rainfall contribution from the C, HYC, E/NE and S/SE circulation types may be because a high proportion of days of a particular type are wet, or it may be because there are only a few wet days, with a large amount of rainfall on each. It is important to distinguish between the two, because the potential impacts are quite different. For example, erosion may be a serious problem where there are a few high-intensity rain days, but not when the rainfall is spread over a large number of low-intensity days. The first possibility, that a relatively high proportion of type days are wet, is indicated by the ratio PROP ct /PROP tot , where PROP ct is the proportion of type days that are wet and PROP tot is the proportion of all days that are wet. The second possibility, that a relatively large amount of rain falls on each type day, is indicated by the ratio PREC ct /PREC tot , where PREC ct is the mean amount of rain that falls on a wet type day and PREC tot is the mean amount of rain that falls on any wet day. These ratios have been calculated for each station for the year as a whole and for each season and used to identify consistent relationships between the circulation types and the daily rainfall regime (Goodess and Palutikof 1998). The relationships for the high-rainfall circulation types are summarized in Table 18.12. The E/NE type is the most consistent high-rainfall circulation type: the probability of rain and the amount of rain per rain day are high in every season and at every station. The C, HYC and S/SE types all have a high probability of rainfall at every station, although only the HYC type is important in every season. The S/SE type is also associated with a higher than average amount of rain per rain day at some stations in every season. These inter-seasonal and inter-station differences are a reflection of the complex topography of the Guadalent´ın Basin.
263
Local and Regional Responses to Global Climate Change 6
5
C Number of days
Number of days
6
4 3 2
F M A M J
J
UC Number of days
4
S O N D J
F M A M J
J
UA
4
S O
N D J
F M A M J
J
8
F M A
M
J
J
A
J
F M A
M
J
J
A
A/HYA
4
S O
N D
W/NW/SW/N
12 8 4
5 Number of days
Number of days
J
S O
N D
J
F M A
M
J
J
A
N D
J
F M A
M
J
J
A
6 E/NE
6 5 4 3 2
S/SE
4 3 2 1
1 0
N D
8
0
A
S O
12
16
8
7
2
0
A
12
0
3
16
8
16
4
0
A
12
0
Number of days
S O N D J
Number of days
Number of days
16
HYC
1
1 0
5
S O N D J
F M A M J
J
A
0
S O
Figure 18.8 Observed and simulated monthly frequencies of the eight circulation types. The solid line shows the mean frequency (number of days per month) calculated from observed SLP data (1956–1989). The shaded area shows the maximum and minimum frequency range observed over any 10-year period. The dashed line shows the mean frequency calculated from UKTR control-run model output. Reproduced by permission of John Wiley & Sons Ltd
264
Mediterranean Desertification 30
25
Percentage
20
15
10
5
0
C
HYC
UC
A/HYA
UA
"W"
E/NE
S/SE
Circulation type
Figure 18.9 The mean annual frequency (shown as the percentage of all days) of the eight circulation types (solid bars); and their contribution to annual rainfall (shown as the percentage of total annual rainfall) for six stations in the Guadalent´ın Basin (open bars). Reproduced by permission of John Wiley & Sons Ltd Table 18.12 High-rainfall circulation types with a higher than average proportion of wet days at every station (+) and a higher than average amount of rain per rain day at every station (*)
Annual Winter Spring Summer Autumn a
C
HYC
+ + +
+ + +∗a + +
+
UC
A/HYA
UA
W/NW/SW/N
+ + *
E/NE
S/SE
+* +* +* +* +*
+ + + +
Rain per rain day is higher than average at every station except Aguilas.
5.4
Validation of the GCM Circulation Types
Before the circulation type–rainfall relationships identified from the observations can be used to interpret GCM output, it is first necessary to validate the simulation of circulation types by the GCM control run for the Iberian Peninsula. Seasonal and monthly frequencies of the eight circulation-type groups have been calculated from 10 years of daily output from the control and perturbed runs of the UKTR GCM. The differences (number of days) between the mean seasonal circulation-type frequencies calculated from the observations and from model output are given in Table 18.13. The model performs better in spring and autumn, and worse in winter and summer. It underestimates the frequency of the C, HYC and UC types in spring and summer, and over the year as a whole. The frequency of the C and HYC types is overestimated in winter. The frequency of the E/NE type is underestimated in every season except summer, when it is overestimated. The frequency of the S/SE types is simulated fairly well, except in autumn when it is overestimated. The
265
Local and Regional Responses to Global Climate Change
Table 18.13 Actual differences (days) between simulated and observed circulation-type frequencies. Differences that are significant at the 5% level are indicated (*)
Type
Annual days ∗
−6 −1 −28∗ +28∗ +2 +8 −5 +2
C HYC UC A/HYA UA W/NW/SW/N E/NE S/SE
Winter days ∗
+2 +1 −2∗ +3 −9∗ +14∗ −6∗ −2∗
Spring days
Summer days
Autumn days
∗
−3 −1 −9∗ +12∗ +2 +4 −6∗ +1
−5 −1 −15∗ +8∗ +6∗ <1 +8∗ −1
0 <1 −1 +6∗ +2 −10∗ −1 +4∗
Table 18.14 Mean seasonal changes (perturbed − control run) in the frequency (number of days) of the eight circulation-type groups. Changes that are significant at the 5% (∗∗ ) or 10% (∗ ) level are indicated
Type C HYC UC A/HYA UA W/NW/SW/N E/NE S/SE
Winter
Spring
Summer
Autumn
−0.5 +0.4 −0.8 0.0 +3.5 −4.5 −0.2 +2.1∗
−0.7 −1.5 +0.3 −0.8 +2.3 +3.1 −1.4∗ −1.1
+4.4∗∗ +3.1∗ +3.6 −4.3∗∗ −7.3∗∗ −1.0 +1.1 +0.4
−0.8 −0.4 +2.4 −2.7 −2.7 +3.5∗ −0.4 +1.2
above circulation types are all high-rainfall types (Table 18.12). The frequency of the low-rainfall A/HYA type is overestimated in every season, as is the frequency of the UA type in every season except winter and the frequency of the W/NW/SW/N group in spring and autumn. The ability of the UKTR GCM model to reproduce the observed circulation-type frequencies has also been assessed by plotting the mean seasonal cycle simulated by the model against the observed seasonal cycle (Figure 18.8). Only 10 years of daily model output are available so the minimum and maximum frequencies observed in any decade during the period 1956–1989 are indicated by the hatched area. Whenever a simulated curve falls within this area the model is considered to be performing well. For example, the model reproduces the seasonal cycle of the S/SE type reasonably well except in autumn but is less successful at simulating the E/NE type. 5.5 Future Changes in Circulation-type Frequency
Seasonal and monthly frequencies of the eight circulation-type groups have been calculated using SLP data from the final decade of the perturbed UKTR model run. The mean seasonal changes (perturbed-run minus control-run) for the eight circulation-type groups are shown in Table 18.14. Only a few of these frequency changes are significant, with the largest changes tending to occur in summer. In this season there is a significant increase in the frequency of the C and HYC types. The latter is a high-rainfall type (Table 18.12). The frequency of the high-rainfall E/NE and S/SE types also increases in summer, but the changes are not significant. The A/HYA and UA groups are both low-rainfall types and show a significant decrease in frequency in summer. In order to investigate how these frequency changes translate into changes in the number of rain days, a weather generator is used, as described in the next section.
266
Mediterranean Desertification
5.6 Using a Weather Generator to Produce Rain Day Scenarios Changes in the number of rain days associated with the circulation-type changes have been investigated using a conditional weather generator (CWG) in which the probability of rain is conditional on the circulation type occurring on each day. Two parameters are calculated for each of the eight circulation types: the probability of rain (PREC ct ) and the cumulative probability of the next day being of a particular type (PROBct 1 – 8 ). The cumulative probabilities (PROBct 1 – 8 ) are taken from a transition matrix for each station and each season. Then, to apply the CWG, on each day a random number between 0 and 1 is selected. This is used to determine the next day’s circulation type. A second random number between 0 and 1 is selected and used to determine whether the day is wet or dry. The CWG is run for 30 days to remove the memory of the arbitrarily selected starting point before starting to record results. For the purposes of the sensitivity experiments described here, we are only interested in the number of rain days. The amount of rain could, however, be simulated by selecting a third random variable for wet days from an appropriate distribution. Three sets of 100 30-year-long simulations have been performed for each of the six Guadalent´ın Basin stations listed in Table 18.1. The sequence of circulation types in each 30-year simulation is dependent on the transition matrix and the random number generator, which means that the 100 sequences making up each set should be different. This allows a wider range of possible future outcomes to be considered than if the circulation types were taken directly from the GCM. The CWG and the output from the three simulation sets are described in detail elsewhere (Goodess and Palutikof 1998). Here, the main results are summarized. In the first simulation set, the CWG parameters (PROBct 1 – 8 and PREC ct ) are calculated from the observations. All available data for 1958–1987 are used in order to maximize the sample sizes. The output from these simulations cannot, therefore, be used for independent validation, but can be used to explore the effect of the CWG on second-order statistics such as variance and persistence. As expected, the mean values are well simulated. In common with other weather generators, variance and persistence tend to be underestimated by the CWG. In the second and third simulation sets, the PROBct 1 – 8 parameter is calculated from the control and perturbed runs respectively of the UKTR GCM. PREC ct is calculated from the observations because the GCM cannot realistically simulate the occurrence of daily rainfall. The second simulation set allows further investigation of the GCM’s ability to simulate the circulation-type frequencies and to determine how this affects the simulated number of rain days (Goodess and Palutikof 1998). The differences between the simulated and observed circulation-type frequencies closely follow those shown in Table 18.14. These errors are propagated through to the simulated number of rain days, which are underestimated by the CWG. The largest rain-day errors occur in winter and spring. The frequency of the high-rainfall UC and E/NE types is underestimated in these seasons, while the frequency of the low-rainfall A/HYA and W/NW/SW/N types is overestimated. Given the discrepancies between the control-run CWG simulations and the observations, the climate-change scenarios are calculated as perturbed-run minus control-run differences. The change (averaged over 100 simulations) in the number of rain days for each season and each station is shown in Table 18.15. The pattern of change is consistent for all stations. These scenarios indicate that the Table 18.15 Mean change (perturbed run − control run) from the 100 simulations of the mean number of rain days simulated by the CWG for six stations in the Guadalent´ın Basin. Significant changes are indicated (*)
Aguilas Alcantarilla Alhama de Murcia Fuente Alamo Lorca Totana
Winter
Spring
Summer
Autumn
+0.3∗ +0.3∗ +0.3∗ +0.2∗ +0.3∗ +0.3∗
−0.7∗ −0.7∗ −1.0∗ −0.5∗ −0.8∗ −0.7∗
+0.3∗ +0.7∗ +0.6∗ +0.6∗ +0.6∗ +0.4∗
00.0 +0.3∗ +0.1 +0.1∗ +0.1 00.0
Local and Regional Responses to Global Climate Change
267
average number of rain days in summer in the Guadalent´ın Basin could increase significantly (by 10–18%), and decrease significantly (by 5–9%) in spring. A small but significant increase (2–4%) is indicated in winter, and little change in autumn (0– 2%).
6 IMPLICATIONS OF THE TEMPERATURE AND RAIN-DAY SCENARIOS ´ BASIN FOR THE GUADALENTIN The temperature scenarios indicate a reduction in the number of frost days and freezing degree days for Alcantarilla (Table 18.9). These changes would be beneficial, but the scenarios also indicate a substantial increase in the number of heat stress days (defined as days ≥ 35 ◦ C). The pattern of rainfall change is unlikely to be beneficial for the Guadalent´ın Basin. Fewer rain days are indicated during spring (Table 18.15) when water is required for groundwater recharge and crop growth. An increase in the number of rain days is indicated in summer. This is the period of non-growth and also the period when the soil surface is most vulnerable to erosion. The scenarios presented here are intended as illustrative results rather than as reliable predictions but are, none the less, considered to be more plausible than the “raw” grid-point output from the GCMs. In some cases, the downscaled climate changes are of the opposite sign to the grid-point changes. For example, grid-point rain-day output from the UKTR model for the square closest to the Guadalent´ın Basin indicates little change or an insignificant decrease in the number of rain days in winter and autumn, a greater decrease in spring (in general agreement with the downscaled scenarios presented in Table 18.15), and the largest percentage decrease (about 60%) in summer, whereas the downscaled scenarios indicate a significant increase in this season. The downscaling methods developed and tested in the Guadalent´ın Basin are both capable of providing climate scenarios at the necessary spatial (i.e. individual station level) and temporal (i.e. daily) scales for input to the various process models developed during the MEDALUS project. These methods are parsimonious of computer time, do not require large amounts of observed data, and have the potential to be applied in other Mediterranean regions that may be vulnerable to desertification.
ACKNOWLEDGEMENTS The UKTR GCM data have been supplied by the Climate Impacts LINK Project (Department of the Environment Contract EPG 1/1/16) on behalf of the Hadley Centre and UK Meteorological Office. We express our appreciation to the Canadian Climate Centre for making their model output readily available to impacts researchers. We thank Susan Adcock and Sarah Watkins for their help in the preparation of Figure 18.2 and Figures 18.5–18.7 respectively. The gridded rainfall data plotted in Figure 18.3 were provided by Mark New.
REFERENCES Boer GJ, McFarlane NA and Lazare M (1992) Greenhouse gas-induced climate changes simulated with the CCC second-generation general circulation model. Journal of Climate 5, 1045–1077. Cubasch U, von Storch H, Waszkewitz J and Zorita E (1996) Estimates of climate change in southern Europe derived from dynamical climate model output. Climate Research 7, 129–149. Giorgi F, Marinucci MR and Visconti G (1990) Use of a limited-area model nested in a general circulation model for regional climate simulation over Europe. Journal of Geophysical Research 95, 18 413–18 431. Giorgi F, Marinucci MR and Visconti G (1992) A 2 × CO2 climate change scenario over Europe generated using a limited area model nested in a general circulation model. 2 Climate change scenario. Journal of Geophysical Research 97, 10 011–10 028. Glahn HR (1985) Statistical weather forecasting. In AH Murphy and RW Katz (eds) Probability, Statistics and Decision Making in the Atmospheric Sciences. Westview Press, Boulder, Colorado, pp. 289–366. Goodess CM and Palutikof JP (1998) Development of daily rainfall scenarios for southeast Spain using a circulation-type approach to downscaling. International Journal of Climatology, 18, 1051–1083. Hewitson BC and Crane RG (1992a) Regional-scale climate prediction from the GISS GCM. Global and Planetary Change 97, 249–267.
268
Mediterranean Desertification
Hewitson B and Crane RG (1992b) Regional climates in the GISS global circulation model: synoptic-scale circulation. Journal of Climate 5, 1002–1011. Hewitson BC and Crane RG (1996) Climate downscaling: techniques and application. Climate Research 7, 85–95. Houghton J, Meiro Filho LG, Callander BA, Harris N, Kattenberg A and Maskell K (eds) (1996) Climate Change 1995. The Science of Climate Change. Cambridge University Press, Cambridge. Hulme M, Briffa KR, Jones PD and Senior CA (1993) Validation of GCM control simulations using indices of daily airflow types over the British Isles. Climate Dynamics 9, 95–105. Jenkinson AF and Collinson P (1977) An Initial Climatology of Gales over the North Sea. Synoptic Climatology Branch Memorandum No 62, Meteorological Office, London. Jones PD, Hulme M and Briffa KR (1993) A comparison of Lamb circulation types with an objective classification scheme. International Journal of Climatology 13, 655–663. Jones RG, Murphy JM and Noguer M (1995) Simulation of climate change over Europe using a nested regionalclimate model. Part 1: Assessment of control climate, including sensitivity to location of lateral boundaries. Quarterly Journal of the Royal Meteorological Society 121, 1413–1449. Karl TR, Wang W-C, Schlesinger ME, Knight RW and Portman D (1990) A method of relating general circulation model simulated climate to the observed local climate. Part 1: Seasonal statistics. Journal of Climate 3, 1053–1079. Kattenberg A, Giorgi F, Grassl H, Meehl GA, Mitchell JFB, Stouffer RJ, Tokioka T, Weaver AJ and Wigley TML (1996) Climate models – projections of future climate. In JT Houghton, LG Meira Filho, BA Callander, N Harris, A Kattenberg and K Maskell (eds) Climate Change 1995, The Science of Climate Change. Cambridge University Press, Cambridge, pp. 285–357. Klein WH (1982) Statistical weather forecasting on different time scales. Bulletin of the American Meteorological Society 63, 170–177. K¨oppen W (1936) Das geographische system der klimate. In W K¨oppen and R Geiger (eds) Handbuch der Klimatologie 3 . Gebr¨uder Borntraeger, Berlin. Lamb HH (1972) British Isles Weather Types and a Register of Daily Sequence of Circulation Patterns 1861–1971 . Geophysical Memoir 116, HMSO, London. McFarlane NA, Boer GJ, Blanchet JP and Lazare M (1992) The Canadian Climate Centre second-generation general circulation model and its equilibrium climate. Journal of Climate 5, 1013–1044. McGuffie K and Henderson-Sellers A (1996) A Climate Modelling Primer. John Wiley, Chichester. McKinlay DA (1996) Country by country listing of extreme climatic events in 1996. Climate Monitor 25, 182–200. Mitchell JFB, Johns TC, Gregory JM and Tett SFB (1995) Climate response to increasing levels of greenhouse gases and sulphate aerosols. Nature 376, 501–504. Murphy JM (1995) Transient response of the Hadley Centre coupled ocean–atmosphere model to increasing carbon dioxide. Part I: control climate and flux adjustment. Journal of Climate 8, 36–56. Murphy JM and Mitchell JFB (1995) Transient response of the Hadley Centre coupled ocean–atmosphere model to increasing carbon dioxide. Part II: spatial and temporal structure of response. Journal of Climate 8, 57–80. New M, Hulme M and Jones PD (1998a) Representing twentieth-century space–time climate variability. I: Development of a 1961–90 mean monthly terrestrial climatology. Journal of Climate 12, 829–856. New M, Hulme M and Jones PD (1998b) Representing twentieth-century space–time climate variability. II. Development of 1901–96 monthly grids of terrestrial surface climate. Journal of Climate 13, 2217–2238. Palutikof JP, Conte M, Casimiro Mendes J, Goodess CM and Espirito Santo F (1996) Climate and climatic change. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 43–86. Palutikof JP, Winkler JA, Goodess CM and Andresen JA (1997) The simulation of daily temperature time series from GCM output. Part I: comparison of model data with observations. Journal of Climate 10, 2497–2513. Schubert S and Henderson-Sellers A (1997) A statistical model to downscale local daily temperature extremes from synoptic-scale atmospheric circulation patterns in the Australian region. Climate Dynamics 13, 223–234. von Storch H, Zorita E and Cubasch U (1993) Downscaling of global climate change estimates to regional scales: an application to Iberian rainfall in wintertime. Journal of Climate 6, 1161–1171. Wigley TML, Jones PD and Kelly PM (1986) Empirical climate studies: warm world scenarios and the detection of climate change induced by radiatively active gases. In B Bolin, BR Doos, J J¨ager and RA Warrick (eds) The Greenhouse Effect, Climatic Change and Ecosystems. SCOPE 26, John Wiley, New York, pp. 271–322. Winkler JA, Palutikof JP, Andresen JA and Goodess CM (1997) The simulation of daily temperature time series from GCM output. Part II: Sensitivity analysis of an empirical transfer function methodology. Journal of Climate 10, 2514–2532.
19
The Impact of Land Abandonment on Regeneration of Semi-natural Vegetation: A Case Study from the Guadalent´ın
J.A. OBANDO
Department of Geography, Kenyatta University, Nairobi, Kenya
1 INTRODUCTION In this chapter changes in the semi-natural vegetation in the Guadalent´ın Basin in south-east Spain are discussed in the context of changing climate and socio-economic activities. The impact of abandonment of agricultural land on the regeneration of semi-natural vegetation is presented using Rambla del Chortal as a case study.
2 LAND ABANDONMENT IN SOUTH-EAST SPAIN Land use over much of the Mediterranean and indeed most parts of the world has been changing over the past 50 years, sometimes at very rapid rates. The vegetation changes that have occurred have been induced by a combination of climatic stress, and natural and human disturbances. As a result, some plants have been able to adopt survival strategies including resilience to invasion (Naveh and Vernet 1991). Agriculture, pastoralism, logging and hunting all made their demands on the land during the 20th century. Deforestation to clear land for agriculture and later overgrazing of the diminished vegetation are the main events that have contributed to the badlands so characteristic of areas in south-east Spain (McNeill 1992). The increasing environmental degradation has been accompanied by deteriorating water quality and a general decline in freshwater resources (Hamdy et al. 1995). According to McNeill (1992), the imprint of human activity remained minimal until the ecological pressures generated by population growth and market integration concentrated the increase in demand. Thereafter, the degradation of vegetation and landscapes declined sharply, and continues to do so today wherever these pressures remain. The area of land directly used for agriculture has been decreasing in Spain over the past few decades. The changes in the landscape have been mainly due to social and economic changes in Spain, from both internal and external forces. McNeill (1992) has given evidence of landscape change and argues that the physical landscapes of the Mediterranean are often recent creations. Fernandez Ales et al. (1992) have shown that the economic development in south-west Spain over the period 1950–1984 led to intensive agricultural systems being concentrated in the more fertile areas, while marginal areas were abandoned, altering the landscape structure and function of the region. The tendency has been towards the abandonment of the less fertile areas and intensification of farming in the remaining areas as the increase in population has occurred in most regions in Spain. The landscape in Rambla del Chortal Basin is dominated by patches of field that have been abandoned for different lengths of time, as shown by a variety of vegetation types. Abandonment refers to land that has been converted from any form of agricultural production or from areas that have been heavily grazed and allowed to revegetate naturally. Land is also considered abandoned when grazing pressure on heavily grazed land is reduced. Complete abandonment implies that the Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
270
Mediterranean Desertification
land has been left to return to its natural state without any human influence, directly or from livestock. In ecological studies (Baudry 1991), land abandonment is used in a much broader sense to refer to change in land use from the traditional or recent pattern to another, less intensive pattern. Large areas of the Guadalent´ın have been abandoned due to both climatic aridity and changing socio-economic aspects of the population.
3 3.1
IMPACT OF LAND ABANDONMENT ON REGENERATION OF VEGETATION Field Experimental Set-up
The impact of land abandonment on semi-natural vegetation is discussed using field data collected in Rambla del Chortal in the Guadalent´ın Basin in south-east Spain. The field sites were selected to represent five major classes of abandonment on two lithologies as shown in Table 19.1. The selected sites within the basin have been abandoned for different lengths of time and were assumed to represent a temporal sequence. However, differences occur along the transects due to varying aspect, position on slope, slope angle and soil depth. This chronosequence approach (Miles 1979; Glenn-Lewin and van der Maarel 1992) has been used by ecologists to describe regeneration in the field. Although an ideal chronosequence requires the areas to have undergone a similar history since abandonment, it is rare to find such conditions in reality. Often the sites may not have experienced the same microclimate, soil conditions, history and availability of plant propagules at the time of abandonment. However, for the length (50 years) and the spatial area (20 km2 ) of study under investigation here, the assumptions of similar soil conditions and availability of propagules are considered to be acceptable. Also, the interpretation of the contemporary plots of vegetation representing different stages of abandonment was aided by existing historical information of the area and local knowledge from the farmers in order to minimize the errors arising from the assumptions. Transects (each one representing an area of 30 m2 ) were selected randomly from the top, middle and footslope (bottom) sections of the slope profile. These profiles were selected from the footslopes along the steepest path to the top of the slope, and varied in length from 29 to 190 m. Every effort was made to select similar sites on the two lithologies chosen, although this proved to be difficult mainly because one site still had human occupation and heavy grazing was evident. The sampling scheme employed for fieldwork is shown in Figure 19.1. Further details on the field area are given in Obando (1996). Field data were collected during two main periods to coincide with minimum and maximum vegetation peaks in the catchment. Measurements were made of surface cover characteristics along selected transects. The first data set was collected in the dry early autumn, just before the onset of the rains, when it is assumed that the vegetation cover is at a minimum. The second field campaign was carried out in the spring of 1994, after the rains and when the vegetation cover is usually at its maximum. Vegetation data were collected on surface cover characteristics. Standard measurements for cover were made in reference to Kershaw (1985) and Mueller-Dombois and Ellenberg (1974). Identification of species was done using several texts including Garcia-Guardia Table 19.1 Categories of abandonment (years), number of transects and legend for sites
Category
Age of abandonment
Shale lithology No. of transects
Early fallow Young fallow Mature fallow Old fallow Very old fallow
<5 5–10 10–20 20–40 >40
None 6 8 12 4
Site
G2 G3 G4 G5
Phyllite lithology No. of transects
Site
3 6 6 12 12
S1 S2 S3 S4 S5
Impact of Land Abandonment on Vegetation Regeneration
271
Age of abandonment Lithology Aspect
Slope profile
Topslope
Figure 19.1
Midslope
Footslope
The selection of sampled areas
(1988) and Polunin and Smythies (1973). The surface cover measurements on the abandoned sites were made along selected transects measuring 30 m in length. The measurements included seasonal attributes of vegetation cover such as plant height and crown, species composition and plant life forms. The crown measurements were made perpendicular to the transect line for each of the plant life forms. The heights for the plants on the transects were also measured, all measurements being made to the nearest centimetre. In the case of trees or canopies above 2 m in height, estimates were made from ground level. The species richness and diversity was assessed within the sites of different ages of abandonment by noting along the transects their absence and presence. In addition, quadrat measures supplemented the transect data for the species composition. The species composition was then calculated as percentages of the total species sampled in the study area. Species diversity and richness measures are frequently seen as indicators of the well-being of ecological systems (Magurran 1987) and can therefore be used to determine the recovery of the vegetation following abandonment. The surface elevations were measured using a micro-relief meter along and across the selected transects. Surface stoniness was measured using 1 m2 meshed quadrats with 36 interception points. Soil samples were analysed to determine the soil properties that influence the hydrology and eventually influence the vegetation as it regenerates on the abandoned sites. These included the particle size distribution of the soil, soil moisture and bulk density. However, only data relating directly to vegetation regeneration are important here. 3.2 Vegetation Characteristics
A lot of research has focused on the vegetation types and cover of the Mediterranean ecosystems, for example by Godron et al. (1981), Di Castri et al. (1981), Guillerm (1991), Le Hou´erou (1991) and Di Castri (1991). The vegetation in the Mediterranean region is characterized by a dominance of woody shrubs with evergreen leaves that are broad and small, stiff and sticky (Di Castri 1991). The dominant natural vegetation in Chortal at the time of study was matorral scrub occurring with scattered trees of Pinus halepensis and bushes of Quercus coccifera in the higher areas of altitude. The scrub area was dominated by Anthyllis cytisoides and Rosmarinus officinalis. Local changes in economic structure and policies have been the main reasons for increases in the area abandoned. Generally, changes in land use in Chortal from 1956 to the present have seen an increase in the land under afforestation from 1.8% to 21.3% (L´opez-Berm´udez et al. 1996). The dry land crops are mainly cereals and almond trees. The matorral vegetation measured on the transects in Rambla del Chortal had an average height of 1.5 m. The species count revealed differences in the species numbers on the transects sampled with an average of 20 species. The commonly occurring species are mainly the shrub and herb types. The main shrub species include Rosmarinus officinalis, Anthyllis cytisoides, Thymus zygis, Helianthemum
272
Mediterranean Desertification
almierse, Cistus albidus and Artemisia campestris. Carline species also occur, mainly in the recently abandoned sites. The grass species were not very common, with the main species sampled being Brachypodium dystachion. There were a few trees found in the old abandoned sites. The most common was Pinus halepensis. Quercus coccifera, even in areas abandoned for 20–40 years, tends to be very young and shrubby in structure rather than reaching tree height. The species composition differed both within and between sites of the same age of abandonment. The species composition and surface cover characteristics were influenced by the topography (aspect and slope), seasonal rainfall variations as well as length of time since abandonment. 3.3
Vegetation Regeneration The results obtained from the transects of the per cent vegetation cover illustrate an increase in the vegetation cover following abandonment (Figure 19.2) for both the autumn and spring data. The assumption is that a bare surface exists at the time of abandonment, following agricultural use. Clearly, the vegetation cover rapidly increases in the initial years before maintaining a cover that tends towards an unspecific equilibrium. The equilibrium is easily shifted by seasonal climatic variations of rainfall and temperature. Such variations can be on a very local scale, and will be further affected by other physical site characteristics, such as soil depth and aspect. As expected, percentage vegetation cover was greater in spring, following the wettest months, than in autumn at all sites. There were interesting differences between sites on shales and the sites on phyllites (Table 19.2). Examination of the ranges of percentage cover between spring and autumn could suggest that plant colonization is easier on the phyllites than on the shales, but in fact 40% of the transects on the shales had experienced overgrazing and consequent reduction of their potential vegetation cover. Mean percentage cover on the shales in spring was only marginally more than on the phyllites in autumn. The percentage cover measured varies (Figure 19.3 and Table 19.2) due to lithology, with a higher mean from the phyllite sites than from the shale sites. A mean seasonal change in the vegetation cover over the period for both sites was 22%, although the seasonal change on the shale sites is higher than on the phyllite sites (Table 19.2). The vegetation cover on the phyllite (Figure 19.3(a)) increases to a mean value of 48% on the sites abandoned after 40 years, and slightly decreases in the last stage of abandonment to 43%. On the phyllite sites, for example, the increase in plant cover with age of abandonment was found to be rapid in the early stages. Nevertheless, the very old sites had less cover than the old sites, which have been abandoned for a shorter time period. The reasons for this could relate to the selection of the sites. The vegetation on the very old sites may have died back due to a decrease in the rainfall 70 60
% cover
50 40 30 20 10 0 0
5
10
15
20
25
30
35
40
45
50
Years since abandonment Autumn 93
Spring 94
Figure 19.2 Increasing cover with abandonment; measured in Rambla del Chortal for two seasons
273
Impact of Land Abandonment on Vegetation Regeneration Table 19.2 Vegetation cover percentage measured from the sites
Vegetation cover %
Mean
All sites, both seasons Autumn 1993 Shales Phyllites Spring 1994 Shales Phyllites
42.5 35.6 27.43 41.83 49.8 45.35 53.66
Mean seasonal change
Standard deviation
Minimum
Maximum
No. of transects
29% 40% 22%
14.3 12.7 8.39 11.89 12.2 9.87 12.82
9.8 9.8 13.1 9.8 23 23 32.8
80 59 44.3 59 80 62.3 80
134 69 30 39 65 30 35
(a) Phyllite sites 70 60 % cover
50 40 30 20 10 0 0
5
10
15
20
25
30
35
40
45
50
Years since abandonment (b) Shale sites 70 60
% cover
50 40 30 20 10 0 0
5
10
15
20
25
30
35
40
45
50
Years since abandonment
Figure 19.3 (a) Vegetation cover percentage on the phyllite sites in autumn 1993. The bold line represents the mean for 39 transects. (b) Vegetation cover percentage on the shale sites in autumn 1993. The bold line represents the mean for 30 transects. The discontinuous lines indicate incomplete data from the overgrazed transects (shown as open triangles)
reducing the cover to low levels. On the shale (Figure 19.3(b)), an increase also occurs in the per cent vegetation cover in the early stages before approaching a steady-state value after 20 years of abandonment. The shale sites have a much lower vegetation cover than the phyllite sites. This may be attributed to the heavy overgrazing observed on 40% of the transects on the shale sites. The vegetation cover on the shale sites beyond 40 years of abandonment increases slightly, while on
274
Mediterranean Desertification
the phyllite sites a slight decrease occurs. This demonstrates the dynamic nature of the vegetation and shows that a steady-state value is not necessarily constant, but can change with, for example, seasonal elements of climate. Figure 19.3 shows the effects of land abandonment on autumn percentage vegetation cover for transects on phyllites and on shales. The minimum autumn values are used for this comparison because they vary less between years compared to the spring values, which are more dependent on the previous incidence of effective rainfall. The means for the transects suggest that the equilibrium or asymptote percentage cover reaches a higher value (around 45%) over the phyllite than over the shales, but it takes longer, up to 20 years, for such an equilibrium to be achieved. Over the shales, the percentage vegetation cover at equilibrium is lower, at around 25% cover, but the equilibrium may be achieved within 10 years. As explained previously, the equilibrium or asymptote is very fragile, and large seasonal and annual differences attributable to other factors frequently occur (Miles 1979; Peet 1992). In semi-arid areas such as the Rambla del Chortal, availability of soil moisture is the main factor limiting plant productivity. The significance of limiting factors varies according to the age of the vegetation. For example, young vegetation with a high growth rate may be more affected by drought than more established vegetation. This was observed following a period of drought in the 1960s. The asymptote or quasi-equilibrium is also affected by the length of the life cycles of individual species. All plants go through growth, maturity and senescent phases, with maximum cover observed at the mature stage. The onset of senescence will be determined by the genetic make-up of the plant as well as environmental factors. Therefore loss of plant cover cannot be related to any one factor such as drought. Figure 19.2 suggests that on an area abandoned for more than 30 years, senescence of some of the original plants may reduce the percentage plant cover from the values seen 10 years previously. The results obtained are consistent with the general models proposed for vegetation succession, for example by Godron et al. (1981), Miles (1979), Crawley (1986), Peet (1992), Glenn-Lewin and van der Maarel (1992) and Veblen (1992). Godron et al. (1981) show a logarithmic increase in total above-ground phytomass following disturbance by fire. Although the models proposed by Peet (1992) relate to vegetation succession in forests, they are broadly similar to the field data obtained here, with an initial increase in the vegetation approaching an asymptotic value. The differences occur in that having reached an asymptote, the biomass may decrease having reached a peak associated with maturity. According to Peet (1992), it is likely that frequently burned shrublands such as Californian chaparral exhibit patterns similar to those documented for forests. As biomass approaches steadystate levels, biomass increment and thus uptake of nutrients drops, while losses through death and decomposition increase. 3.4
Vegetation Regeneration and Plant Strategies
Effective regeneration of vegetation following abandonment first requires short-lived species that can grow quickly, providing a ground cover to help conserve soil moisture that will allow more long-lived species to become established. This is perceived as a plant strategy, with the earliest plants at a site referred to as r, and the following, more woody, species as K (Grime 1979). The tendency is for r type species to dominate in the early years of abandonment, as demonstrated using the autumn field data in Figure 19.4 for both sites. The more woody K species type gradually replaces the r type as the competition ensues for limited resource (Grime 1979; Miles 1979; Crawley 1986). Competition occurs because of limited resources (Whittaker 1967; Crawley 1986; Tilman 1988; Peet 1992). According to Tilman (1988), the two major factors that determine the availabilities of a limiting soil resource and light in a habitat are the rate of supply of the soil resource and the loss or mortality rate that plants experience. Whittaker (1967) used the distributions of species along an environment gradient to explain the effect of interactions and competition for a limited resource. This concept can be used to explain the difference between the shallow-rooted annuals and herbs (r) and the deeper rooting woody shrubs (K) competing for soil moisture. This result is useful in modelling vegetation dynamics for the different species types. The r type species include the herbs
275
Impact of Land Abandonment on Vegetation Regeneration 100
% cover
80 60
r
40
K
20 0
0
10
20
30
40
50
Years
Figure 19.4 Vegetation cover percentage from sites in Rambla del Chortal differentiated into r and K type species
and annuals while the K type species include the shrub species, which are woody in nature. The r type species tend to make use of water on the surface or in the upper layers of the soil while the K type species have the advantage of the deeper soil water. Thornes and Brandt (1994) use this differentiation in modelling vegetation competition. Tilman (1988), on the other hand, has used a model for allocation which assumes that plant growth is determined by the pattern of allocation of carbon to roots, leaves, stems and seeds. Such models are useful in understanding the vegetation changes in the context of changing climate. Understanding the interactions between vegetation and erosion and their response to changes in climate has important implications for management of land systems.
4 SUMMARY AND CONCLUSIONS The semi-natural vegetation in south-east Spain is in a state of constant change. The dynamic nature of this vegetation is affected by both climatic and socio-economic changes. Some factors, such as drought, deforestation and overgrazing act towards decreasing vegetation cover. However, it has been shown that land abandonment can play an important role in restoring vegetation cover. Increases in plant productivity and biomass generally lead to an improvement of the hydrological conditions and hence a decrease in soil erosion and land degradation in the long term. The changes seen following land abandonment occur not only in the percent vegetation cover, but also in variety and abundance of species. Other factors which affect the semi-natural vegetation include lithology, rainfall, aspect and slope. Aspect exerts a major influence on the variations of the surface cover: within Rambla del Chortal, a lower plant cover surface was found on the south-facing slopes than on the north-facing slopes on both lithologies. Further, human disturbances through agriculture practices and grazing also affect the semi-natural vegetation especially in semi-arid areas.
ACKNOWLEDGEMENT The author wishes to acknowledge the Commonwealth Association of Universities in the UK for awarding the scholarship under which this research was undertaken.
REFERENCES Baudry J (1991) Ecological consequences of grazing extensification and land abandonment: role of interactions between environment, society and techniques. In J Baudry and RGH Bunce (eds) Land Abandonment and its Role in Conservation. Options M´editerran´eennes-S´erie S´eminaires 5, CAB International, pp. 13–19. Crawley MJ (1986) Life history and environment. In MJ Crawley (ed.) Plant Ecology. Blackwell Scientific, Oxford, pp. 253–290.
276
Mediterranean Desertification
Di Castri F (1991) An ecological overview of the five regions with a Mediterranean climate. In RH Groves and F Di Castri (eds) Biogeography of Mediterranean Invasions. Cambridge University Press, Cambridge, pp. 3–16. Di Castri F, Goodall DW and Specht RL (eds) (1981) Mediterranean-Type Shrublands. Elsevier, Amsterdam. Fernandez Ales R, Martin A, Ortega F and Ales Enrique E (1992) Recent changes in landscape structure and function in a Mediterranean region of SW Spain (1950–1984). Landscape Ecology 7(1), 3–18. Garcia-Guardia G (1988) Flores Silvestres de Andalucia. Madrid. Glenn-Lewin DC and van der Maarel E (1992) Patterns and processes of vegetation dynamics. In DC GlennLewin, RK Peet and TT Veblen (eds) Plant Succession: Theory and Prediction. Chapman & Hall, London, pp. 11–59. Godron M, Guillerm JL, Poissonet J, Poissonet P, Thaiault M and Trabaud L (1981) Dynamics and management of vegetation. In F Di Castri, DW Goodall and RL Specht (eds) Mediterranean-Type Shrublands. Elsevier, Amsterdam, pp. 317–344. Grime JP (1979) Plant Strategies and Vegetation Processes. John Wiley, Chichester. Guillerm JL (1991) Weed invasion in agricultural lands. In RH Groves and F Di Castri (eds) Biogeography of Mediterranean Invasions. Cambridge University Press, Cambridge, pp. 26–992. Hamdy A, Abu-Zeid M and Lacirignola C (1995) Water resources in the Mediterranean. Water Resources Development 11(4), 515–526. Kershaw KA (1985) Quantitative and Dynamic Plant Ecology. Edward Arnold, London. Le Hou´erou HN (1991) Plant invasions in the rangelands of the isoclimatic Mediterranean zone. In RH Groves and F Di Castri (eds) Biogeography of Mediterranean Invasions. Cambridge University Press, Cambridge, pp. 393–404. Lop´ez-Berm´udez F, Romero D´ıaz MA, Mart´ınez-Fern´andez J, Mart´ınez-Fern´andez JM, Alonso-Sarria F and Belmonte-Serrato F (1996) Field site: Murcia, Spain MEDALUS II Project 1, Basic Field Programme, Final Report (1991–1995). EV5V-CT92-0128, pp. 38–60. Magurran E (1987) Ecological Diversity and its Measurement . Croom Helm. McNeill JR (1992) The Mountains of the Mediterranean World: An Environmental History. Cambridge University Press, Cambridge. Miles J (1979) Vegetation Dynamics. Outline Studies in Ecology, Chapman & Hall, London. Mueller-Dumbois D and Ellenberg H (1974) Aims and Methods of Vegetation Ecology . John Wiley, New York. Naveh Z and Vernet JL (1991) The palaeohistory of the Mediterranean biota. In RH Groves and F Di Castri Biogeography of Mediterranean Invasions. Cambridge University Press, Cambridge, pp. 19–31. Obando JA (1996) Modelling the impact of land abandonment on runoff and soil erosion in a semi-arid catchment. PhD Thesis, Kings College, University of London. Peet RK (1992) Community structure and ecosystem function. In DC Glenn-Lewin, RK Peet and TT Veblen (eds) Plant Succession: Theory and Prediction. Chapman & Hall, London, pp. 103–151. Polunin O and Smythies BE (1973) Flowers of South-west Europe: A Field Guide. Oxford University Press, Oxford. Thornes JB and Brandt J (1994) Erosion–vegetation competition in a stochastic environment undergoing climatic change. In AC Millington and K Pye (eds) Environmental Change in Drylands: Biogeography and Geomorphological Perspectives. John Wiley, Chichester, pp. 305–320. Tilman D (1988) Plant Strategies and the Dynamics of Structure of Plant Communities. Monographs in Population Biology 26, Princeton University Press, Princeton, New Jersey. Veblen TT (1992) Regeneration dynamics. In DC Glenn-Lewin, RK Peet and TT Veblen (eds) Plant Succession: Theory and Prediction. Chapman & Hall, London, pp. 152–187. Whittaker RH (1967) Gradient analysis of vegetation. Biological Review 49, 207–264.
20
Lithology and Vegetation Cover Mapping in the Guadalent´ın Basin as Interpreted through Remote Sensing Data
1 ´ 1 M.A. GILABERT,1 F.J. GARCIA-HARO ´ M.T. YOUNIS,1 J. MELIA, AND A.J. BASTIDA2
1 2
` Remote Sensing Unit, Universitat de Valencia, Spain ` Departamento de Geolog´ıa, Universitat de Valencia, Spain
1 INTRODUCTION AND OBJECTIVES Desertification processes in the environment have stimulated many global monitoring programmes to examine the possible consequences of vegetation cover variation, which is one of the most important ecosystem parameters (Tucker et al. 1985). Vegetation cover is the variable controlling soil erosion activity that is most subject to human manipulation and is therefore an important component of any predictive model (Trimble 1990). Accordingly, there was a requirement for reliable land cover mapping in the MEDALUS II project. Both lithology and vegetation play an important role in studies of erosion and desertification. Aspects such as the use of vegetation cover parameters in modelling (e.g. spectral mixture models for the study of the vegetation dynamics), the response of the lithology and vegetation types to runoff and erosion, their relation to the basin hydrogeomorphology and, finally, the determination of Environmentally Sensitive Areas (ESAs) have been monitored. The lithology–soil cover has a fundamental importance in the determination of both sustainability for desertification processes and vegetation properties in arid and semi-arid areas. In the vegetationorientated applications using remotely sensed data, lithology–soil optical properties influence the radiometric response of canopies since the soil is the final layer or background (Meli´a et al. 1993). For sparse canopies, widely represented over the Earth’s surface, reflectance is very sensitive to the soil optical property changes (Baret et al. 1993). Thus the spatial variation of the lithology and soil types is very important in order to predict the vegetation properties. The derivation of biophysical properties of the Earth’s surface by direct inversion of remotely sensed data is ultimately the most desirable objective. Basic data regarding the extent and dynamics of vegetation are still needed, and better assessment of natural or human-caused changes in the vegetation cover of the Earth is crucial to understand the role of plant communities in climatic, hydrological and geochemical cycles (Malingreau 1986). Vegetation cover maps constructed using field observation and traditional methods of field survey and aerial photographs lack, in most cases, provision for updating the map as change happens. However, this is essential for models treating the desertification process and environmental changes. On the other hand, geological maps mostly indicate structural features and group lithology types in formations according to their geological age. These maps offer a general idea about surface composition and consequently an improved composition map is of great importance in desertification studies. This raises the need for updating the vegetation cover and lithology maps to deduce changes in the state of the vegetation cover or land use. This is in addition to the use of other available maps as an input for some of the models developed within the MEDALUS II project. The possibility of using satellite data for vegetation and lithological mapping and land cover classification has Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
278
Mediterranean Desertification
been explored by many researchers (Norwine and Greegor 1983; Townshend et al. 1987). The opportunity is potentially extremely useful for the synoptic coverage of the remotely sensed data. The multitemporal radiometric response of the vegetation communities improves the detection of seasonal changes and aids in mapping vegetation cover. Monitoring vegetation cover is one of the important applications of remotely sensed data. The selection of the spectral and spatial resolution of such data is mainly dependent on the scale of the study. Low-resolution data were found to be adequate for global monitoring whereas high-resolution data such as Landsat Thematic Mapper (TM) data were found to be useful for regional and local mapping, forest inventory and monitoring of diverse agriculture and vegetation types, especially in semi-arid regions such as the Mediterranean Basin (Hill and Sturm 1991; Olsson 1995). Therefore, Landsat Thematic Mapper data were selected for the present study. In the case of the study area of the Guadalent´ın Basin, the available vegetation cover maps were made in 1985 (MAPA 1985) and updated in 1995 (ICONA 1995) whereas the geological map was surveyed in the 1970s (IGME 1974a). The ICONA map split the vegetation cover into general classes according to major differences in plant species and densities, such as pine forests and matorral (natural vegetation matorral ), whereas the detailed vegetation cover and lithology maps derived from remote sensing data provide differentiation between the natural vegetation and cultivated areas in addition to the sparse vegetation types and pine forests. The lithology map provides composition variation in the study basin independently of the geological structure. The updating technique is based on the use of multitemporal Landsat-5 Thematic Mapper (TM) imagery acquired in the spring and autumn of 1993.
2
STUDY AREA
The Guadalent´ın Basin is located in south-east Spain, shared between Murcia and Almeria provinces (UTM coordinates: Xmin 565000, Xmax 645010, Ymin 4152970 and Ymax 4213000). The climate conditions are semi-arid with hot, dry summers and mild winters. The average rainfall is about 300 mm with a high inter-annual and spatial variability, while the average temperature ranges between 18 and 30 ◦ C. There are commonly prolonged dry periods followed by heavy rainfall, particularly in autumn (L´opez-Berm´udez et al. 1996). Geologically, the Guadalent´ın Basin is located in the eastern sector of the Cordilleras Beticas, which are in the northern segment of the European Alpine Belt (IGME 1974a). The basin is divided into two structural domains, the external and internal zones, differentiated by their paleogeographic and tectonic–metamorphic evolution. The area shows a wide range of lithological exposures (gypsum, marl, clayey conglomerates, limestones and dolomites, greywackes, red slates, quartzite and phyllite) belonging to the internal, intermediate and external zones. The conglomerates are highly variable and their constituents reflect the full range of betic rock formations. The Guadalent´ın Basin also shows a high variability of vegetation types ranging from the natural vegetation to cultivated lands and forests. The natural vegetation includes semi-natural shrubs such as Rosmarinus officinalis and Juniperus oxycedrus, Anthyllis cytisoides and Stipa tenacissima (ICONA 1995). Pinus species are frequently found within the studied basin. A major part of the Guadalent´ın Basin shows low natural vegetation cover, short-lived green shrubs and grasses that disappear in the dry season. Cultivation dominates the plains covering the clayey conglomerate, for irrigated and dry crops. The cultivated areas also include widespread almond and citrus groves. 2.1
Definition of the Dominant Vegetation and Lithological Classes
Classes for the vegetation cover and lithology mapping were selected according to their sustainability in terms of the desertification process and the dominance of each class in the basin. The classes (Table 20.1) included seasonal grasses (mainly bare soil with seasonal grasses and dispersed bushes), dispersed matorral, rosemary (Rosmarinus officinalis), Stipa, irrigated areas, dry cultivated areas, almond plantations and pine forest. The seasonal grasses, dispersed matorral, Rosmarinus and Stipa classes belong to the “natural vegetation cover” that could be differentiated either by the vegetation
Lithology and Vegetation Cover Mapping
279
Table 20.1 Percentage area covered by vegetation types in the Guadalent´ın Basin derived from the classification of the multitemporal TM images (spring and autumn 1993)
Class Seasonal grasses Dispersed matorral Stipa Rosmarinus officinalis Dry cereals Irrigated areas Dense pine forests Almond groves
Area (%) 12 30 14 14 9 3 11 7
Table 20.2 Percentage area covered by lithological classes in the Guadalent´ın Basin derived from the classification of the TM images acquired on 14 September 1993. Values in parentheses indicate areas covered by crops and pine forest and assigned to each class according to ground truth checking
Class
Area (%)
Gypsum Marl Clayey conglomerate (Quaternary) Carbonate Phyllite Greywacke
5 5 37 (3) 26 (7) 11 16 (1)
type as in the case of Rosmarinus and Stipa, or by the density of the natural vegetation as in the case of dispersed matorral and seasonal grasses. The selection of the lithological classes (Table 20.2) was based on similar criteria to those of the vegetation cover mapping. The lithologies were selected to involve the composition properties, vegetation cover, consolidation and relief (Younis and Meli´a 1992). The selected lithology classes show different resistance to the desertification process. The selected classes included gypsum (gypsum and marly gypsum), marl (gypsiferous marls and marls), carbonate (limestones, calcarenite and dolomite), clayey conglomerate (variety of soil types, conglomerates and sands), phyllite and greywacke (greywacke and quartzite). In both classifications, the existing cartographic sources (geological: IGME 1974a,b, vegetation cover and the forest inventory maps: ICONA 1995) in addition to the aerial photographs of the area were used to define the representative training areas of each considered class. These areas were then checked using the aerial photographs (scaled 1:18 000) and a final check was carried out in the field.
3 REMOTE SENSING DATA AND IMAGE PROCESSING 3.1 Image Selection and Processing
In order to provide reliable vegetation and lithology maps by this classification method, multitemporal TM images were used. The TM imagery was associated with reasonable resolution of the TM images
280
Mediterranean Desertification
(30 m pixel size) suitable for the detection of lithology–composition and vegetation type–density variation in semi-arid areas. Differences in the nature and phenological evolution of the vegetation cover provide great advantages in mapping vegetation types and densities as they maximize the differences between classes in any one season. The dates of previous lithology maps were noted to establish when the vegetation cover was in the minimum phenological state. Images from spring and autumn are the best seasons for enhancing the differences between vegetation classes and finally improving the spectral separation in both vegetation and lithological classifications. Two Landsat-5 TM images were selected belonging to spring and autumn (7 April and 14 September 1993). The analysed TM images covered a major part (about 90%) of the Guadalent´ın Basin, defined by a window of 60 × 128 km2 centred on the north of Lorca city (scene 199/34). First, the September 1993 TM image was geometrically corrected using ground control points and the cubic convolution method. The ground control points (about 150 ground control points and topographic maps of scale 1:50 000 and 1:25 000) were selected to be homogeneously distributed over the whole image to ensure a reliable geometric correction of the whole image. The correction was then checked through the selection of another set of points, not used in the corrections, and the UTM coordinates obtained from both were then compared. Both the geometric correction error and the check comparison results showed less than one pixel error which ensures a very good geometric correction of the image. Once the September 1993 image was corrected, the April 1993 image was also corrected using image-to-image correction and the cubic convolution method. The correction was carried out by the selection of more than 200 points distributed over the whole of the image and the resultant error in the geometric correction was lower than one pixel in all cases. The real problem in deriving accurate cover mapping information from multitemporal satellite imagery, however, is the processing of such data prior to extracting features of change. Analysis of the direct linkage between data and the calibration of digital image data to reflectance units is absolutely necessary prior to the use of the multitemporal or multi-image sets (Duggin and Robinove 1990; Coppin and Bauer 1992). Hence, both Landsat-5 TM image sets (spring and autumn) were atmospherically normalized using a method based on a simple atmospheric radiative transfer model (Gilabert et al. 1994) that allowed quantitative comparisons of spectral target signatures. This method requires only inputs that are commonly available (date, latitude, height, aerosol type) and the presence of some dark surfaces in the TM1 band (blue) and TM3 band (red). Normalized Difference Vegetation Index (NDVI) images were then constructed ((TM4 − TM3)/(TM4 + TM3)) for both spring and autumn. The full scheme of the digital image processing is shown in Figure 20.1. 3.2
Band Selection for Classification Purposes
The problem of band selection for the classification process is an important factor for the accuracy of the classification result. The objective of the input band selection is to avoid the use of the spectral bands where the individual classes might show similar spectral signature and accordingly produce inaccuracies in the resulting output map. The input bands were selected after a keen examination of the spectral signature of the selected classes and with the aid of interband statistical information. In vegetation cover classification, the visible TM bands (TM1, TM2 and TM3) and the middle infrared bands (TM5 and TM7) are highly correlated, which indicates a similar information content between them (R 2 = 0.91–0.93). TM2 and TM5 bands show lower correlation coefficient values with TM4 and NDVI (R 2 = 0.62–0.67). The spectral signature of the selected classes showed considerable spectral confusion in TM1, TM3 and TM7 bands. These results are consistent with those obtained by Benson and De Gloria (1985), Horler and Ahern (1986) and Leprieur et al. (1988). Accordingly, TM2, TM5 and the NDVI of both spring and autumn were considered as an input for the process as shown in Figure 20.1. The same method was applied for the band selection in the lithological classification. The interband correlation coefficient of the September image and the examination of the spectral signature of the chosen classes revealed less confusion compared with the case of the vegetation cover classification. Hence the six optical bands of autumn image were used and the band with possible spectral confusion
Lithology and Vegetation Cover Mapping
281
Acquisition of Landsat-5 TM images in spring and autumn 1993. Floating window 199/34
Image corrections: ∗ Geometric coding ∗ Conversion to reflectance values
Lithology mapping
Vegetation mapping
Band selection
Band selection
TM2, TM5 and NDVI for spring and autumn
The six optical bands of autumn images (14-9-1993)
Classification Maximum likelihood method and 95% confidence limit
Checking results ∗ Comparison with the published data ∗ Comparison with aerial photographs ∗ Ground truth
Lithology and vegetation cover maps
Figure 20.1
Scheme of imagery processing
between some of the classes (e.g. TM1 band in the case of gypsum, marl and phyllite) was weighted less than other bands. 3.3 NDVI and Spectral Variation of the Selected Vegetation Classes
Once the dominant vegetation cover types and the input bands were well defined, the correspondent training areas were selected by aid of ground truth information, the available vegetation cover map (MAPA 1985; ICONA 1995) and photo-interpretation of the aerial photographs (for the Velez Rubio and Velez Blanco sheets, acquired in 1985 at a scale of 1:18 000). The selected training areas were chosen to fit certain conditions: large area (more that 200 pixels), homogeneous vegetation cover and low relief.
282
Mediterranean Desertification
The spectral signatures (TM2, TM5 and NDVI) of the selected vegetation cover classes in both seasons showed higher dispersion in the seasonal grasses and dispersed matorral classes. The high dispersion is mainly attributed to the nature of the classes as they consist of different dispersed vegetation types (see Figures 20.2 and 20.3). The spectral signatures of the other classes show good separation between both seasons. Reflectance in TM2 and TM5 for the spring and autumn seasons showed high spectral discrimination between all the vegetation classes except for the case of the seasonal grasses where the dispersion of both seasons was higher (Figure 20.2). On the other hand, the majority of the vegetation classes (except seasonal grasses and pine forest) showed considerable spectral separation between both spring and autumn images. Differences between NDVI in spring and autumn were higher in the case of the crop classes (irrigated and dry cereals) than the natural and semi-natural vegetation classes. Rosmarinus showed higher NDVI values in spring than Stipa, while both showed less significant differences in autumn (Figure 20.3). Seasonal grasses and dispersed matorral classes showed a higher standard deviation in TM2 and TM5 bands and for both seasons’ images, which is explained by the nature of these classes to include
70
TM2 TM5
Spring TM image
Reflectance (%)
60 50 40 30 20 10 0 SG
DM
DC IA Ros St Vegetation classes
Al
Pin
70
TM2 TM5
Autumn TM image
Reflectance (%)
60 50 40 30 20 10 0 SG
DM
DC IA Ros St Vegetation classes
Al
Pin
Figure 20.2 Spectral signatures of the selected vegetation cover classes as extracted from spring and autumn TM images in the Guadalent´ın Basin. SG, seasonal grasses; DM, dispersed matorral; IA, irrigated areas; Ros, Anthyllis cytisoides ‘‘Albaidal’’ and Rosmarinus officinalis ‘‘Romero’’; St, Stipa tenacissima ‘‘Espartal’’; Al, almond groves; Pin, pine
Lithology and Vegetation Cover Mapping
283
0.7 NDVI (spring) NDVI (autumn)
0.6 0.5
NDVI
0.4 0.3 0.2 0.1 0 −0.1 SG
DM
DC
IA
Ros
St
Al
Pin
Vegetation classes
Figure 20.3 Seasonal variation of the NDVI of the selected vegetation cover classes in the Guadalent´ın Basin. SG, seasonal grasses, DM, dispersed matorral; IA, irrigated areas; Ros, Anthyllis cytisoides ‘‘Albaidal’’ and Rosmarinus officinalis ‘‘Romero’’; St, Stipa tenacissima ‘‘Espartal’’; Al, almond groves; Pin, pine
different types of very dispersed bushes and annual grasses. Finally, the use of the multitemporal images for the vegetation cover classification showed clear, efficient spectral discrimination between the vegetation classes as they showed different seasonal variation in the TM2, TM5 and NDVI values. The NDVI values of the seasonal grasses, dispersed matorral and pine forests areas showed a high similarity in the autumn NDVI values. Nevertheless, these classes showed lower standard deviation as the spectral discrimination between them is clearer. Irrigated areas, dry cereals, Rosmarinus–Anthyllis areas and Stipa areas are clearly discriminated by their NDVI values according to their nature and phenology. 3.4 Classification of Lithological Units
Landsat-5 TM images from September 1993 were used for lithology classification of the Guadalent´ın Basin. The six optical bands of the selected image were used in the classification procedure. September was selected to correspond with the lowest phenological activity of the vegetation cover over these lithologies (Younis 1993). In addition to the selected lithology classes, two vegetation cover classes (pine forests and cultivated areas) were included in the classification due to the dense cover of the underlying lithologies. These classes were assigned to the underlying lithological classes according to their spatial distribution and field truth determinations. Cultivated areas (dry cereals and irrigated areas) are associated with the clayey conglomerate (Quaternary sediments) and the pine forests are associated with the carbonate, marl and greywacke classes. The spectral signature derived from the image showed good separation between the selected classes and low standard deviation values, as shown in Figure 20.4.
4 CLASSIFICATION METHOD AND RESULTS The classification was carried out using the I2 S (International Imaging System “Vista”) with “SOLARES” mounted on a Spark 10 Workstation. The supervised maximum likelihood method was used for the lithology and vegetation classification of the Guadalent´ın Basin. The supervised method was selected in order to have good control of the spectral and ground properties of the selected classes and due to the intensive and widely available field information that this method requires.
284
Mediterranean Desertification 50
50
Marl
40
Reflectance (%)
Reflectance (%)
Gypsum
30 20
40 30 20 10
10
0
0 TM1 TM2 TM3 TM4 TM5 TM7
TM1 TM2 TM3 TM4 TM5 TM7 TM bands
TM bands 50
Clayey conglomerate
40
Reflectance (%)
Reflectance (%)
50
30 20 10
40 30 20 10
0
0 TM1 TM2 TM3 TM4 TM5 TM7 TM bands
TM1 TM2 TM3 TM4 TM5 TM7 TM bands 50
50 Graywacke
Phyllite Reflectance (%)
40 Reflectance (%)
Carbonate
30 20
40 30 20 10
10
0
0 TM1 TM2 TM3 TM4 TM5 TM7 TM bands
TM1 TM2 TM3 TM4 TM5 TM7 TM bands
Figure 20.4 Spectral signatures in the TM bands of the selected classes for the lithological classification using the autumn TM image
The classification was based on a maximum likelihood approach which provides best estimates of the probabilities of each selected class and is predicted using training sites whose spectral and ground characteristics are known (Corves and Place 1994). The final classification images of vegetation and lithology mapping are shown in Plate 3 in the colour plate section. The vegetation cover classification shows the dominance of the dispersed matorral (∼30%; areas covered by dispersed matorral included Stipa tenacissima “Espartal”, Anthyllis cytisoides “Albaidal” and Rosmarinus officinalis “Romero”) and the seasonal grasses areas (∼12%) including exposures belonging to the different lithologies and soil cover, as shown in Table 20.1. The same result also shows the dominance of the dry cereals areas (∼9%) within the cultivated areas in comparison with the dry cereals areas of the irrigated areas (∼3%).
285
Lithology and Vegetation Cover Mapping
5 ASSESSING ACCURACY AND RELIABILITY OF THE LITHOLOGY AND VEGETATION MAPS In order to ensure the precision of the spatial distribution of the vegetation types classes, various accuracy tests were carried out. Errors are mainly of two types. The first type is due to the spectral overlap of two or more user-defined land covers. This error can be quantified by the classical sample-based methods of accuracy: the error matrix and confidence estimates. The second type is the mislabelling that occurs in situations where there is incomplete knowledge about the land cover types. Therefore, this may result in missing a number of land cover types when classes are defined (Corves and Place 1994). The first type involves the accuracy test of the classification results (error matrix) while the second type involves the comparison of the classification result with the ground truth and the available vegetation cover maps and information at different scales. 5.1 Accuracy Assessment (Error Matrix)
The overall classification accuracy (OCA) is around 89% for the vegetation cover and 91% for the lithological classification, which is high enough to indicate a reliable classification and, consequently, acceptable mapping (Table 20.3). Table 20.3 Error matrix of the training sites used in the (A) vegetation cover and (B) lithology classification of the Guadalent´ın Basin (A) Vegetation cover a
Class
SG
DM
DC
SG DM DC IA Ros St Al Pin
144 17 12
18 176 15
8 31 272 12
3 2 5
(B) Lithology
IA
Ros 3
7 440
21 9 4
St
Al
Pin
TOT
Ni
ni
nj
4 3 5 525
170 228 309 469 447 379 207 547
84.7 77.2 88.0 95.3 88.6 87.5 93.3 96.0 88.8
26 52 37 29 51 47 14 22
39 67 51 7 50 27 23 12
1 3 16
1 396 31
23 332
15
3
193 4
b
Class
Gyp
Marl
CC
Carb
Phy
Grey
TOT
Ni
ni
nj
Gyp Marl CC Carb Phy Grey
105 13 1 3 3 2
4 267 3 4 1 3
1 3 355 5 3 2
8 15 338 2 11
1 3 2 3 200 2
3 7 12 19 185
110 297 380 365 228 215
95.4 89.9 93.4 92.6 87.7 86.0 90.8
5 30 25 27 28 20
22 15 14 36 11 41
TOT, total pixels; Ni, accuracy of the class; ni, number of pixels classified in another classes; nj, number of pixels of other classes classified in the class. ∗ Confidence limit 95%. a SG, seasonal grasses; DM, dispersed matorral; IA, irrigated areas; Ros, Anthyllis cytisoides “Albaidal” and Rosmarinus officinalis “Romero”; St, Stipa tenacissima “Espartal”; Al, almond groves; Pin, pine forests. b Gyp, Gypsum; CC, clayey conglomerate (Quaternary deposits); Carb, carbonate; Phy, phyllite and Grey, Greywacke.
286
Mediterranean Desertification
The confusion matrix of the selected training sites (chosen classes) indicates that there are two types of confusion: the first is among the seasonal grasses, dispersed matorral and dry cereals. This type of confusion is attributed to the nature of these three classes which are characterized by bare soil and very dispersed vegetation (natural and/or cultivated). The error is around 20% (see Table 20.3), but such a high degree of error is expected among these classes as they present high dispersion in the spring season NDVI data due to their photosynthetic activity. The second type of confusion (error) is found among the natural vegetation cover classes (Anthyllis cytisoides “Albaidal” and Rosmarinus officinalis “Romero”, Stipa tenacissima “Espartal” and the disperse matorral classes) as shown in Table 20.3. The spectral confusion among these classes was expected as they represent similar plant communities, with different densities and mixtures of the three dominant types in some areas. The rest of the spectral confusion was found to present an error of less than 5%. This is well observed between pine forest and the other natural vegetation classes, irrigated areas and dry cereals classes, almond plantations and the pine forest classes. The error matrix of the lithological classification reveals confusion between gypsum and marl, greywacke and carbonate, and phyllite and greywacke classes. The spectral confusion is related to the field and composition similarities between the class pairs. In all cases spectral confusion is lower than 5% of the total examined pixels. 5.2
Ground Truth and Visual Interpretation Checking
The ground truth checking of the spatial distribution of the resultant classification was carried out by two different methods. The first method involved checking with the available geological and vegetation cover maps (IGME 1974a,b; MAPA 1985; ICONA 1995) and the second method compared aerial photographs of part of the Guadalent´ın Basin (Velez Rubio and Velez Blanco sheets). The ground truth checking was also carried out by comparison of the classification result with the field measurements, to ensure the correspondence of the lithological units in the spatial distributions and the contacts with the adjacent units. All the applied methodology shows good correspondence between the classification results and those of the published maps and the field data. The obtained vegetation map was compared with the existing ICONA (1995) and MAPA inventory maps of 1:25 000 and 1:50 000 scales respectively. The accuracy of the obtained vegetation map was evaluated by visual interpretation in addition to the selection of representative pilot areas in both sources. The comparison showed very high coincidence in the pine forest class, matorral classes and the cultivated lands. The ground truth checking involved evaluation of the classification results through the comparison with ground truth to examine the reliability of the derived thematic maps. The nature of the studied area and the experience acquired through the fieldwork provided a wide knowledge that facilitated the comparison of both sources of information (available maps and the obtained thematic map). The comparison revealed that there is a high degree of correspondence for almond groves and pine forests. This result is in harmony with that derived from the error matrix in which a low error was found for both classes. The dry cereals and the irrigated classes showed good correspondence with the cultivated areas mentioned in MAPA (1985) as cultivated areas. The natural vegetation showed a mixture of different plant communities but the comparison showed good correspondence for the Rosmarinus–Anthyllis areas and Stipa areas. In the majority of the examined cases, the classified areas showed the dominance of the class vegetation type. Lower correspondence was found in the case of the dispersed matorral and the seasonal grasses classes due to the similarities in the density and nature of the vegetation cover of both classes. However, the error of pixel assignation between both classes and the rest of the vegetation cover types was found to be very low in the majority of the examined cases.
6
CONCLUSIONS
The classification of the vegetation types in the Guadalent´ın Basin has resulted in a reliable and improved map of the vegetation cover in the area, taking into consideration the most important
Lithology and Vegetation Cover Mapping
287
natural vegetation and cultivated area types. The improvement of the available vegetation cover maps included the consideration of new classes that are not mentioned in these earlier sources of information, and the separation of the matorral into various classes. These classes (Rosmarinus, Stipa and dispersed matorral ) are of great importance to desertification process research and the determination of the age of abandoned lands. The present vegetation cover classification was checked by comparison with the ICONA inventory map and there was good correspondence between the information sources. Moreover, the present vegetation cover classification also differentiates well between three classes within the matorral class – Rosmarinus–Anthyllis areas, Stipa areas and dispersed matorral – as well as the seasonal grass areas which are dominated by very poor seasonal vegetation cover. In addition, the vegetation cover classification discriminates between the irrigated and dry cultivated cereals within the cultivated areas, which are not considered on the ICONA map. The lithological classification provides a composition map independent of the geological formation (age, structure and paleo-environmental conditions). These units are grouped in a relative manner of susceptibility to the erosion and degradation process according to the consolidation of the lithological exposures, nature of the vegetation cover and relief, and have provided five major LDRUS units (lithological desertification units). On the other hand, the combination of composition (mineralogy and consolidation) and field properties (topography and density of vegetation cover) of the lithology classes show different resistance to the erosion process, and lead to the final mapping of different desertification units (LDRUs). Friable gypsum and marl exposures are characterized by sparse vegetation cover, due to the abundance of salts, and moderate relief, and consequently could mark the class as having the greatest sensitivity to erosion or desertification processes. The Quaternary deposit class shows a higher density of vegetation cover which increases the resistance to erosion with respect to the first unit. Similarly, volcanic and metamorphic classes (greywacke and small basalt outcrops) also show high consolidation and relief but with sparse matorral, and higher resistance to desertification. Carbonate and sandstone lithologies, which show very high consolidation, relief and natural vegetation density (mainly covered by dense matorral and pine forest), accordingly show very high resistance to desertification processes in the Guadalent´ın Basin.
REFERENCES Baret F, Jacquemoud S and Hanocq FJ (1993) The soil line concept in remote sensing. Remote Sensing Reviews 7, 65–82. Benson AS and De Gloria S (1985) Interpretation of Landsat-4 Thematic Mapper and Multispectral Scanner data for forest survey. Photogrammetric Engineering and Remote Sensing 51, 1281–1289. Coppin PR and Bauer ME (1992) Processing multitemporal Landsat TM imagery to optimize extraction of forest cover change features. IEEE Transaction on Geoscience and Remote Sensing 32, 918–927. Corves C and Place CJ (1994) Mapping the reliability of Stellite-derived landcover maps – an example from Central Brazilian Amazon Basin. International Journal of Remote Sensing 15, 1283–1294. Duggin MJ and Robinove CJ (1990) Assumptions implicit in remote sensing data acquisition and analysis. International Journal of Remote Sensing 11, 1669–1694. Gilabert MA, Conese C and Maselli F (1994) An atmospheric correction method for the automatic retrieval of surface reflectances from TM images. International Journal of Remote Sensing 15, 2065–2086. Hill J and Sturm B (1991) Radiometric correction of multitemporal Thematic Mapper data for use in agriculture land-cover classification and vegetation monitoring. International Journal of Remote Sensing 12, 1471–1491. Horler DNH and Ahern FJ (1986) Forestry information content of Thematic Mapper data. International Journal of Remote Sensing 7, 405–428. ICONA (1995) El segundo inventario forestal nacional, Region de Murcia. Ministerio de Pesca y Alimentacion, Madrid. IGME (1974a) Mapa geol´ogico de Espa˜na E: 1/50 000, Hoja No 953 (Lorca). Servicio de Publicaciones del Ministerio de Industria y Energ´ıa, Madrid. IGME (1974b) Mapa geol´ogico de Espa˜na E: 1/50 000, Hoja No 952 (Vel´ez Balnco). Servicio de Publicaciones del Ministerio de Industria y Energ´ıa, Madrid. Leprieur CE, Durand JM and Peyron JL (1988) Influence of topography on forest reflectance using Landsat Thematic mapper and digital Terrain data. Photogrammetric Engineering and Remote Sensing 54, 491–496.
288
Mediterranean Desertification
L´opez-Berm´udez FL, Romero D´ıaz A, Mart´ınez-Fern´andez J and Mart´ınez-Fern´andez J (1996) The El Ardal field site: soil and vegetation cover. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 112–134. Malingreau JP (1986) Global vegetation dynamics: satellite observations over Asia. International Journal of Remote Sensing 7, 1121–1146. MAPA (Ministerio de Agricultura, Pesca y Alimentacion) (1985) Mapa de Cultivos y aprovechamientos (1:50 000), hoja No 953 (Lorca). Meli´a J, Gilabert MA and Younis MT (1993) Remote sensing contribution to the study of Alcantarilla (MEDALUS area). Proceedings of the 3rd International Conference in Geomorphology, Hamilton, 23–29 August 1993. Norwine J and Greegor DH (1983) Vegetation classification based on Advanced Very High Resolution Radiometer (AVHRR) satellite imagery. Remote Sensing of Environment 13, 69–87. Olsson H (1995) Reflectance calibration of Thematic Mapper data for forest change detection. International Journal of Remote Sensing 16, 81–96. Townshend JRG, Justice CO and Kalb V (1987) Characterisation and classification of South American land cover types using satellite data. International Journal of Remote Sensing 8, 1189–1207. Trimble SW (1990) Geomorphic effects of vegetation cover and management: some time and space considerations in prediction of erosion and sediment yield. In JB Thornes (ed.), Vegetation and Erosion. John Wiley, Chichester, pp. 87–112. Tucker CJ, Townshend JRC and Goff TE (1985) African land cover classification using satellite data. Science 227, 369–375. Younis MT (1993) Caracterizaci´on y discriminaci´on espectral, mediante radiometr´ıa de campo e im´agenes Landsat-5 TM, de unidades litol´ogicas en el a´ rea de Alcantarilla, Murcia. Thesis Doctoral, Universidad de Zaragoza, December 1993. Younis MT and Meli´a J (1992) Interpretation of Landsat TM images for mapping erosion susceptibility in the area of Murcia, SE Spain. G´eoobserver, special issue for Marisy 92, 353–356.
21
Changing Social and Economic Conditions in a Region Undergoing Desertification in the Guadalent´ın
1 ´ ROMERO DIAZ, ´ ´ ASUNCION PEDRO TOBARRA OCHOA,2 FRANCISCO 1 3 ´ ´ ´ ´ LOPEZ-BERMUDEZ AND GONZALO GONZALEZ-BARBERA
1
Department of Physical Geography, University of Murcia, Spain Department of Fundamentals of Economical Analysis, University of Murcia, Spain 3 CEBAS-CSIC, Universitario de Espinardo, Murcia, Spain 2
1 INTRODUCTION The Guadalent´ın Basin, located in the south-east of the Iberian Peninsula, covers an area of 3300 km2 . The climate is predominantly semi-arid, and the basin is affected by desertification processes (Lo´ pez Berm´udez et al. 1998). Some of the desertification processes observed in the Guadalent´ın are produced by natural causes, especially climatic change, but the effects of human activity also have far-reaching effects. Deforestation, ploughing of agricultural land, overgrazing, land abandonment, irrigation with saline waters, and intensive cultivation are examples of activities that may promote desertification. The Guadalent´ın Basin has sustained human pressure and inherent degradation for thousands of years. However, in recent history, changes in socio-economic aspects have occurred and these have profoundly affected the situation. Within this socio-economic framework there are two fundamental aspects. First, there are changes in the dynamics of the population, the occupational activity and the level of income; and second there are changes in the use of the soil as a consequence of economic policies, changes in availability of water (Tobarra Ochoa 1995), degradation of soil, land abandonment, and other physical factors.
2 CHANGES IN THE POPULATION Research has been undertaken into the evolution of the demography of the Guadalent´ın Basin. This has involved explanation of the historic evolution of the population, and also development of a methodology for demographic projection with the aim of forecasting future population composition. 2.1 Evolution of the Population Change in the population of the Guadalent´ın Basin has been less dynamic than in the region of Murcia as a whole (Figure 21.1 and Table 21.1). For the period from 1900 to 1998, the population of the basin grew in total by approximately 27%, while that of Murcia increased by 92%, and for the whole of Spain the figure is 106.4% (S´anchez and P´erez 1989; S´anchez and Ort´ın 1993). The pattern of annual average growth rate is similar. For the period 1900–1998, the population of the Guadalent´ın Basin grew at a rate of approximately 0.08% per year, for the region of Murcia the rate was 0.88% annually, while for Spain as a whole the figure was 1.17%. This smaller population growth of the Guadalent´ın Basin is explained by the slower internal dynamic, and also migratory Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
290
Mediterranean Desertification 250 Population
200 150
Guadalentín Murcia Spain
100 50 1998
1991
1981
1970
1960
1950
1940
1930
1920
1910
1900
0
Year
Figure 21.1 Population growth Table 21.1
Year
Population growth (1900 = 100)
Population Guadalent´ın Basin
1877 1887 1897 1900 1910 1920 1930 1940 1950 1960 1970 1981 1991 1998
73 190 80 863 83 792 96 083 100 693 104 275 89 490 100 399 101 747 97 047 100 493 106 705 115 270 122 480
Index
Region of Murcia 452 082 489 770 519 149 581 455 620 926 654 436 651 979 731 221 755 850 803 083 832 047 955 487 1 045 601 1 115 068
Guadalent´ın Basin
Region of Murcia
76 84 87 100 105 109 93 104 106 101 105 111 120 127
78 84 89 100 107 113 112 126 130 138 143 164 180 192
Source: Instituto Estad´ıstico de Andalucia (1982–1998); Estad´ıstica Regional de Murcia (1986–1991a,b, 1986–1998). processes which led to the loss of a large number of inhabitants in the recessionary years, during 1920–1930 and 1950–1960. The internal dynamics of the demographic increase are determined by the birth rate and the death rate. With respect to the birth rate, the Guadalent´ın Basin has shown some decreasing values, and always values below those corresponding to the region of Murcia and the whole of Spain. Specifically, and using Natural Population Movement data, the birth rate fell from 23.7 per thousand in the decade of the 1950s, to 8.9 per thousand in 1997. For the region of Murcia, the rate fell from 23.1 to 11.8 per thousand in the same period of time. In Spain, the rate fell from 20.5 to 9.5 per thousand. The death rate in the Guadalent´ın Basin has remained practically constant from the 1950s (8.7 per thousand) until 1997 (8.0 per thousand), with some oscillations in the intermediate years. For the region of Murcia this rate has been similar to that for the Guadalent´ın Basin, decreasing from 8.7 per thousand in the 1950s to 8.04 per thousand in 1997. For Spain as a whole, for the same period, the death rate fell from 9.4 to 8.2 per thousand. Considering the average death rate from 1950 to 1997, the Guadalent´ın Basin shows the highest rate: 8.7 per thousand. For the region of Murcia, the average rate was 7.9 per thousand, and for Spain it was 8.2 per thousand.
Changing Social and Economic Conditions
291
Table 21.2 Population in three age groups expressed as a percentage of the total population
Age group <15
15–65
>65
Area
1981
1986
1991
1998
Guadalent´ın Basin Region of Murcia Spain
28.98 25.70
22.90 25.99 22.47
21.01 22.73 19.38
16.98 18.27 18.68
Guadalent´ın Basin Region of Murcia Spain
60.81 63.05
63.92 63.51 65.34
64.88 65.43 66.81
67.37 67.69 67.58
Guadalent´ın Basin Region of Murcia Spain
10.21 11.24
13.18 10.50 12.19
14.12 11.84 13.82
15.65 14.04 13.74
Source: Estad´ıstica Regional de Murcia (1986–1991a,b) The combination of low birth rates and high death rates in the Guadalent´ın Basin has resulted in a slower increase in population since 1877 than that seen for the region of Murcia. However, the increase in total population in the Guadalent´ın Basin is slightly greater than for the whole of Spain. Specifically, for 1998, in the Guadalent´ın Basin the increase was 2.1 per thousand while in the region of Murcia the corresponding figure was 4.6 per thousand. The average increase rates from 1950 to 1998 were 6.7 per thousand for the Guadalent´ın Basin, 8.8 per thousand for the region of Murcia, and 6.0 per thousand for the whole of Spain. Disaggregating the population by age provides a more detailed picture of the evolution of the total population (Table 21.2). Similar analysis to that above was performed for years 1986 and 1998, with the population being split into three age categories: <15, 15–65 and >65 years old. Note that the proportion of the population in the age group >65 is greater in the Guadalent´ın Basin compared with figures for the region of Murcia and the whole of Spain. Figures for the age group of 15 or less in 1986 and 1998 show a smaller percentage in the Guadalent´ın Basin than in Murcia, but the figures are higher than in Spain as a whole. Therefore, we can conclude that at all three geographical scales there has been a general ageing of the population, and this process is more accentuated in the Guadalent´ın Basin than in the wider scale of the region of Murcia. It is interesting to note trends such as those in the region of Murcia, where the population increase showed peaks in 1979 and 1984 and a strong fall during 1986 to 1990. Then it recovered, only to fall again one year later (Bell Adell 1981; Colino 1993). 2.2 Demographic Projections Having examined some aspects of the historic evolution of the population of the Guadalent´ın Basin, the next step was to carry out demographic projections, with the aim of obtaining future population estimates disaggregated by age. As before, comparisons were made with the region of Murcia as well as Spain as a whole. The methodology for obtaining the projection is based on the survival of cohorts, with lagged (later) corrections. The basic variables which intervene in the projection are the fertility rate and the “step” rate (survival probability). Therefore, the key is to obtain the fertility and step rates, which in the case of the Guadalent´ın Basin have been calculated from the information corresponding to each municipality. Adjustment coefficients need to be developed to correct the estimates on the basis of historical population change. To create the adjustment coefficient, data for 1986 were used to project the population pyramid for 1991. Next, each age group of the projection was compared with the 1991 Population Census data (last census published). This adjustment coefficient was then applied to the future projections for 2001, 2006, 2011 and 2016. The results of the projections are summarized in Table 21.3. It is clear that the age group of 65 or older maintains a greater weight in the Guadalent´ın Basin than in the region of Murcia. In contrast, the age group of between 15 and 65 years (i.e. the group that includes the potentially active working population) exhibits a smaller weight in the Guadalent´ın Basin than in Murcia.
292
Mediterranean Desertification
Table 21.3 Projected population for three age groups (as a percentage of the total population)
Age group
Area
2001
2006
2011
2016
<15
Guadalent´ın Basin Region of Murcia Spain
17.76 18.82 15.36
17.94 18.89 15.08
19.38 18.91 15.16
19.07 18.24 14.65
15–65
Guadalent´ın Basin Region of Murcia Spain
65.27 66.71 67.51
65.49 66.34 67.43
64.41 66.34 66.81
65.08 67.07 66.53
>65
Guadalent´ın Basin Region of Murcia Spain
16.97 14.48 17.12
16.57 14.77 17.49
16.21 14.75 18.03
15.85 14.70 18.82
% of population
68 66 64 62 60 58 2016
2011
2006
2001
1996
1991
1986
1981
56
Year Guadalentín
Murcia
Spain
Figure 21.2 Comparison of the percentage of population of working age in the Guadalent´ın Basin, Murcia and Spain, including projections to 2016
It should be noted that figures for the projected working-aged population are lower in the Guadalent´ın Basin than in the region of Murcia or in Spain as a whole, in relative terms (Figure 21.2). It is also clear that the “ageing” rate is higher in the Guadalent´ın Basin than in Murcia, although smaller than for Spain as a whole. The “dependence” rate (young population plus old population divided by middle-aged population) will be higher in the Guadalent´ın Basin than in the region of Murcia or in the whole of Spain. 2.3
Working Population in the Different Economic Sectors
Information for each of the seven municipalities in Murcia has been obtained from the Regional Statistics Centre of Murcia. There are no data for the Guadalent´ın Basin in 1999. The purpose was to compare the dynamics of the working population in each of the economic sectors with respect to the Guadalent´ın Basin. The Guadalent´ın Basin also includes four municipalities belonging to the province of Almer´ıa in the region of Andalucia, making 11 municipalities in total.
Agriculture In the Guadalent´ın Basin there is a high proportion of the working population in the agricultural sector, although there has been a significant reduction since the early 1980s. In the 1980s the annual change was −21.6%, while nowadays it is −7.4%. This reduction is particularly dramatic in comparison to the national average (Figure 21.3).
Changing Social and Economic Conditions
293
Percentage of working population
Agriculture 50 40 30 20 10 0
1981
1986
1991
1999
Year
Percentage of working population
Industry and energy 25 20 15 10 5 0
1981
1986
1991
1999
1991
1999
Year
Percentage of working population
Building 14 12 10 8 6 4 2 0
1981
1986
Percentage of working population
Year Service industries 60 50 40 30 20 10 0
1981
1986
1991
1999
Year Guadalentín Basin
Region of Murcia
Figure 21.3 Percentage of total population working in agriculture, industry and energy, building and service industries
294
Mediterranean Desertification
Industry and Energy The working population in the industry and energy sector has shown an upward trend since the 1980s, very similar to the change in the regional mean. It is very likely that these increases result from young people moving away from the agricultural sector.
Building As a consequence of the economic crisis at the beginning of the 1980s, the working population in the building sector fell until 1985 with a trend similar to that of the regional and national mean. From 1985, a recovery similar to the rest of the region was initiated, although the rate of increase remained smaller in the Guadalent´ın Basin than in the region of Murcia.
Services An upward trend in the per cent working population in the services sector has been observed, with an annual growth rate of 11.85% in the Guadalent´ın Basin. The growth rate for the region was 6.1%, so the differential that existed in previous years has been reduced. As in the case of the industrial sector, this sector probably receives a flow of workers from the agricultural sector.
3
CHANGES IN THE LEVEL OF INCOME
3.1 Disposable Family Income Figure 21.4 compares the income per person in the agricultural sector for the Guadalent´ın Basin unfavourably against income per person for the region of Murcia for the period 1980 to 1994. Since 1981, the income per person in towns in the Guadalent´ın Basin has not been very different to the income in the region of Murcia. The most important component of this indicator comes from the “service sector” where there is growth, while the “agricultural income” component continues to decline in regional and national terms. The indicator also shows a reduction in the purchasing power of the inhabitants of the Guadalent´ın Basin. This may result in lower local and regional consumption levels. 3.2 Agricultural Income Generated by Agricultural Workers The trend in agricultural income is similar to the regional mean. This can mainly be attributed to farm subsidies, as farm subsidies have become a major component of agricultural income. If farm subsidies are deducted from the agricultural income, the income generated by agricultural Pts × 1000 3000 2500 2000 1500 1000 500 1994
1992
1990
1988
1986
1984
1982
1980
0
Year Guadalentín Basin
Region of Murcia
Figure 21.4 Comparison of income per person in the agricultural sector for the Guadalent´ın Basin and the whole region of Murcia
295
Changing Social and Economic Conditions Pts × 1000 (1982 = 100%)
1994
1992
1990
1988
1986
1984
1982
1980
4000 3500 3000 2500 2000 1500 1000 500 0
Year Income
Farm subsidies
Figure 21.5 Income and farm subsidies per person in the agricultural sector for the region of Murcia, 1980–1995
workers shows a downward trend (Figure 21.5). The effect of increasing farm subsidies since 1990 is very marked.
4 LAND-USE CHANGE Land-use change can be considered to be a socio-economic indicator of land degradation if the reasons for change (or not) are considered in a historical context. Often, in Mediterranean lands, the use and management of natural resources has not adapted to changing demands, resulting in significant disturbances in the ecological balance. Such disturbances cause problems such as increased erosion and loss of soil, degradation of vegetation and a reduction in available water resources (Barber´a et al. 1997; L´opez Berm´udez et al. 1995). 4.1 Historical Changes
In the Guadalent´ın Basin there have been significant land-use changes in the last 300 years (GilOlcina 1971; P´erez-Picazo 1980, 1990; ICONA 1990; Mart´ınez-Carri´on 1990). Perhaps the greatest changes took place around the end of the 18th century, when the laws of “confiscation” were enforced. This led to large-scale deforestation. The sale of “common lands” from 1859 onwards resulted in continuous felling of forest lands and the development of different land uses. In fact, the degradation of the forest had begun years before, under ancient ordinances whereby the farmers could cut pines and other trees for many purposes, without the need for authorization. In the ploughed lands, cereals were the main crops on better soils, but where there were steep slopes and rock outcrops, olive and almond trees were cultivated and Stipa tenacissima L. was the prevalent shrub vegetation. At the beginning of the 18th century the floodplain of the Guadalent´ın Basin was totally converted to cereal crops, because it was easy to plough here and irrigation was possible where necessary. Meanwhile, the sides of the valley and the higher land in the north remained practically without any cultivation. In these areas the possibility of irrigation was limited to small areas located near water sources or ramblas (ephemeral channels), where channelling systems known as boquera were used (Gil-Olcina 1971; G´omez Esp´ın 1989). Plant species associated with different types of land use may also be used as socio-economic indicators. For example, plants known as barrilleras (Salsola kali L. and Salsola longifolia L.), were widely cultivated in the second half of the 18th century. When the plants had been burned the ashes could be used in the manufacture of sodas and soaps. Later, the cultivation of these plants was no longer profitable because new chemical procedures were easier and cheaper. Another example is the esparto grass (Stipa tenacissima L.), once used for a textile fibre and harvested particularly
296
Mediterranean Desertification
from the upper and middle Guadalent´ın Basin until the 1960s. In the 19th century mulberries were traditionally cultivated in the huerta of Murcia. Then vines were introduced in response to the disease Phylloxera in the French vineyards. Later the Spanish vineyards were also decimated by Phylloxera. In historical times various natural disasters such as drought (records begin with 1811, 1817–1818, 1926–1928), flooding (e.g. 1921, 1923, 1939, 1942, 1943, 1949, 1959), and disease epidemics (cholera in 1834) contributed to changes in land use. Sometimes the wrong choice of crops has caused financial ruin and compelled the farmers to abandon their land, especially semi-arid land, and migrate towards areas where irrigation is possible. Irrigation together with new intensive and automated methods of horticulture and fruit-growing has provided a much greater economic potential. This trend continues today, using water brought in by the link between the River Segura and the River Tajo, and increasingly exploiting groundwater resources. 4.2
Recent Changes
Recent changes are considered to be those that have occurred in the latter half of the 20th century. Data for this period are fairly reliable.
General Distribution of Land Use
% land cover
In general, in the region of Murcia, as well as in the Guadalent´ın Basin, the biggest change has been the increase in the area of irrigated land. In the region of Murcia the biggest increase in irrigation took place in the 1950s, as a consequence of the regulation of the headwaters of the Segura River and by the initiation of groundwater extraction (Barber´a et al. 1997). In the Guadalent´ın Basin, the Puentes Dam has provided water for irrigation since the end of the 19th century (Bautista Mart´ın and Mu˜noz Bravo 1986). Between 1947 and 1972 there was an increase in the area of irrigated land (Figure 21.6), but it has been since 1973, when the prohibition on the opening of new wells was lifted (Tobarra Ochoa 1995), that the area of irrigated land has particularly increased. Around the end of the 1970s and during the 1980s, the arrival of supplementary water from the Tajo River (in central Spain), together with the huge increase in the extraction of groundwater, meant that by 1985 the area of cultivation under irrigation was 38% (compared to 10% in 1947). By 1995, 39.5% of the area of cultivation was being irrigated, but this is the limit because there is no further supply of surface water, and groundwater supplies are becoming exhausted. The extension of the irrigated area has occurred mainly on the dry coastal plains, with a negative variation rate for the period 1947–1998 of 69.4% and an annual mean variation of 1.4% in dryland (Table 21.4). In the period 1972–1985, the increase in the irrigated area included areas formerly covered by scrub.
70 60 50 40 30 20 10 0 1947
1972
Dryland
1985 Year Irrigated land
1995
1998
Other
Figure 21.6 Changes in land-use type in the Guadalent´ın Basin between 1947 and 1995
Table 21.4 Coefficient of variation (CV) for land use in the Guadalent´ın Basin
1947–1972 Mean Annual Variation (ha) Dryland Irrigated land Other
−47 941 6 091 42 207
CV (%)
CV (%)
−55.9 −2.2 125.1 5 134.9 5.4
1972–1985 Mean Annual Variation (ha)
CV (%)
CV (%)
18 575 130.6 10 17 455 154.4 11.9 −36 034 −77.9 −6
1985–1995 Mean Annual Variation (ha) −4 975 730 3 998
CV (%)
CV (%)
−93.7 −9.4 101.5 10.2 103.1 10.3
1995–1998 Mean Annual Variation (ha) 1 072 −180 7 457
CV (%)
CV (%)
101.4 33.8 −99.6 −33.2 106.1 35.4
1947–1998 Mean Annual Variation (ha)
CV (%)
CV (%)
−33 269 −69.4 −1.4 24 096 199.2 3.9 2 608 99.0 1.9
297
298
0
200
−50 −100
150
−150 100
−200
1995
1993
1991
1989
1987
1985
1983
1981
1979
−300 1977
−250
0 1975
50
Water table depth (m)
250
1973
No. of wells and groundwater pumping volume (Hm3 year−1)
Mediterranean Desertification
Year No. of wells
Groundwater pumping
Water table depth
Figure 21.7 Number of wells, groundwater pumping volume and water-table depth in the aquifer of the high Guadalent´ın
The huge increase in the area of land available for agriculture and horticulture through irrigation has allowed the income of some of the population to increase, but only at the expense of diminished water resources and increased risk of desertification. Near the coast, groundwater resources have been over-exploited to the extent that the extracted water is now saline. In the upper Guadalent´ın Basin the number of wells increased from 25 to 234 between 1973 and 1996. This has resulted in the water table being lowered by 227.5 m in 23 years (Figure 21.7). The volume of extraction was 24 hm3 year−1 in 1973, rising to 56 hm3 year−1 in 1990, but fell to 30 hm3 year−1 in 1996 because with the water table at a depth of 290 m, water extraction had become very expensive. This semi-arid environment receives 300 mm mean annual rainfall, and therefore irrigated lands depending on groundwater extraction are being abandoned, especially where water extracted from wells is becoming saline. The salinity and the effects of intensive cultivation have left the soils impoverished, and this is a first step towards desertification. As a consequence of the availability of new technology and the use of new machinery, there have been great changes in the landscape, especially in the foothills of the mountains. Unfortunately changes in practice have often been made without any consideration of soil properties, and this has undoubtedly resulted in increased soil erosion.
Distribution of Crops The distribution of land-use types has significantly changed since the 1950s in terms of the area of forest or the area under irrigation, and the types of crops grown have also changed. Currently the crops grown in the Guadalent´ın Basin, in order of importance, include woody crops (almond trees, vines and olive trees), herbaceous crops (wheat, barley and oats), citrus crops (orange, mandarin orange and lemon) and other fruit trees. It is interesting to see how the area devoted to cereals fell from 110 000 ha in the 1950s, to only 13 675 ha by 1998. Meanwhile the area covered by almond trees has doubled, and the area covered by citrus crops has increased from 900 ha in 1947 to more than 5600 ha in 1998 (Table 21.5). Barley is the most abundant cereal crop, and there have been big inter-annual variations in the area sown, as a result of droughts and bad harvests. Since 1984 the area sown has declined steadily. Woody crops, especially almond, were grown instead, but the area covered by this crop has also declined since 1989. On land that can be irrigated, the most notable change has been the extension of citrus cultivation. This is particularly so on the foothills of the mountains, located in the centre and south of the basin. The lemon has traditionally been the most abundant citrus species in the region of Murcia, but since 1989 lemons have declined while the area devoted to oranges has increased. In 1982 the area of
Changing Social and Economic Conditions
299
Table 21.5 Area covered by the main crops in the municipalities of Murcia (Guadalent´ın Basin)
Year
Cereal crops
Woody crops
Citrus crops
1947 1972 1975 1980 1985 1990 1995 1998
110 114 37 698 27 215 44 614 35 430 30 278 19 556 13 675
13 213 16 425 19 784 25 045 25 023 25 572 21 853 21 403
898 1663 2597 3971 5979 5804 5581 5601
Source: Ministerio de Agricultura, Ganader´ıa y Pesca (1982–1994); Estad´ıstica Regional de Murcia (1986–1998). fruit trees exceeded 2000 ha, but then there was severe drought and the area cultivated was reduced by half by 1984. The crops being grown are directly dependent on the availability of water. There have also been big changes on non-cultivated land, the forest and scrub areas. On the foothills, scrub was ploughed for cultivation under irrigation and also for afforestation. Some of the ploughed scrub has since been abandoned. Where the vegetation cover has been reduced, where unsuitable cultivation techniques have been used, and where terrace systems have fallen into disrepair, soil is very vulnerable to erosion. The area of forest in the Guadalent´ın Basin decreased from 116 500 ha in 1947 to 75 500 ha in 1989, but by 1998 it had increased slightly to 80 000 ha. This increase is not due to afforestation, but rather to regeneration of the natural forest. Where land has been abandoned, scrub has grown up. It is important to mention that different crops affect the physical and chemical characteristics of the soil in different ways. Investigations in the Rambla Salada, adjacent to the Guadalent´ın, have shown how land use affects soil organic matter content, structural stability, and the water retention capacity of different soils. Studies have demonstrated high structural stability under forest and scrub (3.4% and 2.1% respectively), compared to lower structural stability under cereals (0.7%) and almond trees (0.6%). The highest values for water retention capacity were found under forest and scrub, intermediate values under olive trees and very low values under cereals and almond trees. These values also apply to the Guadalent´ın Basin and, therefore, changes in cultivation practices are affecting the trend towards the degradation of soils here also.
5 CONCLUSIONS From our study of the evolution of the population, the level of income and changes in land use, it is possible to observe important recent changes in the Guadalent´ın Basin. Linkages between these factors and desertification processes remain unproven, but in our opinion the current degradation in the Guadalent´ın Basin is closely related to the socio-economic changes observed and the decline in the rural population. According to our projection there will be a reduction in the population of working age between now and 2016. A breakdown of the figures in the working sector suggests a shift from agriculture to the industrial and service sectors. The overall trend is a greater reduction in the working population in the Guadalent´ın Basin compared to in the region of Murcia as a whole. The level of income in the Guadalent´ın Basin is lower than the average income in the region of Murcia and in Spain as a whole. This denotes economic stagnation with respect to the region. It is necessary to mention the importance of subsidies conceded to agriculture from the European Union. These have greatly increased since 1991, and now form a large part of the agricultural income. With respect to land-use changes, the most important aspect is the increase in the area of irrigated land; in former forest, in dry lands and along the coast, wherever aquifers could be exploited.
300
Mediterranean Desertification
There is no doubt that this use of water resources will have adverse consequences in terms of land degradation, although irrigation has allowed much greater profitability to be had from the land. Initially it seemed economically profitable to improve cultivation by using irrigation, and the possibility of degradation of that land was not seriously considered. Now we see that the groundwater has been over-exploited, some wells have run dry and others only provide water of very poor quality. In these areas cultivation will have to be abandoned and the soil will be left in a poorer state than it was in before it was irrigated. There have been cases where the ploughing of land for irrigation or the planting of almond trees on former scrub land has definitely accelerated soil erosion processes. Modern machinery has allowed cultivation on steep or difficult soils where it was not possible before, but this has had detrimental effects on soil structure and other physical characteristics. The ecosystem of the Guadalent´ın Basin is fragile, and some degradation processes may be irreversible. Therefore it is critical to use policies that will conserve natural resources, whether they be water resources or soil. Afforestation carried out on former scrub areas has not always been successful. For example, in some places the scrub has been cut and ploughed out and pine seedlings planted, but few of the seedlings survived and the exposed soil was rapidly eroded. In contrast, the naturally growing scrub vegetation which is adapted to the prevailing climatic conditions covers the soil and protects it from erosion. Activities leading to land abandonment should be discouraged. It is clear that land abandonment has partly been caused by agricultural policies, especially those resulting in reduced biological productivity and where over-exploitation of water resources means that irrigation is no longer viable. Agriculture easily becomes uneconomic in these circumstances. Therefore it is essential that, in the light of the socio-economic changes we have highlighted, future planning and management of resources must act to restore conservation practices in the Guadalent´ın Basin.
REFERENCES Barber´a GG, L´opez Berm´udez F and Romero D´ıaz A (1997) Cambios de usos del suelo y desertificaci´on en el Mediterr´aneo: El caso del Sureste Ib´erico. In JM Garc´ıa Ruiz and P L´opez Garc´ıa (eds) Acci´on Humana y Desertificaci´on en Ambientes Mediterr´aneos. Instituto Pirenaico de Ecolog´ıa, Zaragoza, pp. 9–39. Bautista Mart´ın J and Mu˜noz Bravo J (1986) Las Presas del Estrecho de Puentes. Confederaci´on Hidrogr´afica del Segura, Murcia. Bell Adell MC (1981) Estructura y din´amica reciente de la poblaci´on murciana. Universidad de Murcia, Secretariado de publicaciones. Colino J (ed.) (1993) Estructura Econ´omica de la Regi´on de Murcia. Civitas, Madrid. Estad´ıstica Regional de Murcia (1986–1998). Anuarios Estad´ısticos de la Regi´on de Murcia. Consejer´ıa de Econom´ıa y Hacienda, Comunidad Aut´onoma de la Regi´on de Murcia. Estad´ıstica Regional de Murcia (1986–1991a) Movimiento Natural de la Poblaci´on de la Region of Murcia. Consejer´ıa de Econom´ıa y Hacienda, Comunidad Aut´onoma de la Regi´on de Murcia (annual report). Estad´ıstica Regional de Murcia (1986–1991b) Estad´ıstica Hist´orica de la Poblaci´on de la Regi´on de Murcia. Consejer´ıa de Econom´ıa y Hacienda, Comunidad Aut´onoma de la Regi´on de Murcia (annual report). Gil-Olcina A (1971) El campo de Lorca. Estudio de Geograf´ıa Agraria. Departamento de Geograf´ıa, Universidad de Valencia. G´omez Esp´ın JM (1989) Los Caminos del Agua. Caminos de la Regi´on de Murcia. Consejer´ıa de pol´ıtica Territorial y Obras P´ublicas, Murcia, pp. 527–556. ICONA (1990) Clasificaci´on General de los Montes P´ublicos. Ministerio de Agricultura Pesca y Alimentaci´on, Madrid (new edition). Instituto Estad´ıstico de Andaluc´ıa (1982–1998). Sistema de Informaci´on Municipal de Andaluc´ıa (SIMA). Junta de Andaluc´ıa, Sevilla. L´opez Berm´udez F, S´anchez Fuster MC and Romero D´ıaz A (1995) Incidencia de los modelos de gesti´on socioecon´omica (siglos XIX y XX) en la degradaci´on del suelo en el Campo de Lorca (Cuenca del Guadalent´ın, Murcia). Papeles de Geograf´ıa 22-II, 123–142. L´opez Berm´udez F, Romero D´ıaz A, Cabezas F, Rojo Serrano L, Mart´ınez Fern´andez J, Boer M and Del Barrio G (1998) The Guadalent´ın Basin, Murcia, Spain. In P Mairota, JB Thornes and N Geeson (eds) Atlas of Mediterranean Environments in Europe. John Wiley, Chichester, pp. 130–143.
Changing Social and Economic Conditions
301
Mart´ınez Carri´on JM (1990) Las transformaciones agrarias en Lorca durante el siglo XIX y comienzos del XX . Lorca Pasado y Presente, Lorca, pp. 129–148. Ministerio de Agricultura, Ganader´ıa y Pesca (1982–1994). Estad´ıstica Agr´aria de la Regi´on de Murcia. Comunidad Aut´onoma de la Regi´on de Murcia (annual report). P´erez Picazo MT (1980) Aspectos Socioecon´omicos del Siglo XIX en Lorca. Ciclo de Temas Lorquinos, Caja de Ahorros de Alicante y Murcia, pp. 157–176. P´erez Picazo MT (1990) Econom´ıa agraria y desarrollo industrial en Lorca, 1845–1936 . Lorca Pasado y Presente, Lorca, pp. 119–127. S´anchez P and Ort´ın J (1993) Estructura de la poblaci´on. In J Colino (ed.) Estructura Econ´omica de la Regi´on de Murcia. Civitas, Madrid, pp. 221–242. S´anchez P and P´erez C (1989) Demograf´ıa y Recursos Humanos. Papeles de Econom´ıa Espa˜nola 7, 115–134. Tobarra Ochoa P, (1995) Estudio del alto Guadalent´ın desde la perspectiva econ´omica de la gesti´on del agua subterr´anea. Caja de Ahorros del Mediterr´aneo, Murcia.
22
Management Plan to Combat Desertification in the Guadalent´ın River Basin
´ ROBREDO,2 J.A. MARTINEZ ´ L. ROJO SERRANO,1 F. GARCIA ARTERO3 AND ´ A. MARTINEZ RUIZ2 1
DGCONA, Ministerio de Medio Ambiente, Madrid, Spain ´ Universidad Empresa de Murcia, Escuela de Negocios de la Region ´ de Fundacion Murcia, Espinardo, Murcia, Spain 3 DGCONA, Ministerio de Medio Ambiente, Murcia, Spain 2
1 INTRODUCTION The Guadalent´ın Basin is an area characterized by the presence of all the symptoms associated with desertification, particularly because of its climatic and geological features. Also, over the centuries the effects of human intervention have been significant in triggering or accelerating the desertification process. When the balance between human activities, mainly concerned with agriculture, and the maintenance of natural processes was upset, the inhabitants started to feel the effects of the desertification phenomenon on their land. The decrease in agricultural productivity due to soil erosion, the damage caused by flooding, and, more recently, the problems of water scarcity, groundwater over-exploitation and soil salinization, have shown the need to decelerate the advance of the desertification processes and to reverse the current trends through the development and implementation of mitigation techniques. There has also been consolidation of a social awareness of the need for natural resource conservation. However, the call for the implementation of desertification control programmes is not a new concern in the Guadalent´ın Basin. At the end of the 19th century a large part of the population demanded reafforestation of large areas in the basin as the only way to control storm discharge in some ravines. They had been experiencing severe floods that led to loss of lives and possessions (Martinez-Artero 2001). There was clearly a need for a desertification control plan including technical mitigation actions, particularly regarding vegetation restoration and the provision of a proper policy framework.
2 OBJECTIVES The Management Plan to Combat Desertification in the Guadalent´ın River Basin has been designed to offer possible solutions to the problems of desertification. Finding such solutions was one of the objectives of the MEDALUS projects: “To build on the practical experiences of government and commercial agencies, through the interfacing of science and policy, in mitigating the effects of desertification in specific regions” (MEDALUS 1995). The accomplishment of this goal required the formulation of a number of particular objectives related to different stages of the work. These included the compilation of data on environmental Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
304
Mediterranean Desertification
factors such as climate, soil and vegetation, and a collection of data on the previous management of the area. The assessment and quantification of soil erosion, and the effect on soil and vegetation under desertification conditions, were also included among the objectives of the work, along with the assessment of the specific impacts related to human activities. Other important objectives were the production of detailed maps of environmental characteristics such as slope, erosion and land use, and the design of plans and strategies for the mitigation of desertification. These elements have been integrated in the production of the Management Plan to Combat Desertification in the Guadalent´ın Basin (Rojo Serrano et al. 1995). This plan has gone beyond the traditional goals of a mitigation plan (the technical design of restoration measures) to enter the scope of policy formulation in an attempt to understand the connection between the physical aspects of desertification and the consequences of European Union policy.
3
METHODOLOGY
According to the formulated objectives, the Management Plan to Combat Desertification in the Guadalent´ın River Basin is composed of three major units: 1. assessment of the current status of the desertification processes, 2. restoration technical design, and 3. policy design and implementation. The design of the watershed management plan has been dominated by certain relevant elements and principles. These include the depth of regional field experience in desertification control, the awareness of the complexity and importance of the socio-economic and human dimensions of the problem, and the attempt to reconstruct the original vegetation communities as the main principle guiding afforestation. The methodology applied to the elaboration of the Management Plan to Combat Desertification in the Guadalent´ın Basin is depicted in a flow-chart (Figure 22.1), and is described below. 3.1
Assessment of the Desertification Status
The first part of the Plan provides the necessary basic understanding and assessment of the desertification process in the Guadalent´ın Basin to formulate the measures of control. This assessment has focused on the vegetation status and the soil erosion conditions in the basin, as two essential indicators of the desertification process in natural areas. The natural vegetation status depicts the historical evolution of the socio-economical rural context, and both vegetation and soil erosion are the integrated results of human activities and land use over time. The study of a widespread phenomenon such as desertification necessitates the collection and management of a large amount of locational data, and the efficient use of these data by means of a geographical information system (GIS). A geographical database, consisting of four basic layers – rainfall erosivity, erodibility of the parent material, slope and vegetation – was developed. A number of derived layers were generated later. From the available data sources, several manuscripts at a 1:200 000 scale were prepared and introduced into the system by manual digitizing. The data on rainfall erosivity were extracted from an existing publication on Iso-R contour maps (ICONA 1988a), while the raw data on soil erodibility, slope and vegetation came from two primary sources: the Erosion Status Map of the Segura Basin (ICONA 1988b) and the Wildland Vegetation Map of Spain (ICONA 1990). The Wildland Vegetation Map is a key piece of evidence in understanding the present desertification status through improved knowledge of the natural vegetation. This map depicts the spatial situation, extent and specific composition of the different areas with natural vegetation, as well as the kind of potential vegetation structure (climatic–structural types of natural vegetation) of the different areas of the basin and the evolutionary stage (maturity level) reached by the present living vegetation community. Therefore, the information contained in this map is not restricted to the current vegetation in the basin, but it includes information relative to
305
Management Plan to Combat Desertification Forest hydrological restoration projects database
Collection of existing cartographic data
Climate, soils, slope and vegetation maps
Attribute assignment to geographical features
Digitization
Analysis of the accumulated experience
Topology building and attribute linking
Collection of existing legislation related to desertification
Policy and law at the European, national and regional Level
Ascertainment of the most effective management actions Identification of the regulations currently in effect Typology of the selected management actions
Guadalentín Basin geographical database
Rainfall erosivity coverage
Classification of the different policy interventions Soil erodibility coverage
Slope class coverage
Geographical analysis: overlay and reclassification
Soil loss coverage
Soil loss map
Vegetation coverage
Polygon aggregation
Homogeneous vegetation units coverage
Description of the vegetation in the basin
Land use map
Evaluation of their incidence and application degree
Proposal of complementary measures
Geographical overlay Desertification control policy package
Reclassification Socioeconomic information Management actions coverage
Assessment Restoration technical design Desertification control policy package
Figure 22.1
Management actions map
MANAGEMENT PLAN TO COMBAT DESERTIFICATION IN THE GUADALENTÍN BASIN
Management plan to combat desertification in the Guadalent´ın Basin
climate (climatic layers), limiting soil characteristics, and maturity level (level of evolution), which is especially useful in the design of the desertification mitigation measures, and for the restoration of the natural vegetation. Once the stage of data collection and automation was completed, topology and attributes were added to the digitized maps and everything was incorporated into the GIS database. A hillslope and
306
Mediterranean Desertification
rill erosion map (soil loss map) was generated by geographical analysis (overlay + reclassification) using the four basic layers of information, following a methodology designed by ICONA in 1986. This is very similar to that of the EC CORINE Project in the mapping and generalization of USLE erosion factors. Despite the necessary simplifications, as in any erosion mapping exercise, the procedure is valid and produces a practical erosion map where the affected areas are delimited and grouped into categories. Such a map is a useful tool to assist in the assignment of priorities and the definition of strategies for the planning of the erosion control measures. The study of the vegetation status in the basin has made use of the Wildland Vegetation Map of the Guadalent´ın Basin. Its polygons were grouped into 44 homogeneous vegetation units, which are homogeneous not only from an ecological point of view, but also consider the predominant kind of human exploitation of the resources. A land-use map (Plate 4 in the colour plate section) was produced by grouping the vegetation polygons into 10 vegetation or land-use categories.
3.2
Restoration Technical Design
The restoration of the vegetation in the Guadalent´ın Basin is not a new idea. Since 1889, the date of the General Survey of the Guadalent´ın Basin, 62 watershed restoration projects covering either the whole basin or sub-basins within the Guadalent´ın Basin have been carried out. At least one project covered the whole basin, although most projects covered partial areas of interest, and some are replicate studies undertaken at later dates. The information included in these 62 projects has been gathered and analysed, using as a data source the Watershed Restoration Projects Database (ICONA 1993), which is the result of collecting, classifying and systemizing the information contained in 525 projects undertaken in Spain since 1885. Each project is summarized in a file containing the most relevant data and stored in a computer database for efficient use. This is a very valuable data source since it holds the details of more than a century of work in natural vegetation restoration. In order to analyse this accumulated information, and check the consequences of the actions proposed in those projects, a field survey was done. This focused on the applied afforestation techniques, and allowed us to identify the most effective actions. It provided a first-hand understanding of the evolution of the vegetation and the effects of the former management and restoration actions in the area. In addition, valuable information regarding natural vegetation adaptation was collected, which is very useful in ascertaining the suitability of different species and afforestation methods used in restoration processes. Important aspects of the restoration technical design were thoroughly analysed, from species selection, site preparation techniques and planting procedures to silvicultural treatments and other regeneration practices. Techniques for biodiversity improvement of the existing artificial forests of Pinus halepensis were also studied and alternatives for the most arid areas in the hyperxerophilous zone were formulated. The information provided by the Wildland Vegetation Map, along with the bioclimatic diagrams (Montero de Burgos and Gonz´alez Rebollar 1983), has been of great help in defining the species to be used in the restoration process, and particularly the natural scope for the use of Pinus halepensis in afforestation and the areas suitable for the recovery of the mature Quercus rotundifolia vegetation. The analysis of all this information allowed selection of a set of effective management actions, consisting of a number of land-use allocation and restoring activities (afforestation, natural vegetation improvement practices and extensive grazing management, among others), and led to the design of a systematic classification of the different action types (Table 22.1). The methodology used to identify the areas affected by the management actions made use of the GIS database, particularly the vegetation and the slope coverage, the available socio-economic information and the typology of the selected management actions. The Management Actions Map (Plate 5 in the colour plate section) was obtained by overlaying the Wildland Vegetation Map with the Slope Map and by reclassifying the resulting polygons according to the criteria in Table 22.1.
Management Plan to Combat Desertification
307
Table 22.1 Typology of management actions in the Guadalent´ın Basin
Vegetation description Forest
Evergreen oak woodland (Quercus rotundifolia) Juniper woodland (J. oxycedrus and J. phoenicea) Mediterranean pine woodland (Pinus halepensis)
Height (m)
CS typea
>7
Any
1.5–7
S, E
1.5–7
3–7
1.5–3
Slope Maturity (%) levelc Any
1. Forest management
Any
4. Silvicultural practices
E
12, 13
Any
3, 4
4. Silvicultural practices
S, E
3, 15
Any
3–5
3. Introduction of Q. rotundifolia
24, 32 16, 25, 32
Any Any
3–5 2
4, 17
Any
3
4. Silvicultural practices 5. Silvicultural practices and/or canopy coverage improvement 3. Introduction of Q. rotundifolia 5. Silvicultural practice and canopy improvement 4. Silvicultural practices 5. Silvicultural practice and canopy improvement 6. Grazing management 2. Reforestation with P. halepensis 6. Grazing management 6. Grazing management
S, E
2
H, G
Medium brushwood
High mountain low brushwood Low brushwood
Atochar (Stipa tenacissima)
Meadow or pastureland Halophilous vegetation Vegetation on stony soil
Management action
1, 7, 10, Any 14, 21, 22, 23, 32 2, 4, 8, 9, Any 10, 11
H, G E, H, G
Mediterranean pine (Pinus halepensis)
HVUb
26, 33
Any
<1.5 H, G 27, 35 Any 0.5–1.5 S, E, H, G 5, 18, 28, >12 34 <12 <0.5 S 6 Any <0.5
E, H, G
19, 29, 35
<0.5
E, H
20, 30
3 2
2, 3 2–4 2–4 2–4
>12
2–4
<12 >12
2–4 2, 3
31
<12 Any
2 (Ls)d 2, 3 1, 2
2. Reforestation with P. halepensis 6. Grazing management 2. Reforestation with P. halepensis 6. Grazing management 6. Grazing management 6. Grazing management
<0.5
H
<3
L, X
36–38
Any
2–4
6. Grazing management
<7
F, J
39, 40
Any
0–4
6. Grazing management
308
Mediterranean Desertification
Table 22.1 (continued)
Vegetation description Agricultural crops with arboreal vegetation
Agricultural crops (alone or with other non-agricultural species)
Irrigation crops in the low Guadalent´ın Valley
Height (m)
CS typea
–
–
–
–
–
–
HVUb 41, 42
43, 44
44
Slope (%)
Maturity levelc
>20
–
7. Reforestation in agricultural areas
<20
–
>20
–
9. Agricultural management and preservation of existing trees 7. Reforestation in agricultural areas
<20
–
Any
–¸
Management action
8. Agricultural management and soil conservation practices 10. Restoration of saline soils and improvement of irrigation practices
a
Climatic structural type. Homogeneous vegetation units. c Maturity level or level of evolution. d Areas with presence of Lygeum spartium. b
3.3
Design of a Desertification Control Policy
Regarding the development of the desertification control policy package, the basic question to be answered is, in the light of present knowledge on the desertification process in Mediterranean countries, what policy proposals should be formulated for the mitigation of this problem? Looking for an answer, the existing agricultural and forestry policies were thoroughly examined, as well as the incentive programmes concerning aid for particular activities having an impact on desertification. For each sector or subject considered, the genesis, evolution and current status of the statutory provisions were studied, trying to analyse and quantify, as far as possible, the degree of accomplishment in the territory of the Guadalent´ın Basin, as well as the investment projections. At a second stage, these policies and programmes were assessed in terms of effectiveness, coordination degree, acceptance, and effects on desertification. Throughout the whole process, the participants related to the implementation of those policies (government officials, representatives of agrarian organizations, etc.) were contacted and asked for their opinion. This analysis, combined with the study of the socio-economic situation and human activities and positions relevant to desertification, has allowed the formulation of guidelines for the implementation of direct restoring activities, as well as incentive programmes to prevent policy failures and contradictory policies. Finally, a set of complementary policy measures were proposed and a number of recommendations on current policy were made.
4 4.1
RESULTS OBTAINED
Assessment of the Desertification Status The first outcome of the assessment stage was the GIS database which consists of information layers on rainfall erosivity (iso-R contour lines), lithology, soil erodibility, slope, sheet and rill erosion,
Management Plan to Combat Desertification
309
vegetation and land use. One more layer was added to the database at the end of the restoration stage: the Management Actions Map (Plate 5). This database has been used in the assessment of the desertification processes, the identification of critical areas, and the design of the restoration actions. As the vegetation communities show, the main climatic feature of the Guadalent´ın Basin is the aridity. The mean annual rainfall ranges from 200 mm year−1 in the low areas of the basin to 800 mm year−1 in Sierra de Mar´ıa, while the rainfall regime is characterized by high intensities and short duration (Cabezas Calvo-Rubio 1995). The relief of the basin is uneven: half of the basin area (49%) has slopes greater than 12% (considered to be the average slope angle for erosion initiation), and almost a third of the basin area (32%) has slopes of over 20% (generalized scouring slope angle). Both the dryness and the steep slopes of the basin characterize it as an area of high vulnerability to desertification. The natural vegetation of the basin was studied in relation to desertification processes. Three main domains, referred to as climatic–structural types, have been identified and delimited in the natural vegetation areas of the Guadalent´ın Basin: subesclerophilous, esclerophilous and hyperxerophilous. The domain of the subesclerophilous forest comprises those areas capable of sustaining a permanent forest of Acer granatense and Quercus faginea, with the presence of Sorbus aria, Sorbus torminalis and Acer monspessulanum. This domain is very scarce in the Guadalent´ın Basin (1.7% of the basin area), only being present in the upper part of the relatively wet Sierra de Mar´ıa and in the highest altitudes of Sierra de Espu˜na. Those areas where the weather and soil conditions allow survival of Quercus rotundifolia forests, with Olea europaea and other species, belong to the esclerophilous domain. In a large area of this domain the forest status may not be reached because of limiting soil conditions, having Quercus rotundifolia bush-like formations instead. The potential community of most of the western highlands is the “encinar” (Quercus rotundifolia) which belongs to the esclerophilous and subesclerophilous types and is still present in some areas of the western ranges of the watershed, mainly mixed with Pinus pinaster and Pinus nigra. In the hyperxerophilous domain, the forests of broad-leaf species cannot survive because the climate is too dry. Broad-leaf tree species are only able to reach shrub status or, at the most, bush form, with the presence of Olea europaea sylvestris and Quercus coccifera as the most evolved potential vegetation. The forests of Pinus halepensis are very well represented in this domain. The hyperxerophilous type characterizes the lowlands of the central and eastern part of the watershed, with the “atochar” (Stipa tenacissima), “pinar” (Pinus halepensis) and, to a lesser degree, the “romeral” (Rosmarinus officinalis), “albaidar” (Anthyllis cytisoides) and other mixed shrubs, dominating a good part of the landscape. The current land-use distribution in the Guadalent´ın Basin (Table 22.2 and Plate 4) shows that 45% of the basin area can be considered as wildland, with the remaining 55% devoted to agriculture. Only 25% of the basin area is covered by forest, whereas different shrub and bush communities cover the other 20%. The status of the vegetation cover does not guarantee the necessary soil protection, leading to a severe erosion problem. In fact, 60% of the basin area is affected by soil losses exceeding the sustainable level, and the surface erosion in 10% of the territory accounts for a 4 mm decrease in topsoil depth each year. Water scarcity and groundwater over-exploitation rank among the most important environmental problems in the basin. The current demand for water exceeds the available resources, and the estimated water reserves in the Guadalent´ın aquifer show a decreasing trend throughout the last 25 years. Traditionally, the Guadalent´ın Basin has been an area characterized by a significant population density in the countryside. In the last few decades, this has decreased due to emigration and concentration of more of the population in urban locations. 4.2 Technical Design for Restoration The design stage of the Management Plan to Combat Desertification in the Guadalent´ın River Basin consists of the definition and description of the actions to be taken and the methods involved, and the location of these actions on a Management Actions Map.
310
Mediterranean Desertification Table 22.2
Current land use distribution within the Guadalent´ın Basin
Land-use class 1. 2. 3. 4. 5. 6. 7.
Area (ha)
Area (%)
Mixed species forest Single species forest High and medium brushwood Low brushwood Stipa tenacissima (Atochar) Crops with trees scattered or in small groups Crops with other species (non-agricultural, non-arboreal) 8. Pure agricultural crops 9. Vegetation on stony soil 10. Reservoirs and water bodies
11 239.6 72 233.5 45 547.3 11 803.1 8608.4 16 674.9 27 479.3
3.31 21.28 13.42 3.48 2.54 4.91 8.10
140 968.4 3215.7 1586.3
41.54 0.95 0.47
Total area
339 356.5
100.00
The survey of past restoration efforts indicates that the mechanized afforestation techniques, either linear or continuous (e.g. terracing and subsoiling), have been more effective than manual ones (e.g. holes, bench terraces and strips) in meeting the critical objectives of decreasing hillslope runoff, retaining and storing as much water and moisture as possible, and digging up the substratum to stimulate root development. Recent reforestation work shows very clearly the distinct response of seedlings to climatic stress when planted in similar areas where the only difference was the site preparation technique used. In the arid and semi-arid lands of Spain, such as the Guadalent´ın Basin, Pinus halepensis is the fundamental afforestation species because of its ecological features which, along with its genuine Mediterranean origin, make it the ideal element to initiate the recovery of the natural vegetation. Its heliophilous (non-tolerant) character, especially in the early stages, makes it a colonizer of spaces without existing tree cover. Some Pinus halepensis afforestation in the basin, such as that of Sierra Espu˜na dating from 1890, is among the most successful episodes of natural vegetation restoration in Spain and can be considered a good example to follow. However, other afforestation using Pinus halepensis in the worst climatic and edaphic situations is unable to reach the forest status. This is why there is concern that, because of its extraordinary performance in the afforestation of arid and semi-arid lands, Pinus halepensis might be over-used in some cases. New developments in the understanding of natural vegetation dynamics and the relationship with climate in the Guadalent´ın Basin have allowed a better definition of the natural potential areas for the use of Pinus halepensis in afforestation. Afforestation with Pinus halepensis is viable in the subesclerophilous and esclerophilous domains. It is also viable in most of the hyperxerophilous area, but special attention must be paid to limiting conditions on soil and climate. Afforestation is generally possible on chalky substratum, particularly in areas facing north, but it must be avoided in the haloxerophilous zones and the driest areas of the hyperxerophilous type. Such situations can be deduced from the presence of some species, which indicate saline soils and/or very dry conditions. As with any colonizer species, the presence of Pinus halepensis is theoretically ephemeral. Once it has created the shadowy, wetter and cooler conditions on the soil and microclimate, other species, typically those of the gender Quercus in Spain, are able to germinate and persist. The survey has identified the procedures for Quercus rotundifolia regeneration and the techniques involved, bearing in mind that the actions aiming at the recovery of the mature Quercus rotundifolia vegetation can only be undertaken in the subesclerophilous and esclerophilous domains. The hypothesis that it is usually more efficient to favour the spontaneous progress of such species than to afforest directly
311
Management Plan to Combat Desertification
with them can be tested in the future. In this context, the vegetation restoration process is not confined to the afforestation itself, but includes the subsequent silvicultural treatments in the created forest to facilitate the described natural evolution towards the mature forest. According to the methodology described in section 3.2, a number of effective management actions were selected and classified (Table 22.1). This classification established the guidelines for the afforestation programme as well as the improvement programme for currently forested sites, and the formulation of alternatives for the other natural areas. The actions proposed included carrying out afforestation and silvicultural treatments. Generally, the plants to be used in afforestation must be 1 + 0 or 2 + 0 seedlings cultivated in containers and having a well-developed rootball. In most of the basin, and depending on the area to be reforested, the species must be Quercus rotundifolia and/or Pinus halepensis. In some areas it will be possible to use Pinus pinaster or Pinus nigra, mixed with Quercus rotundifolia. A wide variety of site preparation methods have been traditionally used in the Guadalent´ın Basin, showing different success rates. As stated above, mechanized methods seem to attain better results than manual ones, and should be encouraged whenever possible. The silvicultural treatments include cleaning (weeding), thinning, pruning, planting under cover to improve density, and stem cutting under the soil surface for Quercus rotundifolia regeneration. The incidence of the proposed management actions in the Guadalent´ın Basin is depicted in the Management Actions Map (Plate 5). This map stratifies the territory of the basin in ten categories, with the eleventh class including the water bodies in the basin. The areal distribution of the management actions is shown in Table 22.3. The proposed management actions would cause a transformation of the land uses in the basin. Thus, the reforestation work, affecting 15% of the basin area, would produce an increase in the forest area and a decrease in the brushwood and agricultural areas. Similarly, silvicultural treatments in the immature forest stands would favour the succession of large areas to the forest category. Table 22.4 shows a comparison between the areal distribution of land use in the basin before and after the proposed management actions. The agricultural area would decline from 55% to 48% of the basin area, and this 12% decrease would affect marginal land with a low agricultural suitability. It should be noticed from the map that reforestation in agricultural areas would take place along the Table 22.3 Area of the Guadalent´ın Basin affected by management actions shown on the Management Actions Map
Management actions 1. Forest management 2. Reforestation with Pinus halepensis 3. Reforestation and improvement of Quercus rotundifolia 4. Silvicultural practices in existing forest 5. Silvicultural treatments and/or tree planting depending on density conditions 6. Grazing (range) management 7. Reforestation in agricultural areas 8. Agricultural management and soil conservation practices 9. Preservation of existing arboreal vegetation 10. Restoration of saline soils and improvement of irrigation practices 11. Reservoirs and water bodies Total area
Area (ha)
Area (%)
39 973.4 28 896.2 11 707.4
11.78 8.51 3.45
28 093.4 13 445.8
8.28 3.96
30 531.5 23 180.4 86 744.2
9.00 6.83 25.56
13 352.6 61 845.3
3.93 18.23
1586.3
0.47
339 356.5
100.00
312
Mediterranean Desertification
Table 22.4 Land use distribution in the Guadalent´ın Basin before and after the proposed management actions
Land-use distribution
1. 2. 3. 4. 5. 6. 7.
Area (ha) Present
Projected
Mixed species forest Single species forest High and medium brushwood Low brushwood Stipa tenacissima (“Atochar”) Crops with trees scattered or in small groups Crops with other species (non-agricultural, non-arboreal) 8. Pure agricultural crops 9. Vegetation on stony soil 10. Reservoirs and water bodies
11 239.6 72 233.5 45 547.3 11 803.1 8608.4 16 674.9 27 479.3
19 299.1 126 208.7 16 347.2 5683.5 5285.1 13 352.6 22 687.6
140 968.4 3215.7 1586.3
125 902.0 3004.4 1586.3
Total area
339 356.5
339 356.5
borders of the agricultural polygons, in areas adjacent to wildland or forest polygons. The wildland area would rise from 45% to 52% of the total basin area, a 15% increase. The most important change would correspond with the relative representation of the forest within the wildland area, which would increase from 55% to 83%. 4.3
Desertification Control Policy Package
Traditionally, the policies that have played a role in desertification have been agricultural and environmental. However, since 1986, with the approval of the Single European Act, the policy aiming at economic and social cohesion attains a special relevance. It is strongly connected with Common Agricultural Policy in aspects related to rural development, and both policies share concern for natural resource conservation and environmental protection as priority objectives. This relationship is consistent with the trend towards the integration of environmental issues into other policies, and explains why, except for some specific projects or initiatives, environmental policy does not consist of a legislative body with a direct development in specific programmes of action, at least in relation to the problem of desertification. Cohesion and agricultural policies are dealt with in turn below.
Policy of Economic and Social Cohesion Two main instruments were developed by the EC to reinforce the economic and social cohesion among the regions of Europe. These were the Structural Funds (basically the European Agricultural Guidance and Guarantee Fund-Section Guidance and the European Regional Development Fund), and the Cohesion Fund. In Spain, the Structural Funds have played an important role in financing investments for erosion control. The application of the Community Support Framework (CSF) 1989–1993 in the Guadalent´ın Basin was satisfactory and entailed an average investment of 1.03 million euros year−1 . A similar result is expected for the CSF 1994–1999 since the contribution of the Structural Funds to the operative programmes related to desertification has been maintained, in accordance with their new functions regarding the protection and conservation of natural resources. Investments in the field of erosion control and vegetation cover restoration in Spain charged to the Cohesion Fund and its predecessor the Cohesion Financial Instrument have been important and are expected to continue to be important in the future.
Management Plan to Combat Desertification
313
In the Guadalent´ın Basin, during the 1993–1994 campaign, 699 000 euros were invested, charged to the Cohesion Financial Instrument (CFI). From 1995 to 1999, the Cohesion Fund was expected to contribute approximately 3.61 million euros year−1 to the total investment in the basin. Nevertheless, the expectations are not being fulfilled since there has been a shift in the National Administration priorities and the available Cohesion Funds have been directed elsewhere, especially in the field of large infrastructures.
Agricultural Policy: 1992 CAP Reform The important reform of 1992 produced a revolution in the mechanisms used previously by the CAP. A shift was made from an agricultural protection system based on aid to maximum production (which favoured the intensification of agriculture) to a system based on direct aid to income, aimed at reducing agricultural surpluses, but at the same time acknowledging the importance of keeping enough farmers in the countryside as the only way of preserving the environment. Thus, among the most important measures are the trend towards extensive cultivation, the abandonment of herbaceous crops and, generally, the use of agricultural techniques in tune with the environment. The analysis of the current legal status has focused on the following sectors of the CAP. Agricultural structures policy From its beginning more than 20 years ago, this policy has given rise to many measures that rank among the most important actions in fighting erosion. These include set-aside, environmental protection measures in agriculture, afforestation on agricultural land, sensitive areas, and aid to unfavoured areas. However, CAP reform has removed part of the contents of EEC Regulation 2328/91 on the improvement of the agricultural structures effectiveness, since the set-aside programme was incorporated in the Herbaceous Crops Common Market Organization (CMO), while the extensive cultivation promotion and the aid to sensitive areas were included in EEC Regulation 2078/92 on environmentally sound agricultural practices (ECC 1991, 1992a). Production reconversion has developed into afforestation in agricultural land, and the possibility of growing non-food crops in the set-aside areas. These production shifts are a priori more beneficial for the objective of erosion control. The only measure relevant to desertification still ascribed to EEC Regulation 2328/91 is the aid given to investments in improvement plans, which can include among their objectives the protection and improvement of the soil and vegetation cover. The aid given to investments in improvement plans could be another way of fighting erosion, especially if the regional governments in charge of these matters encouraged the implementation of improvement plans that included erosion control among their objectives, instead of leaving everything to the initiative of individual farmers. Unfavoured agricultural areas according to Directive 75/268/EEC (EEC 1975) Some of the municipalities included in the Guadalent´ın River Basin qualify as unfavoured areas, and receive aid aimed at the maintenance of a minimum population level and the conservation of the natural environment. One of the specific aids for these areas, the Compensation Indemnity (CI), is an income support measure – an amount of money received by those farmers who maintain agricultural activity and fulfil certain requirements. The poor acceptance and effectiveness of this measure can be explained by the low amount of the allowances. There are a decreasing number of beneficiaries. Common Market organizations The sectors of herbaceous crops, nuts, and sheep and goats are particularly related to erosion, and the evolution of these land uses, and the associated agricultural and grazing practices, are important in explaining land degradation in the basin. Herbaceous crops The set-aside derived from the new policy has not greatly affected the Guadalent´ın Basin, and is not expected to change in the future. Despite the low acceptance and incidence of this policy, there is an incentive to increase fallow land, and fallow practices have an important effect on desertification. Therefore, the set-aside in the Guadalent´ın Basin should be subject to the fulfilment of specific requirements on environmentally sound fallow practices in arid
314
Mediterranean Desertification
zones. Something is being done under EEC Regulation 2078/92, but its voluntary character reduces its effectiveness. Nuts In the recent past the aid given to nut producers has provided significant support to almond cultivation in the Guadalent´ın Basin. Although aid is directed towards existing crops, it is allowing recovery of some previously semi-abandoned marginal almond groves. Some of these marginal crops are located in areas with steep slopes and are causing severe erosion problems. Unfortunately no environmental considerations are included in this regulation. The concession of the aid should be subject to the fulfilment of a series of requirements on maximum allowable slope (around 20%, for example) and on the execution of soil conservation practices in the areas with slopes between 12 and 20%. Fragile areas not suitable for almond cultivation (slopes > 20%) should not qualify for those aids, and should be reforested. Afforestation in these fragile areas is being obstructed by this policy, which has halted the abandonment of almond crops on unsuitable marginal land, and thus has a negative influence on the desertification process in the basin. Sheep and goats In the region of Murcia and the Guadalent´ın Basin the subsidies have not affected the sheep and goats census. Overgrazing is not a matter of concern in the Guadalent´ın Basin as a whole. However, in the specific areas where the problem may arise, there are no regulations to control the grazing load. In order to prevent subsidized overgrazing, the CAP should include, in the subsidy concession process, the requirement of a balance between grazing load and carrying capacity, a requirement already in effect for cattle. Accompanying measures Since EEC Regulation 233/94 (ECC 1994) states that the member countries can establish environmental protection measures depending on the status of the areas to be used for subsidized sheep and goat production, the Spanish national and regional administration should require the construction of a stock-breeding plan from the producers, identifying the areas to be used and including an assessment of their maximum carrying capacity (Boza et al. 1994). The CAP reform included a package of accompanying measures with three types of programmes: aid to anticipated retirement in agriculture (EEC 1992b); aid to agricultural production methods compatible with environmental protection and natural environment conservation (EEC 1992a); and aid to forest measures in agriculture (EEC 1992c). These last two programmes have an important role in erosion control, but their implementation has been inconsistent. Agricultural practices compatible with environmental protection (EEC Regulation 2078/92) In Spain, the development of this regulation has experienced considerable delays. The agro-environmental programmes did not start until 1996 in most Spanish regions. Therefore the evaluation of the effectiveness and acceptance of this policy is somewhat premature, although it can be said that its implementation has not been as satisfactory as expected. One of the main causes of this situation has been a lack of budgetary allocations, which has led to the scarcity of financial contributions. However, it seems obvious that these programmes can have a positive influence on desertification control if we take into account that they include the following measures:
• •
Agricultural practices in fallow land aiming at erosion control. Practices include the prohibition of burning stubble fields, the restriction of the livestock load, and minimum tilling practices. Specific measures for erosion control in abandoned cultivated areas and steppes (Andaluc´ıa), or measures directed towards soil conservation in the protected natural areas (Murcia).
Unfortunately, these programmes are voluntary, and it would be more effective if the environmental measures were integrated in the CMO regulations, which are compulsory. Reforestation in agricultural areas (EEC Regulation 2080/92) The development of this regulation in Spain has established a maximum subsidy for farmers who decide to reafforest their land which is half the amount allowed by the European Regulation (ECC 1992c). Therefore this has attained a poor acceptance level. At the expected rate of investment, the reafforestation of
Management Plan to Combat Desertification
315
23 000 ha of agricultural land, as recommended by the Management Plan, will take more than 75 years. However, the instrument itself could be very useful if more funds were allocated and the subsidies increased. The development of this policy must guarantee that a higher priority be assigned to the areas where the desertification problem is most serious. The possibility of increasing the aid offered to farmers in order to attain the reforestation of these fragile areas should be studied. In Spain, this directive was developed together with EEC Regulation 1610/89 (ECC 1989) on forest development and management actions in rural areas, and was presented as a National Reafforestation Plan, which raised great expectations. However, this plan is mainly directed towards land owners and the reafforestation of agricultural land. In fact, aid to forest development and management in wildland areas is considerably lower than aid available for reafforestation in agricultural areas. This discrimination between agriculture and forestry means that remedial action on forest land is not sufficiently encouraged. In all Mediterranean countries, and particularly in the Guadalent´ın Basin, the largest areas subject to desertification are wildland areas, and consequently are left out of the scope of the present regulations. The underlying problem is the lack of a Common Forest Policy at the European level, which in the Mediterranean semi-arid areas should not be orientated towards production, but to natural vegetation restoration instead. At the request of the European Parliament, the development of a European Forestry Strategy is in progress, but it is expected to be a slow and difficult process. With regard to the agricultural land afforestation programme, after several campaigns, it seems that the initial projections will not be met. The lack of application has been induced by several problems such as drought, budget shortage, and some management deficiencies. The application of the aid programme for forest investments in rural areas is even more delayed, since it has not been properly developed yet in many regions of the country. Despite these problems, the available public funds for erosion control investments are considerable. The executed investment in the Guadalent´ın Basin under the current policy instruments had allowed the reafforestation of 574 ha year−1 and the carrying out of 773 ha year−1 of silvicultural treatments by 1994. The projected investments are expected to quicken this pace. In the near future, the projections should allow the reafforestation of more than 1000 ha year−1 and the execution of silvicultural practices in over 2400 ha year−1 . If these projections come true, and are maintained in the future, the period needed to complete the actions proposed by the Management Plan in the wildland areas (28 896 ha of reforestation and 53 247 ha of silvicultural treatments) will be around 25 years. Finally, there is an ongoing process of reform of future European Policy. It began with the presentation by the Commission, in July 1997, of Agenda 2000. The objective of this reform is to deepen and extend the 1992 CAP reform while assigning at the same time a new priority to rural development. Among the proposals are the reinforcement of agro-environmental measures, the encouragement of silvicultural activities in rural areas, the redirection of aid to less favoured areas to provide sustainable agriculture, and the integration of environmental concerns in the common market organizations.
5 CONCLUSIONS A new approach has been used in the design of a desertification mitigation plan in a natural area. The global conception of the mitigation plan itself stresses two new features. First, the technical design of the restoration measures relies mainly on knowledge of the natural vegetation and its degree of degradation/evolution, considering that natural vegetation status depicts the present and past impacts of climate and human land use. Second, existing policy and its incidence on desertification will be analysed. The following conclusions can be formulated: • The Guadalent´ın River Basin is subject to a severe erosion problem, since nearly 60% of the basin area is affected by soil losses, which exceeds an acceptable level.
316 • •
• •
• •
•
•
•
• •
•
Mediterranean Desertification
The main way to control desertification is the regeneration of vegetation cover in the basin. More than 100 years of experience allow identification of the actions to be taken and the methods to apply in order to attain this objective. The species to be used in the restoration process should be Quercus rotundifolia in the subesclerophilous and esclerophilous domains, and Pinus halepensis in the esclerophilous and most of the hyperxerophilous domains. Part of the chalky soils can also be reforested with Pinus halepensis, but reforestation is discouraged in the haloxerophilous zones and the driest areas of the hyperxerophilous domain. The most effective restoration methods have proved to be those reforestation methods that improve water retention capacity and infiltration by means of careful preliminary site preparation. According to the Management Plan to Combat Desertification in the Guadalent´ın Basin, and particularly, the Programme of Proposed Management Actions, 52 077 ha in the basin should be reforested (28 896 ha in wildland areas and 23 181 ha in agricultural areas), and 53 247 ha should be subject to silvicultural practices. Those actions, which imply a direct intervention on vegetation, affect more than 30% of the total basin area. Other actions, such as forest, range and agricultural management, as well as soil conservation practices, will cover the remaining 70% of the basin area. The proposed management actions would cause a transformation of the land uses in the basin based on a decrease in the agricultural area, the resulting increase in the wildland area, and a significant enlargement in the relative representation of the forest within the wildland area. Only a small percentage of the area is under the control of the Public Administration. Therefore, most management actions will fall within private ownership. It is essential to implement a set of policy instruments that allow the proposed actions to be undertaken as a result of private decisions, thus minimizing the risk of further conflicts. Existing European Policy related to desertification can be split into Cohesion Policy and Agricultural Policy. Some of the activities favoured by the CAP may conflict with the objective of desertification control. The aid directed towards almond cultivation on steep slopes is causing one of those conflicts. It seems that the regulations on almonds in the basin should include constraints on maximum allowable slope and soil conservation practices in order to qualify for the aid, and should facilitate the reafforestation of the areas not fulfilling the requirements. As for sheep and goat production, a balance between grazing load and livestock carrying capacity is necessary. The national or regional administration could require from the producers the construction of a stock-breeding plan identifying the areas to be used and their maximum carrying capacity. Such a requirement could also be incorporated by the CAP in the sheep and goats CMO. The Management Plan advises the reforestation of 23 181 ha of agricultural land, a 75-year task at the expected rate. However, it seems that the current aid aimed at reforestation of agricultural lands is a good way of attaining this goal, if more funds are allocated and the subsidies are increased. The projected investments to carry out the necessary actions in the wildland areas will allow the completion of this work over a period of approximately 25 years. However, the investment is subject to the fulfilment of the projections and the availability of land. Subsidies and financial aid for forest owners (both private owners and municipalities) should be reinforced, and the funds allocated should be used for – development of forest management plans, – execution of silvicultural treatments, – reforestation and vegetation improvement. These actions, aimed at mitigating the discrimination between agricultural and forest owners, should be integrated in a wider framework, e.g. watershed or regional management plans. The implementation of policy measures should be subject to the creation of a management plan determining the capacity and vulnerability of the territory to accommodate different activities. Additionally, this plan will improve the efficiency of policy measures by concentrating investments wherever they are most needed.
Management Plan to Combat Desertification
317
• The lack of a Common Forest Policy (CFP) underlies many of the problems related to desertification. The development of future European Forestry Strategy must take into account that Forest Policy should be understood not only in terms of forest product procurement, but mainly in the sense of wildland resource conservation and restoration, as is intended in Spain. The development of the European Policy reforms proposed in Agenda 2000 must be a good opportunity for the encouragement of desertification control actions, aiming at the sustainable development of arid and semi-arid environments.
REFERENCES Boza J, Gonz´alez Rebollar JL and L´opez Alados MC (1994) La ganader´ıa extensiva en el sudeste ib´erico. Evaluaci´on de la capacidad sustentadora en pastos semi´aridos. Propuestas metodol´ogicas. (Extensive stockbreeding in the Iberian South-East. Assessment of the carrying capacity in semiarid pastures. Methodological proposals). Planificaci´on Ganadera del Sureste Ib´erico. Proyecto LUCDEME. Granada. Cabezas Calvo-Rubio F (1995) Guadalent´ın Basin: Water Resources Policy. In: MEDALUS II Project 4 Final Report: Research and Policy Interfacing in Selected Regions, pp. 168–192. ECC (1975) Council Directive 75/268/EEC of 28 April 1975 on mountain and hill farming and farming in certain less-favoured areas. In Official Journal L 128, 19/05/1975, pp. 0001–0007. ECC (1989) Council Regulation (EEC) No 1610/89 of 29 May 1989 laying down provisions for implementing Regulation (EEC) No 4256/88 as regards the scheme to develop and optimally utilize woodlands in rural areas in the Community. In Official Journal L 165, 15/06/1989, pp. 0003–0004. ECC (1991) Council Regulation (EEC) No 2328/91 of 15 July 1991 on improving the efficiency of agricultural structures. In Official Journal L 218, 06/08/1991, pp. 0001–0021. ECC (1992a) Council Regulation (EEC) No 2078/92 of 30 June 1992 on agricultural production methods compatible with the requirements of the protection of the environment and the maintenance of the countryside. In Official Journal L 215, 30/07/1992, pp. 0085–0090. ECC (1992b) Council Regulation (EEC) No 2079/92 of 30 June 1992 instituting a Community aid scheme for early retirement from farming. In Official Journal L 215, 30/07/1992, pp. 0091–0095. ECC (1992c) Council Regulation (EEC) No 2080/92 of 30 June 1992 instituting a Community aid scheme for forestry measures in agriculture. In Official Journal L 215, 30/07/1992, pp. 0096–0099. ECC (1994) Council Regulation (EC) No 233/94 of 24 January 1994 amending Regulation (EEC) No 3013/89 on the common organization of the market in sheepmeat and goatmeat and Regulation (EEC) No 3493/90 laying down general rules for the grant of premiums to sheepmeat and goatmeat producers. In Official Journal L 030, 03/02/1994, pp. 0009–0010. ICONA (1988a) Agresividad de la Lluvia en Espa˜na. Valores del Factor R de la Ecuaci´on Universal de P´erdidas de Suelo, Ministerio de Agricultura, Pesca y Alimentaci´on, Madrid. ICONA (1988b) Mapa de Estados Erosivos. Cuenca Hidrogr´afica del Segura, Ministerio de Agricultura, Pesca y Alimentaci´on, Madrid. ICONA (1990) Mapa Forestal de Espa˜na, Sheets 6–10 (BAZA) and 7–10 (MURCIA), Ministerio de Agricultura, Pesca y Alimentaci´on, Madrid. ICONA (1993) Fondo de Proyectos de Restauraci´on Hidrol´ogico-Forestal , Instituto Nacional para la Conservaci´on de la Naturaleza, Madrid. Mairota P, Thornes JB and Geeson N (1998) Atlas of Mediterranean Environments in Europe: The Desertification Context. John Wiley, Chichester. Mart´ınez Artero JA (2001) Repoblaciones Hidrol´ogico-Forestales en la Cuenca del R´ıo Guadalent´ın. In Montes, No 65, pp. 66–69. MEDALUS (1995) MEDALUS III Start-Up Document, Lisbon. (MEDALUS http://www.medalus.leeds.ac.uk). Montero de Burgos JL and Gonz´alez Rebollar JL (1983) Diagramas Bioclim´aticos. Instituto Nacional para la Conservaci´on de la Naturaleza, Madrid. Rojo Serrano L, Mart´ınez Artero JA, Garc´ıa Robredo F, Mart´ınez Ruiz A and Carrera Morales JA (1995) Guadalent´ın Basin: Management Plan. In MEDALUS II Project 4 Final Report: Research and Policy Interfacing in Selected Regions, pp. 130–167.
Section VII
The Agri Basin, Southern Italy
23
General Description of the Agri Basin, Southern Italy
F. BASSO, E. BOVE AND M. DEL PRETE
University of Basilicata, Potenza, Italy
1 INTRODUCTION The Agri Basin is a target area, specifically chosen for the study of desertification processes within the MEDALUS II Project. Land degradation in this area is linked to the fragile geolithological nature of the hilly landscape, susceptibility to water erosion as a result of rainfall events, and the human impact of deforestation (Basso 1995). Throughout history forested land has been converted to arable land, but when the yields from arable crops failed to provide a good living for farmers, arable land has been abandoned or included in set-aside packages (Bove and Quaranta 1996). Land degradation has different forms in each of the three areas that characterize the basin but the consequences of land degradation are visible in the occurrence of frequent landslides. Careful management is needed to limit future soil erosion in these areas, using methods examined in our research (Basso et al. 1996a,b,c, 1998).
2 PHYSICAL ENVIRONMENT The Agri Basin is located in the region of Basilicata in southern Italy. It is economically and socially one of the less favoured areas of Europe and is suffering from land degradation problems that arise from its seasonal extremes of climate (Cantore et al. 1987), and the nature of the geology, with its unstable marine clays and easily eroded sands and conglomerates (Del Prete 1995). It is a marginal region from a climatic point of view, but some parts have already experienced desertification and land abandonment, and the potential results of global warming now threaten other parts. 2.1 Physical Characteristics
The Agri Basin lies between 40◦ 30′ and 40◦ 7′ latitude north, situated in the heart of the Basilicata Apennine Chain in a NW–SE direction and covering about 173 000 ha (Figure 23.1). The Agri River is about 136 km long and begins its course in the Lama Mountains (1200 m a.s.l.) in the municipality of Marsico Nuovo, in the province of Potenza. Its main tributary is the Sauro River and it flows from west to east into the Ionian Sea. Its course is interrupted by artificial reservoirs, the Pertusillo, the Gannano, and the Marsico Nuovo. The Agri Basin includes 29 municipalities in the province of Potenza, between the semi-arid areas of Mediterranean coast and the more humid Apennine mountain range, and can be divided into three sections: Upper Val d’Agri, Middle Val d’Agri and Lower Val d’Agri. The three areas are characterized by different geological, pedological, climatic, and consequently agronomic and socio-economic conditions. The Upper Val d’Agri is situated at around 600 m a.s.l. upstream of the Pertusillo Dam. This area is fairly flat, having formed after the drying up of an ancient Pleistocene lake, and covers a little less than 60 000. Of the 12 municipalities that form the Upper Agri Mountain Community, 10 are located here. Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
322
Mediterranean Desertification
N
Figure 23.1 Topography of the Agri Basin
The Middle Val d’Agri is the area from the Pertusillo Dam downstream to the point where the Sauro torrent flows into the Agri River (Stigliano). Here there are 14 municipalities covering about 91 000 ha, equal to 53% of the total hydrological basin. This area is almost entirely covered by the well-known calanchi badlands. The Lower Val d’Agri, located on the fertile Metaponto soils, includes five municipalities and covers 22 000 ha, which is 13% of the total Agri Basin.
2.2
Geology
The Agri Basin exhibits all the main structural elements of the Apennine Chain and the Bradanic Foreland. The Apennine Chain, present in the Upper Valley, is formed by a complex of nappes made up of numerous tectonic units derived from the fragmentation of larger paleogeographical units. These broke up in the Miocene period to form the following main units: 1. 2. 3. 4. 5.
Liguride complex, comprising pellites, calcareous clays and arenaceous-conglomerates and flysches. Sicilide complex, represented by the Nocara Flysch, variegated clays and the Tusa tuffs. Carbonatic complex, made up of limestones and dolomites deposited in a shallow sea platform environment. Lagonegro complex, comprising siltstones and limestones, radiolarian schists and red flysch. External Flysch complex, almost entirely represented by flysch and sandstones.
The Bradanic Foreland is located to the east of the Apennine Chain, made up of Plio-Quaternary sequences, and is dominated by blue clays of marine origin that pass laterally into the Montemarano sands and Irsina conglomerates. Embedded in the Blue Clays is the allochthonous Metaponto Nappe. The Blue Clays in association with relatively recent tectonic movements and continuing earthquakes give rise to extensive mass failures. These belong to several “cycles” of formation and create a significant hazard in both built-up and rural areas of the region. In the Middle Val d’Agri, calanchi badlands are widespread and contribute to the marginal economic nature of the region (Figure 23.2). The calanchi are developed on the Blue Clays, which have been deeply incised by the Agri tributaries. Two distinctive morphological environments can be identified in these hilly areas: (a) high relief, formed where the Blue Clays are overlain by sandy-conglomerate facies, with spectacular calanchi landscapes of nested gully systems separated by knife-edge ridges; and (b) medium to low energy relief formed exclusively on Blue Clays with
General Description of the Agri Basin
Figure 23.2
323
Badlands or calanchi in the Agri Valley
progressively degraded forms towards the valley axis. Here rounded, isolated forms called biancane occur (Del Prete 1995). 2.3 Tectonics
The early part of the tectonic evolution was characterized by two periods of overthrusting before the lower Miocene and at the end of the lower Miocene. The sea invaded the valley in the upper Miocene and marine deposition continued through the Pliocene and the Calabrian. Uplift occurred simultaneously further west, at the Thyrrenian coast, at the end of the upper Miocene and lower Pliocene, and in the upper Pliocene. As a result, flysch deposits were overthrusted towards the east forming the plastic overthrust sheets that are subject to intense mass movements and landslides (Cotecchia and Del Prete 1997). A relatively uplifted zone borders the Santo Arcangelo Basin to the west, and is crossed by the Agri River in a deep gorge (the location of the Pertusillo Dam). Upstream, faulting and relative subsidence have led to the formation of a structural basin and the accumulation of coarse Pleistocene debris in the lower part of the broad depression of the Upper Agri Basin (Dimase and Galligani 1979; Mancini 1980). 2.4 Landslide and Erosion Phenomena
The Agri Basin area has the highest frequency of landslides in the Basilicata region. In the mountainous and hilly areas landslides are periodically triggered or reactivated by rainfall, earthquakes and unsuitable agricultural techniques. Many rural communities exist on unstable lithology (Ferrara et al. 1995). The areas most prone to seismic activity are not the most prone to landslides. Areas most subject to earthquakes often correspond to zones with a more solid geological constitution (Del Prete and Hutchinson 1988). The Agri River transports about 430 m3 of eroded material per year to the sea. The combined effects of the constitution of the Blue Clays, the tectonic history of the affected areas and the exposure of the slopes to rainfall and solar radiation influence the genesis and development of calanchi and biancane and associated land-forms. 2.5 Climate Conditions
Basilicata as a whole and the Agri Valley in particular can be considered to have a cool temperate Mediterranean climate (Cantore et al. 1987). However, there is a strong gradient from the coastline to the interior mountains. In the coastal areas there is a summer drought (June–September), with rainfall less than 100 mm, and the mean monthly temperature of the warmest months being greater than 23 ◦ C. Inland at higher altitudes (800–1000 m), the climate becomes cooler (20–23 ◦ C) though
324
Mediterranean Desertification
rainfall in summer remains similar. Above 1600 m, rainfall during the critical summer months exceeds 150 mm.
Rainfall Across the Agri Basin the average annual rainfall is 851 mm year−1 , two-thirds of which falls between November and February. The average annual precipitation in the basin increases moving upstream, and from east to west. In the lower part of the basin, and in particular close to the coastal zone, precipitation is less than 600 mm. Over most of the basin the precipitation ranges between 600 and 1000 mm, but in the upper part the average precipitation is higher than 1200 mm. The wettest month of the year is generally December, followed by November; the driest month is July, followed by August. Intense rainfall events capable of causing a lot of damage to the unstable landscape are common. The maximum daily precipitation recorded (1921–1970) was at Marsico Nuovo (29 November 1929: 250 mm) and Aliano (22 February 1931: 230 mm). The maximum hourly precipitation was recorded in Policoro on 24 November 1959 at a rate of 80.0 mm h−1 . This was in association with a cloudburst that occurred over Basilicata and the Ionian coast of Calabria on 24 and 25 November 1959. During the two days of the event, precipitation values totalling more than 300 mm were recorded, with maximum values of 365.6 mm in Pisticci and 322.4 mm in Montalbano Ionico. The resulting widespread flooding was monitored at the hydrological station of Tarangelo where the flood flow rate was 254 m3 s−1 , with a specific discharge of about 0.5 m3 s−1 km−2 .
Temperature Mean annual temperature isotherms (Figure 23.3) illustrate the effect of elevation towards the head of the basin. Across most of the basin the average annual temperature ranges from 12 ◦ C to 15 ◦ C, and the entire Agri Basin is included between the 8 ◦ C and 16 ◦ C isotherms. Climatic indices can be used to summarize annual conditions, such as the De Martonne (1926) Aridity Index P /(T + 100). 2.6
Hydrology
In the upper part of the Agri Basin the most important springs are located at the contact between the calcareous and dolomitic formations of the Apennines, and the clay formations. Springs of smaller size but greater in number are found in the middle part of the basin, while in the coastal area there are a few springs directly supplied by the phreatic aquifer. The three most important hydrometric
Figure 23.3 Mean annual isotherms across the Agri Basin
General Description of the Agri Basin
325
Table 23.1 Discharges of the three hydrometric stations
Station
Le Tempe Grumento Tarangelo
Discharge (m3 s−1 )
No. of observation years
Qmax
Qmed
Qmin
40 14 33
110.00 87.78 288.00
4.18 6.60 10.19
0.35 1.47 0.90
stations are located in the upper part of the basin. One is located at Le Tempe (585 m a.s.l.) 100 km from the river mouth; it has operated since 1929 with a catchment area of 174 km2 . In 1961, the station of Grumento, located 9 km downstream, began taking measurements (525 m a.s.l.; catchment area 278 km2 ). The third station, located in Tarangelo, has been operating since 1925 (470 m a.s.l.; catchment area 507 km2 ). The maximum, mean and minimum discharge values for each of the three stations respectively, with the number of available observation years, are listed in Table 23.1 and reflect water availability. The artificial reservoirs of Pietra del Pertusillo (146.0 × 106 m3 ) and Gannano (2.6 × 106 m3 ) were built with barrages on the Agri River. The Pertusillo Reservoir, with its larger capacity, plays an important role in the Water Management Plan, controlling water for power supply and irrigation. The Gannano Reservoir is located further downstream and is also used to supply irrigation water, as is the Marsico Nuovo upstream. 2.7 Natural History
The natural vegetation of the Agri Basin may be described using the phytoclimatic bands proposed by Pavari. These are based on the following climatic parameters: average annual temperature, average temperature of the coldest and hottest months, average of the minimum and maximum annual temperatures, rainfall distribution, annual precipitation and summer rainfall. The Agri Basin is covered by the Lauretum (all three subzones), Castanetum and Fagetum bands (see Chapter 28). The Upper and Middle Agri Valley have not seen recent deforestation to the same extent as other parts of Basilicata. Arboreal species such as Castanea sativa, Quercus cerris and Quercus frainetto are common. At higher altitudes, such as the slopes of Mount Volturino, impressive beechwoods are found (Corbetta et al. 1991). Throughout the Agri Basin soil characteristics, such as lithology, morphology, soil depth, stoniness and drainage properties, have played as important a part as climate in determining natural vegetation cover. However, the action of humans, in deforestation, agriculture and the abandonment of failed enterprises, has also affected much of the vegetation. Vegetation in the central part of the valley is less disturbed because of the greater soil depth upon which pedogenesis factors have taken place. In the mountainous part of the basin drainage conditions are more variable and allow vegetation communities with a wide range of water requirements. The Agri Basin is an important habitat for some rare animal species, such as the spotted salamander, an amphibian that lives in the karstic caves. The Pertusillo Reservoir has become a place where aquatic birds rest during their long migration flight.
3 SOCIO-ECONOMIC ASPECTS 3.1 Population
At the end of 1991 the population of the Agri Basin reached about 100 000, which was about 16% of the whole Basilicata region. The mean population density was about 50.0 inhabitants km−2 , ranging from 12.7 at Craco to 216.3 inhabitants km−2 in Policoro (ISTAT 1997). In general, as in the rest
326
Mediterranean Desertification 45000 40000 35000
Population
30000 25000
Upper Val d'Agri
20000
Middle Val d'Agri
15000
Lower Val d'Agri
10000 5000 0 1861 1871 1881 1901 1911 1921 1931 1936 1951 1961 1971 1981 1991 Year
Figure 23.4 Residential population trend in the basin 1861–1991
of Basilicata, there have been two strong emigration periods, and a period of population expansion between the two World Wars. Figure 23.4 shows how the population of the three sections of the Agri Valley changed between 1861 and 1991. In the Upper Val d’Agri, Montemurro was the municipality that was most affected by emigration and after 130 years the resident population had more than halved. Also, at the beginning of the 20th century, mass emigration started from Viggiano and Moliterno, mainly directed towards Australia. In the Middle Val d’Agri the resident population decreased by a smaller amount, and Armento was the most dramatically hit by emigration, with only 30% of the resident population still living there. In places such as San Chirico Raparo geographical isolation has contributed to the accelerated mass emigration of young people. The municipalities that registered an increase in resident population were Roccanova, Stigliano and Sant’Arcangelo. In contrast to the upper and middle valleys, the Lower Val d’Agri resident population increased by a factor of four during the 1861–1991 period. Improvements in living conditions have encouraged thousands of people to move, mostly to Policoro where the population increased from a few hundred in 1861 to 15 000 in 1991. Craco was the only municipality here that registered a loss of population. 3.2
Land Use
The agricultural surface of the Agri Basin is 108 549 ha, equal to 17% of the whole of Basilicata. The 13 267 farms, with an average area of 8.2 ha, are quite dispersed. The dominant crops are cereals, predominantly durum wheat, especially in the internal hilly areas. On the coastal plains, from Montalbano Jonico to Policoro, land utilization is more intensive, with irrigated fruits and vegetables. The higher part of the basin, around Moliterno, Marsico Nuovo and Viggiano, is predominantly forested with oak and beech; wild or cultivated fodder crops for livestock are widespread.
Historical and Contemporary Land Use The Upper Val d’Agri is an area particularly rich in water resources, but there are limits to local availability. A large part of the supply is reserved for the population of Puglia, to irrigate the Metapontina area and for hydro-electric stations. Since ancient times, the abundance of water resources has allowed extremely fragmented small family farms to develop on alluvial zones near residential areas (e.g. Marsico Nuovo, Paterno, Sarconi). A widespread network of irrigation canals, mostly managed by the farmers without written agreements, favour and support this kind of small-scale but intensive agriculture, which has always centred around vegetables, herbs, single rows of vines and other scattered plants. It is a
General Description of the Agri Basin
327
type of subsistence farming. Typical cereal crops and fodder largely occupy the remaining soils of the valley. The expansion of irrigation that started in the 1950s has brought about a significant change in farming systems. Originally cattle were kept to produce milk, but then the introduction of irrigation allowed the development of horticultural crops and fruit growing such as apples. Traditional extensive agricultural systems concentrating on wheat cultivation and sheep raising persist in hilly areas, e.g. around Moliterno. Certain products such as “canestrato” (mixed goat and sheep milk) have been successful on the market. A significant area is still occupied by vines (Viggiano), olives (Montemurro), chestnuts and oak coppice. At higher altitudes, pasture, mainly for sheep, and beechwoods or poplars for forestry are the dominant types of land use. In the past, agriculture has been maintained at very high altitudes with the cultivation of rye and potatoes. Figure 23.5 shows how land use changed between 1982 and 1990 in the three parts of the Agri Basin. In the Middle Val d’Agri the agricultural landscape is interrupted by the badland soils, especially in the Sauro sub-basin. Macchia Mediterranea (natural herbs and shrubs) dominate these clay soils though in the past forests occupied a greater area. Today agriculture is based on wheat cultivation (durum wheat) alongside traditional livestock breeding. At higher altitudes, there are pastures for sheep. Olives and figs are important products from some hilly areas, e.g. around Aliano, but this remains small-scale because of the long distances to markets. Vines are ubiquitous. Intensive agriculture and horticulture (citrus fruits, peaches and vegetables) are found only where the Agri River is very wide and the water table is high, near the Sauro torrent and Sant’Arcangelo. In the Lower Val d’Agri, malaria infested the coastal zone for many centuries and it was not until this could be controlled at the beginning of the 1950s that the area could be used for agriculture. Initially the land was used for cereals and winter pasture, but now irrigation and land improvements allow horticulture and fruit growing. Early fruits for foreign markets, such as strawberries, can be produced. Industry has also been attracted to this area of agricultural growth. Figure 23.6 contrasts land use in the three sections of the Agri Basin. Moving from the upper valley towards the coast, less land is used for pasture and more for arable and horticulture. The use of aerial photography has helped to delimit the areas associated with different types of land use. Four classes of land use, dependent on limitations of climate and soils, have been identified (Ferrara et al. 1995; Ferrara and Taberner 1997). Class I land has deep soils and a climate suitable for growing most crops. Class IV land cannot be used for arable crops but may support forestry on stable slopes.
Land Use for the Future The Upper Val d’Agri is currently an area of change, although agriculture remains the fundamental activity, exploiting the constant supply of water from the mountains. One new development is tourism, both in summer and winter, and interest in the region is likely to grow with the development of the Val d’Agri and Lagonegrese National Park. Another change is the expansion into fruit growing. New technology makes it profitable to use the well-irrigated areas to grow small soft fruits such as raspberry, redcurrants and strawberries, besides the traditional vines and olives. There is an international market for these fruits, and also for vegetables such as potatoes and legumes, especially if they can be grown out of season in greenhouses. There have also been advances in systems for rearing livestock and pasture management. Around Stigliano and Aliano, overgrazing when pasture growth is restricted during the summer drought or low temperatures of winter can be a problem, increasing erosion. The fragile badland soils require protection from irreversible damage. For the same reason monoculture with durum wheat is now being restricted, using set-aside regulations imposed by the EC. It is more difficult to produce competitive crops in the Middle Val d’Agri than in the Upper Val d’Agri. Availability of water limits the choice of arable crops, and there are large areas of badlands that cannot be cultivated easily. Therefore a large part of the territory has to be set aside or partially reforested. However, the strategies that are possible on well-irrigated areas remain valid here. Where the water table is high enough fruit can be grown, for example old varieties, such as the peach “Percoca of Sant’Arcangelo”, for which there is a specialist market.
328
Mediterranean Desertification
Area (ha)
Upper Val d’Agri 20000 18000 16000 14000 12000 10000 8000 6000 4000 2000 0 Arable land
Meadows and permanent grassland
Woods
Area (ha)
Middle Val d’Agri 40000 35000 30000 25000 20000 15000 10000 5000 0
1982 1990
Arable land
Meadows and permanent grassland
Woods
Lower Val d’Agri 25000
Area (ha)
20000 15000 10000 5000 0 Arable land
Meadows and permanent grassland
Woods
Figure 23.5 Land-use evolution, 1982–1990
To help stabilize the soils and conserve water, reforestation is widespread throughout the Agri Basin. One particular area is the Fiumara di Corleto which flows into the Sauro River near Guardia Perticara. In the Lower Val d’Agri the favourable climate and abundant water supply allow a wide range of crops to be grown. As in the Middle Val d’Agri, old crops such as “Staccia of Tursi”, an orange variety that ripens late, are being reintroduced. It will also be possible to grow subtropical crops in greenhouses. Now that malaria is no longer a problem in the coastal area, agriculture and horticulture face competition from industry, tourism and urbanization.
General Description of the Agri Basin
329
Upper Val d'Agri 22% 3%
26%
30%
19% Middle Val d'Agri 22%
woods
5% arable land 16%
meadows and permanent grassland
20% permanent tree cultivation other uses
37% Lower Val d'Agri 23% 6% 18%
11% 42%
Figure 23.6
Current land use
4 CONCLUSIONS The history of the Agri Basin has included deforestation to increase the area of agricultural land, unsustainable cultivation practices such as monoculture with durum wheat, the resulting abandonment of land due to soil erosion and low crop yields, and overgrazing by sheep, goats and cattle (Basso et al. 1983). Now only effective management strategies can save the Agri Basin from continuing land degradation. The factors affecting land use in the area today are complex: historical, agronomic and socioeconomic. The needs of the population have to be balanced against conservation of soil and water resources. River flow regulation and irrigation, careful selection of crops and tillage methods, and
330
Mediterranean Desertification
better management of livestock and pasture to avoid overgrazing can address these objectives. Minimum tillage, parallel to the contours in hilly areas, should be encouraged. The most suitable crops here are cereals with an autumn–spring cycle (e.g. wheat, barley, oat and triticale – a crossbreed between spurred rye and wheat), legumes (e.g. horsebean, chickpea and lupin), and forage crops such as sulla, sainfoil and lucerne on improved pastureland. New horticultural techniques with soft fruits, and reintroduced crops such as old varieties of peach, can be developed to take advantage of the international market.
REFERENCES Basso F (ed.) (1995) Difesa del Suolo e Tutela dell’Ambiente. Editrice Pitagora, Bologna. Basso F, Giordano A and Linsalata D (1983) Effetti della pioggia simulata sull’erosione di un terreno declive della Basilicata sottoposta a differenti modalit`a di lavorazioni. Quaderni. 129, Consiglio National Ricerca. Basso F, Bove E, Ferrara A, Pisante M and Basso B (1996a) Progetto MEDALUS III (Mediterranean Desertification and Land Use): Processi di desertificatione nel bacino dell’Agri in Basilicata. Conference on Scientific Research in Basilicata, 29 February–1 March. Basso F, Bellotti A, De Natale F, Ferrara A and Pisante M (1996b) Analisi del rischio di degradazione del suolo in aree agricole della Basilicata: una proposta metodologica. Proceedings XXX Conference of Italian Society of Agronomy, 6–9 June 1996, Perugia. Basso F, Bove E, Ferrara A, Pisante M and Quaranta G (1996c) Land degradation and desertification processes in the Agri Basin: prevention and management methodologies through use of remote sensing, low environment impact techniques and socio-economic issues. Proceedings International Conference on Mediterranean Desertification, 29 October–1 November 1996, Crete, Greece. Basso F, Bove E, Del Prete M and Pisante M (1998) The Agri Basin, Basilicata, Italy. In P Mairota, JB Thornes and N Geeson (eds) Atlas of Mediterranean Environments in Europe: The Desertification Context. John Wiley, Chichester, pp. 144–149. Bove E and Quaranta G (1996) EC Agricultural Policy Impact on Land Use in Southern Italy: The Case of Clay Hill Areas in Basilicata Region. International Conference on Land Degradation, 10–14 June 1996, Adana, Turkey. Cantore V, Iovino F and Pontecorvo G (1987) Aspetti climatici e zone fitoclimatiche della Basilicata. CNR Publication No. 2, Istituto di Ecologia e Idrologia Forestale, Cosenza. Corbetta F, Ubaldi D and Zanotti AL (1991) La vegetazione a Lygeumspartum nei calanchi del Basento, Basilicata. Archivio Botanica Italiana 67, pp. 141–155. Cotecchia V and Del Prete M (1977) Structurally complex formation in Basilicata and their behaviour in relation to landslide phenomena. Paper presented at the International Symposium on the Geotechnics of Structurally Complex Formations, Associazione Geologica, Capri. Del Prete M (1995) Geology. In Basso F (ed.) Difesa del Suolo e Tutela dell’Ambiente. Editrice Pitagora, Bologna, pp. 191–234. Del Prete M and Hutchinson JN (1988) La frana di Senise del 26 luglio 1986 nel quadro morphologico del versante meridionale della collina timpone. Rivista Italiana di Geologia XXII, 1. De Martonne E (1926) Une nouvelle fonction climatologique: l’indice d’aridit´e. La met´eorologie 19. Dimase AC and Galligani U (1979) I suoli del Comune di Montemurro (Potenza). Nota alla Carta Pedologica. CNR Publication No. 62, Centro di Studio per la Genesi, Classificazione e Cartografia del Suolo, Firenze. Ferrara A and Taberner M (1997) A Computer Program for Extracting Spatial Information from Landsat TM Images. MEDALUS III Working Paper No. 74, King’s College, London. Ferrara A, Pisante M, Harrison AR and Taberner M (1995) The use of spatial relationship analysis to study the Agri Basin with remotely sensed images. Medalus II Final Report. King’s College, London. ISTAT (1997) Movimento anagrafico dei comuni, Roma. Mancini F (1980) Ricerche pedologiche in Val d’Agri. Presentazione del lavoro. CNR, Centro Studio della Genesi, Classificazione e Cartografia del Suolo, no. 60, Firenze.
24
The Agri Valley – Sustainable Agriculture in a Dry Environment: Crop Systems and Management
F. BASSO, M. PISANTE AND B. BASSO
University of Basilicata, Potenza, Italy
1 INTRODUCTION Protection of the soil plays an important role in conserving the environment since soil erosion directly contributes to soil degradation and environmental pollution, and can lead to desertification and decreased soil stability. Unless the well-being of the soil is maintained, agronomic activity becomes unproductive. Interest in soil protection has increased in the Basilicata region since many disasters caused by floods have occurred in recent years. There have been many superficial slips, but also mass movements that have devastated entire urban and rural areas bringing about the loss of human lives of all ages with consequent irreparable damage to both the infrastructure and the regional economy. One important factor in this is the complex geolithological structure that makes up the Basilicata region, often characterized by a rough morphology and slope landslide formations mainly made up by varicoloured or scaly clays. In addition, there are the torrential and seasonal regimes of many water courses; the irregular distribution of the rainfall, mainly concentrated during the autumn–winter period; and the increasing rural exodus that has characterized the vast abandonment of mountainous and hilly areas (Basso et al. 1997). Soil protection used to be a farm tradition in many mountainous and hilly environments in the past and were applied at both territory and farm scale, with some success in avoiding soil erosion problems.
2 SUSTAINABLE AGRICULTURE IN THE AGRI VALLEY Various changes in the last few years have deeply affected the Basilicata region. In particular, the continuing exodus of many young farmers, new crop systems and EC regulations have changed the agronomic landscape in the central hilly areas, creating new problems for agriculture. Recently, the situation has become more serious due to the application of the EC regulation 1098/88, better known as “set-aside”. The area of terrain given over to set-aside is about 100 000 ha in Basilicata, equal to 21.1% of the soils destined to be set-aside in southern Italy (Basso et al. 1996). This is a large area, and therefore it is easy to understand the confusion this has brought in terms of management of the crops grown, general land use, environment protection and conservation. Set-aside affects almost all the marginal areas of Basilicata, and in particular those falling within the Agri Basin. A consistent loss of soil fertility has occurred where runoff and mass movement have removed the soil surface layer. To limit the effects of water erosion and intense rainfall events, agronomic techniques have been developed to improve slope stability and reduce the loss of fertile soil. These techniques include soil water management, and the identification of the best cultivation techniques and the most suitable crops. Concerns about the effect of set-aside have led to the formulation of new Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
332
Mediterranean Desertification
management strategies. The concepts of low-input sustainable agriculture and of reduced chemical input (Kirchner et al. 1993) include techniques such as green manuring and burning crop residues, and identifying suitable practices to improve the global fertility of the soil (MacRae and Mehuys 1985). The main aim is to significantly increase the microbial biomass and to increase the unstable pool of C and N (Franzluebbers et al. 1994). This helps to stimulate organic matter turnover and consequently increases the availability of mineral nutrients for crops (Kern and Johnson 1993). It is also helpful to use cultivation techniques, known as conservation tillage, which allow only minimal disturbance of the active biological part of the soil.
3 3.1
SOIL TILLAGE METHODS FOR SUSTAINABLE AGRICULTURE: EXPERIMENTAL RESULTS FROM THE HILLY AREA OF THE AGRI BASIN Introduction
Optimization of soil tillage requires knowledge of the soil type, water relations and the response of the soil to conditions of water stress. The complexity of the problem is evident in the internal hilly areas of Basilicata, where the low amount and uneven distribution of rainfall, and the prevailing clay nature of the soil, favour surface soil erosion processes and great mass movements. Such areas then degrade quickly and are left abandoned (Basso and Postiglione 1994). Increases in tractor fuel costs have made minimum tillage techniques more attractive to farmers. It is possible to substitute conventional tillage (ploughing at 40–50 cm depth + harrowing) with minimum tillage techniques (scarifying, ploughing at 20 cm depth, rotavation, harrowing, etc.) without excessively compromising the yield. Recently, farmers have shown more interest in minimum tillage methods, particularly in association with set-aside practices. The Institute of Agronomy at the University of Napoli carried out a long-term study (1970–1985) of the Agri Basin, and since 1986 the Department of Crop Production at the University of Potenza has also been involved. The objective was to study the effects of different tillage methods on the physical and hydrological properties of the soil, on soil losses and nitrogen losses, on root and biomass growth and on the yield of the two typical crops of the area: horsebean for seed and durum wheat. 3.2
Materials and Methods
The studies, still continuing, have been conducted on the experimental fields at Guardia Perticara (PZ) located in a hilly area at 700 m a.s.l. The site is on a clay-loam soil, with typical vertic characteristics (verticustorthens). Physical and chemical soil characteristics are listed in Table 24.1. Table 24.1 Physical and chemical characteristics of the soil at Guardia Perticara
Sand (0.02 < ∅ < 0.2 mm) (%) Silt (0.002 < ∅ < 0.02 mm) (%) Clay (∅ < 0.002 mm) (%) Carbonate (%) pH Organic matter (%) Total N (%) P2 O5 available (Olsen method) (µg g−1 ) K2 0 exchangeable (Ammonium acetate method) (µg g−1 ) Mean values based on 48 samples.
1976–1980
1981–1984
1984–1988
1991–1993
26.4 23.6 50.0 17.3 8.0 1.6 0.17 22
44.9 20.7 34.4 5.5 7.7 1.4 0.10 56
42.5 22.9 34.4 5.2 7.5 1.3 0.11 58
44.2 23.0 32.8 6.3 7.2 1.3 0.12 58
102
342
345
345
Sustainable Agriculture in a Dry Environment
333
The research studies were carried out during three different periods, each period with different aims and procedures. In the first period (1976–1980), two tillage systems were compared: conventional (ploughing at 40 cm depth + harrowing) and minimum tillage (total weed control + rotavation) on a range of crops (durum wheat, horsebean for seed, Spanish esparcet, meadow, and natural pasture). In order to evaluate soil losses over an area of 1000 m2 , 50 × 20 m plots with an 18% slope and equipped with superficial runoff sampling and measuring equipment (GEIB type) were used. In the second period (1981–1988), four tillage methods (ploughing at 40 cm depth + harrowing; ploughing at 20 cm depth + harrowing; rotavation; direct drilling) were compared on 600 m2 (40 × 15 m) plots with a 14% slope. The plots were hydraulically isolated with rigid plastic down to a depth of 1.5 m, and equipped with superficial and deep runoff measuring equipment and collecting tanks. The following crops were considered: durum wheat cv. “Appulo” in 1981–1984 and a rotation of durum wheat and horsebean for grain cultivar “Chiaro di Torre Lama” in 1984–1988. In the third period (1990–1993), the following tillage methods were compared: ploughing at 40 cm depth + harrowing; scarifying at 50 cm depth + ploughing at 20 cm depth + harrowing; scarifying at 50 cm depth + harrowing and total weed control + rotavation on 600 m2 plots. A rotation of durum wheat and horsebean for seed was adopted. The physical characteristics of the plots (i.e. texture, bulk density, structure stability, humidity, cracking, temperature) have been described by various colleagues and collaborators (Amato et al. 1994; De Franchi et al. 1994). Nitrogen and soil losses were determined by De Falco et al. (1993, 1994), and total soil biomass and C/C02 ratio at two depths by Villani et al. (1991). Root density, measured at the blooming of durum wheat and horsebean for seed using the Newman method, is reported by Ruggiero et al. (1990), and the growth indices of the two crops with respect to tillage methods were calculated by Quaglietta et al. (1994). The crop yield with relative components was evaluated for each period with respect to tillage methods. The meteorological data for the trial period were plotted using SIAP equipment (since 1990 a multifunction Kampus automatic station has been used (Tecno-EL, Rome)). 3.3 Physical Characteristics of the Soil
The effect of soil tillage methods on bulk density has been evaluated at different stages over the period 1982–1993. During 1982–1983 the measured differences due to tillage did not achieve statistical significance, even though bulk density was greater under minimum tillage than under conventional tillage (Figure 24.1(a)). There were no significant differences between the three soil depth layers considered, although bulk density tended to increase with depth. After five years (Figure 24.1(b)) the minimum tillage treatment showed a lower bulk density with respect to other tillage methods in the 10–20 cm layer (1.37 versus 1.52 g cm−3 ). This was probably due to an increase in organic matter. No appreciable differences were found in bulk density between conventional and minimum tillage treatments in the 1992–1993 period (Figure 24.1(c)). 3.4 Chemical Characteristics of the Soil
The content of organic matter in the soil after being subjected to the different tillage methods varied both with time and soil depth. In 1988, the greatest organic matter content was found in the 0–10 cm layer with direct drilling (2.01%) (Figure 24.2(a)). At greater depths differences were less important. However, at 40–50 cm the percentage of organic matter was 1.27% for conventional tillage and 0.80% for direct drilling. In 1992, in the 0–20 cm layer the organic matter content showed no differences between the different tillage methods (Figure 24.2(b)), while at 20–40 cm, rotavation showed the lowest percentage (1.1%). A close relationship exists between soil nitrogen content and soil organic matter, and they are similarly affected by tillage. Differences in the C/CO2 ratio (indicating microbial biomass) associated with the four tillage methods in 1988 were not statistically significant, but the highest value was obtained where there was no tillage. The highest value for total biomass was obtained from the 20–40 cm level, below ploughing to 20 cm.
334
Mediterranean Desertification 1982−83
(a) 1.6 Bulk density (g cm−3)
1.4 1.2 direct drilling rotavation ploughing at 20 cm ploughing at 40 cm
1 0.8 0.6 0.4 0.2 0 10−20
20−30
40−50
Soil depth (cm) 1987−88
(b) Bulk density (g cm−3)
1.6 1.4 1.2 direct drilling rotavation ploughing at 20 cm
1 0.8 0.6
ploughing at 40 cm
0.4 0.2 0 10−20
20−30
40−50
Soil depth (cm) 1992−93
Bulk density (g cm−3)
(c) 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
rotavation ploughing + harrowing
10−20
20−30
40−50
Soil depth (cm)
Figure 24.1 The effect of four tillage methods on bulk density 3.5
Soil Erosion
Between 1976 and 1995 soil erosion where the soil tillage methods were being compared was at reasonable rates, between 1 and 5 t ha−1 year−1 . Surface runoff was related to rainfall events, mostly during the autumn–winter period, and rainfall intensity (Table 24.2). The effect of soil tillage methods on soil erosion was very limited in the 1976–1980 period (Table 24.3): 1.40 t ha−1 for minimum tillage and 1.14 t ha−1 for conventional tillage. The 0.30 t ha−1 difference equal to 19% is a negligible quantity. In the 1981–1984 period, the effects of the different
335
Sustainable Agriculture in a Dry Environment 1987−88
(a)
% Organic matter
2.5 2 direct drilling rotavation ploughing at 20 cm ploughing at 40 cm
1.5 1 0.5 0
0−10
10−20
20−30
40−50
Soil depth (cm) 1992−93
(b) % Organic matter
2.5 2
rotavation ploughing + harrowing
1.5
scarifying + ploughing + harrowing scarifying + harrowing
1 0.5 0 0−20
20−40 Soil depth (cm)
Figure 24.2
The effect of four tillage methods on organic matter content
Table 24.2 Rainfall registered during 1976–1993 in the experimental field
Year
1975/76 1976/77 1977/78 1978/79 1979/80 1981/82 1982/83 1983/84 1984/85 1985/86 1986/87 1987/88 1988/89 1989/90 1990/91 1991/92 1992/93
Total rainfall (mm)
Rainfall autumn–winter
No. of events with runoff
611.8 730.3 652.5 580.7 526.9 467.2 421.0 704.9 1073.2 606.0 512.4 569.4 610.0 429.0 811.0 618.0 774.6
335.8 573.9 372.8 324.3 388.6 368.0 244.0 509.0 713.0 396.0 317.0 407.0 278.0 201.0 317.0 318.0 575.4
10 8 12 11 10 21 23 19 23 16 17 16 18 12 30 24 44
336
Mediterranean Desertification Table 24.3 Soil losses in experimental plots under different tillage systems (1976–1980)
Weeding + rotavation
Ploughing + harrowing
Period
t ha−1
mean t ha−1 year−1
t ha−1
mean t ha−1 year−1
1976–1980
6.28
1.40
5.13
1.14
Table 24.4 Hydrological data registered in experimental plots under different tillage methods on continuous durum wheat (mean, 1981–1984)
Hydrological parameters Runoff (mm) Soil losses (t ha−1 )
Direct drilling
Rotavation
Ploughing at 20 cm
Ploughing at 40 cm
62.66 1.35
54.20 1.23
48.60 1.17
40.10 1.12
Table 24.5 Hydrological data registered in experimental plots under different tillage methods (annual mean) Hydrological parameters
Crop Horsebean for seed (1984–1985, 1986–1987)
Durum wheat (1985–1986, 1987–1988)
Direct Rotavation Ploughing Ploughing Direct Rotavation Ploughing Ploughing drilling at 20 cm at 40 cm drilling at 20 cm at 40 cm Runoff (mm) Soil losses (t ha−1 )
106.90 4.15
85.45 3.27
77.70 3.66
73.85 3.88
67.10 1.73
58.55 1.63
52.65 1.59
45.25 1.58
soil tillage methods on durum wheat (Table 24.4) were modest. In fact, soil losses were 1.35 t ha−1 for direct drilling and 1.12 t ha−1 for conventional tillage, while runoff was 62.7 mm for direct drilling and 48.6 mm for conventional tillage. In the 1984–1988 period, the soil losses ascertained for the soil tillage methods described above on a durum wheat–horsebean rotation (Table 24.5) were clearly greater for horsebean, with values between 4.15 t ha−1 for no tillage and 3.27 t ha−1 for minimum tillage (harrowing only). Instead, soil losses were negligible for durum wheat and the highest value (1.73 t ha−1 ) was reached with no tillage compared with the minimum value (1.58 t ha−1 ) for conventional tillage. Table 24.6 compares tillage systems for horsebean and durum wheat, measured during the 1990–1992 period. Minimum tillage helped to reduce soil losses by a small amount in each case. 3.6
Runoff and Nitrogen Losses
Leguminous crops help to improve the nitrogen status of the soil, and therefore fodder from legume crops may be produced without the use of nitrogen fertilizers (Power et al. 1983). In environments prone to erosion this may have important consequences on the limitation of nitrogen losses due to runoff. The amount of soluble nitrogen lost with surface runoff water is not related to the amount of nitrogen in the whole soil profile, but rather to the amount localized at the surface, especially where this has been applied to the soil as a top dressing (Baker and Laften 1983). Details of experiments to determine the effects of different tillage systems on soluble nitrogen in runoff are given below.
Sustainable Agriculture in a Dry Environment
337
Table 24.6 Hydrological data registered in experimental plots under different tillage methods (mean, 1990–1992)
Hydrological parameters Horsebean Runoff (mm) Soil losses (t ha−1 ) Durum wheat Runoff (mm) Soil losses (t ha−1 )
Rotavation
Scarifying at 50 cm depth
Scarifying at 50 cm, ploughing at 40 cm
Ploughing at 40 cm depth
26.80 0.76
21.60 0.73
23.90 0.93
23.80 0.92
26.30 0.49
22.70 0.50
23.50 0.58
26.80 0.67
Materials and Methods The research was conducted for two years (1991–1993) on faba bean crop (cv. “Chiaro di Torre Lama”) in rotation with durum wheat. The location was a hilly area of Basilicata, at 700 m a.s.l. on a clay-loam soil with an average slope of 14%, and having the chemical and physical characteristics illustrated in Table 24.1. Four tillage methods were compared: ploughing at 40 cm + harrowing (P + H); scarifying at 50 cm + ploughing at 20 cm + harrowing (S + P + H); scarifying at 50 cm + harrowing (S + H); chemical weed control + rotavation (R). Each tillage method was tested on 600 m2 plots, hydraulically isolated, with two replications tilled along the maximum slope. Surface runoff was collected to test the effect of soil tillage on runoff and losses of solutes. Deep runoff was sampled through a drainage tile placed at one end of each plot, perpendicular to the maximum slope at 1 m depth. Mineral nitrogen (N-NH4 and N-NO3 ) were determined with a DR 2000 spectrophotometer on water samples collected after each runoff event.
Results and Discussion Cumulative precipitation reached 446 mm in the period from December 1991 to October 1992, and 629 mm during the period November 1992 to October 1993. The seasonal distribution was more uniform in the first year when 51 rainfall events occurred and 20 of these gave surface runoff. In the second year, rainfall was concentrated in the autumn–winter period and most of it was snow. Of 44 precipitation events only 12 resulted in surface runoff. Runoff events were distributed quite uniformly during the first year following rainfall distribution. Events with runoff values higher than 1 mm occurred at the end of December and in October and corresponded to highintensity rainfall. In the second year, events with a runoff value higher than 1 mm were recorded in December, January and March. Rainfall was very limited during the summer, and no runoff events were recorded. In both years the ploughing and harrowing (P + H) treatment resulted in the greatest loss of mineral nitrogen (N). Scarifying, ploughing and harrowing (S + P + H) also resulted in significant losses. The recorded concentrations are of the same order of magnitude of those reported by other authors (Baker and Laften 1983). The time-course of mineral N concentration in surface runoff showed that in both years values were generally greater during the spring period and were lower in events that occurred after abundant rainfall. Very few events of deep runoff were recorded (only two events in the first year and seven events in the second year), generally occurring at, or soon after, an intense or prolonged precipitation event. Figure 24.3 compares differences between treatments in terms of runoff (surface and deep), and nitrogen loss. There were no simple relationships. Rotavation (R), or ploughing and harrowing (P + H), tended to result in the greater losses, but all losses were small during the period under study.
338
Mediterranean Desertification
(a)
16
Runoff (mm)
14 12 10 8 6 4
surface runoff
2
deep runoff
0 P+H
R
S + P+ H S +H
P+H
1991−92
R
S+P+H S+H
1992−93 P+H R S+P+H S+H
Ploughing + Harrowing Rotavation Scarifying + Ploughing + Harrowing Scarifying + Harrowing
(b)
N-NO3 and N-NH4 (kg ha−1)
0.25 0.2 0.15 0.1 surface runoff 0.05 deep runoff 0 P+H
H
S+P+ H
S+H
1991−92
P+H
H
S+P+ H
S+H
1992−93
Figure 24.3 Total amount of (a) runoff and (b) N losses 3.7
Root Systems
It emerged for the 1982–1988 period that soil tillage modifies root distribution along the profile favouring its increased presence in the tilled layers. The average root density was not significantly influenced by tillage methods, but the root distribution changed down the profile. The root density during the 1982–1983 period, expressed as a percentage, decreased with depth with all treatments, while in 1988 it decreased with rotavation and direct drilling. Instead, with conventional tillage the root density reduction was found in the deep layers, starting at 40 cm. Root distribution during the 1990–1992 period showed a higher density in the surface layers for the minimum tillage methods for both durum wheat and horsebean for seed. 3.8
Crop Growth Analysis and Yield
Figures 24.4 and 24.5 compare the different tillage treatments in terms of the performance of durum wheat and horsebean for seed. The crop growth index and the leaf area index suggest that variation in tillage treatment had very little effect on crop growth. The average grain yield values of the experimental crops for the period 1976–1992 were as follows: 2.3 t ha−1 for durum wheat with conventional tillage, 2.0 t ha−1 with shallow ploughing, 1.8 t ha−1 with minimum tillage and 1.3 t ha−1 for direct drilling, while yields for horsebean were respectively 2.2, 1.8, 1.5 and 1.3 t ha−1 .
339
Sustainable Agriculture in a Dry Environment 1991−92 5
4
4 Leaf area index
Leaf area index
1990−91 5
3 2
3 2
1
1
0 130 134 146 160 172 188 208 232 246 260
0 100
102
110
Days after sowing
122
142
156
1990 −91
184
204
220
204
220
1991−92
60
40
20
0
−20 130 134 146 160 172 188 208 232 246 260
Crop growth index (g m−3 day−1)
Crop growth index (g m−3 day −1)
172
Days after sowing
60
40
20
0
−20 100
102
110
S+H
Figure 24.4
122
142
156
172
184
Days after sowing
Days after sowing Scarifying + Harrowing
S+P+H
Scarifying + Ploughing + Harrowing
P+H
Ploughing + Harrowing
R
Rotavation
The effect of four tillage methods on growth parameters in durum wheat
4 CROP SYSTEMS FOR SUSTAINABLE AGRICULTURE The research aimed to evaluate the agronomic, environmental, energy requirement and economical implications of selected crop systems and to experiment with other agronomic techniques, such as different tillage methods that influence the physical–chemical characteristics of the soil and affect soil fertility in typical marginal environments (Bonciarelli et al. 1986; Toderi and Bonari 1986; Amato et al. 1994; Basso and Postiglioni 1994; De Falco et al. 1994; De Franchi et al. 1994). At a hilly site in the Middle Val d’Agri, evaluation of productive potentiality, water reserves and soil structure were studied, in relation to environmental factors. 4.1 Materials and Methods
The experimental field located in Guardia Perticara in the Province of Potenza is 700 m a.s.l. on calcareous clay. Table 24.1 illustrates the physical and chemical characteristics of the terrain in the 1991–1993 period. The crop systems that were compared consisted of three two-year successions: (a) durum wheat (cv. Appio)–fallow land; (b) durum wheat–chickpea (cv. Sultano); (c) durum wheat–vetch/oat (cv. Pietranera and cv. Sonar). Two energy input tillage methods were included, one high energy (traditional method with ploughing and harrowing once or twice) and the other low energy (rotavation only). An experimental plot scheme was adopted, with three repetitions with tillage methods on the larger plots (2700 m2 ), crop systems on the subplots (900 m2 ) and crops on the elementary plots
340
Mediterranean Desertification 1991−92 5
4
4 Leaf area index
Leaf area index
1990−91 5
3 2 1
3 2 1
0 130 134 146 160 172 188 208 232 246 260
0 120
130
144
186
1990−91 60
212
222
246
1991−92 Crop growth index (g m−3 day −1)
Crop growth index (g m−3 day−1)
198
Days after sowing
Days after sowing
40
20
0
−20 130 134 146 160 172 188 208 232 246 260
60
40
20
0
−20 120
130
144
Days after sowing
186 198 212 Days after sowing
S+H
Scarifying + Harrowing
S+P+H
Scarifying + Ploughing + Harrowing
P+H
Ploughing + Harrowing
R
Rotavation
222
246
Figure 24.5 The effect of four tillage methods on growth parameters in horsebean for seed
(450 m2 ). The research began in autumn 1993 on land previously used for pasture and alfalfa. In October 1993 superficial scarifying was performed in order to make the surface uniform and to eliminate all crop residues.
Soil The chemical and physical analyses of the samples were carried out according to the methods of Page et al. (1982). The tillage methods to be compared were as follows: minimum tillage (MT); rotavation at 20 cm depth and conventional tillage (CT); ploughing at 40 cm depth and harrowing at 20 cm depth.
Microbiological Analysis and Enzymatic Activity The samples were carefully and uniformly air-dried until they reached a humidity that allowed sifting at 2 mm and then they were left for five days at 20 ◦ C, being at 60% of their field water content. Soil respiration was evaluated after the incubation period according to Dumontet and Mathur (1989), as well as carbon (MBC) and nitrogen (MBN) content in the microbial biomass. The β-glucosidase and arylsulphatase, and phosphatic activity (acid and alkaline) were determined according to Page et al. (1982), while FDAase and nitrate reductase were evaluated according to Abdelmagid and Tabatai (1987).
Sustainable Agriculture in a Dry Environment
341
Soil Moisture Soil moisture readings were taken during the biological cycle from five successive soil layers for each trial using a gravimetric method in both experimental years.
Growth Indicators During the first year, dry matter yield was determined for all comparisons during the phenological stages, divided according to the plant parts and carried out by sampling 25 cm of each row. The leaf area index (LAI) was also determined using an automatic Area Meter LI-COOR. 4.2 Climatic Trend
Annual rainfall was 629.4 mm during 1993–1994 (October to September) and 414.2 mm during 1994–1995 (October to May). Scarce rainfall characterized the summer periods, concentrated in few events. There were long cold periods of 0 ◦ C during the winter of 1994, with the lowest temperature reaching −4 ◦ C in February. The winter of 1995 was milder, having a minimum temperature of about −2.9 ◦ C. Summer temperatures exceeded 26 ◦ C from the end of May 1994 and exceeded 24 ◦ C for the same period in 1995. Maximum temperatures of around 30 ◦ C were reached in June 1994. 4.3 Results and Discussion
Microbiological Analysis and Enzyme Activity Results outlined in Table 24.7 suggest that minimum tillage preserves a greater level of microbial biomass than does conventional tillage, especially in autumn. The type of crop rotation, whether vetch/oat–wheat, or fallow–wheat, made no particular difference. The 40 kg ha−1 N added to the wheat crops in November 1994 and the 20 kg ha−1 N added in March 1995, with mineral fertilization, sustain the microbial biomass. Our results are in agreement with those reported by Franzluebbers et al. (1994, 1995), who found an increase of up to 31% in microbial biomass carbon in soils that underwent no tillage compared with the same soils tilled conventionally. It is interesting to note that the specific respiration activity in September was significantly greater under conventional tillage than under minimum tillage. The intense respiration activity relative to the microbial biomass for the conventional treatments in September indicates a stress condition, with maintenance of the microbial population rather than growth, and is often seen as a response to disturbance (Fliebach and Reber 1992).
Soil Moisture Figure 24.6(a) shows that the tillage method used hardly affected soil moisture, the results being very close throughout the two successive years. Figure 24.6(b) demonstrates that soil moisture under fallow land was usually greater than under the crops where transpiration was greater. All the crops behaved similarly. The lower soil moisture value under fallow land in March 1994 reflects a period when there was sparse vegetation cover and evaporation was facilitated. These data confirm that fallow land may improve the conservation of water with respect to sown crops, but the ecological and technical importance of this function is variable in different environments.
Physiological and Yield Results During 1993–1994 there were no statistically significant differences between the two tillage systems for wheat. Weed control did not have any affect on grain production but did affect the weight of 1000 seeds and the hectolitre weight (Table 24.8). There were greater differences between the two years. The results for 1994–1995 can be attributed to both the more normal seasonal weather trend, characterized by more uniform rainfall, and to the
342
Table 24.7 Microbial biomass content and enzymatic activity of the soils in relation to different agronomic practices
Plota
Date
C microbial biomass (mg C per kg dry matter)
N microbial biomass (mg N per kg dry matter)
Acid phosphate
Alkaline phosphate (µmoli)
β glucosidase (PNPg−1 )
Arylsulphatase d.m. h−1
MT–VO/W
04/94 09/94 04/94 09/94 04/94 09/94 04/94 09/94
425d 521f 340c 284a,b 325b,c 261a 448d,e 476e
77b,c 83b,c 57a 51a 57a 58a 75b 86c
23.5e (±1.7) 20.9d (±1.0) – 17.2c (±0.8) 8.9a (±0.2) 8.8a (±0.5) 11.9b (±0.4) 11.5b (±1.3)
16.0g 14.0e 7.1a 7.8b 10.0c 10.9c 12.2d 14.8f
2.5c 3.5e 2.1a,b 1.9a 2.1a,b 2.2b 2.8c,d 2.8c,d
1.9d 2.5f 1.7c 1.4a 1.7c 1.5b 2.1e 1.9d
CT–VO/W CT–F/W MT–F/W a
(±35) (±34) (±37) (±12) (±17) (±22) (±32) (±14)
(±2) (±11) (±5) (±5) (±4) (±5) (±4) (±8)
(±0.3) (±0.4) (±0.2) (±0.2) (±0.2) (±0.3) (±0.5) (±1.0)
MT, minimum tillage; CT, conventional tillage; VO, vetch–oat; W, wheat; F, fallow. The values followed by the same letter are not statistically different (p < 0.05; n = 3) using the Duncan Test.
(±0.2) (±0.1) (±0.1) (±0.1) (±0.1) (±0.1) (±0.1) (±0.1)
(±0.1) (±0.2) (±0.1) (±0.1) (±0.1) (±0.1) (±0.1) (±0.1)
0−60 cm layer
(a)
Soil moisture %
28 23 conventional minimum tillage
18 13 8 Sample dates 0−60 cm layer
(b)
Soil moisture %
27 22 fallow durum wheat vetch−oat chickpea
17 12 7 Sample dates
Figure 24.6
Soil moisture in the 0–60 cm layer for (a) two tillage methods and (b) four crops 343
344
Mediterranean Desertification
Table 24.8 Maximum leaf area index, total dry matter and yield for wheat, chickpea and vetch–oat in 1993–1994
Tillage
Maximum LAI
Total dry matter (g m−2 )
Yield (t ha−1 )
1000 seeds (g)
Hectolitre weight (kg)
Dry matter (t ha−1 )
1.7 1.8
332 329
2.9 2.9
36 38
78 79
– –
1.6 1.8
249 166
1.9 1.7
335 330
– –
– –
3.5 3.2 n.s.
671 545 n.s.
– – n.s.
– – n.s.
– – n.s.
7.0 7.7 n.s.
Wheat CT MT Chickpea CT MT Vetch–oat CT MT LDS P = 0.05
CT, conventional tillage; MT, minimum tillage; n.s., not significant. Table 24.9 Yield of wheat, chickpea and vetch–oat for 1994–1995
Tillage Wheat Conventional Minimum LDS P = 0.05 Previous crop Fallow Chickpea Vetch–oat Chickpea Conventional Minimum LDS P = 0.05
Vetch–oat Conventional Minimum LDS P = 0.05
Grain yield (t ha−1 )
Harvest index (%)
Weight 1000 seeds (g)
Hectolitre weight (kg)
5.0 4.7 n.s.
42.4 42.0 n.s.
40 41 n.s.
78 78 n.s.
4.6b 5.2a 4.8ab
40.5a 42.7a 43.5a
1.84 1.78 n.s.
37.7 39.3 n.s.
333 329 n.s.
– – n.s.
50.3 58.3 n.s.
– – –
– – –
– – –
40a 40a 40a
78a 78a 78a
n.s., not significant. The values followed by the same letter are not statistically different (p < 0.05; n = 3) using the Duncan Test. earlier sowing. The lack of weed control had no influence on grain yield but did on the number of ears per square metre, hectolitre and 1000 seed weight (Table 24.9).
5
CONCLUSIONS
The purpose of this research was to be able to make some recommendations to farmers in the Agri Basin about tillage methods. So far we have found that the vertic character of the soil is an important factor, and that the soil structure and other characteristics of the soil are only slightly
Sustainable Agriculture in a Dry Environment
345
affected by different tillage methods. Microbial biomass was greater under minimum tillage than under conventional tillage. The higher organic matter content where there was little or no tillage seemed to have a favourable effect on the development of the microbial flora, and on some physical characteristics of the soil. The influence of tillage methods on soil erosion losses was shown to be modest even though surface runoff was favoured by the absence of tillage. The average soil content collected in suspension was shown to be minor with respect to minimum and conventional tillage. Nitrogen losses were also modest with no differences between soil tillage systems for both durum wheat and horsebean for seed. As for plant development, the influence of soil tillage methods was studied for both top and root growth, though the differences found were not always statistically significant. With conventional tillage, crop development, yield and weed control were higher than in the absence of tillage. However, considering that the differences in production between traditional and minimum tillage were modest, and also considering economic and energy cost reductions, the adoption of minimum tillage systems compared to conventional tillage may contribute to the development of more sustainable agriculture. Analytical results of organic C and total N have not shown any clear variations after two years of study. After an experimental period of 10 years it might be possible to illustrate significant increases in organic C in soils subjected to minimum tillage compared with conventional tillage. Higher values of runoff and nitrogen losses were recorded in the first year, affected by rainfall distribution. In the second year, most precipitation consisted of snow, so that conditions were more favourable for infiltration than for surface runoff, and deep runoff and soil losses were relatively higher. Nitrogen losses (N-NO3 + N-NH4 ) were limited because of the small amount of runoff. The highest nitrogen losses were found to be associated with just rotavation, and the treatment of scarifying plus ploughing plus harrowing proved to be the most effective in limiting runoff and consequent losses. Growth analysis results indicated higher values in conventional tillage for chickpea while for wheat and vetch/oat no significant differences were found. Vetch/oat production was different during the two years, showing a higher LAI in the second year.
REFERENCES Abdelmagid HM and Tabatai MA (1987) Nitrate reductase activity of soil. Soil Biology Biochemistry 19, 421–427. Amato M, Pardo A and Landi G (1994) Effetti delle modalit`a di lavorazione di un terreno declive sull’accrescimento radicale del favino (Vicia faba minor beck) e del frumento (Triticum durum Desf.) in rotazione. Rivista di Agronomia XXVIII(4), 407–412. Baker JL and Laften JM (1983) Water quality consequences of conservation tillage. Journal of Soil and Water Conservation 38(3), 186–193. Basso F and Postiglione L (1994) Aspetti agronomici della conservazione dei terreni in pendio: sistemazioni e lavorazioni. Rivista Agronomia XXVIII(4), 273–296. Basso F, De Franchi AS, De Falco E, Amato M and Landi G (1996) Valutazione di sistemi colturali a diverso impatto energetico inseriti in una realt`a aziendale della collina interna lucana che pratica il set-aside. Agricoltura e Ricerca 164, 381–392. Basso F, Pisante M, De Franchi AS and Basso B (1997) La difesa dai fenomeni erosivi nell’Italia meridionale. Informatore Agrario 43, 1–13. Bonciarelli F, Archetti R, Farina G and Battistelli A (1986) Effetto di nuovi sistemi di lavorazione su alcune propriet`a chimiche e meccaniche del terreno. Rivista di Agronomia 20(2–3), 172–177. De Falco E, Amato M and Basso B (1993) Ricerca su accrescimento e produzione del cece (Cicer arietinum L.) in ambiente collinare della Basilicata. Rivista di Agronomia 1, 23–33. De Falco E, De Franchi AS, Basso F and Postiglione L (1994) Effetti delle modalit`a di lavorazione di un terreno declive a rotazione favino-frumento sull’erosione e sulla qualit`a dei deflussi. Rivista di Agronomia XXVIII(4), 348–355. De Franchi AS, Amato M, Pisante M and Graziano FS (1994) Osservazioni sulla variazione di alcune caratteristiche fisiche di un terreno declive (struttura e umidit`a) in relazione a diverse modalit`a e colture. Rivista di Agronomia XXVIII(4), 427–432.
346
Mediterranean Desertification
Dumontet S and Mathur SP (1989) Evaluation of respiration based methods for measuring microbial biomass in metal-contaminated acidic mineral and organic soils. Soil Biology Biochemistry 21, 431–436. Dumontet S, Coppola E, Perucci P and Bufo SA (1996) Modificazioni indotte da differenti pratiche agronomiche su alcuni parametri biochimici e microbiologici del suolo. Agricoltura e Ricerca 166, 165–166. Fliebach A and Reber HH (1992) Effect of long-term sewage sludge application on soil microbial parameters. In JE Hall, DR Sauerbek and P L’Hermite (eds) Effects of Organic Contaminants in Sewage Sludge on Soil Fertility, Plants and Animals. Office for Official Publications of the European Communities, Luxembourg. Franzluebbers AJ, Hons FM and Zuberern DA (1994) Long-term changes in soil carbon and nitrogen pools in wheat management systems. Soil Science Society of America Journal 58, 1639–1645. Franzluebbers AJ, Hons FM and Zuberern DA (1995) Tillage and crop effects on seasonal soil carbon and nitrogen dynamics. Soil Science Society of America Journal 59, 1618–1624. Kern JS and Johnson MG (1993) Conservation tillage impacts on national soil and atmospheric carbon levels. Soil Science Society of America Journal 57, 200–210. Kirchner MJ, Wollum IIAG and King LD (1993) Soil microbial populations and activities in reduced chemical input agroecosystems. Soil Science Society of America Journal 57, 1007–1012. MacRae RJ and Mehuys GQ (1985) The effect of green manuring on physical properties of temperate areas. Advances in Soil Science 3, 71–94. Page AL, Miller RH and Keeney DR (1982) Methods of Soil Analysis. American Society of Agronomy, Madison, Wisconsin. Power JF, Follett RF and Carlson GE (1983) Legumes in conservation tillage systems: a research perspective. Journal of Soil and Water Conservation 38(3), 217–218. Quaglietta Chiaranda F, De Falco E and Graziano FS (1994) Effetto delle modalit`a di lavorazione di terreno declive sull’accrescimento e sulla produzione di favino (Vicia faba minor Beck) e di frumento (Triticum durum Desf.) in rotazione. Rivista di Agronomia XXVIII(4), 438–447. Ruggiero C, Basso F and Amato M (1990) Effects of different soil tillage methods on root density of wheat (Triticum durum desf.) in a hilly area of southern Italy. NATO Advanced Research Workshop on Mechanism and Related Processes in Structured Agricultural Soils, 13–16 September, University of Minnesota, St Paul, Minneapolis. Toderi G and Bonari E (1986) Lavorazioni del terreno: aspetti agronomici. Il. Lavorazioni e pianta coltivata. Rivista di Agronomia 20(2–3), 106–133. Villani F, Capilongo V and Amato M (1991) Effect of tillage on the microbial biomass of a clay soil. Agricoltura Mediterranea 121, 130–135.
25
Soil Erosion and Land Degradation
F. BASSO, M. PISANTE AND B. BASSO
University of Basilicata, Potenza, Italy
1 INTRODUCTION Poor farm management is an important factor for soil degradation and erosion, as are the uncontrolled destruction of forests and pollution of water resources. It was estimated in the 1980s that by the year 2000 the global loss of arable soil due to degradation would reach 10 million hectares per year (Dudal 1982). Natural climate cycles and changes, such as the drought across the Sahel in the 1970s, are also important factors, sometimes causing irreversible damage to the agricultural potential of the soil. Good farm management should include the selection of the most suitable tillage techniques for the terrain. It is essential to preserve an equilibrium that minimizes soil erosion. Knowledge of soil formation processes combined with studies of the soil–plant–atmosphere relationships will be necessary in the future to maintain healthy agricultural systems (Aru 1991). Unfortunately demographic forces and pressure to exploit natural resources are often at odds with the sustainable course.
2 SOIL EROSION 2.1 Soil Erosion in the Mediterranean Environment
The Mediterranean environment provides many examples of instances where the natural equilibrium in the environment has been upset, leading to degradation and erosion. Important factors include climate (especially volume and incidence patterns of rainfall), pedological conditions and surface features such as degree of slope, slope aspect and length, and stoniness (Poesen and Bryan 1990). The geology and the lithological constituents of the soil play a large part in determining soil stability. Soils developed on crystalline metamorphic substrates are particularly erodible, as are acid volcanic rocks and compact carbonate rocks where soil development is too slow to compensate if the shallow overlying soil is removed. Soils developed on tufa or basic volcanic rocks are less vulnerable (Aru 1991). Within the soil, the mechanical composition, the proportion of clay particles, the form of individual clay minerals and the soil organic matter content all affect susceptibility to erosion (Toderi 1991). Soil erosion can cause a loss of soil fertility. The surface soil horizons are generally those most rich in organic matter, elements required for plant nutrition, and micro-organisms beneficial to plant growth. The eroded soil may concentrate at the bottom of a slope, and Verity and Anderson (1990) noted that horsebean seed yield was greater from the base of slopes. Alternatively, eroded soil may be carried and dispersed further afield, by water or wind. The soil depth may be reduced by erosion to such an extent that the previous vegetation cover can no longer be supported (Landi 1984). Sometimes removal of the upper horizon reveals soil layers rich in toxic elements, or sands that drain too quickly to supply the needs of the vegetation (Schertz 1983). There is evidence that cultivation can increase soil erosion even on shallow slopes (Smith and Stamey 1965). Reported differences between cultivation and non-cultivation may be as much Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
348
Mediterranean Desertification
as 17 t ha−1 (Ruhe 1969). In the Mediterranean region, intensive cropping or intensive livestock farming can cause extensive, irreversible problems. In Italy, soil erosion is mainly caused by water, and reaches 165 Mg ha−1 year−1 on the marine Pliocene soils of the Tacina Basin in Calabria (Raglione et al. 1980). Recently, other Italian research has measured soil erosion on clay terrain in Toscana at 0.35–59.0 t ha−1 year−1 (Panicucci 1972; Chisci and Tellini 1973). In scaly clay soils on the central area of the southern Apennines the values ranged from 3.0 to 9.0 t ha−1 year−1 (Boschi and Chisci 1978), and from 1.0 to 10.4 t ha−1 year−1 on Quaternary clay soils on the hilly Apennine chain of Basilicata (Postiglione et al. 1983). 2.2
Principles of Reducing Soil Erosion on Agricultural Soils
The principles that help soil protection and therefore soil preservation are strictly linked to those improving water management and the hydrological budget. Therefore the location of areas suitable for agriculture, soil tillage methods, crop selection and crop management techniques are very important. Research is being carried out to resolve problems arising from overcultivation and from increasingly unpredictable hydrological conditions. Methods to reduce or to eliminate erosion can be described according to their influence. These include the following methods: 1.
organizing superficial or underground drainage systems to reduce the amount of water, especially in soils upstream from basins; 2. favouring soil permeability, increasing the water infiltration capacity in order to impede or reduce superficial runoff; 3. favouring a suitable vegetation cover reducing surface wind speed and superficial water runoff, in order to contain soil particle detachment and soil particle transportation. A good vegetation cover is fundamental in preventing the detachment of soil particles that are successively transported away by physical agents. The vegetation is a source of soil organic matter which ensures a greater water infiltration capacity, and the presence of stable colloids avoids the detachment of fine earth particles during or after rainfall. Organic matter can be directly added to limestone or loose sandy soils to favour stable soil aggregates resistant to detachment. Solute transport in soil water removes valuable plant nutrients and occurs when the infiltration capacity of the soil is less than water inflow from rainfall or from upslope. Water or wind needs to have a certain velocity in order to transport soil material. This velocity, defined as the transport limit velocity, has to overcome the natural resistance of the soil particles to transport them. This depends on soil particle size and weight, and the prevalence of soil aggregates, as well as the degree of vegetation cover at the surface, and the slope angle. 2.3
Soil Conservation in the Agri Basin: Results of a 25-year Research Study
Introduction Soil conservation is a problem of major concern in Italy, because of the unfavourable geological features of its steep hilly and mountainous landscape, which makes up 78% of the Agri Basin peninsula. The climate is Mediterranean, characterized by dry, hot periods and by particularly intense rainfall events. Human activity in the past has allowed widespread deforestation as a consequence of the necessity to enlarge the area of arable land. Subsequently much of the cleared land situated on slopes has become difficult to farm and has been abandoned (Barbieri and Basso 1973). Fires, whether natural or lit by shepherds to encourage a temporary supply of nutritious forage for their herds, cause a lot of damage and conditions favourable to erosion are increased. The dramatic consequences of land degradation are made manifest by landslides and intensive sheet erosion, which have hit this country repeatedly during the past 40 years (Basso and Linsalata 1983; Postiglione and Marzi 1983; Basso et al. 1986; Postiglione 1988, 1993). These are the reasons why there must be action to combat soil erosion. Among the most important agronomic methods of soil conservation are the
Soil Erosion and Land Degradation
349
control of water reaching the soil, rationalized land use and the application of other cultivation techniques useful to limit the process of erosion. It is well known that ill-chosen farming techniques or unsuitably intensive land use favour the degeneration of the soil and consequently harm the environment (Basso and Postiglione 1994). Therefore it is of the utmost importance to examine the different possibilities for land use, aiming for soil conservation. Since 1970 the CNR (National Research Centre) has financed research programmes on the problems of soil conservation in the Agri Basin in Basilicata, a region that is particularly prone to erosion phenomena and hydrogeological instability. These research programmes were set up and carried out by the Institute of Agronomy and Hydraulics, University of Naples–Portici, the Institute of Agronomy and the University of Bari, and since 1986, by the Department of Crop Production, University of Potenza.
Materials and Methods The initial experimental programme was planned to compare field tillage systems that follow contour lines with field tillage systems that follow the direction of slopes and at the same time observe the influence of both conventional and minimum tillage systems on different crops (durum wheat, horsebean for seed, Spanish esparcet, mixed meadow and natural pasture) during the study period (1970–1980). The aim was to evaluate the degree of soil erosion from experimental plots of 1000 m−2 (50 × 20) with an 18% average slope and from Wischmeier plots of 60 m−2 (20 × 3) with a 12% average slope (Wischmeier and Smith 1965). The experimental field was equipped with a weather station, surface runoff collectors and automatic recorders (Cavazza et al. 1983). The study is still in progress on an experimental field located in the Municipality of Guardia Perticara (Province of Potenza, Basilicata Region). The area is in the Sauro sub-basin, part of the Agri Basin, between 554 and 700 m a.s.l. on 30 ha of land having a south-west aspect. The soil is classified by the USDA (1975) as a “vertic ustorthens” type. The vertic nature of the soil is clearly seen during the summer period. The soil is made up of limestone particles and clay with a high calcareous content giving an alkaline reaction and a high infiltration capacity of 12 cm h−1 (Marzi et al. 1983). Research was conducted on 16 experimental plots as shown in Figure 25.1. During the second study period (1980–1995), each plot was 15 m wide and 40 m long, giving a 600 m2 surface area. The mean slope was 14%. Eight of the plots were hydrologically isolated in order to measure runoff and erosion. This isolation was carried out using fibreglass-reinforced plastic panels in the soil up to a 1.30 m depth around each plot. The plots were protected from the runoff and erosion that could have occurred on the slope above them by a diversion canal. Each plot had a zinc canal placed parallel to the slope to gather superficial runoff. The canal was linked to the plot by a rubber sheet thus avoiding the inclusion of water from outside. The superficial runoff was brought by 0.14 m diameter PVC tubes to a measurement point. The infiltrated runoff was collected by PVC drains located perpendicular to the line of maximum slope, at a depth of 1 m, at the lower end of each plot. The water collected was then brought to an isolated 50-litre collection tank for measurement. Tables 25.1 and 25.2 summarize the physical and chemical characteristics of the soils in the experimental fields. Superficial and infiltrated runoff measurements for each plot were taken after every rainfall event by directly measuring water height in the collection tanks with a metric pole. Soil erosion was evaluated from a 1-litre water sample taken after mixing the water in the collection tanks. The dry weight of the sample was determined after oven drying the sample at 105 ◦ C until a constant weight was achieved. The dry weight value per sample volume was multiplied by the total water quantity collected in the tank to give the total mass of eroded soil present in the infiltrated runoff. This value, together with the material transported with the superficial runoff collected by the rubber sheet and canals, gave the total mass of soil eroded after each rainfall event. During the 1980–1988 period the following tillage methods were tested: 1. ploughing at 40 cm depth + harrowing (P4 + H) 2. ploughing at 20 cm depth + harrowing (P2 + H)
350
Mediterranean Desertification
Plots Contour Lines (m)
N
735
2
1
3
4
3
3
2
4
1
1
2
2
4
Protection Canals
1
3
4
Collection Canals
730
72
5 5 71 Collection Tanks
720 Measuring Equipment
71
0
Figure 25.1 Layout of the experimental field system Table 25.1 Physical and chemical characteristics of the soil in the first experimental field (in % unless otherwise stated) Field
Coarse fragments (>2 mm)
Size class and particle diameter
Pebble Gravel Little Coarse Fine gravel sand sand Upper part 0–25 cm 25–50 cm Middle part 0–25 cm 25–50 cm Lower part 0–25 cm 25–50 cm
Silt
Carbonate pH Organic Total Total Total as CaCO3 matter N P2 O5 K2 O (ppm) (ppm)
Clay
3.7 2.3
1.9 1.2
3.0 2.2
6.84 4.39
13.21 40.25 39.60 13.46 40.25 41.90
16.3 20.6
7.8 8.0
1.81 1.31
0.18 0.12
0.22 0.21
1.08 1.02
3.7 5.3
2.4 2.4
3.6 3.6
7.71 6.67
18.69 23.55 50.05 16.13 27.95 49.25
17.3 22.0
8.0 8.1
1.61 1.26
0.17 0.12
0.22 0.23
1.02 1.40
3.9 11.2
1.9 1.8
1.9 3.0
4.72 5.37
11.68 34.90 48.70 12.78 30.80 51.05
11.5 12.5
8.0 7.9
1.53 1.31
0.13 0.12
0.16 0.17
1.44 1.44
Mean values based on 100 soil samples.
Table 25.2 Physical and chemical characteristics of the soil in the second experimental field
Sand Silt Clay Carbonate pH Organic matter Total N Available P2 O5 a Exchangeable K2 Ob (ppm) (ppm) (%) (%) (%) (%) (%) (%) 44.9 20.7 34.4 a b
5.5
Olsen method. Ammonium acetate method.
7.7
1.4
0.1
56
342
351
Soil Erosion and Land Degradation
3. harrowing (H) 4. no tillage (NT) In the period 1990–1995 different tillage methods were used: 1. harrowing (H) 2. ploughing at 40 cm depth + harrowing (P + H) 3. scarifying at 50 cm depth + ploughing at 20 cm depth + harrowing (S + P + H) 4. scarifying at 50 cm depth + harrowing (S + H) A random distribution method was used to distribute combinations of tillage method and crop to the plots in the experimental field. The leaf area index (LAI) of two crops (chickpea and durum wheat) cultivated in a biennial rotation was determined during the 1993–1995 period in order to evaluate the relationship between vegetation cover and soil erosion. LAI was determined by collecting plant leaves of 0.10 m2 with 3 replications every 15 days from the beginning of germination (second half of March) to harvest, and then measuring leaf area with an Area Meter LI COR mod. LI-3100. The climatic data were collected using a multifunction Kampus station equipped with sensors to measure rainfall, wind direction and velocity, soil and air temperature, humidity and radiation. The thermopluviometric trend of the experimental plots was determined by looking at the results over 25 years, starting in 1970. Figure 25.2 is a climate diagram of the experimental field according to the method of Walter and Leigh (1960). The diagram is drawn with a temperature scale equal to twice the rainfall scale, to indicate the extent of the period of insufficient water availability for plant growth. On the diagram this is when the rainfall curve is below the temperature curve, between May and September. The climate diagram reflects a typical Mediterranean climate with relatively low temperatures and abundant rainfall during the autumn–winter period and with a dry summer with scarce rainfall and high temperatures. Most rainfall fell during spring (mean March value 68 mm) and winter (mean December value 73 mm). Rainfall was much lower during the summer period, particularly during July, the hottest month (mean 21 ◦ C), with 28 mm of rainfall on average. Therefore, the best growth conditions are limited to two periods, from the last days of winter to the end of spring, and autumn whenever temperatures are high enough for plant growth and there is sufficient water. 90
45 671.8 mm
60
30
45 15
30
13.7 °C
Precipitation (mm)
Temperature (°C)
75
15 0
J
F
M
A
M
J J Month
A
S
O
N
D
0
Figure 25.2 Walter and Leigh’s climate diagram for the experimental area, Guardia Perticara, showing means for 1970–1997
352
Mediterranean Desertification Table 25.3 Rainfall registered during 1970–1995 in the experimental fields
Years
1970/71 1971/72 1972/73 1973/74 1974/75 1975/76 1976/77 1977/78 1978/79 1979/80 1980/81 1981/82 1982/83 1983/84 1984/85 1985/86 1986/87 1987/88 1988/89 1989/90 1990/91 1991/92 1992/93 1993/94 1994/95
Rainfall (mm)
No. of events with runoff
Total
Autumn–Winter
Superficial
Deep
764.2 773.4 839.2 653.0 545.4 611.8 730.3 652.5 580.7 526.9 469.3 467.2 421.0 704.9 1073.2 606.0 512.4 569.4 610.0 429.0 811.0 618.0 774.6 737.9 661.2
446.0 685.9 686.6 354.1 418.6 335.8 573.9 372.8 324.3 388.6 368.0 367.7 244.2 509.0 713.2 396.0 317.0 407.0 278.0 201.0 622.0 335.0 575.4 527.0 342.0
18 22 19 14 7 10 8 12 11 10 21 23 19 23 16 17 16 18 12 30 27 27 12 26 24
– – – – – – – – – – – – – – – – – – – – 15 8 7 13 12
Table 25.3 shows that rainfall varied between 421 and 1073 mm year−1 during the period 1971–1995. The mean value was 650 mm during the whole 25 years, which is quite different from the mean value of 820 mm recorded during the previous 50 years (1920–1970). A particular change over the period 1971–1995 was that precipitation was concentrated mainly during the autumn–winter period, which during 1976–1977 accounted for 78.58% of the total annual precipitation. The meteorological trend in the two-year (1993–1995) experimental period reflected the trend for the full 25 years of observation. In particular, rainfall was 737.9 mm in the first year and 661.2 mm in the second; rainfall distribution during the seasons was similar to the total period except for March 1994 which was characterized by no rainfall and August 1995 when rainfall was much greater than the average, with 97.6 mm against an average 29.5 mm over 25 years. The lowest temperature of −3 ◦ C was recorded in March 1994 and January 1995, while the hottest temperatures were recorded in July, reaching 35 ◦ C during the first year and 32 ◦ C in the second.
Results and Discussion The effect of direction of tillage on soil erosion The results (Table 25.4) show that soil water flow control was more effective in the layout along contour lines than in the layout according to slope. Soil loss was related to number of incidences and intensity of the rainfall during the year. During 1971–1980, the average volume of soil loss for the layout following the contour lines was 1.25 t ha−1 while it was 1.54 t ha−1 for the same period in the layout running downslope. This
353
Soil Erosion and Land Degradation Table 25.4 Influence of surface layouts on soil losses (1971–1980)
Layouts
1971–1973
Following contours Following slope
1974–1976
1976–1980
1971–1980
Total (t ha−1 )
Mean (t ha−1 year−1 )
Total (t ha−1 )
Mean (t ha−1 year−1 )
Total (t ha−1 )
Mean (t ha−1 year−1 )
Total (t ha−1 )
Mean (t ha−1 year−1 )
4.90 5.79
1.63 1.93
3.36 4.15
1.12 1.38
4.28 5.41
1.07 1.35
12.54 15.35
1.25 1.54
Table 25.5 Influence of different crops on soil losses (1971–1980, on 1000 m2 plots)
Crops
1971–1973
1976–1980
1971–1980
Total (t ha−1 )
Mean (t ha−1 year−1 )
Total (t ha−1 )
Mean (t ha−1 year−1 )
Total (t ha−1 )
Mean (t ha−1 year−1 )
Total (t ha−1 )
Mean (t ha−1 year−1 )
6.87 6.18 5.49 4.65
2.29 2.06 1.83 1.55
7.01 5.43 3.85 2.74
2.34 1.81 1.28 0.91
6.42 5.00 – –
1.43 1.11 – –
20.30 16.61 9.34a 7.39a
2.03 1.66 1.56 1.23
3.46
1.15
1.26
0.42
1.40
0.31
6.12
0.61
Horsebean Durum wheat Sweet vetch Mixed hay field (with alfalfa) Natural pasture a
1974–1976
Period 1971–1976.
indicates that tillage following the contours reduced soil erosion by 23.2% compared to tillage following the direction of slope. The effect of different crop systems on soil erosion The influence of different crops on soil loss was also considerable (Table 25.5). Natural pasture always showed the lowest soil loss values, the mean value for the period 1971–1980 being 0.6 t ha−1 year−1 . Horsebean for seed was the crop associated with the highest soil loss among the crops used in the experiment. Although there were large differences between soil losses for individual study periods, it is clear that there is a series of increasing erosion risk as follows:
natural pasture < mixed meadow < Spanish esparcet < durum wheat < horsebean (Barbieri and Basso 1973; Basso and Linsalata 1983; Postiglione et al. 1983). Similar results were found for the 60 m2 Wischmeier plots for the same crops. Here the slope angle was lower, and so soil loss was also lower. The mean values for soil loss recorded during the 10 years of observation ranged from 0.18 t ha−1 year−1 with natural pasture, to 1.15 t ha−1 year−1 with horsebean for seed (Marzi et al. 1983). The effect of different tillage systems on soil erosion With regard to the influence of different tillage systems on the amount of runoff and soil loss, it was found that the mean soil loss after conventional tillage (ploughing followed by harrowing) was 1.14 t ha−1 year−1 during the period 1976–1980. After minimum tillage (harrowing and chemical weed control) the mean soil loss amounted to 1.40 t ha−1 year−1 . This is a 19% difference in soil loss (Postiglione et al. 1990). During the following 10-year observation period, particular attention was paid to the influence of different tillage systems and crops. From the comparison of four different tillage systems applied on a layout according to slope, and cultivated with durum wheat, it emerged that the tillage systems influenced both volume of surface runoff and the amount of soil loss. After ploughing at 40 cm
354
Mediterranean Desertification
depth, the turbidity of runoff water was higher, whereas the amount of runoff was smaller. In contrast, turbidity was much lower from the unploughed soil, but the amount of runoff was higher. Consequently there was a higher soil loss for the non-tillage system (1.35 t ha−1 year−1 was the mean value for the period 1981–1984) than after ploughing at 40 cm depth (1.12 t ha−1 year−1 ). Decreasing intermediate values were found after roto-tillage and ploughing at 20 cm depth (Basso et al. 1986). These results were subsequently confirmed (Tables 25.6 and 25.7) by the trials carried out during the following four-year period, both for durum wheat and horsebean. For the latter crop, as in the previous trial, much greater soil loss was measured. Furthermore, ploughing was found to have a positive effect on weed control, helping to increase horsebean yield, due to weaker competition (Basso et al. 1987). During the years 1990–1993, tillage systems that were slightly different from those used previously were compared (Table 25.8). Roto-tillage alone or scarifying at 50 cm depth plus harrowing showed smaller amounts of soil loss compared to scarifying plus ploughing at 20 cm depth with Table 25.6 Hydrological data measured from experimental plots under different tillage methods under continuous durum wheat (average 1981–1984)
Hydrological parameter
Direct drilling
Rotavation
Ploughing at 20 cm
Ploughing at 40 cm
Runoff (mm) Runoff coefficient (%) Turbidity (g l−1 ) Soil losses (t ha−1 )
62.66 15.43 2.16 1.35
54.20 13.36 2.26 1.23
48.60 11.93 2.43 1.17
40.10 9.86 2.80 1.12
Table 25.7 Hydrological data measured from experimental plots under different tillage methods (annual mean) Hydrological parameters
Horsebean for seed (1984–1985, 1986–1987)
Durum wheat (1985–1986, 1987–1988)
Direct Rotavation Ploughing Ploughing Direct Rotavation Ploughing Ploughing drilling at 20 cm at 40 cm drilling at 20 cm at 40 cm Runoff (mm) Runoff coefficient (%) Turbidity (g l−1 ) Soil losses (t ha−1 )
106.90 13.65
85.45 10.85
77.70 9.85
73.85 9.20
67.10 11.50
58.55 10.00
52.65 8.95
45.25 7.70
3.65 4.15
3.75 3.27
4.60 3.66
5.15 3.88
2.80 1.73
2.85 1.63
3.05 1.59
3.60 1.58
Table 25.8 Hydrological data measured from experimental plots under different tillage methods (average over 1990–1993)
Hydrological parameters
Horsebean Runoff (mm) Soil losses (t ha−1 ) Durum wheat Runoff (mm) Soil losses (t ha−1 )
Rotavation
Scarifying at 50 cm depth
Scarifying at 50 cm + ploughing at 40 cm
Ploughing at 40 cm depth
19.9 0.76
13.8 0.73
14.2 0.92
15.9 0.91
19.8 0.39
17.1 0.41
17.0 0.48
18.6 0.55
355
Soil Erosion and Land Degradation
Table 25.9 Soil organic matter and total nitrogen values of the Guardia Perticara experimental field
Depth (cm)
Organic matter (%) Direct
0–10 10–20 20–40 40–50
2.01 1.57 1.42 0.80
Rotavation
1.80 1.49 1.23 0.68
Total nitrogen (%)
Ploughing 20 cm
40 cm
1.50 1.54 1.26 0.95
1.43 1.25 1.37 1.27
Direct
1.42 1.22 1.09 0.81
Rotavation
1.45 1.15 1.05 0.91
Ploughing 20 cm
40 cm
1.04 1.02 1.01 0.94
1.03 0.98 0.99 0.96
Mean values based on 20 soil samples. harrowing or ploughing at 40 cm depth alone. This was true both for durum wheat and for horsebean, the latter showing greater soil loss. Organic matter and total nitrogen were determined in all experimental plots at the end of the trials. Organic matter content (Table 25.9) of the surface layers was higher in the plots that had been left unploughed for years, whereas it had decreased in the plots treated with roto-tillage and in those ploughed at 20 cm depth, and was lowest in those ploughed at 40 cm depth. The 0–10 cm layer had the highest value (2.01%) after no ploughing, and 1.43% after ploughing at 40 cm depth. Similar trends were measured for total nitrogen content (Postiglione et al. 1990).
Superficial and Deep Runoff Superficial runoff occurred after 27 individual rainfall events in the year 1993–1994, and after 23 events during the year 1994–1995. An event was defined as rainfall that fell within a six-hour interval (Linsalata et al. 1983). Of all the rainfall events, 48% produced superficial runoff for both durum wheat and horsebean during the year 1993–1994 compared to 43.2% in the following year. The rainfall events resulting in runoff were divided up according to the period in which they occurred each year. The years were then divided into four periods with respect to the dates on which the main cropping technique was carried out and to a measure of vegetation cover (LAI): • Period 1 was characterized by little or no vegetation cover from the date on which the main cropping technique was performed to the date when LAI values were ≤0.5. This was at the beginning of spring, when the low winter soil temperatures slow plant growth down. • Period 2 included almost all of spring, when vegetation cover was at a maximum, with LAI values >0.5. • Period 3 was when the crops were ripening and becoming senescent rather than growing. The period between the LAI dropping to below 0.5 to the time of harvest was the first 10 days of July for both crops in both years. • Period 4 was between harvest and the first tillage operations to clear crop residues (during the first half of October). These divisions did not indicate differences between the crops. The distribution of the rainfall events in both years gave similar results, with the greatest number of events resulting in superficial runoff occurring during the first period. In general, the occurrence of superficial runoff was infrequent and only after a few events did it exceed 1 mm resulting during the first and last periods (Figure 25.3). The first rainfall event after tillage resulted in low runoff values from the chickpea plot in both years. The durum wheat residues probably played a part in this. Likewise, the rainfall events that occurred just before harvest (events 24 and 25) indicated the more positive effect of vegetation cover with durum wheat compared
356
Mediterranean Desertification Wheat
Chick-pea
700
−1
Soil eroded (kg ha )
600 500 400 300 200 100 0 1993−94
1994−95
1993−94
1994−95
Wheat 1993−94
1994−95
10
Runoff (mm)
8 Harrowing Scarifying + Harrowing
6
Scarifying + Ploughing + Harrowing Ploughing + Harrowing
4 2 0 superficial runoff
deep runoff
superficial deep runoff runoff
Chick-pea 1993−94
1994−95
10
Runoff (mm)
8 6 4 2 0 superficial runoff
deep runoff
superficial runoff
deep runoff
Figure 25.3 Surface, deep runoff and soil losses for wheat and chickpea during the 1993–1995 period
to chickpea. In fact, the latter showed higher runoff probably due to the poor vegetation cover before harvest compared to the vegetation cover of durum wheat. The effect of crop presence on runoff during the two-year period was negligible since the most runoff was recorded when there was no or little vegetation cover (70% for durum wheat and 67% for chickpea). Scarifying (S) + ploughing (P) + harrowing (H) produced the lowest amount of runoff for both crops (10.2 mm for durum wheat and 12.7 mm for chickpea). Periods b and c resulted in
Soil Erosion and Land Degradation
357
limited superficial runoff in both years; a mean value of 0.95 mm was obtained for both crops during 1993–1994 while in the following year the mean superficial runoff was 0.92 mm for durum wheat and 1.7 mm for chickpea. Thirteen deep runoff events occurred for durum wheat and eight for chickpea during 1993–1994. Twelve events occurred for both crops during the following year. These events occurred at the same time or just after the heaviest rainfall events. The measured volumes were greater than those that had occurred in the preceding years with durum wheat and horsebean on the same experimental field in response to the greater volume and intensity of rainfall (De Falco et al. 1994; Pisante et al. 1994). The vegetation structure of the crops also influenced runoff near harvest time. Most rainfall events were only associated with limited soil erosion and only those that occurred during summer and the beginning of autumn caused significant erosion. This is probably due to the high intensity of the rainfall that occurred during these events (between 11.9 and 52.0 mm h−1 ) falling on the sparse vegetation cover. Only one event in each year resulted in large amounts of eroded soil during the winter period, recorded in February 1994 (event 17) and March 1995 (event 37), corresponding to very high runoff measurements. Recorded rainfall intensity during these periods was only between 2 and 4.6 mm h−1 , therefore only superficial runoff produced a high amount of eroded soil. Erosion from under durum wheat gave lower values, with a two-year mean value of 1.131 t ha−1 compared with 1.617 t ha−1 . The lowest values were recorded for durum wheat with tillage methods that excluded ploughing (0.863 t ha−1 for S + H and 0.938 t ha−1 for H). The lowest values for chickpea were found with S + H (1.069 t ha−1 ) and P + H (1.248 t ha−1 ).
3 CONCLUSIONS From the results above it is clear that slope angle is a significant factor for soil erosion in the experimental fields, and carrying out tillage following the land contours rather than tillage downslope is important in reducing the soil erosion risk. When different tillage systems were compared, ploughing at 40 cm depth followed by harrowing resulted in the smallest amount of soil loss. Similar results were obtained after ploughing at 20 cm depth followed by harrowing. In addition, these two tillage systems produced the highest yields and better weed control was obtained. With regard to the protection against erosion afforded by the most common crops of the area, it emerged that medium- and long-term fodder crops give higher soil protection than annual crops, but among annual crops durum wheat is better than horsebean. Taking all the trials together, the amount of runoff water under different tillage systems made up around 6% of the rainfall on average. Soil loss in these hilly areas of southern Italy was limited (generally, slightly more than 1 t ha−1 year−1 , within a range of 0.40–4.15 t ha−1 year−1 ) even if some soil loss occurred every year, as a continuous process. Much more serious are the locally common phenomena of destabilization and landslides, which occur periodically and are the main cause of land degradation. The data obtained in the experimental plots during the period 1990–1995 show that the most significant soil losses corresponded to a small number of rainfall events. These events were characterized by an intensity of between 13.2 and 52 mm h−1 in summer or between 2 and 4.6 mm h−1 in winter when the soil surface was exposed, before the crops had grown to provide an effective cover. The typical Mediterranean climate in the internal hilly areas keeps the soil cold in winter and limits the growth of these crops to the spring months. Superficial runoff was similar under durum wheat and chickpea, but soil erosion was greater under chickpea after all rainfall events. This was due to the poor vegetation cover of the crop in some phases, i.e. at harvest, but also due to other characteristics of the plant that influence the soil surface conditions. In fact, the chickpea crop is small in stature with a poor root system, as well as being sown in rows with a larger space left between crops compared with durum wheat, which fully covers the land on which it is sown. This space allows for the direct erosive action of runoff, leading to a greater superficial runoff.
358
Mediterranean Desertification
REFERENCES Aru A (1991) Il suolo, parte fondamentale dell’ecosistema per l’agricoltura. Agricoltura e Ambiente I, 34–49. Barbieri R and Basso F (1973) Problemi agronomici della conservazione del suolo. Il campo Sperimentale nel Bacino dell’Agri. Annali Facolta Scienze Agrarie dell’ Universita di Napoli-Portici , Serie IV, VII, 1–35. Basso F and Linsalata D (1983) Influenza delle sistemazioni superficiali e delle colture sull’erosione del terreni declivi del bacino dell’Agri. Quaderno Consiglio National Recerca 129, pp. 75–95. Basso F and Postiglione L (1994) Aspetti agronomici della conservazione dei terreni in pendio: sistemazioni e lavorazioni. Rivista di Agronomia XXVIII(4), 273–296. Basso F, Postiglione L and Carone F (1986) Lavorazione del terreno in un ambiente collinare dell’Italia meridionale. Un triennio di prove sull’erosione del suolo e sulla produzione del fumento. Rivista di Agronomia 20(2–3), 218–225. Basso F, Postiglione L and Carone F (1987) Influenza delle modalita di lavorazione di un terreno declive sottoposto a rotazione: favino da seme-frumento duro. Erosione e risultati produttivi. Rivista di Agronomia 21(4), 237–243. Boschi V and Chisci G (1978) Influenza delle colture e delle sistemazioni superficiali sui deflussi della erosione in terreni argillosi di collina. Genio Rurale 41, 7–16. Cavazza L, Linsalata D and De Franchi AS (1983) Nuovi modelli di misuratori per la determinazione della erosione idrica. Quaderno 129, Problemi agronomici per la difesa dai fenomeni erosivi, Consiglio National Recerca, Rome, pp. 45–57. Chisci G and Tellini M (1973) Indagini sperimentali sugli aspetti della conservazione del suolo in piccoli bacini. Ann Ist Sper Studio e Difesa del Suolo, Firenze IV, 39–52. De Falco E, De Franchi AS, Basso F and Postiglione L (1994) Effetti delle modalit`a di lavorazione di un terreno declive a rotazione favino (Vicia faba minor Beck.) frumento (Triticum durum Desf.) sull’erosione e sulla qualit`a dei deflussi. Rivista di Agronomia 28(4), 348–355. Dudal R (1982) Land degradation in word perspective. Journal of Soil and Water Conservation 37, 245–249. Landi R (1984) Regimazione idraulico-agraria e conservazione del suolo. Rivistia di Agronomia XVII, 147–174. Linsalata D, De Franchi AS, Marchione V and Basso F (1983) Un decennio di osservazioni sull’erosivit`a della pioggia in Basilicata. CNR Quaderno 129, 113–124. Marzi V, Linsalata D and De Franchi AS (1983) Primi risultati sull’impiego dei misuratori di erosione del terreno. Quaderno 129, Problemi agronomici per la difesa dai fenomi erosivi, Consiglio National Recerca, Rome, pp. 58–74. Panicucci G (1972) La difesa del suolo. Conv.: La difesa del suolo: le sistemazioni montane e fluviali. CNR, Milan. Pisante M, De Falco E and Basso B (1994) Losses of mineral nitrogen in surface and deep runoff from a durum wheat crop (Triticum durum Desf.) on a sloping soil with different tillage methods. Proceedings of 13th International Conference, 24–29 July, Aalborg, Denmark. International Soil Tillage Research Organization, pp. 341–345. Poesen JWA and Bryan RB (1990) Influence de la longueur de pente sur le russellement: role de la formation de rigoles et de croutes de sedimentation orstoty. Ses Pedol XXV(12), 71–80. Postiglione L (1988) Esperienze di sistemazione nella collina meridionale – Sistemare la collina per difendere il suolo e tutelare I’ambiente. Associazione Nazionale delle Bonifiche. Soceit´a Editore, 11 Mulino, pp. 281–285. Postiglione L (1993) Agriculture and environmental problems in the Mediterranean area (with particular reference to Italy). Medit IV, 35–42. Postiglione L and Marzi V (1983) Prefazione. Quaderno 129, Problemi agronomici per la difesa dai fenomi erosivi, CNR, Rome, pp. 7–9. Postiglione L, Basso F and Linsalata D (1983) Influenza delle sistemazioni superficiali e delle modalita di lavorazione su terreno decisive a rotazione biennale: favino da semefrumento duro. Erosione e risultati produttivi. Quaderno 129, Problemi agronomici per la difesa dai fenomi erosivi , CNR, Rome, pp. 146–161. Postiglione L, Basso F, Amato M and Carone F (1990) Effect of soil tillage methods on soil losses, on soil characteristics and on crop production in a hilly area of Southern Italy. Agricoltura Mediterranea 120, 148–158. Raglione M, Sfalanga M and Torri D (1980) Misura dell’erosione in un ambiente argilloso della Calabria. Annuali Istitute Spermentale Studio e Difesa del Suolo XI, 159–191. Ruhe RV (1969) Quaternary Landscapes in Iowa. Iowa State University Press, Ames, Iowa. Schertz DL (1983) The basics for soil loss tolerances. Journal of Soil and Water Conservation 44, 10–14. Smith RM and Stamey WL (1965) Determining the range of tolerable erosion. Soil Science 100, 414–424.
Soil Erosion and Land Degradation
359
Toderi G (1991) Problemi della conservazione del suolo in Italia. In Agricoltura e Ambiente, Edizione Bologna, 50–59. Verity GE and Anderson DW (1990) Soil erosion effects on soil quality and yield. Canadian Journal of Soil Science 70(3), 471–484. Walter H and Leigh H (1960) Klimadiagramm. G. Fisher, Jena, Germany. Wischmeier WH and Smith DD (1965) Predicting Rainfall Erosion Losses from Cropland East of the Rocky Mountains. Agriculture Handbook 282, USDA, Washington, DC.
26
Social and Economic Conditions of Development in the Agri Valley
E. BOVE AND G. QUARANTA
University of Basilicata, Potenza, Italy
1 INTRODUCTION The Agri Valley may be divided geographically and socio-economically into three distinct parts: Upper, Middle and Lower (Basso et al. 1998). Each of the three can be considered roughly homogeneous in terms of physical environment, natural resources, social conditions and economic development (Quaranta 1997). However, the influences on socio-economic development may come from a much wider area, from the mountains to the alluvial lands of the coastal plain.
2 DISTRIBUTION OF POPULATION 2.1 Contemporary Distribution of Population
The Agri Valley is part of Basilicata, which covers an area of 10 000 km2 in southern Italy, with slightly more than 600 000 inhabitants (ISTAT 1997). The area is divided in two provinces, Potenza (the chief town) and Matera, and 131 municipalities of which about 40 are partially or totally within the Agri Valley. This study concentrates on 29 of the municipalities (Table 26.1). Basilicata is predominantly mountainous or hilly. The woodland and pasturelands were once the summer location of transhumant flocks of sheep and goats that spent the winter near the coast. During the summer, people would live in small farms scattered throughout the woodlands and pasturelands. What distinguishes the hilly areas today is the widespread presence of badlands, shrubby areas, and arable lands cultivated with durum wheat (Bove and Quaranta 1996). The summer drought is extreme, the population is very small and the desolate countryside is liable to frequent landslides. In contrast, the coastal plain has recently been characterized by a rapid and continuous population growth and a big increase in the cultivation of intensive crops such as vegetables, strawberries and citrus fruits. 2.2 Historical Features
Historical settlements in the Agri Basin have been described by Boenzi and Giura Longo (1994). Recently, a Neolithic settlement has been discovered in the Upper Agri Valley (Bianco and Cataldo 1994) a short distance from the source of the Agri River, known in ancient times by the name of Akiris (Adamesteanu 1995). There is also evidence of extensive Roman civilization (Soprintendenza Archeologica della Basilicata 1981). In the Middle Agri Valley archaeologists have uncovered a remarkable monastic settlement (Fonseca 1995) indicating ancient contact between the local population and Greek colonization on the coastal margin of the Lower Agri (Ministero per i Beni Culturali e Ambientali 1996). Later the coasts became infested with malaria, driving the population out (Rossi Doria 1963). Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
362
Mediterranean Desertification
Table 26.1 Total surface, resident population and density in 1995
Communes
Total area km2
Population %
Total
%
Density (inhabitants km−2 )
Upper Agri Valley Marsico Nuovo Paterno Marsicovetere Viggiano Tramutola Grumento Nova Moliterno Sarconi Spinoso Montemurro
592.25 101.03 39.25 37.82 89.03 36.48 66.17 97.65 30.46 37.82 56.54
28.80 4.91 1.91 1.84 4.33 1.77 3.22 4.75 1.48 1.84 2.75
32 245 5468 4246 4443 3181 3253 1919 4966 1389 1831 1549
32.55 5.52 4.29 4.49 3.21 3.28 1.94 5.01 1.40 1.85 1.56
54.44 54.12 108.18 117.48 35.73 89.17 29.00 50.86 45.60 48.41 27.40
Middle Agri Valley Castelsaraceno San Chirico Raparo San Martino d’Agri Armento Corleto Perticara Guardia Perticara Gallicchio Missanello Gorgoglione Cirigliano Rocccanova Aliano Sant’Arcangelo Stigliano
960.18 74.18 83.00 50.25 58.50 88.98 52.95 23.48 22.30 34.23 14.93 61.63 96.32 89.47 209.96
46.66 3.61 4.03 2.44 2.84 4.32 2.57 1.14 1.08 1.66 0.73 3.00 4.68 4.35 10.21
29 766 1932 1537 1085 889 3243 789 1071 685 1326 502 1998 1425 7082 6202
30.06 1.95 1.55 1.10 0.90 3.27 0.80 1.08 0.69 1.34 0.51 2.02 1.44 7.15 6.26
31.00 26.04 18.52 21.59 15.20 36.45 14.90 45.61 30.72 38.74 33.62 32.42 14.79 79.16 29.54
504.94 76.28 156.93 132.94 71.50 67.29 2 057.37 9 992.27 0.21
24.55 3.71 7.63 6.46 3.48 3.27 100.00
37 017 894 5812 8594 6578 15 139 99 028 6 09 238 0.16
37.38 0.90 5.87 8.68 6.64 15.29 100.00
73.31 11.72 37.04 64.65 92.00 224.98 48.13 60.97
Lower Agri Valley Craco Tursi Montalbano Jonico Scanzano Jonico Policoro Total Agri Valley Basilicata Total Agri Valley/Basilicata
Source: elaboration on ISTAT data.
3 3.1
DEVELOPMENTAL ASPECTS OF THE POPULATION
Demographic Dynamics The Agri Valley has experienced social and economic problems in common with the rest of southern Italy, mainly concerned with isolation. Mountains with only limited areas worth cultivating, arid hilly areas and malaria-infested lowlands have discouraged investment so that there are few roads and inadequate services. These factors, together with the absence of secure employment opportunities, have been a barrier to socio-economic progress until recently. A hundred years ago the standard of living was particularly low, soil erosion was widespread following deforestation, and there was the
363
Social and Economic Conditions in the Agri Valley
added threat of natural disasters such as earthquakes, so many southern Italians emigrated overseas (Villani and Massafra 1968). The exodus so worried the mayor of Moliterno that he welcomed the Italian president of that time with the words: “I greet you in the name of my eight thousand citizens, of which three thousand have emigrated to America, and five thousand are preparing to follow them” (Sereni 1968). Between the First and Second World Wars the politics of the Fascist regime in Italy largely prevented further migration, but after that more Italians emigrated, particularly to South America and Australia. From the mid-1950s to the end of the 1970s, migration patterns shifted more to northern Italy and other European countries. Since then emigration has become insignificant but within the Agri Valley there have been changes in the distribution of the population, with the population from isolated rural areas tending to move to new centres on the coast. The total population of the Agri Basin rose from 80 000 in 1861 to approximately 100 000 in the mid-1990s. However, the population of the isolated Upper Agri Valley declined by about 20% over the same period. The population of the historic centre of Marsicovetere, at an altitude of 1000 m a.s.l., has declined to only a few hundred, while down in the valley bottom the new centre of Villa d’Agri has a population of more than 4000. Likewise, in Montemurro, the population has halved since 1861 to around 1550 today. The same trends are occurring in the Middle Agri Valley. While the centre of San Brancato, in the territory of Sant’Arcangelo, continues to expand along the principal road, the population continues to decrease in the old historic centres. Between 1861 and 1995 the population declined by over 70% in Armento and Cirigliano. In contrast, the population throughout the Lower Agri Valley has been expanding. With the exception of Craco, where the population halved over the period 1861–1995, the other municipalities have demonstrated exceptional expansion. In this relatively densely populated sub-area the small town of Policoro has emerged with more than 15 000 inhabitants. 3.2 Migration Consequences
The high numbers of people who have felt forced to migrate from the Agri Basin have greatly modified the demographic structure. The elderly persons index is calculated as a percentage ratio between the resident population over 65 years old and the population under 6 years old (in 1991). In the Agri (Figure 26.1) this index shows clearly the demographic fragility and impoverishment. It is clear that many municipalities are inhabited only by elderly retired people. In Figure 26.1, note
Cirigliano Marsico Nuovo
Stigliano
Craco Corleto Perticara Gorgoglione Guardia Perticara Marsicovetere Paterno Montalbano Jonico Viggiano Gallicchio Aliano Montemurro Scanzano Jonico Tramutola Armento Missanello Grumento Nova Tursi Sant'Arcangelo Spinoso SarconiSan Martino D'Agri Roccanova Moliterno
Policoro
San Chirico Raparo Elderly persons index (%)
Castelsaraceno 93−249 249−341 341−457 457−1138
Figure 26.1
Distribution of the elderly persons index for the Agri Basin in 1991
364
Mediterranean Desertification
Table 26.2 Major socio-economic indicators, 1991 (percentages are shown in parentheses)
Demography
Under 6 years Over 6 years 6–14 years 15–65 years More than 65 Elderness index (65/5 × 100) People with university degree/total population Illiteracy in population >6 years old Retired/total population Total migration value Total population variation Active population Employed Unemployed Agriculture Industry Other activity Agricultural Sector Farms Total surface (ha) Farms <2 ha Farms 2–5 ha Farms 5–10 ha Farms 10–20 ha Farms 20–50 ha Farms >50 ha Industry No. of firms Employed
Upper Agri Valley
Middle Agri Valley
Lower Agri Valley
Total
1.903 30.176 4.223 21.218 4.735 249 2
1.595 29.605 3.711 20.006 5.888 369 2
2.495 33.928 5.577 24.756 3.595 144 2
5.993 93.709 13.511 65.980 14.218 237 2
7
10
6
8
16 −123 1.824
23 −209 1.554
12 −34 2.447
17 −366 5.825
12.286 8.788 (72) 3.498 (28) 2.042 (20) 3.578 (35) 4.650 (45)
12.387 8.373 (68) 4.014 (32) 3.039 (30) 2.978 (29) 4.106 (41)
15.101 10.885 (72) 4.216 (28) 3.752 (30) 3.427 (27) 5.310 (43)
39.774 28.046 (70) 11.728 (30) 8.833 (27) 9.983 (30) 14.066 (43)
4.792 50.851 2.246 (47) 1.108 (23) 662 (14) 401 (8) 260 (5) 111 2
6.178 80.863 2.487 (40) 1.474 (24) 783 (13) 642 (10) 483 (8) 307 5
4.550 41.590 1.311 (29) 1.334 (29) 1.216 (27) 334 (7) 237 (5) 116 3
15.520 173.304 6.044 (39) 3.916 (25) 2.661 (17) 1.377 (9) 980 (6) 534 3
2236 4698
2032 3753
Source: elaboration on ISTAT data. Percentage values are in brackets.
2222 5408
6490 13 859
Social and Economic Conditions in the Agri Valley
365
how the index decreases through the three sub-areas, passing from an extreme of more than 1100 at Cirigliano (Middle Agri Basin), to 93 at Policoro (Lower Agri Basin). One of the consequences of having a predominantly elderly population is the difficulty in arranging schooling for the remaining children, who may be dispersed over a wide area. The situation appears particularly alarming in the marginal centres of the Middle Agri Valley where poverty is rife. In these municipalities, the unemployment rate is very high. For the entire basin it was 30% in 1991, which was almost three times the national figure. Employment in agriculture accounted for nearly 30% of the basin’s workforce in 1991 (Table 26.2).
4 ECONOMIC ACTIVITIES Farming has always played a very important role in the economic system of the Agri Basin. In the high part of the basin there is fertile land in the valley bottoms (about 10 000 ha) and an abundant water supply. The availability of water has favoured agriculture, particularly dairy farming, and recently horticulture and fruit growing, aided by irrigation. In the 1920s rice was introduced but the results were not encouraging (Azimonti 1929). It was in the early 1950s that a programme of expansion and reorganization of irrigation was begun. This enterprise has produced excellent results in terms of productivity and revenue, not only for the big farms but also for the numerous small part-time family farms on the right bank of the Agri River. These farms are noted for high productivity of the cropping system and especially for high quality beans. Recently this rather labourintensive crop (Bove 1993) has been recognized by the European Union as a product with Protected Geographical Indication (PGI). Since this prestigious recognition, acreage of beans has increased by 300% in the last seven years. Beginning with this typical agricultural product, marketed with the name “Fagioli di Sarconi ” (Sarconi’s beans) the possibility of introducing a collection of typical, specialist products of the Upper Agri Valley gradually emerged. A number of factors helped to make this idea a success. There was increasing unemployment as industrial ventures and subsidies failed but here was an opportunity to make money out of tourism (Caneva 1996). This area is rich in archaeological treasures and farm holidays are popular. The number of winter tourists is also increasing. People are interested in rediscovering old traditions (Larotonda 1996), and traditional local foods, such as ham, apples, wine and cheeses from goat and sheep milk. In addition, the National Park of Val d’Agri and Lagonegrese was created. In the Middle Agri Valley one can pass with amazing rapidity from the most desolate uninhabited stretches to tracts that are prodigiously fertile. In fact, only in the beds of Agri River and its tributary the Sauro are there the environmental conditions that permit flourishing forms of agriculture, such as the horticulture and fruit growing on the small-holdings known as “Sant’Arcangelo’s Gardens”. Small-holdings are widespread around the populated centre, where on the steeper slopes generations of hardworking peasants have tended olive trees. Another important characteristic of this area lies in the widespread cultivation of durum wheat on land that is far from naturally suitable. Subsidies have encouraged production, even when the return has proved uneconomic. In the area as a whole, uneven annual incomes and hidden unemployment are common. All the new plans and ventures for specialist products were going well until a multinational consortium recently discovered large oil deposits in the Middle and Upper Agri Valley (Figure 26.2). Mining of the crude oil deposits has begun and could completely change the economy of the region, but it is not clear how much the resident population will benefit. The exploitation of the oil will involve the creation of at least 50 wells, which will surely have a negative impact on the image of the quality of the agricultural products and on the environment (Figure 26.3). In the Lower Agri Valley land reclamation and land reform since 1950 have created systems of production that are second to none in the Mediterranean Basin. Here advanced structures and farming practices produce commodities highly competitive in both the national food processing industry and in the international markets for fresh products. However, these farms are under pressure because the demand for land for urban use and recreational activities has increased dramatically.
366
Mediterranean Desertification
Figure 26.2 The valley bottom in the Upper Agri Valley
Figure 26.3 Oil well in a vineyard
Social and Economic Conditions in the Agri Valley
367
In conclusion, the Agri Basin is a region of widely divergent extremes. Farming will always play an important role in the rural areas, but it must be adapted to prevalent social and environmental conditions.
REFERENCES Adamesteanu D (1995) Fiumi e torrenti nella Lucania antica. In Le vie dell’acqua in Calabria e Basilicata. Carical, Cosenza. Azimonti E (1929) La colonizzazione dell’Alta Val d’Agri. In La colonizzazione in Basilicata. Tipografia del Senato, Roma. Basso F, Bove E, Del Prete M and Pisante M (1998) The Agri Basin, Basilicata, Italy. In P Mairota, JB Thornes and N Geeson (eds) Atlas of Mediterranean Environments in Europe. John Wiley, Chichester, pp. 144–151. Bianco S and Cataldo L (1994) L’insediamento “appenninico” Civita di Paterno (Potenza). Galatina. Boenzi F and Giura Longo R (1994) La Basilicata: i tempi, gli uomini, l’ambiente. Edipuglia, Bari. Bove E (1993) La Montagna lucana. In Indagine sui lavoratori agricoli dipendenti nelle zone interne del Mezzogiorno. Edizioni Scientifiche Italiane, Napoli, pp. 99–125. Bove E and Quaranta G (1996) Desertification in Southern Italy: The Case of Clay-Hill Areas in Basilicata Region. ICALPE, Corte, Corse. Caneva G (1996) Le risorse naturali. In Omaggio alla Val d’Agri . Ars Grafica, Villa d’Agri. Fonseca CD (1995) ‘Et habitavit secus flumen. . .’: i percorsi fluviali di Basilicata in et`a medioevale. In Le vie dell’acqua in Calabria e Basilicata. Carical, Cosenza, pp. 239–276. ISTAT (1997) Popolazione e movimento anagrafico dei comuni . ISTAT, Roma. Larotonda A (1996) Le tradizioni popolari. In Omaggio alla Val d’Agri . Ars Grafica, Villa d’Agri. Ministero per i Beni Culturali e Ambientali (1996) I Greci in Occidente. Electa Napoli. Quaranta G (1997) Interazioni tra strumenti di politica agraria e politica economica: un’ipotesi interpretativa del loro impatto su famiglie – aziende dell’Alta Val d’Agri. In A Cioffi and A Sorrentino (eds) Le piccole aziende e la nuova politica agricola dell’Unione Europea: problemi economici e strutturali . Franco Angeli, Milano. Rossi Doria M (1963) Memoria illustrativa della carta dell’utilizzazione del suolo della Basilicata. Consiglio Nazionale delle Ricerche, Roma. Sereni E (1968) Storia del paesaggio agrario italiano. Editori Laterza, Bari. Soprintendenza Archeologica della Basilicata (1981) Grumentum: la ricerca archeologica in un centro antico. Congedo Editore, Galatina. Villani P Massafra A (1968) Scritti sulla questione meridionale. Laterza, Bari.
27
Characterization of Soil Hydraulic Properties in a Desertification Context
ALESSANDRO SANTINI AND NUNZIO ROMANO
Department of Agricultural Engineering, University of Naples ‘‘Federico II’’, Portici, Naples, Italy
1 INTRODUCTION Evaluating the impact that practical land management applications can exert on hydrological processes is important when solving hydrological, environmental and soil conservation problems. Mathematical models that describe the basic hydrological processes and interactions over time can represent valuable tools to help agencies and private firms to identify problems, for decision-making, or to give guidance to farmers. Such models have become widely available with the increase of cheap, powerful computers. In the past, several approaches based on empirical or semi-empirical concepts have been proposed. The major aims were to sort the large amounts of measured data, and identify the principal variables of the problem and new relationships. Some scientists endeavoured to give the coefficients of empirical or conceptual models, which basically have the character of fitting parameters, also a physical meaning. However, the effectiveness of the related results depends on the input data sets as well as on the mathematical and statistical techniques employed. Classic examples are the Kostiakov (1932) or the Horton (1940) equation for infiltration, or the Universal Soil Loss Equation (USLE) of Wischmeier and Smith (1978), or simple water budget models such as that proposed by Chopart and Vauclin (1990). Physically based, distributed-parameter hydrological models have recently been developed to overcome the intrinsic limitations in current empirical models. Models of this kind focus their attention mainly on the mathematical description of the most significant processes taking place and structural characteristics. Many hydrological models include descriptions of processes such as saturated–unsaturated flow, evaporation, overland flow and channel flow. These physical processes are expressed as non-linear, partial differential equations which for practical interest have to be solved by employing numerical methods and adopting properly designed algorithms to reduce errors due to the discretization of the flow domain. In some cases, however, not all the equations describing such processes are clearly known or it is not always possible to express mathematically the laws that govern the behaviour of a physical process. Therefore, even sophisticated models often employ empirical or semi-empirical relationships. Furthermore, these models are “distributed”, in the sense that they allow for the spatial description of the system characteristics. As the scale increases, for example to embrace a whole catchment, the natural spatial variability of the soil characteristics becomes an important factor that can influence the assessment of the overall system’s response to specific conditions (Wood et al. 1988). The efficiency of the distributed approach is dependent on spatial variability being properly addressed and factors causing spatial variations being correctly modelled. The problem of spatially integrating at a large scale the processes operating at a small scale can represent a typical problem. When faced with land degradation and sustainability problems, understanding the mechanisms with which water moves from the land surface to the groundwater table through the unsaturated zone is of primary importance in predicting catchment hydrological responses, rainfall erosivity, and Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
370
Mediterranean Desertification
sediment deposition. Because of its boundary position between atmosphere and groundwater, soil represents a crucial element within the hydrological cycle as it determines the partitioning of incident water, either rainfall or irrigation water, into runoff and infiltration. The infiltrated water is in turn subject to evaporation at the land surface, transpiration from plants, and percolation processes, which are all affected by changes in the regime of the unsaturated zone, by the status of vegetation, and by climatic conditions (Santini 1992). Whichever mathematical model is used to solve a particular problem, for example the common parametrization of unsaturated flow processes offered by the Richards equation (Richards 1931), or a detailed stochastic approach (Yevjevich 1987), or the comprehensive and sophisticated SHE model (Abbott et al. 1986), there is the recognition that one of the major limitations to applying a model is tied to the availability of information to correctly assess the soil hydraulic behaviour. Furthermore, soil hydraulic characteristics are highly non-linear functions of the moisture regime in soil and in most cases the existence of a complex structure of spatial variations is shown. Therefore, the reliability of model predictions is extremely sensitive to the accuracy with which the hydraulic properties characterizing the soil are determined. Laboratory and field investigations to determine the soil hydraulic properties at larger scales can result in laborious and very expensive investigations since the inherent spatial variability of such properties should also be accounted for. It can thus be better to compromise between accuracy in experimental evaluations and cost-effectiveness of the investigations by applying simplified methodologies. However, this brings out two important questions: the accuracy in estimates from simplified methods, and the sensitivity of the model to the soil hydraulic parameter data. Greater accuracy can be gained only through specific calibration of the simplified method employed with respect to the soil types and local conditions. Once the soil hydraulic properties have been determined, the influence of variability of these properties on model predictions should be assessed. Effort should thus be devoted to developing accurate and cost-effective methods to measure typical variables affecting flow and transport processes in the unsaturated zone. The main objective of this chapter is to review briefly some methods for determining the soil characteristics related to soil hydrological processes and to summarize some recent results obtained to characterize the soil hydraulic behaviour at different scales in the landscape. Special emphasis is devoted to those methods that have proved a tendency towards successful simplifications, so that a trade-off between efficiency at the scale of interest and accuracy in calculated values can be attained. The first example will deal with a parameter estimation method that was specifically developed to reduce laboratory experimental efforts without sacrificing the accuracy of the estimated soil hydraulic parameters. This method is applied to hydraulically characterize differently tilled field plots. Attention is given to the reliability of the results with respect to the type of soil being investigated, which shows typical features of the soils more easily subject to degradation phenomena. A second example will concern the evaluation of different pedo-transfer rules that permit, in a relatively simple way, prediction of the soil hydraulic properties for which there are no measured data, from available information on basic physical properties of soils. This study has been undertaken as part of a larger, collaborative project, the MEDALUS project, which is improving our understanding of the processes that are responsible for land degradation and developing related sustainability issues in Mediterranean environments. One of the project target areas is the Agri River Basin, which is located in southern Italy and has a total drainage area of approximately 1700 km2 . There is a tendency towards land degradation in this basin, especially in its central part. This is mainly due to the fragile lithological structures of the hilly relief, which are susceptible to erosion by rainfall and uncontrolled human activities. The experiments discussed in this chapter were conducted in the catchment of the Sauro River, the Sauro River being the main tributary flowing into the Agri Basin. The Sauro catchment is located within the Middle Agri Valley and represents an interesting area for site-specific studies of surface erosion and land degradation. The soils of the Sauro catchment develop mainly in a xeric moisture and mesic thermic regime, with parent materials mainly consisting of clayey components. The environment has a very dynamic geomorphology. Slides, slide terraces and accumulation glacis dominate the landscape, and such processes thus affect the soils. Where the landscape is quite stable, layering occurs with
Soil Hydraulic Properties in a Desertification Context
371
horizon formation; these soils are well developed and occur only on few areas of the Sauro. As slope gradient increases, chaotic young soils occur with minimum horizon formation. These soils are widely distributed on the slopes of the valley. The hydrology of this catchment is affected by the seasonality of the precipitation. The streamflow regime thus depends strongly on seasonal variations, with low or no flow during most of the year and high discharge peaks of short duration around autumn or early winter.
2 HYDRAULIC CHARACTERIZATION OF SOIL Water flow in soil is typically described by the Richards equation (Richards 1931) whose model parameters are the water retention, θ (h), and hydraulic conductivity, K(θ), functions. These functions are usually referred to as soil hydraulic properties and describe the relationships between the volumetric soil water content, θ , the pressure potential head, h, and the hydraulic conductivity, K, for unsaturated porous media. Comprehensive reviews of existing techniques for measuring soil hydraulic properties are available, but no suitable single method has been developed which performs well in a wide range of circumstances and for all soil types. Laboratory methods entail measurements being taken under controlled conditions by employing complex measuring devices, thus yielding accurate results. They can also turn out to be relatively rapid as it is possible to gather many samples, even originating from different locations, and then run the tests on several of them simultaneously. However, they require extraction of undisturbed samples from soil and this can pose limitations to the validity of laboratory methods in some cases (e.g. structured or cracked soils). Field methods avoid compression of the soil inside the cylinder used to collect the sample and do not lead to changes in soil structure due to sampling and test preparation procedures, but require skilled operators and intensive measuring campaigns, especially if a large number of points need to be characterized. The water retention function θ (h) is usually determined directly in the laboratory on undisturbed soil samples by inducing a series of wetting and drainage events and taking measurements at equilibrium conditions, or in the field by measuring simultaneously water contents and pressure potentials during a transient flow (Bruce and Luxmoore 1986; Klute 1986). Implementation of direct methods to determine the conductivity function is far more difficult, as unsaturated hydraulic conductivity varies over many orders of magnitude not only among different soils, but also for the same soil as the water content varies from saturation to very dry conditions. However, the numerous direct methods for assessing the K(θ) function usually involve measurements of state variables, such as water content and pressure potential, which are well-documented technologies commonly used by different types of users (Dirksen 1991). In general, direct methods are cumbersome and the associated costs of measurement are thus relatively high. Experiments based on these methods often need several stages of steady-state or equilibrium conditions to be reached or require rather restrictive initial and boundary conditions to be imposed in performing the transient flow. Even though they provide very reliable results, their use is limited to specific situations or types of investigations. Some authors have therefore proposed to estimate simultaneously the water retention and hydraulic conductivity functions from a transient flow experiment by using the inverse problem methodology in the form of the parameter optimization technique (Kool et al. 1987). Following this approach experimental operations can be simplified, the employment of sophisticated measuring devices can be avoided, and in most cases the total duration of the experiment can be significantly reduced compared to conventional techniques. Moreover, parameter estimation methods can allow a detailed error analysis of the estimated parameters to be incorporated in the numerical procedure. Both direct and indirect inversion methods perform well when facing water flow problems at a small scale, whereas their use may become inefficient or practically impossible if unsaturated soil hydraulic characterization should involve large land areas or even whole catchments. A valuable attempt to overcome such difficulties was made by introducing predictive methods that estimate the soil hydraulic properties from more easily measured soil attributes. Within these predictive methods, regression equations enabling the unsaturated hydraulic properties to be estimated from
372
Mediterranean Desertification
physico-chemical soil properties such as texture, bulk density, clay mineralogy and organic matter are referred to as pedo-transfer functions, PTFs (Bouma 1989), and are becoming very popular among researchers, soil physicists and field practitioners. 2.1
Parameter Estimation Method
Solving the inverse problem of determining soil hydraulic properties by the parameter optimization approach basically entails minimizing a suitable objective function which expresses the discrepancy between measured values of certain variables during a transient flow experiment and the simulated system response. In this study we present a laboratory inverse method developed to determine simultaneously the water retention and hydraulic conductivity functions of undisturbed soil samples. The laboratory test entails subjecting an initially saturated soil sample of length L to an evaporation process and starts from a hydrostatic equilibrium, with the pressure potential head at the bottom of the soil sample, hL , nearly equal to zero. The evaporation flow is then performed by draining the sample with a small fan placed near the top and with the lower end completely sealed. At the specific time t during the transient flow event, the following variables are measured: total weight of the soil sample, W , and pressure head, h, at different soil depths, assuming z = 0 at the top of the soil sample. The evaporation process is simulated by numerically solving the Richards equation (Richards 1931), which is written here in its pressure head based form: ∂ ∂h ∂h = −1 (1) k(h) C(h) ∂t ∂z ∂z and where C(h) = dθ/dh is the soil water capacity. Equation (1) is subjected to the initial condition: h(z, t) = hL + z − L
t = 0, 0 ≤ z ≤ L
and the following boundary conditions: ∂h K(h) − 1 = E(t) ∂z ∂h −1=0 ∂z
t > 0, z = 0
(2)
(3)
t > 0, z = L
which prescribe at the upper boundary the evaporation rate E(t) derived from sample weight measurements, and the zero-flux condition at the lower boundary. Because of the non-linearity of the partial differential equation (1), due to the strong dependence of K and θ on h, the solution to problem (1) and (2)–(3) is sought numerically using a Crank–Nicolson-type finite-difference scheme (Romano et al. 1998). Unsaturated soil hydraulic properties are usually described by relatively simple, closed-form analytical expressions for the water retention and hydraulic conductivity functions. For the case study reported in the next section, we decided to adopt the following monotonic hydraulic model: −m
θ (h) = θr + (θs − θr )[1 + |αh|1/(1−m) ]
(4a)
k(θ ) = k0 exp[β(θ − θs )]
(4b)
which couples van Genuchten’s θ (h)-function (van Genuchten 1980) with the conductivity function K(θ) as described by an exponential relation (Ciollaro and Romano 1995). Parameters θs and θr represent the saturated and residual values of soil water content, respectively, K0 is the hydraulic conductivity when h = 0, and α(α > 0), m(0 < m < 1) and β are empirical parameters. Parameter β chiefly depends on soil pore size distribution. The unsaturated hydraulic model (4) has been shown to provide reasonable descriptions of the unsaturated behaviour of different types of soils, especially those showing higher percentages of clay contents (Romano and Santini 1999). In this study, the
Soil Hydraulic Properties in a Desertification Context
373
saturated water content θs was measured independently in the laboratory, whereas θr was set at zero. Soil hydraulic characteristics are thus defined through the following four-element parameter vector b = {α, m, log(K0 ), β}. The decimal logarithm of the parameter K0 is optimized in this study. The unknown parameter vector b is determined by minimizing the following objective function: O(b) = [h − h∗ (b)]T W[h − h∗ (b)]
(5)
where h is the observation vector, whose elements are pressure heads hij measured at time ti and soil depth zj , and h∗ is the simulated response vector, whose elements are the simulated pressure heads hij (b) at the same space–time co-ordinate as computed by the numerical model for a given parameter vector b. The matrix W is a weighing matrix. Details about specification of the objective function and solution of the optimization problem can be found in the paper by Romano and Santini (1999).
Application of the Laboratory Inverse Method at Plot Scale The laboratory inverse method presented in the previous section was applied to determine the unsaturated soil hydraulic properties at an experimental farm operated by the University of Basilicata and University of Naples “Federico II”. The site is located in the Sauro catchment, near the village of Corleto Perticara, at about 700 m a.s.l., and has an annual average temperature of 12 ◦ C and an annual average precipitation of about 790 mm. The soil at the study area was classified as Vertic Ustorthent, with an Ap horizon of 0.3 m in thickness, overlying a Cca horizon that extends to a depth of approximately 1 m below soil surface. The present investigation was conducted on the four field plots depicted in Figure 27.1 and hereafter identified as Plots P1, P2, P3 and P4. All experimental plots consisted of an area 40 m long by 15 m wide, with an average longitudinal slope of 14% and a nearly zero lateral slope, and were used for two-year rotations of winter wheat (Triticum durum, Desf.) and horsebean for seed (Vicia faba minor, Back.) for seven years before soil sampling. The plots received the following treatments: zero tillage for Plot P1; conventional ploughing (ploughing to a depth of about 20 cm + harrowing) for Plot P2; deep ploughing (ploughing to a depth of about N
735
730
725
P1
P2
Contour lines (m)
g ch uar an d n sa el po mp int lin
P3 P 4
g
dr
ain
g rin su ea e m um fl
720
sto tan rag k e
715
710
Figure 27.1 Plan of the ‘‘Corleto’’ experimental farm, illustrating the plots under investigation. Solid circles identify core sampling locations within each plot
374
Mediterranean Desertification
40 cm + harrowing) for Plot P3; and minimum tillage (scarifying + ploughing to a depth of about 20 cm + harrowing) for Plot P4. Tillage was usually performed in the autumn. The data used in the present investigation were collected in March of the seventh year of tillage treatments when all plots were under winter wheat. Undisturbed soil cores were taken from each of the four plots with a sampling design consisting of 10 sampling points systematically distributed throughout each plot, as shown in Figure 27.1. Two soil cores were taken from the uppermost layer at each sampling point, giving a total of 80 samples. Each soil core was analysed to determine the particle-size distribution, bulk density, saturated water content and hydraulic conductivity values. The laboratory inverse method was employed to estimate simultaneously the water retention and hydraulic conductivity functions of soil cores of a group of 40 samples. Each soil core pertaining to this group had a length L of 10 cm and was 8.5 cm in diameter. Transient evaporation experiments were performed in a constant-temperature room using an apparatus that can test up to eight soil cores simultaneously, as shown schematically in Figure 27.2. Input data for the inverse optimization procedure, such as soil water pressure heads at the two depths of 3 cm and 6 cm below the top of the core and total soil weights, were monitored using a data-logger and a computer. To evaluate the accuracy of the predictions of the proposed inverse method, measured water retention characteristics up to pressure potential heads of −2.5 m were obtained from the other 40 soil cores by a suction table apparatus (Romano et al. 2002). Each soil core from this group was 8 cm in diameter and 5 cm in height. As an example, Figure 27.3 compares the results for Plot P3. In this figure the circles refer to the mean water retention characteristics measured under equilibrium conditions by the sandbox apparatus, whereas the squares represent the mean values of the optimized water retention curves according to equation (4a). It is apparent that while the two methods are virtually indistinguishable in almost all the investigated range of measured pressure heads, they only lead to some discrepancy close to saturation conditions. The different sizes of the cores from the two groups, as well as various levels of disturbance during sampling and some small fan
base pressures
soil sample tensiometer
load cell
control system data-logger
stepper motor
pressure transducer
Figure 27.2 Schematic representation of the system used for the evaporation experiments
375
Soil Hydraulic Properties in a Desertification Context 0.5 equilibrium
0.4
Water content
transient
0.3 P3
−4
−3
−2 Pressure head (m)
−1
0
0.2
Figure 27.3 Mean water retention characteristics for Plot P3. Circles are the means of the independently measured retention data points, whereas squares are the means of the inverse estimated water retention curves. Vertical bars are ±1 standard deviation 0.5
0.4
Water content
P1 P2 P3 P4
0.3
−4
Figure 27.4
−3
−2 Pressure head (m)
−1
0
0.2
Mean inverse estimated soil water characteristics for the four investigated plots
different degrees of complete saturation, can account for the relatively small differences observed for h greater than about −0.5 m. Notice that the width of the vertical error bars illustrates that both methods have reproduced the same range of variability, thus indicating a very good efficiency of the proposed inverse method. Figure 27.4 shows the mean water retention curve calculated by averaging the 10 optimized θ (h)-curves for each of the four plots. With the exception of the mean retention curve of Plot P4,
376
Mediterranean Desertification
the shape of the mean θ (h)-curves are similar for Plots P1, P2 and P3, and the three tillage treatments do not appear to affect the retention characteristics significantly. It is interesting that the mean values of water content at saturation are nearly identical among the plots, suggesting that effects of longterm tillage on the structure of the soil being investigated are negligible. Even though the four mean saturated water contents are very close together, the mean θ (h)-curve associated with Plot P4 clearly follows a different pattern with respect to the other three plots. The expectation would be that tillage would exert a greater influence for Plot P3, since this plot was subjected to deeper ploughing, but an explanation of the behaviour of the retention characteristics of Plot P4 can be found by looking at the particle-size distributions of the collected soil cores. Inspection of Figure 27.5, which depicts the mean particle-size distribution (PSD) curves derived by arithmetically averaging all the observed PSD points for each plot, reveals that the mean PSD curves of Plots P1, P2 and P3 do not differ significantly, although some differences are noticeable within the fine sand fraction (according to the ISSS classification). On average, the soil at Plot P4 shows a higher percentage of particles greater than 0.02 mm and a smaller percentage of particles less than 0.002 mm. Thus, the occurrence of remarkable differences between the soil water retention curve at P4 and those of the remaining plots chiefly can be attributed to the observed differences in the particle-size distributions. The rapid decrease in water content as pressure potential increases shown by the θ (h)-curve of Plot P4 (Figure 27.4) can be explained by the fact that on average the soil of this plot has the smallest percentage of clay (Figure 27.5). To summarize the results and further show the effectiveness of the proposed inverse method, in the following all calculations were referred to the data sets from Plots P1, P2 and P3, as they can be regarded as a statistically homogeneous sample. Figure 27.6 shows the mean θ (h) and K(θ) hydraulic properties of the soil under investigation. The mean water retention characteristics (Figure 27.6(a)) obtained with the two methods follow the same pattern: averaging the data from Plots P1, P2 and P3 has made discrepancies between measured saturated water contents of the two different groups of soil cores almost negligible. Figure 27.6(b), which depicts the mean optimized hydraulic conductivity curve, also shows the average values of measured Ks or log(Ks ). These values are both higher than the estimated hydraulic conductivity at saturation obtained by extrapolation. This tendency is consistent with the findings of Ciollaro and Romano (1995), who pointed out the presence of a possible bias between the optimized and independently measured saturated hydraulic conductivity.
Coarse sand
% Particle < d
100
Fine sand
Silt
Clay P1 P2 P3 P4
50
0
2
0.2
0.02
0.002
Particle diameter, d (mm)
Figure 27.5 Mean particle-size distributions for the four investigated plots
0.0002
Soil Hydraulic Properties in a Desertification Context
377
(b) 102
0.5
(a) equilibrium
0.4
0.3
−4
−3
−1 −2 Pressure head (m)
0
Water content
transient
Hydraulic conductivity (cm h−1)
mean value of Ks
0.2
mean value of log Ks 100
10−2
10−4
10−6 0.3
0.35 0.40 Water content
0.45
Figure 27.6 Mean soil hydraulic properties for the ‘‘Corleto’’ experimental farm: (a) soil water retention function and (b) hydraulic conductivity function. Dots represent the mean of 30 values using data from plots P1, P2 and P3
Predictive Methods Based on Pedo-transfer Functions A pedo-transfer function is essentially a regression equation which relates (or transfers) some available information of soil physical and chemical properties, i.e. texture, bulk density and organic matter, to soil water retention and hydraulic conductivity characteristics. Results from evaluating two different PTFs proposed in the literature to predict the soil water retention functions are presented here. Romano and Santini (1997) gave more comprehensive validations of widely used PTFs. The PTFs selected here were defined as continuous pedo-transfer functions and developed following two different approaches. The first PTF proposed by Gupta–Larson (Gupta and Larson 1979) θ (hi ) = (ai × Sa) + (bi × Si) + (ci × Cl) + (di × OM) + (ei × ρb )
(6)
which predicts values of water content at specific pressure potential heads, follows the so-called “point regression approach” (Tietje and Tapkenhinrichs 1993). The symbols Sa, Si and Cl represent, respectively, percentages of sand, silt and clay according to the FAO definitions (FAO/UNESCO 1994), OM is organic matter expressed as a percentage of the <2000 µm soil fraction, and ρb is oven-dry bulk density. Note that originally the authors developed this PTF using air-dried values of bulk density. The regression coefficients depend on the specific pressure heads considered and are listed in Table 27.1 for hi equal to −1 and −10 m. The other PTF was proposed by Rawls and Brakensiek (1989)
1 −1 (λ+1) λ+1 h (7) θ (h) = θr + (θs − θr ) 1 + hb and follows the “functional parameter regression approach”, i.e. it predicts the parameters in a closed-form analytical relation for the water retention function. Table 27.1 reports the regression equations used to predict the four parameters of this relation. Note that equation (7) is structurally similar to van Genuchten’s equation (4a) with α = (hb )−1 and m = λ/(λ + 1).
Application of Predictive Methods at Hillslope Scale Data presented in this section were obtained from undisturbed soil samples collected at 50-m spacing along a 5-km-long transect from the topsoil of a hillslope of the Sauro catchment. This transect ran
378
Mediterranean Desertification
Table 27.1 Regression coefficient values and parameter equations for the PTFs (6) and (7)
PTF
Regression coefficients or parameter equationsa
Pressure head (m)
Gupta–Larson −1.0 −10.0
a × 103 5.018 1.563
b × 103 8.548 3.620
c × 103 8.883 7.154
d × 103 4.996 2.388
Rawls–Brakensiek θs = φ θr = −0.0182 + (0.000873 Sa) + (0.00513 Cl) + (0.0294 φ) − (0.000154 Cl2 )
e × 102 −24.230 −5.759
−(0.00108 Sa φ) − (0.000182 Cl2 φ 2 ) + (0.000307 Cl2 φ) − (0.00236 φ 2 Cl) ln(hb ) = 5.340 + (0.184 Cl) − (2.484 φ) − (0.00214 Cl2 ) − (0.0436 Sa φ) − (0.617 Cl φ) + (0.00144 Sa2 φ 2 ) − (0.00855 Cl2 φ 2 ) − (0.0000128 Sa2 Cl) + (0.00895 Cl2 φ) − (0.000724 Sa2 φ) + (0.0000054 Cl2 Sa) + (0.500 φ 2 Cl) ln(λ) = −0.784 + (0.0177 Sa) − (1.062 φ) − (0.000053 Sa2 ) − (0.00273 Cl2 ) −(0.00000235 Sa2 Cl) + (0.00799 Cl2 φ) − (0.00674 φ 2 Cl) a
Sa, % sand; Cl, % clay; ρb , oven-dry bulk density; φ, porosity; θs , saturated water content; θr , residual water content; OC, % organic carbon.
right through the “Corleto” experimental farm described previously. All soil samples were subjected to standard laboratory measurements to determine oven-dry bulk density (ρb ), particle density (ρs ), particle-size distribution (PSD), organic carbon content (OC) and soil water retention curves θ (h) (Blake and Hartge 1986; Gee and Bauder 1986; Klute 1986; Nelson and Sommer 1986). Total porosity φ was calculated from the measured oven-dry bulk density and particle density using the relation φ = 1 − ρb /ρs , while organic matter was calculated from the measured organic C using the empirical factor 1.724. The collected samples comprised a relatively wide range of soil texture classes, with variations from about 3 to 58% for sand contents and from about 14 to 53% for clay contents, while silt contents were within a narrower range from about 24 to 54% (Romano and Santini 1997). Water retention data points were measured up to a pressure head of about −2.5 m using the sand-kaolin boxes, and at −30, −60 and −120 m by a membrane plate apparatus. To allow more objective comparisons between the observed and predicted retention characteristics, for each soil sample the measured θ (h)-values were fitted with van Genuchten’s closed-form expression (4a). This analysis was carried out by fixing the parameter θr at zero and estimating the remaining model parameters θs , α and n = 1 − 1/m using the RETC software package (van Genuchten et al. 1991). Therefore, the selected pedo-transfer functions are evaluated by using the fitted water retention data sets as a reference for comparisons. The frequency distributions were statistically analysed for all fitted and predicted data sets. Data were tested for normality at 95% significance level using the procedure proposed by D’Agostino et al. (1990). This test is based on the K 2 statistic, which accounts for the combined effects of skewness and kurtosis. The criterion used in this study for assessing the performance of the two different PTFs is to evaluate the maximum error (MXE), the mean error (ME) and the root mean square error (RMSE): N N (err i ) N × 100 ME = MXE = maxi=1 [abs(err i )] i=1
RMSE =
N i=1
1/2 2 (err i ) N × 100
379
Soil Hydraulic Properties in a Desertification Context
where N is the number of observations and err i represents the error between the fitted and predicted water retention values. Table 27.2 shows comparisons between fitted and PTF-estimated water content θ at −1 and −10 m and summarizes the efficiency of using a pedo-transfer rule. In terms of sample means and standard deviations, the PTF of Rawl–Brakensiek provides better estimates of water contents along the hillslope than the PTF of Gupta–Larson. This outcome may have important implications on the effect exerted when using a simplified estimation procedure on reliably assessing the mean value of a variable in an area. From a broader point of view of using hydrological models at large scales, the choice of a method, which enables the accurate determination of the mean values of certain soil hydraulic properties, becomes a matter of primary importance. Water contents predicted by the PTF of Gupta–Larson have a much larger variability in comparison with the fitted data sets, illustrating qualitatively a lack of efficiency of this pedo-transfer function. However, despite statistically significant differences among individual means and standard deviations, the CVs of the PTF-estimated water contents compare closely with the CVs calculated for the fitted water contents. This indicates a high similarity in relative dispersion between the water retention data sets. In addition, the CVs in Table 27.2 increase from wetter to drier regions of the water retention functions. The K 2 values indicate that the considered data sets are normally distributed. However, if we compare the computed Kc2 of fitted and predicted data, the related frequency distributions show differences of both skewness and kurtosis characteristics. The maximum error, mean error and root mean square error values listed in Table 27.2 provide a quantitative evaluation of the greater efficiency of the Rawls–Brakensiek PTF than that of the PTF of Gupta–Larson, and this agrees with evidence from Romano and Santini (1997) where the functional parameter regression approach is superior to the point regression approach. Figure 27.7 shows fitted data against estimated water contents at h = −1 m. Figure 27.8 depicts parameters α and n = 1/(1 − m) of van Genuchten’s relation θ (h) as computed by the RETC program and the corresponding parameters predicted by the pedo-transfer rule of Rawls–Brakensiek. In Figure 27.8(a), the computed αVG and predicted αPTF are mostly contained in a relatively narrow band, which is very different from the 1:1 straight line, while at larger values the points spread in a fan-shaped way. For the examined soil samples, the PTF of Rawls–Brakensiek has a systematic overestimation of parameter αVG . As for parameter n, Figure 27.8(b) shows an undefined correspondence between the computed, nVG , and predicted, nPTF , data points. For values of nVG greater than about 1.5, the variable nPTF is essentially constant and thus the PTF of Rawls–Brakensiek is unable to interpret the soil behaviour at larger n values, especially for soils with a predominant sand fraction. On the other hand, Rawls and Brakensiek (1989) warned against the use of their
Table 27.2 Statistics of fitted and PTF-predicted water contents
Summary statisticsa
Variable µ
σ
CV
Kc2
0.390 0.286
0.0463 0.0482
11.9 16.9
0.15 0.58
PTF-estimated water retention data Gupta–Larson θ−1 0.432 0.255 θ−10 0.372 Rawls–Brankensiek θ−1 0.279 θ−10
0.0576 0.0517 0.0441 0.0507
13.3 20.3 11.9 18.2
4.1 0.87 2.1 1.3
Fitted water retention data θ−1 θ−10
Maximum error
Mean error
RMS error
0.154 0.148 0.0970 0.121
−4.26 3.09 1.77 0.755
5.95 5.91 3.79 4.95
µ, Mean; σ , standard deviation; CV, coefficient of variation (%); Kc2 , K 2 statistic for normality 2 = 5.99). test computed from the data (Kα=0.05 a
380
Mediterranean Desertification
0.5
0.5
PTF-estimated q−1
(b) 0.6
PTF-estimated q−1
(a) 0.6
0.4
0.3
0.4
0.3
PTF-G&L
PTF-R&B
0.2
0.2 0.2
0.3
0.4
0.5
0.6
0.2
0.3
Fitted q−1
0.4
0.5
0.6
Fitted q−1
Figure 27.7 Scatter diagrams of fitted versus predicted water content θ−1 for the pedo-transfer function proposed by (a) Gupta–Larson and (b) Rawls–Brakensiek. Dots represent 100 samples collected along the transect (b) 2.5 PTF-R&B 0.06
PTF-estimated n
PTF-estimated a, 1/cm
(a) 0.08
0.04 0.02
2.0
1.5
PTF-R&B 1.0
0.00 0.00
0.02
0.04
0.06
0.08
1.0
VG Parameter a, 1/cm
1.5 2.0 VG Parameter n
2.5
Figure 27.8 Comparisons between the shape parameters α and n of the water retention curve (4a) as estimated by the RETC code, αVG and nVG , and by the pedo-transfer rule of Rawls–Brakensiek, αPTF and nPTF
pedo-transfer function for soils with a percentage of sand greater than 70%. Table 27.2 shows that overall the degree of efficiency of the PTF proposed by Rawls and Brakensiek is relatively very high, even though there is a relatively poor relationship between the computed and PTF-predicted parameters α and n. Therefore, because of the high non-linearity of the closed-form expressions (4a) and (7) it does not seem important to seek a good correlation between the computed and predicted shape parameters featuring in the relationships for the unsaturated soil hydraulic properties.
3
ISSUES RELATING TO SOIL SPATIAL VARIABILITY AND CONCLUDING SUMMARY
The selection of a mathematical model to understand natural phenomena or to predict the behaviour of a particular physical system is never simple; it is unavoidably reliant upon economic questions, personal biases and modeller’s experience and expertise, as well as depending on hydrological issues, the scientific rigour to be applied, and the available data. In practice, distributed models do not need long time series of meteorological and hydrological data for their calibrations, yet they demand an
Soil Hydraulic Properties in a Desertification Context
381
enormous amount of information to describe the spatial distributions of catchment parameters and inputs, such as topographic characteristics, vegetation coverage, land use and soil properties. Owing to their physical relevance, some parameters of the system under consideration should be determined in the field independently of the hydrological model being used. However, as the area of interest increases and the computational grid scale becomes larger than the scale of variability, field campaigns become very expensive to carry out and model parameters may begin to lose their physical significance (Woolhiser 1996). It is thus evident that two basic problems affect the practical implementation of a hydrological model: spatial variability and scale. These two concepts also interact dynamically. Therefore, assessing spatial variability characteristics over different scales of interest is currently receiving much attention and is becoming an important research topic in hydrology (Wood et al. 1990; Grayson et al. 1992; Santini et al. 1996; Romano and Santini 1997). The problem outlined above applies particularly to the hydraulic parameters of soil, mainly because of the significant non-linearities of the unsaturated flow processes involved. Beven (1989) raised important issues about the parametrization of land surface hydrology and the meaning and use of effective parameters at the model grid scale. However, some answers have been promptly supplied not only by Beven (1993, 1995), but also by other investigators with a slightly different point of view, such as the research group led by Feddes (Feddes et al. 1993; Kabat et al. 1997). These studies reinforce the need for detailed, but limited in number, local soil hydraulic characterizations that would set up a good basis of information to allow estimation of effective soil hydraulic parameter values at different grid scales. It has been recognized that indirect inverse methods based on parameter optimization techniques can offer an attractive tool for assessing large-scale effective soil hydraulic parameters. The indirect inversion approach involves the numerical solution of an equation, generally the Richards equation (Richards 1931), which is assumed to govern a basic flow process (such as infiltration, drainage or evaporation), subjected to imposed initial and boundary conditions. Estimating the unsaturated soil hydraulic properties from data collected during a basic flow process provides invaluable information to derive a set of effective parameter values for various soil types or, if a simplified procedure is required, for the different areas associated with different soil–landscape units of the catchment. A map of such values can then be established. An analogous framework can be applied to build up a map of values of a certain functional variable that is assumed to characterize specific behaviour of a physical system. Yet, in some cases a detailed soil hydraulic characterization provided by the inverse methods can be difficult to carry out because of the lack of means or shortage of skilled technicians, but one can use simplified approaches which have shown reasonable efficiency in predicting soil hydraulic properties. Some of the proposed pedo-transfer rules can meet those requirements (Romano and Santini 1997). For certain types of practical applications, this problem is also closely linked to the use of a method that is efficient in determining the system’s parameters at the scale of interest. With specific reference to the parametrization of unsaturated flow processes, a successful method should be able to evaluate the soil hydraulic properties easily and inexpensively, but without losing valuable information for the subsequent analysis of soil spatial behaviour (Romano 1993). The approaches presented in this chapter for determining soil hydraulic properties can be carried out by means of simple instrumentation techniques and hence may offer substantial savings in terms of experimental efforts. We have reviewed results obtained from an integrated work carried out on hydraulically characterizing unsaturated porous systems and have shown that certain measurement techniques can be successfully modified, by introducing appropriate simplifications, to satisfy requirements of time-efficiency together with accuracy in evaluation.
ACKNOWLEDGEMENTS Primary funding for this research was provided in part by the Commission of European Communities, DG XII, under the Environment and Climate Research Programme (“MEDALUS II: Project 4, Research and Policy Interfacing in Selected Regions”, contract EV5V-CT92-0166, and “MEDALUS III: Project 1, Core Project”, contract ENV4-CT95-0115). We would also like to acknowledge the support of the CEC-Human Capital and Mobility Programme “Using Existing
382
Mediterranean Desertification
Soil Data to Derive Hydraulic Parameters for Simulation Models in Environmental Studies and in Land-use Planning” (contract CHRX-CT94-0639). Part of the work described was published in Journal of Hydrology and Soil Science. The authors wish to thank Elsevier Science BV and Williams & Wilkins for granting permission to reproduce here some material reported in these papers.
REFERENCES Abbott MB, Bathurst JC, Cunge JA, O’Connell PE and Rasmussen J (1986) An introduction to the European Hydrological System–SHE 1. History and philosophy of a physically-based, distributed modelling system. Journal of Hydrology 87, 45–59. Beven K (1989) Changing ideas in hydrology – the case of physically-based models. Journal of Hydrology 105, 157–172. Beven K (1993) Estimating transport parameters at the grid scale: on the value of a single measurement. Journal of Hydrology 143, 109–123. Beven K (1995) Linking parameters across scale: subgrid parameterizations and scale dependent hydrological models. Hydrological Processes 9, 507–525. Blake GR and Hartge KH (1986) Particle density. In A Klute (ed) Methods of Soil Analysis, Part 1: Physical and Mineralogical Methods, 2nd edition. Agronomy Monograph 9, ASA and SSSA, Madison, Wisconsin, pp. 377–382. Bouma J (1989) Using soil survey data for quantitative land evaluation. Advances in Soil Science 9, 177–213. Bruce RR and Luxmoore RJ (1986) Water retention: field methods. In A Klute (ed) Methods of Soil Analysis. Part I: Physical and Mineralogical Methods, 2nd edition. Agronomy Monograph 9, ASA and SSSA, Madison, Wisconsin, pp. 663–686. Chopart JL and Vauclin M (1990) Water balance estimation model: field test and sensitivity analysis. Soil Science Society of America Journal 54, 1377–1384. Ciollaro G and Romano N (1995) Spatial variability of the hydraulic properties of a volcanic soil. Geoderma 65, 263–282. D’Agostino RB, Belanger A and D’Agostino RB (1990) A suggestion for using powerful and informative tests of normality. The American Statistician 44, 316–321. Dirksen C (1991) Unsaturated hydraulic conductivity. In KA Smith and CE Mullins (eds) Soil Analysis–Physical Methods. Marcel Dekker, New York, pp. 209–269. FAO/UNESCO (1994) Soil Map of the World . Revised legend, with corrections. ISRIC Technical Paper 20, International Soil Reference and Information Centre, Wageningen, The Netherlands. Feddes RA, Menenti M, Kabat P and Bastiaanssen WGM (1993) Is large-scale inverse modelling of unsaturated flow with areal average evaporation and surface soil moisture as estimated from remote sensing feasible? Journal of Hydrology 143, 125–152. Gee GW and Bauder JW (1986) Particle-size analysis. In A Klute (ed) Methods of Soil Analysis. Part 1: Physical and Mineralogical Methods, 2nd edition. Agronomy Monograph 9, ASA and SSSA, Madison, Wisconsin, pp. 383–411. Grayson RB, Moore ID and McMahon TA (1992) Physically based hydrologic modeling: 2. Is the concept realistic? Water Resources Research 26, 2659–2666. Gupta SC and Larson WE (1979) Estimating soil water retention characteristics from particle size distribution, organic matter percent, and bulk density. Water Resources Research 15, 1633–1635. Horton RE (1940) An approach towards a physical interpretation of infiltration capacity. Soil Science Society of America Proceedings 5, 399–417. Kabat P, Hutjes RWA and Feddes RA (1997) The scaling characteristics of soil parameters: from plot scale heterogeneity to subgrid parameterization. Journal of Hydrology 190, 363–396. Klute A (1986) Water retention: laboratory methods. In A Klute (ed) Methods of Soil Analysis Part 1: Physical and Mineralogical Methods, 2nd edition. Agronomy Monograph 9, ASA and SSSA, Madison, Wisconsin, pp. 635–662. Kool JB, Parker JC and van Genuchten MTh (1987) Parameter estimation for unsaturated flow and transport models – a review. Journal of Hydrology 91, 255–293. Kostiakov AN (1932) On the dynamics of the coefficient of water percolation in soils and on the necessity of studying it from a dynamic point of view for purposes of amelioration. Transactions of the 6th Commission of the International Soil Science society, Russian, Part A, pp. 17–21. Leij F, Russell WB and Scott ML (1997) Closed-form expressions for water retention and conductivity data. Ground Water 35, 848–858.
Soil Hydraulic Properties in a Desertification Context
383
Nelson DW and Sommer LE (1986) Total carbon, organic carbon, and organic matter. In AL Page (ed) Methods of Soil Analysis. Part 2, 2nd edition, Agronomy Monogr 9, ASA and SSSA, Madison, Wisconsin, pp. 539–579. Rawls WJ and Brakensiek DL (1989) Estimation of soil water retention and hydraulic properties. In HJ MorelSeytoux (ed) Unsaturated Flow in Hydrologic Modeling – Theory and Practice. NATO ASI Series, Vol 9, Kluwer Academic, Dordrecht, pp. 275–300. Richards LA (1931) Capillary conduction of liquids through porous mediums. Physics 1, 318–333. Romano N (1993) Use of an inverse method and geostatistics to estimate soil hydraulic conductivity for spatial variability analysis. Geoderma 60, 169–186. Romano N and Santini A (1997) Effectiveness of using pedo-transfer functions to quantify the spatial variability of soil water retention characteristics. Journal of Hydrology 202, 137–157. Romano N and Santini A (1999) Determining soil hydraulic functions from evaporation experiments by a parameter estimation approach: experimental verifications and numerical studies. Water Resources Research 35, 3343–3359. Romano N, Brunone B and Santini A (1998) Numerical analysis of one-dimensional unsaturated flow in layered soils. Advances in Water Resources 21, 315–324. Romano N, Hopmans JW and Dane JH (2002) Water retention and storage: suction table. In JH Dane and GC Topp (eds) Methods of Soil Analysis, Part 4: Physical Methods. SSSA, Madison, Wisconsin. Santini A (1992) Modelling water dynamics in the soil–plant–atmosphere system for irrigation problems. Excerpta 6, 133–166. Santini A, Romano N and Coppola A (1996) Geostatistical analysis of soil spatial variability in a hillslope of the Agri river basin. Proceedings of the Conference on “Problems with Large Irrigation Districts” East-Sesia Farmers’ Union. SIREA, Torina, pp. 281–293 (in Italian, with English abstract). Tietje O and Tapkenhinrichs M (1993) Evaluation of pedo-transfer functions. Soil Science Society of America Journal 57, 1088–1095. Van Genuchten MTh (1980) A closed form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Science Society of America Journal 44, 892–898. Van Genuchten MTh, Leij FJ and Yates SR (1991) The RETC code for quantifying the hydraulic functions of unsaturated soils. USEPA Report 600/2-91/065. RS Kerr Environment Research Laboratory, US Environmental Protection Agency, Ada, Oklahoma. Wischmeier WH and Smith DD (1978) Predicting Rainfall Erosion Losses. Agriculture Handbook 537, USDA Agricultural Research Service, Washington, DC. Wood EF, Sivapalan M, Beven K and Band L (1988) Effects of spatial variability and scale with implications to hydraulic modeling. Journal of Hydrology 102, 29–47. Wood EF, Sivapalan M and Beven K (1990) Similarity and scale in catchment storm response. Reviews in Geophysics 28, 1–18. Woolhiser DA (1996) Search for physically based runoff model – a hydrologic El Dorado. ASCE Journal of Hydrological Engineering 122, 122–129. Yevjevich V (1987) Stochastic models in hydrology. Stochastic Hydrological Hydraulics 1, 17–36.
28
Aspects of Forestry in the Agri Environment
AGOSTINO FERRARA,1 VITTORIO LEONE1 AND MALCOLM TABERNER2 1 2
Universita` degli Studi della Basilicata, Potenza, Italy c/o Institute for Environment and Sustainability, Ispre, Italy
“Forests come before people and deserts follow them” (Chateaubriand)
Symbols of mystery and wild landscapes; places of refuge and worship; sources of knowledge and irreplaceable resources – forests are now threatened, more than ever before, by degradation phenomena arising from misuse and land abandonment. Proper management of forests is crucial but such is the scale of the problem that priority must be given to forests in territories prone to desertification where they act as a protective entity and a biological filter.
1 INTRODUCTION The Val d’Agri has one of the most interesting landscapes in the Basilicata Region on account of its structural and environmental characteristics, its extent, its evolutionary aspects, and its socioeconomic peculiarities. The valley is a microcosm of all the features that make the Basilicata territory unique. Diversity is a particular characteristic of the woodland in the Agri where many interesting features are exhibited. Consider the range, for example, from the high stands of Fagus sylvatica to the residual, and rare, Abies alba forests; from the diffused, yet dense, Quercus cerris forests to the scrubby Mediterranean macchia, the most widespread form of natural vegetation in the most typically Mediterranean parts of the basin. Unfortunately, this scenario is contrasted with the increasing proportion of barren areas where accelerated soil deterioration and long-term decreases in biological productivity are evident. Intensive human activities and sudden extreme climatic events cause stress and desertification, threatening the landscape and its diversity. Forests can play a major role in this context by allaying the deterioration in vegetation cover and preventing further soil and land degradation (Mainguet 1994). The aim of our work has been to analyse the forestry aspects of the Agri Basin, to illustrate the main problems encountered and propose solutions, and discuss the possibilities for improving management.
2 THE AGRI ENVIRONMENT The Val d’Agri is one of the biggest hydrographic basins in the Basilicata Region of southern Italy. It follows a NW–SE direction extending from the eastern sector of the Lucanian Apennine Chain to the Ionian Sea (Ferrara et al. 1996). The valley has an irregular triangular form with its base to the west (see Figure 28.1). It is confined by the Basento and Cavone basins to the north, the Sele to the west, and the Sinni and Noce to the south. Orographically it is delimited by the Volturino massif, Monte Viggiano, Monte Raparo and Monte Sirino. Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
386
Mediterranean Desertification
The valley, from structural, evolutionary and environmental points of view, consists of distinct geographical identities. On the basis of environmental and physical characteristics the area can be divided into three large sub-areas whose features have also affected the forestry evolution: the Upper, Middle and Lower Val d’Agri (Table 28.1). The mountain landscape of the upper part of the Agri valley is formed by limestone and dolomite where herbaceous vegetation and tree cover limit erosion by water despite the steep slopes. A milder landscape is found in the middle region with flysch and sandstone soils. Here, water erosion and landslides are more frequent because of the clay marly nature of the substrata which have become exposed following extensive deforestation. Towards the sea, calanchi badlands, which are areas of extreme environmental sensitivity (Basso et al. 1997, 2000; Kosmas et al. 1999), are found on the hilly Pliocene soils (Figure 28.2). These are succeeded by large sandy and pebble marine surfaces strongly incised by transverse valleys in which residues of riparian vegetation exist. From a climatic point of view the highest rainfall is found in the south-west part of the valley and is due, in part, to its W–E orientation and, in part, to the higher altitudes found there (Cantore et al. Table 28.1 Main structural parameters of the Agri Valley
Parameters
Upper
No. of municipalities Area (ha)a 1991 populationb Population density (inhabitants ha−1 )b Mean altitude (m a.s.l.) a b
12 59 824 46 059 0.57 900
Middle
Lower
17 91 126 36 978 0.32 600
Total Agri Valley
6 21 718 38 179 0.67 250
35 172 668 121 216 0.47 –
Falling within the Agri hydrographic basin. Referring to the administrative area (ISTAT 1991).
Lauretum (hot) Lauretum (intermediate) Lauretum (cold) Castanetum Fagetum
49
Naples
48
47
46
445 Coordinates: UTM 0 56
57
58
59
60
61
62
10 km 63
64
65
Figure 28.1 Agri Valley forest phytoclimatic classification, based on Pavari (Cantore et al. 1987)
Aspects of Forestry in the Agri Environment
387
Figure 28.2 Calanchi badlands, showing a deeply incised drainage system in the underlying silty clays, a common feature of the Middle Agri Valley
1987). This area, which corresponds to the Lagonegrese and Thyrrenic slope, is influenced by humid wind currents coming directly from the Atlantic Ocean (Bonin 1978) producing a mean annual rainfall that can reach 2000 mm. The mean annual rainfall in the Middle and Lower Val d’Agri decreases with altitude to the Ionian coastal band where minimum values of 260 mm year−1 are common. This area is identified by a strict Mediterranean pluviometric regime, having an accentuated seasonal distribution with high pressure common in winter everywhere except near the Ionian coast where it peaks in autumn. Concerning temperature, the thermal gradient increases from west to east along the basin, and the entire valley lies between the 8 ◦ C (Monte Volturino) and 16 ◦ C isotherms (Montalbano Jonico). The sea also influences this gradient because, despite Monte Volturino being 75 km away from the Ionian Sea, it is only 40 km from the Thyrrenian Sea. The forest physiognomy of the area can be well defined using the forest phytoclimatic classification proposed by Pavari (Cantore et al. 1987), which is based on a comprehensive set of parameters: annual mean temperatures; mean temperature of the coldest and hottest months; minimum and maximum mean annual temperatures; rainfall distribution; annual rainfall; and summer rainfall. As shown in Figure 28.1, the basin is characterized by a rapid altitudinal sequence of the forest phytoclimatic zones from east to west: the Lauretum hot subzone in the coastal and subcoastal areas in Lower Val d’Agri; the intermediate and cold Lauretum subzone extending up to the Middle Val d’Agri with a deep intrusion of the cold subzone between S. Arcangelo, Tursi, and the higher part of the valley near Marsiconuovo. Finally, the higher parts of the valley can be assigned to the Castanetum zone, and the highest altitudes to the Fagetum zone.
3 THE FOREST ENVIRONMENT Unlike other parts of Basilicata where, over the years, the forests have been considerably degraded, the Val d’Agri still possesses interesting and large forest stands. The stands are mainly localized in the middle and upper zones, especially where steeper slopes have discouraged conversion to farmland. A forest can only be considered well preserved when an adequate canopy density is maintained as this is a fundamental factor of its functionality in a Mediterranean environment. A reduction in canopy density, below critical thresholds, has often given way to degradation phenomena through
388
Mediterranean Desertification
steppe development processes and increased water and wind erosion which expose bedrock in a short time. The Middle and Upper Val d’Agri still have large forests with Fagus sylvatica (common beech); Castanea sativa (sweet chestnut); Quercus cerris (turkey oak); and Quercus frainetto (Hungarian oak). Minor species are more widely diffused: Acer campestre (field maple); Acer monspessulanum (Montpellier maple); Carpinus betulus (hornbeam); Alnus cordata (Italian alder); Corylus avellana (common hazel); Prunus spp. (cherries, etc.); and Malus sylvestris (crab apple). Some extremely important residual formations of the once widely distributed virgin Abies alba (common silver fir) forest exists, amongst which the Laurenzana forest is a very famous and characteristic example. The large pure beech (Fagus sylvatica) forests are found at the higher altitudes where grand old trees testify to the past application of the Borbonic Law of 21 August 1826, which prescribed clear-cutting with seed trees. These extensive beech forests are found on the slopes of Monte Sirino and Monte Volturino, and near the springs of Sella Cessuta in Moliterno, with old standards 30 m high mixed with Acer obtusatum (Neapolitan maple) and Ilex aquifolium (holly). Other interesting, mature, high beech stands are also found on the main chain surrounding Marsiconuovo where the Agri River rises (Figure 28.3). Grazing lands and bare surfaces are found at higher altitudes fringed with pure beech forests or mixed with Alnus cordata, Ilex aquifolium and Taxus baccata (common yew). This latter species is mainly spread over the western slopes and is clearly influenced by the winds from the Atlantic. It could be said that the beech forest has been pushed down the mountain by the high mountain climate favouring Astragalus spp. and by humans requiring an increased area of mountain pasture. Downslope, mixed oak forests correspond to the transition into the lower climate band. These forests, of which the Lata forest is a typical example, can be extensive, consisting of purely, or predominantly, Quercus cerris. At these altitudes the high forests gradually give way to coppices of Quercus cerris which are often associated with Acer spp., Ostrya carpinifolia (European hophornbeam) and Quercus pubescens (downy oak). Along streams and in humid areas these are replaced by Salix spp. (willows), Acer spp., Alnus cordata and Populus nigra (black poplar). There has also been extensive reforestation with Pinus nigra (Corsican pine), Pseudotsuga menziesii (Douglas fir),
Figure 28.3 High stand beech forests at the source of the Agri River
389
Aspects of Forestry in the Agri Environment
Pinus radiata (Monterey pine) and, rarely, Cedrus spp. (cedars), Abies alba and Picea abies (Norway spruce) on the northern limestone slopes of the Middle and Upper Val d’Agri. The lower part of the valley, down to the sea, is characterized by less consistent, discontinuous and impoverished forest areas essentially dispersed amongst farmland. They have often been degraded by grazing and burning and mainly consist of Quercus pubescens and Quercus ilex (holm oak) coppices. In this part of the valley forests are often reduced to small patches or to vestigial vegetation groups. Coppices, when too quickly rotated or subject to uncontrolled grazing, exhibit degradation phenomena due to soil compaction and to a deterioration in their species composition. Conifer forests, which are sporadically present and have all been planted, are mainly represented by Pinus halepensis (Aleppo pine), Pinus pinea (stone pine), and Cupressus arizonica (Arizona cypress) stands, sometimes mixed with Eucalyptus spp. (Figure 28.4). Using the concept of a “vegetation belt” (sensu Famiglietti and Schmidt (1969) in which a belt contains species having equal or similar ranges in their distribution, both horizontally and vertically, i.e. species with similar climatic and edaphic needs), the following “belts” can be distinguished in the Agri Basin: • Evergreen Mediterranean oaks: Quercus ilex (Q. ilex) typically having an understorey with subtropical characteristics; more than 200 species can be assigned to this zone. • Deciduous oaks: Quercus pubescens (Q. pub.) occupies the zone between the Quercus ilex band and the superior mixed broadleaf band (QTA); the ecological forms are markedly xeric and typically Mediterranean. Mountain vegetation and forests are determined by the climate; rainy and cold winters with dry summers. It is by far the richest belt with over 600 species in the Basilicata region. • Deciduous mixed forests: Quercus–Tilia–Acer (QTA) is typical of temperate zones. It has fewer species than the FA belt immediately above it because of the effects of long-term anthropic forces. • Beech and silver fir: Fagus–Abies (FA) marks the upper boundary of the forest formations in the area. It includes species needing a humid and cool climate where, locally, mixed beech and European silver fir forests find optimal conditions intermixed with evergreens such as Ilex aquifolium, Taxus baccata, Lonicera spp. (honeysuckle) and Dafne laureola. The land above this belt is dominated by summit grasslands and rocky meadows which belong to the belt named “Mediterranean mountain grasslands”.
48
47
46
445 Coordinates: UTM 0 56
57
58
59
60
61
62
63
10 km 64
Figure 28.4 Landsat satellite image of the Agri Basin (darker areas indicate dense vegetation and lighter areas are almost bare soil or rock)
390
Mediterranean Desertification
•
Mediterranean mountain grasslands: includes short-lived grassland plants capable of surviving dry summers and cold winters; small shrubs, autumn germinating, short stem plants, often with a rosette of leaves and having the typical adaptations of plants growing on rocky and shallow soils. In some of these zones these species are accompanied by others belonging to a rare belt called “Vaccinium and Loiseleuria” (VL).
4
PHYSIOGNOMY AND STRUCTURE OF THE FORESTS
4.1
Qualitative Aspects
The consistency of forest vegetation, its typology and physiognomy, are strongly influenced by anthropic activities. In Basilicata, unchecked exploitation and periodical and partial deforestation have considerably disrupted the forest environments. The seriousness of anthropic pressure is alarming. From 1872 to 1968 more than 25% of the forest area in the Upper Val d’Agri was lost. Currently, the most important forest typologies found in the valley are Quercus cerris and Quercus pubescens forests (high forests and coppices), Fagus sylvatica forests (high forests and coppices), mixed mesophile forests with Quercus cerris prevalent (coppices), Castanea sativa forests (coppices) and conifer reforestation (Mediterranean and mountain pine). Quercus cerris, accompanied by Quercus frainetto in favourable conditions, are the most extensive species and can be considered as a xerophile accentuation of mixed forests which originated from mesophile species and were induced by anthropic actions in the form of repeated and intense felling, which was often alternated with grazing. Under these conditions, Q. cerris and Q. frainetto have prevailed over the more mesophile species such as Acer spp. and Tilia spp. which encouraged a well-balanced association rich in species composition and a good structure that was particularly well stratified. The present oak forests can be considered a paraclimax plateau with more basic and adaptable species, dominated by Quercus cerris. In the upper part of the valley, high forests dominate which are mainly managed under a uniform shelterwood system (Figure 28.5). In other parts of the valley, high forests have been replaced with coppices with dense stands (100–150 per hectare) that
Figure 28.5 Quercus cerris forests are widespread throughout the basin
Aspects of Forestry in the Agri Environment
391
are rotated every 15–18 years. These coppices are frequently subject to intensive grazing which has a deleterious effect on species composition and leads to a decrease in stand efficiency. The Quercus pubescens formations can also be interpreted as a xerophile alteration of previously mixed associations that tended, in that area, to develop only into bushes or stunted coppices. As before, the maintained stand density is high (100–150 standards per hectare) and rotations are from 15 to 18 years. The Fagus sylvatica forests, dislocated from large residential areas, fall into two distinct typologies: either pure, or almost pure, forests in cool locations at high elevations; or forests where beech prevails, or is predominant, at less favourable altitudes where the forest is in transition towards the oak belt. In this second group the species are clearly less shade tolerant and, on clay soils, are fiercely competitive. Maples are frequent in the contact belt and, in particular, Acer lobelii (Lobel’s Maple) has an important role as it is the only exclusively indigenous element endemic to the central–southern Apennine chain and is recognized in the association Acer lobelii–Fagetum. At the lower margin of the beech forests we can also find residual stands of Abies alba of which the Laurenzana high forest is one of the most important examples in all the central–southern Apennine chain. Anthropic pressures have also blighted some species in these formations: Acer pseudoplatanus (Sycamore), Alnus cordata, Tilia platyphyllos (large-leafed linden), Ulmus montana (wych elm) once characterized the primitive mixed forests making them more mesophile. The almost total disappearance of Abies alba from the mixed mesophile association can be attributed to excessive and selective human exploitation. With regard to the silvicultural systems, high forests with uniform shelterwood systems prevail in this area as well, but the beech forests in the less favourable and transition zones near the upper limits of the forest cannot be ascribed to any defined silvicultural system. Of particular note are the chestnut coppices which are mainly used to produce vineyard posts. They are found in the upper part of the valley and cultivated in 12-year rotations. Lastly, it should be noted that reforestation in the Val d’Agri is mainly associated with the forest management plan arising from the Zanardelli Law of 31 March 1904, which promoted a large reforestation programme in the southern regions. From 1911 to 1927, about 570 ha of the basin were reforested as a result of this law, representing about 10% of regional reforestation. Few of these forests remain today and those that are left have been severely damaged by fire and grazing. In the upper part of the Val d’Agri they can be found at, for example, Marsiconuovo (Tumulo, 60 ha), Tramutola (Monticello, 40 ha) and Viggiano (Monte di Viggiano, 10 ha). These reforestations were carried out mainly with Pinus nigra, Abies alba, Pinus halepensis and Cupressus sempervirens (Italian cypress), according to local climatic conditions and altitude. Following a further law, dated 10 August 1950, which promoted a huge programme of public works in the so-called “Mezzogiorno of Italy” (southern regions plus the islands of Sicily and Sardinia), with the purpose of re-animating the local economy, parts of the upper basin were again reforested, mainly on municipal land using government funds. Primarily, reforestation was to improve the level of employment and for soil preservation on bare soils using terraces (gradoni or small bench terraces) and lay-byes on lime soils and in strips on clay soils. In the first case, 1800–2000 linear metres per hectare of terraces were used along the contour lines, having a mean width of 70 cm and a 40 cm working depth, inclined towards the mountain. In the second case, the bands were 50 cm wide and 40 cm deep, having a 1.5–2.0% slope towards the valley. Different species were used; the trees were mainly deciduous and predominantly beech. The beech plants used were either natural wildings or seedlings transplanted from nurseries. Oak trees were planted using direct seeding at 1500–2000 kg of acorns per hectare and using mainly local species, though Quercus rubra (red oak) was sometimes planted. Good results, in terms of rooting, have been obtained using Alnus cordata transplants and/or seedlings (S1 and T1) as well as other broadleaf species. Some exotic conifers also gave very good results: Pinus radiata, Cupressus arizonica and Cedrus atlantica (Atlas or Algerian cedar). Planting was often at a very high density: up to 2500 plants per hectare were used at altitudes lower than 1000 m, mainly using Pinus halepensis (40%) and broad-leaved species (27%); above 1200 m, a lower density of around 2000 plants per hectare was adopted using about 36% Pinus nigra and 18.5% Alnus cordata with a significant number of Pinus sylvestris. This last species was introduced as a substitute for Pinus nigra when it was realized
392
Mediterranean Desertification
that P. nigra was experiencing grave problems as indicated by a widespread forest decline, and it was also particularly susceptible to attack by insects. 4.2
Quantitative Aspects
Table 28.2 summarizes the present forest situation in the Val d’Agri as indicated by the main forest parameters. Analysis of the data in Tables 28.1 and 28.2 shows that the percentages of forest areas are not highly related to those municipalities having the least favourable ratio of land area to residents. Over the years 1973–1986 the forest index increased where there has been a decrease, or only a slight increase, in population (Figure 28.6(a)). As would be expected, the municipalities located in the lower valley had the greatest opportunities for production and employment and, correspondingly, an increase in population and a similar increase in farmed land which resulted in a concurrent decrease in forest area. The forest index is also influenced by altitude: low values are found in those municipalities where land use is mainly orientated to farm use, but on the other hand, higher values are found in the municipalities at higher altitudes where the high forests are found (Figure 28.6(b)). The forest index, in relation to the silvicultural system, is slightly less well correlated with the total values. Table 28.2 Main forest parameters
Parameters
Upper valley
Middle valley
Lower valley
Total
Coniferous high forests (ha) Broad-leaved high forests (ha) Mixed high forests (ha) Total high forests (ha) Total coppices (ha) Total forests (ha) Forest indexa (%) Change in forest index 1973–1986 (%) Surface damaged by fires (mean 1993–1994) (ha)
3100 3800 700 7600 7300 14 900 25 +13 232
1700 7600 1500 10 800 5600 16 400 18 +3 642
2000 400 100 2500 400 2900 13 +5 87
6800 11 800 2300 20 900 13 300 34 200 20 +7 961
12 10 8 6 4 2 0 −2 −4 −6 −8 −20
(b)
−10
0
10
20
Population change (%)
30
Forest index change 1986 (%)
(a)
Forest index change (%)
Data refer to the Agri hydrographic basin, derived from reprocessing MiPAF data (Ministero delle Politiche Agricole e Forestali) 1973, 1986, 1993, 1994. Data rounded to 100 ha. a Forest index = % forest area relative to total area 50 45 40 35 30 25 20 15 10 5 0 0
1500 500 1000 Prevailing altitude (m a.s.l.)
Figure 28.6 (a) Forest index change versus population change (1973–1986) and (b) forest index versus altitude (data refer to 1986)
Aspects of Forestry in the Agri Environment
393
From Tables 28.1 and 28.2 it is possible to see that, in the case of the high forests, the forest index is quite constant with altitude, while with the coppices, the forest index decreases significantly with altitude. Fires, in terms of both frequency of occurrence and areal extent, are a serious problem in an environment that is already subject to degradation and desertification phenomena. They follow a clear pattern which is found in many other Mediterranean regions and certain conclusions can be drawn: the highest number of fires are found in areas having the lowest forest index; in the Val d’Agri they occur in areas having indices up to 20%, corresponding, seemingly, to those areas dominated by coppices.
5 PROBLEMS AND PERSPECTIVES OF INTERVENTION: A HYPOTHESIS There are two important issues that face the Val d’Agri. First, a rationalization of management systems in the upper part of the territory where forestry is practised and where the forest represents an important economical resource is necessary. Various forest management systems which are adapted to the particular local environments of the region can be adopted owing to its extensive, mainly publicly owned forests, the forest physiognomy and species composition, and, in general, the high fertility. Second, the lower and middle parts of the valley, where forest cover is already less widespread and the climatic conditions less favourable, face a slow and continuous reduction in forest area. Forest is being replaced by farmland and is therefore becoming limited to marginal areas. Table 28.3 shows the fundamental characteristics of the Val d’Agri forests. The approach must depend on the particular characteristics of the local environment. Options include forest management improvement; increasing stand density; introduction of indigenous species in over-simplified or degraded artificial ecosystems; conversion of coppices to high forests; increasing coppice rotation cycles; grazing regulation; and fire control (prevention and protection against fires). Forest management includes many different initiatives of which one of the most urgent is to eliminate the high forests managed under the even-aged uniform shelterwood system. These particular forests are widespread and are derived from the extensive felling carried out in the past, over a large part of the land. Examples can be found in the municipality of Marsico Nuovo. The new forest management systems stress the importance of biodiversity. They should promote a gradual substitution of even-aged stands with stable and functional uneven-aged stands (either strictly or in groups or strips). Particular attention should also be made to protect accessory species of mixed stands which have been often sacrificed, or completely eliminated, by silvicultural practices which favoured a privileged principal species. Many of the coppices in the area (mainly Fagus sylvatica and Quercus cerris) could easily be converted into high forests and, where environmental pressures have outweighed economical ones, this has already started to happen. The main focus in the middle and lower parts of the valley should be on recovering the degraded forests. Here, thinning, which improves forest function, and timing are important, especially in cultivated forests where delays in cutting can compromise stand survival. In this sense, regeneration and intermediate and final cutting should be carried out regularly to promote stand “renaturalization”. This gives the forest optimum conditions for functional stability and improvement of its diversity and species composition. Particular attention should also be paid to fire control. Fire is a major disruptive force in rural communities and a key agent of change. It is insufficient to install fire prevention systems, even if they are technologically advanced. A well-defined programme of co-ordinated actions should be planned, including care of the tree crop as mentioned above, in a management framework for fire prevention. These guidelines cover the general situation, but they can be translated into more precise options to develop specific interventions creating a fundamental instrument to ensure that a correct balance between intervention and the needs of the environment is regained. Furthermore, this also guarantees the integration, rationalization and co-ordination of policies, funded from different sources, in the most beneficial manner.
394
Mediterranean Desertification
Table 28.3 Fundamental characteristics of the Val d’Agri forests
Upper Canopy cover
Structure
Main species
Forest state
Ownership/land tenure Particular problems Suggested intervention
Middle
Lower
Continuous and in large Discontinuous, with Low and units, with clear some large units where discontinuous in increase from east to the canopy is closed residual forests west Even-aged high forest, High forests at higher Residue forests, often old-aged, altitudes bands, isolated managed by a uniform groups, riverbank shelterwood system forests, forest on rocky outcrops, Mediterranean macchia, reforestation mainly with conifers Beech, Turkey oak, Turkey oak, Italian oak, Turkey oak, Holm Italian oak, maple, evergreen oak, black oak, Mediterranean black pine, chestnut pine, Aleppo pine macchia, Aleppo pine, cypress, eucalyptus, honey locust Lack of final felling, Lack of intervention, Abandonment, spontaneous forest ageing, conflicts with ageing, conflicts advancement grazing, fires with grazing Private < public Private = public Private > public Spring control, grazing regime, depopulation Management plans, recovery, re-naturalization, conversion from coppices to high stands
Grazing, fire Coppice rationalization: lengthening rotations, grazing control, fire control
Fire, environmental problems Re-naturalization, recovery, fire control, thinning
REFERENCES Basso F, Bellotti A and De Natale F, Ferrara A and Pisante M (1997) Analisi del rischio di degradazione del suolo in aree agricole della Basilicata: una proposta metodologica. Rivista di Agronomia XXXI(3), 864–871. Basso F, Bove E, Dumontet S, Ferrara A, Pisante M, Quaranta G and Taberner M (2000) Evaluating environmental sensitivity at the basin scale through the use of geographic information systems and remote sensed data: an example covering the Agri basin (southern Italy). Catena 40, 19–35. Bonin G (1978) Contribution a la connaissance de la vegetation des montagnes de l’Apennin centro-meridional. These de Docteur-des-sciences, Univ. de Droit, d’Economie et des Sciences Aix-Marseille III. Cantore V, Iovino F and Pontecorvo G (1987) Aspetti climatici e zone fitoclimatiche della Basilicata. Annali dell’Istituto di Ecol, e Idrologia Forestale, Public N 2, Cosenza. Famiglietti A and Schmidt E (1969) Fitocenosi forestali e fasce di vegetazione dell’Appennino Lucano centrale (gruppo del Volturino e zone contermini). Annali del Centro di Economia Montana delle Venezie, vol. VII, CEDAM, Padova.
Aspects of Forestry in the Agri Environment
395
Ferrara A, Bellotti A, Faretta S, Mancino G and Taberner M (1996) The Agri Basin Environment. Website: http//:www.unibas.it/agrimed/ ISTAT (1991) Annuario Statistico Italiano. ISTAT, Rome. Kosmas C, Ferrara A, Briasouli H and Imeson A (1999) Methodology for mapping Environmentally Sensitive Areas (ESAs) to desertification. In C Kosmas, M Kirkby and N Geeson (eds) The Medalus Project: Mediterranean Desertification and Land Use. European Union, pp. 3147. Mainguet M (1994) Desertification: Natural Background and Human Mismanagement . Springer-Verlag, Berlin.
29
Modelling Large Basin Hydrology and Sediment Yield with Sparse Data: The Agri Basin, Southern Italy
J.C. BATHURST,1 J. SHEFFIELD,2 C. VICENTE,3 S.M. WHITE4 AND N. ROMANO5 1
Water Resource Systems Research Laboratory, School of Civil Engineering and Geosciences, University of Newcastle upon Tyne, Newcastle upon Tyne, UK 2 Department of Civil and Environmental Engineering, Princeton University, Princeton, New Jersey, USA 3 C/Cafetos #4, Col. Campestre, Cordoba, Veracruz, Mexico 4 Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire, MK45 4DT, UK 5 Istituto di Idraulica Agraria, Universita` di Napoli ‘‘Federico II’’, Portici, Italy
1 INTRODUCTION Through its programmes of research into desertification in the Mediterranean Region, the European Commission has sought to develop a thorough understanding of the desertification phenomenon and to provide guidelines for protection, management and rehabilitation. The MEDALUS project has taken on these tasks in part through an emphasis on large river basin scale studies and mathematical modelling. Models provide an important means of integrating the knowledge obtained from experimental studies, of developing an understanding of basin response mechanisms and of supporting the decision-making process for land and water management. In particular, physically based models provide the means for predicting the impacts of possible future changes in land use and climate, and thence for adopting appropriate measures for protection, management and rehabilitation. The large basin scale approaches the regional scale for which management strategies and planning decisions are relevant, while still forming a well-defined hydrological unit. As its contribution to MEDALUS, the Water Resource Systems Research Laboratory (WRSRL), at the University of Newcastle upon Tyne in the UK, applied its SHETRAN hydrological and erosion modelling system to the upper 1532 km2 of the 1700-km2 Agri Basin in southern Italy, one of the MEDALUS focus basins. There have been few tests of physically based catchment models at such scales (Refsgaard et al. (1992) report an example) and the study therefore had the following aims: 1. Extension of the tested SHETRAN scale of application beyond the 700 km2 already modelled in the MEDALUS project (Bathurst et al. 1996). 2. Exploration of SHETRAN applicability and limitations, including model parameter evaluation, at the scale of 1000–2000 km2 and with a simulation grid scale (2 km × 2 km) at the limit of the model’s physical basis. 3. Exploration of the contribution that modelling can make to planning at the large basin scale. The Agri Basin was chosen by the MEDALUS project as a focus for study because of its state of incipient desertification, which could degenerate into serious land degradation without the appropriate control measures. However, as is usually the case except in small intensively monitored research Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
398
Mediterranean Desertification
basins, many of the data that would ideally be used for model validation were not available. In particular, there was no measured river discharge record. A further aim of SHETRAN’s application to the Agri Basin therefore became a demonstration of model use with sparse data, especially quantification of the associated uncertainty in model output. Aspects of the application described here include data assembly, definition of uncertainty limits, and the hydrological and sediment yield simulations.
2 2.1
SHETRAN SHETRAN Description
SHETRAN is a physically based, spatially distributed modelling system for water flow, sediment transport and contaminant migration, applicable at the scale of the river basin (Ewen 1995; Ewen et al. 2000). It incorporates the major elements of the land phase of the hydrological cycle (interception, evapotranspiration, snowmelt, overland and channel flow, unsaturated and saturated zone flow) and the sediment transport component accounts for soil erosion by raindrop impact and overland flow, and transport by overland flow and channel flow (Bathurst et al. 1995; Wicks and Bathurst 1996). Each of the processes is modelled either by finite-difference representations of the partial differential equations of mass and energy conservation or by empirical equations derived from independent experimental research. The spatial distribution of catchment properties, rainfall input and hydrological response is achieved in the horizontal direction through the representation of the catchment by an orthogonal grid network and in the vertical direction by a column of horizontal layers at each grid square. SHETRAN is continually evolving as new process descriptions and solution schemes are introduced. The version used here (v3.4.2) is distinguished by improved numerical stability in the surface flow calculations relative to earlier versions and by the representation of the subsurface as a one-dimensional (vertical flow) unsaturated zone overlying a two-dimensional (lateral flow) saturated zone. 2.2
SHETRAN Parameters
Within each model grid square, each physical characteristic is represented by one parameter value. As long as the grid square is small compared with the distances over which there is significant spatial variability in basin properties and hydrological response, this does not compromise the model’s ability to represent local variations in response. However, as grid scales increase, the local spatial variability in properties and response becomes subgrid. There are then difficulties in applying the equations of small-scale physics which make up SHETRAN and evaluating their parameters, at the grid scale (e.g. Beven 1989). In particular, the field measurements that form the basis of parameter evaluation are most easily carried out at the point or plot scale, which may not be representative of the large grid scales (of the order of 1 km) used in modelling river basins. The solution has been to use “effective” parameter values, which represent the subgrid spatial variability, to give a grid scale response. However, this is a pragmatic approach and it is recognized that the concept may not allow an accurate reproduction of the observed response in all circumstances (as shown for example by Binley et al. 1989). The principal soil parameters and functions in SHETRAN are the soil depth, the saturated zone conductivity (for lateral flow), the saturated values of conductivity and moisture content for the unsaturated zone (for vertical flow) and the water retention (moisture content–tension) and moisture content–conductivity relationships for the unsaturated zone. These characteristics do not vary through a simulation. The proportion of ground covered by vegetation at the grid scale (i.e. the proportion of the grid square that is not bare soil) is accounted for by a proportional index on a scale from 0 to 1 and can be varied in a predetermined manner through the simulation. The vegetation parameters are canopy drainage and storage terms used in calculating interception, properties affecting evapotranspiration, and root distribution, and are mostly time invariant. Overland flow resistance is quantified by the Strickler resistance coefficient (the reciprocal of the Manning coefficient). The coefficient is specified by the modeller, usually according to land use, and does not vary through the simulation.
Modelling Large Basin Hydrology and Sediment Yield
399
The ease with which the soil can be eroded is quantified by two coefficients, representing raindrop impact erodibility and overland flow erodibility respectively. A coefficient is also used to represent channel bank erodibility. These coefficients cannot yet be determined directly from measurable soil properties and therefore require calibration. Topographic elevations are determined from appropriate maps or Digital Elevation Models. Channel characteristics are quantified in terms of the channel cross-sectional shape, elevation and Strickler resistance coefficient. All the above parameters and functions are spatially variable between grid squares (or channel links as appropriate), as are also the specified time-varying rainfall, the specified time-varying meteorological variables determining potential evapotranspiration, and the simulated time-varying hydrological responses. 2.3 Uncertainty in Model Output
An important constraint on the accuracy of physically based, distributed modelling is the uncertainty surrounding the evaluation of the model parameters, which arises from scale effects, sparse field data and the scope for multiple calibrations of apparently equal validity (Beven and Binley 1992). Consideration of this uncertainty is therefore increasingly acknowledged to be an important feature of the modelling process (e.g. Beven and Binley 1992; Ewen and Parkin 1996; Quinton 1997). In particular, uncertainty in model parameter evaluation needs quantification as a basis for evaluating uncertainty in model output. For the Agri simulations, the following method, based on Ewen and Parkin (1996), was applied. Using hydrological judgement, upper and lower bounds are set on the values of the model parameters, reflecting uncertainty in the values. A series of simulations is carried out so that each parameter takes the different values assigned to it. The number of simulations depends on the number of parameters involved, the number of values assigned to each parameter and the number of combinations of different parameter values investigated. The various simulation outputs are then superimposed on each other and the overall time series of maximum and minimum output bounds extracted. These bounds may be composed of contributions from several of the simulation outputs. The bounds on the model parameters thus translate into bounds on the model output and conclusions on model performance are drawn according to the width of the resulting output envelope and the degree to which it contains the measured data.
3 FOCUS BASIN STUDY 3.1 Data Requirements and Data Assembly
The data required for validation of SHETRAN are the property data which describe the catchment (topography, soils and vegetation), the time series of input data which drive the simulation (rainfall and potential evapotranspiration), and the time series of hydrological variables which provide the basis for comparing observed and simulated basin response (e.g. channel discharges and groundwater levels). As far as possible the information should be available on a spatially distributed basis while rainfall and channel discharge should also be available as continuous records (or at least as hourly averages). Assembling and processing such a demanding data set is a time-consuming and intensive activity. Initially contacts have to be developed with the relevant data collection agencies and with local hydrologists familiar with the response of the basin in question. An iterative procedure then unfolds in which data are assembled from the various agencies, the overall availability of data is examined (including coincidence of the meteorological input and hydrological variable time series) and eventually a simulation period is selected. Often this period represents a compromise between availability of data and other requirements such as a particularly interesting sequence of hydrological conditions. Although data are now typically logged in electronic form, many past data records are still provided as hard copy (e.g. raingauge charts and topographic maps) and the next stage is therefore to digitize them for use on a computer. This can take several weeks, even months. Following this the data are checked for quality (e.g. discrepancies in time series data from different sources), and gaps in the time series records are filled (usually by a correlation procedure). Many of the data then need
400
Mediterranean Desertification
processing to put them in a form that has meaning for the model application. For example, rainfall data are often provided as daily values whereas simulation of the dynamic processes responsible for runoff requires the data to be input at hourly or even smaller intervals. Temporal disaggregation of the daily values, usually by comparison with the continuous record from the nearest autographic gauge, is therefore carried out. Finally the data are converted to the format specified in the model software for computer processing. Undertaking this task for the Agri Basin was a major operation which involved several MEDALUS groups (see the Acknowledgements). 3.2
Agri Basin Data Assembly and Processing
The simulation area consists of the 1532-km2 basin upstream of the Gannano barrage (an irrigation offtake structure) (Figure 29.1). It was represented by 383 grid squares of dimensions 2 km × 2 km, this being the finest possible resolution bearing in mind SHETRAN’s current computation requirements (Figure 29.2). The simulation period is 1985–1988 inclusive, which offers the maximum combined availability of autographic rainfall data and reservoir water balance data (used to generate a channel discharge record).
Topographic and Channel Data Topographic elevations in the basin range from 104 m to 1976 m. Contour data for the basin (at 25-m intervals on a 1:25 000 scale map) were obtained from the Istituto Geografico Militare (Italiano) and converted into a 50-m resolution Digital Elevation Model (DEM) by the MEDALUS project. This was then used to derive the model grid network and the elevations characterizing each square (Figure 29.2). The DEM was also used to generate a basin wetness index map from which the channel system was derived. Channel dimensions were obtained by interpolation between the headwaters and outlet (based on the number of upstream channel links at each link) and checked against a number of field measurements. Strickler resistance coefficients for channel and overland flow were evaluated on the basis of past experience with SHETRAN simulations and data in the literature (e.g. Engman 1986; Wicks et al. 1992).
Soil Property Data SHETRAN requires a soil distribution map and values for the soil parameters (section 2.2). However, there is no national soils map of Italy. On the basis of a 1:150 000 scale lithology map, soil type N ITALY
Pertusillo
Gannano
10 km
Figure 29.1 Location map of the Agri Basin
401
Modelling Large Basin Hydrology and Sediment Yield
Elevation (m)
1 2
Channel network
Outlets: 1 Gannano, 2 Pertusillo
100− 300 300− 500 500− 700 700−900 900−1100 1100−1300 1300−1500 1500−1700
Grid resolution = 2000 m x 2000 m
Figure 29.2 SHETRAN grid network, channel system and elevation distribution for the Agri Basin. Grid spacing is 2 km
was therefore determined according to the distribution of the three principal rock types in the basin: flysch, sandstone and limestone. Geostatistical analysis of soil physical and hydraulic property data collected on a hillslope in the basin indicate a correlation scale of 0.4–1.3 km (Santini et al. 1996; Romano and Santini 1997). While ideally this scale would define the maximum distance between evaluations of the model soil parameters, it is reasonably similar to the 2-km grid scale actually used. At each model grid square, the soil column was divided into a topsoil layer and an underlying (fractured) rock layer. A representative, albeit approximate, size distribution was then determined for the area of each lithology by the MEDALUS project, from which hydraulic property data in the topsoil (including retention curves and porosities) were derived using the Brooks and Corey (1964) formulation. The topsoil saturated conductivity was set according to the underlying rock type: 0.0005 m day−1 for soil over flysch, and 2 m day−1 for soil over sandstone and limestone. For the fractured rock layer, the required hydraulic property data were taken from the literature (e.g. Dunne and Leopold 1978; Rawls and Brakensiek 1989; Bras 1990). The overall soil column thickness was set at 10 m throughout the basin, on the basis of a survey of well depths (made with the assistance of the local MEDALUS groups) which indicated a range of thicknesses between 2 and 20 m. Topsoil thicknesses were distributed according to approximate data in the guidebook attached to a 1:100 000 scale land systems map, with magnitudes of 0.05, 0.1, 0.25, 0.5, 0.75 and 1.25 m. The approximate nature of the evaluations was a major source of uncertainty in the simulations.
Vegetation Property Data SHETRAN requires a vegetation distribution map and values for the vegetation parameters (section 2.2). Through the MEDALUS project, a 50-m resolution land-use map was digitized from analysis of multitemporal Landsat Thematic Mapper images. Nine land uses were specified: open water, bare soil, rock, urban, field crops (without distinguishing between different crops), pasture, macchia, deciduous woodland and coniferous woodland. The map was aggregated to a 2-km resolution for use with SHETRAN and for this scale the open water and urban classifications were removed. The main crops are durum wheat (with some maize grown with irrigation) and, in the valleys, fruit and vegetables. Overall, 60% of the basin is arable (30% seedbed, 10% tree crops, including poplar, and 20% permanent pasture), 20% is woodland and 20% is under other uses. Parameter values
402
Mediterranean Desertification
for each vegetation type (relevant to the evapotranspiration and soil erosion calculations) were based on previous applications of SHETRAN in Mediterranean Basins, data in the literature and the plant parameter values from an agricultural research station managed jointly by the Universities of Basilicata, Naples and Bari at Corleto Perticara, inside the basin (e.g. Lukey et al. 1995; Bathurst et al. 1996).
Rainfall Fifteen daily raingauge records were used to represent the basin above the Gannano barrage. In addition, autographic raingauge charts were obtained and digitized for five stations (Castelsaraceno, Nova Siri Scalo, Guardia Perticara, Tramutola and Senise). The autographic records were then used to disaggregate the totals for the daily gauges to hourly amounts, matching each daily gauge to the nearest autographic gauge. In this way a basin record of 11 years (1978–1988) of hourly rainfall was generated for the 15 gauges, whose areal coverages were determined using Thiessen polygons. (Only the period August 1983–December 1988 was actually used in the simulations.) An example of the time series is shown in Figure 29.3. An excellent spatial and temporal distribution was therefore obtained for the principal model input. Generally, rainfall records for the Agri Basin extend over about 70 years. Annual rainfall varies from 530 mm at the coast to 1100 mm in the mountains. However, the distribution of rainfall through time is erratic and uneven, with daily rainfalls in excess of 200 mm occasionally recorded in some areas. Snow may fall on the higher mountains in winter. Further details of the rainfall regime are given in Mazzanti et al. (1998).
Evaporation Daily pan evaporation data and automatic weather station data for calculating evaporation by the Penman formula were available for the simulation period for an experimental site at Policoro, downstream of the Gannano barrage. In addition, daily temperature data were available for the simulation period for this site, the agricultural research station at Corleto Perticara (within the basin) and at the Pertusillo dam. Daily potential evapotranspiration was therefore generated from the daily mean temperature measurements at the three sites using the Blaney–Criddle formula: PE = p (0.46 T + 8.13)
(1)
where PE is daily potential evapotranspiration (mm day−1 ); p is the percentage of the annual hours of daylight each day, expressed as a mean daily value for each month (%); and T is mean daily temperature (◦ C). As the formula tends to overestimate winter evapotranspiration and underestimate summer evapotranspiration, the calculated data were corrected according to the relationship between the ratio of the Blaney–Criddle and Penman derived monthly evapotranspirations and the mean monthly temperature, for the Policoro station. The resulting potential evapotranspiration data were distributed spatially according to the elevations of the three temperature measurement sites (Policoro, 31 m; Pertusillo, 533 m; Perticara, 750 m). Building on Denmead and Shaw (1962) and Feddes et al. (1976) and with the advice of Professor Ian Calder (University of Newcastle upon Tyne), actual evapotranspiration was calculated from a relationship between the ratio of actual to potential evapotranspiration and soil moisture potential (Figure 29.4). Different relationships were proposed for crops (without distinguishing between crop types), native vegetation and bare soil conditions.
Discharge For the simulation period there are no river gauging data. Discharge could therefore be determined only from daily water balance data for the Pertusillo reservoir (Figure 29.1), giving daily inflow to the reservoir from its catchment area of 585 km2 . However, the reservoir outflow to the river downstream (which forms an essential input to the model channel network) could be calculated with only monthly resolution, too coarse to permit satisfactory representation of typical storm events. It
Modelling Large Basin Hydrology and Sediment Yield
403
1988 Rainfall (mm/hr)
20
10
0 Jan
Feb Mar
Apr May Jun
Jul
Aug Sep
Oct
Nov Dec
Aug Sep
Oct
Nov Dec
Aug Sep
Oct
Nov Dec
Aug Sep
Oct
Nov Dec
1987 Rainfall (mm/hr)
20
10
0 Jan
Feb Mar
Apr May Jun
Jul 1986
Rainfall (mm/hr)
20
10
0 Jan
Feb Mar
Apr May Jun
Jul 1985
Rainfall(mm/hr)
20
10
0 Jan
Feb Mar
Apr May Jun
Jul
Time (hours)
Figure 29.3 Example of hourly rainfall time series for the simulation period, January 1985–December 1988
was therefore simulated as a small percentage of the calculated inflow (5% as an initial estimate). The remainder of the inflow was removed from the simulation since (in reality) it is transferred out of the basin by pipeline. It had been expected that water balance data could similarly be used to determine a discharge time series at the Gannano barrage (hence the choice of this site for defining the simulation area). However, the data were found to be incomplete and the Pertusillo record therefore provides the only means of checking the simulation results. Essentially it allows an internal validation of the Agri model, for the upper part of the basin. Water from springs and groundwater sources is used for irrigation in the upper Agri Basin and the coastal zone. In addition there are large water transfers out of the Agri Basin to supply irrigation schemes in neighbouring basins.
404
Mediterranean Desertification
Actual/potential evapotranspiration
1.0 Sparse vegetation/bare ground Native vegetation Crops
0.8
Upper Bounds
Lower Bounds 0.6
0.4
0.2
0.0 −150.0
−100.0
−50.0
0.0
−150.0
−100.0
−50.0
0.0
Soil moisture potential (m)
Figure 29.4 Uncertainty bounds for the relationship between the ratio of actual to potential evapotranspiration and soil moisture potential applied in SHETRAN, for three land covers
4 4.1
SHETRAN SIMULATIONS Simulation Constraints
The limited availability of the discharge data and the approximate nature of the soil parameter evaluations meant that a comprehensive, fully validated application of SHETRAN to the Agri Basin was not possible. The applications were therefore carried out with the more limited aim of demonstrating that it is feasible to apply SHETRAN to a basin the size of the Agri and achieve physically reasonable results. Consequently the simulation results should be viewed as illustrations of modelling potential rather than definitive descriptions of the Agri Basin response. Reflecting the uncertainty in the model parameter values, the results are presented in the form of uncertainty envelopes as described in section 2.3. 4.2
Hydrological Simulations
The first stage in applying the model was establishment of parameter baseline values. These are not necessarily the most accurate or most representative values. Instead, they are best-estimate values, based on available data and the modeller’s own hydrological judgement, although there may also be a degree of calibration in their evaluation. They form the basis for selecting the parameter bound values, which in turn enable the output uncertainty envelope to be determined. In the case of the Agri, a number of preliminary simulations assisted in the evaluation of those model parameters and functions for which there was the greatest uncertainty and to which the model output was most sensitive. These were soil depth, the saturated hydraulic conductivity for the fractured rock layer, the Strickler overland flow resistance coefficient and the variation of the ratio of actual to potential evapotranspiration with soil moisture potential. All other model parameters and functions were set directly from measured data or data in the literature as described in section 3.2. Through the preliminary simulations, empirical adjustment of the fractured rock conductivities was found necessary to represent the winter baseflows. The adjusted conductivities are rather higher than might be expected for the local soil types and lithology and are an example of the “effective” values referred to in section 2.2.
405
Modelling Large Basin Hydrology and Sediment Yield
Bound values for the four critical parameters and functions were set by defining maximum and minimum values to bracket the baseline values, based on past experience, hydrological judgement and information in the literature. In the case of the soil thickness, bounds were set only for the total column length; the topsoil thicknesses were not varied. Similarly, bounds for the saturated conductivity were set only for the fractured rock layer which, occupying most of the soil column, exercised more influence on the groundwater simulation results than did the topsoil. For the Strickler coefficient, the baseline and bound values were set on the assumption that land use controls the surface roughness. Similarly, for the evapotranspiration relationship, the bound values varied with land cover (Figure 29.4). Table 29.1 shows the baseline and bound values. To determine the output uncertainty bounds, simulations were carried out for each combination of the four sets of maximum and minimum parameter or function values, giving 24 or 16 runs. Each simulation was started in August 1983 so that the results for the period of interest, 1985–1988, did not show a dependence on the initial conditions. (The length of the “run-in” period was selected from preliminary tests.) Output bounds for the discharge into the Pertusillo reservoir were obtained by superimposing the 16 simulated time series of daily discharge to create an uncertainty envelope. Model application was then completed by determining the proportion of time for which the observed daily discharge record was contained within the envelope. For the period 1 January 1985–31 December 1988, the containment was 79% when calculated for daily flows and 74% when calculated for monthly runoff. (There are 39 days of record in 1986 that are missing from the measured discharge time series, so these results refer to 1422 days out of the full simulation period of 1461 days.) Figure 29.5 compares the simulation bounds with the observed monthly discharge record for the basin to the Pertusillo reservoir for the full simulation period. Figures 29.6 and 29.7 make the same comparison for daily discharge for 1985 and 1987 respectively. (The 1988 discharge time series is similar to 1985 in having several high flow events at the end of the year, while the 1986 time series Table 29.1 Values for the model parameters and functions used in determining the simulation bounds
Parameter/function Soil column thickness (m) Saturated hydraulic conductivity for fractured rock (m day−1 ) Flysch Sandstone Limestone
Minimum value 5
0.1 10 10
Baseline value
Maximum value
10
20
0.2 20 20
0.4 40 40
15 20 10 5 5 2 2
Strickler overland flow resistance coefficient Bare soil Bare rock Field crops Pasture Macchia Deciduous forest Coniferous forest
3.75 5 2.5 1.25 1.25 0.5 0.5
7.5 10 5 2.5 2.5 1 1
Evapotranspiration function: maximum value of actual/potential evapotranspiration ratio (see Figure 29.4) Crops Native vegetation Bare soil
0.6 0.4 0.2
– – –
1.0 0.6 0.2
406
Mediterranean Desertification
observed simulation bounds simulated
Monthly discharge volume (m∗∗3)
1.0e + 08
8.0e + 07
6.0e + 07
4.0e + 07
2.0e + 07
0.0e + 00 0
12
24 Time (months)
36
48
Figure 29.5 Comparison of observed monthly discharge into the Pertusillo reservoir with the simulated uncertainty envelope for 1 January 1985–31 December 1988
2.0e + 07
Daily discharge (m∗∗3)
observed simulation bounds simulated 1.5e + 07
1.0e + 07
5.0e + 06
0.0e + 00
0
50
100
150
200
250
300
350
Time (days)
Figure 29.6 Comparison of observed daily discharge into the Pertusillo reservoir with the simulated uncertainty envelope for 1985
follows 1987 in having no such events.) Table 29.2 shows the measured annual water balance for the basin to the Pertusillo reservoir and an annual summary of the uncertainty bound simulations for the basin to the Pertusillo reservoir and the full 1532-km2 basin to the Gannano barrage. At the annual scale, as also for the full four-year period, the simulation bounds contain the measured runoff totals for the basin to the Pertusillo reservoir. In Table 29.2 simulation results are shown both for the 1422-day period with measured reservoir water balance data (to allow comparison of measured and simulated runoffs) and for the full 1461-day simulation period (for completeness of summary).
407
Modelling Large Basin Hydrology and Sediment Yield 1.5e + 07
Daily discharge (m∗∗3)
observed simulation bounds simulated 1.0e + 07
5.0e + 06
0.0e + 00 731
781
831
881
931
981
1031
1081
Time (days)
Figure 29.7 Comparison of observed daily discharge into the Pertusillo reservoir with the simulated uncertainty envelope for 1987 Table 29.2 Hydrological mass balance and flow simulation results for the Agri Basin above the Pertusillo reservoir and the full basin above the Gannano barrage
Year
Measured rainfall (mm)
Measureda Total runoff Best-r 2 potential runoff/rainfall Best-r 2 Upper Lower ratio evapoMeasuredb transpiration (mm) simulation simulation simulation (mm) (mm) bound bound (mm) (mm)
Pertusillo reservoir 1985 1257 1986 1038 1987 957 1988 929 Average 1045
1348 1302 1332 1353 1334
536 320 313 321 372
Gannano barrage 1985 1016 1986 832 1987 853 1988 811 Average 878
1363 1327 1348 1364 1350
– – – – –
a
655 467 (366)c 283 238 411 (386)c
851 613 (477)c 429 383 569 (535)c
436 302 (223)c 162 150 263 (243)c
0.52 0.45 (0.35)c 0.30 0.25 0.39 (0.37)c
262 99 79 83 131
335 149 139 140 191
198 63 42 52 89
0.26 0.12 0.09 0.10 0.15
Calculated from measured temperature record; see section 3.2. Calculated from measured reservoir water balance data; see section 3.2. c Calculated only for days with measured reservoir water balance data; all other results refer to all simulated days. b
408
Mediterranean Desertification 1.0e + 07
Daily discharge (m∗∗3)
observed simulated
5.0e + 06
0.0e + 00
0
50
100
150
200
250
300
350
Time (days)
Figure 29.8 Comparison of observed daily discharge into the Pertusillo reservoir with the best-r 2 simulation for 1985
In the presence of uncertainty, any single simulation is only one out of the range of possible basin representations accounted for by the uncertainty envelope. Single simulations should therefore be presented, if at all, with care. However, a single flow simulation was required as input to the sediment transport simulations. Out of the 16 simulations described above, the one with the closest agreement with the observed hydrograph (as indicated by the r 2 value) was therefore selected. At r 2 = 0.564, that agreement is by no means perfect: visual comparison of observation and simulation shows both overestimation and underestimation of flow peaks (e.g. Figure 29.8). However, the general pattern is well represented. An annual summary of the best-r 2 simulation is shown in Table 29.2. 4.3 Soil Erosion and Sediment Yield Simulations Simulations were carried out to demonstrate the capability of SHETRAN for representing soil erosion and sediment yield at the large basin scale. There are no relevant sediment yield data for the Agri Basin and the simulation results were therefore examined by comparison with measured sediment yields in other Mediterranean basins. The values of the additional model parameters for the sediment simulations were estimated from the soil and vegetation data used in the hydrology simulations, from the literature and from experience with previous applications. In particular, the three soil erodibility coefficients were based on the values used in simulating catchments in Mediterranean France and Portugal (Lukey et al. 1995; Bathurst et al. 1996). Information on the sediment size distribution was obtained from a survey along the Agri channel carried out during the spring of 1994. Five representative sizes (for the hillslope and channel) were selected, ranging from 0.1 mm (fine sediment) to 256 mm (channel bed boulders). The Engelund–Hansen equation was used to determine sediment transport capacity for overland flow. To represent uncertainty, upper and lower bound values were set for the erodibility coefficients (10 and 0.1 J−1 for the raindrop impact erodibility coefficient and 20 and 1 mg m−2 s−1 for the overland flow and the bank material erodibility coefficients). An envelope of sediment yield results was then produced, its upper bound based on the combined use of the upper bounds for the erodibility coefficients, its lower bound corresponding to the lower coefficient values. The sediment simulations were driven by flow data from the best-r 2 hydrology model. In this way, the uncertainty, as
409
Modelling Large Basin Hydrology and Sediment Yield
represented by the bounds, is due only to the uncertainty in evaluating the erodibility coefficients. However, inaccuracies in the flow simulation have a direct impact on the accuracy of the sediment yield simulation. The errors apparent in the best-r 2 simulation (Table 29.2) mean that the sediment yield simulations are similarly in error compared with how they would appear if based on a completely accurate hydrological simulation. The simulation results are given in Table 29.3 and the simulated variation in daily sediment yield for 1985 for the basin to the Pertusillo reservoir is shown in Figure 29.9. In Table 29.3 the sediment yields for the Gannano barrage refer to the full 1532-km2 basin. Table 29.3 Sediment yield simulation results for the Agri Basin above the Pertusillo reservoir and the full basin above the Gannano barrage
Best-r 2 simulated runoff (mm)
Upper bound (t ha−1 year−1 )
Lower bound (t ha−1 year−1 )
Pertusillo reservoir 1985 1257 1986 1038 1987 957 1988 929 Average 1045
655 467 283 238 411
12.2 4.0 3.3 2.9 5.6
11.1 3.4 2.9 2.6 5.0
Gannano barrage 1985 1016 1986 832 1987 853 1988 811 Average 878
262 99 79 83 131
7.0 9.1 6.0 5.1 6.8
4.0 5.3 3.3 2.8 3.9
Year
Rainfall (mm)
Simulated sediment yield
1.0e + 07
8.0e + 07 simulated water discharge 6.0e + 07
5.0e + 06
4.0e + 07
2.0e + 07
0.0e + 00
0
50
100
150 200 Time (days)
250
300
350
Daily sediment yield (kg)
Daily discharge (m∗∗3)
simulated sediment yield bounds
0.0e + 00
Figure 29.9 Simulated sediment yield bounds and best-r 2 simulation of daily water discharge for inputs to the Pertusillo reservoir for 1985. In this case the sediment bounds are almost indistinguishable
410
5
Mediterranean Desertification
DISCUSSION OF RESULTS
5.1 Hydrological Simulations The estimated nature of important model parameters and the lack of good quality validation data limit the conclusions that can be drawn regarding the performance of SHETRAN in the Agri simulations. However, the simulations, as far as they go, are encouraging concerning the feasibility of applying SHETRAN to basins of the size and nature of the Agri. The 79% containment of the measured Pertusillo discharges by the simulation envelope is considered to be a good result, comparable with figures achieved elsewhere, e.g. around 80% for the Rimbaud catchment in southern France (Parkin et al. 1996; J.C. Bathurst and J. Sheffield unpublished data) and for the Reynolds Creek catchment in Idaho, USA (J.C. Bathurst unpublished data). At the annual scale the containment is 100%, suggesting an ability to represent, within the limits of uncertainty, the annual water balance. It is particularly satisfactory that this ability is demonstrated for a range of annual rainfall totals: from 1985 to 1988 the annual rainfall for the Pertusillo basin fell by 31% relative to the average annual value for the four-year period (Table 29.2). An important indicator of simulation success is the extent to which the simulation bounds represent the pattern of observed hydrograph variability. The uncertainty envelope should be wide enough to contain most of the observed hydrograph but not so wide that its representation of the various features of the hydrograph (individual peaks, recessions, baseflow) is meaningless. In this case, for the basin to the Pertusillo reservoir, the bounds for the daily discharge hydrograph are not unreasonable (Figures 29.6 and 29.7). Some of the winter peaks are overestimated but most are within the bounds and their general shape and timing are well represented. The bounds describe particularly well the late part of the recessions and emphasize the difference between the winter and summer baseflow magnitudes. Envelope width is therefore appropriate. In Figure 29.5, the fluctuations associated with individual events are absorbed into a smoother monthly variation. In this case the bounds represent well the month-to-month variation, with an excellent description of the summer low-flow periods. The timing of the peak monthly runoff each year is well represented but the simulated magnitudes change in accuracy through the simulation. Reflecting the decrease in annual rainfall from 1985 to 1988, both simulated and measured peak discharges decline from year to year. Good containment of the measured peaks within the uncertainty envelope is achieved in 1987 and 1988 but in 1985 and 1986 the simulations are too high. A number of simulation approaches were attempted but it did not prove possible to find one that could represent all four annual peaks with similar accuracy. The reasons for this are unknown but could include the following:
1.
errors in the rainfall data or in the reservoir water balance data from which the observed discharge hydrograph is calculated; 2. some catchment characteristic that acts to reduce the impact of an annual rainfall change on discharge and which is not represented in SHETRAN; 3. unknown human intervention in the Pertusillo basin, including water diversion or transfer. Figure 29.5 shows SHETRAN to be capable of predicting monthly discharge with an uncertainty envelope that is relatively narrow and describes well the observed variation. Such a capability is of particular use to the water resources planner, providing, in the absence of measured runoff data, a basis for setting upper and lower limits on water allocation. Table 29.2 shows that the simulated annual runoff in the catchment to the Gannano barrage is less than that for the basin to the Pertusillo reservoir. The two principal reasons for this are that, in the simulation, 95% of the inflow to the Pertusillo reservoir is removed from the system (to represent transfers from the reservoir by pipeline to a neighbouring basin) and that the mean annual rainfall for the basin to the Gannano barrage is lower than that for the basin to the Pertusillo reservoir (Table 29.2). The simulations thus illustrate an ability to model spatial variability in internal basin response as a function of annual rainfall. This is an important requirement for simulating the impacts of land-use changes (which are typically spatially distributed) and climate changes.
Modelling Large Basin Hydrology and Sediment Yield
411
Because of the extraction of the Pertusillo water, it is not possible to comment on the simulated discharge hydrograph at Gannano as the representation of a natural system. 5.2 Scale Effects
One of the aims of the SHETRAN application was to investigate the appearance of any scale effects in parameter evaluation associated with the use of the 2-km model grid. Because the parameters are evaluated as “effective” values, representative at the scale of the grid square, they must account for different integrations of subgrid processes depending on the grid scale. Previous studies with SHETRAN and other models suggest that the same model parameter values can be applied at the plot (1–100 m2 ) and microbasin (order 1 ha) scales, using small model grid spacings (20 m or less) and with a good availability of field data (Wicks et al. 1988; Connolly and Silburn 1995; Figueiredo 1998). With a 2-km grid, though, there is likely to be significant subgrid variability and it would not be unexpected for the effective parameter value to differ from the measured value, typically obtained at the point scale. For example, the saturated zone conductivity may increase to compensate for a reduction in simulated groundwater gradients caused by the use of large grid squares. Similarly the overland flow resistance may decrease to account for the inclusion of subgrid channel flow within a large grid square. Previous experience has suggested that scale effects in evaluating saturated zone conductivity are not significant as long as basin topography is subdued and there is a general homogeneity of land use, soil characteristics and hydrological response within the basin. For example, applications of the SHE modelling system (SHETRAN’s precursor) to large basins in India (area 800–5000 km2 ) (Jain et al. 1992; Refsgaard et al. 1992) and to the Cobres Basin in Portugal (area 701 km2 ) (Bathurst et al. 1996) suggest that conductivities evaluated at the point or small scale can be successfully applied with a model grid spacing of 2000 m. Figueiredo (1998) similarly found no evidence of a scale effect when modelling a 137-km2 basin in north-east Brazil, although in this case the basin did not typically have a saturated zone in the soil column. However, an application to a more hilly basin in Idaho (area 234 km2 ) shows an increase in the calibrated value of saturated zone conductivity as the grid spacing increases from 50 m to 1000 m (J.C. Bathurst, unpublished data). The results for the Agri (a hilly basin) agree with the latter finding, since the conductivities required for a satisfactory baseflow simulation are large compared with the expected measured values (Table 29.1). For example, a typical conductivity for flysch is 0.0001 m day−1 , much lower than the model baseline value of 0.2 m day−1 . For overland flow resistance, the picture is less clear. In previous applications the Strickler coefficient has been evaluated as 1 (a relatively high resistance) for small basins but also for the Idaho Basin. For the India basins it was in the range 3–7 while for the Cobres it was set at 6. Figueiredo (1998) applied values of 15 and 25 at the basin scale as a function of the amount of bare ground. The Agri baseline values of 1–10 are largely consistent with the previously applied range but vary as a function of land use. It remains possible, therefore, that the Strickler coefficient increases slightly (i.e. resistance decreases) as grid scale increases but the effect does not appear to be large. Other factors such as the type of ground roughness may have a greater effect. 5.3 Soil Erosion and Sediment Yield Simulations
Table 29.3 shows the difference between the upper and lower bounds on the simulated annual sediment yields to be small. This suggests, for this particular case, a relative insensitivity to the soil erodibility coefficients (used to derive the uncertainty bounds), which in turn suggests that the simulated sediment yield is dominated more by limitations in the ability of overland flow or channel flow to transport eroded soil rather than by the hillslope or channel erosion itself. Such limitations could themselves be related to a relative infrequency of simulated overland flow and to the form of the overland flow and channel flow capacity transport equations. The erodibility coefficients provide a clearly defined basis for setting uncertainty bounds, using available experimental results and previous experience (e.g. Wicks et al. 1992). However, given that
412
Mediterranean Desertification
uncertainties in sediment yield simulation are usually larger than uncertainties in water flow simulation, their use in this case produces an unrealistically narrow envelope. Other areas of uncertainty that are known to affect the simulations and which might form an alternative basis for setting uncertainty bounds are the soil size distribution and the choice of transport capacity equation (Norouzi Banis 1998). However, with current data availability and process understanding, such a basis would be more subjective than one based on the erodibility coefficients. It should also be remembered that the sediment simulations are based on the best-r 2 hydrology simulation, with its associated errors. If the flow uncertainty were to be incorporated, the sediment bounds would be rather wider. The wettest year (1985) produces the highest simulated sediment yield. This is consistent with greater raindrop impact and overland flow erosion and a greater amount of runoff to transport the eroded material. Higher river flows also cause more bank erosion. For the other three years (1986–1988) there is rather less inter-annual variability, despite variations in the rainfall and best-r 2 runoff. This may be consistent with the general observation that annual sediment yield is affected more by a few events than by the annual rainfall and runoff totals. However, whether the true sediment yield followed this pattern is of course unknown. Because of the extraction of the Pertusillo water, it is not possible to comment on the simulated sediment yield at Gannano as the representation of a natural system. As the true sediment yield for the Agri Basin is unknown, the simulated sediment yields (in the range 2–13 t ha−1 year−1 ) can be assessed only in the context of measurements made elsewhere in the Mediterranean or other semi-arid regions. The yields are low relative to the long-term yield of 93 t ha−1 year−1 measured for a 804-km2 subcatchment of the neighbouring Sinni River (Blasi et al. 1991). However, they are comparable with the long-term yields of 2–11 t ha−1 year−1 measured for 13 basins of area 150–2400 km2 in a high erosion area of south-east Spain (Romero D´ıaz et al. 1992), and with an approximate range of 1–10 t ha−1 year−1 for catchments of area 1000 km2 in the south-east USA, derived from Walling (1983, Figure 3). The simulations are probably therefore of the generally correct order of magnitude but perhaps err on the low side. It may be noted also that 1985 was relatively wet compared with the 1951–1971 mean annual rainfall and the largest of the simulated yields may therefore also be relatively high for the Agri Basin. In 1985, the simulated sediment yield is higher for the basin to the Pertusillo reservoir than for the basin to the Gannano barrage. This agrees with the general observation that sediment yields tend to decrease as basin area increases (e.g. Walling 1983). However, the reverse is true for the simulated yields of 1986–1988. This is in spite of the lower runoff at the Gannano scale and the removal of transported sediment in the water transfer at the Pertusillo reservoir. (The simulation assumes the same sediment concentration in the water discharged from the reservoir to the Agri as in the water flowing into the reservoir.) One explanation for this pattern may lie in the balance between the sediment yield generated on the hillslopes and the yield derived from in-channel sources (bed and banks). From test simulations it was found that the simulated yield from the in-channel sources tended to mask the contribution from the hillslopes. Even for the hypothetical case where the whole of the Agri Basin was covered with the same vegetation (to ensure a uniform level of protection against soil erosion) and the Pertusillo reservoir was eliminated, the simulations still showed a downstream increase in sediment yield. Only by eliminating the in-channel sources and deriving sediment entirely from the hillslopes did the simulations show a downstream decrease in yields. Thus for the relatively dry period of 1986–1988 it is likely that little hillslope erosion was simulated and that yields were derived mainly from in-channel sources. By contrast, the greater rainfall of 1985 enabled more hillslope erosion to be simulated, with a consequent effect on the downstream variation in sediment yield. Analysis of the simulation data shows that the Gannano yield is higher than the Pertusillo yield during the period May–December; during January–April they are similar. The difference follows closely the different summer growths of vegetation simulated at the two scales. Some 31.5% of the basin to the Pertusillo reservoir is covered by deciduous trees but at the scale of the full basin to the Gannano barrage this cover is only 21.4%. In the Gannano Basin excluding the Pertusillo Basin the cover is 15.2%. In the simulation, as the trees increase their leaf area during the summer, they
Modelling Large Basin Hydrology and Sediment Yield
413
provide increased protection against soil erosion by raindrop impact. When the leaf area decreases at the end of the year, the erosion rates converge at the two basin scales. An additional explanation for the simulated downstream increase in yields may therefore be that the relatively greater extent of summer vegetation growth in the basin above the Pertusillo reservoir provides greater protection against erosion, thus reducing the sediment yield relative to that at the Gannano scale. Without validation data it is not known if the true variation of sediment yield along the Agri system is as simulated, nor if the above explanations are generally correct. However, the analysis shows how SHETRAN output can be examined to explain in physical terms an apparently unexpected result. The simulations indicate that careful account should be taken of in-channel and hillslope contributions when modelling sediment yield. They also raise the possibility that basins with large in-channel sediment supplies (from bed and bank erosion) may not show the conventional downstream decrease in sediment yield.
6 CONCLUSIONS A number of positive conclusions can be drawn concerning the application of SHETRAN to basins of the size of the Agri (1000–2000 km2 ) and use of the model in the management of basins threatened with desertification. However, because of the poor quality of the soil and flow data, it was not possible to validate the model comprehensively and the simulation results should be viewed as an illustration of potential rather than a definitive description of the Agri Basin response. 1. From comparison with the generated Pertusillo flow data, SHETRAN reproduces the overall water balance well, at the monthly and annual scales and within the limits of uncertainty. The daily discharge time series is also reproduced within reasonable uncertainty limits. The simulated sediment yield is similar to yields in high erosion areas; however, the simulation bounds could be revised by incorporating uncertainty from the flow calculations and changing the basis for representing uncertainty in the sediment calculations. Evaluation of the saturated hydraulic conductivity is consistent with previous suggestions that this parameter may show some dependency on model grid scale in hilly basins. This effect needs to be quantified but in general there does not appear to be any fundamental obstacle to applying SHETRAN to basins of the scale of the Agri. The simulation thus doubles SHETRAN’s tested scale of application, from 700 km2 in the MEDALUS Phase I project, increasing its relevance to the larger scales at which planning decisions are typically made. 2. The results show how SHETRAN can be applied to problem solving, even with sparse data, by defining uncertainty envelopes. Lack of data does not stop decisions from being made. The value of SHETRAN in such cases lies in its ability to quantify the potential consequences (i.e. the associated uncertainty) of making decisions in the absence of data. The uncertainty envelope quantifies the range of possible basin responses as determined from the available data. Decision makers can then design their projects to accommodate this range. Alternatively it may be more cost-effective to fund a data collection programme that enables the uncertainty to be reduced. 3. The Agri simulation demonstrates the superiority of physically based models for applications to catchments with poor or non-existent records of output data. Their parameters have a physical meaning and can therefore be specified from field measurements or information in the literature. More traditional models could have been calibrated for the basin to the Pertusillo reservoir using the observed discharge record but would have had no basis for extension to the scale of the basin to the Gannano barrage. 4. In a certain respect, mathematical models have the ability to form the concluding output and the principal practical deliverable of an interdisciplinary project such as MEDALUS. This is because they are the means by which the results of other components of the project can be drawn together to provide an overall view of the central problem. SHETRAN, for example, can incorporate knowledge gained from the small spatial scale experimental studies of soil physics, vegetation growth patterns and the effect of land use on hydrological response. It takes as input the results of meteorological surveys and climate scenario generation. It can be applied to
414
Mediterranean Desertification
examine the basin-scale impacts of trends identified by socio-economic studies and its output can provide information relevant to the mapping of desertification indicators. SHETRAN can be used to develop an understanding of basin response mechanisms and to highlight areas of poor understanding. Specifically, it provides a means of predicting the impacts of possible future changes in land use and climate, and thence for adopting appropriate measures for protection, management and rehabilitation of river basins. Further application of SHETRAN to the Agri Basin will seek to improve the soil property database and to predict the impacts of possible future changes in land use and climate on water and sediment yields.
ACKNOWLEDGEMENTS The authors are most grateful to the MEDALUS group of Professor F. Basso (University of Basilicata) for help in assembling the Agri Basin data set. They also thank the MEDALUS teams of Professor M.J. Kirkby (University of Leeds), Professor J. Thornes (King’s College London) and Dr J. Palutikof (University of East Anglia) for their help with data processing. Dr O. Hamad, Dr J. Sherwood and Dr J. Stunell (formerly postgraduate students at the University of Newcastle upon Tyne) are similarly thanked for digitizing and processing map and chart data. Professor Ian Calder (University of Newcastle upon Tyne) provided valuable advice on evapotranspiration relationships. The work described here was funded by the European Commission through the MEDALUS II (contract number EV5V-CT92-0164) and MEDALUS III (contract numbers ENV4-CT95-0115 and 0119) projects, and this support is gratefully acknowledged. The participation of Ms Vicente (at Departamento de Hidr´aulica y Medio Ambiente, Universidad Polit´ecnica de Valencia, Spain, at the time of the study) was funded by the European Commission’s ERASMUS international exchange programme for students.
REFERENCES Bathurst JC, Wicks JM and O’Connell PE (1995) The SHE/SHESED basin scale water flow and sediment transport modelling system. In Singh VP (ed) Computer Models of Watershed Hydrology. Water Resources Publications, Highlands Ranch, Colorado, pp. 563–594. Bathurst JC, Kilsby C and White S (1996) Modelling the impacts of climate and land-use change on basin hydrology and soil erosion in Mediterranean Europe. In CJ Brandt and JB Thornes (eds) Mediterranean Desertification and Land Use. John Wiley, Chichester, pp. 355–387. Beven K (1989) Changing ideas in hydrology – the case of physically-based models. Journal of Hydrology 105, 157–172. Beven K and Binley A (1992) The future of distributed models: model calibration and uncertainty prediction. Hydrological Processes 6, 279–298. Binley A, Beven K and Elgy J (1989) A physically based model of heterogeneous hillslopes 2. Effective hydraulic conductivities. Water Resources Research 25, 1227–1233. Blasi L, Cassano G and Grauso S (1991) Valutazione dell’entit`a della sedimentazione nel bacino artificiale di M te Cotugno (media valle del fiume Sinni, Basilicata). Geologia Applicata e Idrogeologia XXVI, 111–139 (in Italian with English abstract). Bras RL (1990) Hydrology: An Introduction to Hydrologic Science. Addison-Wesley, Reading, Massachusetts. Brooks RH and Corey AT (1964) Hydraulic Properties of Porous Media. Hydrology Paper No. 3, Colorado State University, Fort Collins, Colorado. Connolly RD and Silburn DM (1995) Distributed parameter hydrology model (ANSWERS) applied to a range of catchment scales using rainfall simulator data II: application to spatially uniform catchments. Journal of Hydrology 172, 105–125. Denmead OT and Shaw RH (1962) Availability of soil water to plants as affected by soil moisture content and meteorological conditions. Agronomy Journal 54, 385–390. Dunne T and Leopold LB (1978) Water in Environmental Planning. Freeman, San Francisco. Engman ET (1986) Roughness coefficients for routing surface runoff. Proceedings of the American Society of Civil Engineers, Journal of Irrigation and Drainage Engineering 112, 39–53.
Modelling Large Basin Hydrology and Sediment Yield
415
Ewen J (1995) Contaminant transport component of the catchment modelling system SHETRAN. In Trudgill ST (ed.) Solute Modelling in Catchment Systems. John Wiley, Chichester, pp. 417–441. Ewen J and Parkin G (1996) Validation of catchment models for predicting land-use and climate change impacts: 1. Method. Journal of Hydrology 175, 583–594. Ewen J, Parkin G and O’Connell PE (2000) SHETRAN: distributed river basin flow and transport modeling system. Proceedings of the American Society of Civil Engineers, Journal of Hydrologic Engineering 5, 250–258. Feddes RA, Kowalik P, Neuman SP and Bresler E (1976) Finite difference and finite element simulation of field water uptake by plants. Hydrological Sciences Bulletin 21, 81–98. Figueiredo EE (1998) Scale effects and land use change impacts in sediment yield modelling in a semi-arid region of Brazil. PhD thesis, University of Newcastle upon Tyne, UK. Jain SK, Storm B, Bathurst JC, Refsgaard JC and Singh RD (1992) Application of the SHE to catchments in India. Part 2. Field experiments and simulation studies with the SHE on the Kolar subcatchment of the Narmada River. Journal of Hydrology 140, 25–47. Lukey BT, Sheffield J, Bathurst JC, Lavabre J, Mathys N and Martin C (1995) Simulating the effect of vegetation cover on the sediment yield of Mediterranean catchments using SHETRAN. Physics and Chemistry of the Earth 20(3/4), 427–432. Mazzanti B, Preti F, Romano N and Santini A (1998) Characterization of climatic evaluation by analysis of rainfall time series: the Agri Basin case study. Proceedings of the XXVI Congress of Hydraulics and Hydraulic Constructions, vol. II, Catania, Italy, 9–12 September CUECM, pp. 259–271 (in Italian with English abstracts). Norouzi Banis Y (1998) Data provision and parameter evaluation for erosion modelling. PhD thesis, University of Newcastle upon Tyne, UK. Parkin G, O’Donnell G, Ewen J, Bathurst JC, O’Connell PE and Lavabre J (1996) Validation of catchment models for predicting land-use and climate change impacts: 2. Case study for a Mediterranean catchment. Journal of Hydrology 175, 595–613. Quinton JN (1997) Reducing predictive uncertainty in model simulations: a comparison of two methods using the European Soil Erosion Model (EUROSEM). Catena 30, 101–117. Rawls WJ and Brakensiek DL (1989) Estimation of soil water retention and hydraulic properties. In MorelSeytoux HJ (ed.) Unsaturated Flow in Hydrologic Modeling Theory and Practice. Kluwer Academic, Dordrecht, The Netherlands, pp. 275–300. Refsgaard JC, Seth SM, Bathurst JC, Erlich M, Storm B, Jørgensen GH and Chandra S (1992) Application of the SHE to catchments in India. Part 1. General results. Journal of Hydrology 140, 1–23. Romano N and Santini A (1997) Effectiveness of using pedo-transfer functions to quantify the spatial variability of soil water retention characteristics. Journal of Hydrology 202, 137–157. Romero D´ıaz MA, Cabezas F and L´opez-Berm´udez F (1992) Erosion and fluvial sedimentation in the River Segura basin (Spain). Catena 19, 379–392. Santini A, Romano N and Coppola A (1996) Geostatistical analysis of soil spatial variability in a hillslope of the Agri river basin. In Problems with Large Irrigation Districts. Proceedings of the East-Sesia Farmers’ Union Conference, Novara, Italy, 6–7 June, pp. 281–293 (in Italian with English abstracts). Walling DE (1983) The sediment delivery problem. Journal of Hydrology 65, 209–237. Wicks JM and Bathurst JC (1996) SHESED: a physically-based, distributed erosion and sediment yield component for the SHE hydrological modelling system. Journal of Hydrology 175, 213–238. Wicks JM, Bathurst JC, Johnson CW and Ward TJ (1988) Application of two physically-based sediment yield models at plot and field scales. In Bordas P and Walling DE (eds) Sediment Budgets. International Association of Hydrological Sciences Publication No. 174, Centre for Ecology and Hydrology, Wallingford, UK, pp. 583–591. Wicks JM, Bathurst JC and Johnson CW (1992) Calibrating SHE soil-erosion model for different land covers. Proceedings of the American Society of Civil Engineers, Journal of Irrigation and Drainage Engineering 118, 708–723.
Section VIII
Conclusions
30
Emerging Mosaics
J.B. THORNES
Department of Geography, King’s College London, UK
1 INTRODUCTION “Any policy oriented measures directed at the monitoring and control of land degradation and desertification in the Mediterranean need to recognize that there is no simple panacea for the achievement of sustainable land management.” (Ghazi 1999)
From reading the previous 29 chapters in this book, what emerges most clearly is that there is no magical underlying truth that pulls it all together for the reader. The essential diversity of the landscape arising from physique and culture, and the palimpsest character of this mosaic that arises from its history, are such that the search for universal truths about causes and remedies for desertification and the appropriate actions to be taken are as diverse as the mosaic of landscape itself. As a result, the wheel turns on another axis. It is to identify elements of the mosaic that are homogeneous enough to justify common approaches to management. These can then be supported at national and transnational levels, and policies developed that are robust enough to satisfy the enormous diversity of the Mediterranean environment, thus avoiding the problem that local anomalies undermine the policy by producing unacceptable outcomes. Such outcomes may lead not only to ridicule, but to social injustice among the recipients of the policy and to ultra-conservatism among the policy makers. Unfortunately the anomalies and injustices emerge only in the implementation and then only after 30 or 40 years of struggling on both sides. It would be better to design policies or implementation mechanisms that are flexible enough to cope with the diversity that arises from the mosaic and the palimpsest. To some extent this has been achieved by the sharing of power and costs between the central authority (the European Commission) and the national governments or their autonomous regions. The devolution of power in this form encourages recognition of the spatial mosaic, even at the very local level, and calls for a clearer identification and resolution of the mosaic. This was attempted in MEDALUS I through the concept of desertification response units and in MEDALUS II by examining the desertification of environmentally sensitive areas. More recently, efforts have involved key indicators of desertification, as described by Imeson and Cammeraat in Chapter 14.
2 DIFFERENT PROBLEMS AND CAUSES IN THE TARGET AREAS In MEDALUS II, in an attempt to stimulate interdisciplinary approaches, it was decided to focus on several target areas and this approach served the project well. The target areas were chosen because they were known to be significantly affected by desertification, but at the same time offered marked between-area differences to capture the main contrasts within the Mediterranean region as a whole. They include the Guadalent´ın Basin in south-east Spain, the Agri Valley in southern Italy, north-west Sardinia and the island of Lesvos. All are “dry” Mediterranean areas and come within the ICCD (United Nations 1994) definition of desertified regions. By focusing thematic and modelling efforts, it was possible to draw on the extensive existing knowledge of these areas and Mediterranean Desertification: A Mosaic of Processes and Responses. Edited by N.A. Geeson, C.J. Brandt and J.B. Thornes 2002 John Wiley & Sons, Ltd
420
Mediterranean Desertification
on the earlier monitoring efforts of the MEDALUS Programme, as indicated by Brandt and Thornes (1996). Descriptions of these target areas and the sample thematic and modelling efforts carried out within them are given in Chapters 17 to 29. In this section I take the opportunity to reinforce the point that, although desertification is the recurrent theme, the main thematic problems are different in character in each area: water in the Guadalent´ın, grazing in Sardinia and Lesvos, and tillage methods in the Agri Valley, Basilicata. In the Guadalent´ın Valley water resources, both surface and groundwater, dominate the desertification problem through intensive irrigated agriculture and its associated problems. An increase in irrigated agriculture has been the main form of change in the Guadalent´ın over the last 50 years, using mainly groundwater and water from the inter-basin transfer which brings water some 300 km from the Tajo River in central Spain. In 1973, 24 Hm3 year−1 were extracted locally, rising to 56 Hm3 year−1 in 1990, but reducing to 30 Hm3 by 1996 because, with the water table at a depth of 290 m, water extraction had become very expensive. Irrigated lands that depend on local groundwater are being abandoned, especially where water extracted from wells is becoming saline. These combined effects have led to a wave of land abandonment. The salinity and the effects of intensive cultivation have left the soils impoverished and this is the first step towards desertification. In Chapter 21, the authors track longer-term demographic and economic changes since the end of the 18th century. In developing the Plan to Combat Desertification in the Guadalentin Basin (Chapter 22), Rojo Serrano et al. review the early demands for reafforestation of large areas to avoid damage and loss of life from flooding and debris flows. Afforestation, as the universal palliative to land degradation, is not a new concept, but there was renewed clamour after the disastrous floods in the basin in 1973 which led to great damage of property and loss of life. Chapter 22 gives a careful account of the development of the Management Plan and reveals that 62 watershed restoration projects distributed across the basin have been completed since 1885. The survey of past restoration efforts indicates that the mechanized afforestation techniques, such as terracing subsoiling, have been more effective than manual ones (holes, bench terraces and strips) in cutting hillslope runoff, and retaining and storing as much water and moisture as possible. This leaves some doubt as to the dogma that asserts that traditional knowledge leads to best practice. The authors come implicitly to the conclusion that the ideal element to initiate recovery of the natural vegetation is Pinus halepensis which can act as a nursery species for Quercus rotundifolia. They use an ecologically based classification of the mosaic of land-use patches to propose the required action in all parts of the basin. In most of the basin, but depending on the area to be reforested, the species must be Quercus rotundifolia or Pinus halipensis. However, in some areas it will be possible to use Pinus pinaster or Pinus nigra mixed with Quercus rotundifolia. Elsewhere it has been proposed that the use of non-tree species constitutes a valuable alternative (Francis and Thornes 1990) and that the success or otherwise of afforestation in reducing runoff and soil loss depends on the sequence of climatic conditions in the early years of regeneration and the pattern adopted for planting the new trees (Obando 1997). In a nutshell, very varied rainfalls in the years succeeding planting should lead to little or no reductions, and planting in the lower part of the basin at a rate that is linear with time is most effective in reducing sediment yield. A case study of the Sardinian target area is given in Chapter 6, where the emphasis is on livestock agriculture and the impacts of the reduction of the traditional rotation of cereal–grazing–fallow and the associated abandonment of arable lands and reduction in forage availability. The authors engage in the important debate about overgrazing as a major cause of land degradation. As late as 1996, Seligman claimed that “Amongst all the factors that contribute to land degradation in the Mediterranean Basin, high stocking rates must be placed low on the list.” In Chapter 6, it is shown that this problem is a very complicated one, involving the effects of changing structure and composition of the pastures and their impact on soil porosity. Significant changes in structural porosity of the soil were revealed over a single year and high stocking rates were shown to be important in this particular case. The impact of the combination of fire and overgrazing occurring on the Greek island of Lesvos (Chapter 7) is limited to the survival of phrygana (maquis) through its adaptation and through
Emerging Mosaics
421
management practices. It is shown that attempted complete exclusion of fire only brings extra problems of biomass accumulation, higher temperature burns and excessive damage to vegetation and seed banks. The embedded study of the island of Chios shows how fires reduce the capacity to produce livestock feed, leading to rises in fodder imports in the decade 1981–1991 and this has led to a feedback spiral: to minimize import costs, natural ecosystems have been “completely” overgrazed. The combination of grazing and fire and its impact on community structure have also been studied in detail as a thematic topic by Dalaka et al. (Chapter 9), where the hypothesis of convergent evolution in Mediterranean environments is also discussed at length. They conclude that, while grazing mainly affects individual activities, it is fire or “corrective activities” (such as reforestation) that change species composition and alter the community status. They also conclude that the duration of that action, rather than specific environmental parameters, seems to be the cause of divergence between communities, indicating how important the planning of land use is when acting to reverse desertification. In the Agri Valley of Basilicata, Italy (Chapters 24–26) the contrast between the upper, middle and lower sections reveals, within the local-scale mosaic, contrasting behaviour and contrasting approaches to desertification mitigation. Unlike the other two target areas, the problems here arise largely from the physiographic conditions, with clay soils, mountainous topography and a history of deforestation giving rise to severe degradation. Here long-term field experiments at Guardia Perticara on appropriate cropping systems for sustainable agriculture have examined how soil tillage practices change the chemical, organic and physical structures of the clay soils under the dominant crops, durum wheat and horsebean for seed. Here, too, the traditional practice is crop rotation with fallow, durum wheat/chickpea, durum wheat, vetch/oats. The experimental results confirm that fallow improves the conservation of water, but the ecological and technical importance of this response varies in the different environments that occur within the Basin. They also conclude that minimum tillage does not produce great differences from traditional tillage practices in terms of economic and energy costs. In the Agri Valley, human activity in the past allowed widespread deforestation as a consequence of the need to enlarge the area of arable land. Subsequently, much of the cleared land situated on slopes has become difficult to farm and has consequently been abandoned. This is a recurrent theme in all target areas and in other areas studied. The onset of abandonment for a variety of reasons leads to a downward spiral of less intensive and poorer labour inputs in the form of conservation measures, leading inexorably to a reduction in the quality of rural life. In the Guadalent´ın, it was the increase in available irrigation resource that led to population increase and extension onto unsuitable territories which subsequently led to abandonment and neglect of land, which in turn produced salinized and eroded soils and unacceptable standards of living. As in other parts of Europe, the push for productionist agriculture, with its associated system of national and European Community subsidies, has sustained agriculture beyond the limit. Today in the Guadalent´ın, “much of the agriculture is now dependent on subsidy” (Chapter 21). To this must be added the over-extension resulting from new irrigation and the associated massive crop changes (the area of “forest” that includes matorral and other scrub forms decreased by 35% between 1947 and 1989). In the Guadalent´ın, the biggest change was in the area devoted to cereals and almonds on dry lands and the area used for citrus on irrigated land. Scrub land has been ploughed for cultivation and afforestation and some of the ploughed land has since been abandoned. It is the view of the authors of Chapter 21 that the current degradation in the Guadalent´ın Basin is closely related to the socio-economic changes they observed and the decline in the rural population. The main thrust of the revised Common Agriculture Policy (CAP), in the form of Agenda 2000, is towards sustaining rural populations by non-productionist methods. This can only be interpreted as implicitly positively beneficial from a desertification point of view. Nevertheless, because the agriculture of the Guadalent´ın Basin, and most of the agriculture of the Mediterranean, is dependent on subsidies, the removal of those subsidies (as proposed in the McSharry reforms to the 1992 GATT round and the restraints of World Trade Organization agreements) could precipitate an economic collapse of marginal incomes that might do far more than climate change in exacerbating the existing land degradation.
422
3
Mediterranean Desertification
THEMATIC STUDIES
Vegetation as a control on erosion pervades almost all the chapters in this book. Quinton et al. (Chapter 8) provide an analysis of the bio-engineering properties of different species for mitigation, based on rainfall simulations accompanied by measures of soil loss. Their contribution culminates in a table of semi-natural vegetation treatments that will provide an important source for future mitigation activities, whilst recognizing that more work is needed on developing ecological successions and re-vegetation methods that promote a sustainable, high value canopy or ground cover. The decline in the rate of erosion with cover is greatest for shrub and bush covers than for grasses. Improvements in the capacity to identify vegetation covers from remotely sensed imagery in sparse vegetation, where the soil transmits an important part of the received signal (Chapter 10), will make it possible to plan mitigation actions more effectively in future than at present, through the ease of survey of the cover type that this development brings. The tight coupling of vegetation and climate (Chapter 20) means that the normalized difference vegetation index (NDVI) can be inverted to provide indicators of climate and climate change at a regional scale (INDVI). Increases in the INDVI based on a composite monthly NDVI for selected months correspond to decreases in the aridity levels, though there is a high spatial variability of the index, mainly due to seasonal contrasts. The authors conclude that their method “should work in all areas where the thermal factor is limiting in summer”. The recognition that most Mediterranean soils involve a higher percentage of rock fragments on, and in, the soil surface, necessitates a re-thinking of conventional runoff hydrology and hydraulics and its related implications for soil erosion. The extensive experimental work on the impact of rock fragments on soil degradation and water conservation (Chapter 11) indicates the importance of rock fragments to productivity. The relative biomass production of rainfed wheat can be related to the percentage of rock fragments as well as the evapotranspiration conditions. When all rock fragments were removed from the surfaces of 32 plots, the biomass production of rainfed cereals decreased by 2–30%. Another scourge of Mediterranean soils in the context of desertification is the problem of salinization caused by the salt content of the underlying parent material, excessive evaporation, sea-water infiltration, irrigation and other anthropogenic factors (Chapter 12). Again the first-order impact is abandonment of land engendering the downward spiral of rural economies. In Chapter 13, Postiglione also outlines the management option and techniques for this blight.
4
MODELLING
Much of the work in this phase of the project has involved direct observation in the form of field trials, plot experiments and archival sources. These have provided new information and empirical results (in the form of statistical models) that have both added to the armoury of management tools and thereby expanded the range of alternative management options. This volume provide a synthesis of these empirical results, but other products too are disseminating the results to the wider public, such as the Atlas of Mediterranean Environments: The Desertification Context (Mairota et al. 1998) and more than 1000 articles and chapters in books and scientific journals. The limitations of direct observations make them rather inadequate for longer-term decision making. They are but snapshots in time and space. As samples, there are the added problems of representativeness and transportability. Plot experiments are tiny fractions of the landscape and rather artificial at that. Their use is mainly to expand and confirm our best guesses about what is happening and to inform better bases for prediction. To try to overcome these uncertainties, the project has adopted a deterministic model building and realization for hillslope processes at the local and regional scale (as described in Chapter 16) and for catchment and local runoff and sediment yield in the specific context of the Agri target area (Chapter 29). Here, as is almost invariably the case in model applications, the main constraint is data for both calibration and validation. Indeed, the co-operation and complementarity between field observers and laboratory modellers of different disciplines, cultures and philosophies has been one of the most heartening successes of the project,
Emerging Mosaics
423
which will have a lasting impact as the local ICCD plans develop in the context of this co-operation. It is through this channel that the complex physics of soil hydraulics can enter the decision-making process (Chapter 27). The bridging of science and society, a prerequisite for successful management, is better achieved through mutual co-operation in projects than by decree from paymasters, just as satisfying the needs of end users can only meaningfully occur by equal sharing of knowledge provided by involvement of local scientists who have the necessary insights to local problems and these need not only be the tillers of the soil. Optimally they will be broadly based scientists with wisdom and political influence at the many different decision-making levels. A primary requirement of the planning process is to have a clear knowledge of what has gone before (historical depth) and the capacity to recognize the likelihood of different responses of the different elements of the mosaic (geographical intuition), again stressing the importance of local experience as revealed by the development of mitigation needs in the Guadalent´ın and the Agri (Chapters 22, 24 and 25). The current format for this is to develop indicators that identify the degree of desertification and its spatial distribution as a shortcut to bypassing the modelling approach. Planners are always seeking the “single number” that will tell them what to do and where. The work that has been presented here on the basis of mainly empirical evidence, shows that the desertification problem is simply too complicated to be represented in the desired indicator fashion. This problem is recognized and discussed by Cammeraat et al. (Chapter 15). They follow Rapport and others in defining key indicators as follows: “An environmental attribute that, when measured, quantifies the magnitude of stress, habitat characteristics, degree of exposure to the stressor or degree of ecological response to the exposure.” Cammeraat et al. believe that a key indicator should also reflect linkages to other biotic and abiotic processes both at the same and higher scale levels. In this way, key indicators can be used for upscaling. The literature on indicators is almost as large as the list of indicators that are, or could be, used to identify desertification propensity or changes in desertification tendency.
5 EMERGENCE AND STABILITY Prompted by the chapters on indicators (Chapters 14 and 15), the final section of this concluding chapter attempts to think through this problem to identify a basis for the choice of indicators. The usual approach is to define the core requirement in such a way that it can be identified. Thus the ICCD defines desertification in terms of the climatic parameters of dry sub-humid, arid and semi-arid environments, by the ratio of rainfall to evapotranspiration. This is a kind of “legal” definition. If the mosaic component of interest does not fall into one of these classes, it is technically not covered by the Convention. Alternatively, in the Boolean type of definition, several indicators are used to classify the mosaic elements in a linear programming-type approach. These approaches invite the use of a hotchpotch of misunderstood, unknown, unmeasured or very subjective, often qualitative indicators that fail to capture either the dynamic or complicated character of the phenomenon. The substantive conceptual underpinning of the search for indicators must involve the following questions: • Is there any substantive value at which the rate of change of the process of degradation increases dramatically? • Is there any specific value of a variable at which there is a change from a negative feedback (constraining) to a positive feedback (unconstrained) behaviour in land degradation? The first is a step or catastrophe. The second is a bifurcation. Figure 30.1 shows schematic representations of these two cases and explicitly implies that vegetation cover is absolutely the key indicator that is required, probably in conjunction with measures of rainfall (both population and intensity) and soil water receptivity capacity. We can be sure that, in the mosaic of land uses, these variables do control the thresholds of erosion and the bifurcation of feedback type. They are easy, if tedious, to measure and can be captured at, and related to, different spatial scales. The first can be measured by remote sensing, the second by relating soil properties to geological properties, for which good maps are usually readily available.
424
Mediterranean Desertification
(a)
0% er
v
Soil erosion rate
co
Ve
t ge
at
io
n
30%
60% Rainfall /
rain day
(b)
Biomass (or plant cover)
Critical bifurcation
Trees Critical bifurcation Critical erosion thresholds 30% plant cover Matorral High erosion rates
Critical bifurcation
Degraded matorral Bare soil
Figure 30.1 (a) An example of a catastrophic representation of key indicators. The horizontal plane is the ‘‘control space’’ involving two key indicators for soil erosion (vertical axis). Note that the vegetation axis is decreasing away from the origin and that the rainfall intensity is increasing in the conventional manner. The surface is a three-dimensional representation of the equilibrium values of soil erosion with these key indicators. The whole surface tips down from right to left and is split in two surfaces, separated by a step (catastrophe) running from left to right. In the shaded area there are two possible rates of erosion, very high or very low. Throughout history, variations in the two indicator values move us through trajectories on the ‘‘control space’’ providing different values of erosion. These trajectories are called the ‘‘slow-dynamic’’ of the system. The response, in terms of soil erosion, is the ‘‘fast-dynamic’’. (b) Representation of the critical indicator actual evapotranspiration over rainfall and the vegetation response by biomass (cover) and type. As Eat /R increases to the right, the total biomass falls and the different types separate out at critical key values of the indicators
Emerging Mosaics
425
One issue that concerns us greatly is that the landscape mosaic we see today has emerged from some primitive initial condition that itself has been evolving over the millennia. The complicated network of political boundaries most readily illustrates the point. These emerged as the interfaces of political or tribal competitive forces from the “spatial soup of prehistory”. This emergence is stimulated and controlled by the pre-existing surface, rather in the way that specialized cell functions are mapped out in the developing foetus (Wolpert 1998). This attracts the question, were the mosaics mapped out by some precursors (chemical in developmental biology, geological in drainage basin, physical features in political boundaries) or does the emergence reflect some underlying forces that give rise to spatial patterning, as with the development of crystals? The 1930s model of desertification as the “advancing Sahara” has been abandoned in derision and is being replaced by a contagion model. In this theory, desertification is a process more akin to measles: an otherwise unblemished surface is broken into patches identified as degradation hotspots which then expand to produce widespread desertification, moving out from many centres, representing a coupled erosion–deposition patchwork (Pickup and Chewing 1986). The key question then becomes, given a relatively uniform and healthy ecological surface, how will new patches susceptible to degradation emerge, and what are the key indicators for this emergence? If the desertification hotspots do emerge as a mosaic, we would like to know how stable they are, when subjected to perturbations, natural or anthropogenic, such as fire, climate change or intensive grazing and are there indicators of this stability? Brunsden and Thornes (1979) argued that this could be developed through a Transient Form Ratio, an idea that has been further developed by Phillips (1999) for desertification. Brunsden and Thornes (1979) suggested that the state of the system could be defined by a Transient Form Ratio (TFR), TFR = ta /tf , analogous to the safety factor in engineering. Here ta and tf are the mean relaxation and recurrence times respectively for the perturbations: ta is the recovery time and tf the average time between the events. If recovery takes a long time and the events are quite frequent, TFR is greater than 1 and transient forms will prevail, but if TFR is less than 1, stable forms prevail. This ratio has the advantage of incorporating the effects of lagged response to the perturbation. In the Mediterranean, the inter-annual fluctuations of rainfall are very great. In the Guadalent´ın, a year with 160 mm of rainfall may be followed by one with 500 mm of rain. In an unstable system, the cover will “track” the rainfall. In a stable ecology, the mean vegetation adjusts to the mean long-term rainfall even though there are violent inter-annual variations. This concept is particularly important for the study of change. In the natural state of the Mediterranean environment, patches are continually subjected to perturbations and those that can resist change are said to be resilient. Systems that are not resilient not only respond to change but may do so in a non-linear fashion. A small pulse can lead to a large change, perhaps even to the destruction of an entire ecosystem. Or the production of many new stable states may occur and this may further complicate the mosaic of land uses. There are many possible pathways and many possible destinations for the trajectories of change. The management skill is to know which trajectory will be followed and where it will lead. Destabilizing an otherwise stable condition is the most serious outcome, for the ripple effect so produced could engulf an entire nation. Once the downward spiral of rural depopulation starts it is progressively more difficult to arrest. The identification of the thresholds between stable and unstable systems and of the trajectories that will be followed after the threshold is crossed should form the basis of indicators that are relevant to the management. There is little to be gained from forcing the system back into an unstable state. The application of these principles has been demonstrated in relation to land degradation, the impacts of grazing, the effects of climate change on dry Mediterranean plant communities and most recently to the restructuring of vegetation cover and erosion along climatic gradients.
6 IMPERATIVES AND PRIORITIES Toulmin (2001) has issued a clear call asking if the ICCD should be substantially reformed. The main shortfalls she outlines are that the International Convention model is inappropriate; that desertification is still a poorly understood problem; that no clear link has been established between desertification
426
Mediterranean Desertification
and climate change or proof found that desertification leads to global climate change; and, above all, that there is no serious funding specifically linked to the Convention. The unwillingness of the World Bank to engage in the problem has limited its attraction for all audiences. Add to these the serious criticism of the Convention made by Thomas and Middleton (1995) that it was promoted as a device to bolster UNEP’s political clout on the world stage, and we must clearly question its relevance to the southern European Annex IV countries. In this book, we have shown that the problem in Europe is quite different from that in Africa, even if the essential causes and indicators are the same. The Convention has, then, a different context and status. Because of the complicated spatial, cultural and historical mosaic, serious cases of land degradation with their associated vicious downward spiral of land abandonment, desertification and unsustainable agriculture are embedded in economically advanced economies with alternative buffering capacities. Nevertheless, unemployment, rural impoverishment and land abandonment can still be linked to land degradation. Following its recognition of the problem at the Mytilene Conference in 1985 (Fantechi and Margaris 1986), through a further conference in Crete (Balabanis et al. 1999) and through debate in the European Parliament as well as its research projects in Research Frameworks IV and V, the European Commission has recognized and supported the Convention, both indirectly and directly. Indeed it could be argued that without the Convention, the interests of the European Commission might not have emerged to push the governments of the Mediterranean European states into action. Here, too, the prospect of financial return appears to be a powerful incentive to recognize the regional problems in the states that are directly bound to desertification, as understood by the Convention. Moreover, as reflected by the European Parliament’s action, the people of Europe understand the problem’s complex and complicated dimensions and are prepared to support direct action through Agenda 2000, the revised CAP. There remains the need for a flexible adaptation of the policy to the mosaic, but the role of the Convention in promulgating change is not in question. If “the lights are dimmed” (Toulmin’s phrase), progress in this direction in Europe will be more difficult to achieve.
7
CONCLUSION
Desertification is a major issue in the Mediterranean environment. Intensification of production has caused agriculture to extend well beyond degradationally stable patches and, as the post-productionist ethos (reduction of subsidies, internalizing of environmental costs) takes hold under a new EU Common Agricultural Policy, these areas will become economically even more marginal, abandonment will occur and desertion will lead to desertification. The imperative is for nations and the EU to recognize this and the priority is to establish machinery to deal with it. Desertification is a serious issue even in the advanced economies of southern Europe, but its special characteristics have to be recognized. Action will be needed up and down the decision chain on an information basis that varies with the level. Farmers’ priorities and national priorities differ, though the common goals of intergenerational equity, stable land-use mosaics and political empowerment need to be addressed at all levels. Throughout, the strong spatial differentiation that is characteristic of the Mediterranean and the conditional stability of its elements will determine how the mosaic can and will evolve.
REFERENCES Balabanis P, Peter D, Ghazi A and Tsogas M (eds) (1999) Mediterranean Desertification. Research Results and Policy Implications. Proceedings of the International Conference, 29 October–1 November 1996, Crete, Greece. Volume 1, EUR 19303, Directorate General for Research, Luxembourg, pp. 5–16. Brandt CJ and Thornes JB (eds) (1996) Mediterranean Desertification and Land Use. John Wiley, Chichester. Brunsden D and Thornes JB (1979) Landscape sensitivity and change. Transactions of the Institute of British Geographers 4, 463–484. Fantechi R and Margaris NS (eds) (1986) Desertification in Europe. Reidel, Dordrecht. Francis CF and Thornes JB (1990) Matorral: erosion and reclamation. In J Albaladejo, MA Stocking and E Diaz (eds) Soil Degradation and Rehabilitation Under Mediterranean Conditions. CSIC, Madrid, pp. 87–117.
Emerging Mosaics
427
Ghazi A (1999) Biodiversity and global change. In P Balabanis, D Peter, A Ghazi and M Tsogas (eds) Mediterranean Desertification. Research Results and Policy Implications. Proceedings of the International Conference, 29 October–1 November 1996, Crete, Greece. Volume 1, EUR 19303, Directorate General for Research, Luxembourg, pp. 5–16. Mairota P, Thornes JB and Geeson N (eds) (1998) Atlas of Mediterranean Environments in Europe: The Desertification Context. John Wiley, Chichester. Obando JA (1997) Modelling the impact of land abandonment on runoff and soil erosion in a semi-arid catchment. PhD thesis, King’s College London. Phillips JD (1999) Earth Surface Systems: Complexity, Order and Scale. Blackwell, Oxford. Pickup G and Chewing VH (1986) Random field modelling of spatial variation in erosion and deposition in flat alluvial landscapes in arid central Australia. Ecological Modelling 33, 269–296. Seligman NG (1996) Management of Mediterranean grasslands. In J Hodgson and AW Illius (eds) The Ecology and Management of Grazing Systems. CAB International, Wallingford, UK, pp. 359–391. Thomas DSG and Middleton NJ (1995) Desertification: Exploding the Myth. John Wiley, Chichester. Toulmin C (2001) Lessons from the theatre. Should this be the final call for the Convention to Combat Desertification? International Institute for Environment and Development, for the World Summit on Sustainable Development (Johannesburg 2002), London. United Nations (1994) International Convention on Combating Desertification. United Nations, Paris. Wolpert L (1998) Principles of Development. Oxford University Press, Oxford.
Glossary
Ablation rate: quantity of soil eroded from slopes and transported by running water. Generally expressed in t ha−1 year−1 . Absorption feature: a range of wavelengths (or frequencies) in the electromagnetic spectrum within which radiant energy is absorbed by a substance. Aggregate distribution: distribution of classes of soil aggregates (particles adhering to one another) according to their size. Albedo: proportion of incident solar radiation reflected by the clouds surrounding the Earth. Anchor station: an observation site at which quantities are measured that are needed to calibrate measurements made from satellites and to validate the information inferred from these measurements. Aqueduct: an artificial surface channel for conveying water. Aquifer: a geological formation of water-bearing rock with sufficient porosity and permeability to yield economic supplies of groundwater. Badland: an area where gullies adjoin each other and cover all, or nearly all, of the surface. Barrilla: halophytic plants (mainly Halogeton sativus). The burned ashes were used as raw material for soap production. It was an industrial crop of great importance in the 17th century right across the Guadalent´ın Basin. Biancane: a form of erosion with a typical dome-shaped configuration and a radial drainage network (Italy). Blown sand dunes: a hill or ridge of blown sand piled up by the wind. Boqueras: a traditional system of south-east Spain to take sporadic flow from ramblas (ephemeral flow channels) to crops. Bradiseysm: slow movement, either raising or lowering of the soil in localized areas of the Earth’s crust. Calanchi (badlands): erosion form that occurs in blue clay without vegetation (Italy). Calcixeroll: mollisol (i.e. with a dark organic-rich surface horizon) in a xeric soil moisture regime with a calcic or gypsic horizon within 150 cm below the soil surface (Soil Survey Staff 1975). Canestrato: cheese made from the milk of goats or sheep. Catena: a repeated sequence of soil profiles that is geographically related to and associated with relief features. Climate scenarios: internally consistent pictures of a plausible future climate; not predictions of future climate. Coltura mista: different crops (annual and perennial) cultivated on the same plot. Compaction: the development of a dense, compact surface soil layer (e.g. due to cultivation with heavy machinery, or overgrazing), characterized by a much lower permeability so impeding the movement of water and air, and the growth of plant roots. Confined aquifer: groundwater reservoir overlain and underlain by impervious or almost impervious rock formations. Coppice (ceduo): forestry stand originating primarily from sprouts (Italy). Coppice with standards (ceduo matricinato): method of reproduction in which selected trees arising from either seedlings or sprouts are maintained as standards above a simple coppice stand (Italy). Cortijo: farmhouse in southern Spain. Crusting: development of a surface layer on soils ranging in thickness from a few millimetres to a few centimetres, which is more compact, hard and brittle when dry than the material immediately beneath it (see also compaction).
430
Mediterranean Desertification
Desamortizaci´on: laws promoted in Spain by liberals in order to sell municipal, communal and church lands to private owners. Differing-aged stand (soprassuolo disetaneo): stand where trees of at least three age classes are intimately intermingled in the same area (Italy). Digital elevation model: an ordered collection of topographic elevation data for a particular area. Drill sowing (semina a righe): sowing of seeds in parallel, uniform bands that run the length of the seedbed (Italy). Esparto: the common name of the perennial grass Stipa tenacissima, an indigenous plant used for fibre production since at least the Bronze Age. Eustatic movements: raising or lowering, on a global scale, of the average sea level, mainly due to the melting or freezing of the polar ice-caps. Evapotranspiration: the combined loss of water from a given area and during a specified period of time, by evaporation from the soil surface and by transpiration from plants. Even-aged stand (soprassuolo coetaneo): stand where all trees are the same age or at least in the same age class (Italy). Field capacity: the percentage of water remaining in the soil following saturation and free drainage. Fine earth bulk density (Bdfe): the mass of dry soil per unit of volume excluding the volume of rock fragments. Forest index (indice di boscosit`a ): the percentage of the total surface covered by forest (Italy). General circulation models (GCMs): complex, three-dimensional computer-based models of the atmospheric circulation developed from numerical forecasting models and used to investigate future climate change. Geographical information system (GIS): system with digitized computer maps; these maps can be combined with other maps, or can be processed. Greenhouse effect: greenhouse gases in the atmosphere, such as carbon dioxide, are largely transparent to short-wave solar radiation, but absorb long-wave radiation from the Earth and so maintain the Earth at a temperature higher than it would be in their absence. This natural effect is enhanced by the release of greenhouse gases from human activities such as the burning of fossil fuel. Groundwater: subsurface water that occurs beneath the water table, occupying the pores of the soils and geological formations that are fully saturated. Gullies: deep water-worn channels, cutting through soil into weathered material and/or rock. Heat wave: thermal event during which the air temperature increases to several degrees above the normal value. High forest (fustaia): stand originating from seed (Italy). Infrared region: portion of the electromagnetic spectrum just beyond the red end of the visible spectrum, such as radiation emitted by a hot body. Interception: fraction of rainfall lost to evapotranspiration, due to retention of the canopy. Interception is equal to rainfall minus throughfall. LAI: leaf area index. The area of leaves above a given area of ground (usually one square metre). Land abandonment: land that has been converted from any form of agricultural production or from areas that have been heavily grazed, and then left and allowed to revegetate naturally. Complete abandonment implies that the land has been left to return to its natural state without any human influence, directly or from livestock. LANDSAT: acronym indicating a series of Earth resources scanning satellites; the data recorded by the sensors are widely used for land resources assessment. Latent heat flux: the heat of evaporation that is carried with the flow of moist air. Leaf water potential (): the energy status of the water contained in leaves. Loess: fine-grained, permeable, unstratified Pleistocene aeolian deposit. Macchia mediterranea: bushy vegetation made up of shrubs and low trees (Italy). See also matorral, maquis, Mediterranean scrubland. Maquis: a vegetation type of the Mediterranean area, mainly composed of evergreen broadleaved shrubs, less than about 5 m high (France).
Glossary
431
Matorral : type of shrubby vegetation cover found in south-east Spain, comparable to the French maquis and garrigue and macchia mediterranea. Matorral arbusto is thicket and matorral matas is brushwood. Mediterranean scrubland: Mediterranean scrublands have resulted from the interaction between natural factors and very ancient human disturbance. The main control on the ecosystem is the annual summer drought. This intense hydric stress imposes a set of adaptations (such as sclerophyllous leaves) and characteristic structures of the plants. Meteorological bomb: depression with a rate of pressure fall of 1 hPa h−1 or more, lasting at least 24 hours at the latitude of 60 ◦ N (17 hPa in 24 hours at the average Mediterranean latitude of 38 ◦ N). Montado: savanna in Portugal, typically with cork oaks (Quercus suber), Quercus rotundifolia and some cultivated fields. Phreatic aquifer: porous water-bearing formation in which the groundwater table forms the upper boundary. Phrygana: undershrubs or dwarf shrubs, e.g. thyme, sage, the many species of broom, and species of Cistus and Phlomis. Undershrubs are not potential trees; they are short-lived and reproduce by seed. Rambla: ephemeral water course (Spain). Recharge: infiltration of the rain, first into the soil, then deeper towards aquifers. Rock fragments: mineral particles larger than 2 mm in diameter including all sizes that have horizontal dimensions less than the size of a pedon (Poesen and Lavee 1994). Rotation (turno): the period of years required to grow a crop of timber to its specific condition of economic or natural maturity (Italy). Saladeres: ecosystems typical of arid and semi-arid environments that can be termed cryptowetlands, as water is very rare over the soil surface but is the driving force responsible for their origin and the accumulation of salts in the soil which is an essential characteristic. Halophytic plants are characteristic (Spain). Salt water intrusion: penetration of salt water into freshwater aquifers under the influence of groundwater development. Savanna: grassland and/or shrubs with scattered trees that do not form a complete canopy. Scenario: internally consistent picture of a plausible future state (e.g. a model simulation of a possible future state, a projection of future climates). A scenario is not a prediction of a future state. Seed tree (riserva): tree left standing singly or in groups for the purpose of furnishing seed to restock the cleared area naturally (Italy). Silvicultural system (forma di trattamento): a planned programme of silvicultural treatment during the whole life of a stand (Italy). Spacing (sesto d’impianto): number and distribution of individual trees in artificial reproduction (Italy). Strip (group) shelterwood method (tagli successivi a strisce o a gruppi ): application of the shelterwood method in strips and groups or patches (Italy). Tending fellings (cure colturali, tagli colturali ): various cuttings with the object of the improvement of the existing stand (Italy). Terrace: narrow surface plane or with shallow slope build-up following the contour lines, with the aim of increasing the water-holding capacity and accumulating fertile soil. Thinnings (diradamenti ): intermediate cuttings aimed primarily at controlling the growth of stands through adjustments in stand density; cuttings in immature stands in order to stimulate growth of the trees that remain (Italy). Total bulk density (BDt): the mass of dry soil per unit of volume including the volume of rock fragments. Uniform shelterwood method (trattamento a tagli successivi uniformi ): cuttings uniformly applied over the entire stand (Italy). Wildings (selvaggione): natural seedling (Italy). Woodlands: portion of farm area devoted to tall forest, coppices and maquis (ISTAT 1982).
432
Mediterranean Desertification
REFERENCES ISTAT (1982) XIX Censimento Nazionale dell’ Agricoltura Italiana. Fascicoli provinciali Matera e Potenza. Poesen J and Lavee H (1994) Rock fragments in topsoils: significance and processes. Catena 23: 1–28. Soil Survey Staff (1975) Soil Taxonomy: A Basic System of Soil Classification for Making and Interpreting Soil Surveys. USDA-SCS Agricultural Handbook 436, US Government Print Office, Washington, DC.
Index
Abandoned fields, vulnerability to erosion 11 Abbott, M.B. 370 Abies alba 385, 391 Abruzzo, remote-sensing characterization of climate 47–54 Acer lobelii 391 Adams, R.M. 164 Adaptation of Mediterranean ecosystem to survive fire 83–86 Aegean islands 87 Afforestation in Guadalent´ın Valley 393–394, 420 Agenda 21, 10 Agenda 2000 6, 7 Agri Valley 321–330, 419, 420, 422 data assembling and processing for SHE model 400–404 forestry in 385–395 Ganno Barrage 402 geology of 322–323 modelling basin hydrology 397–422 modelling hydrology and sediment yield 397–416 physical environment of 321–325 population 361–363 socio-economic aspects of 325–330, 361–367 soil erosion and land degradation 347–359 sustainable agriculture 331–346 Agriculture agricultural income, Guadalent´ın basin 294 Common Agricultural Policy 6, 7 MEDRUSH model application 203, 215 Alcantarilla climate station 38 Alentejo, Portugal 178–179 Algeria, forest fires 85 Almeria 112 Alqueria functions in 4–6 indicators and response units 178, 180–183 soil erosion 7–9 water supply 6 yield of major crops 5 Anderson, G.W. 347 Animal husbandry 181 Anthyllis cytisoides 99, 114 Apulia salinization 167, 171 Araus, J.L. 38 Arbutus unedo 34, 45 Archaeomedes Project 5
Archibold, O.W. 33 Arianatsu, M. 85 Aridification in Hungary 152–162 Aridity, increasing 38 in Val d’Agri National Park 365 Aridity indices, De Martonne 49 Aru, A. 347 Ashcroft, G.L. 137 Asphodel deserts 86 Asphodelus microcarpus 86 Aswan Dam, and irrigation 161 Atmospheric CO2 change, impacts on vegetation 33–46 Ayers, R.S. 168 Azimonti, D. 365 Balabanis, P. 10 Barbera, G.G. 295 Barbieri, R. 168, 348 Barilleras 295 Basilicata 361 Batha 83 Baudry, J. 7, 10, 270 Bautista, Martin, J. 296 Beerling, D.J. 44 Begon, M. 119 Bell Adell, M.C. 291 Below ground impacts of plants on bioengineering properties 98–99 Betts, R.A. 33 Bevan, K. 381, 399 Bianco, S. 361 Bifurcation theory 423, 424 Binley, A. 399 Biodiversity 168, 393 Bioengineering principles and desertification mitigation 93–105 Biomass 35 above ground in asphodel deserts 86 effects of rock fragments 141 simulated 38, 42 Birdlife International 6 Biswell, H. 84 Blaney Criddle formula 402 Blevins, R.L. 132 Boer, M. 11 Bond, J.J. 137 Bonin, G. 386 Bouma, J. 372
434
Index
Bray–Curtis ordination method Bretherton, F.P. 206 Brown, L.C. 96 Bruce, R.R. 371 Brunsden, D. 425 Bultot, F. 164 Burke, S. 5, 177
110, 112
C++ used in MEDRUSH 203 Calanchi 386 Calder, I. 414 Caliandro, A. 171 Caliandro, F. 167 California 109 sites for landscape characteristics 110 Cameraat, L.H. 423 Ca˜nada de Cazorla, response units 188–190 Ca˜nada Hermosa, response units 188 Caneva, G. 367 Cantore, V. 386 Carbon dioxide (CO2 ), simulated effects of elevated on monthly plant parameters 41 Caroti, L. 49 Carpathian Basin, soil salinization 169 Cataldo, L. 361 Catastrophe theory 423, 424 Cavazza, L. 169, 349 Celicio, P. 166, 167 Cenological surveys, in Hungary 158 Chanduvi, F. 169 Channel routing component in MEDRUSH model 219–221 Chaparral 274 Chewing, V.H. 425 Childs, S.W. 132, 137 Chile 109 Fundo Santa Laura 110 Chios, fires 88 Choresh 83 Chortal, rambla de 269 Ciollaro, G. 372, 376 Cistus species 85 Clark, S.C. 87 Climate change potential effects of rising CO2 on Mediterranean vegetation 33–44 CO2 -I impacts on vegetation 33–46 inter-annual seasonal variability on NPP 43 Coker, P. 110 Colino, J. 291 Commonwealth Association of Universities 275 Conese, C. 53 Convallario-Quercetum 157 Coppices 390, 393 Corleto Perticara 373, 402 Crop yield in Alqueria area (Guadalent´ın) 181 Crop yield reduction due to salinity 168
Crops in Guadalent´ın Basin Cyprus, saline soils 167
298
Dargie, T. 112 De Cillis, E. 169 De Falco, E. 357 De Martonne Aridity Index 324 De Miguel, J.M. 8 De Pascale, S. 168 De Ploey, J. 132 Deforestation 269, 275 and salinization 165 Val d’Agri 325, 362, 390 Degradation paths, indicated by plant species 114 Dehesa 7 Del Prete, M. 323 Denmead, O.T. 402 Desertification, definition of 6 Desertification response units 184 Diamond, S. 8 Digital elevation component in MEDRUSH model 204 Dimase, A.C. 323 Dimitrakopoulos, A. 89 Dirksen, C. 371 Dombois, E. 111 Doorenbos, J. 142 Droughts 8 and soil salinity 164 drought, deciduous shrubs 33, 42 in Greece 8 in Guadalent´ın 296 in Hungary 152 partially offset by rising CO2 38 Dudal, R. 347 Durum wheat 326, 333, 354, 365, 373 Earthquakes 363 Economic development in south-west Spain 269 Egypt, Nile delta and salinity 169 Ehleringer, J.R. 33 Ellenberg, S. 111 Elwell, H.A. 96 Emigration 363 Val d’Agri 326 Environmental risk of erosion 12–13 Environmentally sensitive areas, 177–185, 394, 395 comparing different target areas 178–180 sensitivity to land degradation 178–180 Erosion rates, see Soil; Soil erosion rates Escadafel, R. 169 ESP exchangeable sodium percentage 164, 170 Esparto grass 295 see also Stipa tenacissima Euphorbia acanthothamus 85 European Commission xv
Index Evapotranspiration 424 in field with rock fragments 138–141 in laboratory with rock fragments 137 Ewan, J. 399 Experimental field plots and sites Athens, and rock fragment studies 135–137 Guadalent´ın, for bioengineering experiments 93–105 on impact of increased CO2 33 Sele plain, soil salinity 170 Val d’Agri 333, 350, 370, 373 Fagus sylvatica 385, 391, 393 Feddes, R. 381, 402 Ferrara, A. 323, 324 Fire disturbance regimes, fire and grazing 113–116 grazing and fires in Greece 109–118 landscape protection from 83–92 in Val d’Agri 348, 391, 393 on Lesbos 420 plant species along fire/grazing intensity gradient 115–116 recovery time 8 Flint, D.L. 132, 137 Fonseca, C.D. 361 Forest Borbonic law of 1826 (Italy) 388 canopy density 387 forest index change 294 Guadalent´ın Basin 299 phytoclimatic classification 387 Francis, C.F. 93, 96 Franzluebbers, A.C. 332 Frere, M. 142 Functional performance indicators 181 Functional trends, of nitrogen and phosphorous 113 Functions resisting desertification 181 Galillee, saline soils 167 Galligani, U. 323 Gannano Barrage 403–409 Gatt, Uruguayan round 6 General Circulation model, and Hungary 152 Geographical information system 304 for environmental sensitivity 394 for land capability in Hungary 156–157, 159 Ghazi, A. 419 Gil-Olcina, A. 295 Glenn-Lewin, D.C. 270, 274 Godron, M. 274 Gonzalo Rebollar, J.L. 306 Goodess, C. 8, 44 GRASS, integrated with MEDRUSH model 203 Grayson, R.B. 381 Grazing and fire, Greece 86–92, 118–119 effects on shrub morphology 114
435
grazing level, responses to 114 intensity gradient 114 intensity-plant species 115–116 Larissa, field study 112–113 overgrazing 88 trampling 85 Greece Chios, fires in 88–90, 421 grazing and fires 88, 109–118 Larissa, study of grazing on composition 112–113 Lesbos 8, 87 saline soils 167 Greenhouse effect 161 Griesbach, A. 84 Groundwater Guadalent´ın 296–298 Hungary 153–155 Guadalent´ın Basin 179, 419, 420, 421 agriculture 292 changing social and economic conditions 289–301 changes in income 294 economic conditions 292–299 indicators of desertification 177–187 land-use changes 295, 296 plan to combat desertification 303–317, 420 population 289–292 compared to Murcia region 289 history 289–290 recent (20th century) changes 296 sensitivity analysis 178 water resources 298 Guardia Perticara 328, 332, 349 Gupta–Larson method 379–380 Hamad, O. 414 Hamdy, A. 269 Hanks, R.J. 137 Hanson, C.T. 132 H¨attenschwiler, S. 34, 35, 40 Herbaceous plants, post-fire 85 Hillel, D. 137 Horsebean crop 333, 354 Horton equation for infiltration 369 Hudson, N. 93 Hungary aridification of 152–162 climate change 152–153 Great Hungarian Plain Danube–Tisza interfluve 156, 158 Kiskunag National Park 158 land use changes 159–161 groundwater changes in 153–156 impact on land capability 156–158 index of continentality 158 index of relative heat demand 158 soil dynamics 158–159
436
Index
Hunting 181 Hutchinson, J. 323 Hydraulic conductivity, and rock fragments 132 Hydraulic roughness, and rock fragments 132 Hydrology characterization of soil hydraulic properties 369–383 hillslope scale variability 377–381 in the Agri Valley, hydrological data for different tillage 336, 343 runoff 356, 357 I.C.C.D., see UNEP, International Convention on Combatting Desertification ICONA, Wildland Vegetation Map of Spain 304, 306 Imeson, A. 99 Indicators, desertification environmental characteristics underpinning selection of 177–187 infiltration rate in MEDRUSH model 206 Ingelmo-Sanchez, F. 132 IPPC, International Panel on Climate Change 38 Irrigation drip171 Egypt 9 Greece 9 Guadalent´ın Basin 181, 296, 298, 300 Italy 9 Puglia 326, 327 soil salinization 165, 166, 169, 171 Val d’Agri 329 Isotope 18 O 166 Israel, saline soil areas 167 Johnson, M.G. 332 Jones, M.B. 38 Kabat, P. 381 Kamar, M.J. 137 Kemper, W.D. 132 Kent, M. 110 Kern, J.S. 332 Key indicators 176 and stability 423 Kirkby, M.J. 179 Klute, A. 371 Kool, J.B. 371 Kosmas, C. 8, 178 Kostiakov equation 369 Land abandonment 7 different degrees of 269 Guadalent´ın Basin 269–276, 300 impact on regeneration of semi-natural vegetation 269–276 Land capability change under aridification, Hungary 156–167 Land care 6
Landi, R. 347 Landscape characteristics for MEDALUS II sites and sites in Chile and California 110 Landscape functions 179–182 Landslides 323, 357, 386 Lang, R.D. 96 Larotonda, A. 365 Lavee, H. 136, 137 Le Houerou, H.N. 85, 86 Leaching requirement 170 Leaf area index 35, 37, 345 Leggett, J. 38 Lesbos, 180, 419, 420 Libya, soil salinization 169 Linsalata, D. 348, 355 Litter 42 Lukey, B.T. 402, 408 Luxmore, R.J. 371 Machia 385 MacRae, R.J. 332 Magier, J. 132 Mairota, P. 422 Malaria 361, 362 Management actions, map 306 Management actions, typology of in Guadalent´ın 307 Mancini, F. 323 Maquis 83 and soil salinity 165, 168 Margaris, N.S. 85, 88 Martinez-Carrion, J.M. 295 Marzi, V. 348 Mass movement 332 Massafra, A. 363 Mazzanti, B. 402 McCaffrey, L.A. H. 96 McSharry, approach to reform of CAP 6 MEDALUS II xv MEDRUSH model 203–227 conceptual basis of 203 microtopography in 204, 207, 208 Mehuys, G.Q. 332 Mesopotamia, salt accumulation 161 Mezzogiorno (Italy) 391 Middleton, N.J. 426 Miglietta, F. 34 Migration, consequences in Val d’Agri 363 Miles, J. 274 Mitchell, J.F. B. 38 Modelling large basin hydrology and sediment yield with sparse data: the Agri Basin 397–416 MEDRUSH basin-scale physically based model for forecasting runoff and sediment yield 210–213, 227 changes in surface roughness over time 213
Index construction of sub-basins and representation of flow strips 214, 223 grain-size effects in 209 implementation 224 sediment transport by wash processes 211 sediment transport in general 210 sub-basin component 205–206 mountain grassland 390 SHE model 397–416 uncertainty in 398–399 vegetation simulation models 35–40 water balance model of Doorenbos and Pruit 142 Montero de Burgos, J.L. 306 Mooney, H.A. 33, 39 Moustakas, N. 136 Mulching with gravel experimentally 142–143 Muller, M.J. 35 Munoz Bravo, J. 296 Murcia, climate change 35, 36 Naveh, Z. 8, 85 Negev, central saline soils 167–168 Neolithic soil salinization 163 Net primary productivity 35
437
Pinus spp. 85, 110, 117 Pinus halepensis 389, 391, 420 used in restoration 306 Pinus nigra 388, 391, 392, 420 Pinus pinaster 420 Pinus pinea 389 Pinus radiata 389 Pinus sylvestris 391 Pistacea lentiscus 34 Plantago sp. 99 Plough layer, simulated 133, 137 Policy, desertification control 308 Population by economic sectors 292 demographic projections 291 Guadalent´ın Basin 289–290 Postiglione, L. 163–175, 348, 349, 422 Pre-dawn water potential 34 Pressures responsible for desertification 181 Prez, C. 289 Primary productivity, elevated CO2 39 Pruit, W.J. 142 Puigdefabrigas, J. 114
Oechel, W.C. 85 Oil, recent discovery of in Val d’Agri 365 O’Riordan, T. 5 Ortin, J. 289 Overgrazing and fire 83 Guadalent´ın 269, 272, 275 salinization 165 Overland flow 206
Quaternary, eustatic movement and saline springs 166–167 Quercetum, Hungary 157 Quercus cerris 385, 388, 390, 393 Quercus coccifera 112, 113 Quercus ilex 34, 45, 389 Quercus pubescens 388, 389, 390 Quercus rubra 391 Quercus suber, elevated CO2 33 Quercus sylvatica 393 Quinton, J. 399
Palimpsest 10 Palutikof, J. 8, 38 Pannonian (endemic) species 158 Papadopoulos, I. 167 Parkin, G. 399 Parry, M. 131 Peet, R.K. 274 Penman, H.L. 142 Perniola, M. 168 Perpignan, climate change 35 Pertusillo Dam, Val d’Agri 321, 323, 325, 403–409 Perz-Picazo, M.T. 295a Peter, D. 10, 131 Phlomis 85 Phlomis fructicosa 85 Photosynthesis 34, 40, 42, 85 Phrygana 83, 85, 110, 420 Phyllirea angustifolia 34 Phyllite, vegetation regeneration 272 Phyloxera 296 Pickup, G. 425
Ragab, R. 171 Rain-fed crops 181 Rain-splash 206 Rainfall and rising CO2 38 characteristics of Mediterranean 8 in Val d’Agri 323 inter-annual variability and N.P.P. 42 simulator in bioengineering studies 96 Rainflow 206 Ravina, I. 132 Rawls and Brakensiek method 379 Rebeiro 6 Reforestation and changes in plant species composition census in Greece 88, 116 Refsgaard, J.C. 397 Regeneration of plant cover 269–276 Remote sensing 394, 422 application to Guadalent´ın Target Area 127–128
116
438
Index
Remote sensing (continued) aridity maps from 53 Lambertian reflectance model 123 LANDSAT-5 TM 124 LANDSAT image of Agri Basin 389 linear spectral mixture modelling 119–122, 124 modelling physical scenes 122 NDVI 47, 119 NDVI, new index and De Martonne and Thornthwaite indices 52 NOAA-AVHRR 47–54 use of NOAA-HVRR NDVI data for climatic characterization 47–54 vegetation cover assessment in Mediterranean semi-arid landscapes 47–55 vegetation in relation to NDVI 51 Renard, K.G. 98 Representative flow strips 215 in MEDRUSH model 204 Reproductive tillers, and grass establishment 112 Resilience 177 Respiration in plants 42 Response unit methodology 178 Restoration technical design 306, 309 Retama sp. 114 Reynolds Creek catchment, Idaho 410 Rice 365 Richards, I.D. 93 Richards equation 370, 372, 381 Rill-wash 206 Rimbaud catchment, France 410 Rio Conference 5 River Agri 321, 322, 365, 370 Ebro 9 Guadalent´ın 420 Nile 161 Rhˆone 9 Sauro 327, 365, 370, 373 Segura 9, 296 Sele 166 Tajo 9, 296, 420 Rock fragments 131–145, 422 effects on cereals 142 effects on evapotranspiration 134 in soils on conglomerates 142 in soils on shales and sandstones 142 in soils on marls 142 laboratory tests on soils 138 Rodrigues, V. 6 Rojo-Serrano, L. 420 Roman settlement, Val d’Agri 361 Romero Diaz, A. 412 Romkens, M.J. 133 Root density 333 Root studies in relation to infiltration 99 Rossi Doria, M. 361
Routing of water and sediment in MEDRUSH model 204, 209 distribution of overland and subsurface flow 206 Ruggiero, C. 168 Ruhe, R.V. 348 Runoff 356 from plots 136 Sacropoterium spinosum 85 Salinization causes 165–167 definition, of sodic and saline and sodic-saline soils 164 defloculation of soils 167 extent of problem and impact 167–168 historical perspective 163–164 in Hungary 159 in the Mediterranean 163–175 management of sodicity and salinity 168–171 properties of sodic and saline soils (table) 164 Salsola kali 295 Salsola longifolia 295 primary 165 Sanchez, P. 289 Sardinia 419, 420 Sarno (southern Italy) 1998 disaster 165 Scafati (Camagna region, Italy) evapotranspiration and rainfall 166 Scale effects in SHETRAN applications 411 Scarascia-Mugnozza, G.E. 34 Schertz, D.L. 165, 347 Schimel, D. 38 Sclerophyll tissue 109 Sclerophyllous shrubs and CO2 33–46 Sea water evaporation 164 Sea water infiltration 165 Seasonal variations, response of vegetation to higher CO2 36, 40 Segal, M. 38 Sele river plain (southern Italy) salinization 166–167, 170 Set-aside 331, 332 Seville, climate change 35, 36 Shales, vegetation regeneration 272 Shalhevet, J. 171 Shantz, H.L. 84 Shaw, R.T. 402 SHE/SHETRAN 397–416 Sheep 361 Sherwood, J. 414 SHETRAN model 397–416 data requirements and assembly 399 description 398 parameters of 398 Simanton, R.J. 97 Similarity (vegetational) index 110 Simulated rainfall in laboratory 133
439
Index Skourtos, M. 89 Smith, D.D. 349, 369 Smith, R.M. 347 Smith, T.R. 206 Socio-economic functions and indicators 178 Soil aggregate stability 132, 178 bulk density of fine earth 133 characteristics of in Val d’Agri 333–336 compaction, with rock fragments 132 cracking 170 dynamics under aridification 158 effects of tillage systems 353–355 erosion and land degradation in Val d’Agri 347–359 erosion and rock fragments 131, 132–137 erosion and sediment yield modelled in SHETRAN 408–409 erosion in Alqueria area 186–188 hydraulic properties in Val d’Agri 369–383 loss of fertility due to erosion 347 moisture regime with rock fragments 140, 141 organic matter in field plots 355 pedo-transfer functions 377 physical degradation with rock fragments 132–135 properties in MEDRUSH model 204 quality, indicators of 178 roughness with rock fragments 133 salinization 167–168 vertic soils 349 water potential 34, 37, 42 Soil erosion rates 178 effects of land use on, under Mediterranean conditions 57–71 effects of plant properties 96 Italy 348 plots under durum wheat 357 simulated with SHETRAN 408 socio-economic risk of erosion of the landscape unit 188 with rainfall simulation experiments under different vegetation types by season 97 Solonchak 167 Somez, B. 167 Sørensen index 110, 111, 114 values of 116 Spain Aragon 9 Extremadura 7 Spanish National Hydrological Plan 9 Stability criterion, Smith and Bretherton 206 Stamey, W.L. 347 Stewart, D. 110, 111 Stipa bromoides 112 Stipa tenacissima 95–99, 295 Stocking, M.A. 96
Stocking rate 86 Stomatal closure 34 Stomatal conductance 34 Strickler overland flow resistance coefficient 404, 411 Stunell, J. 414 Sub-basins in MEDRUSH model 204 Subsidy of fodder costs 91 Surface sealing prevented 136 Sustainability 5 sustainable land management 419 Szabolcs, I. 165 Target areas Agri Basin, southern Italy 319–417 Guadalent´ın Basin, Spain 231–303 Tedescchi, P. 168 Terraces 7 abandoned 93–94 accumulation of eroded sediments 9 agriculture 67 reforestation and erosion 12 small bench terraces 93, 391 Thomas, D. 426 Thornes, J.B. 93, 96, 177 Thornthwaite index of humidity 49, 50 Thymus capitata 85 Tillage methods impacts on hydrology, erosion and soil chemistry 333–345 impacts on erosion 352–355 Tilman, D. 274 Toderi, G. 347 Topographic wetness index 206 Toulmin, C. 425 Tourism 327 Trabaud, L. 92 Tractor fuel costs 332 Transfer functions for time of water flow through a reach in MEDRUSH model 221 Transient form ratio 425 Transpiration 34, 36 Trunk volume 34 Tunisia, salinity 169 Turkes, M. 9 Turkey anticyclonic activity 9 forest fires 85 saline soils 167 UNCED 5 UNEP 5 International Convention on Biodiversity International Convention on Combatting Desertification 423 Universal soil loss equation 369 Van der Leeuw, S. 5 Van der Maarel, E. 270, 274
6
440
Index
Van Genuchten, M.Th. 372, 379 Van Wesemael, B. 133 Vegetation adaptive strategies to help post-fire recovery 85 and erosion 422 at bioengineering study sites in Guadalent´ın 95 bioengineering principles 93–105 changes, Guadalent´ın 269 changes on Danube–Tisza interfluve, Hungary 157–159 data collection 270 equilibrium after abandonment 272–274 general scheme for vegetation in relation to climate regime 51 growth component in MEDRUSH model 204, 216–219 in SHE model 42, 401–402, 412 regeneration after abandonment 272–275 root density and infiltration rates 99 sclerophyllous shrubs 34, 83 selection of 100 species for revegetation (table) 100–104 senescence 274 soil and water conservation, indicators of ecosystem function and structure 184 structure 33–46 surface cover characteristics, Guadalent´ın 96, 271–273 used for restoration 306 wildland vegetation map of Spain 304 Verity, G.E. 347
Villani, P. 363 Voisey, H. 5 Volatile oils in phrygana
84
Walling, D.E. 412 Water availability and elevated CO2 39, 42 Water conservation strategies, and rock fragments 131 Water potential of Stipa bromoides and Quercus coccifera in August 113 Water resources Guadalent´ın 296, 298, 300 in Val d’Agri 326 planning with SHE model 410 water budget parameters 152 water deficit in Hungary 152 water retention curves, modelled 375 watershed restoration projects database 306 Water use efficiency of plants 34, 35, 38 Water vapour adsorption with rock fragments 141 Webber, D.J. 110, 111 Weed control 354 Westcot, D.W. 168 Whittaker, R.H. 8, 274 Wicks, J.M. 411 Willis, W.O. 137 Wischmeir, W.H. 349, 369 Wolpert, L. 425 Wood, E.F. 369 Woodward, I. 8 Yevyevich, V. 370 Zarzilla de Ramos 179
Plate 1MFalse-colour composition derived from the LSMM corresponding to 7 April 1993. It depicts the fractional coverage in the Guadalentín Basin of soil (red), crops (green) and natural vegetation (blue) (see Chapter 10)
Plate 2MLocation of the Guadalentín Basin in the south-east of the Iberian Peninsula. This satellite image in false colour clearly emphasizes the aridity of the area (see Chapter 17)
Plate 3M(A) Vegetation and (B) lithology classifications of the Guadalentín Basin as interpreted from spring and autumn Landsat TM imagery (see Chapter 20)
Plate 4MVegetation and land-use map, part of the management plan to combat desertification for the Guadalentín Basin (DGCONA 1995). Reproduced by permission of John Wiley & Sons Ltd, from Mairota, P. et al. (1998) (see Chapter 22)
Plate 5MManagement actions map, part of the management plan to combat desertification for the Guadalentín Basin (DGCONA 1995). Reproduced by permission of John Wiley & Sons Ltd, from Mairota, P. et al. (1998) (see Chapter 22)
