Perspectives for Agronomy (Developments in Crop Science, Volume 25)

PERSPECTIVES FOR AGRONOMY Adopting Ecological Principles and Managing Resource Use Developments in Crop Science Volum...

Author: M.K. van Ittersum | S.C. van de Geijn

34 downloads 1074 Views 24MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

PERSPECTIVES FOR AGRONOMY Adopting Ecological Principles and Managing Resource Use

Developments in Crop Science Volume 1 Oil Palm Research, edited by R.H.V. Corley, J.J. Hardon and B.J. Wood Volume 2 Application of Mutation Breeding Methods in the Improvement of Vegetatively Propagated Crops, by C. Broertjes and A.M. van Harten Volume 3 Wheat Studies, by H. Kihara Volume 4 The Biology and Control of Weeds in Sugarcane, by S.Y. Peng Volume 5 Plant Tissue Culture" Theory and Practice, by S.S. Bhojwani and M.K. Razdan Volume 6 Trace Elements in Plants, by M.Ya. Shkolnik Volume 7 Biology of Rice, edited by S. Tsunoda and N. Takahashi Volume 8 Processes and Control of Plant Senescence, Y.Y. Leshem, A.H. Halevy and C. Frenkel Volume 9 Taigu Genic Male-Sterile Wheat, edited by Deng Jingyang Volume 10 Cultivating Edible Fungi, edited by P.J. Wuest, D.J. Royse and R.B. Beelman Volume 11 Sugar Improvement through Breeding, edited by D.J. Heinz Volume 12 Applied Mutation Breeding for Vegetatively Propagated Crops, by C. Broertjes and A.M. van Harten Volume 13 Yield Formation in the Main Field Crops, by J. Petr, V. Cern~, and L. Hrugka Volume 14 Origin of Cultivated Rice, by H. Oka Volume 15 Nutritional Disorders of Cultivated Plants, edited by W. Bergmann Volume 16 Hop Production, edited by V. Ryb~t~ek Volume 17 Principles and Methods of Plant Breeding, by S. Borojevi~ Volume 18 Experimental Morphogenesis and Integration of Plants, by J. Seb/mek, Z. Sladl~ and S. Proch~tzka Volume 19 Plant Tissue Culture: Applications and Limitations, by S.S. Bhojwani Volume 20 Weather and Yield, edited by J. Petr Volume 21 Plant Physiology, edited by J. Sebfinek Volume 22 Reproductive Adaption of Rice to Environmental Stress, by Y. Takeoka, A.A. Mamum, T. Wada and P.B. Kaufman Volume 23 Natural Rubber: Biology, Cultivation and Technology, edited by M.R. Sethuraj and N.M. Mathew Volume 24 Irrigated Forage Production, by A. Dovrat Volume 25 Perspectives for Agronomy, edited by M.K. van Ittersum and S.C. van de Geijn v

Developments in Crop Science 25

PERSPECTIVES FOR AGRONOMY Adopting Ecological Principles and Managing Resource Use Proceedings of the 4th Congress of the European Society for Agronomy, Veldhoven and Wageningen, The Netherlands, 7-1 1 July 1996 Edited by

M.K. VAN ITTERSUM Wageningen Agricultural University, Department of Theoretical Production Ecology, P.O.Box 430, 6700 AK Wageningen, The Netherlands

S.C. VAN DE GEIJN Research Institute for Agrobiology and Soil Fertility (AB-DLO), P.O. Box 14, 6700 AA Wageningen, The Netherlands

ELSEVIER

Amsterdam - Lausanne - New York - Oxford - Shannon - Singapore - Tokyo 1997

ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 P.O. Box 21 l, 1000 AE Amsterdam, The Netherlands

Partly reprintedfrom the European Journal of Agronomy, Vol. 7/1-3

ISBN 0 444 82852 4 © 1997, ELSEVIER SCIENCE B.V. 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 or otherwise, without the prior written permission of the publisher, Elsevier Science B.V., Copyright & Permissions Department, P.O. Box 521, 1000 AM Amsterdam, The Netherlands. Special regulations for readers in the U.S.A.-This publication has been registered with the Copyright Clearance Center Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the U.S.A. All other copyright questions, including photocopying outside of the U.S.A., should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified. No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. This book is printed on acid-free paper Transferred to Digital Printing 2006

Preface Since the Second World War, agricultural development has been characterised by a strong increase in land and labour productivity in large parts of Europe (Porceddu and Rabbinge, 1997), strongly stimulated by policy (e.g., Common Agricultural Policy of the EU). The level of self-sufficiency has been surpassed for nearly all agricultural products. The strong increase in productivity leading to the present situation characterised by over-production, has been attended by changes in farm structure, a decrease in agricultural employment, social, budgetary and economic problems at the regional and European Union level and by environmental deterioration. This situation contrasts with a situation of relatively low productivity, socio-economic and environmental problems in mainly Southern and Eastern parts of Europe, and with a major problem and at the same time a challenge for the next century at a global level: food production and security for a doubled population with a more affluent diet. Agricultural development has evolved from an activity with mainly one-dimensional, productivity aims, into a multi-dimensional issue with environmental, agricultural, economic and social objectives. Agriculture should adopt ecological principles and it should optimize the use of resources, i.e., agriculture should 'ecologize', and at the same time it should meet social and economic objectives. Agricultural development is an important issue within the framework of what is called sustainable development. The notion of sustainable development calls for explicit consideration of each of the mentioned objectives and for consideration of the problem at several aggregation levels. The situation in (a part of) Europe cannot be isolated from the situation at country or global scale; problems and solutions at field level should also consider, for instance, plant and crop rotation level.

Agriculture with such broadened objectives requires new systems at a range of aggregation levels. It requires different analyses, synthesis of knowledge and a different type of intervention at the policy level. Agriculture with broadened objectives requires a different agronomy. It calls for detailed knowledge concerning the functioning and production of agricultural plants and crops and their ecological relationships. In addition, it calls for synthesis and design of new ideotypes and genotypes, new production technologies, cropping systems, farming systems and agroecological land use systems. Basic knowledge at field, plant and lower levels of integration should be used and synthesized in the design of new systems at higher levels that meet a set of explicit objectives. These new systems should then be evaluated for their effectiveness at the various levels. This type of agronomic research will often be of an interdisciplinary nature with agronomists working together with breeders, physiologists, soil scientists, economists and sociologists. To fulfil this new role, agronomy has a range of sophisticated tools at its disposal. To fully exploit the potential of these tools, they should not be used separately, but in combination. A new agronomy should tailor the tools to the type of questions and benefit from the synergism of: empirical and experimental research, be it in laboratories, climate chambers, greenhouses or in the field to diagnose and analyse problems and to test new designs; mathematical modelling techniques to summarize knowledge, to test hypotheses and to identify knowledge gaps; - prototyping on experimental and commercial farms to design and implement new crop rotations and farming systems; and - model-based explorations to improve systems -

-

vi

understanding and identify a wide range of options. During the Fourth Congress of the European Society for Agronomy, held in Veldhoven-Wageningen, The Netherlands, 7-11 July 1996, the new perspective for agronomy emerged. Various keynote addresses, session themes, and oral and poster contributions demonstrated the need for a new role of agronomy and its tools (Van Ittersum et al., 1996). The special issue of the European Journal of Agronomy and the Proceedings Book of the Fourth ESACongress (Van Ittersum and van de Geijn, 1997) present a set of case studies illustrating the various agronomic tools that can be used for specific questions. The case studies are grouped in sections illustrating relevant subquestions in developing an agriculture with broadened objectives. The papers were selected such that the various subquestions were represented in the Proceedings. This implies a non-random sample from the contributions during the Congress, since the number of contributions addressing the level of cropping system, farm and agricultural land use was limited. Nevertheless, we think that agronomy should consider these levels of scale in its analysis and design because questions of stakeholders often concern these levels. After an introductory paper on the role of agronomy in research and education in Europe, the second section presents case studies addressing issues concerning agricultural land use, food security and environment. The next set of papers addresses crop physiological aspects in relation to growth factors such as radiation, CO 2, temperature and water. Experimental research and simulation modelling are used in mutual interaction. One important outlet for the generated and integrated knowledge is the ideotyping of crops. Improving resource-use efficiency in agriculture positively affects economic, environmental and agricultural objectives. Many papers presented during the Congress have directly or indirectly addressed this issue. A set of papers particularly focusing on nutrients and organic matter is presented in this volume. Again, a combination of experimental and modelling research is used to enhance understanding of the system and identify options for improvement. The final section addresses the design of integrated and ecological arable farming systems. Prototyping is

put forward as a promising tool to design and implement new farming systems. Indicators are presented that support evaluation of farming systems. Finally the contribution of model-based explorations in developing new farming systems is considered. A discussion follows on the notion that development of sustainable farming systems is not a matter of one-way research delivery, but rather a process in which researchers and target groups should cooperate, learn and develop in true interaction. This applies especially as the operationalisation of sustainability requires value-driven choices calling for a continuous interaction between society, its organisations and farmers on the one hand, and the scientists and designers on the other. We hope that the activities of the European Society for Agronomy and the Proceedings of its Fourth Congress will stimulate to serve the new perspectives of agronomy, i.e., to adopt ecological principles, to optimally manage the use of resources and to meet social and economic objectives.

Martin K. van Ittersum Wageningen Agricultural University Department of Theoretical Production Ecology P.O. Box 430 6700 AK Wageningen The Netherlands

Siebe C. van de Geijn Research Institute for Agrobiology and Soil Fertility (AB-DLO) P.O. Box 14 6700 AA Wageningen The Netherlands

References Porceddu, E. and Rabbinge, R., 1997. Role of research and education in the developmentof agriculture in Europe. Eur. J. Agron., 7: 1-13. Van Ittersum, M.K., Venner, G.E.G.T., van de Geijn, S.C. and Jetten, T.H. (Editors), 1996. Book of Abstracts, Volume I and II. Fourth Congress of European Society of Agronomy, 7-11 July, 1996, Veldhoven, The Netherlands, 736 pp. Van Ittersum, M.K. and van de Geijn, S.C. (Editors), 1997. Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use. Elsevier Science, Amsterdam, in press.

vii

Acknowledgements The editors gratefully acknowledge logistic and technical assistance of Loes Helbers, Irene Gosselink and Guido Venner in processing the flow of manuscripts. We also sincerely thank the reviewers of the manuscripts for contributing to the maintenance of the scientific standard of the papers: D. Auclair L. Bastiaans F. Bonciarelli C.J.H. Booij M.G.R. Cannell J.G. Conijn J.B. Dent P. Dijkstra M. Donatelli P.A.I. Ehlert B. Gerowitt J. Goudriaan D.J. Greenwood P. Gregory J.J.R. Groot J. Hassink A.J. Haverkort G. Hoogenboom C. Jambert B.H. Janssen S.C. Jarvis J. Kubat P.J. Kuikman E.A. Lantinga D.W. Lawlor J.F. Ledent D.K.L. MacKerron L. 't Mannetje H. Meinke K. Mengel J.M. Meynard S. Mikkelsen M.I. Minguez

G.M.J. Mohren J.D. Mumford H. Naber J.J. Neeteson B. Nicolardot S.E. Ogilvy J.E. Olesen F.W.T. Penning de Vries J.R. Porter G. Russell M. Schenk H. Schnyder J.J. Schr6der C.A. Shand A.L. Smit J.-F. Soussana J.H.J. Spiertz E.A. Stockdale P.C. Struik H.F.M. Ten Berge N. Van Breemen P. Van Halteren H. Van Keulen E.N. Van Loo M.A. Van Oijen W.H. Van Riemsdijk J.A. Van Veen J. Vos D.C. Whitehead A.P. Whitmore F.G. Wijnands D. Younie J.C. Zadoks

This Page Intentionally Left Blank

ix

Table of contents Preface ............................................................................................................................................................. v Acknowledgements ................................................................................................... ................................ vii Section I INTRODUCTION

Role of research and education in the development of agriculture in Europe E. Porceddu and R. Rabbinge .......................................................................................................................

3

Section 2 AGRICULTURAL LAND USE, FOOD SECURITY AND ENVIRONMENT

Land use transformation in Africa: three determinants for balancing food security with natural resource utilization P.A. Sanchez and R.R.B. Leakey .................................................................................................................... 19

Agro-ecological characterisation, food production and security P. Bullock. ..................................................................................................................................................... 29

The potential benefits of agroforestry in the Sahel and other semi-arid regions H. Breman and J.J. Kessler. .......................................................................................................................... 39

Chemical crop protection research and development in Europe R. Neumann .................................................................................................................................................... 49

Emissions of CO2, CH4 and N20 from pasture on drained peat soils in the Netherlands C.A. Langeveld, R. Segers, B.O.M. Dirks, A. van den Pol-van Dasselaar, G.L. Velthof and A. Hensen .......... 57

Section 3 CROP PHYSIOLOGY AND IDEOTYPING

Effects of CO2 and temperature on growth and yield of crops of winter wheat over four seasons G.R. Batts, J.I.L. Morison, R.H. Ellis, P. Hadley and TR. Wheeler. ........................................................... 67

Use of in-field measurements of green leaf area and incident radiation to estimate the effects of yellow rust epidemics on the yield of winter wheat R.J. Bryson, N.D. Paveley, W.S. Clark, R. Sylvester-Bradley and R.K. Scott. ................................................... 77

Simulating light regime and intercrop yields in coconut based farming systems 3'. Dauzat and M.N. Eroy ................................................................................................................................. 87

Improving wheat simulation capabilities in Australia from a cropping systems perspective: water and nitrogen effects on spring wheat in a semi-arid environment H. Meinke, G.L. Hammer, H. van Keulen, R. Rabbinge and B.A. Keating. .................................................... 99

Comparison of CropSyst performance for water management in southwestern France using submodels of different levels of complexity C. O. Stockle, M. Cabelguenne and P. Debaecke ......................................................................................... 113

Root growth of three onion cultivars A.D. Bosch Serra, M. Bonet Torrens, F. Domingo Oliv~ and M.A. Melines Pagbs ...................................... 123

Interspecific variability of plant water status and leaf morphogenesis in temperate forage grasses under summer water deficit J.-L. Durand, F. Gastal, S. Etchebest, A.-C. Bonnet and M. Ghesqui~re .................................................... 135 Evaluation of sunflower (Helianthus annuus, L.) genotypes differing in early vigour using a simulation

model F. Agiiera, F.J. Villalobos and F. Orgaz ...................................................................................................... 145

Options of breeding for greater maize yields in the tropics A. Elings, J.W. White and G.O. Edmeades .................................................................................................. 155

Section 4 MANAGING RESOURCE USE

Nitrogen budgets of three experimental and two commercial dairy farms in the Netherlands J.J. Neeteson and J. Hassink ........................................................................................................................ 171

Resource use at the cropping system level P. C Struik and F. Bonciarelli ...................................................................................................................... 179

The efficient use of solar radiation, water and nitrogen in arable farming: matching supply and demand of genotypes A.J. Haverkort, H. van Keulen and M.I. Minguez .......................................................................................... 191

Soil-plant nitrogen dynamics: what concepts are required? E.A. Stockdale, J.L. Gaunt and J. Vos ......................................................................................................... 201

Modeling crop nitrogen requirements" a critical analysis C 0. Stockle and P. Debaeke ....................................................................................................................... 217

Maize production in a grass mulch system - seasonal patterns of indicators of the nitrogen status of maize B. Fell, S.V. Garibay, H.U. Ammon and P. Stamp ........................................................................................ 227 Nitrogen transformations after the spreading of pig slurry on bare soil and ryegrass ~5N-labelled ammonium T. Morvan, Ph. Leterme, G.G. Arsene and B. Mary ...................................................................................... 237 Size and density fractionation of soil organic matter and the physical capacity of soils to protect organic matter J. Hassink, A.P. Whitmore and J. Kub6t ....................................................................................................... 245 Characterization of dissolved organic carbon in cleared forest soils converted to maize cultivation L. Delprat, P. Chassin, M. Lin~res and C. Jambert. ...................................................................................... 257 Analysis of impact of farming practices on dynamics of soil organic matter in northern China H.S. Yang and B.H. Janssen .......................................................................................................................... 267 Agronomic measures for better utilization of soil and fertilizer phosphates K. Mengel. ................................................................................................................................................... 277 Section 5 DESIGNING FARMING SYSTEMS

A methodical way of prototying integrated and ecological arable farming systems (I/EAFS) in interaction with pilot farms P. Vereijken ................................................................................................................................................. 293

The Logarden project: development of an ecological and an integrated arable farming system CA. Helander. ............................................................................................................................................. 309

xi

Integrated crop protection and environment exposure to pesticides: methodes to reduce use and impact of pesticides in arable farming F. G. Wijnands ..............................................................................................................................................

319

Use of agro-ecological indicators for the evaluation of farming systems C. Bockstaller, P. Girardin and H.M.G. van der Werf ................................................................................ 329

Model-based explorations to support development of substainable farming systems: case studies from France and the Netherlands W.A.H. Rossing, J.M. Meynard and M.K. van lttersum ............................................................................... 339

Learning for substainable agriculture B.M. Somers .................................................................................................................................................

353

Author Index ............................................................................................................................................... 361 Subject index ............................................................................................................................................... 363

This Page Intentionally Left Blank

Section 1 INTRODUCTION Role of research and education in the development of agriculture in Europe E. Porceddu and R. Rabbinge ........................................................................................................................ 3 Reprinted from the European Journal of Agronomy 7 (1997) 1-13

This Page Intentionally Left Blank

© 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

Role of research and education in the development of agriculture in Europe E. Porceddu a'*, R. Rabbinge b aDepartment of Agrobiology and Agrochemistry, University of Tuscia, Via S. Camillo De Lellis, O1100 Viterbo, Italy bDepartment of Theoretical Production Ecology, WageningenAgricultural University, P.O. Box 430, 6700 AK Wageningen, The Netherlands

Accepted 16 June 1997

Abstract Agricultural research and education in Europe has played a major role in the advancement of agriculture and land use during the last century. The scientific basis of agriculture has been strengthened and the use of insight, knowledge and expertise in farmers' fields is widely adopted. As a result of this development, productivity per hectare, efficiency and efficacy of use of external inputs has increased considerably. During the last decades, objectives of agriculture and land use have broadened and this illustrates the need for further ecotechnological knowledge and insight to reach, in a balanced way, multiple goals of agriculture, productivity, protection of the environment, nature conservation and development. Research and education have to be developed in that direction to make agricultural science and technology more responsive to changing societal demandsl The disciplinary scientific quality and depth should develop in tandem with integrating problem-oriented multidisciplinary research activities. Systems approaches may serve as an instrument to that goal. © 1997 Elsevier Science B.V. Keywords: Systems approaches; Reorientation agricultural research; Broadened objectives

I. Introduction Since mankind started to exploit plants and other organisms to fulfill the changing demands for food and other needs, agriculture has been an important activity of mankind. The continuous evolution and manipulation of organisms and their function has been based on empiricism, knowledge, insight and expertise. The development has accelerated during the last century, when the scientific basis for agriculture strengthened and its simultaneous implementation became possible. The last decades have seen a broadening of aims and objectives of agriculture and

an increased importance of various environmental and socio-economic constraints. The ways that development took place, the change from old concepts to new perspectives and the challenges of agricultural research in general and more specifically of agronomy, are described in this introductory article to the Proceedings of the 4th ESA conference. The possibilities to reach those aims, and the new institutions and concepts which recently were introduced and adopted, are described.

2. Historical setting

* Corresponding author. Reprinted from the European Journal of Agronomy 7 (1997) 1-13

Scientific developments in agriculture during the

19th century were dominated by two scientists: Justus von Liebig and Gregorius Mendel. Their discoveries in the field of plant nutrition and the laws of heredity opened entirely new roads. Even today they still inspire scientific progress in fields very distant from the ones in which they originally made their contributions. A careful examination of their scientific experiences leads us to understand that, contrary to what is often reported, they worked in a lively and stimulating environment. It laid the basis for agricultural sciences and the institutionalization of research in agriculture; an event that Whitehead (1925) considered as the greatest innovation of the 19th century. The experiments carded out by private farmers, including the introduction of new crops, the establishment of chairs of agriculture at several university level teaching institutions, and the initiation in 1786 of the first public experimental farm near Braunschweig and a farm school in Hamburg, gave birth to agricultural research and experimental institutes in most European countries. The forestry school and the pomology institute of Weihenstephan were set up in 1803, followed by the research and teaching institutes at Hohenheim in 1818 and at Moglin in 1819. By the end of the century, Germany could boast as many as 87 research institutes, many of them in Prussia (Nomisma, 1996). The movement soon spread to France, where, at the initiative of private societies, farm schools were set up in Roville, near Nancy in 1822, Le Saussaie in 1828, and Bechelbon in Alsace in 1834. The first public institution became eventually operational in Versailles in 1848. In total there were 61 experimental stations in France at the end of the century (Nomisma, 1996). In the United Kingdom, the first experimental institution was set up in Rothamsted in 1844 as a private foundation, a form it was to retain until the end of the century; it was later joined by similar organizations in Woburn and Pumpherstone. By the end of the century agricultural colleges had been established at Bangor in 1889, Leeds in 1890, Aberystwyth in 1891, Nottingham in 1893, Reading in 1893, Cambridge in 1894 and Wye in 1894 (Speedy, 1994). In Italy, a private school opened in Tuscany in 1834 with a program of theoretical and practical lectures. Six years later it was transformed into a university level institution attached to the University of Pisa

(Coppini and Volpi, 1991). Within a few years of the unification of the country, in 1866, the University of Naples started planning a Faculty of Agriculture. It eventually became operational in 1872, though by that time it had been preceded by a Higher School of Agriculture, set up in 1870 by the University of Milan. Three agricultural experimental stations were set up in the same year. The last institution set up during last century in Italy was the Higher School of Agriculture at the University of Perugia in 1896. In Spain teaching started in 1855 at the Agricultural Central School in Aranjuez, close to Madrid, which had been promoted by the Ministry of Agriculture (J.A. Cubero, in litteris). In many countries of Eastern Europe similar developments took place, although in most cases much later. During 18501900 many National Agricultural Research Systems (NARS), including academic education and extension, were developed. However, during the last decade of the 19th century, many of these systems suffered due to considerable contraction of the agricultural sector. The story of the agricultural research institutes was very different in Denmark and the Netherlands, where the stimulus for extensive teaching and experimentation activities was provided by the great agricultural crisis of the last quarter of the 19th century. While other countries adopted protectionistic measures, these two countries reacted by rising the level of cultivation and agronomic techniques of their farmers. As a result, their competitive capacity increased. Agricultural research in Denmark was entrusted to four experimental stations, while teaching was concentrated at the Higher School of Agriculture in Copenhagen. In the Netherlands, the Faculty of Agriculture, then still a Higher School, was founded at Wageningen in 1876. Very quickly it gave rise to four distinct bodies: the Higher School of Agriculture, the School of Horticulture, the Agricultural Secondary School, and the Agricultural Colonial School. Agricultural extension was started and an intensive system of experimental stations and farmer field experiments initiated (Eveleens and Rabbinge, 1994). This is not the place for attempting a full-scale historical review of Europe's agricultural research, experimentation, and teaching institutions, but the few words devoted to their origin demonstrate not only the fervor that existed in Europe for agricultural

research and teaching, but also that this fervor made it possible to develop and exploit certain processes. In no more than a century these lead to greater transformations in agriculture than the ones that had occurred in the two preceding millennia. Both in arable farming and dairy farming, increases in land and labor productivity in the last 100 years were dramatic, compared with all ages before. In Italy, wheat production, which at the beginning of the century was of the order of 1 t/ ha, is today four times as great; the cultivation of 1 ha of wheat today absorbs 4 man days or 30 h of labor, while as many as 70 man days were needed only 70 years ago. Similar developments occurred in other industrialized countries. Average yield level or wheat in the Netherlands increased from 1.5 t/ha to 8.5 t/ha and labor requirement decreased from 300 to 15 man hours/ha (de Wit et al., 1987). At present yield levels of 10 t/ha are not exceptional. Similar developments can be observed in other crops. The increased productivity coupled with an expansion of the agricultural land allowed most of the European countries to produce enough food commodities and meet the increasing demand through population growth and diet change. Severe famine crisis could thus be prevented. Also, emigration, which drained

more than 40 million Europeans during the 19th century, could be prevented in this century.

3. Recent developments in agricultural productivity and employment Events during the last quarter of the present century, a period even the youngest scientists have experienced at first hand, are well documented. At the end of the 1960s the number of people served by a single European Union farmer was only half the present number (Table 1), and the amount of produce provided by such a farmer was half of what it is today (Table 2). This has left an ever larger number of people free to be engaged in other activities, to leave the rural areas and settle in towns (Table 1). In the same 25 years, agricultural land has decreased and within this acreage the arable crop land has increased, the yield per unit of land has increased by approximately 50%, and yield per unit of labor has tripled (Table 2). These facts made it possible for agricultural commodity prices to lag behind the increase in the general cost of living, enabling an ever larger number of people to enjoy

Table 1 Total and active population in EU countries. Situation in 1973 and trends 1970-1993. (Source: Eurostat, 1994; FAO, 1995) Country

Denmark Finland Sweden Austria Belgium France Germany Ireland Netherlands United Kingdom Greece Italy Portugal Spain

1993

1993 vs. 1970

Total popul, (1000)

Rural Rural/ T o t a l Agric. Agric./ popul, total labor labor total (1000) popul. (1000) (1000) popul. labor labor (%) (%)

Inhab/ arg. labor (#)

Total popul, (%)

Rural Total popul, labor (%) (%)

Agri./ tot. % labor (%)

Inhab./ arg. labor (%)

5155 5017 8633 7846 10251 56848 63694 3555 15 158 57908

644 2011 1356 3199 243 14400 7704 1512 1716 6341

12 40 16 41 2 25 12 43 II II

2567 2030 3912 3570 3744 21908 36 Ill 1149 6640 25 348

131 174 139 245 99 1195 1272 157 256 522

5l 8.6 3.6 6.9 2.6 5.5 3.5 13.7 3.9 2.1

39 29 62 32 104 48 50 23 59 Ill

105 109 107 105 103 100 105 121 ll6 104

64 88 89 89 37 98 68 106 95 97

111 93 100 Ill 98 105 134 103 145 100

48 34 41 41 55 41 62 52 78 73

202 320 260 270 190 270 170 230 150 140

10055 57 812 10300 39390

3671 17604 6732 8134

37 30 65 21

3715 20267 4464 11 868

791 1488 516 1212

21.3 7.3 11.6 10.2

13 39 20 33

ll4 107 ll4 117

88 92 100 71

98 105 139 96

45 40 53 32

260 260 220 360

Table 2 Total land and utilized agricultural area in EU countries. Situation in 1973 and trends 1970-1993. (Source: Eurostat, 1994; FAO, 1995) Country

1993 Total area (1000 ha)

Denmark Finland Sweden Austria Belgium France Germany Ireland Netherlands United Kingdom Greece Italy Portugal Spain

1993 vs. 1970 (2)/(1) (%)

Crop area/ (2) (%)

1

Utilized agric, area (1000 ha) 2

(2) (%)

(4) (%)

(3) 1970 (%)

(5) (%)

Output/ area (%)

Output/ labor (%)

4

(2)/ labor (ha) 5

3

4306 33 699 44 759 8385 3050 54 883 24871 7031 3694 24419

2751 2610 3359 3482 1412 30 217 17 162 4450 1997 17 178

64 8 8 42 46 55 69 63 54 70

92 96 83 41 58 59 68 17 46 35

19,2 11,6 19,2 15,6 17,8 21,3 10,1 29,1 8,3 30,3

92 87 97 89 88 91 126 92 91 89

94 94 91 85 96 94 145 65 105 84

90 91 89 43 53 58 60 24 40 37

183 190 173 208 207 207 177 162 128 121

153 137 135 144 154 136 143 158 187 152

282 263 232 297 320 280 255 255 239 165

13 196 30 098 9202 50 508

5785 16 800 3829 26 398

43 56 42 52

51 54 58 58

9,3 9,3 5,6 18,2

65 87 78 73

81 64 51 75

41 73 89 56

155 202 140 194

151 145 102 180

233 292 143 348

higher food standards and a better lifestyle. Today there are problems of overproduction. The increasing yield per unit of land, as result of research innovations and use of external inputs, which have been progressively introduced and adopted by farmers, is continuing. Greater income has also led to a greater demand for goods and services of high elasticity with respect to income, among them a cleaner environment, stricter requirements for food quality and the way food is produced. Environmental quality is nowadays considered as a 'superior good'. Its demand rises in proportion to income growth, while food production is an 'inferior good', the demand for which falls in proportion to increases in income (Engel' s Law). Demand on food quality increasingly dominates food quantity while concern for environmental health and safety grows. Natural resources are no longer perceived by policy communities as merely the medium to produce more and cheaper food but rather in terms of local and global ecosystem functioning. Ecosystem maintenance, bio-diversity, recharge of ground water, clean air and bequeath value are important topics, as agronomists well know. Agriculture is also a service to society, for example, as a means to preserve the landscape. However, there

is no doubt that such a service to society requires greater attention in planning and financing structures, and can only play an additive role for the farmer community. It cannot replace farming as an economic activity. Interventions using farmers income compensation, as it was recently experimented in the Netherlands, where a number of farmers in certain areas are subsidized to maintain landraces and primitive varieties, utilizing traditional technologies, may be seen as examples. National attitudes are nowadays dominated by the effects of market globalisation, which modifies trade flows, reallocates production and consumption at the world level, thus influencing agricultural activities and natural environment and, quite generally, setting the scene for any development in rural areas.

4. The influence of agricultural policy and trade agreements At present, the situation in EU agriculture can be depicted as follows: agricultural employment decreased by 2% per year during the last 25 years, but in a number

of member states agricultural employment still exceeds 5% (Table 1); land and labor productivity are still limited in some member countries (Fig. 1), in spite of progress during the last 25 years (Table 2); land per unit of labor is still limited in a number of countries (Table 1), and a high percentage of agricultural laborers remains (Table

•

•

25

• Gr

20 ,P • IRL

-110

,E

,I **

2); •

pro capite GPD (Gross Domestic Production) in member states is inversely proportional to the percentage of labor in agriculture (Fig. 2).

Taking into account these facts, the restructured EU-CAP (Common Agricultural Policy) envisages an agriculture with: •

more technology, but also with greater diversity; a smaller cultivated acreage, but also with more land dedicated to special uses; a more stable production, but also with a greater variety of produce; a smaller number of farmers, who, however, perform several and more diversified activities.

• • •

*Dk

*UK,B

The reform of the EU-CAP was followed by a further extension of trade agreements in the form of General Agreement on Tariffs and Trade (GATT). The most recent economic analyses (Anania et al., 1996) indicate that these agreements would not distort the EU-CAP, although the impact may not be negligible. Analyses also underscore the need that the EU

--

0

I ....

5000

I

I

10000

15000

20000

25000

30000

pro cap/h) GDP

Fig. 2. Relationbetween pro capite GrossDomesticProductionand labor force in agriculture in EU countries during 1993. agricultural system maintains its high competitiveness in terms of low cost in combination with high product quality, this in the domestic market as well as in the international one. The need represents the nodal point of the European agriculture in the near future: lower costs, including environmental costs, better quality and toxicologically safe products. These involve the efficiency of production processes and farm technologies. That requires a policy presently adopted by the European Union. Yet, it is not sufficient. The relative weight of various objectives has to be reconsidered. The choice of such a weight factor may result in considerable differences in land use in the future, as shown in an explorative study on reformulation of the Common Agricultural Policy of the European Union (Rabbinge and van Latesteijn, 1992; WRR, 1992). Using modem simulation and systems approaches, and incorporating appropriate technical information, a series of land use scenarios were generated, based on different priorities for objectives that dominate agricultural development: o

jA

B

k,

~m

•

O I&l i

]mooo

F

•

i

o 1000

2000

~

•

4000

Output per hectare (ECU's per ha)

Fig. 1. Trends in agricultural productivityin EU countries during the period 1970-1993. Lower left points 1966-1970, upper fight points 1986-1990. (Source: Eurostat, 1994; FAO, 1995).

•

free market and free trade, based on economic efficiency, i.e. maximum yield at minimum cost; regional development, with the aim of maintaining employment in agriculture at the highest possible level but within the constraints of a productive agriculture; nature and landscape conservation, by the creation of natural reserve areas separated from the agricultural ones and thus minimiz-

ing land use for agriculture and maximizing land productivity; environmental protection, minimizing the negative effects of agriculture; in other words, minimum use of pesticides and emission of nitrate, and other environmental side effects of agriculture, per unit of product or per unit of land. The results provided a framework for assessing the strategic options on which policy makers may base their decision. Indeed, the exploration of options showed very marked differences: •

•

•

as far as land use is concerned, the maximum acreage used by agriculture in the case of regional development is three times as great as the minimum associated with the free market hypothesis, but 40 out of the present 127 million hectares will nevertheless have to be taken out of use in all scenarios (Fig. 3a); as far as employment is concerned, all scenarios give rise to a further downturn of agricultural employment, but the size of the exodus may vary between 3 and 4.5 million man power units depending on the chosen scenario (Fig. 3b); quite independent of the scenario actually adopted, policy measures can successfully promote more environmentally friendly production methods by limiting the use of fertilizer and reducing the large scale use of crop protection agents, avoiding adverse effects on the environment (Fig. 3c).

The outcomes of the scenarios demonstrated the enormous challenge and chances for reorientation of the CAP, and for agricultural research. It becomes even more important to do such explorative studies when the European Union will be further extended in the near future. The considerable increase of land area, the tremendous possibilities for agriculture and the needed readjustment of the common agricultural policy in those types of situations require extensive explorations. Therefore, explorative studies on land use and agrotechnologies should become major issues on the future research agenda.

5. Agronomy and multiple goals in agriculture and land use The changes of agriculture from a purely production-oriented activity into a science based production sector, trying to meet productivity, efficiency and efficacy aims, has been of considerable importance during the last decades. Agriculture has broadened and diversified its objectives, rendering explicit and important a number of economic, environmental, social and nature-protection roles that in the past were simply passed over in silence. The concept of a good agricultural practice - where farmers produce in an economic and socio-political environment conducive to long-term conservation and use of natural resources, and do not want their source of survival and generation of income to disappear - has always been the foundation of agronomic teaching. Indeed, the same Justus von Liebig, often blamed as the father of chemistry in agriculture, way back in 1855 wrote as follows: 'The task of the farmer is not to achieve high crop yield to the detriment of the field, which only causes it to impoverish earlier. Rather, it should be in his own interest, as well as society's, to achieve high yields that are constantly increasing on a permanent basis.' (von Liebig, 1855). The concept is echoed in the 'sustainability' idea, entered into a common use thanks to Our Common Future, the report of the World Commission on the Environment and Development (1987), and defined by the Consultative Group for International Agricultural Research (TAC, 1989) as the successful management of resources for agriculture to satisfy changing human needs while maintaining or enhancing the quality of the environment and conserving natural resources. The question of sustainability, consequently, arises when the resources used for production are placed under stress, as is now widely happening in industrialized countries due to excessive irrigation, fertilizer and pesticide application, and in developing countries where increasing population pressure continues to strain product on resources. An agricultural production system is not sustainable if it leads to declining productivity, degrades the resource base, or is not economically viable. The given description of unsustainability does not necessarily lead to a precise definition of sustainabilty. This is not easy as

.AND USE 150-

"~" 100 r,-

o..,

E t~

2 <:

50

0

A

B

C

D TODAY

SCENARIOS BdR.OYIdB~ ,,,

A

_,: E

.~

f

,.,

6-

W C

|

4"

g C

J

2-

, _ t~2~.~'.-:

I i F~//~ A

B

,~,~~-',',;.~,,:~|~!:..:~::: C

D TODAY

SCENARIOS CROP PROTECTION

E

v

300-

O C ¢g

]~ 200oo .~ 1000A

B

C

D TODAY

SCENARIOS

Fig. 3. Land use, employment, and crop protection in different scenarios (WRR, 1992): (A) Free market and free trade. (B) Regional development. (C) Nature and landscape. (D) Environmental protection.

sustainability comprises both normative and scientific notions that may be weighted differently, but in all cases the 'unsustainability spiral' (Rabbinge, 1997) should be broken. Agriculture, indeed, makes use of nature's productive capacity, tapping outputs from the system in the form of harvest, and the very continuity of this process presupposes that certain inputs must be added to the system to compensate the tapped outputs. It is in any case well known that inputs, no matter what they might be, cannot be 100% converted into outputs or, put differently, that the efficiency of the system cannot be 100%. Part of the inputs is always lost into the environment as leakage. These losses can be reduced but not entirely avoided, irrespective to the farmer aiming at maximizing the output or seeking input efficiency. In this context now, the role of the agronomist (literally: he who governs the fields) is to optimize the system in such a way as to obtain a maximum efficiency. The modern agronomist is no longer a pure empiricist. More and more science based interventions and measures are developed and implemented. White peg agronomy is replaced by production ecology (Rabbinge, 1997). Poor agricultural efficiency, and the environmental pollution that may derive from it, is often associated with the law of diminishing returns. It states that the relationship between the amount of a production factor and the yield level is not linear, but levels off and more and more external inputs are needed in order to push up the yields to their potential level. High input doses are blamed to be responsible for inefficiency and pollution. While the law is certainly valid, it should be noticed that it has been formulated for situations where all the other production factors remain constant. This is obviously not the case when changes over time occur. Using data of maize yield and nitrogen fertilizer use in the United States for the period 1945-1982 (Fig. 4), de Wit (1992) showed that at the high end of the yield range the fertilizer efficiency is at least equal to, if not greater than the efficiency at the low end. The relationship flattens for some years in the 1960s, as indicated by the law of diminishing returns, but when agricultural practices improved again in the 1970s - possibly under influence of the oil shock, but there may also have been an impact of the change of the varieties' cytoplasm, as a consequence of susceptibility of the T cytoplasm to Hel-

10

Grain yield (kg ha -1) 8000 -

1982 OO

•

6000 •• •

4000

2000

•

1970

O o 00 ° -O

1945 I

0

40

,,

I

I

I

80

120

160

Nitrogen (kg ha -1) Fig. 4. Plot of annual average maize yields versus nitrogen fertilizer rate in USA during the period 1945-1982. (From: Sinclair, 1990).

the fertilizer response returned to its previous level. These relationships become even more evident when expressed in energy terms (de Wit, 1979). Elaborating on the law of optimum (Liebscher, 1895) - a production factor, which is in minimum supply, contributes more to production, when the other production factors are closer to their optimum - de Wit (1992) contended that most production resources are used more efficiently with increasing yield level due to optimization of growing conditions. According to this vision, strategic agronomic research, that is to serve both agriculture and its environment, should be directed towards the search for the minimum of each production resource that is needed to allow maximum utilization of all other production resources in the farming system considered. The problems - inefficiency and pollution - therefore derive from the fact that the processes governing agricultural production are not yet properly understood. This may lead to the adoption of measures and practices that eventually do not contribute towards efficient use of resources and control of environmental effects. The broadened objectives of agri-

minthosporium

-

culture have also broadened the mission of agronomy as a scientific discipline. Agronomy is the science that, taking advantage of ecological principles, devices and tests new approaches, rules and means to govern the relationships between the different production factors in order to obtain an appropriate harvest. The study of single production factors is entrusted to such propaedeutic disciplines as agrometeorology, soil physics, crop physiology, genetics, pathology, entomology, and land use, while the results are offered to other disciplines, such as economy, sociology, production ecology, and human nutrition and health, that provide the strategic options of the agricultural production. Agronomy, therefore, plays an altogether central role in agricultural research. It has a revitalized and crucial role after some decades of obsolescence and crisis due to the difficulties it faced in devising innovative solutions and/or sustainable options.

6. Approaches in agronomic research

The study of agronomic problems has been traditionally faced by means of two different types of research that could be called general and specific or specialist approaches (L. Cavazza, in litteris). The general approach consists essentially in verifying the reaction of the entire agro-ecological system i.e. soil, plant, atmosphere, other organisms - to well defined experimental conditions containing one or a combination of few changed technological factors or doses of factors, where a factor may be an input (ploughing, crop rotation, fertilization, irrigation, weed, pest, disease control etc.), its dose, and/or its application time. The qualifying components of the general approach rest on the identification of the factors capable of modifying the system and the analysis of the responses to the doses of the factor(s) under test. The results are strongly in line with the experimental conditions, rich in unsuspected information about the system behavior in response to the specific experimental practices and of considerable technical value. Nevertheless, the system is considered as a single entity, or as a set of a limited number of single entities, and the approach does not provide any direct, simple information on what occurs inside the system or the

11

system components. The theoretical knowledge of single components does not enter into the logic of the experiment; rather it is utilized to explain and justify the results. Findings are, therefore, essentially descriptive, site specific, closely tied to experimental conditions, without wide opportunities for generalization. Real innovations are rare, although they are promptly entered into agricultural practice. It should be pointed out, however, that most of the progress in crop and animal production has been attained in this way and no new technique should be entered into actual agricultural practice without having been tested in this way. The specific approach has its starting point in the analysis of the system components, such as root system, canopy structure and hormonal system. One or few of those will then become the subject of research to gain as much insight as the state of scientific knowledge will enable. The smaller and more clearly defined the component, the better it can be analyzed, the more readily principles and rules can be formulated, even in deterministic terms. The results are not bound to the experimental conditions. The difficulty in this type of approach derives from having to assess the behavior of the system from some detail. The risk of a loose re-assembling of various components, that have nevertheless been very precisely analyzed, is always there. Another risk is that only part of the system is analyzed, considering the other parts as fixed. One relies on either a single or just a few technological factors to push up the entire system. Failure in optimizing the system, promotion of diminishing returns and pollution of the environment may become real consequences. Both the general and specific approaches were very productive in scientific terms. The first by the implementation and derivation of practical rules, the second by gaining understanding in some of the basic processes involved. The distinction between the two approaches has its counterpart in education and training. A scholar who has shared his energies over extensive scientific bases, though without getting insight in any one of them, will be inclined to the general approach, but will not get the results typical of the other approach, unless he enrolls on specific postgraduate preparation, like a Ph.D. or similar project. On the other hand, a scholar who, with identical commitment, concentrates his education on some specific

aspect, will prefer the specific approach, even though he will find it difficult to insert his findings into the system and to translate them directly into advice and recommendations at the system level. The characterization of the two approaches does not seek to place them in opposition to one another, or to underscore their own relative shortcomings. It rather aims at highlighting the opportunities for integrating them, on the ground that the thorough scientific insight and the creativity of the specific approach, joined with the description and understanding of the system behavior and its problems typical of the general approach, show to be very productive and will ultimately lead to real progress in agronomic science. Research and education approaches are really modem and productive when they combine and utilize in a proper balance the most positive components of the general and specific approach. Actually, in the last decades we have witnessed the integration of process specific knowledge into very precise, widely accepted relationships between processes in the system and driving factors from outside the system. Systems approaches and simulation were very functional in that integration. Simulation models, capable of synthesizing different pieces of knowledge, were developed. These models are simplifications of the real system functioning. They function in the soil, plant, atmosphere, or other organisms systems, but only represent the major elements of the system. The fact that models require well defined hypotheses on systems behavior and an adequate interpretation of the results needs to be stressed. Once formulated, the model helps in a heuristic way to gain insight that may be used to derive general insights and rules. The model, when confronted with experimental evidence, is, therefore, the true reviewer of our knowledge. When a model produces unacceptable results, it means that input data were not reliable or, even worse, hypotheses and assumptions underlying the model were not correct, that knowledge is insufficient, and that we must consequently review the simplifications made about the system to better identify its character and define the relations existing among the different components. In this way, systems analysis and simulation constitute an excellent opportunity in agricultural research, in the advancement of knowledge and the optimization of relations among factors. Optimi-

12

zation itself requires creativity and innovation, in addition to a well structured education, where deepening in basic science is embedded in the broadened perspective of agriculture and agricultural problems. Modem agricultural research means finding solutions to whole-system deficiencies as well as to component problems. It is futile to attack specific technical problems without addressing the overall pattern. Modem agronomic research must aim at all levels, including farmers' fields, cropping systems, farming systems, and regional and supra-regional levels. Systems analysis enables the agronomist to go beyond the farm bounds. It enlarges his field of activity to policy, social and economic grounds, that may sometimes heavily affect his research activity. Crop and/or farm systems are, in fact, embedded in the economic-social-political system, and this requires a widening of those considerations, levels, constraints, and results that guide farmers in implementing agronomic rules and technologies. Two new dimensions, among those in front of us, call for immediate and adequate attention: (i) the shifting of interests from farm to the regional dimension, and (ii) the widening of effects and aims of farm activity to include, in addition to production, also those of labor conditions, commodity quality, agricultural impact at farm and district levels, and more generally all those interactions between farm activities and the surrounding environment. The new dimensions require the commitment of scientists from different fields, well beyond those traditional in agricultural research. It represents an additional real challenge for agronomic research.

7. Challenges for agronomic research

Agriculture has to meet at global level a rising demand for marketable outputs, while satisfying ever tighter constraints with respect to safety of products and impact of production techniques on man, nature, environment and landscape. The attainment of these goals requires a comprehensive and integrated research approach. It is the synthesis of knowledge from various disciplines into a coherent framework, subsequently used to develop, implement and evaluate location specific farming and management options.

Crop harvest is the result of interactions among components such as growth-determining factors, growth-limiting factors, and growth reducing factors (Fig. 5), where: •

•

•

growth-determining factors are those which determine the growth potential of a crop under proper supply of water and nutrients, such as radiation and crop characteristics; growth-limiting factors comprise abiotic resources, such as water and nutrients, the suboptimal supply of which limits crop growth and yield; growth reducing factors are those, like pests, diseases, weeds, that reduce attainable growth to actual growth (van Ittersum and Rabbinge, 1997).

Three research themes at the interaction between growth determining, growth limiting, and growth reducing factors deserve urgent attention and may represent useful ground for international co-operation among European Countries. The first research theme concerns precision farming. Smart farming technologies aiming at both productivity and efficiency gains are vital under either well endowed or marginal conditions. There is a considerable need to manipulate the defining, limiting and reducing factors, in such a way that all individual factors are used with nearly maximal efficiency. Actually, the role of growth limiting factors and agronomic practices - such as fertilizers, irrigation and crop density - has been extensively investigated in many different parts of the world to determine actual yield. Yet, much research is still of a descriptive nature at systems level and does not consider the causal factors behind it. A clear cut distinction between cause and effect is absent while processes with long time coefficients are not used to understand systems behavior. The analysis of systems will result in a better understanding of the role of specific components and their interactions, and will allow to more appropriate farm technologies to be devised. In particular, improvements in efficiency under suboptimal growth conditions would produce a tremendous impact on agriculture in Southern Europe. The optimization of external inputs and the way they affect the growth and production of agro-ecosystems under suboptimal conditions is a major task for production ecologists.

13

I.

AttainableOrowll~//~, i

(

n~n~gen phosi~xxus "

•d i ~ s ~

:~o , •

g:

,

~ 1

!

i

Production level

Fig. 5. Production levels and corresponding growth factors. (From: Rabbinge, 1993; van Ittersum and Rabbinge, 1997).

The second theme is closely linked to these aspects and concerns the water resource use efficiency. In a situation dominated by an ever increasing competition between agriculture and urban areas for the use of water, agricultural research has the duty of optimizing the utilization efficiency of water. Optimization of irrigation aimed at maximizing water use efficiency, performing irrigation in crop critical development and production stages would greatly help in economizing this important resource (Bonciarelli, 1996). Studies at field level and the underlying processes of water flow, water uptake and use may lead to better use of this important external input. However, studies at the watershed level are also needed. Suboptimal use at the field or farm level does not necessarily lead to low efficiency at the watershed level. The contrary may be true in some situations (Seckler, 1996). In combination, field and watershed studies may help to design optimal irrigation systems, techniques and policies at watershed, district, farm and field levels. The third theme concerns the soil. In Mediterranean as well as in Central European areas the study of interactions among soil components and system management is an urgent issue, an absolute imperative. Processes such as microbial activity, mineralization and immobilization events may be studied within a well defined framework of cropping systems research. Old findings on no tillage agriculture may also find their place within this framework. Examples of studies at various integration levels are scarce. The majority of research is at field level or at lower integration levels, limiting the knowledge basis for changes in the agro-ecosystem management. Through wider rotations, growth and yield reduction,

due to soil born diseases, pests and weeds, are limited and better use of mineralization processes and a reduction in nitrate or other plant nutrient losses are possible. Cropping systems research and farming systems research are still in their infancy. Results from descriptive research are available, but well defined explanatory research in connection with agronomic practices in general is lacking. A renaissance of mixed farming systems may even be considered. In some places in Europe, mixed farming systems never disappeared, but they are in general low in efficiency and much less economically feasible than specialized farms. However, new techniques, increased insight and better agronomic methods may help to redesign such farming systems that may lead to a technology jump in efficiency. Some preliminary results of such systems are very encouraging (Lantinga and Rabbinge, 1997). 8. C o n c l u s i o n s a n d p e r s p e c t i v e s

The chances and challenges for agronomy as a discipline are increasing. New perspectives, and new problems are identified. The solution of the problems mentioned is directly connected to the capability of generating scientific knowledge and technological innovations. The systems approach, identified as suitable to provide valuable solutions to the existing problems, requires that innovations proceed in a very organized and structured way. The success of the agricultural knowledge innovation systems depends, among others, on: •

•

•

national cohesion and institutional infrastructure; homogeneity in objectives among parties (farmers' organizations, research institutions, agricultural and environmental authorities, consumers etc.); incorporation of system-oriented, instead of single production-oriented, objectives in research policies; careful monitoring of signals from the outside world; more output-financing research; interest in basic science; co-operation among institutions in different European countries, extended to Eastern and Southern (Mediterranean) countries.

14

The list highlights very clearly that the present role of agricultural research is different from its classical role. The empirical approach, that has been so successful in the past decades, has to be complemented with a systems approach, which bridges the gap between detailed process-oriented research explaining the functioning of crops on basis of the physical, chemical and physiological processes, and the application of that insight and knowledge at systems level. The new approach requires that research teams have a multi-disciplinary nature and they need to work together in a problem-oriented approach; traditional boundaries between disciplines are counterproductive. Disciplinary contributions are fully used only when they are of high quality and tailored to the specific questions, dictated by the problem formulated at system level. An example of such a research organization is the C.T. de Wit Graduate School of Production Ecology of the Wageningen Agricultural University. It has made the systems approach and its development an important part of its mission. A research approach using old concepts and not using the new possibilities and opportunities is inadequate in answering problems raised in a modem society. The described systems approach may help to fulfill the new mission of agronomy as an interdisciplinary field. Productivity rise, efficiency increase, efficacy of input use, maximization of biological control mechanisms, and protection of land and natural resources require a reorientation of agronomic research. The techniques and possibilities are there, why not use them?

Acknowledgements We would like to thank Luigi Cavazza for the useful discussion on the subject, Martin van Ittersum and Siebe van de Geijn, the editors, and Alexandra Nagel for the valuable editorial comments on the manuscript.

References Anania, G., De Filippis, F. and Scoppola, M., 1996. Implicazioni delraccordo GATT per l'agricoltura e le politiche agrarie deU'Unione Europea. In: G. Anania and F. De Filippis (Editors),

L'accordo GATT in Agricoltura e l'Unione Europea. Collana RAISA, Franco Angeli Editore, Milano, pp. 218-339. Bonciarelli, F., 1996. L'evoluzione degli obiettivi ed il ruolo della ricerca agronomica. La Rivista di Agronomia (in press). Coppini, R.P. and Volpi, A., 1991. La Fecolt/i di Agronima dell Universit~t di Pisa. Istruzione agraria e transformazione economicail ruolo delle scuole di agricoltura nella Toscana della prima parte dell'Ottocento. In: A. Benvenuti, R.P. Coppini, R. Favilli and A. Volpi (Editors), La Facolt/l di Agraria delrUniversit~t di Pisa, Pisa, pp. 43-87. de Wit, C.T., 1979. The efficient use of labour, land and energy in agriculture. Agric. Systems 12: 279-287. de Wit, C.T., 1992. Resource use efficiency in agriculture. Agric. Systems, 40: 125-151. de Wit, C.T., Huisman, H. and Rabbinge, R., 1987. Agriculture and its Environment: Are there other ways? Agric. Systems, 23: 211-236. Eurostat, 1994. Serie storiche 1960-93. Eurostat. Eveleens, K.G. and Rabbinge, R., 1994. Current trends in post graduate education at Wageningen Agricultural University: innovation, organisation and internationalisation. Eur. J. Agric. Ed. Extens., 1: 9 l - 102. FAO (Food and Agricultural Organisation of the United Nations), 1995. Country Tables. FAO, Rome. Lantinga, E. and Rabbinge, R., 1997. The renaissance of mixed farming systems: a way towards sustainable agriculture. In: Proceedings of Conference on Nitrogen Emission from Grasslands', 20-22 May, 1996. Institute of Grassland and Environmental Research, North Wyke, United Kingdom, in press. Liebscher, G., 1895. Untersuchungen uber die Bestimmung des Dungerbeduffnisses der Ackerboden und Kulturpflanzen. J. Landwirtsch., 43: 49. Nomisma, 1996. Rapporto 1995 sull'Agricoltura Italiana. Agra Editrice. Rabbinge, R., 1993. The ecological background of food production. In: D.J. Chadwick and J. Marsh (Editors), Crop Protection and Sustainable Agriculture. Ciba Foundation Symposium 177, Wiley, Chichester, UK, pp. 2-29. Rabbinge, R., 1997. Integrating policy and technical issues for international research on agriculture and the environment using systems approaches. In: Teng, P.S., Kropff, M.J., ten Berge, H.F.M., Dent, J.B., Lansigan, F.P. and van Laar, H.H. (Editors): Applications of Systems Approaches at the Farm and Regional Levels. Kluwer, Dordrecht, pp. 249262. Rabbinge, R. and van Latesteijn, H.C., 1992. Long term options for land use in the European Community. Agric. Systems, 40: 195210. Seckler, D., 1996. The new era of water resources: from 'dry' to 'wet' water savings. IIMI, Colombo, 17 pp. Sinclair, T.R., 1990. Nitrogen influence on the physiology of crop yield. In: Rabbinge, R., Goudriaan, J., van Keulen, H., Penning de Vries, F.W.T. and van Laar, H.H. (Editors), Theoretical Production Ecology: Reflections and Prospects. Simulation Monographs 34. Pudoc, Wageningen, pp. 41-55. Speedy, A.W., 1994. Development and future prospects for agricultural Education in Great Britain. In: International Conference

15 on Integration of Agricultural Science Education. EC Countries. Bologna, Italy, October 27-29, pp. 73-86. TAC (Technical Advisory Committee)CGIAR, 1989. Sustainable Agriculture production. Implications for International Agricultural Research. FAO Research and Technology Paper 4, Rome. van Ittersum, M.K. and Rabbinge, R., 1997. Concepts in production ecology for analysis and quantification of agricultural input-output combinations. Field Crops Res., 52: 197-208. yon Liebig, J., 1855. Die Grundsaetze der Agricultur-Chemie: mit Ruecksicht auf die in England angestellten Untersuchungen - 2.

Durch einem Nachtrag verm. Auflage. Vieweg, Braunschweig. Whitehead, A.N., 1925. Science and the modern world: Lowell Lectures, 1925. Glasgow, 252 pp. World Commission on the Environment and Development, 1987. Our common Future. Report by the Bruntland Commission. Oxford University Press, Oxford. WRR (Netherlands Scientific Council for Government Policy), 1992. Ground for choices: Four perspectives for the rural areas in the European Community. The Hague, The Netherlands, p. 144.

This Page Intentionally Left Blank

Section 2 AGRICULTURE LAND USE, FOOD SECURITY AND ENVIRONMENT Land use transformation in Africa: three determinants for balancing food security with natural resource utilization P.A. Sanchez and R.R.B. Leakey .................................................................................................................... 19 Reprinted from the European Journal of Agronomy 7 (1997) 15-23

Agro-ecological characterisation, food production and security P. Bullock ......................................................................................................................................................

29

The potential benefits of agroforestry in the Sahel and other semi-arid regions H. Breman and J.J. Kessler .......................................................................................................................... 39 Reprinted from the European Journal of Agronomy 7 (1997) 25-33

Chemical crop protection research and development in Europe R. Neumann .................................................................................................................................................... 49

Emissions of CO2, CH4 and N20 from pasture on drained peat soils in the Netherlands C.A. Langeveld, R. Segers, B.O.M. Dirks, ,4. van den Pol-van Dasselaar, G.L. Velthof and A. Hensen .......... 57

Reprinted from the European Journal of Agronomy 7 (1997) 35-42

This Page Intentionally Left Blank

t~ 1997 ElsevierScience B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

19

Land use transformation in Africa" three determinants for balancing food security with natural resource utilization Pedro A. Sanchez*, Roger R.B. Leakey International Centerfor Research in Agroforestry, P.O. Box 30677, Nairobi, Kenya Accepted 8 April 1997

Abstract

The continued threat to the world's land resources is exacerbated by the protracted food crisis in sub-Saharan Africa. Per capita food production continues to decrease even though this region compares favorably with other tropical regions in terms of climate and soil resources. The main determinant of this situation is the widely recognized need for an enabling policy environment that favors smallholder rural development. However, there are two other key determinants to food security and environmental sustainability in Africa that have not received sufficient attention in the past and are the focus of this contribution: (1) the need to tackle soil fertility depletion as the fundamental biophysical constraint to food security and (2) the need for more intensive and diverse land use, based on the domestication of indigenous trees to produce high value products while increasing agroecosystem resilience. Approaches that include these three issues will transform smallholder farming in Africa into productive and sustainable enterprises and will contribute greatly to food security and environmental conservation, in a win-win situation. © 1997 Elsevier Science B.V. Keywords: Food security; Environmental sustainability; Soil fertility depletion; Domestication of indigenous trees; Agroindustry policy; Profitable tree crops; Soil fertility replacement; Nutrient recapitalization; Irvingia gabonensis; Prumus africana

1. Introduction

The continued threat to the world's land resources is now exacerbated by looming food shortages in the Third World, increased food prices globally and wider climatic variations resulting from global warming. During the last decade food security was not a global priority, but recent studies such as 2020 Vision (IFPRI, 1996) show that food security should be one of the main global concerns of our time. Food inse* Corresponding author. E-mail: [email protected]

curity encompasses food scarcity as well the inability to purchase food, a poverty-related issue. Although food insecurity occurs throughout the developing world, it is most acute in Sub-Saharan Africa (hereinafter referred to as Africa), where the attainment of food security is intrinsically linked with reversing agricultural stagnation, safeguarding the natural resource base and reducing population growth rates (Cleaver and Schreiber, 1994). The purpose of this contribution is to raise three determinants of food security and environmental sustainability in Africa for the next 25 years.

Reprintedfrom the European Journal of Agronomy 7 (1997) 15-23

20

2. The problem Per-capita food production continues to decrease in Africa, in contrast to increases in other parts of the developing world (Fig. 1). Analysis of trends indicate that the green revolution and enabling policies drastically increased per-capita food production in Asia and Latin America, where the race between food production and population growth is steadily being won. In contrast, the race is being lost in Africa, where percapita food production has been declining at about 2% per year since 1960 (World Bank, 1996a). In addition, Africa has about half of its population classified as absolute poor (those subsisting on per-capita incomes of less than one US dollar per day), as well as the highest proportion of undernourished children and the highest rates of population growth of any region in the world (Badiane and Delgado, 1995; World Bank, 1995a). Most of the poor in Africa live in rural areas. A 4% annual sustained growth rate of agricultural production in Africa is an absolute requirement for reversing this situation by the year 2020 (Badiane and Delgado, 1995). Improved governance, macroeconomic stabilization, and more support to the agricultural sector are widely considered the necessary conditions to reversing the trend of per-capita agricultural production (Cleaver and Schreiber, 1994). Attempts to bypass the development trajectory of Europe, North America and Asia which started by developing a strong agricultural base have simply not succeeded in Africa or elsewhere. Agriculture (in its broadest sense, which includes food crops, tree crops, livestock, forestry and fisheries) is the engine of economic growth (Brown and Haddad, 1994). Agriculture meets the country's basic needs for human 130 125 120 115 110

S. America

105 100

Africa

95 90 1980

1985

1990

1995

Fig. 1. Per capita food production indices (1979-81 = 100).

survival and provides rural employment which leads to the development of an agroindustrial sector that processes food, timber, paper etc. This in turn leads to marketing enterprises, and often to an export sector. Industrialization and service sectors arise out of this foundation. Mining of natural resources, tourism and other sectors contribute significantly to growth but are not the engine that agriculture is because these sectors do not address the basic needs of the majority of the population, nor do they provide sufficient employment opportunities for the expanding rural labor force. African countries that are relatively better off like South Africa and Zimbabwe have made agricultural development a high priority, in spite of their abundant mineral resources and a strong tourism sector. In addition to the general need for an enabling policy environment, other factors affecting African agriculture's productivity over the last 20 years are low crop productivity, low growth rate of agricultural production (2% per year), the highest rate of population growth of any region in the world (2.9% per year), and the highest rates of land degradation (30%) of usable land (Cleaver and Schreiber, 1994). Indeed, the population to land resource ratio in Africa has been highlighted as a major factor contributing to Africa's low per-capita food production, but is widely debated. While at the continental scale, Africa is not densely populated, large areas of Africa have population densities approaching that of Asia, but with a major difference in that Africa does not have the irrigation infrastructure Asia has (John Lynam, personal communication). In addition, Africa is certainly faced with the large problem of increasing population that Asia no longer has, which exacerbates the decline in per-capita agricultural production. However, Africa compares favorably with other tropical regions in terms of climate and soil resources (Sanchez and Logan, 1992; World Resources Institute, 1992). The biophysical endowment of tropical regions of Africa, Latin America and Asia is summarized in Table 1. In terms of rainfall, tropical Africa has a similar proportion of areas with drought stress as tropical Asia, while tropical America is better endowed. Tropical Africa compares favorably with the other two regions in terms of the proportion of its soils with fertility problems, such as high soil acidity resulting in

21

Table I Areal extent of soil-related constraints in the tropics of Africa, Asia and Latin America (calculated from a data base by W. Couto, North Carolina State University, based on Sanchez et al., 1982) Soil-related constraints

Tropical Africa

Tropical Asia

Tropical America

Total Tropics

Total area (million ha) % long drought stress % low in weatherable minerals % aluminium toxicity % poor drainage % steep (> 30% slope) % low cation exchange capacity % high phosphorus fixation (by Fe) % sandy % vertic (black cotton soils) % gravel layer % calcareous (Fe and Zn deficiencies) % saline % alkaline % shallow to rock (<50 cm)

1555 67 31 26 22 22 II 11 7 3 3 2 I I 1

1205 72 27 24 16 51 i 20 2 4 7 8 2 1 2

1879 45 46 43 20 32 4 32 4 I 1 I 1 I 3

4638 60 36 32 19 34 5 22 4 3 3 3 1 1 2

aluminum toxicity, high phosphorus fixation, micronutrient deficiencies in calcareous soils, salinity and alkalinity. It compares less favorably in terms of its higher proportion of soils low in weatherable minerals and low cation exchange capacity, which are largely located in the desert margins of Africa. In terms of physical soil constraints such as poor drainage, shallow soil depth, steep slopes and gravelly soils, tropical Africa ranks well. At the continental scale therefore, the climate and soils of tropical Africa compare well with other tropical regions where development has advanced further. The main determinant of Africa's position at the bottom of the development scale, therefore is the need for an enabling policy environment, rather than an inferior biophysical endowment. Two other determinants are not so widely recognized in Africa, but the authors believe to be necessary conditions for sustainable development. They are: (1) the need to tackle soil fertility depletion as the fundamental biophysical constraint and (2) the need for more intensive and diverse land use based on profitable tree crops.

declining per-capita food production of the continent (Sanchez et al., 1995; World Bank, 1995b, 1996b; IFPRI, 1996). The magnitude of nutrient mining is huge, as evidenced by nutrient balance studies. Table 2 shows large per-ha losses of nitrogen, phosphorus and potassium during the last 30 years in about 100 million ha of cultivated land in Africa. In contrast, commercial farms in North America and Europe have averaged even larger but positive nutrient balances during the last 30 years is over four times the cultivated land of Africa. This has often resulted in groundwater and stream pollution in the developed world. Nutrient mining in Africa, therefore, contrasts sharply with nutrient buildups in temperate regions. How did this situation come about? Everywhere in the world, people have settled first in high potential areas with fertile soils, adequate rainfall and moderate Table 2 Net nutrient balances in cultivated land of Africa and Europe + North America (calculated by Sanchez et al., 1995 based on studies by Smaling, 1993 and Frissei, 1978) Region

3. Soil fertility r e p l e n i s h m e n t

Soil fertility depletion in smallholder farms of Africa is beginning to be recognized as the fundamental biophysical limiting factor responsible for the

Nitrogen (kg N/ha per 30 years)

Africa -700 Europe + North +2000 America

Phosphorus Potassium (kg P/ha (kg K/ha per 30 years) per 30 years) -100 +700

--450 +1000

22

temperatures, such as in parts of the highlands of Eastern and Central Africa and the plateau of Southern Africa with fertile soils derived from basic rocks. These areas, such as the Lake Victoria basin, now have some of the highest population densities in the world. Originally such populations were supported by the high soil nutrient capital of the predominant high fertility soils (classified as Nitisols in the FAO legend and as Alfisols and eutric Oxisols in soil taxonomy). This nutrient capital has gradually been depleted, primarily through successive crop harvest removals. Leaching and soil erosion have also contributed to these losses which have exceeded nutrient inputs such as biological nitrogen fixation, manures and inorganic fertilizers (Smaling, 1993). A similar situation is now occurring in the inherently less fertile, sandy soils of the West and Southern African subhumid savannas and the Sahel, because of population increases in these more marginal areas. Static or decreasing crop yields in Africa are largely due to nutrient depletion (Borlaug and Dowswell, 1994). In addition to marked reductions in crop productivity, nutrient depletion triggers several negative side effects on farm such as less fodder for cattle, less fuelwood for cooking, smaller amounts of crop residues, and less manure from the cattle. These in turn further increase runoff and erosion losses because there is less plant cover to protect the soils from wind and water erosion. There are also major environmental externalities. Deforestation in search of the few remaining pockets of fertile land in densely populated areas often results in an almost total removal of trees from the landscape, as seen in parts of Ethiopia. Lack of tree protection at the upper parts of watersheds severely affect their functioning. Increasing soil erosion from unproductive cropland, communal grazing lands and denuded watersheds leads to silting of reservoirs, lakes and coastal areas and can lead to the eutrophication of fresh waters. Food shortages and famines become more acute during drought years. Lack of opportunities for cash income pushes people off the land and into urban areas where many cannot find productive jobs are, further taxing the limited urban infrastructure. Urban poverty, crime, and political unrest commonly follow (Homer-Dixon et al., 1993).

Fertilizers have been the traditional tool to overcome soil fertility depletion, and indeed fertilizer use is responsible for a large part of the food production increases worldwide, including in the commercial farm sector in Africa (Mokwunye and Vlek, 1986; Borlaug and Dowswell, 1994). Fertilizer use is viewed as a recurring cost of production, that must be paid for by the increased crop yields farmers obtain. Attempts to introduce this approach to smallholder farming in Africa have met with limited success, even with input subsidies and credit schemes. Current thinking on natural resource management, however, leads us to propose an alternative approach for situations where the traditional one has not worked. Development activities that supply water for agriculture, such as reservoirs and irrigation systems have long been considered as capital investments, paid for by governments and development banks. Users pay the recurrent costs such as maintenance of canals and drainage ditches on the farm. Replenishing plant nutrients cansh also be viewed as a capital investment (Sanchez et al., 1995). Restoring nitrogen and phosphorus, the two most limiting nutrients, to their original levels in the soil, in a way that maintains them and allows them to be used for many years, is a capital investment. Nutrient capital refers to the soil reserves of nutrients that will be released gradually over a time scale of years or decades (Sanchez and Palm, 1996). Nitrogen capital consists of the active and slow pools of soil organic nitrogen. Phosphorus capital consists of the same active and slow pools of soil organic phosphorus plus inorganic phosphorus fixed at the surface of iron and aluminum oxides on clay particles. Replenishment is not feasible with nutrients that are not held by soil organic matter or clay particles, such as potassium in most soils, but mechanisms exist to build up the nitrogen and phosphorus capital in soils where these elements have been depleted. The 'interest' from such capital is used for crop production for years, and with good management, the 'principal' can remain at a high level. Nitrogen needs of most cereal crops at a grain yield level of 4 t/ha can be met by the appropriate use of organic inputs in Africa (Palm, 1995). Nitrogen capital can be built-up gradually through additions of organic inputs, because much of the nitrogen not taken up by the crop plus the carbon present in the

23

organic inputs can be converted into soil organic matter. For higher yields, inorganic fertilizer nitrogen must be added (Sanchez, 1995). Phosphorus capital, however, can only be built through the additions of inorganic phosphorus fertilizers, including rock phosphates and more soluble forms, because organic sources are unable to meet more than half the crop needs (Palm, 1995). The World Bank is beginning to incorporate soil fertility replenishment as a capital investment into their lending policies in Africa (World Bank, 1996b). There is, however, a need for a substantial research agenda that addresses new issues such as the interactions between organic and inorganic sources of nutrients, and the role of trees as recyclers of the 'interest' not utilized by crops, to further increase the returns from the investment. The type of policies required to enable nutrient replenishment need also to be determined. They include conventional returns to investment studies as well as how to determine the social and environmental benefits at the national and global scales, and therefore who should pay for such capital investments. The various benefits of recapitalization resulting from the restoration of all the service flows of nutrient capital are in essence multilayered. One layer consists of the on-farm benefits, another of the national benefits and a third of the global benefits. The issue of who should pay for this recapitalization thus becomes relatively straightforward, in principle at least. Using the fundamental principle that those who benefit from a course of action should incur the costs of its implementation, three layers of costs can be distinguished, corresponding to the three layers of benefits (Sanchez et al., 1995). On-farm, maintenance costs should be borne by farmers, whereas the national and global societies should share the more substantial costs of actual phosphorus applications. This sharing should reflect the ratio of national to global benefits. Research is needed for evaluating these various layers of costs and benefits on the basis of actual field measurements.

4. Intensifying and diversifying land use Soil fertility replenishment can go a long way in boosting agricultural production and restoring food

security in Africa, but it is a necessary but not sufficient condition for sustained development. Numerous other factors have to come together as well. For example, the reduction of post-harvest losses, pests and disease attacks, soil losses by water and wind erosion, as well as the declining size of land holdings and declining human health. The last two impact on the availability of field labor which is also a consequence of family members moving to the town to make offfarm income. Clearly what is needed is a paradigm shift from policies directed only at increasing yields of the few staple food crops. This Green Revolution approach has played, and will continue to play, an important part in meeting the needs of the rural poor, but additional steps also have to be taken. Dewees and Scherr (1996) have indicated that policy scientists need 'to stretch their conceptual framework.., to consider more carefully the links between markets, the environment, household production and household welfare'. It is in this vein that the authors suggest that a further transformation is needed in the long run; intensifying and diversifying land of smallholder farms in Africa as the next determinant. President Yoweri Museveni of Uganda, in his opening address to the SPAAR meeting in Kampala, February 6th, 1996 articulated this idea very clearly. He stated 'it does not make sense to grow low-value products (maize and beans) at a small-scale; instead highvalue products should be grown at a small-scale, while low value products should be grown on a large-scale'. The obvious implication is that small-scale farming in Africa must diversify by producing a combination of high-value, profitable crops along with the basic food crops. Examples of this strategy occur in Western Kenya, where small patches, in the order of 100 m 2, of french beans are grown by smallholders contracted by an exporting company for fresh consumption in Europe. The market is assured and farmers intensively water, fertilize and weed these islands of wealth among their lower value crops. But the largest opportunities for diversification come from tree products. Traditionally people throughout the tropics have depended on their indigenous plants for fruits and everyday products of the household from medicines to fibres. The products have also provided the essential vitamins and minerals for family health and

24

through local and regional trading have generated the cash to meet the needs of the household for purchased products. Maybe it is here, in peoples' own backyard, that the solution lies. But sadly, through deforestation, the forest or woodland that used to be in the backyard is now all but disappeared for the vast majority of people in Africa. In 1992, a conference pulled together the growing amount of biophysical information on the techniques available to domesticate the wide range of the wild and overlooked species, many of which are trees (Leakey and Newton, 1994a,b; Newton et al., 1994). These 'Cinderella' species (so called because their value has been largely overlooked by science although appreciated by local people) include indigenous fruit trees and other plants that provide medicinal products, ornamentals or high-grade timber. Examples are the bush mango (Irvingia gabonensis) a nutritious fruit from the humid tropics of West Africa (Ladipo et al., 1996), Uapaca kirkiana and Sclerocarya birrea, from the Miombo woodlands of Southern Africa. Techniques are being developed to convert some of these wild species into domesticated crops in agroforestry systems including vegetative propagation and clonal selection that capture genetic diversity (Leakey and Jaenicke, 1995). Domestication involves the formulation of a genetic improvement strategy for agroforestry trees and a strategy on the use of vegetative propagation to capture the additive and non-additive variation of individual trees in tree populations (Simons, 1996). Furthermore, guidelines have been developed for determining the species priorities of farmers (Franzel et al., 1996; Jaenicke et al., 1996). The domestication strategy for these indigenous fruit tree species, as well as for Prunus africana and Pausinystaliajohimbe, two priority trees for medicinal products, is to conserve their genetic resource in living-germplasm banks and subsequently to develop cultivars for incorporation into multistrata agroforests. High-value trees can fit in specific niches on farms while leaving more open land to staple food crops, or other profitable crops such as vegetables. Timber trees can also be grown on farm boundaries with leguminous fodder trees under them. Similarly fuelwood trees can be grown on field boundaries or as contour hedges on sloping lands. In such a scheme, improved fallows become a crucial part of the crop rotation

scheme. In such farms, income is increased and diversified, providing resiliency against weather or price disruptions. Soil erosion is minimized, nutrient cycling maximized and above- and below-ground biodiversity enhanced. The farm truly approximates a functioning ecosystem. The latest definition of agroforestry summarizes this approach: a dynamic, ecologically-based, natural resource management system that, through the integration of trees in farm and rangeland, diversifies and sustain smallholder production for increased social, economic and environmental benefits (Leakey, 1996). Through domestication these tree crops could become higher yielding, have higher quality products, be more attractive commercially and diversify diets. Such progress could improve household welfare by providing traditional food and health products, boosting trade, generating income and diversifying farming systems, both biologically and economically, beyond the production of basic food crops. Generally tree crops have lower labor requirements than basic food crops and could thus allow farmers to also seek offfarm income. A new paradigm for smallholder farming in Africa emerges: one which instead of being based on a limited number of highly domesticated crops, often grown in monoculture, is based on a much greater diversity of plants that together produce food and high-value products (Leakey and Izac, 1996). What evidence is there that it will work? The African experience with homegardens provides some examples of traditional intensification. In Nigeria, for example, they are important areas with high population densities (1000 people/km2). In such areas up to 29% of the cultivated area can be in compound gardens and these produce 59% of the crop output (Okafor and Fernandes, 1987). In monetary terms the output of these gardens is 5-10 times greater than crop fields, with returns to labor also 4-8 times greater (Watson, 1990). But the concept we are advocating is a total farm approach, not just a comer of it as in the case of homegardens. A better model is found in Southeast Asia, where the productive diversity is extended from homegardens to complex agroforests. In south-west Sumatra, near Krui, farmers have for more than a century been planting damar trees (Shoreajavanica) for resin production, in mixture with a range of crops and indigen-

25

ous fruit tree species. Originally, farmers cleared the forest by slash and burn methods and planted upland rice, along with coffee, fruit and damar trees. Lowland rice patches are also part of the farming system. The area in complex damar-based agroforests is contiguous and now exceeds 10000 ha and 65-80% of the households are involved in damar production each averaging 1-5 has of these agroforests. The system insures that the farm is productive at all stages of its growth with food crops in years 1-3, coffee and bananas in years 3-8, fruits and fuelwood in years 2-20, resins, fruits and timber from year 20 onwards (De Foresta and Michon, 1994), and paddy rice all the time. In the case of damar, the resins are utilized by industries in Indonesia or exported worldwide. In 1984, the export market represented one third of the harvested volume, a trade rising from 250-400 t/year between 1972 and 1983 (Michon and de Foresta, 1994). In 1994, the damar production was expected to reach 10000 t (Dupain, 1994), at a value of US $300-400/t. Eighty percent of this trade is met by the damar agroforests. The economic value of the damar trade and its associated activities is of major significance to the villages around Krui. In 1993, the profits from damar production were US $7.2 million from sales, US $2.6 million from added value and US $1.4 million from wages. To this is added US $ 0.3 million in profits made by Krui traders (De Foresta and Michon, 1994). This analysis excludes the locally consumed products from these agroforests, e.g. fruits, vegetables, spices, fuelwood, timber, palm thatching, rattan, bamboo, fibres, as well as paddy rice. Systems similar to the damar agroforests are also practiced in Sumatra with rubber and cinnamon. The 'jungle rubber' agroforests covered more than 2.5 million ha, while cinnamon agroforests covered 42 600 ha in 1989 (Aumeeruddy, 1994). These damar, cinnamon and jungle rubber complex agroforests are probably the ultimate example of an alternative to slash and burn agriculture, being highly productive, biodiverse, and very similar in structure to natural tropical high forest. In short, they are the nearest thing to a domesticated ecosystem (Michon and de Foresta, 1996). While land use intensification caused by demographic pressure is generally associated with envir-

onmental degradation, the long-term relationship between land resource degradation and demographic pressure is not necessarily negative and linear. Harwood (1994) has described a U-shaped curve, with an initial fast rate of land degradation with subsistence agriculture, increasing with increasing population pressure. With further increases in population pressure, however, a point is reached where degradation is reversed with further land intensification and incorporation of trees within the farm. This has happened in the semiarid Machakos District of Kenya, where despite increasing population pressure since the 1930s farmers were able to reverse land degradation through an indigenous soil conservation technology that improved both crop and livestock productivity (Pagiola, 1994; Tiffen et al., 1994). Furthermore, in the more heavily populated areas of the Central Province of Kenya, where farm size is extremely small, the number of trees on farms is also expanding as farmers increasingly recognize their value (Cleaver and Schreiber, 1994).

5. Enabling policies Current policy recommendations place high priority on the revitalization of the agricultural sector in Africa (IFPRI, 1996), and some success stores are beginning to emerge (Cleaver and Schreiber, 1994). The fact that most food in Africa is produced by smallholders, often female farmers is frequently considered a major constraint to agricultural development. In contrast, these authors believe that smallscale farms can be an asset rather than a liability when supported by appropriate policies, the agricultural production boom in Asia is a product of smallholder farms and not of a shift from small to largescale farming. Such policies include improvements in land tenure, infrastructure, marketing information, credit, research, extension and access to inputs and markets at reasonable prices. Public investments to expand access to primary education to girls and improve public health services in rural areas also play an important role in this transformation process. Policy reform to seize opportunities for smallholder development and to eliminate policies that discriminate against the smallholder agricultural sector therefore remains top priority. Policy reform, therefore, is a

26 necessary but not a sufficient condition for food security and environmental conservation. In order for enabling policies to work in most of Africa the twin issues of soil fertility depletion and land-use intensification and diversification have to be tackled. Thus the vision now is of agroforestry as an integrated land use policy that combines increases in productivity and income generation, with environmental rehabilitation and the diversification of agro-ecosystems. Such a vision can be fitted to the range of situations found in the major ecoregions of Africa. According to Cooper et al. (1996), the realization of this vision, however, is going to be dependent on: •

•

•

the appreciation by the international community and the donors, of the importance of high value indigenous species in the lives and welfare of local people, as well as incentives (or the removal of disincentives) for local people to plant trees on their farms, the domestication of commercially-important indigenous tree species producing high-value products, and the development of processing and marketing infrastructure.

For this latter step, it will be important for scientists involved in domestication to work closely with the food and pharmaceutical industries since the agroforester needs to know that there will be a market, while the industry which develops the market wants to know that there is a regular supply of a uniform and high quality product, before committing capital to developing that market (Leakey and Izac, 1996). Commercialization is both necessary and potentially harmful. It is necessary because without it the market for products is small, and the opportunity for rural people to make the money would not exist. A degree of product domestication is therefore essential. On the other hand commercialization is potentially harmful to rural people, if it expands to the point that outsiders with capital to invest, come in and develop large-scale moncultural plantations. However, from the experience of the Southeast Asian complex agroforests, smallholder units producing non-timber forest products, which are also biologically-diverse, do seem to be economically viable indicating that the intensification and diversification of land uses is not a pipe-dream.

6. Conclusion

The dynamic tension between increasing population, declining farm-size, declining farm labor supply, diversification of income sources and the management of natural resources varies with regions. For SubSaharan Africa, where most of the food is produced by smaUholders this tension should encourage farmers and policy makers to replenish the fertility of exhausted soils and plant high-value trees as a diversified but intensive agroecosystem, based on profitable tree crops, as the way forward socially, economically and environmentally.

References Aumeeruddy, Y., 1994. Local representations and management of agroforests on the periphery of Kerinci Seblat National Park, Sumatra, Indonesia. People and Plants Working Paper 3, UNESCO, Paris, 46 pp. Badiane, O. and Delgado, C.L., 1995. A 2020 vision for food, agriculture and the environment in sub-Saharan Africa. Food, Agriculture and the Environment Discussion Paper 4, IFPRI, Washington, 56 pp. Borlaug, N. and Dowswell, C.R., 1994. Feeding a human population that increasingly crowds a fragile planet. Supplement to Transactions 15th World Congress of Soil Science, Acapulco, Mexico. International Soil Science Society, Chapingo, Mexico, 15 pp.. Brown, L. and Haddad, L., 1994. Agricultural growth as the key to poverty alleviation. 2020 Brief No. 7, IFPRI, Washington, 2 pp. Cleaver, K.M. and Schreiber, G.A., 1994. Reversing the spiral; the population, agriculture and environment nexus in sub-Saharan Africa. World Bank, Washington, 293 pp. Cooper, P.J.M., Leakey, R.R.B., Rao, M.R. and Reynolds, L., 1996. Agroforestry and the mitigation of land degradation in the humid and sub-humid tropics of Africa, Exp. Agric., 32: 000-000. Cunningham, A.B. and Mbenkum, F.T., 1993. Sustainability of harvesting Prunus africana bark in Cameroon. A medicinal plant in the international trade. People and Plants Working Paper 2, Paris, UNESCO, 28 pp. De Foresta, H. and Michon, G., 1994. Agroforests in Sumatra: where ecology meets economy. Agrofor. Today, 6: 12-13. Dewees, P.A. and Scherr, S., 1996. Policies and markets for nontimber tree products. EPTD Discussion paper, IFPRI, Washington, 76 pp. Dupain, D., 1994. Une regione traditionaliment agroforestiere en mutation: Le Pesisir. CNGARC, Montpellier, France. Franzel, S., Jaenicke, H. and Janssen, W., 1996. Choosing the right trees: setting priorities for multipurpose tree improvement. ISNAR Research Report No. 8., ISNAR, The Hague, The Netherlands, 87 pp.

27

Frissel, M.J., 1978. Cycling of Mineral Nutrients in Agricultural Ecosystems. Elsevier, Amsterdam, 356 pp. Harwood, R.R., 1994. Agronomic alternatives to slash-and-burn in the humid tropics. In: P.A. Sanchez and H. van Houten (Editors), Alternatives to Slash and Burn Agriculture. Symposium ID-6, 15th World Congress of Soil Science Acapulco, Mexico. International Society of Soil Science, Chapingo, Mexico, pp. 93106. Homer-Dixon, T.F., Boutwell, J.H. and Rathfens, G.W., 1993. Environmental change and violent conflict. Sci. Am., 268 (2): 16-23. IFPRI, 1996. Feeding the word, preventing poverty and protecting the Earth: a 2020 vision. International Food Policy Research Institute, Washington, WA, 28 pp. Jaenicke, H., Franzel, S. and Boland, D.J., 1996. Towards a method to set priorities among species for tree improvement research: a case study from West Africa. J. Trop. For. Sci., 7: 490-506. Ladipo, D.O, Fondoun, J.M. and Nganga, N., 1996. Domestication of bush mango (lrvingia spp. ): some exploitable intraspecific variations in West Africa. In: R.R.B Leakey, A.B Temu and M. Meinyk (Editors), Domestication and Commercialization of Non-timber Forest Products in Agroforestry Systems. NonWood Forest Products 9, FAO, Rome, (in press). Leakey, R.R.B. and Newton, A.C., 1994a. Domestication of 'Cinderella' species as a start of a woody plant revolution. In: R.R.B. Leakey and A.C. Newton (Editors), Tropical Trees: Potential for Domestication and the Rebuilding of Forest Resources, HMSO, London, pp. 3-6. Leakey, R.R.B. and Newton, A.C., 1994b. Domestication of Timber and Non-timber Forest Products, MAB Digest 17, UNESCO, Paris, France, 94 pp. Leakey, R.R.B. and Jaenicke, H., 1995. The domestication of indigenous fruit trees: opportunities and challenges for agroforestry. In: K. Suzuki, S. Sakurai, K. Ishii and M. Norisada (Editors), Proc. 4th International BIO-REFOR Workshop, BIO-REFOR, Tokyo, Japan, pp. 15-26. Leakey, R.R.B., 1996. Definition of agroforestry revisited. Agrofor. Today, 8 (l): 5-7. Leakey, R.R.B. and Izac, A.-M.N., 1996. Linkages between domestication and commercialization of non-timber forest products: implications for agroforestry In: R.R.B. Leakey, A.B. Temu and M. Melnyk (Editors), Domestication and Commercialization of Non-timber Forest Products for Agroforestry. NonWood Forest Products, 9, FAO, Rome, Italy, (in press). Michon, G. and de Foresta, H., 1994. Damar agroforests in the Passisir, Sumatra. Rainforest Alliance, May 1994, New York, 45 pp. Michon, G. and de Foresta, H., 1996. Agroforests as an alternative to pure plantations for the domestication and commercialization of NTFP's. In: R.R.B. Leakey, A.B. Temu and M. Melnyk (Editors), Domestication and Commercialization of Non-timber Forest Products in Agrofor. Syst. Non-Wood Forest Products 9, FAO, Rome, (in press). Mokwunye, A.U. and Vlek, P.L.G., 1986. Management of Nitrogen and Phosphorus Fertilizers in Sub-Saharan Africa. Martinius Nijhoff, Dodrecht, 362 pp. Newton, A.C., Moss, R. and Leakey, R.R.B., 1994. The hidden

harvest of tropical forests: Domestication of non-timber products, Ecodecision, 13 (July): 48-52. Okafor, J.C. and Fernandes, E.C.M., 1987. Compound farms of southeastern Nigeria: a predominant agroforestry homegarden system with crops and small livestock. Agrofor. Syst., 5: 153168. Pagiola, S., 1994. Soil conservation in a semi-arid region of Kenya: rates of return and adoption by farmers. In: T.L. Napier, S.M. Camboni and S.A. EI-Swaify (Editors), Adopting conservation on the farm. Soil and Water Conservation Society, AIkeny, IA, pp. 171-187. Palm, C.A., 1995. Contribution of agroforestry trees to nutrient requirements of intercropped plants. Agrofor. Syst., 30: !05124. Sanchez, P.A., 1995. Science in agroforestry. Agrofor. Syst., 30: 5-55. Sanchez, P.A. and Logan, T.J., 1992. Myths and science about the chemistry and fertility of soils in the tropics. SSSA Special Publication, Soil Science Society of America, Madison, Wl, 29: 35-46. Sanchez P.A. and Palm, C.A., 1996. Nutrient cycling and agroforestry in Africa. Unasyiva, 185 (47): 24-28. Sanchez, P.A., Couto, W. and Buol, S.W., 1982. The fertility capability soil classification system: interpretation, applicability and modification. Geoderma, 27: 283-309. Sanchez, P.A., lzac, A.-M.N., Valencia, I. and Pieri, C., 1995. Soil fertility replenishment in Africa: a concept note. ICRAF, Nairobi, 5 pp. Simons, A.J., 1996. ICRAF's strategy for domestication of nonwood tree products. In: R.R.B. Leakey, A.B. Temu and M. Melnyk (Editors), Domestication and Commercialization of Nontimber Forest Products in Agroforestry Systems. Non-Wood Forest Products 9, FAO, Rome, (in press). Smaling, E., 1993. An agroecological framework for integrated nutrient management with special reference to Kenya. Doctoral thesis, Agricultural University, Wageningen, The Netherlands, 250 pp. Tiffen, M., Mortimer, M. and Gichuki, F., 1994. More People, Less Erosion. Environmental recovery in Kenya. Wiley, Chichester, UK, 311 pp. Watson, G.A., 1990. Tree crop and farming systems development in the humid tropics. Exp. Agric., 26: 143-159. World Bank, 1995a. Towards Environmentally Sustainable Development in Sub-Saharan Africa: a World Bank Agenda. Report No. 15111-AFR, World Bank, Washington, WA, 114 pp. World Bank, 1995b. Feasibility of Phosphate Rock Use as a Capital Investment in Sub-Saharan Africa: Issues and Opportunities. Africa Technical Division, World Bank, Washington, WA, 72 PP. World Bank, 1996a. African Development Indicators 1996. World Bank, Washington, WA, 431 pp. World Bank, 1996b. Natural Resource Degradation In Sub-Saharan Africa: Restoration of Soil Fertility. Africa Region. World Bank, Washington, WA, 12 pp. World Resources Institute, 1992. World resources 1992-1993, A Guide to the Global Environment. Oxford University Press, New York, pp. 28 ! - 284.

This Page Intentionally Left Blank

© 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van Ittersum and S.C. van de Geijn (Editors)

29

Agro-ecological characterisation, food production and security P. Bullock * Soil Survey and Land Research Centre, Cranfield University, Silsoe, Bedford MK45 4DT, United Kingdom Abstract

There has been a large increase in food production in the last 30 years which has kept pace with the growing population. There is, however, concern that increased production has been at the expense of increasing degradation of the natural resources and hence is not sustainable. There is growing belief that agricultural development has paid insufficient attention to basic ecological principles and that agriculture needs to be subject to a paradigm change which incorporates these principles into current and future developments while at the same time being cognisant of the continued need for external inputs to achieve adequate production. The FAO Agro-Ecological Zones Project is a step towards agro-ecological harmony. It provides a basis for informing planners about land potential, the suitability and yield potential for some 11 major crops important in the developing world and the potential sustainability of the various cropping systems. It was one of the first schemes also to quantify the amount of potentially cultivable land in developing countries. New developments, which now allow the agroecological zones approach to be strengthened and made more available, include the availability of well organised national and international land-related databases, the development of crop suitability and production models of varying sophistication, models for environmental risk assessment including risks from the adoption of particular agricultural practices, and geographical information systems that now allow information to be put to policy makers and others in a simple and useable format. To feed the predicted future world population is a challenge on an unprecedented scale. It needs to be met with a combination of a sound agro-ecological approach to food production, biotechnological development, carefully controlled and understood inputs, improved technology, more investment in deprived areas and well thought out strategies for food production, access and trade.

Keywords: Agro-ecological zones; Crop growth models; Environmental risk assessment; Sustainable land use and management; Food production; Food security

1. Introduction

The last few decades have seen a number of major changes in agriculture and food production. Prominent among these has been the Green Revolution which has had a major impact on agriculture worldwide. Using improved plant and animal breeding, large fertiliser inputs, improved technology and other external inputs, productivity has increased greatly in many parts of the world enabling the population to

* Fax: +44 1525 863253. Tel: +44 1525 863251. E-mail: [email protected]

be better fed than at any time in history. Agricultural production grew faster than population with per caput production about 18 per cent above that of 30 years ago (Alexandratos, 1995). However, this success is moderated by two facts: (i) approaching one billion people suffer from hunger, largely because of problems of getting surplus food to where it is needed, drought and warfare and (ii) the increased food production in some areas has been at the expense of increasing degradation of our natural resources. Thus, it is estimated that some 6-7 million hectares of land are lost each year to soil erosion, about one quarter of the world's land is suffering from desertification and some 20 million hectares

30

Table 1 Sustainable land use and management principles for future food production Economic viability Strong productivity Working with indigenous land users In harmony with the soil and prevailing climate According to ecological principles Environmentally benign Preserving and preferably enhancing soil quality Careful water use and management have been largely lost through salinisation (World Resources Institute, 1992). Not all agriculture has been marked by increased productivity. Several countries and whole regions, mainly in Africa, failed to make progress. Many subsistence farmers of the developing world are experiencing decreasing yields, caught in a poverty trap which precludes them from using fertiliser to at least keep pace with the loss of nutrients from the land (Stoorvogel and Smaling, 1990; van der Pol, 1992; FAO, 1996). Most such farmers are farming in extremely fragile circumstances with, in addition to poverty, an unreliable climate and soils of depleting fertility. In Africa, food production per person has fallen by 20% since the 1960's (Pretty, 1996). Thus with both intensive and subsistence agriculture there is concern over the sustainability of agricultural production. The historical association of expansion and intensification of agriculture with pressures leading to resource degradation and adverse impacts on the wider environment is now well documented (FAO, 1996). Such pressures are likely to increase in the future. How to minimise the negative effects on the land resources will be a major challenge for future agricultural production. It is time to review, learn from history and develop a strategy for a more soundly based agriculture. That

there is a degree of will to do this is suggested by the Rio Summit and Agenda 21. Sustainable development issues are fundamental to a strategy for sound adequate food production and food security, guided by the principles outlined in Table 1. Meeting food and fibre demands of a future population will require that land currently in production produces more, that land currently not in production is brought into production and, where reasonable that degraded land is rehabilitated to a reasonable productive capacity (Pierce and Lal, 1991). There is a growing belief that agricultural development needs to be more mindful of ecological principles and rather than pursuing external inputs alone as the panacea for increased production, a better balance involving external inputs, natural processes and resource-conserving technologies is required. The concept of agro-ecological zones, if followed, would lead to a more balanced, more harmonious approach to agricultural production, necessary for future sustainability.

2. Agro-ecological principles Pierce and Lal (1991) and others have highlighted the importance of applying ecological principles to agricultural systems as part of the sustainable development strategy. Cox and Atkins (1979) and Pierce and Lal (1991) point out that agricultural systems, though managed systems, function with the same group of physical and biological laws as other ecosystems but differ from the latter in both structure and function in a number of ways (Table 2). A better understanding of the importance and implications of ecological processes in the sustainable agricultural paradigm needs to be developed. The concept of the agro-ecological approach recognises that crop

Table 2 Comparisons of Agricultural and Natural Ecosystems(Pierce and Lal, 1991) Agrosystems

Natural Ecosystems

Highly productive with huge energy and material subsidies Lower productivity: few or no subsidies Simple ecosystems, resembling early stages of succession; hence unstable More complex, higher diversity and higher stability Artifically regulated; little continuity Self-regulating and self perpetuating; considerable continuity Not adaptive Adaptive Not very diverse Diverse Artificial management Nature's management

31

production worldwide is influenced by a complex interaction of several factors, particularly the species and cultivar of the crop being grown, the nature of the climate, the soil type in which the crop is grown and the use of inputs, such as fertilisers and pesticides (FAO, 1991). It does not preclude the use of external inputs to maintain production levels but attempts to follow ecologically sound concepts and cultural practices in doing so. In the view of Sehgal et al. (1990), the need for increased food production to support the growing population 'demands an appraisal of our soil and climatic resources to recast an effective and alternative land use plan. Since soils and climatic conditions of a region largely determine the suitability of different crops and their yield potential, efforts in mapping agro-ecological regions may go a long way in identifying optimum cropping patterns for increasing agricultural production on a sustainable basis.' The food security of future populations will be dependent on developing sustainable agriculture. A combination of an agro-ecological approach and a better understanding of the significance of ecological processes and their application in crop production systems are essential if this is to be achieved.

3. The Agro-Ecological Zone (AEZ) concept

0FAO 197S, ~991) The Agro-Ecological Zone concept is an important one in bringing world agriculture back towards ecologically based, more sustainable principles. FAO has been one of the chief developers of the concept and some 15 years ago introduced the AEZ system in support of developing countries. The types of question that the establishment of the FAO agro-ecological zones aimed to address were (FAO, 1991): how much potentially arable land is available and its location what crops can be grown, and what are their potential yield what level of inputs are required to produce the necessary yields what is the risk of land degradation and how can it be minimised

which crops in which areas would produce the greatest return given more resources what would be the cost of realising these returns what are the maximum returns available where should future research be concentrated

3.1. The parameters for delimiting agro-ecological zones

A number of sets of information were assembled for the FAO agro-ecological zone delimitation: Crop Requirements: To define, with respect to climate and soil, the needs of the following crops: wheat, paddy rice, maize, pearl millet, sorghum, soybean, cotton, phaseolus bean, white potato, sweet potato and cassava (Fig. 1). Climate Information" Emphasis was given to temperature and water availability, the combination of which is expressed in the growing period. The length of the growing period is assumed to be the continuous period from the time when rainfall is greater than half the potential evapotranspiration until the time when the rainfall is less than the full potential evapotranspiration, plus a number of days required to evaporate an assumed 100 mm of soil moisture reserve when available (FAO, 1978, Sehgal et al., 1990). Soil Information: The AEZ project used the FAO 1 "5 million soil map as its information base. Four basic variables are included on the soil map. The soils themselves are divided into 26 major groups and further subdivided to give a total of 106 individual soil units. The map also specifies three categories of texture and of slope. There is also additional information in the form of different soil phases, such as stoniness and salinity. The Land Inventory: In the FAO project, the climatic inventory was combined with the soil map to produce a land inventory map which contains the accumulated data on climate, thermal regime, length of growing period zones, soil associations, slope, phase and texture. The map represents the information base of the study. Calculations were then made of the maximum potential yield of each crop in each zone, taking into consideration how ideal the soil conditions were, problems of soil management where

32

Choice of crops and identification of their climatic and soil requirements Assembly of agroclimatic data Assembly of soil data Assessing agroclimatic suitability Assessing land potential for particular crops

Potential population supporting capacity

Fig. 1. Agro-ecological zones production stages.

climatic conditions were less than ideal, to produce a land suitability assessment for each particular crop. The AEZ approach can provide answers to questions that are an important aid to planning and development for food security. These include (FAO, 1991):which crops are best suited to the soil and growing conditions in any country? what agricultural policies would make best use of land resources of a particular country? where are the potential food-deficit areas that cannot attain self-sufficiency? where are the best prospects for examining production including cash crops? which areas, with high production potentials, could provide food for, or support people moving from food-deficit areas? what levels of farming inputs and soil conservation need to be developed to meet specific targets for sustainable self-sufficiency and crop exports? what are the future requirements for seed, fertilizer, pesticide and agricultural power to meet future population demands? The AEZ project was one of the first to quantify systematically the extent of potentially cultivable land in developing countries, to provide estimates of crop yields under varying levels of inputs and helped to provide a basis for identifying the popula-

tion-supporting capacity of developing lands (Higgins and Kassam, 1981 ; FAO, 1991. It has been used to inform planners about land potential, which crops are most suitable for a given area and what levels of sustainable production can be expected under different levels of input. It also provided a balance sheet to test whether agricultural strategies were compatible with land resource potentials. The conclusions reached by the agro-ecological zone project were that in 1975 some 55 out of the 117 countries studied could not support their populations from their own land resources with low-input production systems. The area of these countries was 32.5% of the entire developing world. However, the situation changed dramatically if the level of inputs was improved. By the year 2000 the number of countries unable to support their own populations was calculated to be 65 if low inputs were applied, 36 at intermediate inputs, 19 even if high inputs were applied. The project was important also in estimating that there was potential for about 300 per cent increase in the existing arable area in the developing world, but more importantly, it enabled the potential land available to be determined on a country or regional basis. In particular, it showed that while some countries have large amounts of land currently unused for agriculture but suitable for cultivation, e.g. many African and south America countries, in other parts of the world such as southern Asia and the Near East/North Africa, the amount of currently unculti-

33

Table 3 Land With Crop Production Potential in Developing Countries - Land in Use and Land Balance (million ha) (Alexandratos, 1995)

Sub-Saharan Africa Near East/North Africa East Asia (exc. China) South Asia Latin America and Caribbean

Land in Use 1988-90

Balance

Land in Use 2010

Balance

213 77 88 191 190

797 16 97 38 869

255 81 103 195 217

754 13 82 34 842

vated land available to feed increasing populations is relatively small. Predictions of the land use in 19881990 and in 2010 and the balance of land available have been given by Alexandratos (1995), (Table 3).

sustainable agricultural development and this involves adopting the principles of the agro-ecological zones.

3.2. Adoption of the AEZ approach

4. Developing on the AEZ Foundation

The AEZ approach has much to offer in terms of assessing the capability of a country to feed its current and future population. It has been promoted by FAO who have applied the system in, for example, Mozambique (FAO, 1982), Bangladesh (FAO, 1988) and Kenya (FAO 1991). The Technical Advisory Committee of the Consultative Group on International Agricultural Research (CGIAR) has adopted the FAO AEZ approach to subdivide some of the main climatic zones into rainfed moisture zones and thermal zones (TAC/CGIAR, 1992). Their approach was weakened by the exclusion of soils and landform data on the grounds of keeping the number of subdivisions to manageable proportions but provides valuable information on land use patterns and major food crops in the different regions of the developing world. Recently, Sivakumar and Valentin (1997) have examined the crop production potential of different developing region AEZs including soil and landform constraints. However, despite the information it is able to provide and the message it carries, few countries in the developing world have integrated the agro-ecological zone approach into their strategic agricultural development. In view of the fact that the AEZ approach is in harmony with strategies for sustainable development, surely it is only a matter of time before its approach becomes commonplace. To date, rather little attention has been given to using the AEZ approach in developed countries, the implication being that much more is known about the carrying capacity and potential of land there. But developed countries also need to pay attention to

In the last decade there have been at least four major developments which permit significant advances beyond the AEZ project.

4.1. National and international databases The first of these advances relates to improvements in national and international databases. Generally, climatic databases are of long standing and although the detail and types of measurement varies from country to country, the databases are well established. Developing countries fall short of adequate climatic information and this and the vagaries of the climate make crop suitability modelling and yield predictions difficult. Concern in the last decade about climate change and the need to improve General Circulation Models (GCMs) is impacting positively on the collection and management of climatic data worldwide (Houghton et al., 1996) and better datasets are likely to become available. Soils databases have a much shorter history than those relating to climate but there have been major developments in these in recent years. At an international level there are now several databases" the FAO soil database, the one used in the AEZ project; GRID, a UNESCO-organised database including both soils and climatic information; and the International Geosphere-Biosphere Programme is also promoting an international soil database involving inputs of data from FAO, the US Soil Conservation Service, ISRIC and SSLRC on behalf of the European Soil Database. Within Europe there is spatial information about

34 soils available at a scale of I'IM, including a digital version of the map supported by an analytical database (Breuning-Madsen and Jones, 1994), which together constitute the European Soil Database. The amount and scale of information about national soils is variable. That for many countries of Europe is among the most comprehensive and best organised in the world. Le Bas and Jamagne (1996) provide an update of the mapping and supporting databases available for each of the countries of the European Commission.

4.2. Crop growth and crop suitability models The second significant improvement of recent years has been the development of models of crop growth and crop suitability. Models are now available for most of the major crops and these continue to be refined. Several such models predict the growth and production of annual crops according to crop species, soil type, moisture conditions and weather during the growing season. Some are concerned with a single crop, others are broadly based across a number of crops. Some of the principal models include:CRIES"

Comprehensive R_.esourceIndustry and E_valuation System. Includes a yield model which allows prediction for some 30 different crops (Schultink, 1987). DSSAT (D...ecisionSupport S_ystem for Agrotechnology Transfer). Contains a number of crop models (Singh et al., 1990). WOFOST (W...~orldF..~oodStudies). Models the growth and production of annual crops (van Diepen et al., 1989). Now in Version 6.0. ALMANAC (Agricultural Land Management _Alternatives with Numerical Assessment Criteria). A predictive and process orientated model simulating hydrology and erosion; able to simulate competition between 2-10 species (Jones and O'Toole, 1987). EPIC (Erosion Productivity Impact Calculator). Predicts the long term effects of various components of soil erosion on crop production (Williams et al., 1990). ACCESS (Agroclimatic Climate Change and European Soil Suitability). Models the impact of climate change on the suitability of land for strategic European crops (Loveland et al., 1994).

A recently produced Register of Agro-Ecosystems Models (Plentinger and Penning de Vries, 1996) lists

some 200 different models currently available, the number giving a clear indication of the efforts being put into the development of simulation and other models. The potential and attainable levels of food production in the different agroclimatic regions using a modelling approach has been discussed by Penning de Vries (1997).

4.3. Environmental risk assessment Although problems of land degradation associated with some agricultural systems have been long acknowledged, a much wider group of environmental problems are now recognised e.g. water quality, air pollution, soil contamination. There are now a large number of environmental risk assessment models available with which a particular agricultural land use and management can be assessed from the point of view of likely damage to the environment (Plentinger and Penning de Vries, 1996). Examples include:LEACH M

EPIC EUROSEM SUNDIAL

(Leaching Estimation and Chemistry Model). The model includes sub models for water flow, pesticide movement, nitrogen dynamics and salinity (Wagenet and Hutson, 1989). (see above) for soil erosion. (European Soil Erosion Model) (Morgan et al., 1994). (Simulation of Nitrogen D_.ynamicsin Arable Land) (Bradbury et al., 1993).

4. 4. Use of Geographical Information Systems (GIS) The fourth major advance is that of the development and widespread use of GIS in predictive modelling for both crop suitability and environmental risk. The beneficial effects of being able to integrate many different types and sources of information and provide output according to selected models is huge. One of the reasons that uptake of the AEZ approach by national planners has been disappointing is the lack of availability in the past of suitable inexpensive tools to handle multiple databases. Perhaps more than any other single tool in recent years, GIS has revolutionised the opportunities for expert systems which can be integrated into decision support systems through which agroecological zones can be better defined, crops and their management systems se-

35

recognition that food production is not only about economics. While economics are very important, agricultural systems will need to be managed in the future in a different way to what they have been in the past, with more consideration of the land resource itself and the wider environment (Fig. 2). It needs to be recognised that much land in the future that is marginal to agriculture may need to be used for food production. Such lands are very vulnerable to damage and will need to be managed with care to achieve both adequate productivity and protection of the land resource. a better understanding of the needs and inputs of agricultural systems within the environment in which they are being applied. Thus for a given crop, multiple crops or rotational systems, it is reasonable to need to know their impact on the land resource after 5 years, a decade and so on, i.e. what is the anticipated lifespan of the system in that particular agro-ecological zone and what changes can be made to prolong that life and still

lected and by which a sustained land use strategy can be developed. Not only has GIS improved the technological handling and management of data but has provided a means of information transfer to the policy maker on the one hand and the land user on the other in a simple, understandable form.

5. An agro-ecological approach to future food production The Agro-Ecological Zones (AEZ) project of FAO provides an important contribution to the attempt to build sustainable agricultural production for the future. The following are now needed:delimitation of agro-ecological zones world wide using the latest databases, models and information about the agricultural systems likely to be involved in future production. Policy makers in every country need to be made aware of the importance and urgency of developing food production strategies based on agro-ecological principles.

MANAGING THE LAND RESOURCE

MAXIMISIN PROFIT AND PROFITABILITY

PREVENTING SOIL DEGRADATION

'REDUCING ENVIRONMENT POLLUTION

REDt rCING EMISSIONS FROM SOIL TO ATMOSPHERE

•

Economic viability

•

Appropriate land use

•

Soil erosion control

•

Monitoring land use change

•

Technology development

•

Conservation technology

•

Crop nutrient management

•

Carbon sequestration

•

Growing the soil to grow the crop

•

Soil enhancing cropping systems

•

Reducing industrial emissions

•

Managing CH4 source and sink

•

Farming by soil, landscape and climate

•

Managing the noncrop period

•

Catchment management

•

Catchment management

•

•

Fanning by soil type, landscape and climate

Fig. 2. Managing the land resource (after Pierce and Lal, 1991).

Managing fertiliser USe

36

maintain a suitably productive system. a research and development programme that is strongly focused on improving the quality and productivity of agricultural systems. There will be great dependence on biotechnology to play a major part in this, perhaps in the course of time to develop crops specifically for given agro-ecological zones. Advances in biotechnology will need to be supported by research into managing water movement and water storage in soils, manipulation of nutrient dynamics, harnessing micro-organisms, seeking more efficient fertiliser use and improving organic matter levels. greater awareness on behalf of the policy makers, development of national strategies for food production and security and linking of these world wide to manage food production and security. At a global strategic level, special attention and recognition will need to be given to sub-Saharan countries, and to others caught in the trap of subsistence farming. Strategies will need to be developed for increasing their productivity in a sustainable manner. Much of this will depend initially on large scale investment and on adequate mechanisms for technology transfer.

6. Overview of food production and food security

There have been enormous increases in food production in the past 50 years in the developed countries and this has more than kept pace with the population increase in those countries. In some situations there has been overproduction and storage problems of unused food have tended to make the headlines. The problems of transferring the excess food from one area to another with need for food have not been adequately solved. Dominated by subsistence farming, productivity appears to be declining in sub-Saharan countries rather than increasing (Harrison, 1987; FAO, 1993) although pockets of improvement are the basis for restrained optimism (Pretty, 1996). The rehabilitation and development of suboSaharan agriculture is a critical regional problem with international dimen-

sions. The problem of improving food consumption levels in sub-Saharan countries needs to be given centre stage in any debate on the future of the world food economy (Alexandratos, 1995). A number of studies suggest that the lands of the world could provide increased productivity and food production provided current trends in food production supported by improved technology, plant breeding and new agricultural systems can be sustained. There is much concern, however, about the sustainability of current agriculture (e.g. Kendall and Pimentel, 1994), because of the problems of increasing land degradation, dependence on fertiliser and limited water supplies which may ultimately prevent agricultural production keeping pace with the needs of a hugely increasing world population. This challenge to increase food production must be met with a combination of biotechnological development, improved technology, more investment in currently deprived areas and well thought out strategies for food access, trade and production. To feed the predicted future world population is a challenge on an unprecedented scale. Some comfort can be taken from the progress that has been made in the last 30 years but the next 30 years will require an even greater advance. Agricultural production needs to move hand in hand with sustainable development. The principles of farming in harmony with soil and climate rather than against them are fundamental to a future sustainable agricultural production and to food security.

References Alexandratos, N. (Editor), 1995. World Agriculture Towards 2010 - An FAO Study. FAO and John Wiley and Sons, Chichester, 488 pp. Bradbury, N., Whitmore, A.T., Hart, P.B.S. and Jenkinson, D.S., 1993. Modelling the fate of N in crop and soil in the years following application of 15N labelled fertiliser to winter wheat. J. Agric. Sci., 121: 363-379. Breuning-Madsen, H. and Jones, R.J.A., 1994. A Soil Profile Analytical Database for the European Union. Report for the Joint Research Centre, ISPRA, Italy. Cox, G.W. and Atkins, M.D., 1979. Agricultural Ecology: An Analysis of World Food Production Systems. W H Freeman and Co., San Franciso. 721 pp. FAO, 1978. Report of the Agro-Ecological Zones project. Volume 1. FAO, Rome.

37

FAO, 1982. Assessment of land resources for rainfed crop production in Mozambique. FAO, Rome. FAO, 1988. Land resources appraisal of Bangladesh for agricultural development. FAO, Rome. FAO, 1991. How Good the Earth. FAO, Rome. 33 pp. FAO, 1993. Agriculture: Towards 2010. Conference C93/94. FAO, Rome. FAO, 1996. World Food Summit. Volume 1. Technical background documents 1-5. FAO, Rome. Harrison, P., 1987. The Greening of Africa. Paladin Grafton, London. 380 pp. Higgins, G.M. and Kassam, A.H., 1981. The FAO agro-ecological zone approach to determination of land potential. Pedologie, XI: 147-168. Houghton, J.J., Meiro Filho, L.G., CaUander, B.A., Harris, N., Kattenberg, A. and Maskell, K. (Editors), 1996. Climate Change 1995-The Science of Climate Change. Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel On Climate Change. Cambridge University Press. Jones, C.A. and O'Toole, J.C., 1987. Application of crop production models in agro-ecological characterization: simulation models for specific crops. In: A.H. Bunting, (Editor), Agricultural Environments, CAB International, pp 199-210. Kendall, H.W. and Pimentel, D., 1994. Constraints on the expansion of global food supply. Ambio, 23: 198-205. Le Bas, C. and Jamagne, M., 1996. Soil Databases to Support Sustainable Development. Joint Research Centre, European Commission, EUR 16371 EN. 149 pp. Loveland, P.J., Legros, J.P., Rounsevell, M.D.A., de la Rosa, D. and Armstrong, A.C., 1994. A spatially distributed soil, agroclimatic and soil hydrological model to predict the effects of climate change on land use within the European Community. Transactions 15th World Congress of Soil Science. International Society of Soil Science, Commission V, Symposia Volume 6a, 83-100. Morgan, R.P.C., Quinton, J.N. and Rickson, R.J., 1994. Modelling methodology for soil erosion assessment and soil conservation design: the EUROSEM approach. Outlook Agric., 23: 5-9. Penning de Vries, F.W.T., 1997. (in press). Potential and attainable food production levels in different regions. In: D.J. Greenland, P.J. Gregory and P.H. Nye (Editors), Land Resources: On the Edge of the Malthusian Precipice? Philosophical Transactions of the Royal Society, Series A, 355. Pierce, F.J. and Lal, R., 1991. Soil management in the 21st century. In R. Lal and F.J. Pierce (Editors), Soil Management for Sustainability. Soil and Water Conservation Society, USA. pp. 175-179 Plentinger, M.C. and Penning de Vries, F.W.T., (Editors), 1996.

CAMASE - Register of Agro-ecosystems Models. Version II. DLO Research Institute of Agrobiology and Soil Fertility, Wageningen. 411 pp. Pretty, J.N., 1996. Could sustainable agriculture feed the world? Biologist, 43: (3). 130-133. Schultink, G., 1987. The CRIES resource information system: computer aided spatial analyses of resource development potential and development policy alternatives. In: Beek, K., Burrough, P.A., and McCormack, D.E. (Editors), Quantified Land Evaluation Procedures. ITC Publication No. 6. Enschede, The Netherlands. pp 95-99. Sehgal, J., Mandal, D.K, Mandal, C. and Vadivelu, S., 1990. Agro Ecological Regions of India. National Bureau of Soil Survey and Land Use Planning, Nagpur, India. Publication no. 24, 73 Pp. Singh, U., Tsuji, G.Y. and Godwin, P.C., 1990. Planting new ideas in DSSAT: the CERES-Rice model. Agrotechnoi. Transf., 10: 1-6. Sivakumar, M.V.K. and Valentin, C., 1997. (in press). Agro-ecological zones and the assessment of crop production potential. In: D.J. Greenland, P.J. Gregory and P.H. Nye (Editors), Land Resources: On the Edge of the Malthusian Precipice? Philosophical Transactions of the Royal Society, Series A, 355. Stool'vogel, J.J. and Smaling, E.M.A., 1990. Assessment of soil nutrient depletion in Sub-Saharan Africa, 1983-2000. Report 28, The Winand Stating Centre for Integrated Land, Soil and Water Research, Wageningen. TAC/CGIAR (Technical Advisory Committee of the Consultative Group on International Agricultural Research), 1992. Review of CGIAR Priorities and Strategies. Part I. CGIAR, Washington DC. van der Pol, F., 1992. Soil mining, an unseen contributor to farm income in southern Mali. Bulletin 325: Royal Tropical Institute, Amsterdam. van Diepen, C.A., Wolf, J., van Keulen, H. and Rappoldt, C., 1989. WOFOST: a simulation model of crop production. Soil Use Management, 5: (1) 16-24. Wagenet, R.J., and Hutson, J.L., 1989. LEACH-M - A process based model of water and solute movement, transformations, plant uptake and chemical reactions in the unsaturated zone. Continuum - Water Resources Institute, Vol. 2. Version 2. Cornell University, USA. 148 pp. Williams, J.R., Dyke, P.T., Fuchs, W.W., Benson, V.W., Rice, O.W. and Taylor, E.D., 1990. EPIC-Erosion/Productivity Impact Calculator. 2. User Manual. US Department of Agriculture, Technical Bulletin No. 1768. 127 pp. World Resources Institute 1992. World Resources 1992-93. Towards Sustainable Development. Oxford University Press, Oxford. 385 pp.

This Page Intentionally Left Blank

© 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

39

The potential benefits of agroforestry in the Sahel and other semi-arid regions

H. Breman a,l'*, J.J. Kessler b aAB-DLO, Research Institutefor Agrobiology and Soil Fertility, P.O. Box 14, 6700AA Wageningen, The Netherlands bAlD Environment, Donker Curtiusstraat 7-523, 1051 JL Amsterdam, The Netherlands

Accepted 1 July 1997

Abstract This article summarises results of three related studies. The first study is a synthesis and analysis of available knowledge on woody plants in semi-arid regions with an emphasis on the Sahel region, investigating the potential 'added value' of woody plants to improve nutrient and water availability for agricultural crop and pasture production goals. The other two are field studies aimed to validate key issues identified by the synthesis study, through measurements on the influences of woody plants on water and nutrient availability, and investigations of the relationship between soil organic matter quantity and quality and the recovery rate of nitrogen and phosphorus from fertilisers. The field studies support the conclusion that processes leading to an added value of woody plants in agroforestry systems are mainly related to reduced losses of water and nutrients. Such added values are therefore lowest where they are most needed, in resource-poor environments. Specific farmers goals, agroecological and socio-economic conditions have to be taken into account to make optimal use of the potentials of agroforestry. The potential benefits of agroforestry systems are mainly in terms of the improved efficiency of nutrient inputs than as an alternative for fertilisers. This has major implications for design and management of agroforestry systems. © 1997 Elsevier Science B.V. Keywords: Agroforestry; Soil fertility; Water availability; Sahel; Systems analysis

1. Introduction Although drought years brought the Sahel countries into publicity in the early seventies, water scarcity appears not to be the main bottleneck for rural devel-

* Corresponding author. ~Present address: IFDC-Africa, P.O. Box 4483, Lom6, Togo, Africa.

opment in this region. Extremely poor soil fertility, extreme aridity of a long dry season causing high fragility of perennial plant communities, rapid turnover of organic matter, and unfavourable socio-economic conditions are more important constraints (e.g. Penning de Vries and Djit~ye, 1982; Van Keulen and Breman, 1990). A Malian-Dutch research project called 'Production Soudano-Sah61ienne (PSS)' aimed to alleviate these constraints (Production Soudano-Sah61ienne, 1991), recognising:

Reprinted from the European Journal of Agronomy 7 (1997) 25-33

40

• • •

the urgent need to improve soil fertility (Penning de Vries and Djit~ye, 1982); the relative land scarcity and shortage of organic fertilisers (Breman et al., 1990); the unfavourable price relations of chemical fertilisers and agricultural products, given the low average recovery rate of nutrient inputs (Wooning, 1992).

Agroforestry is defined as a land-use system in which woody plants are grown in association with agricultural crops, pastures or livestock. Apart from diversification of the system with tree crops, agricultural crop and/or livestock production are expected to be improved by the interaction with woody plants, mainly through their capacities of deep rooting, nitrogen fixation and soil conservation, all leading to improved soil fertility at the benefit of agricultural crops and pastures (e.g. Young, 1989). Agroforestry has been widely promoted as a more sustainable agricultural production system, and would be particularly attractive for developing countries where the use of external inputs is not feasible (e.g. Winterbottom and Hazelwood, 1987). More recently, evidence has been generated to support the hypothesis that 'ecological agriculture technologies' such as agroforestry are important mainly in view of their potential to improve soil organic matter status (improved quantity and quality), which will improve the economic feasibility of using fertilisers through a higher recovery rate. Improved soil management is crucial for sustainable intensification of agriculture in the Sahel region (Kon6 and Groot, 1996; T6m6 et al., 1996; Breman and Sissoko, 1997). The PSS project also analysed the potential 'added value' of woody plants in terms of improved water and nutrient availability for plant production, in other words their increased accessibility within the rooting zone. Potential benefits of agroforestry in the Sahel region were identified, based on farmers' goals and Sahelian site conditions. Two aspects were characteristic for the studies executed: •

•

the role of woody plants as an option to increase the recovery rate of fertilisers, not as an alternative for fertilisers; the use of a systems approach to quantify the potentials of woody plants to improve water

and nutrient availability for primary production in relation to variable site conditions. Following presentation of the methodology of the synthesis study and field validations, this article presents the results, the consequences of the findings for potentials of agroforestry, and conclusions in terms of constraints and opportunities. As a result of the systems approach, using the absolute and relative availability of nutrients and water as main variables to determine potentials for specific regions, the conclusions are not only applicable to the Sahel region, but to semi-arid regions in general.

2. Methodology This article summarises results of three related studies. The first concerns a synthesis and analysis of available knowledge on woody plants in semi-arid regions (Breman and Kessler, 1995). More than 500 publications on woody plants in semi-arid agro-ecosystems were analysed, with an emphasis on the Sahel region, using insights on eco-physiological processes of primary production as a guiding framework (Penning de Vries and Djit~ye, 1982). This study investigated the potential 'added value' of woody plants to improve nutrient and water availability in rangelands and agricultural lands. Estimates were made of the maximum possible benefits from woody plants, by quantifying all relevant eco-physiological processes influenced by woody plants (Tables 1 and 2). Maximum use was made of data from the Sahel region, and other (semi-)arid regions (Australia, India and Northern America mainly) to fill in gaps of knowledge, while taking into account the differences in ecological conditions, e.g. through modelling and simulation. Amongst others, simulation models were used to assess the influence of woody plants on the microclimate, to quantify light interception by woody plants, and to quantify organic matter dynamics. Nitrogen (N) availability for woody plants and herbaceous plants in a mixed vegetation, was estimated in relation to variable woody plant canopy cover and the number of leaf layers of their crowns, using a formula which had been developed for a vegetation with annual and perennial plant species (Breman and De Ridder, 1991).

41

Table 1 The maximum effects of woody plants on water availability in a mixed vegetation in the Sahel zone (150--600 mm) and the Sudan zone (600-1200 mm), in comparison to a vegetation without woody plants Process

Sahel zone

Sudan zone

Stem flow: deep infiltration Improved soil structure: less run-off

+ +

+4++÷

Improved soil storage capacity Micro-climatic changes Hydraulic lift uptake by deep roots: less percolation losses

0 + 0 +

+ + + ++

0, negligible effects; +, ++, ++, water availability increases 10-50, 50-100 or > 100 mm/year.

The synthesis study was necessary to focus the other two field studies as part of the PSS project. One of the research clusters studied the possibility to increase the recovery of fertiliser nutrients in agriculture through the improved organic matter status of the soil, amongst others through the integration with woody plants. Since the duration of the project was too short to develop comparable situations with and without well developed woody plants, two alternative study approaches were applied. Study A aimed to quantify the processes by which woody plants can influence water and/or nutrient availability for primary production and soil fertility changes. Study B aimed to quantify the recovery of N and phosphorus (P) for soils with variable organic matter status. Study A was executed on the Niono ranch, 15 km East of Niono in central Mali (Southern Sahel climate zone, 550 mm annual rainfall, altitude 300 m, deep sandyloam soils); study B was executed on the same ranch and more to the South on the Cinzana and the N'Tarla research stations (Northern and Southern Sudan climate zones, 800 and 1000 mm annual rainfall, respectively, similar altitude and soils). Study A included measurements of key tree parameters of Acacia seyal and Sclerocarya birrea (Soumar6, 1996; Breman and Sissoko, 1997). Studies on the area where water and nutrient availability were influenced by woody plants, were executed for Acacia

senegal, Acacia seyal, Balanites aegyptiaca, Combreturn ghasalense, Commiphora africana and Sclerocarya birrea. Within a circle with a radius of 2 m

around the stem the whole root system was excavated. In addition, some lateral roots were entirely excavated; the total length of all lateral roots was determined using the ratio between base diameter and length of excavated roots. The water balance was monitored during 1 year, using eight rain gauges: four under the crown (in each principal direction) to measure through fall, and four at a distance of twice the crown diameter to measure real rainfall. Stem flow was measured by a stem spanned collector. Water interception by the crown was calculated as rainfall minus through fall and stem flow. Evaporation under the crown and beyond the crown area was measured by a microlysimeter, and the calculated water balance was verified by monitoring soil humidity (weighing soil samples before and after drying). The influence of woody plants on soil fertility was determined by measuring foliage and litter production (Soumar6, 1996) and by taking soil samples to determine the enrichment factor for carbon (C), N and P (Radersma, 1996; Soumar6, 1996; Breman and Sissoko, 1997). For the six woody species mentioned, soil samples were taken (at 0 - 2 0 , 2 0 - 5 0 and more than 50 cm soil depth) under the tree crowns (three in each principal direction) and at a distance of two and

Table 2 The maximum effects of woody plants on nutrient availability in a mixed vegetation in the Sahel zone and the Sudan zone, in comparison to a vegetation without woody plants Process Spatial concentration Uptake by lateral roots Capture of wind blown material Deposition by animals Reduction of losses Decreased wind erosion Decreased water erosion Decreased leaching Decreased fire (volatilisation) Internal recycling External recycling Enrichment of the system Uptake by deep tap roots Nitrogen fixation P-uptake through mycorrhiza

Sahel zone

Sudan zone

+ + +

++ + +

+ + + + + +

+ ++ ~,~, + ++ ++

0 0 0

+ + +

0, negligible effects; +, ++, +++, N increase I-5, 5-10 or > 10 kg/ ha per year; P increase 10% of N increase.

42

three times the crown radius (one in each principal direction). The enrichment factor is one way to express the influence of woody plants on soil fertility, and is defined as the ratio of nutrient concentration under a woody canopy and in the open field. Per tree species, three trees were studied, but eight trees for Acacia seyal and Sclerocarya birrea. Light interception was measured with a photometer (trade-mark DECAGON), starting in April, when trees are without leaves, until maximum tree foliage biomass in August. Foliage biomass was estimated using the allometric relations established per species by Hiernaux et al. (1992). Measurements were done for four trees of both Acacia seyal and Sclerocarya birrea, twice a month, at 8, 10, 12, 14 and 16 h, with 44 observations per tree (11 in each principal direction at equal distances from each other: five under the crown, five beyond the crown and one in the open field). Study B investigated the relationship between soil organic matter quantity and quality and the recovery rate of N and P from fertilisers. In each of the climate zones mentioned above, three soil types were studied: sand, loam and clay (Kon6 and Groot, 1996; Breman and Sissoko, 1997). The 'three quadrant methodology' (Van Keulen and Wolf, 1986) was used to measure the recovery of N and P. Comparisons between sites with a low and a relatively high soil organic matter status were made in three ways: (1) using land that had just been abandoned by farmers after years of production, and a more than 10 year old fallow; (2) using a degraded crop land, and a productive land densely covered for more than 40 years with the perennial grass Andropogon gayanus; and (3) following annual fertiliser application, monitoring changes in N and P by recovery by grasses (the annual Pennisetum pedicellatum and the perennial Andropogon gayanus) and leguminous herbs (Stylosanthes hamata and Vigna unquilata).

increase nutrient and water availability, as compared to a herbaceous vegetation only. Tables 1 and Table 2 present results for the Sahel zone (150-600 mm/year of rainfall) and for the subhumid Sudan ('savannah') zone (600-1200 mm). The effects by each indicated process refer to the maximum effects for a mixed vegetation with woody plants as a whole, and not to the area covered by woody plants (fraction of the soil surface occupied by the projection of their crowns). These effects are expected for a cover of woody plants as found in the 'natural vegetation' of good condition before the drought of the seventies, generally considered as the maximum cover possible: 2-20% in the Sahel zone and 15-35% in the Sudan zone (variation according to soil type). On the basis of the results of the analyses carried out as part of the synthesis study, it was concluded that the added value of woody plants largely depends on the processes that reduce losses of nutrients and water, mainly through soil improvement. The added value increases with increasing rainfall, from the Sahel to the Sudan zone. The following woody plant species characteristics were identified as having maximum potential to increase water and nutrient availability for primary production by the processes indicated:

3. Results

As regards site conditions, maximum benefits in terms of improved nutrient availability can be expected on well-drained, deep and fertile soils. Maximum benefits on water availability can be expected on fine-textured soils with high moisture contents and where potential evapotranspiration is high (e.g. in fine-textured valleys in arid zones).

3.1. Synthesis study To analyse the potential 'added value' of woody plants, Breman and Kessler (1995) defined and quantified the processes through which woody plants can

• • • • • • • •

the capacity to generate stem flow; deep rooting, preferably reaching the ground water table; a tall, closed canopy and an evergreen phenology; unpalatibility of foliage and fruits, for maximum contribution to soil improvement; a slow growth rate and long life span; high efficiency of internal nutrient cycling; the capacity for nitrogen fixation and presence of mycorrhiza; foliage properties allowing maximum contribution to soil organic matter status (quantity and quality).

43

The effects of the separate processes cannot be simply added to obtain an overall effect of the mixed vegetation, because several processes are more or less incompatible in view of the required properties of the woody species. Examples of incompatible processes are 'increased moisture retention of the soil through improved soil structure' and 'uptake by deep roots of percolated water' (Table 1), or 'decreased leaching of nutrients' and 'nutrient uptake by deep roots' (Table 2). Based on the relative availability of water and nutrients, water is the factor limiting plant growth in the Northern Sahel zone, in areas with run-off in the Southern Sahel zone, and elsewhere in areas with strong run-off. In other words, in the Sahel zone the limited improvement of water availability, in absolute terms, due to the influences of woody plants, may be relatively important. On the other hand, in the Sudan zone the impacts of any improved water availability will be limited as long as nutrients are the main factor limiting primary production. Here, the most important added value of woody plants is the influence on nutrient availability, to be benefited from in agroforestry systems. The situation becomes more complicated where irrigation or fertilisers are being applied. Where water availability is adequate following irrigation, several processes improving nutrient availability (Table 2) could become important, but, unfortunately, most woody species do not survive and produce well when inundated. In addition, their shading effects and the supply of nesting sites for graniverous birds are generally considered unacceptable. On the other hand, where water is the main limiting factor the processes that improve water availability (Table 1) become useful, and synergistic effects can be expected when applying fertilisers. The nutrient improving processes (Table 2) will only remain effective as long as they do not show a negative feedback with increased soil fertility (e.g. N-fixation and internal recycling). Some of these processes, particularly those decreasing nutrient losses become more important under conditions of improved fertilisation. From the synthesis study, it can be concluded that agroforestry has the highest potential (i.e. to improve the potential to use available nutrients and water for primary production purposes in a sustainable way) where production conditions are already most favourable!

Table 3 Comparison of agroforestryparameters from the synthesis study of available knowledge and form field studies in the Southern Sahel zone

Synthesis study Fieldstudies Water balance Stem flow (%) Growing season Nutrient balance Foliage production (g/m2) Enrichment factor Ca Enrichment factor Na Enrichment factor P~ Light interception Isolated tree (%) Cereal yield increase Faidherbia x milletc (kg/ha)

0--20 Extension rare

2-5 No extension

average230 1-3 1-3 l-3x

310, 155b Average 1.3 0.9-1.3 0.8-3.1

Average 47

50, 65b

Maximum 200--400

Maximum300

aln 20--30 cm topsoil under woody plant canopy in comparison to open field; bAcaciaseyal and Sclerocaryabirrea,respectively;CFor Sudan zone.

3.2. Validation by field work The two field studies, aimed to test some of the above conclusions on the added value of woody plants, showed the positive influence of soil organic matter status on recovery rate of fertilisers (Kon6 and Groot, 1996). Soil organic matter quality was improved by annual phosphate inputs on a cowpea crop on different textured soils during successive years, in the 800 mm rainfall zone. The P recovery rate increased from 15% in the first year to 50% in the fourth year. The highest effects were obtained on a sandy soil, particularly in relatively humid years. The perennial grass Andropogon gayanus showed a much higher N recovery rate (70%) than the annual Pennisetum pedicellatum (40%). The N recovery rate of Pennisetum pedicellatum was higher when cultivated on a plot where Andropogon gayanus had previously been grown (45%), as compared to a plot with previous cultivation of annual crops (35%). The root systems of the studied woody species vary strongly, determining their capacity to utilise nutrients and water from extended areas and deep soil layers (i.e. redistribution of nutrients and water towards the topsoil and the woody plant). This influences competition for water and nutrients with herbaceous plants. Combretum ghasalense had the most extensive root

44

system (in width and in depth), Sclerocarya birrea had the longest lateral roots (up to 40 m from the trunk), Acacia seyal had the longest tap root (6 m), while Commiphora africana had a very shallow root system. The study confirmed that roots of woody plants are concentrated in the topsoil, where strong overlap occurs with those of herbaceous plants, and where most water and nutrients are found (Soumar6, 1996). The field results on water and nutrient availability and light interception (Table 3) support the conclusions from the synthesis study (Breman and Kessler, 1995). Based upon the enrichment factor, which expresses the importance of soil improvement by woody plants, soil improvement by woody plants on the Niono ranch appears to be very limited, if not negligible, for C, N and P. The only exceptions were the high P enrichment factors (P-Bray, i.e. available P), up to 3.1, for Acacia senegal and Balanites aegyptiaca, but the average enrichment factor for total P was only 1.1. The combination of drought and heavy exploitation rates (cropping and grazing mainly) on the ranch must be the explanation for the overall limited enrichment effects of woody plants. As a result, perennial grasses have disappeared, there has been massive mortality of woody plants, and a decline of C content in the topsoil from 2.5 to 4.0 g/kg in the seventies (Penning de Vries and Djit6ye, 1982) to 2.0-2.5 at present (Radersma, 1996; Breman and Sissoko, 1997). Annually, as much as 90% of the annual production of the herbaceous layer and of litter from woody plants disappears by grazing and fire, while maximum return to the soil of organic material produced is a condition to make optimal use of the added value of woody plants in agroforestry systems (Breman and Kessler, 1995). De Ridder and Van Keulen (1990) predicted an increase of the cation exchange capacity (CEC) of 4.3 mmol/kg per 1 g/kg increase of soil organic matter. The difference in N recovery between Andropogon gayanus and Pennisetum pedicellatum, as mentioned above, was associated with a difference in topsoil C of 3.9 and 2.4 g/kg, respectively, and a CEC of 1.9 and 1.6 mmol/kg, respectively, which is as predicted (Breman and Sissoko, 1997). For woody plants, the average enrichment factor of 1.3 for C in the topsoil (Table 3) would thus imply a CEC increase of only 1.3 mmol/kg. However, the field dynamics of

organic matter appear to be more complicated. Radersma (1996) observed only a slight increase of the CEC in relation to Sclerocarya birrea and Combretum ghasalense, while one Acacia seyal tree showed an important increase and another a slight decrease, without correlation with the C enrichment factor. Breman and Kessler (1995) predicted the maximum increase of cereal production through increased availability of nutrients by woody plants in the Sudan zone to be 200-400 kg/ha (depending upon nutrient use efficiency), equivalent to about 5 kg N/ha. This maximum can be reached with a canopy cover of about 20%, limited shade effects (using homogeneously distributed trees with a high ratio of trunk height to crown diameter, or no foliage during the growing season), and a return to the soil of all organic material produced by the agroforestry system, except for cereal grain yield. If one or more of these conditions is not met, the increase of cereal production will be less. This probably explains why the maximum cereal yield increase of 300 kg/ha, observed for well managed Faidherbia albida park lands in the Sahel region, is nowadays seldomly reached.

4. Consequences The results of the field validations on the role of woody plants in plant production systems by the two field studies show that, in spite of several processes through which woody plants can improve the availability of nutrients and water, the added values of woody plants are generally even less than expected on the basis of the synthesis study, and thus do not easily represent benefits for farmers, because: the added values are lowest where they are most needed, in resource-poor environments; the competition between woody plants and crops or pastures is strong; conditions are generally not suitable for maintenance of an optimal canopy cover and tree canopy structure, and for maximum return to the soil of organic material produced, mainly because of the high exploitation levels and drought events. The field evidence strengthens the arguments of

45

Breman and Kessler (1995) to limit agroforestry (research) to on-farm testing of options that have potentials to meet specific farmers goals, for specific soil-climate and socio-economic conditions. The following represent the best potentials for effective use of agroforestry. 4.1. In sylvopastoral systems

In case of water limited production, woody cover is good for security aims (to stabilise fodder availability during drought) and to reduce resource losses by erosion. In case of nutrient limited production, a canopy cover of 15-20%, of homogeneously distributed trees, with unpalatable foliage, and a high ratio of trunk height to crown diameter, supports animal production goals. To maximise animal production goals, fodder banks of highly palatable woody species are required, at locations representing optimum growth conditions and high niche differentiation for woody plants (e.g. valleys and depressions with deep and fertile soils). 4.2. In cropping systems

Windbreaks are useful to improve crop establishment on sandy soils in dry areas characterised by nutrient limited primary production (Southern Sahel zone), and where shallow ground-water is available. In the Sudan zone, maximum benefits for crop production are obtained where canopy cover is 15-20%, trees are homogeneously distributed, with a high ratio of trunk height to crown diameter (possibly achieved by pruning) and restricted exploitation. Here, the potential benefits are more important for farmers using fertiliser inputs than those without? 4.3. In relation to economic conditions

There is a positive correlation between the potentials for agroforestry and wood and fruit prices, and a negative one with labour wages. Stimulating agroforestry through specific subsidies is justified on slopes and in the upper course of river basins in view of the crucial role in soil stabilisation by woody plants, and in desert margins for protection

of more productive land against desertification. Such subsidies will be more effective and sustainable than to tackle the symptoms of desertification and of fiver basin degradation, provided that the woody plants are not exploited, but used for soil protection. In view of the improved efficiency and recovery rate of the use of water and nutrients (naturally available and external inputs), there will be a less need for subsidies in the agricultural sector in well designed and managed agroforestry systems.

5. Constraints and opportunities Even a focus on the potential options outlined above will not easily generate participation and enthusiasm by farmers, for two main reasons. (1) Farmers should learn how to manage agroforestry systems in such a way that maximum benefits are obtained in terms of increased water and/or nutrient availability for crops or pastures, instead of maximum use of tree crops only. This implies finding the optimum balance between (i) benefits of the added value by woody plants in terms of water and nutrient availability for crops and pastures, and (ii) increased competition between woody plants and crops or pastures as woody cover increases. Based upon farmers goals of productivity (of crops, pastures or woody plants), security and/or sustainability, careful decisions on the design and management of agroforestry systems should involve four interrelated issues: • • • •

selection of tree species with desirable properties (see list of properties in Section 2); woody plant configurations (densities and orientation); tree management issues (pruning, lopping, controlled harvesting, etc.); selection of appropriate site conditions for specific goals.

It will be difficult to find the fight 'package' as regards these four issues, but the amount of 'trial and error' can be substantially reduced by taking the situations with highest potentials for effective use of agroforestry (as indicated above) as starting points for on-farm trials. Benefits can be further improved through judicious use of fertilisers in well designed

46 agroforestry systems, thus making use of the potentials for improved recovery of nutrients. (2) The organic matter status of soils in most semiarid regions is poor, whereas possibilities to improve the soil organic matter status are limited. Such improvements will require a long 'transition period' during which investments (in terms of labour and inputs mainly) are not matched by increased outputs, due to the slow establishment of a woody plant community, and the rapid turn-over of organic matter under tropical conditions, particularly in the Sahel region (De Ridder and Van Keulen, 1990). Land tenure security is one pre-requisite, and subsidies during this transition period may be another. Using simulation models, Groot et al. (1997) estimated that the process of soil improvement through an increased annual supply of organic matter takes 5-15 years. Woody cover required to maintain C concentrations on cropland at a level of 6 g/kg (considered as a minimum for agricultural use of a soil with 20% clay) was estimated at an average of 34% for the Southern Sahel, provided return to the soil of all organic material produced. This unrealistic high level underlines the limited potentials for agroforestry in this zone, and the need for careful management decisions as indicated above. The required 17% woody cover in the Sudan zone is more realistic to achieve. Two specific opportunities for agroforestry were also identified. (1) In the Sahel region (and most likely other semiarid regions), the disappearance of perennial grasses, the replacement of annual species with long growth cycles by short cycled annuals (Breman and De Ridder, 1991) and the massive mortality of woody plants (Breman and Kessler, 1995) created a reduction of water use by vegetation. Whereas at the start of the growing season during the seventies (Penning de Vries and Djit~ye, 1982) water availability was limited to the first 50 cm of the soil profile, the current situation at the Niono ranch (SoumarE, 1996; Breman and Sissoko, 1997) shows a water profile exceeding 250 cm soil depth. This implies potentials for tree planting and reforestation. (2) In Faidherbia albida agroforestry parklands nitrogen availability for crops is improved by at most 5 kg/ha without fertiliser use, corresponding to about 25% of the average natural annual N availability (Penning de Vries and Djit~,ye, 1982). In addition,

the improved organic matter status will improve the feasibility of using fertiliser inputs. Using woody plants that shed their foliage during the dry season, the yield of associated crops (e.g. vegetables) during the dry season, even when irrigated, can be stimulated by benefiting from this added value of the trees in terms of improved nutrient availability and organic matter status, in a similar way as for Faidherbia albida during the rainy season.

References Production Soudano-SahElienne, 1991. Production SoudanoSahElienne (PSS). Exploitation optimale des ElEments nutritifs en Elevage. Projet de cooperation scientifique. IER, Bamako et AB-DLO, Wageningen, 38 pp. Breman, H. and Kessler, J.-J., 1995. Woody plants in agro-ecosysterns of semi-arid regions (with an emphasis on the Sahelian countries). Advanced Series in Agricultural Sciences, Vol. 23. Springer-Verlag, Berlin, 340 pp. Breman, H. and De Ridder, N., 1991. Manuel sur les p~iturages des pays sahEliens. ACCT-CTA-KARTHALA, Paris, 485 pp. Breman, H., Ketelaars, J.J.M.H. and TraorE, N., 1990. Un remade contre le manque de terre? Bilan des ElEments nutritifs, production primaire et 61evage au Sahel, SEcheresse, 2: 109-117. Breman, H. and Sissoko, K., 1997. L'intensification de L'agriculture SahElienne. KARTHALA, Paris (in press). De Ridder, N. and Van Keulen, H., 1990. Some aspects of the role of organic matter in sustainable intensified arable farming systems in West-African semi-arid tropics. Fert. Res., 26: 299310. Groot, J.J.R., KonE, D. and de WiUigen, P., 1997. L'engrais chimique pour une intensification durable en zone soudanosahElienne. In: H. Breman and K. Sissoko (Editors), L'intensification de L'agriculture SahElienne. KARTHALA, Pads (in press). Hiernaux, P., CissE, M.I., Diarra, L. and de Leeuw, P., 1992. Fluctuations saisonni~re de la feuiaison des arbres et des buissons sahEliens, amelioration de l'Evolution des ressources fourrag~res des parcours sahEliens. CIPEA, Document de travail no. 002/92. ILCA, Addis Ababa. KonE, D. and Groot, J.J.R., 1996. EfficacitE d'utilisation de phosphore et de l'azote par Stylosanthes hamata, Vigna unguiculata, Andropogon gayanus et Pennisetum pedicellatum en zone soudano-sahElienne du Mali. Rapports PSS no. 21, IER, Bamako et AB-DLO, Wageningen, 45 pp. Penning de Vries, F.W.T. and Djit~ye, M.A., 1982. La ProductivitE des Paturages SahEliens. Une l~tude des Sols, des VEgEtations et de L'exploitation de Cette Ressource Naturelle. PUDOC, Wageningen, 525 pp. Radersma, S., 1996. Influence des arbres agroforestiers sur le sol et la strate herbacEe du Sud du Sahel. Rapports PSS no. 26, IER, Bamako et AB-DLO, Wageningen, 106 pp.

47

Soumar6, A., 1996. Utilisation des 616ments nutritifs par deux arbres du Sahel: Acacia seyal et Sclerocarya birrea. Rapport PSS no. 22. IER, Bamako et AB-DLO, Wageningen, 120 pp. T6m6, B., Breman, H. and Sissoko, K., 1996. Intensification agricole au sahel: mythe ou r6alit6. Rapport de synth6se des travaux du Colloque International, Bamako 28 novembre-2 d6cembre 1995. IER, Bamako et AB-DLO, Wageningen, 28 p. Van Keulen, H. and Breman, H., 1990. Agricultural development in the West African Sahelian region: a cure against land hunger? Agric. Ecosyst. Environ., 32:177-197.

Van Keulen, H. and Wolf, J., 1986. Modeling of Agricultural Production: Weather, Soils and Crops. Pudoc, Wageningen, 478 pp. Winterbottom, R. and Hazelwood, P.T., 1987. Agroforestry and sustainable development: making the connection. Ambio, 16: 100-110. Wooning, A., 1992. Les prix du b~tail, de la viande, des produits laitiers et des engrais dans les pays sah~liens. Rapport PSS no. 1. IER, Bamako et AB-DLO, Wageningen, 82 pp. Young, A., 1989. Agroforestry for soil conservation. ICRAF, Nairobi, 276 pp.

This Page Intentionally Left Blank

© 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

49

Chemical crop protection research and development in Europe R. N e u m a n n

*

Research and Development Novartis Crop Protection AG, R-1004.8.15, CH-4002 Basel, Switzerland

Abstract The following five theses are discussed: 1. The majority of the crop protection research and development money is spent by European companies. These efforts are targeted at a global market. The relative importance of the European market strongly varies with the crops and the different pests. 2. Crop protection research and development is a very costly process which will accentuate the concentration process in the crop protection industry. 3. Crop yields will have to increase significantly, if we want to meet the growing demand for food without claiming more land from nature. High yielding varieties and new agrochemicals with further improved characteristics will be important pillars. 4. Risks and benefits must be seriously judged and compared. Overemphasising the risks reduces the benefits. Insufficient crop protection is a waste of resources. 5. Crop protection has tremendously improved in the last 50 years and will continue to do so. Chemical synthesis will remain the most important source for new active ingredients. Natural products will not per se play a dominant role in the near future. Keywords: Crop protection; Research and development; New chemical products; Risk and benefit

1. Introduction

2.

Do we really need research and development in crop protection? Don't we have more than enough herbicides, insecticides, and fungicides? Many people would argue that we have too many. But this is wrong. In the past, we had a steady flow of new active compounds from research to the markets and this must continue. We need more and better products for resistance management, for integrated pest control, and for higher yields. Otherwise, we will not be able to satisfy the needs of a growing population without destroying more rainforests and other important natural habitats. This paper is structured around five theses with the following topics"

3.

1.

Europe as part of a global market

* Corresponding author. Phone +41 (61) 697 34 73; Fax +41 (61) 697 84 57.

4. 5.

The increasing expenses in crop protection R and D The need for increasing yields in crop production Risks and benefits in crop protection New developments in crop protection R and D

2. 1st Thesis The majority of the crop protection research and development money is spent by European chemical companies. These efforts are targeted at a global market. The relative importance of the European market strongly varies with the crops and the different pests. In 1994 the 20 biggest crop protection companies invested around 2.5 billion US$ in research and development (Wood Mackenzie, 1995a). Out of these 20 companies 7 have their headquarters in Europe, 6 are based in the US and 7 in Japan. This could in-

50

dicate a more or less equal importance of the three regions for research and development in crop protection. However, the seven European companies spent 1.4 billion US$, corresponding to 58% of the grand total, the six US based companies spent little over half that amount, namely 756 million US$ or further 31%, and the R and D effort of the seven Japanese companies add up to 282 million US$ i.e. the remaining 11%. In relative terms, the industry spent 9.7% of sales for R and D which shows that these companies invest intensely into the future of crop protection. Today, these R and D efforts must be targeted at the global market. Home markets are too small to justify the ever increasing costs of R and D. The size of the respective world markets is a major factor for the setting of priorities in R and D. Due to the globalization, this is valid for all big companies. Smaller companies, on the other hand, look for allies in order to market their new products globally. Home markets are of little special importance for most European crop protection companies. Specific European needs are taken into consideration as far as it can be justified by the size of the respective markets. In Germany, this has led to a situation where more than 40 pest problems are without a permitted pesticide. Products exist for these problems, but the size of the respective markets can-

not justify the cost of their registration (Meinert, 1991). A few figures shall help to sense the weight of Europe in world-wide crop protection. •

•

•

•

Europe represents a quarter of the global market of about 28 billion US$. North America - which is mainly the US - lies at 30%, and Japan accounts for 17%. After the US and Japan, France is number three in a country ranking, followed by Brazil, and Germany (Wood Mackenzie, 1995b). Together, these five most important countries add up to 60% of the global market. Looking at West Europe on its own" France takes a third, Germany 15%, followed by Italy with 14%, the UK with 10% and Spain with 9%. All other countries use the remaining 20% (Wood Mackenzie, 1995b). Almost half of the pesticides used are herbicides (45% in Europe and 47% world-wide); due to the different climatic conditions, fungicides are more important in Europe (29% and 20%) and insecticides more in the rest of the world (18% and 29%) (Wood Mackenzie, 1995c). Europe's importance in the world-wide herbicide markets varies with the crops. It repre-

Table 1 Major companies" the 75% Club 1980

1985

1990

1994

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1 I. 12. 13. 14. 15. 16. 17. 18.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1 I. 12. 13. 14. 15.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

1. Ciba 2. AgrEvo 3. DuPont 4. Monsanto 5. Zeneca 6. Rh6ne-Poulenc 7. Bayer 8. DowElanco 9. American Home 10. BASF 11. Sandoz

Bayer Ciba-Geigy Monsanto Shell ICI Rh6ne-Poulenc BASF Eli Lilly DuPont H6chst Stauffer Dow Union Carbide Am. Cyanamid FMC Rohm&Haas FBC Kumiai

Bayer Ciba-Geigy Monsanto ICI Shell Rh6ne-Poulenc BASF H fchst Dow DuPont Schering Eli Lilly Am. Cyanamid Stauffer FMC

Ciba-Geigy ICI Bayer Rh6ne-Poulenc DuPont DowElanco Monsanto H6chst BASF Schering Sandoz Am. Cyanamid

51

In 1980, 75% of total world-wide sales were made by 18 companies, in 1985 there were 15, in 1990 only 12. Today 11 companies hold 75% of the market and by the year 2000 it should be expected that there will be fewer than 8 research based companies in the 75% club. With Ciba and Sandoz merging to Novartis it already is down to 10. One of the major driving forces behind this concentration process is the need to keep or obtain 'critical mass' as a basis for successful research and development.

sents 43% in cereals, 30% in fruits and vegetables, 16% in maize, and below 5% in rice and soybeans (Wood Mackenzie, 1995c).

3. 2nd Thesis Crop protection research and development is a very costly process which will accentuate the concentration process in the crop protection industry.

4. 3rd Thesis

A total of 148 new active ingredients were introduced between 1983 and 1993, or about 13.5 per year. From this the R and D expenses for a new active ingredient can be roughly calculated to be in the vicinity of 200 million US$. 20 years ago, this figure would have been close to 80 million, or 2.5 times less, after correcting for the inflation. Modem chemical crop protection started after World-War II. Before that time almost no adequate solutions were available. A large number of compounds have since been discovered. The Pesticide Manual (1994) lists 1285 active ingredients that at some time were used for crop protection, out of which 725 still are in use today. This number will decrease with time. Due to the increasing costs of development, registration and re-registration, more products will be withdrawn from the market than newly introduced. As another consequence, the concentration process in crop protection companies will continue (Table 1).

Wheat

Barley

Maize

Rice

World Germany

Crop yields will have to increase significantly, if we want to meet the growing demand for food without claiming more land from nature. High yielding varieties and new agrochemicais with further improved characteristics will be important pillars. Today, almost nobody contests the need for further increases in the production of food and fibres for a growing and prosperous population. FAO and others agree that we will at least have to double if not triple crop production in the next 50 years. We have not only to keep up with the human population increase, but also to provide the feed for the increasing demand for meat. This is easy to say, but will be hard to achieve. Since we cannot afford to take more land from nature, the only way out is to increase the yields. The last 25 years have seen increases in the world-

I+

55 % }+62%

[ . + 18 %

World Germany

1+41%

World Germany

1+70% 1+35%

World

!.... 1

2

3 yield:

I +53% 4

5

t/ha

Fig. 1. Increase of yield 1969/71 - 1994.

52

wide yields, of 55% for wheat, of 18% for barley, even 70% for maize, and 53% for rice (Fig. 1). As e.g. the European countries have shown, further substantial increases are possible (Oerke et al., 1994). The same publication gives an estimate of the contribution made by world-wide crop protection to the production of the eight principal food and cash crops. Of the theoretically attainable production still about 40% is actually lost due to weeds, animal pests, and diseases. The 60% harvested would further be cut in half without crop protection. This alone clearly shows that today's pesticides and the way they are used must be improved. We cannot afford to loose such a high percentage of the attainable production. The increasing importance of integrated pest management (IPM) and the inevitable spread of resistance are further reasons for the need for new products. There are many definitions of IPM around. Most of them give 'minimizing the use of chemicals' as the main goal. Extreme views equate IPM with pesticidefree farming. The definition which Ciba supports reads: Integrated Pest Management is the farmer's best combination of cultural biological and chemical measures that yield the most cost-effective, environmental sound and socially acceptable insect, disease and weed management for crops in a given situation. High yielding varieties and new agrochemicals with further improved characteristics will be important pillars; certainly among other things like e.g. soil fertility or a sufficient supply of water. The main driving forces for better products are in the areas of safety and performance as discussed in the next thesis.

5. 4th Thesis Risks and benefits must be seriously judged and compared. Overemphasising the risks reduces the benefits. Insufficient crop protection is a waste of resources.

Crop protection products have to be very safe" safe for the consumer, safe for the environment, safe for the farmer and safe for the industry worker.

About 40% of the R&D expenses are spent for the safety evaluation of the products, i.e. about 80 million US$ for each successful new active ingredient. Product safety is very important and has always been the major hurdle to receive the official registration. But on the other hand, the benefits from the crop protection chemicals have to be considered, too. They are now included in the guidelines from the EU. Risks and benefits must be seriously judged and compared. They have to be discussed case by case, separately for each use of a specific product. Region specific facts like climate, soil conditions, availability of other solutions, level of education of the farmers, pest pressure, resistance status, just to name of few, also need to be taken into account. Overemphasising the risks reduces the benefits. Insufficient crop protection is a waste of resources. As has been said before, we must increase production. If we cannot do it through increased yields, we will have to claim more land from nature by reducing the area for forests or wildlife. Now-a-days, this is bad politics everywhere, also in a wealthy Europe. Safer products will come from new chemicals with better intrinsic properties like lower use rates, higher selectivity, and lower persistence. And they will also come from innovations in formulation, packaging, application equipment, and the timing of applications. In the area of packaging, a new development from Ciba by the name of LinkPak could be~cited, which is a closed filling system that uses re-tillable containers. So the farmer has no problem with contamination or the disposal of empty packages. Beside safety, performance is the other major driving force: Better performance for the farmer means lower application rates or fewer applications, means better suitability for system approaches like the integrated pest management (IPM) or resistance management, and means - more generally- better reliability, versatility, and ease of handling. In addition, changes in cropping methods, like no-till cropping or rice that is sown and not transplanted, will generate the need for new products with different properties. Better performance for the farmer translates into cheaper products for the consumer: This is not so important for many of us in Europe, but very important for a lot of people as the hunger in the world is directly related to poverty. Lower costs of food

53

means fewer people starving. This is something to remember for those people who extrapolate from our European surpluses to world-wide agriculture. And, last but not least, better performing products will certainly also help to secure the profitability of the crop protection industry and thus continuation of their research activities.

6. 5th Thesis Crop protection has tremendously improved in the last 50 years and will continue to do so. Chemical synthesis will remain the most important source for new active ingredients. Natural products will not per se play a dominant role in the near future.

Research adapted to the new needs and developments by improving the discovery process. Although serendipity still is the main avenue to new and better products, two major changes have become apparent. The first change is a considerable increase in the number of compounds tested, thereby enhancing the chances to find new and better products by luck, and the other exploits new biological targets by making use of the fast growing knowledge about biochemical and physiological processes in important crop plants and in pests and beneficial organisms. Random screening where a large number of molecules is evaluated in selected biological tests has been the standard method for many years. It certainly was improved over time, but will now be drastically expanded. 20 years ago, block screening tests needed gram quantities of chemicals. Therefore, relatively few compounds only were available. Today many more compounds can be screened as 100 mg is sufficient. The aim is to increase the access to new chemistry further by decreasing the amount of material needed. As a consequence, the focus is now on high-speed synthesis, high throughput screening systems and specialised tests. The standard block screen in Ciba tests in stage 1A about 12,000 compounds a year which come in regular amounts of 100 to 200 mg. A much larger number of structures about 100,000 will be evaluated in a 'high throughput screen' (HTS). Such a new test system has to be automated and work with specialised biological and biochemical mini tests that do not

need more than 1 mg. Promising structures will then have to be synthesised in larger amounts for the standard block screen. Chemical synthesis will remain the most important source for new active ingredients. Robots will help to increase the efficiency. Natural products will become more important, especially those from plants and micro-organisms. However, it does not seem probable that they per se will play a dominant role in the near future. Nevertheless, the structural richness of the natural products make them interesting research objects and a valuable source for new ideas for synthesis chemists. Fludioxonil (Gehmann et al., 1990) and pymetrozinc (Fliickiger et al., 1992) are two examples to illustrate this type of approach: The phenylpyrroles are a successful example of a new class of fungicides closely related to a natural bioactive compound. Pyrrolnitrin is a secondary metabolite produced by Pseudomonas and some Myxobacterales. It served as the lead structure for the phenylpyrroles fenpiclonil and fludioxonil. Pyrrolnitrin has a high activity against various plant pathogenic fungi but has the disadvantage of being very photolabile. The new phenylpyrrols are at least as active as pyrrolnitrin and are significantly more stable in light. From the main areas of improvement, as mentioned above, two points should be highlighted: environmental safety and resistance management. Coating of the seeds results in optimal targeting of this fungicide and, as a consequence, in a very low rate of application per hectare, thereby keeping the environmental burden to a minimum. The new mode of action makes fludioxonil a very valuable tool in resistance management. The second example is pymetrozine which represents a novel class of insect control agents with again a new mode of action. The compound affects the behaviour of homopterous insects and causes them to stop feeding which eventually leads to their death. It has an excellent activity against susceptible and resistant aphids, whiteflies and hoppers. Thanks to its selectivity it is an ideal IPM product and, again, its new mode of action makes it a very valuable tool in resistance management programs. More generally, in the field of insect control, new chemicals like pymetrozine or the Bayer product imidacloprid, with new modes of action have already

54 been or will soon be introduced. They reduce the dependence on the small number of chemical classes that presently dominate the markets, e.g. the organophosphates, carbamates, and pyrethroids. And, among others, they will improve our choices in integrated pest management and in resistance management. The second approach for the discovery of new solutions exploits new biological targets by making use of the fast growing knowledge about biochemical and physiological processes in important crop plants and in pests and beneficial organisms. Two examples are given to illustrate this approach. The first is a new method to fight fungal and microbial diseases. In the course of evolution plants have developed natural defence mechanisms that help to protect them from damage by pathogens. One important self defence mechanism is a phenomenon called Systemic Activated Resistance (SAR). When attacked by fungi, bacteria or viruses, plants react with the development of local necrosis which aim to block further development of the pathogen. In addition, plants produce a signal molecule which activates the defence systems throughout the plant - very similar to an immunisation achieved by vaccination and, thereby, helps the plant to help itself. New biological and biochemical test systems made it possible to specifically search for compounds that could imitate this biological process and, thereby, activate the corresponding resistance mechanism. The project already was successful. A molecule (CGA 245'704) was found that stimulates the plant defence mechanisms and, as a consequence, provides effective and long-lasting protection against a wide range of pathogens without being a fungicide, i.e. without having a direct effect against plant pathogens (Kessmann et al., 1996). The other example of using biological knowledge stands under the title: Attract and Kill. This method combines the attracting power of pheromones with insecticides. Male codling moths are attracted to drops of the female pheromone placed on the leaves. If the drop contains also some insecticide, e.g. cypermethrin, the male insect will die instead of enjoying the expected female companionship. By this method, populations are kept low by eliminating the male insects. In field trials in Egyptian cotton against the

pink bollworm Attract and Kill gave much better results than conventional insecticide sprays (Hofer et al., 1995). Manual application of the droplets had to be done which is certainly more laborious than aerial sprays. On the other hand, the high specificity and the extremely low residues have to be put into the cost and benefit evaluation.

7. Concluding remarks To improve crop protection means to optimise in many different directions. It ~is a difficult task with many contradicting objectives and needs of farmers, the industry and the society. It is the task of industry to discover and develop today those products that the farmer will need in 10 to 20 years from now and that will then be accepted by the local society. Crop protection is of public interest and has to be discussed publicly. Industry has to accept its role in these discussions and has to acquire the relevant capabilities. The case of 'Brent Spar' between Shell and Greenpeace has neatly shown how important it is today, not only to know the right arguments but also to communicate them professionally. The last 50 years brought chemical crop protection to a very high standard. The next years will bring various changes, many will be linked to biotechnology. To improve our crop protection methods will remain a basic need for us and our children. No prejudice should impede the implementation of the best solutions. We have to integrate all our knowledge, methods and products to optimise agriculture, be they chemical, biological or biotechnological. There is no reasonable alternative.

References Fliickiger, C.R., Kristinsson, H., Senn, R., Rindlisbacher, A.S., Buholzer, H. and Voss, G., 1992. PYMETROZINE- A novel agent to control aphids and whiteflies. Brighton Crop Protection Conference- Pests and Diseases l: 43-50. Gehmann, K., Nyfeler, R., Leadbeater, A.J., Nevill, D. and Sozzi, D., 1990. CGA 173'506: A new phenylpyrrole fungicide for broad-spectrum disease control. 1990 Brighton Crop Protection Conference- Pests and Diseases 2: 399-406. Hofer, D. and Angst, M., 1995. Control of Pink Boll Worm in Cotton with SIRENE, a Novel Sprayable Attract and Kill

55

Formulation. Proceedings Beltwide Cotton Conferences, San Antonio, TX: 949-952. Kessmann, H., Oostendorp, M., Ruess, W., Staub, T., Kunz, W. and Ryals, J., 1996. Systemic Activated Resistance- A new Technology for Plant Disease Control. Pesticide Outlook, June 1996: 10-13. Meinert, G., 1991. Minor uses - a difficult proplem. Gesunde Pflanzen, 43: 238-240. Oerke, E-C., Dehne, H-W., Sch6nbeck, F. and Weber, A., 1994. Crop Production and Crop Protection. Elsevier, Amsterdam, 808 pp.

The Pesticide Manual, 1994. C. Tomlin (Editor). British Crop Protection Council, Surrey and The Royal Society of Chemistry, Cambridge, UK, 1341 pp. Wood Mackenzie, 1995a. Agrochernical Service, Update of the Companies Section, 307 p. Wood Mackenzie, 1995b. Agrochemical Service, Update of the Countries Section, 167 p. Wood Mackenzie, 1995c. Agrochemical Service, Update of the Products Section, 198 p.

This Page Intentionally Left Blank

© 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geo'n (Editors)

57

Emissions of CO2, CH4 and N20 from pasture on drained peat soils in the Netherlands C.A. Langeveld a'*, R. Segers a, B.O.M. Dirks a, A. van den Pol-van Dasselaar b, G.L. Velthof c, A. Hensen d aDepartment of Theoretical Production Ecology, Wageningen Agricultural University, P.O. Box 430, 6700 AK Wageningen, The Netherlands bDepartment of Soil Science and Plant Nutrition, Wageningen Agricultural University, P.O. Box 8005, 6700 EC Wageningen, The Netherlands CNMI, Department of Soil Science and Plant Nutrition, Wageningen Agricultural University, P.O. Box 8005, 6700 EC Wageningen, The Netherlands dNetherlands Energy Research Foundation ECN, P.O. Box 1, 1755 ZG Petten, The Netherlands Accepted 8 April 1997

Abstract Soils have an important role in the global budgets of the greenhouse gases carbon dioxide ( C O 2 ) , methane (CH4) and nitrous oxide (N20). In particular, peat soils are considered to exhibit relatively high emission rates. The purpose of this study is to make an integrated assessment of the emissions of CO2, CH4 and N20 from pasture on drained peat soils in the Netherlands (almost 10% of the total land area). The study is based on monitoring studies, described or to be described in more detail elsewhere, at the experimental farm 'R.O.C. Zegveld'. Besides emissions from the soil-plant part of the system, CO2 emission from cattle and their excreta, and emissions of CH4 and N20 from excreta in grazed pastures were included. The annual fluxes of CO2, CH4 and N20 were estimated. The estimated net CO2 emission was (l I + 3) x 103 kg/ha per year, CH4 emissions ranged from --0.3 + 0.1 to 0.1 + 0.1 kg/ha per year and N20 emissions from 14:1:1 to 61 + 4 kg/ha per year. By extrapolation we estimated the contribution of the investigated sources on drained peat pastures to the overall national greenhouse gas emission of 244 x 109 kg CO2 equivalents/year, assuming that (l) the emissions at the monitored site were representative for the Netherlands, and (2) the uncertainties found in the monitoring studies were the sole sources of uncertainty. The percentage contributions to the overall national greenhouse gas emission were estimated at 1.3 _.+0.3% for CO2, 0.0% for CH4 and 0.9 + 0.1% for N20. © 1997 Elsevier Science B.V. Keywords: Carbon dioxide; Enhanced greenhouse effect; Groundwater table; Methane; Nitrous oxide; Pasture; Peat

1. Introduction The enhanced greenhouse effect refers to a rise of the equilibrium temperature at the earth's surface due *Corresponding author. Tel.: +31 317 482140; fax: +31 317 484892; email: [email protected]

to anthropogenic emissions of greenhouse gases into the atmosphere. Carbon dioxide, methane and nitrous oxide are major gases responsible for this enhanced greenhouse effect (Table 1). Soils have an important role in the global budgets of these gases. Changes in tropical land use probably constitute about 10% of the global CO2 sources

Reprinted from the European Journal of Agronomy 7 (1997) 35-42

58

(IPCC, 1995). Wetlands and rice paddies may contribute about 30% to the global gross emission of CH4 (Prather et al., 1995). Probably at least 60% of the global gross N20 emission evolves from soils, in particular wet tropical forest soils and cultivated soils (Prather et al., 1995). The main processes underlying exchange of greenhouse gases between soils and the atmosphere are of a biological and physical nature. CO2 is produced by respiration of various organisms and consumed in assimilation by plants. CH4 is produced in methanogenic breakdown of organic compounds and consumed in methane oxidation, both by micro-organisms. Denitrification and nitrification by microorganisms can produce N20; N20 is also further reduced or consumed during denitrification. The presence of organic material drives respiration, methanogenesis and denitrification. Therefore, peat soils which have a high organic matter content could constitute important sources. Drainage probably affects the balance between emissions of CO2, CH4 and N20, and the overall contribution to the enhanced greenhouse effect. Nyk~inen et al. (1995), for example, found higher CO2 and N20 emissions and a lower CH4 emission at a site drained for agriculture than at a corresponding non-drained 'virgin' site, resulting in a higher overall contribution to the enhanced greenhouse effect at the drained site. Besides drainage the fertilisation and grazing regimes affect the emissions of greenhouse gases from drained peat soils used for dairy farming. In the Netherlands, peat soils cover approximately 0.28 × 10 6 ha (F. de Vries, Winand Staring Centre, Wageningen, pets. commun.), almost 10% of the total land area. Most soils are drained, fertilised and used as pasture land. About 30% of the pastures in the Netherlands is situated on drained peat. Thus far, no estimates based on measurements in the Netherlands have been made of CO2 and CH4 emissions from drained peat pasture soils. Kroeze (1994) estimated N20 emissions from this source, but did not account for differences in groundwater tables. This study aimed at integrating the field monitored data available in the Netherlands on emissions of CO2, CH4 and N20 from drained peat pasture soil and at assessing the importance of emissions from this soil type. We focused on the soil-plant-atmosphere part of the system, but included CO2 emissions from

cattle and their excreta, and CH4 and N20 emissions from excreta in the case of grazing. Methane emission by cattle was omitted, however.

2. Materials and methods

2.1. Site description Emissions of CO2, CH4 and N20 were monitored at the experimental farm 'R.O.C. Zegveld', The Netherlands. The soil is a Terric Histosol (FAO classification). Within the refined Dutch soil classification system (De Bakker and Schelling, 1989) the soil is a 'koop' peat soil. The top layer of about 0.2 m has a relatively high clay content of about 300 g/kg soil (Velthof and Oenema, 1995b) and covers a 6-7-m thick layer of wood sedge peat. With respect to greenhouse gas emissions, we considered the site an adequate representative of the peat pasture soils used for intensive dairy farming in the Netherlands for two reasons. 'Koop' peat soil is the most common of the nine peat soil types found in the Netherlands (Hendriks, 1992), while the management like imposed groundwater tables, fertilisation regime, mowing/ grazing regime, composition and stocking density of the cattle, reflects the common farming practice on these soils well. Since 1969 the soil area at the farm has been divided into two sections, one where a high groundwater table was maintained (high GWT; median at 0.3 m depth), and one with a low groundwater table (low GWT, median at 0.5 m depth). The average carbon content in the 0-0.2 m layer of the high GWT plots was 156 g C/kg dry soil at a C/N ratio of 9.6, the corresponding values at the low GWT plots were 223 g C/kg dry soil and 12.0 (Velthof and Oenema, 1995b). N fertiliser was applied in split applications, as calcium ammonium nitrate (CAN). Applications ranged from 161 to 464 kg N/ha per year (Velthof et al., 1996).

2.2. Monitoring studies Instantaneous CO2 fluxes were measured throughout April, May, June, August, September and October, 1994 (Hensen et al., 1995), by an eddy correlation method having a fetch of approximately 500 m. The resolution of the monitor was 0.1 p.p.m, by volume

59

with a reproducibility of the concentration measurements, i.e. a coefficient of variation for repeated measurements of a standard concentration, of 0.03%. The error in a 30-min averaged flux value was about 10% due to the combination of errors in measurements of the vertical wind velocity and CO2. The set-up of the flux measurements implied that treatments differing in GWT, grazing/mowing regime and N fertilisation regime were spatially integrated, and that CO2 emitted by cattle was largely included in the measurements. Besides the CO2 fluxes air temperature (T) and global radiation (R) were recorded during the monitoring period. Local instantaneous CH4 and N20 fluxes at different treatments were monitored with vented closed flux chambers (Van den Pol-van Dasselaar and Oenema, 1997; Velthof and Oenema, 1995a,b), sampling 0.10.2 m 2 weekly or bi-weekly. The specifications for the CH4 monitoring study were: reproducibility of the concentration measurements 0.08%; lower limit detectable fluxes about 0.03 mg CHn/m2 per day. The specifications for the N20 study were: reproducibility of the concentration measurements 5%; lower limit detectable fluxes about 0.4 mg N20/m 2 per day. 2.3. Estimation of annual emissions The instantaneous CO2 flux measurements were used to estimate the annual net CO2 emission. We estimated the instantaneous total CO2 emission by cattle by evaluating their C balances. The C uptake was calculated using the tabulated energy requirement of the cattle (IKC, 1993), ignoring growth. It should be noted that the calculated CO2 emission by cattle is partly derived from C in feed concentrates (about 30% of the C taken up annually by dairy cows in the Netherlands comes from feed concentrates; E.A. Lantinga, Dept. Theoretical Production Ecology, Wageningen, pers. commun.). The pasture's (upward) respiratory CO2 fluxes (Fc.r) were calculated by subtracting the cattle's CO2 emission from the measured night-time CO2 fluxes. Subsequently, an exponential relationship was fitted between T and Fc,r. The instantaneous respiratory CO2 fluxes from pasture and cattle were subtracted from the measured instantaneous day-time CO2 fluxes to obtain the (downward) assimilatory CO2 fluxes (Fc,a). A hyperbolic relationship was fitted

between R and Fc, a. Diurnal patterns of radiation (R) and temperature (T) during 1994 were reconstructed (Goudriaan and Van Laar, 1994) using daily R and the minimum and maximum T for Wageningen, 50 km east of the site. To estimate the annual net CO2 emission, the two obtained relationships - between T and Fc,r, and between R and Fc,a - were applied to the full year's diurnal pattems of T and R. The uncertainty in the estimated annual net CO2 emission associated with the statistical error in the two obtained relationships was calculated using the principles of Monte Carlo simulation: 1. The relationships between T and the coefficient of variation of Fc,~, and between R and the coefficient of variation of Fc,a, were determined considering classes of 100 successive (T, F~,r) and (R, F~,~) pairs. 2. Drawings were done from the normal distribution to obtain stochastic realisations of F~,r and F~,~for each half hour of the year. We assumed that any drawing applied to both Fc,~and F~,a(i.e. the deviations from their fitted values were correlated). 3. Net CO2 emissions for a year were calculated from 17520 (=365 x 48) realised values of Fc.r and Fc,a. 4. The estimated annual net CO2 emission and its uncertainty were obtained as the stabilised average and standard deviation of repeated calculations of the annual emissions, respectively. The annual CH4 and N20 emissions were obtained by trapezoidal integration of the mean fluxes over time (Van den Pol-van Dasselaar and Oenema, 1997; Velthof et al., 1996). An upper limit for the uncertainty of the annual CH4 emission was estimated by applying an error propagation analysis on the trapezoidal integration method. We assumed that (1) the maximum of 0.2 mg CH4/m2 per day in the typical range of standard deviations found in replicated measurements (Van den Pol-van Dasselaar and Oenema, 1997) was relevant during the whole observation period, and that (2) bi-weekly monitoring took place. For N20, we assumed that the differences between the two observation years were an adequate measure for the variability of the annual emissions. The total emissions from peat pasture soils in the Netherlands were estimated by extrapolating the Zegveld data in a straightforward linear way, discussed more extensively below.

60 Table 1 Characteristics of CO2, CH4 and N20 as greenhouse gases Gas

Pre-industrial atmospheric concentrationa (ppmv)

1992-Atmospheric GWPb (kg CO2 concentrationa equivalents/kg) (ppmv)

Radiativeforcing of emissions 1850presenta'c (Wm 2)

Global sum of sources° (109 kg/ year)

Sum of sources in the Netherlandse (109 kg/year)

CO2 CH4 N20

280 0.700 0.275

355 1.714 0.311

1.5 0.5 0.2

26000 + 4000 550 + 90 25

189 1.0 0.06

1 24.5 320

ppmv, parts per million by volume. aIPCC (1995). bDirect Global Warming Potential in 1995 following addition of 1 kg of each gas to the atmosphere, relative to CO2 for a 100-year time horizon (Albritton et al., 1995). CTotal direct radiative forcing 1850-present: 2.4 W/m2 (IPCC, 1995). dFrom budgets for CO2 (IPCC, 1995), and CH4 and N20 (Prather et al., 1995). eEstimates for 1994 (RIVM, 1995).

3. Results

(54.5 + 0.6) x 103 kg CO2/ha -I per year and a respiratory flux of the pasture of (55.8 + 2.5) x 103 kg CO2/ ha -1 per year. Including the respiration by the cattle

3.1. Carbon dioxide

We estimated the instantaneous total CO2 emission by cattle at 27.5 kg CO2/ha per day (Table 2). The obtained regression equation for the pasture's respiratory CO2 flux (in mg CO2/m2 per s) was Fc,r = 0.33 x 2.20 ~r-20)/10 (n = 1009~ r 2 = 0.12; T is temperature, in °C). The regression equation for the assimilatory CO2 flux (in mg C O y m 2 per s) was Fc,a = - 1 . 1 1 x R/(335 + R)+0.03 (n = 2182; r 2 = 0.67; R is global radiation, in W/m2). Using these equations and applying the described m e t h o d to estimate uncertainties we arrived at a cumulative assimilatory flux of the pasture of

of (10.0 + 1.6) x 103 kg C O J h a per year (Table 2), we arrived at an estimate for the total annual net CO2 emission in 1994 of (11 + 3) × 103 kg CO2/ha per year (Table 3). Since the cattle largely feeds on the grass and subsequently respires it, it was considered to be an integral part of the pasture ecosystem. As a consequence, a relevant estimate for the CO2 emission by the pasture sec could not be given. The open nature of intensive dairy farming ecosystems - in terms of flows of matter and energy makes any system boundary and associated CO2 emission an arbitrary choice. Moreover, annual variations in pasture productivity and respiration

Table 2 Calculation of daily CO2 emissions from cattle at Zegveld using C balances Cattle type

C uptakea (kg C/head per day)

C not excreted as manureb (kg C/ head per day)

Milkc (kg C/ head per day)

CO2 emission (kg C/head per day)

Stockingdensityd (number of heads/ha)

CO2 emission (kg CO2/ha per day)

Female adult Female young (>1 year) Calves (<1 year) Total

6.56 + 0.92 3.2 + 0.4

4.92 + 0.77 2.4 + 0.4

1.0 + 0.1

3.92 + 0.78 2.4 + 0.4

1.48 0.41

21.3 + 4.2 3.6 + 0.6

1.9 + 0.2

1.4 + 0.2

1.4 + 0.2

0.50

2.6 + 0.4 27.5 + 4.3

aUsing IKC (1993) and Van Dijk et al. (1995); assumption: C content dry matter 40%. Uncertainties from estimated uncertainties in underlying dairy index numbers. bSome of the herbage taken up (75 + 5%) is digested (Meijs, 1981). CRegional data (Van Dijk et al., 1995). Assumptions: milk dry matter content 13.2% (IKC, 1993), milk dry matter C content 40%. dZegveld data (Van Houwelingen, R.O.C. Zegveld, pers. commun.).

61

Table 3 Estimates of annual emissions of CO2, CH4 and N20, and their uncertainties, for different treatments at drained peat pasture soils in Zegveld, The Netherlands, and annual contribution to the enhanced greenhouse effect Gas

Treatment

Annual emission Annual contribu(kg/ha year) tion enhanced greenhouse effect (kg CO2 equivalents/ha per year)

CO~ Integrated (11 +3) x 103 CH4b HighGWT, grazed 0.1 + 0.1 High GWT, mown -0.2 + 0.1 Low GWT, grazed -0.3 + 0.1 Low GWT, mown -0.3 + 0.1 N20c High GWT, grazed 23 + 4 High GWT, mown 14 +_ 1 Low GWT, grazed 61 + 4 Low GWT, mown 28 + 3

(11 +3) x 103 2+2 -5 + 2 -7 + 2 -7 + 2 (7 -I- 1) × 103 (4.5:1: 0.3) x 103 (20 + 1) x 103 (9 :t: 1) x 103

GWT, groundwater table. a1994; CO2 emitted from cattle and their excreta included; uncertainties estimated via Monte Carlo simulation. bl994; CI-I4 from cattle not included; uncertainties estimated via error propagation analysis. CAverages + absolute deviations from these averages, for the periods March 1992-March 1993 and March 1993-March 1994 (Velthof et al., 1996).

found during the growing season, 1-3 weeks after fertiliser application. During dry periods in the summer and during the winter, fluxes were low. We also conclude that there was a clear difference between the treatments. Higher fluxes in the grazed treatments may have resulted from the stimulation of N20 production by cattle excreta. Higher N20 fluxes at lower groundwater tables might be explained by the combination of increased availability of C and N compounds, enhanced nitrification and an increased NEO/N 2 end product ratio in denitrification.

4. Discussion

4.1. Comparison of Zegveld with another drained peat pasture site

further complicate the establishment of robust emission figures.

We found substantial emissions of CO2 and N20, while those of CH4 were negligible or slightly negative (Table 3). Using the different Global Warming Potentials for the three gases, we conclude that the contributions of the CO2 and N20 emissions to the enhanced greenhouse effect had the same order of magnitude, while the contribution of CH4 emissions was negligible. Nyk~inen et al. (1995) also carded out a study on the emission of C O 2 , CH4 and N20 from a drained peat

3.2. Methane

Table 4

Measured CH4 emissions were negligible or slightly negative (Table 3). This suggests that relatively aerobic soil conditions suppressed methanogenic processes in favour of methanotrophic processes. Besides, the abundance of other electron acceptors than oxygen, like nitrate, probably suppressed methanogenesis during wet periods in winter.

Ranges of estimated average annual emissions of CO2, E l 4 and N20, and annual contribution to the enhanced greenhouseeffect of a hypothetical fen in the Netherlands, based on literature data Gas

Annual emission (kg/ha per year)

Annual contribution to the enhanced greenhouse effect (kg CO2 equivalents/ha per year)

CO2 a

1.5 x 103 to -0.5 x 103 100 to 160 0.07 to 0.7

- 1.5 x 103 to --0.5 x 103

-

3.3. Nitrous oxide

CH4b N20c Total

2.5 x 103 to 3.9 x 103 0.02 x 103 to 0.2 x 103 1 x 103 to 4 x 103

Compared with N20 emissions from mineral soils (Velthof et al., 1996), the N20 emissions in Table 3 were relatively high. The high availability of carbon (C) and nitrogen (N) compounds used in denitrification, combined with high denitrification capacities and locally low partial oxygen pressures, probably promoted N20 fluxes. In general, high fluxes were

aUsing typical average values on C accumulation: 0.1 x 103 to 0.4 X 103 kg C/ha per year (Gotham, 1991; Korhola et al., 1995; Armentano and Menges, 1986). bAssumption: average flux of 87:1:18 mg CH4/m2 per day during an emission season of 150 days, zero fluxes during the rest of the year (Bartlett and Harriss, 1993). CUsing regression lines of N20 flux with water table of Moore (1996) and assuming an average groundwater table at 0.1 m depth.

62 soil. However, grazing cattle was absent at their site in Finland. They observed a high COz emission of 22 x 103 kg/ha per year. This might be explained by their removal of the green vegetation throughout the monitoring period. The N20 emission of 13 kg/ha per year they found is lower than we found in any treatment. This was probably associated with (i) the lower N fertilisation rate, 80 kg N/ha per year, and (ii) the absence of grazing cattle. Compared with our negligible or slightly negative CH4 emissions, they found a small positive CH4 emission.

4.2. Comparison with wetlands Implementation of policy decisions for conversion of the peat soils in the Netherlands to natural wetlands could result in the generation of fens. This would have implications for agriculture (less area for dairy farming) and environment ('nature restoration', change of emissions). To assess the impact of this type of land use change on greenhouse gas emissions, we estimated the emissions of CO2, CH4 and N20 from a hypothetical fen using the typical ranges of data found in literature (Table 4). Compared with the data of Table 3, the N20 and CO2 emissions for fens were much lower and the CH4 emissions much higher. The total contribution to the enhanced greenhouse effect per hectare of pasture on drained peat in the Netherlands is probably larger than the contribution per hectare of fens. The main causes for the differences are the increased aerobicity, the presence of cattle, and N fertilisation at the drained site. To put the CH4 emission from the hypothetical fen in perspective, we estimated the CH4 emission from cattle at Zegveld. Using the specific emissions by cattle under Dutch dairy farming conditions (grazing regime, constitution of diets) of Van Amstel et al. (1993), we estimated this emission at 200 kg CHa/ha per year, which is of the same order of magnitude as the CH4 emission from the fen.

4.3. Emissions from drained peat pastures compared to the total national emissions To assess the importance of emissions from drained peat pastures in the contribution of the Netherlands to the enhanced greenhouse effect we extra-

polated the data from Zegveld. With this we did a number of assumptions that seemed reasonable, taken into account the methodological marginal notes with the estimation procedure of the emissions, discussed below, and the still limited quantitative knowledge on the dependence of the emissions on environmental factors. Further studies on the possibly limited validity of the assumptions could be used to refine to extrapolation procedure. The assumptions were:

1. In the monitoring studies reliable estimates for the instantaneous, average fluxes from the different treatments at Zegveld were obtained. The method for monitoring CO2 implied that relatively large areas were sampled, resulting in representative measurements for these areas. However, as a result, spatial integration over plots differing in management took place and we were not able to identify effects of N fertilisation or groundwater tables in our data set. In the CH4 and N20 monitoring studies, the sampling was confined to small areas. The estimation of fluxes for larger areas from these samples is methodologically difficult, because of the high spatial variability of the fluxes. 2. The calculated annual emissions estimates were representative, despite the temporal confinement of the observations. All fluxes are known to exhibit temporal variability. The use of regression equations for relating the CO2 fluxes to temperature and radiation to estimate the total annual emission is a relatively reliable method, because of the critical dependence of the relevant biological processes on temperature and radiation. Annual CH4 and N20 emissions were estimated by trapezoidal integration of flux estimates obtained from weekly or bi-weekly measurements during day-time. This method did not account for possible effects of (i) a strongly non-linear temporal behaviour of fluxes at a smaller time scale, and (ii) a diurnal pattern of the fluxes. 3. The high and low GWT plots in Zegveld represent pastures on drained peat in the Netherlands in classes II/II* (69% of the total peat pasture area of 0.28 x 106 ha) and III/III* (20%) of the soil map of the Netherlands (F. de Vries, Winand Staring Centre, Wageningen, pers. commun.; classes

63

described in Van der Sluijs, 1992), respectively. The on average lowest GW"r is between 0.5 and 0.8 m depth for class II/II*, and between 0.8 and 1.2 m depth for class III/III*. By this classification, the high GWT plot at Zegveld was classified as II/II*, and the low GWT plot as III/III*. We neglected differences between the types of peat soil and duration of the drainage. 4. The emission per hectare from the remaining pastures on drained peat (11%) equals the average of the emission per hectare in class II/II* and in class III/III*, using the relative areas as weight factors. Quantitative information on the relation between GWT and emissions is scarce. This assumption was introduced as an obvious possibility. We expect that possible errors due to this assumption were small, because of the relative small area of these soils. 5. The effect of Nfertilisation on other peat pastures is comparable to that of the N fertilisation at Zegveld. Most pastures on drained peat are N fertilised. CAN, which was used at Zegveld, is the most important N fertiliser in the Netherlands (LEI, 1996). The CAN application rates at Zegveld corresponded with recommended total N application rates in the Netherlands amounting to 200-400 kg N/ha per year (Unwin and Vellinga, 1994). In the fanning practice in the Netherlands, a part of the N is applied via cattle slurry. The effect of cattle slurry on greenhouse gas emissions is not clear; unpublished exploring studies suggest lower N20 and higher CH4 emissions than for CAN. 6. Effectively half of the drained peat pasture area is used for grazing, the other half for mowing. We assumed that half of the grass production is directly consumed by grazing cattle during the grazing season of about 180 days. The other half is mown and supplied as silage during the rest of the year. This assumption is reasonable for dairy farms where fresh and conserved grass are the only farm-grown roughage components. Using the data in Table 3, we estimated the respective total emissions from pastures on drained peat soils in the Netherlands at (3.1 + 0.8) x 109 kg CO2/year, (-0.03 + 0.02)x 106 kg CHa/year and (6.9 + 0.5) x 106 kg N20/year. The last figure is in

the lower half of the range of 2.6 x 106-13.2 x 106 kg N20/year that can be derived from Kroeze (1994) for N20 emissions from drained peat pasture soils. By implicitly assuming that high and low GWT plots were equally important, Kroeze (1994) probably slightly overestimated N20 emissions from drained peat pasture soils. The emissions we found corresponded to (3.1 + 0.8)x 109 kg CO2 equivalents/ year for CO2, (-0.7 + 0.5)x 106 kg CO2 equivalents/year for CH4 and (2.2 + 0.2)x 109 kg CO2 equivalents/year for N20. We compared these figures with the estimated total greenhouse gas emissions in the Netherlands. In 1994, the emissions of CO2, CH4 and N20 corresponded with 189 x 109, 25 x 109 and 19x 109 kg CO2 equivalents/year, respectively (RIVM, 1995). Including 11 x 109 kg CO2 equivalents/year for other greenhouse gases (Van Amstel et al., 1994), the estimated overall emission of greenhouse gases in the Netherlands in 1994 was 244 x 109 kg CO 2 equivalents/year. Thus, CO2, CH4 and N20 emitted from pastures on drained peat soils could constitute about 1.3 + 0.3, 0.0 and 0.9 __ 0.1%, respectively, of this overall emission. In these figures 02 emitted from cattle and their excreta was included, while CH4 emitted from cattle was not included. It should be noted that the given uncertainties only reflect the uncertainties associated with the respective monitoring studies. An attempt to quantify the uncertainties associated with the assumptions made for the extrapolation does not seem relevant yet. This because neither sufficient knowledge of the processes underlying emissions, nor sufficient field measurements are available to do so in a reliable way.

Acknowledgements Funding for this study was received from the National Research Programme on Global Air Pollution and Climate Change, and the C.T. de Wit Graduate School Production Ecology. We thank J. Goudriaan, M.H.H. Hermsen, M. Hoosbeek, K. van Houwelingen, L. Klemedtsson, C. Kroeze, E.A. Lantinga, P.A. Leffelaar, H. Nyk/inen, O. Oenema, R. Rabbinge and F. de Vries for various contributions.

64

References Albritton, D.L., Derwent, R.G., Isaksen, I.S.A., Lal, M. and Wuebbles, D.J., 1995. Trace gas radiative forcing indices. In: J.T. Houghton, L.G. Meira Filho, J. Bruce, Hoesung Lee, B.A. Callander, E. Haites and K. Maskell (Editors), Climate Change 1994. Cambridge University Press, Cambridge, pp. 205-231. Armentano, T.V. and Menges, E.S., 1986. Patterns of change in the carbon balance of organic soil - wetlands of the temperate zone. J. Ecol., 74: 755-774. Bartlett, K.B. and Harriss, R.C., 1993. Review and assessment of methane emissions from wetlands. Chemosphere, 26: 261320. De Bakker, H. and Schelling, J., 1989. Systeem van Bodemclassificatie voor Nededand: de Hogere Niveaus. PUDOC, Wageningen, 209 pp. (in Dutch). Gorham, E., 1991. Northern Peatlands: role in the carbon cycle and probable responses to climate warming. Ecol. Appl., 1: 182195. Goudriaan, J. and Van Laar, H.H., 1994. Modelling Potential Crop Growth Processes. Kluwer, Dordrecht, 238 pp. Hendriks, R.F.A., 1992. Afbraak en Mineralisatie van Veen. Rapport 199. DLO-Staring Centrum, Wageningen, 152 pp. (in Dutch). Hensen, A., Kieskamp, W.M., Vermeulen, A.T., Van den Bulk, W.C.M., Bakker, D.F., Beemsterboer, B., M61s, J.J., Veltkamp, A.C. and Wyers, G.P., 1995. Determination of the Relative Importance of Sources and Sinks of Carbon Dioxide. Report ECN-C-95-035. ECN, Petten, 68 pp. IKC, 1993. Handboek voor de Rundveehouderij. IKC, Lelystad, 629 pp. (in Dutch). IPCC (Working Group 1), 1995. Summary for policy makers: Radiative forcing of climate change. In: J.T. Houghton, L.G. Meira Filho, J. Bruce, Hoesung Lee, B.A. Callander, E. Haites and K. Maskell (Editors), Climate Change 1994. Cambridge University Press, Cambridge, pp. 7-34. Korhola, A., Tolonen, K., Turunen, J. and Jungner, H., 1995. Estimating long-term carbon accumulation rates in boreal peatlands by radiocarbon dating. Radiocarbon, 37: 575-584. Kroeze, C., 1994. Nitrous Oxide (N20). Emission Inventory and Options for Control in the Netherlands. RIVM, Bilthoven, 163 PP. LEI (Landbouw-Economisch Instituut), 1996. Landbouwcijfers 1995. LEI-DLO/CBS, The Hague. (in Dutch). Meijs, J.A.C., 1981. Herbage Intake by Grazing Dairy Cows. PhD Thesis. PUDOC, Wageningen, 264 pp.

Moore, T., 1996. Trace gas fluxes from northern peatlands. In: R. Laiho, J. Laine and H. Vasander (Editors), Northern Peatlands in Global Climatic Change. Oy Edita Ab, Helsinki, pp. 115-121. Nyk/inen, H., Aim, J., L~ng, K., Silvola, J. and Martikainen, P.J., 1995. Emissions of Cl-h, N20 and CO2 from a virgin fen and a fen drained for grassland in Finland. J. Biogeogr., 22: 351-357. Prather, M., Derwent, R., Ehhalt, D., Fraser, P., Sanhueza, E. and Zhou, X., 1995. Other trace gases and atmospheric chemistry. In: J.T. Houghton, L.G. Meira Filho, J. Bruce, Hoesung Lee, B.A. Callander, E. Haites and K. Maskell (Editors), Climate Change 1994. Cambridge University Press, Cambridge, pp. 73-126. RIVM, 1995. Achtergronden bij Milieubalans 95. RIVM, Bilthoven. (in Dutch). Unwin, R.J. and Vellinga, Th.J., 1994. Fertilizer recommendations for intensively managed grassland. In: L. 't Mannetje and J. Frame (Editors), Grasslands and Society. Proceedings of the 15th General Meeting of the European Grassland Federation. Wageningen Pers, Wageningen, pp. 590-602. Van Amstel, A.R., Albers, R.A.W., Kroeze, C., Matthijsen, A.J.C.M., Olivier, J.G.J. and Spakman, J., 1994. Greenhouse Gas Emissions in the Netherlands 1990, 1991, 1992 and Projections for 1990-2010. RIVM, Bilthoven, 93 pp. Van Amstel, A.R., Swart, R.J., Krol, M.S., Beck, J.P., Bouwman, A.F. and Van der Hoek, K.W., 1993. Methane - The other Greenhouse Gas. RIVM, Bilthoven, 108 pp. Van den Pol-van Dasselaar, A. and Oenema, O., 1997. Effects of grassland management on the emission of methane from intensively managed grasslands on peat soil. Plant Soil. (in press). Van der Sluijs, P., 1992. Hoofdstuk 11: Grondwatertrappen. In: W.P. Locher and H. de Bakker (Editors), Bodemkunde van Nederland. Deel 1. Algemene Bodemkunde. 2nd edn. Malmberg, Den Bosch, pp. 167-180. (in Dutch). Van Dijk, J.P.M., Douma, B.E. and Van Vliet, A.L.J., 1995. Bedrijfsuitkomsten in de Landbouw (BUL). Boekjaren 1990/91 t/m 1993/1994. Landbouw-Economisch Instituut, The Haag. (in Dutch). Velthof, G.L. and Oenema, O., 1995a. Nitrous oxide fluxes from grassland in the Netherlands: I. Statistical analysis of flux-chamber measurements. Eur. J. Soil Sci., 46: 533-540. Velthof, G.L. and Oenema, O., 1995b. Nitrous oxide fluxes from grassland in the Netherlands: II. Effects of soil type, nitrogen fertilizer application and grazing. Eur. J. Soil Sci., 46: 541549. Velthof, G.L., Brader, A.B. and Oenema, O., 1996. Seasonal variations in nitrous oxide losses from managed grasslands in the Netherlands. Plant Soil, 181: 263-274.

Section 3 CROP PHYSIOLOGY AND IDEOTYPING Effects of CO2 and temperature on growth and yield of crops of winter wheat over four seasons G.R. Batts, J.I.L. Morison, R.H. Ellis, P. Hadley and T.R. Wheeler ........................................................... 67 Reprinted from the European Journal of Agronomy 7 (1997) 43-52 Use of in-field measurements of green leaf area and incident radiation to estimate the effects of yellow rust epidemics on the yield of winter wheat R.J. Bryson, N.D. Paveley, W.S. Clark, R. Sylvester-Bradleyand R.K. Scott .................................................... 77 Reprinted from the European Journal of Agronomy 7 (1997) 53-62 Simulating light regime and intercrop yields in coconut based farming systems J. Dauzat and M.N. Eroy ................................................................................................................................. 87 Reprinted from the European Journal of Agronomy 7 (1997) 63-74 Improving wheat simulation capabilities in Australia from a cropping systems perspective: water and nitrogen effects on spring wheat in a semi-arid environment H. Meinke, G.L. Hammer, H. van Keulen, R. Rabbinge and B.A. Keating ..................................................... 99 Reprinted from the European Journal of Agronomy 7 (1997) 75-88 Comparison of CropSyst performance for water management in southwestern France using submodels of different levels of complexity C O. Stockle, M. Cabelguenne and P. Debaecke ......................................................................................... 113 Reprinted from the European Journal of Agronomy 7 (1997) 89-98 Root growth of three onion cultivars A.D. Bosch Serra, M. Bonet Torrens, F. Domingo Oliv~ and M.A. Melines Pages ...................................... 123 Interspecific variability of plant water status and leaf morphogenesis in temperate forage grasses under summer water deficit J.-L. Durand, F. Gastal, S. Etchebest, A.-C. Bonnet and M. Ghesqui&e .................................................... 135 Reprinted from the European Journal of Agronomy 7 (1997) 99-107 Evaluation of sunflower (Helianthus annuus, L.) genotypes differing in early vigour using a simulation model F. Ag~iera, F.J. Villalobos and F. Orgaz...................................................................................................... 145 Reprinted from the European Journal of Agronomy 7 (1997) 109-118 Options of breeding for greater maize yields in the tropics A. Elings, J.W. White and G.O. Edmeades .................................................................................................. 155 Reprinted from the European Journal of Agronomy 7 (1997) 119-132

This Page Intentionally Left Blank

1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

67

Effects of CO2 and temperature on growth and yield of crops of winter wheat over four seasons G.R. Batts a'b'c, J.I.L. Morison d'*, R.H. Ellis b, P. Hadley c, T.R. Wheeler b aDepartment of Meteorology, University of Reading, PO Box 239, Reading, RG6 6AU, UK bDepartment of Agriculture, University of Reading, PO Box 236, Reading, RG6 6AT, UK CDepartment of Horticulture and Landscape, University of Reading, PO Box 221, Reading, RG6 6AS, UK dDepartment of Biological Sciences, John Tabor Laboratories, University of Essex, Wivenhoe Park, Colchester, C04 3SQ, UK

Accepted 1 July 1997

Abstract Crops of winter wheat (Triticum aestivum L. cv. Hereward) were grown in the field in four consecutive seasons from 1991/ 1992 to 1994/1995 at Reading, UK, within polyethylene-covered tunnels along which a temperature gradient was superimposed on the ambient temperature variation at normal atmospheric (ca. 370) or an increased [CO2] (ca. 700 #mol CO2 mol-t air), producing many environments from one sowing date in each season at one location. Mean seasonal temperatures varied by up to 4°C along the temperature gradient. Increased [CO2] had no effect on crop duration, or on the rate of reproductive development, which had the same temperature sensitivity across all years. A 2°C warming, on the 4-year ambient mean temperature (10°C), reduced crop duration by 42 days (from 254), and reduced the reproductive phase by 16 days (from 130). Crop biomass generally declined with increase in mean temperature, and was greater at increased [CO2], with the effect of increased [CO2] varying with temperature and between years (6-34% range in relative stimulation by increased [CO2]). Grain yield was substantially reduced by warmer temperatures, and increased by doubling [CO2], but the effect varied greatly between years and with temperature (7-168% range). There were both positive and negative interactions of temperature and increased [CO2] on biomass and grain yield. In all 4 years, the increase in grain yield from doubling [CO2] was negated by an increase in mean seasonal temperature of only 1.0-2.0°C. Year-to-year variation in the responses of biomass and grain yield to [CO2] and temperature resulted from differences in environmental conditions, influencing biomass partitioning and altering the role of different yield components. © 1997 Elsevier Science B.V. Keywords: Wheat; Growth; Yield; Interannual variation; CO2; Temperature

I. Introduction The potential impacts of an increasing atmospheric CO2 concentration ([CO2]) and warmer global temperatures on crop production are substantial (Watson et al., 1996). Increased [CO2] is expected to increase * Corresponding author. Fax: +44 1206 873416; e-mail: [email protected]

primary biomass production in temperate crops through a stimulation of net photosynthetic rate and/ or a decrease in water use, whereas global warming per se reduces crop duration in determinate crops such as wheat (e.g. Slafer and Rawson, 1994; Batts et al., 1996; Morison, 1996) resulting in lower biomass production and reduced grain yields (Mitchell et al., 1993, 1995; Wheeler et al., 1996a; Batts et al., 1998), at least where only one crop is grown per

Reprinted from the European Journal of Agronomy 7 (1997) 43-52

68

year. Therefore, any increase in yield due to increased [CO2] may be partly or wholly offset by warmer conditions. However, many papers and several observations suggest that the stimulating effect of increased [CO2] on biomass accumulation is larger at higher temperatures (see Rawson, 1992, 1995), so that the final outcome on yield of these two key aspects of global environmental change, temperature and CO2 needs clarification and quantification. Changes in yield of small grain cereals such as wheat can occur through changes in tiller numbers per area and numbers of grain per ear, or through changes in grain size. Wheat plants grown in an increased [CO2] have frequently shown more tillers (e.g., Gifford and Morison, 1993; Mitchell et al., 1993; Rawson, 1995; Batts et al., 1996, 1998) with more grains per unit area from more ears (e.g., Rawson, 1995; Batts et al., 1998; see also review by Lawlor and Mitchell, 1991). Increased temperature can result in reduced smaller grains, fewer grains per ear, and fewer ears m -2 (e.g., Mitchell et al., 1993,

1995; Rawson, 1995), in addition to any effects of high temperatures during particular critical periods on grain fertility (e.g., Mitchell et al., 1993). Clearly, the potential interactions between temperature and [CO2] on wheat growth and yield are complex, and this paper reports a 4-year investigation of these interactions for winter wheat crops (cv. Hereward) grown in our field-based temperature gradient facility (Hadley et al., 1995; Batts et al., 1996). Our previous papers have reported the detailed results of individual years, but the aim here is to compare the responses to temperature and [CO2] between years, and to highlight the variation in responses observed.

2. M a t e r i a l s a n d m e t h o d s

Four field experiments were conducted at University of Reading School of Plant Sciences field unit (51o27 ' N, 0056 ' W). Certified seed of winter wheat cv. Hereward was sown in rows spaced 0.12 m apart

Table 1 Crop husbandry of winter wheat cv. Hereward grown in 4 years in double-wall temperature-gradient tunnels

Sowing date Mean establishment (plants m-2) Soil analysisa Previous crop Pre-emergence herbicide Fungicide fumigantsb Fertilisers (kg ha-I) Seedbed Top dressing 1 Top dressing 2 Top dressing 3 Pesticides Soil applications

Fumigantsb

1991/1992

1992/1993

1993/1994

1994/1995

8 Jan 1992 278

24 Nov 1992 298

8 Nov 1993 296

30 Nov 1994 250

4; 3; pH 6.5 Grass ley Tribunil at 2.25 kg ha-I, 9 Jan Fungaflor

2; 2; pH 6.7 Winter wheat Tribunil, 7 Dec

3; 2; pH 6.8 Maize Tribunil, 12 Nov

3; 2; pH 6.9 Winter wheat Tribunil, 5 Dec

Fungaflor

Fungaflor

Fungaflor

30 N lO0 N double ridge; 31 Mar 40 N late stem extension; 7 May

25 N; 70 P2Os; 110 K20 50 N pre-double ridge; 16 Feb 50 N pre-terminal spikelet; 6 Apr 45 N late stem extension; 4 May

30 N; 70 P205; 110 K20 50 N double ridge; l Mar 50 N terminal spikelet; 24 Mar 40 N mainstem flag leaf; I l May

30 N; 70 P2 05; 110 K20 50 N double ridge; 17 Mar 50 N pre-terminal spikelet; 27 Mar 40 N mainstem flag leaf; l0 May

Gamma-Col,9 Nov

Gamma-Col,8 Nov

Gamma-Col,1 Dec

Fumite 7000

Birlane 24 at 4.2 1 ha-I" 12 Nov Fumite 7000

Fumite 7000

-

Gamma-Col at 1.4 1 hal, 6 Jan

Fumite 7000

ap and K values, respectively, expressed as ADAS scil indices. bUsed following canopy closure until maturity.

69

on 8 January and 24 November 1992, 8 November 1993 and 30 November 1994 into a sandy loam soil to provide uniform stands of 250-300 plants m -2 (Table 1). Four 20 m by 1.75 m areas were sown in each experiment, and a double-wall polyethylenecovered (replaced at each sowing) temperature gradient tunnel was constructed over each crop area (see Batts et al., 1996, 1998). One pair of tunnels was controlled to a target concentration of ca. 360 ttmol CO2 mol -~ air ('normal'; with seasonal mean daily concentrations of 390, 380, 365 and 371 for years 1-4, respectively) and the other pair to a target of 700 /zmol CO2 mol -I air ('increased'; seasonal means of 713, 684, 698 and 691 for years 1-4). The precision of control of the high [CO2] treatment gave concentrations in the tunnel between 630 and 770 /xmol mol -~ for 60%, 86%, 78% and 94% of the duration of the experiment in years 1-4, respectively. The temperature gradient along each tunnel was imposed, and enrichment with CO2 began soon after sowing and maintained until harvest maturity. The temperature gradient facility was relocated ca. 40 m away for the experiments in 1993/1994 and 1994/1995. The crop along each tunnel was divided into five (1991/1992) or six (1992/1993, 1993/1994 and 1994/1995) plots, thereby providing plots grown at different temperatures. Vertical mixing fans between the plot increased the coupling of plants to the air, although these were only installed late in the first experiment. Standard agronomic practices (Table 1) were followed in all 4 years and crops were irrigated to maintain the soil at near field capacity through a network of porous pipe

placed on the soil surface at intervals of 0.32 m. In each year, crops were also grown in ambient field plots near the tunnels. At crop maturity, a 1 m 2 (1991/1992), 0.18 m 2 (1992/1993), 0.19 m 2 (1993/1994) or 0.36 m 2 (1994/ 1995) area was separately harvested from two locations within each plot. Above-ground biomass only was harvested. The number of ears were counted, then separated into straw and seed fractions. The grains were counted and mass recorded. The moisture content of a 5 g sample of grain, and the dry mass of the straw were determined after drying at 80°C for at least 72 h. In summary, these experiments provided a total of 20 (1991/1992) or 24 (1992/1993 to 1994/1995) plots comprising a combination of two [CO2], five or six temperature regimes and two replicates. The effects of mean temperature (between sowing and maturity) and [CO2] on the dependent variables were examined using comparison of regressions in order to determine the significance of differences between the CO2 treatments and whether or not there was any interaction between the effects of [CO2] and temperature.

3. Results 3.1. Growth conditions The mean temperature from sowing to harvest maturity of the ambient plots outside the tunnels ranged from 11.0°C in the late sown 1991/1992 season to

Table 2 Mean temperatures (°C) in ambient plots and in plots at the warm and cool ends of temperature gradient tunnels during different stages of development (SW, sowing; DRS, double ridges formation; MAT, harvest maturity)

Increased Normal Ambient Seasonal ambient mean

SW-DRS DRS-MAT SW-DRS DRS-MAT SW-DRS DRS-MAT

1991/ 1992

1992/1993

1993• 1994

1994• 1995

8.3-7.7 19.5-13.1 7.9-7.8 18.1-15.0 6.5 14.7

6.5-6.3 14.7-11.6 6.6--6.4 14.2-12.4 6.5 14.2

7.0-6. I 13.9-11.0 6.5-5.8 14.2-11.2 5.1 13.2

8.3--6.4 15.4-12.4 8.0-6.4 15.0-12.4 6.1 14.2

I 1.0

9.7

9.2

10.4

Averages of pairs of tunnels maintained at normal (ca. 376 #mol mol-I) or increased (ca. 697/zmol mol-I) CO2 concentration during field experiments with winter wheat in 4 years.

70

9.2°C in the earlier sown and colder 1993/1994 season (Table 2). The temperature gradient in the tunnels varied diurnally and during the season, being typically small in the winter because of low insolation, and the requirement of the air conditioning equipment to deice, and increased markedly from spring onwards. During the whole vegetative development period (from sowing to double ridge formation) the mean daily difference between end plots within the tunnels in each season was <0.5°C except in the last year, when improvements in the system gave larger differences (1.4-1.9°C, Table 2). During the reproductive phase the mean daily difference was typically 3-4°C (Table 2). It should be noted that the effectiveness of the temperature gradient system is larger than suggested by the figures in Table 2 because the temperatures shown are those calculated over a particular phase, not a particular calendar time. As there is a general rising temperature trend during the season, plants developing more slowly tend to 'catch up' in mean temperatures over a particular phase. The 4-year mean [CO2] were 376 and 697 #mol CO2 mol-! in the 'normal' and 'increased' treatments. Mean insolation was similar during reproductive development, with a 4-year mean of 16.6 MJ m -2 per day (Table 3), but varied substantially between years during vegetative development from 2.9 (1993/1994) to 7.4 MJ m -2 per day (1994/1995). A seasonal mean of ca. 65% of this insolation was incident on the crop canopies within the tunnel growing area; although the autumn-spring average was closer to ca. 75%. While there are no spectral changes in the tunnel, except for reduction of UV transmission, the solar radiation inside the tunnels is much more diffuse (see Hadley et al., 1995 for further details of the temperature gradient system).

Table 3 Mean solar radiation flux (MJ m -2 d-j) incident on ambient plots during development between emergence (EM) and double ridges formation (DRS) and between DRS and harvest maturity (MAT) during field experiments with winter wheat in 4 ears

EM-DRS Duration DRS-MAT Duration

1991/1992

1992/1993

1993/1994

1994/1995

5.9 72 16.0 77

3.5 110 15.3 126

2.9 108 16.6 136

7.4 105 18.3 128

Approximate duration of periods (days) are also shown.

320 a

300-= :3 0

•~

280 -

240 -

E

220 -

¢: 0

200

-

180

-

= "1o

o o

260 -

160

8

~

0.012

I

I

9

10

,, .

I

I

I

I

11

12

13

14

.,

c o

15

b ......-'"" "''"'"

0.010

..-" ,.

"10 A • ,c--x .z

.......'"

0.008

"0 O.

""O"

.-'"" "

0..'""" ....'"

0.0O6

0.004

I

10

'

I

12

'

I

14

' "

I

16

'

I

18

'"v

20

Temperature (oc) Fig. 1. Relations between (a) crop duration from sowing to maturity (Yl), and (b) the rate of reproductive development (Y2) and mean temperature (t) during each phase for winter wheat crops cv. Hereward grown at increased (solid symbols) or normal (open symbols) [CO2] during 1991/1992 (i--1), 1992/1993 (A), 1993/1994 (O), and 1994/1995 (O). Results for crops grown in field plots in ambient conditions outside the tunnels are shown for comparison (shaded symbols). The fitted line in (a) is: 1/yl = 3.931 x 10-4t (SE = 1.223 x l0 -6) (re = 0.918, 96 DF) in (b) is: Y2 = 5.492 x 10-4t (SE = 3.00 x l0 -6) (r E = 0.889, 96 DF), with 95% confidence intervals for the prediction (dashed lines).

3.2. Crop development The temperature differences imposed between the warmest and coolest plots within the tunnels altered the durations of crops by as much as 6 weeks over the 4 years. A single relation between the crop duration (days from sowing to maturity) and temperature described all the data for the four seasons;(P < 0.001; Fig. l a), indicating a base temperature of 0°C and a thermal time requirement of 2544°C.d (SE _ 8°C.d). Similarly, the rate of progress (the reciprocal of duration) through the reproductive development phase (double ridges formation to harvest maturity) was positively and linearly related to temperature (P < 0.001; Fig. lb) with a base temperature of 0°C and a

71

1 55 0000

.....

1992/1993 season which had a limited range of temperature (P > 0.1; Fig. 2b). The stimulating effect of increased [CO2] was significant in all 4 years (all P < 0.001). There was a larger effect of increased [CO21 at warmer temperatures in 1991/1992 (P < 0.05; Fig. 2a); no interaction in either 1992/1993 or 1993/1994 (Fig. 2b,c); and much larger increases of biomass with increased [CO2] at cooler than warmer temperatures in 1994/1995 (P < 0.05; Fig. 2d). Therefore, the range in the relative effect of increased [CO2] at different temperatures on biomass varied between years (Table 4).

e

000

50O 0

~

4000

3000

0

~

1992/93

b"

•

4000

~

1500

"

. . . . .

1000

........

1993/94

C

E>, 1 5 0 0 "0

=

~

30oo

2000

3.4. Grain yield

l ooo

o~=

500

1000 0

~

~

i

4000

~

~9a4~5

3000

,

0

*

I

h

1000

,

2000

500

1000 0

"t

1500

d

!

8

! -

10

i

I

I

12

14

16 Temperature

0

! 8

10

i - i 12

14

16

(°C)

Fig. 2. Relations between biomass (left column) and grain dry mass (right column) and the mean temperature from sowing to harvest maturity in 1991/1992 (a,e), 1992/1993 (b,f), 1993/1994 (c,g) and 1994/1995 (d,h) for crops grown at increased (solid symbols) or normal [CO2] (open symbols). All regressions are significant (P < 0.05), and where separate regressions are shown for the two CO2 treatments they are significantly different (P < 0.05).

thermal time requirement of 1821 °C.d (SE + 11 °C.d). There was no significant effect of [CO2] treatment on either of these relations. For both phases of development, the durations of crops grown outside the tunnels were comparable to those grown inside, except in 1991/1992 (Fig. la,b), which is probably because of the poorer coupling of the plants to the air stream in that year prior to the installation of the vertical mixing fans.

3.3. Crop biomass Crop above-ground biomass at maturity declined linearly with increasing seasonal mean temperature (P < 0.001; Fig. 2a,c,d), with the exception of the

In 1991/1992, grain dry mass per unit area declined as a curvilinear function of temperature (P < 0.001; Fig. 2e). Increased [CO2] increased grain yield (P < 0.001) with more effect at warmer temperatures. In all other years, grain yield was a negative linear function of mean temperature (1992/1993 and 1994/1995 P < 0.001, 1993/1994 P < 0.01; Fig. 2fh). In 1992/1993, increased [CO2] increased grain yield (all P < 0.001) with no interaction with temperature (P > 0.05; Fig. 2f). In 1993/1994 and again in 1994/1995, however, this interaction was significant (both P < 0.05); with grain yield less sensitive to temperature at normal [CO2] (Fig. 2g,h), hence, the relative effect of increased [CO2] was greater at cooler than at warmer temperatures, with increases in grain yield of up to 3.0 and 4.5 t ha -~ in the coolest plots, in the respective years. In the last 3 years the grain yield at 10°C varied by only 103 g m -2 in normal [CO2], but by 554 g m -2 at increased [CO2] so that the relative effect of increased [CO2] ranged from 27 to 89% in the three different years (see Table 4). In order to understand the differing patterns of grain yield response to temperature and [CO2] between years the relations between grain yield and biomass or yield components were examined (Table 5). As Fig. 2 indicates, the pattern of grain yield variation with temperature and CO2 was often similar to that of biomass so that biomass changes accounted for a large part of the effect of temperature and [CO2] on grain yield (Table 5). However, harvest index was not constant, showing a large decline with increasing temperature in the first 2 years and showing increases with increased [CO2] in all except the first year,

72

Table 4 Response of above ground biomass and yield to increased CO2 concentration (approximately a doubling) in winter wheat crops grown in temperature gradient tunnels in the field in four seasons Year

1991/1992 1992/1993 1993/1994 1994/1995

Range

Grain yield at 10°C

Biomass (%)

Grain yield (%)

Normal[CO2]

Increased[CO2]

Grainyield (%)

Warming(°C)

6-31 34 8-17 33-17

7-44 72-168 46-7 58-31

b 683 580 623

b 1292 738 994

b 89 27 60

1.0 1.8 2.0 1.2

The range shown is the range in relative effect of increased CO2 between the coolest and warmest plots on biomass and yield. The absolute and relative effect of increased CO2 on yield (g m-2) at 10°C are also showna together with the warming above the ambient seasonal mean which negates the increase of grain yield by increased CO2. a4-year ambient mean temperature. bprediction beyond 1991/1992 temperature range inadvisable. with particularly large effects in the second year (Fig. 3). The yield variation across temperature and CO2 treatment attributable to each of the three primary components (the n u m b e r of ears m -a, number of grains per ear, and individual grain mass) differed substantially between years, and in several cases the increased [CO2] was not a significant factor. This is notable in the contribution of variation in ear number m -2 to variation in grain yield, which was negligible in the second year and very high in the last, but with no effect of increased [CO2] in either case. In 1992/1993, small contributions of variation in ear number m -2 to variation in grain yield were linked to large contributions from numbers of grain per ear, and in 1994/1995 the opposite occurred. In contrast in years 1 and 3, there was a large [CO2] effect on ear number m -2. Individual grain mass was particularly important in determining yield in the warm 1991/1992 season, where total grain mass declined drastically with

increasing temperature (Fig. 2e) and individual grain weight declined from ca. 35 to 15 mg grain -1 over the temperature range (see also Wheeler et al., 1996a;Wheeler et al., 1996b). Similarly in 1994/ 1995 cooler plots had markedly heavier total (Fig. 2h), and individual grain mass, which declined from ca. 33 to 25 mg grain -1 over the 3°C temperature range at normal [CO2] and from 46 to 25 mg grain -I at increased [CO2] (see Batts et al., 1998). The combined effect of grain number per ear and ear number per area resulted in a high correlation of grain yield with number of grains per area in all years.

4. Discussion While some studies have shown effects of increased [CO2] on apical development in wheat and other cereals during particular phases of development (e.g.,

Table 5 Percentage variation accounteda for by linear regression analysis of grain yield (g m-2) on biomass or components of grain yield for all treatments in 4 years Year

Biomass(g m-2)

Ears (m-2)

Grains (ear-~)

Grain mass (mg grain-I)

Grains (m-2)

1991/92 1992/93 1993/94 1994/95

45.1 66.6 (77.3) 49.4 (68.5) 86.1 (87.4)

1.4 (65.6) 6.2 26.6 (69.8) 81.0

79.4 (82.5) 62.4 (71.5) 3.7 (46.4) 0.0 (15.8)

81.9 8.6 (66.0) 1.4 (48.5) 89.5 (91.0)

86.7 80.5 (82.8) 63.4 (78.4) 77.4

Values in parentheses are the correlations when [CO2] is included as a factor in the regression. aPercentage variation accounted for = [(Total MS - Residual MS)/Total MS] × 100. Degrees of freedom for 1991/1992 to 1994/1995 were 38, 42, 46 and 46, respectively. Single figures are presented where correlation values are not significantly different (P > 0.05) from those when [CO2] was included as a factor.

73

Marc and Gifford, 1984; Batts et al., 1996), these are generally small, and were not detectable here over the whole duration from sowing to maturity or over the reproductive period alone (Fig. l a,b). Similar lack of developmental response to increased [CO2] has also been shown in winter wheat by Slafer and Rawson (1994) and Mitchell et al. (1995), (1996). The relations in Fig. 1 can be used to calculate that the effects of a 2 and 4°C warming (in the range of current climate change scenarios for Europe over the next 50100 years) added to the 10°C 4-year ambient mean would result in reductions of 42 and 72 days, respectively, for total duration, and added to the 4-year mean for the reproductive phase (ca. 14°C) would result in reductions of 16 and 29 days, respectively. However, using the AFRCWheat development model, Butterfield and Morison (1992) predicted that a mean warming of 2 and 4°C would reduce the total duration of a winter crop by 20 and 35 days in the southern UK, and the reproductive phase would be reduced by 8 and 20 days, respectively. Part of the discrepancy could be because the simulation used a model parameterised for a different, slower developing cultivar (Hustler) and for a much earlier sowing date, which combined to reduce the mean seasonal temperature in the control simulation to 9°C, and give considerably longer growing seasons. Another reason could be that the seasonal warming in the model was throughout the season, where temperature effects particularly during early development can be affected by photoperiod and vernalisation responses, whereas the characteristic of the temperature gradient tunnels is that the warming is more pronounced in the spring and summer, than in the winter. Therefore, assessing the temperature treatment by the mean season-long value as used in Fig. 1 could be overemphasising the temperature sensitivity. However, until we have more firm information on the likely seasonal pattern of temperature change expressing crop durations as a function of seasonal mean temperature seems the simplest approach. The major objective of the experimental approach developed for these studies was to detect and quantify any interaction between the effects of [CO2] and temperature on grain yield. The negative effect of temperature and the positive effect of [CO2] on grain yield resulted in interactive effects in three out of the four years (Fig. 2e,g,h). The relative effects of increased [CO2] on biomass and grain yield of these crops varied

greatly within and between years (Table 4). For example, there were large differences in the relative effect of increased [CO2] at 10°C (Table 4), indicating that the response of grain yield to [CO2] was influenced by environmental conditions, of which the main variables were probably the temperature and the insolation as the crops were maintained well-watered. While the sowing date in the first year was late, due to technical problems, which may have influenced the results, the other years had similar sowing dates yet showed differences in yield response. For example, the relative effect of increased [CO2] on grain yield (at 10°C) was 89% in 1992/1993 but only 27% in 1993/1994 despite similar mean whole season, vegetative and reproductive phase temperatures (Table 2). However, changes in the exact seasonal time course of temperature and radiation can have a more substantial influence on mean yield and yield stability than changes in the mean conditions (e.g., Nonhebel, 1994). Clearly, each year has a unique seasonal time course of temperature and insolation, and plants in the -

13

40-

1992/93

1991/92

c

3020~X 10-

&

"0

•~- 0

m

!

W

1993•94

~ 60-

b

1994195

d

e-

50-

o•

40-

000° ; ' ~ ; -

-

30208

I

I

t

10

12

14

16

8

I

!

I

10

12

14

16

Temperature (°C)

Fig. 3. Relations between harvest index (grain mass/above ground dry mass) and the mean temperature from sowing to harvest maturity in 1991/1992 (a), 1993/1994 (b) 1992/1993 (c), and 1994/1995 (d) for crops grown at increased (solid symbols) or normal [CO2] (open symbols). All regressions are significant (P < 0.05) except in (b) where the lines indicate the mean values, and where separate regressions are shown for the two CO, treatments they are significantly different (P < 0.05).

74 different temperature plots develop at different rates in these experiments and are therefore subjected to a unique coincidence of weather conditions with ontogenetic stage. There is inevitably a confounding effect in any such experiments using the normal seasonal time course of insolation, such that plants in higher temperatures mature earlier and therefore 'miss' the generally increasing solar radiation (see also Mitchell et al., 1993). Moreover, the coincidence of stress events during sensitive developmental phases, particularly high temperatures following anthesis (Stone and Nicolas, 1995) significantly contributes to seasonal variation in grain yields via influence on the rate and duration of grain growth (Wardlaw and Moncur, 1995; Wheeler et al., 1996b). Therefore, the different magnitudes of response to both increased [CO2] and temperature in each year is not surprising, and it emphasises the difficulty of making yield predictions for future climate and atmospheric conditions, without very detailed scenarios. Seasonal warming of only ca. 1-2°C was sufficient to negate the grain yield stimulation due to increased [CO2] (Table 4). Indeed, in 1993/1994 and 1994/1995 the benefit of doubling [CO2] to grain yield was particularly small at the warmer temperatures as a result of the negative interaction between [CO2] and temperature, so much so that only limited extrapolation of the relations (see Fig. 2g,h) would be required before the increase in grain yield from doubling [CO2] was lost. Thus, in some seasons, and at some temperatures, the response to increased [CO2] was small given the large [CO2] treatment; large compared to scenarios of only approaching twice the pre-industrial [CO2] (viz. 540 ~mol CO2 mo1-1) by the end of next century, (IS92d scenario, IPCC, 1996). The limited warming of only 1-2°C to offset the grain yield from doubled [CO2] is considerably smaller than the 4.5°C estimated to almost negate yield gains in a modelling study by Goudriaan and Unsworth (1990), indicating a greater sensitivity of yield to temperature than characterised by the model. However, the same comments apply here as in the above discussion over predictions of changes in duration that as most warming in these experiments was in the spring and summer expressing the warming expressed on a whole season basis may be overestimating the sensitivity. If Figs. 2 and 3 were shown using the mean temperature during grain filling (from anthesis to maturity) the data would have a span

of ca. 6°C for years 1 and 2 and ca. 4°C in years 3 and 4, rather than the spans shown of between 2 and 3°C (see also Wheeler et al., 1996a,Wheeler et al., 1996b). Using this temperature basis would increase these experimentally derived estimates of the temperature required to negate the CO2 effect to 2-3°C, which still represents a substantially higher temperature sensitivity than the model used by Goudriaan and Unsworth (1990). While such comparisons between the relative effects of mean seasonal temperature and [CO2] increase are useful, the detailed pattern of temperature and other climate variables during the season is important, as shown here and in modelling work elsewhere (Grashoff et al., 1995; Nonhebel, 1994). While above ground biomass and grain yield showed a broadly similar pattern of response to temperature and [CO2] in all years (Fig. 2), there were substantial changes of harvest index with temperature and with increased [CO2] in most cases (Fig. 3). The marked declines in harvest index that occurred in the first 2 years were when yields were affected by high temperature episodes during grain filling (Wheeler et al., 1996a,b). Analyses of which yield components accounted for the observed variation of grain yield between treatments, showed that for all years the number of grain per unit area was a major determinant, and this arose either through variation in ear number m-2 or grain number per ear (Table 5). Grain size and its response to temperature and [CO2] differed between years, and in normal [CO2] was largest in years 2 and 3 (ca. 45 mg at cooler temperatures), compared to ca. 32 mg in equivalent temperatures in years 1 and 4 (see Batts et al., 1998; Wheeler et al., 1996a) clearly indicating different source-sink relations between years. The role of the ear number in determining yield varied between years, and the relative effect of increased [CO2] on ear number at maturity varied (9%, 0%, 12% and 20% increases, in years 1-4, respectively). Many other reports have found substantial stimulation of tillering in wheat with increased [CO2] (e.g., Gifford, 1977; Mitchell et al., 1993, 1995). The propensity of winter wheat to tiller during the period around double ridge to terminal spikelet formation is linearly related to main stem dry mass at the same stage (Batts and Baylis, 1985; Batts et al., 1996), which depends on the cultivar (Batts et al., 1998) and on the environmental conditions from sowing. For example, Mitchell et al.

75

(1996) found that shading a wheat crop near terminal spikelet formation had a substantial effect on reducing the number of ears per unit area, compared with two alternative shading periods between terminal spikelet and the start of grain-fill. The lack of [CO2] promotion of tillering in 1992/1993 (Batts et al., 1996) may have been caused by the low solar radiation flux during the vegetative development period in this year (Table 3), and this effect may have been exacerbated by the reduction in solar radiation within the temperature gradient tunnels. However, there was still a pronounced increase in yield in response to increased [CO2] in this year because of more grain per ear, and it should be noted that the grain filling rate of the cooler treatments at normal [CO2] was the same as that outside (Wheeler et al., 1996b) indicating that the radiation regime in the tunnels was not too low. Clearly, wheat plants have substantial flexibility in the partitioning of the increased photosynthate supply in increased [CO2], and therefore increased tillering may not be a prerequisite for effects on grain yield. Unravelling these complex effects is difficult experimentally in the field. There is a need to explore further the influence of climatic variation within and between years on crop performance and yield, with both experiments and modelling studies (e.g., Semenov and Porter, 1995; Grashoff et al., 1995). Detailed field datasets, from the same location, over several unique environmental patterns such as summarised here are valuable in calibrating and validating crop growth models that are to be used to predict the impacts of climate change on crop production. The variation in response to increased [CO2] we observed here has important implications to any attempt to provide climate change impact predictions, either for changes in the mean climate, or for the equally important possible changes in climate variability (e.g., Kattenberg et al., 1996).

Acknowledgements We thank X. Liu, Pam Nkemka, Marelia Puche, G. Ruffles and M. Cantwell for technical assistance. This research was funded by the UK AFRC Wheat-CO2 Minitopic (GRB, TRW) and the UK BBSRC BAGEC programmes (GRB), The Royal Society and The University of Reading Research Endowment

Fund. This work forms part of the IGBP Global Change and Terrestrial Ecosystems Focus 3 Wheat Network. GRB acknowledges funding from the EU Environment programme under the MODEXCROP contract, no. ENV4-CT95-0142.

References Batts, G.R. and Baylis, A.D., 1985. An investigation of the factors influencing tiller death relating to the growth and development of four varieties of winter wheat. Report Centre, Zeneca Agrochemicals, Jealott's Hill Research Station, Brackneil, UK. Batts, G.R., Wheeler, T.R., Morison, J.I.L., Ellis, R.H. and Hadley, P., 1996. Developmental and tillering responses of winter wheat (Triticum aestivum) crops to CO2 and temperature. J. Agric. Sci. (Cambridge), 127: 23-35. Batts, G.R., Ellis, R.H., Morison, J.I.L., Nkemka, P.N., Gregory, P.J. and Hadley, P., 1998. Yield and partitioning in crops of contrasting cultivars of winter wheat (Triticum aestivum) in response to CO2 and temperature in field studies using temperature gradient tunnels. J. Agric. Sci (Cambridge), 130: in press. Butterfield, R.E. and Morison, J.I.L., 1992. Modelling the impact of climatic warming on winter wheat cereal development. Agric. For. Meteorol., 62:241-26 I. Gifford, R.M., 1977. Growth pattern, carbon dioxide exchange and dry weight distribution in wheat growing under differing photosynthetic environments. Aust. J. Plant Physiol., 4: 99-110. Gifford, R.M. and Morison, J.I.L., 1993. Crop response to the global increase in atmospheric CO2 concentration. Int. Crop Sci., I: 325-331. Goudriaan, J. and Unsworth, M.H., 1990. Implications of increasing carbon dioxide and climate change for agricultural productivity and water resources. In: B.A. Kimball, N.J. Rosenberg and L.H. Allen (Editors), Impact of Carbon Dioxide, Trace Gases, and Climate Change on Global Agriculture. Special Publication Number 53. American Society of Agronomy, Madison, Wl, pp 111-130. Grashoff, C., Dijkstra, P., Nonhebel, S., Schapendonk, A.H.C.M. and van de Geijn, S.C., 1995. Effects of climate change on productivity of cereals and legumes: model evaluation of observed year to year variability of the CO2 response. Global Change Biol., 1:417-428. Hadley, P., Batts, G.R., Ellis, R.H., Morison, J.I.L., Pearson, S. and Wheeler, T.R., 1995. Temperature gradient chambers for research on global environment change. II. A twin-wall tunnel system for low-stature, field-grown crops using a split heat pump. Plant Cell Environ., 18: 1055-1063. IPCC, 1996. In: J.T. Houghton, L.G. Meira Filho, B.A. Allander, N. Harris, A. Kattenberg and K. Maskell, Climate Change 1995: The Science of Climate Change. Cambridge University Press, Cambridge, 572 pp. Kattenberg, A., Giorgi, F., Grassl, H., Meehl, G.A., Mitchell, J.F.B., Stouffer, R.J., Tokioka, T., Weaver, A.J. and Wigley, T.M.L., 1996. Climate Change 1995: The Science of Climate

76

Change. Climate Models - Projections of Future Climate. Cambridge University Press, Cambridge, pp. pp 289-349. Lawlor, D.W. and Mitchell, R.A.C., 1991. The effects of increasing CO2 on crop photosynthesis and productivity: a review of field studies. Plant Cell Environ., 14: 807-818. Marc, J. and Gifford, R.M. 1984. Floral initiation in wheat, sunflower, and sorghum under carbon dioxide enrichment. Can. J. Bot., 62: 9-14. Mitchell, R.A.C., Mitchell, V.J., Driscoil, S.P., Franklin, J. and Lawlor, D.W., 1993. Effects of increased CO2 concentration and temperature on growth and yield of winter wheat at two levels of nitrogen application. Plant Cell Environ., 16: 521529. Mitchell, R.A.C., Lawlor, D.W., Mitchell, V.J., Gibbard, C.L., White E.M. and Porter J.R., 1995. Effects of increased CO2 concentration and increased temperature on winter wheat: test of ARCWheatl simulation model. Plant Cell Environ., 18: 736748. Mitchell, R.A.C., Gibbard, C.L., Mitchell, V.J. and Lawlor, D.W., 1996. Effects of shading in different developmental phases on biomass and grain yield of winter wheat at ambient and increased CO2. Plant Cell Environ., 19:615-621. Morison, J.I.L., 1996. Global Environmental change impacts on crop growth and production in Europe. In: Implications of Global Environmental Change for Crops in Europe, Aspects of Applied Biology, Vol. 45. Association of Applied Biologists, pp. 62-74. Nonhebel, S., 1994. The effects of use of average instead of daily weather data in crop growth simulation models. Agric. Systems, 44: 377-396.

Rawson, H.M., 1992. Plant responses to temperature under conditions of increased CO2 Aust. J. Bot., 40: 473-490. Rawson, H.M., 1995. Yield response of two wheat genotypes to carbon dioxide and temperature in field studies using temperature gradient tunnels. Aust. J. Plant Physiol., 22: 23-32. Semenov, M.A. and Porter, J.R., 1995. Climatic variability and the modelling of crop yields. Agric. For. Meteorol., 73: 265-283. Slafer, G.A. and Rawson, H.M., 1994. Sensitivity of wheat phasic development to major environmental factors: a re-examination of some assumptions made by physiologists and modellers. Aust. J. Plant Physiol., 21: 393-426. Stone, P.J. and Nicolas, M.E., 1995. Comparison of sudden heat stress with gradual exposure to high temperature during grain filling in two wheat varieties differing in heat tolerance. I. Grain growth. Aust. J. Plant Physiol., 22: 935-944. Wardlaw, I.F. and Moncur, L., 1995. The response of wheat to high temperature following anthesis. I. The rate and duration of kernel filling. Aust. J. Plant Physiol., 22: 391-397. Watson, R.T, Zinyowera, M.C. and Moss, R.H., 1996. Climate Change 1995: Impacts, Adaptations and Mitigation of Climate Change: Scientific-Technical Analyses. Cambridge University Press, Cambridge. Wheeler, T.R., Batts, G.R., Ellis, R.H., Hadley, P. and Morison, J.I.L., 1996a. Growth and yield of winter wheat (Triticum aestivum) crops in response to CO2 and temperature. J. Agric. Sci., 127: 37-48. Wheeler, T.R., Hong, T.D., Ellis, R.H., Batts, G.R., Morison, J.I.L. and Hadley, P., 1996b. The duration and rate of grain growth, and harvest index, of wheat (Triticum aestivum L.) in response to temperature and CO2. J. Exp. Bot., 47: 623-630.

© 1997 Elsevier Science B. V. A 11rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

77

Use of in-field measurements of green leaf area and incident radiation to estimate the effects of yellow rust epidemics on the yield of winter wheat R.J. Bryson a'*, N.D. Paveley b, W.S. Clark a, R. Sylvester-Bradley a, R.K.

Scott c

aADAS Boxworth, Battlegate Road, Boxworth, Cambridge, CB3 8NN, UK bADAS High Mowthorpe, Duggleby, Malton, North Yorks, YOI7 8BP, UK CUniversity of Nottingham, Department of Agriculture and Horticulture, Sutton Bonington Campus, Loughborough, Leics., LEI2 5RD, UK Accepted 16 May 1997

Abstract In-field estimates of green leaf area index for treatments with varying amounts of yellow rust (Puccinia striiformis (Westend.)) were directly proportional to laboratory measured green leaf area index (R 2 = 0.75). The field technique depended on shoot counts and a leaf form factor (F = 0.83) which was derived from 20 varieties of winter wheat by relating the product of their leaf lengths and widths to leaf areas measured by a planimeter (R E = 0.95). In two experiments at ADAS Terrington, UK, on the susceptible winter wheat variety Slejpner, epidemics of yellow rust ranged from nil to severe with 60 (1994) and 52 (1995) different combinations of fungicide dose and timing. Assessments of disease severity (%) integrated as the area under the disease progress curve accounted for yield differences within each season, but the relationship differed markedly between seasons. In-field assessments of green leaf area index integrated over time, or healthy area duration, showed a curvilinear relationship with grain yield (1994, R E = 0.63" 1995, R 2 = 0.73), but any healthy area duration value in the brighter year of 1995 related to larger yields than the equivalent value in 1994. Intercepted radiation by green leaf tissue accumulated after flowering (20 June in both years), estimated through the Beer's Law analogy from field-measured green leaf area index and total incident radiation (i.e., healthy area absorption), accounted for more variation in grain yield (1994, R E= 0.80; 1995, R E - 0.92). There was no seasonal difference in the conversion coefficient between grain dry matter and the amount of incident radiation absorbed by green leaf tissue (1.4 g/MJ) but the intercepts of the relationships were sensitive to the date from which integration began. It is suggested that in-field green leaf area index assessments, interpreted through a simple model which provides estimates of differences in intercepted light energy, may prove useful in the analysis of experiments on disease control. © 1997 Elsevier Science B.V.

Keywords: Winter wheat; Yellow rust; Green leaf area; Radiation interception; AUDPC; Radiation use efficiency

I. Introduction The relationship between disease control by fungicides and yield loss varies from site-to-site and from * Corresponding author. Tel.: +44 1954 268242; fax: +44 1954 268268; e-mail: rosie_bryson @adas.co.uk

season-to-season, the value of disease control is therefore difficult to predict (Webster, 1987; Murray et al., 1994). Most empirical models of yield loss due to disease do not take account of host processes nor the wide range of potential yields. Such models may be acceptable for survey estimates of yield loss, especially in crops with a narrow yield range exposed to

Reprinted from the European Journal of Agronomy 7 (1997) 53-62

78

short duration disease epidemics late in the season, but are less helpful in developing disease control strategies. Traditionally, estimates of the effects of fungicides on foliar pathogens have been made by assessing the proportion of leaf area on which symptoms are expressed following treatment, either at a particular growth stage (Murray et al., 1994), or as the area under the disease progress curve (AUDPC) (Fry, 1975; Large, 1952; Teng, 1983, 1985). As AUDPC is an integral of disease severity over time it is often used as a comprehensive measure of the effectiveness of fungicides (Shaner and Finney, 1977). However, measurements of disease severity alone may not fully reflect the effects of disease on the yield-forming processes in the host. A better alternative may be to base disease/yield loss models on crop function (Waggoner and Berger, 1987). Yield is predominantly determined by the crop's capacity to intercept light energy and utilise it for growth. Potential yield is directly related to the amount of photosynthetically active radiation intercepted by green tissue (Monteith, 1977). This can be described formally by an equation derived from Beer's Law (Monteith and Unsworth, 1990): f = 1 - exp(-kL), where f = fraction of light intercepted, k=extinction coefficient (which is dependent on canopy geometry) and L = green leaf area index (GLAI). GLAI is defined as the number of units of planar area of leaves per unit area of ground. The Beer's Law analogy implies that there is an optimal canopy size, considering all green tissues, at which the cost of creating, maintaining or protecting a further increment in canopy size may prove uneconomic in terms of growth (SylvesterBradley et al., 1995). Several studies have confirmed that measurements of canopy size, and in particular the effect of disease on GLAI, correlate more closely to yield loss than estimates of percentage disease severity alone (Lim and Gaunt, 1981; Waggoner and Berger, 1987; Whelan and Gaunt, 1990; Bryson et al., 1995). However, in disease/yield loss studies and fungicide efficacy experiments, measures of GLAI are rare. A possible reason for this is the practical difficulty of measuring GLAI, particularly of diseased leaves. Green area can either be measured in the laboratory using a planimeter, which is time consuming and involves destructive sampling or by the use of hand-held canopy

analysers which measure total vegetation area only (Campbell, 1986; Welles and Norman, 1991). In a much cited paper, Waggoner and Berger (1987), suggested that disease progress should be related to crop growth by taking into account both the amount of green area available for photosynthesis, the 'healthy area duration' (HAD), and the amount of incident radiation absorbed by that healthy area, the 'healthy area absorption' (HAA), using the Beer's law analogy. An attempt has been made in this paper to test whether the HAD and HAA models described by Waggoner and Berger (1987) are applicable to foliar disease epidemics on winter wheat in a temperate environment and whether these models can be supported by simple in-field measurements of disease severity, GLAI and total incident radiation. This paper describes the derivation and testing of a rapid field based method to determine GLAI in healthy and diseased crop canopies of winter wheat using measurements of leaf dimensions. Data are presented from two field experiments where different epidemic levels of yellow rust were achieved using combinations of fungicide dose and timing. These were used to test whether the effect of yellow rust on the yield of winter wheat could be better accounted for by calculating loss of green leaf area and hence reduction in radiation interception, than by disease severity measurements.

2. Materials and methods

2.1. Determination of winter wheat leaf form factor To allow accurate determination of leaf area from length and width measurements taken in the field it was necessary to determine a leaf form factor. Leaf samples were taken at GS39 (Tottman, 1987) from the fungicide treated plots of a variety experiment at ADAS Terrington, Norfolk. The experiment was a fully randomised block design with three replicate blocks of 20 varieties; Admiral, Andante, Avalon, Beaver, Brigadier, Cadenza, Estica, Flame, Galahad, Haven, Hereward, Hornet, Hunter, Hussar, Longbow, Mercia, Norman, Riband, Rialto and Zodiac. A 0.5 m 2 sample was taken randomly from each plot by cutting plants from the sample area at ground level. Green

79

leaves were randomly selected from each sample to represent leaves at different layers in the crop canopy. They were cut from the stem and leaf sheath immediately above the ligule (50 leaves plot -l) and wrapped in moist paper towel prior to measurement in the laboratory. The total area (cm 2) of the 50 leaves from each plot was determined using a calibrated leaf area meter (Delta-T Devices Ltd., Burwell, Cambridge). The length and width of each individual leaf was then measured using a grid measuring 350 x 50 mm, and delineated into 5 mm units on the length-axis and 1 mm on the width-axis.

2.2. Comparison of GLAI using a leaf area meter and linear leaf measurements in healthy and diseased winter wheat canopies GLAI is a function of the amount of green leaf area per shoot and the number of shoots per unit area of ground. Leaf length and width measurements taken in the field were used in conjunction with a form factor and shoot counts to calculate GLAI. These estimates were then compared with laboratory based leaf area meter measurements on both healthy and diseased crop canopies. In order to test the method rigorously, wheat canopies of different sizes were measured weekly from GS39 until canopy death. Crop sampling for both methods was carded out on an experiment at ADAS Terrington. The experiment was a fully randomised block design with 4 replicate blocks each of 6 treatments. The treatments were the factorial combination of three levels of nitrogen (60, 180 and 270 kg ha-l), with and without yellow rust (Puccinia striiformis (Westend.)), on a susceptible variety Hornet (Bryson et al., 1995). Yellow rust plots were inoculated at GS33 by the introduction of yellow rust infected, potgrown wheat plants. Plots without yellow rust were prophylactically treated with the fungicide tebuconazole (as c.p. Folicur-Bayer).

2.2.1. Determination of GLAI in the field using leaf length and width measurements Measurements were carded out weekly on 10 randomly selected, destructively sampled shoots per plot. Leaf length and width were measured as described above. When the leaf was not fully expanded (i.e., no ligule visible), the length measurement was taken from just the emerged portion of the leaf. Leaf width

was measured to the nearest millimetre by flattening the widest part of the leaf (generally just above the ligule) on the base scale. Assessments of the percentage of leaf area expressing disease symptoms (Anonymous, 1976) and the percentage of remaining green area were made on the same shoots and at the same time as leaf dimensions. Each leaf layer was assessed for the presence of yellow rust (P striiformis), brown rust (Puccinia recondita), Septoria spp. (S. nodorum and S. tritici) and mildew (Erysiphe graminis f.sp. tritici). As shoot number remains constant after GS 55, an assessment of the number of shoots m -2 was only recorded on one occasion prior to harvest when fertile shoots could be most easily identified. Shoot number m -2 was calculated from counts of the number of fertile shoots in four, randomly selected 1.0 m length rows per plot (row spacing = 12.5 cm).

2.2.2. Determination of GLAI in the laboratory using a leaf area meter A 0.75 m 2 sample was taken from each plot from pre-determined areas to avoid local bias. A gap of at least 20 cm was left between sample areas with at least 50 cm from the end of the plots and/or tramlines. Plants were cut from the sample area at ground level with scissors. All the above ground material was placed in plastic bags and stored for a maximum of 3 days in a cold room at 4-6°C prior to growth analysis. In the laboratory, the total sample fresh weight was recorded. A randomly selected sub-sample (approx. 10%) was taken, weighed and analysed in the laboratory to determine green leaf areas and shoot numbers m -2. The total green leaf area was measured using a calibrated leaf area meter. For the diseased and senescing leaves it was necessary to make a subjective judgment as to how much of the leaf was green. If dead, non-green or diseased areas of the leaf were patchy it was necessary to assess the percentage of the leaf area affected; that amount was removed from the leaf and the remaining area was then classed as green and measured.

2.3. Fungicide dose response experiment 2.3.1. Experiment design In 1994 and 1995 at ADAS Terrington, Norfolk, experimental plots of the yellow rust susceptible winter wheat variety, Slejpner were arranged in a fully

80

randomised block design with two replicate blocks of 60 (1994) and 52 (1995) treatments. The fungicide mixture tebuconazole (as c.p. Folicur-Bayer) plus fenpropidin (as c.p. Patrol-Zeneca) was applied at either full, 0.5 or 0.25 of the label recommended dose (1 1 c.p. ha -l plus 0.7 1 c.p. ha -l respectively). Sprays were applied as a combination of timings at eventual leaf 3 fully emerged (typically GS32), eventual leaf 2 fully emerged (typically GS33), eventual leaf 1 (the flag leaf) fully emerged (GS 39) and ear fully emerged (GS59) (Tottman, 1987). As this paper deals specifically with the yield of the winter wheat crop as a result of differing levels of disease epidemic and not the efficacy of the fungicide timings and doses, full treatment details are not given.

2.3.2. Assessments Percentage disease and green leaf area assessments were carried out on 10 randomly selected shoots per plot as described above. Leaf length and width measurements were carried out on two of the 10 randomly selected shoots and used to determine GLAI as detailed previously. 2.3.3. Additional measurements All plots were combine harvested using a 'Sampo' plot combine harvester. One combine strip was taken through the centre of each plot to avoid edge effects. A 1 kg sample from each plot was retained for the determination of moisture content. All yields were corrected to 85% dry matter. Total incident radiation was measured using a dome solarimeter (Delta-T Devices Ltd., Burwell, Cambridge) placed above the crop. Readings were taken every minute and then averaged every hour to determine a value of total incident radiation per day (MJ m -2 day-l). Data handling and statistical analysis was carried out using Excel 5.0 and MLP (Ross, 1987). Unless otherwise stated, all significance is quoted at the 95% confidence level.

wheat at flag leaf emergence stage (GS 39) was from 50 leaves from each of three replicate plots of 20 varieties. The intercept of the line was found not to be significantly different from zero by regression analysis. Therefore, the relationship between leaf area obtained by the leaf area meter and by the product of length and width was determined using the equation y = Fx where y (leaf area meter leaf area) and x (leaf length x width area) were the means of their respective populations and F, the leaf form factor, where the intercept was equal to zero (Fig. 1) (Bhan and Pande, 1966). Length and width measurements did not vary greatly between varieties. Mean width ranged from 1.4 cm (Cadenza, Estica) to 1.7 cm (Admiral, Longbow, Norman) and mean length, from 17.5 cm (Longbow) to 21.5 cm (Hornet). The relationship y = 0.83x was derived from all the data points over the 20 varieties tested giving a leaf form factor of 0.83 (R2= 0.95; Fig. 1). There was no significant difference in leaf form factor between varieties. The form factor was not assessed at growth stages before or after GS39 but the consistency that was found, irrespective of varying relationships between length and width, supports the use of a single value throughout crop growth. Inspection of individual leaves showed that there was little evidence of significant variation in rectangularity. The difference from rectangularity which the form factor represents occurs mainly at the leaf tip. It appears that ...., 1800

~ 16oo y = 0.83x = 0.95

i

141111

n

"~ 121111

[

k ~' 10110 °.

I 1000

3. R e s u l t s

3.1. Winter wheat leaf form factor (F) The data used to derive a leaf form factor for winter

tl,,,,m~,~

R z

I 1200

I 1400

I 1600

I 1800

Leaf length x leaf width area- (cm 2 for 50 leaves)

Fig. I. The data used to derive a leaf form factor (/7) for winter wheat at flag leaf emergence stage (GS39). Points are from 50 leaves from each of three replicate plots of 20 varieties. F, the slope of the relationship was 0.83.

81

this portion of the leaf is a consistent proportion of the whole. Most modem varieties of winter wheat, despite their genotypic differences, generally have similar canopy morphology if compared with older varieties (Gale and Youssefian, 1985). F derived in this study was not only consistent between varieties but was similar to F values of 0.85 and 0.82 determined by Owen (1968) for the older wheat varieties Gabo and Mexico. Also Bhan and Pande (1966) reported that F = 0.80 for several different varieties of rice. This suggests that the relationship of leaf length and width, and thus leaf shape, may be consistent across species of cereals possibly allowing the use of a single figure for F in future cereal crop leaf area studies.

3.2. Comparison of field and laboratory measured GLAI Field GLAI was calculated in both yellow rust inoculated and fungicide treated plots, F was taken as 0.83. Shoot numbers were assessed in the field at GS 75 and were not significantly different from the shoot numbers obtained in the laboratory from the quadrat samples. A single figure for shoot numbers m -2 was used to calculate field GLAI at each nitrogen rate (N1 = 330, N2 =410, N3 = 4 1 8 shoots m-2). Laboratory calculated GLAI was determined using both green area and shoot number measurements determined for each individual quadrat at each sampling time and in each nitrogen rate. Comparisons of GLAI determined from field and laboratory measurements are shown in Fig. 2 for both the fungicide treated and yellow rust inoculated plots. Field and laboratory GLAI did not differ significantly from a one to one proportionality, y = 0.93x (R 2 = 0.75). However, the field measurement of GLAI was slightly overestimated compared with the laboratory measurements. A possible reason for this was that length and width measurements in the field were made on main shoots (i.e., those shoots selected for disease assessments) whereas quadrat samples consisted of a range of shoot sizes, which may have resulted in smaller mean leaf areas. It is also possible that although care was taken to minimise sample damage, sampling and processing of the quadrat samples may have resulted in dehydration or accidental loss of green area, resulting in a decrease in GLAI in the laboratory

as compared with field measurements. Laboratory and field determination of GLAI in diseased samples both rely on the visual assessment of the percentage of leaf area expressing symptoms. In this study, percentage disease and green area assessments were made by the same individuals in both the field and laboratory and should not have contributed to the small differences found between the two techniques.

3.3. Fungicide dose experiment Although the grain yield, disease severity and other crop measurements presented in this section were obtained following the application of fungicide treatments, the aim here is not to consider the efficacy of these treatments but to describe the relationship between the yellow rust epidemic, the physiology of the host and final grain yield. In both years, the predominant foliar disease was yellow rust; the combined amount of other foliar diseases did not exceed 2% on any leaf layer in either year. The yellow rust epidemic was more severe in 1995 than 1994 with a maximum of 80% (SE 7%) (1995) and 62% (SE 4%) (1994) of the area of leaf 2 affected by symptoms at GS 75 in untreated plots.

AA O , O

O

me

o •

o O .at,'

e~ifb

~4 AD-

y = 0.93x R 2 : 0.75

~3

~

A 4L

O

A &

0

1

2

3

4

5

6

Field measured G L A I

Fig. 2. Comparison of green leaf area index (GLAI) calculated from field measurements of green leaf area (leaf length x width x form factor) and fertile shoot counts, with GLAI assessed in the laboratory from green leaf area measured with a meter and fertile shoot counts. Samples were from 4 replicate plots of three nitrogen treatments taken on four sampling occasions from yellow rust inoculated plots (D, N l; O, N2; A, N3) and fungicide treated plots (1, Ni; O, N2; &, N3).

82

with AUDPC in 1994 and 1995 (Fig. 3), the equations of the regression lines are given below. • "O..

~.e,~-

|

~ e~

a a'o. *

~

~..

oo

o

......

o a""

6

O

".. O" O

4

i

~

a

2

I

I

I

I

I

1500

3000

4500

6000

7500

AUDPC (% disease severity days)

Fig. 3. The relationship between the sum of the areas under the disease progress curve (AUDPC) for leaves 1 (14th June-25th July), 2 (31st-May 25th June) and 3 (31st May-19th July) and grain yield (tha-I at 85% dry matter) in 1994 (I-1, full line) and 1995 (O, dotted line).

3.3.1. Disease severity based yield models To aid interpretation of the effects of foliar disease epidemics on crop yield, quantitative techniques have been developed employing the disease progress curve as a starting point (Teng, 1983). Disease epidemics are usually quantified by measuring disease severity, in this case the percentage of the leaf area expressing visible yellow rust symptoms. Empirical models for estimating yield loss caused by a single disease may be categorised into single point (one independent variable, representing the entire epidemic, to reflect yield loss), multiple point (yield loss estimated from several severity assessments) or integral models (loss prediction from input variables that represent disease over a defined epidemic duration, such as area under the disease progress curve (AUDPC)). The AUDPC model is the most comprehensive of the three and is commonly used in disease/yield loss studies. It is used here to relate the yellow rust epidemics to grain yield, within and between experimental seasons. The AUDPC values reported here are the sum of AUDPC for leaves 1, 2 and 3 calculated by the trapezoidal rule (Campbell and Madden, 1990) (Fig. 3). For leaf 1, the percentage of yellow rust severity was integrated from 14th June to 25th July. For leaves 2 and 3, percentage severity was integrated from 31st May to 25th and 19th July, respectively. These dates were the same in both years. Yield was negatively correlated

(1994) y= 10.08-0.00142x (R2=0.69)

(1)

(1995) y=10.85-0.00089X (R2=0.90)

(2)

The relationship between yield and AUDPC was close in 1995 (eq. (2)) but less so in 1994 (eq. (1)). The data in 1994 were very variable with AUDPC values of zero resulting in a yield difference of over 3 tha -~ (Fig. 3). When the regressions for both years were compared, both the slopes and intercepts were found to be significantly different. Although the effect of the fungicide treatments on yield could be explored further within each experimental year, no useful summary could be made between years. The relationship between disease and yield loss is inherently more complex than is implied by disease/ yield loss models using linear regression (Teng, 1983). A single AUDPC value may describe a severe epidemic for a short time or a minor epidemic for a long time. To account for the variable effects of disease severity, Madden et al. (1981), developed a nonlinear model based on the Weibull distribution. The advantage of this model was that it allowed for the definition of a minimum threshold of disease below which no loss occurred. However, experimental determination of the minimum threshold was found to be difficult as, in the early phases of an epidemic, not all plants were infected and sampling a diseased population with a low level of disease gave highly variable data (Teng, 1985). Single point, multiple point and integral models all rely on a measure of disease severity, i.e. the proportion of host tissue showing symptoms (James and Teng, 1979), and as such have only an indirect link to the productivity of the host plant. Also, they do not take account of environmental conditions such as incident radiation, rainfall and soil fertility which will all affect yield potential. Hence, relationships between disease severity and yield have generally proved poor over seasons, as demonstrated here (Fig. 3), and over sites and seasons as shown by other workers (James, 1974; Waggoner, 1977; Murray et al., 1994; Bryson et al., 1995).

3.3.2. Crop based yield loss models The HAD values reported here are the sum of the

83 integrals of GLAI through time for leaves l, 2 and 3 (from 31st May (GS39)) until no green area remained (19 July, leaf 3 and 25 July, leaves l and 2) for both 1994 and 1995. Following the precedent of the definition of HAD and HAA (Waggoner and Berger, 1987), no account was taken of ear green area or interception; in these experiments disease did not affect the ears. HAD from 31st May gave a curvi-linear relationship with yield in both 1994 and 1995 (Fig. 4). A simple exponential curve was fitted to the data giving:

(SE 0.28) and 1995 of 8.29 (SE 0.34) and total incident radiation from 31 st May in 1994 of 1200 MJ m -2 and 1995 = 1336 MJ m -2. In Fig. 5 an estimate of the healthy area absorption (HAA), i.e., accumulated intercepted radiation (MJ m -E) by green leaf tissue, was calculated from 20th June (approximately GS 61) until no green area remained (18th July) assuming an extinction coefficient of 0.45 (Sylvester-Bradley et al., 1990). In both years HAA related directly to yield. The regression equations are given below.

(1994) y= 11.22-47.52exp(-0.017x) (R E=0.63)

(1994) y = 0 . 8 5 - 0 . 1 7 x (R2=0.82)

(3) (1995) y= 11.80-29.57exp(-0.01 lx) (R 2 =0.73)

(4) Within each experimental year HAD and yield related reasonably well (eq. (3) and eq. (4)). However, between years all three parameters of the curves were significantly different. In particular, the curve for 1995 was horizontally displaced in relation to 1994 (Fig. 4). The HAD model does not take account of either the way light is attenuated by crop canopies of different sizes, the amount of total incident radiation available or interception by ears and other nonleaf organs. These parameters differed between the two seasons with a maximum GLAI in 1994 of 6.03

"G 10

l

S

& s

[] ,'" o

o..i;),"

@ 6

I~ v 0 o o o o

13

&

0

//"

i 2

'

0

.SO

,,I

100

I

150

I

I

200 2.50 HAD (days)

I

I

I

300

350

4100

Fig. 4. The relationship of grain yield to healthy area duration (HAD) after 31st May of leaves l and 2 (until 25th July) and leaf 3 (until 19th July) in 1994 (l:], full line) and 1995 (O, dotted line).

(1995)

y=l.81-O.O16x (R2=0.91)

(5) (6)

The relationship of yield with HAA in both seasons was closer than the relationship with HAD (eq. (5) and eq. (6); Fig. 5). There was no significant difference in slope over the two seasons but the intercepts were significantly different. It was found that if intercepted radiation was accumulated from earlier than 20th June the slope of the line did not change significantly but the intercept became increasingly negative. For example, the equations of the regression lines of intercepted radiation accumulated from the end of May were as follows: (1994) y = - 4 . 0 6 + 0 . 0 1 5 x (R2=0.80)

(7)

(1995) y= -3.67+0.015x (R2=0.92)

(8)

The significant difference in intercept between the two years probably relates to the growth stages at which the effects of disease on intercepted radiation began to relate to grain growth. Since maximum yields were similar in both seasons (Fig. 5) the period relating disease effects to yield effects is likely to have started and finished earlier in 1995 than 1994. This is supported by the observation that the yellow rust epidemics were more severe earlier in 1995 than 1994 but were checked sooner by the high temperatures in July and August of 1995. From the slopes of eq. (5) and eq. (6) (Fig. 5) the mean radiation use efficiency (RUE) by green leaf tissue from 20th June was calculated as 1.41 g grain dry matter MJ -I of total intercepted radiation. This is in line with the findings of Monteith (1977) who found that for crops such as barley, beet, apples and potatoes, RUE in unstressed situations was approximately 1.4 g dry matter MJ -I of total intercepted

84

Ii

!

10

S

,9"" j ~ a

o

o

° It1 O0

@ 6

ooO'6

J

0

0 oo°°" -°°

~

13 I:! 0

2

0

I 700

I 800

I 900

I 1000 Accumulated Intercepted radiation (MJ m"z)

I 1100

Fig. 5. The relationship of grain yield in 1994 (i--1, full line) and 1995 (O, dotted line) with radiation intercepted from 20th June to canopy senescence, by green area of leaves 1, 2 and 3 (healthy area absorption, HAA). Interception was calculated by the Beer's Law analogy, assuming an extinction coefficient of 0.45.

radiation. This suggests that in these experiments the primary effect of yellow rust was on radiation interception, not RUE.

4. Discussion Models which rely solely on the quantification of visible disease symptoms do not take account of variations in growing conditions which occur between geographically dispersed sites and between seasons, nor do they adequately recognise the period when the disease is actually causing yield loss (Teng, 1985). Several workers have already emphasised the need for a more crop based approach to develop disease control strategies, combining an understanding of crop growth with knowledge of disease development (Waggoner and Berger, 1987; Whelan and Gaunt, 1990; Bryson et al., 1995). However, there is still a reluctance amongst pathologists to incorporate crop growth measurements into disease/yield studies and fungicide experiments. This paper has demonstrated that it is possible to measure GLAI in the field at the same time as making conventional disease assessments. The use of length and width measurements, together with the leaf form factor, allowed rapid determination of green leaf area.

If this in-field method is to be adopted elsewhere it is important to appreciate its dependence on precise estimates of shoot number. It is also necessary to recognise that assessments of green leaf area either in the laboratory or in the field, particularly in diseased crop canopies, are based on subjective judgements of the percentage area of the leaf that is green; care must therefore be taken to standardise these assessments (Parker et al., 1995). As yet, no cheap and quick alternative exists, although several workers have suggested that canopy area measurements may be possible using spectroradiometers (Asrar et al., 1984; Whelan and Gaunt, 1990; Hansen, 1991). As the technology of spectral component analysis of absorbed and reflected wavelengths becomes more advanced, remote in-field assessments of both healthy and diseased crop canopies may become feasible and could potentially be used to develop and support cropbased disease/yield models. In-field measurements of GLAI were used to test whether the HAD and HAA models described by Waggoner and Berger (1987) could improve the explanation of yields over sites and seasons as compared with AUDPC models. The data on peanut plants collated by Waggoner and Berger (1987) for HAD gave a curvilinear relationship with yield. It was perhaps surprising that all of their data for 78 crops of peanuts over 14 years fitted one curve. It may be that, although the peanut crops were grown over several years, environmental conditions were relatively similar and the peanut canopies were not very different prior to defoliation. When the wheat canopies described here differed by a large amount of green area (>2 GLAI units between 1994 and 1995) and total incident radiation levels differed (by 136 MJ m -2 between 1994 and 1995), the HAD curves were significantly displaced. In this study, HAD did not give a consistent explanation of yield over the two seasons although it did demonstrate a decreasing return from increasing green area, suggesting that yield is more closely related to absorption of solar radiation than to leaf area alone. As an integral over time, HAD has a similar disadvantage to AUDPC in that it does not differentiate between large GLAI for a short period and small GLAI for a long period (Johnson, 1987). Nor does it account for the diminishing effect of increasing canopy size on the proportion of light intercepted (Monteith and Unsworth, 1990). This

85

has implications for disease control strategies in that a large canopy may be able to tolerate some loss of green area without an economically significant effect on yield. On the other hand, any loss of green area from a small canopy could have a serious effect on yield, making protection an economic necessity. In this study, different fungicide treatments resulted in very different values of HAD but equivalent yields. The relationship of yield to HAA gave a better correlation than that with HAD in both experimental years. With relatively large canopies, only two seasons to provide variation in incident radiation, and a large proportion of the treatments giving good disease control, most of the data points in this relationship are clumped. A more thorough test of the predictive power of HAA must depend on data from a broader range of circumstances. Nevertheless, the RUE of green leaf tissue determined here was not only consistent between the two seasons, but was similar to the RUE determined from separate disease control experiments at this site (1.2 g MJ-i; Bryson e t al., 1995) as well as to the radiation conversion coefficients reported for several different crops (Monteith, 1977) Although HAA gave the best estimate of harvested yield of the three models tested, the intercept of the relationship was found to be highly sensitive to the start date taken for the period of integration. For instance, when the relationship of yield to HAA was tested from the end of May the intercepts of the equations became negative, but the slopes were unchanged. Waggoner and Berger (1987) obtained negative intercepts when they related peanut yield to HAA. They concluded that the negative intercept indicated that no peanuts were set at very small HAA values. The values of HAA presented in this paper were calculated from 20th June in both years so that they could be compared on a common basis. On 20th June both crops were assessed as being at the start of anthesis (GS61; Tottman, 1987). However, assessment of growth stages can be prone to assessor error and the distinction between the beginning and middle of anthesis is uncertain when assessments are made on a weekly basis. It is therefore possible that the wheat crops in 1994 and 1995 were at different developmental stages on this date. It would appear that the integration of intercepted radiation over time must be

combined with precise and accurate records of growth stages if the approach, based on HAA, is to lead to an improved capacity to model absolute yield, rather than just yield 'loss'. Waggoner and Berger (1987) suggest that the amount of solar radiation intercepted by the green portion of a crop canopy is all that is needed to predict crop loss. This was obviously not the case in this study. Johnson (1987) pointed out that, when used over an entire season, HAA/yield models may not account for different source-sink relationships at different crop stages. Whilst restriction of the HAA model to the period when the harvested portion of the crop was developing provided a consistent relationship in this case, there are likely to be circumstances in which sink limitation will reduce RUE. The period before flowering is particularly important in determining the sink capacity of wheat (Evans and Wardlaw, 1996) and it may be necessary to monitor growth during this earlier period if a cropbased explanation of yield variation is to prove sufficiently robust to support commercial decision-taking. HAD and HAA, as originally defined by Waggoner and Berger (1987), do not take account of ear green area. Diseases other than yellow rust may affect yield by effects on ears (Jones and Odebunmi, 1971). In conclusion, definition of crop productivity as the product of radiation interception by green leaf tissue and RUE provides a framework for understanding disease induced reductions in yield. The approach described here is not intended as a predictive tool but it is envisaged that it will lead to the development of models which may be utilised in crop management decisions. Integrals such as HAA are an improvement over the more conventional forms of disease progress analysis, but further work is needed, particularly to account for the timing of disease effects in relation to the processes of crop development.

Acknowledgements The contribution of the Home-Grown Cereals Authority and the UK Ministry of Agriculture Fisheries and Food to the funding of this research is gratefully acknowledged. Thanks are due to Dr Alan Gay for the statistical analysis, and Miss Diane Moss, Mr

86

Jess Hunt and other ADAS colleagues for technical support.

References Anonymous, 1976. Manual of plant growth stage and disease assessment keys. Ministry of Agriculture Fisheries and Food, Harpenden. Asrar, G., Fuchs, M., Kanemasu, E.T. and Hatfield, J.l., 1984. Estimating absorbed photosynthetic radiation and leaf area index from spectral reflectance in wheat. Agron. J., 76: 300-306/ Bhan, V.M. and Pande, H.K., 1966. Measurement of leaf area of rice. Agron. J., 58: 454. Bryson, R.J., Sylvester-Bradley, R., Scott, R.K. and Paveley, N.D., 1995. Reconciling the effects of yellow rust on yield of winter wheat through measurements of green leaf area and radiation interception. Aspects Appl. Biol., 42: 9-18. Campbell, G.S., 1986. Extinction coefficients for radiation in plant canopies calculated using an ellipsoidal inclination angle distribution. Agric. For. Meteorol., 36: 317-321. Campbell, C.G. and Madden, L.V., 1990. Temporal analysis of epidemics I: Description and comparison of disease progress curves. In: C.G. Campbell and L.V. Madden (Editors), Introduction to Plant Disease Epidemiology. Wiley, New York. Evans, L.T. and Wardlaw, I.F., 1996. Wheat. Chapter 22:501-518. In: E. Zamski and A.A. Schaffer (Editors), Photoassimilate Distribution in Plants and Crops: Source:Sink Relationships. Marcel Dekker, New York, 905 pp. Fry, W.E., 1975. Integrated effects of polygenic resistance and a protective fungicide on development of potato late blight. Phytopathology, 65:908-911. Gale, M.D. and Youssefian, S., 1985. Dwarfing genes in wheat. In: G.E. Russell (Editor), Progress in Plant Breeding, Vol. 1. Butterworths, Oxford. Hansen, J.G., 1991. Use of multispectral radiometry in wheat yellow rust experiments. Bulletin OEPP/EPPO 21, 651-658. James, W.C., 1974. Assessment of plant diseases and losses. Annu. Rev. Phytopathol., 12: 27-48. James, W.C. and Teng, P.S., 1979. The quantification of production constraints associated with plant diseases. Appl. Biol., 4: 201267. Johnson, K.B., 1987. Defoliation, Disease, and Growth: A Reply. Phytopathology, 77: 1495-1497. Jones, D.G. and Odebunmi, K., 1971. The epidemiology of Septoria tritici and S. nodorum IV: The effect of inoculation at different growth stages and on different plant parts. Trans. Br. Mycol. Soc., 56: 281-288. Large, E.C., 1952. The interpretation of progress curves for potato blight and other plant diseases. Plant Pathol., 1: 109-117. Lim, L.G. and Gaunt, R.E., 1981. Leaf area as a factor in disease assessment. J. Agric. Sci., 97: 481-483. Madden, L.V., Pennypacker, S.P., Antle and C.E., Kingsolver,

C.H., 1981. A loss model for crops. Phytopathology, 17: 685689. Monteith, J.L., 1977. Climate and the efficiency of crop production in Britain. Philos. Trans. R. Soc. London Ser. B, 281: 277294. Monteith, J.L. and Unsworth, M.H., 1990. Principles of Environmental Physics. Edward Arnold, London. Murray, G.M., Ellison, P.J., Watson, A. and Cullis, B.R. 1994. The relationship between wheat yield and stripe rust as affected by length of epidemic and temperature at the grain development stage of crop growth. Plant Pathol., 43: 397-405. Owen, P.C., 1968. A measuring scale for areas of cereal leaves. Exp. Agric., 4: 275-278. Parker, S.R., Shaw, M.W. and Royle, D.J., 1995. The reliability of visual estimates of disease severity on cereal leaves. Plant Pathol., 44: 856-864. Ross, G.J.S., 1987. Maximum Likelihood Program. Release 3.08. Numerical Algorithms Group, Oxford. Shaner, G. and Finney, R.E., 1977. The effect of nitrogen fertilisation on the expression of slow-mildewing resistance in Knox wheat. Phytopathology, 67: 1051-1056. Sylvester-Bradley, R., Stokes, D.T. and Scott, R.K., 1990. A physiological analysis of the diminishing response of winter wheat to applied nitrogen - I. Theory. Aspects Appl. Biol., 25: 277287. Sylvester-Bradley, R., Goodlass, G., Paveley, N.D., Clare, R.W. and Scott, R.K., 1995. Optimising the use of fertiliser N on cereals and parallels for the development of fungicide use. In: H.G. Hewitt, D. Tyson, D.W. Hollomon, J.M. Smith, W.P. Davies and K.R. Dixon (Editors), A Vital Role for Fungicides in Cereal Production. Bios, Oxford, pp 43-56. Teng, P.S., 1983. Estimating and interpreting disease intensity and loss in commercial fields. Phytopathology, 73:1587-1590. Teng, P.S., 1985. Construction of predictive models II. Forecasting crop losses. In: C.A. Gilligan (Editor), Mathematical Modelling of Crop Diseases, Academic Press, London, pp. 179-206. Tottman, D.R., 1987. The decimal code for the growth stages of cereals with illustrations. Ann. Appl. Biol., 110: 441-454. Waggoner, P.E., 1977. Simulation of modeling of plant physiological processes to predict crop yields. In: J.J. Landsberg and C.V. Cutting (Editors), Environmental Effects on Crop Physiology. Academic Press, New York, pp. 351-363. Waggoner, P.E. and Berger, R.D., 1987. Defoliation, disease and growth. Phytopathology, 77: 393-398. Webster, J.P.G., 1987. Decision theory and the economics of crop protection measures. In: K.J. Brent and R.K. Austin (Editors), Rational Pesticide Use. Proceedings of the 9th Long Ashton Symposium. Cambridge University Press, Cambridge, 348 pp. Welles, J.M. and Norman, J.M., 1991. Instrument for indirect measurement of canopy architecture. Agron. J., 83: 818-825. Whelan, H.G. and Gaunt, R.E., 1990. Yield loss:disease relationships in barley crops with different yield potentials. Proceedings of the 43rd NZ Weed and Pest Control Conference 1990, pp. 159-162.

© 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

87

Simulating light regime and intercrop yields in coconut based farming systems J. Dauzat a'*, M.N. Eroy b aCIRAD/GERDATPlant Modeling Unit, P.O. Box 5035, Montpellier, France bDavao Research Center~Philippines Coconut Authority, Philippines, Philippines Accepted 16 June 1997

Abstract Intercropping experiments of corn and mungbean have been conducted at the Davao Research Center of the Philippines Coconut Authority under coconut stands at different densities. Yields obtained in these experiments are more or less linear functions of the photosynthetically active radiation measured under the trees. In order to extrapolate these results for other palm ages and densities, the following steps have been achieved: (1) measurement and modeling of the architecture of 5, 20 and 40 year old palms, (2) generation of virtual coconut stands, (3) simulation of light transmission using these virtual stands, (4) prediction of intercrop yields by combining the results of intercropping experiments and the simulated light transmission. The simulated light transmission under 5, 20 and 40 year old coconut stands were close enough to field measurements to consider that both computerized coconut mock-ups and radiative models are valid. Radiative simulation experiments could thus be performed in order to assess the effect of coconut density on photosynthetically active radiation (PAR) transmission as well as the effect of frond pruning. Results exhibit a nearly linear relationship between light transmission and tree density. Pruning also appears as an effective mean of increasing the light permeability of coconut stands. These results are interpreted in terms of corn and mungbean yields by combining radiative simulations and field intercropping experiments. © 1997 Elsevier Science B.V. Keywords: Plant architecture; Radiative transfers modeling; Coconut; Intercropping; Cocos nuciferal

1. Introduction An accurate modeling of the photosynthetically active radiation (PAR) regime is essential to predict the behavior of intercrops in agroforestry systems such as coconut based farming systems (CBFS) where PAR received by intercrops is commonly 1/4 to 1/3 of the PAR in open field. Intercropping experiments under coconuts in the Philippines demonstrated * Corresponding author.

that, in the absence of strong water deficit and with a proper fertilization supply, the intercrop yields are more or less linearly related to the available PAR (B6nard et al., 1996). Thus optimizing CBFS can be achieved mainly through the choice of coconut density or by frond pruning in order to get sufficient light for intercropping. Few coconut density trials exist because they are lengthy and expensive. Moreover, the density is not the only factor to be taken into account: the development of palms, their planting pattern and the radiative

Reprinted from the European Journal of Agronomy 7 (1997) 63-74

88

conditions affect the intercrops potentialities. A radiative model accounting for all these factors is thus essential for understanding and optimizing CBFS. Classical radiative modeling represents plants as simple shapes (e.g. spheres, cones, cylinders...) without taking into account the actual plant geometry inside these shapes (Brown and Pandolfo, 1969; Chiapale, 1975; Charles-Edwards and Thorpe, 1976; Li and Strahler, 1985; Riou et al., 1989) or uses global statistical canopy properties but disregard spatial arrangement between plant items (Kimes, 1984; Sinoquet, 1989). Recent 'architectural' models such as the coconut models used in this study offer a much more realistic representation of plants because they are based on their botanical description, taking into account the precise shape of plant organs as well as their spatial or geometrical organization in threedimensional space (Reffye et al., 1988; Goel et al., 1991; Aries et al., 1993; Dauzat, 1995). The recent development of software generating realistic threedimensional models of plants opens new possibilities for the radiative transfer modeling. This elicited the interest of the Plant Modeling Unit of CIRAD to develop a specific radiative software exploiting the three-dimensional information attached to the computerized plant mock-ups. Initial studies on oil palm and coconut in Ivory Coast (Girard, 1992; Dauzat, 1994) showed that radiative climate can be assessed acutely on these computer models and plant architecture variations having significant bearing on transmission can be identified.

Moreover, climatic factors, especially the quantity and variation on sky condition, can be assessed. This enables prediction as to radiative climate in a given stand considering its density, age, planting pattern and seasonal fluctuations of radiation. The horizontal distribution of light at the soil level is also assessed as illustrated in Fig. 5.

2. Material and methods 2.1. Experimental site and plant material 2.1.1. The site Field experiments have been conducted at the Davao Research Center (DRC; 07 ° 05 N, 125 ° 57' E) of the Philippines Coconut Authority (PCA). The climate is characterized by average annual rainfall of 2400 mrn/year fairly well distributed throughout the year, a relative humidity of 73-82%, a mean temperature of 27°C and annual sunshine duration of 2350 h. The gently sloping soils are well drained and their average composition is 28% sand, 31% silt and 41% clay. The pH (H20) is 6.6. 2.1.2. The coconut stands Tree description and radiative measurements have been done within three stands of LAGUNA TALL coconuts. The first stand is composed of 5 year old trees planted in a 9 x 9m triangular pattern with rows oriented North-South. The second is composed of 20

Table 1 Corn and mungbean varieties used in intercropping campaigns at the PCA Davao Research Center Crop

lntercropping campaigns 1st

2nd

3rd

4th

Corn

USM var. 6 USM var. 2 SMC 357

Mungbean

Pag-asa 7 Pag-asa 3 BPI Mg 7 BPI Mg 9 BPI Mg 60 BPI glabrous 3

USM var. 6 USM var. 2 USM var. 10 SMC 357 IPB H921 P3246 Pag-asa 7 Pag-asa 3 BPI Mg 9 Candelaria Local, Davao Sariaya

USM var. 6 USM var. 5 USM var. 8 IPB H921 P3246 XOF 62 Pag-asa 7 Pag-asa 3 BPI Mg 9 Candelaria Local, Davao Sariaya

USM var.6 USM var. 2 USM var. 5 P3246 CPX 3007 P3022 Pag-asa 3 BPI Mg 7 BPI glabrous 3 BPI Mg 9 PAEC 5 Local, Bansalan

89

year old trees planted with the same pattern as above. The third stand was planted 40 years ago in square design, with a 8 x 8m spacing. Rational felling of certain palms within the 20 year old stand 1 m above the ground created a density and lighting gradient which determined the intercropping treatments (Fig. 7): • • • •

rows, leaving a free corridor either side of the palm rows. They received the fertilizers and phytosanitary treatments usually practiced in the region. Different combinations of varieties were tested for each campaign as indicated in Table 1. More details about these intercropping trials are given in B~nard et al. (1996).

2.2. Description of coconut stands

(L1) standard interrow (control) (L2) standard interrow with greater lateral lighting (L3) thinned interrow (L4) very thinned interrow

The stand description included the description of the individual trunks (diameter, height, projection and azimuth; see Fig. l) as well as of the number of green fronds per tree. In order to assess the inter-trees variability, 50 palms were sampled within the neighborhood of the plots used for radiative measurements for each age group. Ten palms were sampled to get the phyllotaxic angle from leaf scars along the trunk. The value was controlled later on other trees by angle measurement between fronds of rank 9 and 14. The frond length (petiole and rachis) was measured

2.1.3. The intercrops Four intercropping campaigns have been practiced for corn and for mungbean (Mungo radiata) within the thinned 20 year coconut stand between 1991 and 1994. The land was tilled prior to sowing with a disc plough and ridged. The crops were planted in NorthSouth strips down the middle of the coconut inter-

level of ,~q lower frond

I t 1 I I I t I I

diameter

I •

= i

0

I ! i I

~\e'*, / ~ - r a c h i s . ~

I 1 t t I I I

an~]le i J

"trunk projection"

Fig. 1. Definition of some coconut architectural parameters.

/e-

90 on 100 dry fronds from different trees for each age group. In addition, to study intra-tree variability, all the dry fronds produced during a year have been measured on five trees per group. The frond inclination at the junction of petiole and rachis was measured on the most number of leaves on five trees per age group using an electronical clinometer. In order to model frond curvature, the height of some points along the rachis was measured. All accessible fronds (i.e. fronds of rank above 6 or 7) of 10 young palms (under 10 years old) were sampled. The leaflet number was counted on both sides of three fronds taken from the five trees selected per age group. This counting was done on 50 cm long sections of rachis in order to assess the spacing of the leaflets. Three unbroken leaflets were taken on each side of the 50 cm sections in order to measure their length, width and surface area. These measurements were obtained by an opto-electronical planimeter (LICOR leaf area meter) after masking the gaps within the lamina if ever. The vertical and horizontal angles of leaflet insertion (see Fig. 1) were measured on photos of 10 cm long segments of rachis. All the fronds of an apparent phyllotaxic spire were taken from two trees of each age group. Segments were then cut approximately every 40 cm for photos.

2.3. Modeling of coconut architecture The morphogenetic growth pattern of coconut adheres to Corner's architectural model (Hall6 et al., 1978), which is characterized by the existence of a single leaf axis with lateral inflorescence like in oil palm and papaya. Thus the description of the plant topology is quite simple, but an accurate geometrical description is needed for our purpose. Some palm features were conveniently characterized by a mean and a SD, assuming a Gaussian distribution (Dauzat and Eroy, 1995). It is the case for the parameters of the trunk (height, diameter, projection, azimuth), the frond number and their phyllotaxy, the number of leaflets on each side of the fronds. Other features have been fitted with simple functions (Dauzat and Eroy, 1995). For instance, a power function was used to fit the frond inclination at petiole end against the frond rank, a quadratic function to fit the leaflet spacing against their rank and a sinusoidal function to fit the leaflet length against their position on the rachis. The leaflet area has been fitted with a power function of their position on the rachis. The frond curvature is modeled as a flexing beam subjected to gravity by a sub-program. The two parameters of this sub-program (the Young modulus and

Fig. 2. Simulated mock-upsof 20 (unpruned~pruned),5 (unpruned) and 40 year old (unpruned/pruned)coconut trees.

91 •

•

•

• •

•

• I ,': i ~iei iiii~ii~iiiiiiiiiiiiii~iieiii~iiiiiiiiiiiiiii!iiiiiiiiiiiiiiii~i I

•

•

•

•

•

•

6i!i!ilili!iiiiiiiiiii!iiiii!jiiiiiiiiiiiiiii!ii!iiiiiiii!iiiiii!iijiiiij •••:ii!!••t::!!!!!i!!•••ii]i !•ii •!!•!!!ii !•ii•!•ii•i i•!i!•!i!i i i!i !i!i!!i!i!i !• O O I:~'~,!:~eiiiiiiii:iliti~ili iiiiilO~ii!ii~iii~iiiiiiiiii!iliiiii!i!i!ii !

O

!ii!iiiiiiiiiiii!iiiiiiiiiiii!iiii!i!ii!ii:iiii!iiiiiiii!iiii•iiiiiiiii!ii!i!iiiii iii!i t i ii i i i! i i iii iiiii iii i ii ii!iiiii! iii ! i ii!i i iioi ii i ii ii i!!i i i i

to::iiii iiiiiiieii!iiiiiiiiii!i!iii!ii!iiii!iiiiii iiiiiiiiii!iiiiiiiiiiiiil • • J": ,:::' ~O ~!~!~!!!!iiiiiiliQiiiiiiiiiiii~i!i~ii!iiiiiiiiiii!Qiii] /:e~ ~: i!J!ii!iii!ii!!ii!iiii!iiiiiiiiilJliiiiiiiiiiiiiiiiiii!i~il !• • I:_: ::!

O

iiii;i;iiiiii:i!ii!iiii)iiii :iii i iiiiiii !iiiiii ;?iiiiiiii!!i iiiiiiiiiiii!iiiiii o i!ii!;i!iiiii:iiiiii; !iii !i!i

O

:i!iiii!iiiiii!i,iiiiiiiiiiii,iiiiiiiiii:ii?i!ii?iiiii:i:,iiiii o i::i,ii!iiiiii

•

•

lei i!i !i i !i !!!ii iiiiiiiii!!!ii!i!i!iiiii!iiiiiljiliiiiii!i!iiiii!!iiiii!iiiiiii !i!! I•

O

O t:: i

:iii~i~ii:i~ii:~:ii~.i!ii~ii:ii~iii!iiiii~:~.i:i~iii~:!.:~:ii:~;~ii~ii:i~i~:~i!?:.i~:i!igi:i~i~.~:i~i~.i~i.:~i@.~ii:i~.i:i!i~i:,!~di!;i~ii!!iii

:el;ii!i!i~!!?iiiiiiiiiiii!ili!;i;!iiiiiiii!iiiiiiiii!iiiiii!iijiiioil •

•

O

•

•

•

•

•

•

•

•

Fig. 3. Simulated scenes with triangular design (left) and square design (right). Circles indicate coconut positions and inner rectangle represent the effective area used for radiative simulations (assuming that the stand portion within this area is surrounded by identical stand portions all around). The triangular design was used with a distance between trees of 8.5, 9, 10 and 11 m and the square design was used with a distance between trees of 8, 8.5, 9 and 10 m.

meters, like the frond length or the number of leaflets, are given with the intra-tree variability. One parameter file was created for each age group. In order to test the effect of pruning on transmitted radiation we also simulated pruned trees using the same parameter files but limiting the number of fronds to 18 (Fig. 2). To analyze the effect of the planting patterns and of the tree density, we created scenes with square and triangular designs, at different spacing (Fig. 3). Scenes in square design had 56 trees with eight rows of seven coconuts. Four densities were used, 100, 123, 138 and 156/ha, corresponding to distances between trees of 10, 9, 8.5 and 8, respectively. Four

the conicity) have been fitted using a specific interactive program. 2.4. Simulation of virtual coconut stands The coconut generator program is a software which computes the tree geometry through the functions chosen for modeling. The palm generation is stochastic, i.e. restitutes the observed tree variability (Reffye et al., 1995). Thus the input parameters are specified with their variability. Some parameters pertain to the global features of the observed population with the inter-tree variability: phyllotaxic angle, height and diameter of trunk, number of fronds. Other para•

•

•

•

0

•

.... ~

•

iii! il ii ::~!!i~~: ::~::: .... ......

O

::oii:~::!::i i::i~il i iiiiilil il i~iiiii!i~iii:,ilitiiii • '::~ :' ~iiii!i.i! iii:i!i;iii!ili!!ii!ii.....~,ii~,~e', o::: t:!iii!Oi;;ii iii.:iiii.~ii!.i.i!iii.li~ •

:~~,ii:i;~e i~ i~i~i~i :',:iQ:,::ii ~i:,:si~eiiiiiiii;~i~iii; : i;;:~c.,.:!:1

fO!!!ili~s~t::!iiiii~Oiii:.ii~!~i~i i i:iOi!i!~ii !!il2::¢!:!ii ~i i i i:.i~:i:i i~it:i ,~!~

O

O O

~,~o,~: , i ..~eiil;.i~..~i!i .li ~'i !i i......... i!','~

0

%..:

O

0

iiii :::i:~i:::i:i::~":!ii!::iii:::!i:i:i: O

~.:!:::i:~:~:eii.i :!;!]:;i!.L.ii !i:i!.O!.i!:i.i:.ii£1~i!:!i!i;i :O!ii:] ~e:~i:i!,!~i~:;i~!:,:!i ! ,~e!i!i!i i i'.i:!.i:i~i !i :i~!:!:i~!iO!ii•!i i i i:i!!:i!!!;i:!

0

:el ~:~ i i! :~e~,i!iliiiili ii!i iiiii~ili~iiiiiiiii • •~::~:i~,:::?,:~iei i lf:iii!i!~ii!i.i!iilji~iiiiiiiii!ii!!ii!i@i!iiii i !i!!i~:!i

O

• .. 1:...

:i:.:

... :..:..:..: :::::::::::::::::::::::::::::::::::::::::::::::::::::::::

:.:

i~:i::Si:.?!`!:::i:!i!]:i:i~if~!i!~]!?i!]:!i.::~!!?i]fi!ii~!i!ii~!!]i!i!~i~fi!]ii!]i):~!i!iii!]i:]i!i!i!~ii[i:~!~.~!:

Oi,: ~::::!:',::i::!ilt:::i~.iO ':: !~i~i !tl! ;!ti,:ii £iti':lii• i:,'~i:i', 'tti, 'i'::!i',! • 0

f"

""

:.o! t;t:~::~ii i!~ili !6i!!!i!! i!ii! i ~ii~,ii i i ! t~t o ! !! ii eiii~i!!i','~i i!~',i !i i i~i i!i!!~!!iiii!~i!!~!!~'~1!

!.:

:!iii!iii:i:ii-:ii:i!iiiiiiiii:i.i!~i!!!:iiiii!~ii:i!:iiii!]iii!iiiii~!il

...... :iiiii!iiiiii!:!iiii!iiiii~:i!!i::i!i

...... :!:::!iiiii::iii!'!i? i!i'i!:

.i:;i~Tq:-;i::!;:;.:~ii;~i:i:f;::; i::i~:!:ii; ::i:-ii!!!!i[!i!i t::i~:i!~i. !::;i!!;~ ~if!~ i!.:~i~.~i~!if~i:~t.ii;;!::~!;i!i!~: /: i!ii.!;.:h~!.::;.:~;!1

0 ~i!:i!~~i~:.ii!~i~!ii~!0ii~i~i~!~]:!~i~ii!!ti~!!!~i~!~i!~ii~f:iii~i~i~i~i!i~.ii~:i0!~!~!ii~i::i~i~:]i~:.~::i~::~:~ii]:~ •

0

•

0

•

Fig. 4. Simulated thinned stands. Same as legend in Fig. 4 with felled trees represented by empty circles. Thinning original stands at 143 trees/ ha leads to densities of 107 (left) and 72 trees/ha (right).

92

Fig. 5. Simulation of the light transmitted under a 20 year old coconut stand at density 143 during a day. rows of seven coconuts and four rows of six coconuts were included in scenes with triangular planting. Four densities were used: 81, 115, 143 and 160 trees/ha, corresponding to planting distances of 11, 10, 9 and 8.5 m, respectively. Three other designs were simulated by removing some palms from the triangular design at density 143. The first one represents the DRC experiment (Fig. 7). The other two result from a thinning down to 107 and 72 palms/ha (Fig. 4). The same computerized trees are used for simulating stands at a given age, assuming that the coconut density and planting design do not deeply affect the tree development. Data previously collected in a coconut density experiment in C6te d'Ivoire with densities ranging from 115 to 180 trees/ha showed that this statement is acceptable (Girard, 1992). 2.5. Radiative measurements

In characterizing the radiative conditions in CBFS,

the main concern is to assess the quantity of transmitted PAR, the part of the solar spectrum used by the chloroplasts for photosynthetic conversion. The transmitted PAR under the 5, 20 and 40 year old coconut stands was measured by a set of quantum sensors at the soil level in the absence of intercrops while the incident radiation was recorded by a reference sensor placed above the canopy or in open ground. The sensors used for measuring the transmitted radiation were manufactured by the CIRAD because of their low cost as compared to the commercial sensors. They were made with amorphous silicon cells, so-called SLAMs, which have a spectral sensitivity within the 400-700 nm waveband, though a very slight overlapping with adjacent wavebands (UV and NIR) occurs. The cells were equipped with a precision resistor. The whole was sprayed with a waterproof varnish and inserted in a black case made within a nylon bar. A white cover acting as a diffuser was placed above the cell. A commercial sensor was used as a reference for the calibration of these sensors.

93

To determine not only the mean transmission rates but also to map the distribution of light, the protocol called for the use of 32 SLAM sensors placed in two adjacent elementary triangles n. The sensors were connected to a Delta-T logger programmed to read the signals every 5 s and to integrate them every 5 min. Each of the three stands had at least 3 days of continuous logging from approximately 0600 until 1700 h. 2.6. Simulation of radiative climate under coconut stands

The software developed split the sky hemisphere in 46 sectors according to the Den Dulk's 'TURTLE' model (Den Dulk, 1989). The quantity of PAR incoming from each direction is calculated in two steps: •

•

First direct and diffuse components of global radiation are calculated from the ratio of global radiation on extra-terrestrial radiation using de Jong formulas (cited by Spitters et al., 1986). The distribution of diffuse radiation within the 46 sectors is then calculated by combining the formulas of Dogniaux (1973) describing the brightness of a clear sky and of Anderson (1966) describing the brightness of a standard overcast sky. Instantaneous direct radiation is distributed into three neighboring sectors according to the spherical distance of their center to the sun direction.

The sector brightness can be calculated for a given moment or integrated on several hours or several days with a 30 mn time step. Radiative transfers within the canopy are simulated by two programs developed at the CIRAD Plant Modeling Unit (to be published): for each of the 46 sectors defined: •

The MIR program calculates the interception of the incident radiation by the vegetation elements and by the soil. Light interception can be output for each canopy element (e.g. each leaflet), for element classes (e.g. for fronds according to their rank) or for indivi-

. ,

i An elementaryMangle is the space delimitedby three neighboring trees.

•

dual plant. A map of the radiation reaching the soil can be obtained (Fig. 5). The MUSC program calculates the multiple scattering of the intercepted light at different levels of the canopy, i.e. the mutual lighting between the soil and horizontal layers of vegetation. Resulting light interception is output for each layer defined by the user and for the soil.

These two complementary programs provide a detailed radiative balance of the canopy illuminated from each direction. The RADBAL module then combines the results obtained for the different directions according to the instantaneous or integrated sector brightness. Owing to this procedure, the directional radiative exchanges have to be calculated only once and the total radiative balance can then be obtained rapidly for any radiative condition (Rapidel, 1995).

3. Results 3.1. Radiative simulations 3.1.1. In situ measurement and validation of radiative simulations The measured rate of light transmission substantially varies with age (Fig. 6). This results from a crown development (frond length, declination and number of fronds) increasing from early stages to reach its maximum on the 15th year and decreasing gradually beyond 30 years. Simulations have been run for coconut stands with bare soil, i.e. without intercrop. In the absence of data about soil and coconut leaf optical properties in the PAR domain, plausible values have been tested:

•

10, 20 and 25% for the leaf reflection and transmission coefficients (assumed to be equal)

•

5, 10 and 15% for the soil reflection coefficient.

Simulations showed that PAR at soil level depends very slightly on the chosen optical properties (Dauzat and Eroy, 1995); actually, the light fraction impinging on the soil without interception by the vegetation is much more important than the fraction scattered by

94

field data (Fig. 6, left). The diurnal evolution of PAR transmission is also correctly simulated (Fig. 6, right). Likewise, the radiative model was able to simulate the transmitted light in the complex situation of the DRC experiment (Fig. 7). The simulated values for the different treatments are quite close to the measured ones. All these results are satisfactory enough to consider that both the coconut models and the radiative models are valid and can be used for radiative simulation experiments.

50 v

c

.o .~

E

40 ! 30

m r-

20

"" IZ

10

< n

0 40

20

5

age groups measured

3.1.2. Radiative simulation experiments PAR transmission under 20 and 40 year old coconut stands has been simulated on wide range of tree density and with different planting patterns as presented in Section 2.4. For a given age, the PAR transmission is linearly related to coconut density, irrespective of the planting design (Fig. 8). The regressions of %PAR transmission vs. density give:

l l simulated

600 X.

X.

X.

'7

ffl

o

x

400t

... for 20 year old stands /X

E

=L

%PAR = 79.97-0.386 density (R 2 = 0.996)

200

%PAR=85.79-0.312 density (R 2 = 0.984)

IZ

x x.

0

. !

5

7

.

.

.

.

.

... for 40 year old stands

x

r

,

l

!

9

11

13

15

17

time of day (h) x

measured

......... simulated

Fig. 6. Upper, Comparisonof simulatedand measuredPAR transmission under coconuts for different age groups. Lower, Daily evolution of the PAR transmitted under a 20 year old coconut stand.

the vegetation toward the soil. Thus we chose arbitrarily a reflection-transmission coefficient of 20% for the leaves and a reflection coefficient of 10% for the soil. One could expect that the presence of an intercrop with an albedo higher than the albedo of the soil would modify the radiative balance of the canopy. In practice this modification is negligible: the presence of an intercrop with an albedo of 20% would barely increase the downward PAR radiation under coconuts by about 1%. The PAR transmission has been simulated for 5, 20 and 40 year old observed stands on the whole period of in situ radiative measurements. The average simulated PAR transmission is in good agreement with

It appears that removing some fronds of the trees (to limit their number at 18) can increase the light penetration by about 25-40%. As a result, the quantities of PAR transmitted under 20 year old pruned stands with densities of 143 and 156 palms/ha are more or less comparable to those obtained under unpruned palms having densities of 95 and 100 palms/ha. Enhancement of light transmission is less marked for 40 year old than for 20 year old coconuts.

3.2. Corn and mungbean yield predictions The corn and mungbean yields largely differed from one intercropping campaign to another (Fig. 9). This can be mainly attributed to the varieties grown and, to a lesser extent, to temporary water logging (BEnard et al., 1996). Besides these differences, it can be noticed that the yield response for both crops is more or less a linear function of the PAR received. In order to simulate the expected yields of corn and mungbean grown under coconuts at different densi-

95

ties, we simulated first the PAR available under coconuts. The yields were then obtained by interpolation using the experimental yield responses. For corn under 20 year old unpruned coconuts, decreasing the tree density to 50%, may mean doubling or tripling the yield (Fig. 10). The yield response of mungbean to the density is proportionally smaller and varies among the campaigns due to the shade tolerance of the varieties (Fig. 11). The pruning of the palms has a drastic effect on the corn and mungbean yield. Globally, the density and pruning effects 0

@

,L'~

©

•

60

OC:: 50 tx . <_... thinned stands,, o~

I

40 t 30 t

e,. tw

n<

0

®

0

0

0

0

0

®

0

70

600

L4

L4

L3

.

.

.

.

.

L2

~'

90

"11

13

"A. "-.I

t

l

110

!

i

130

!

-

150

Coconut density (kees ha "1 )

6O

L1

•

,

pruned trees El

&I

V

c0 50 I/) E 40] c

..,.,

11. "I.

"If.& • triangular design I square design

20

L3

&

,...,.

i'-=. 1

*-. 30

L2

"" X

20

A

L1

",.*

•

@

0

pruned trees

"-.

•

90

]

1'

"1

110 130 150 170 Coconut density (trees ha 4 )

Fig. 8. Simulated PAR transmission vs. density under 20 (upper) and 40 year old (lower) coconut stands.

/

.'7"

-'o

~n

E

are smaller for 40 year old stands than for 20 year old stands; because the light permeability is higher for younger trees, the competition for light is less intensive than for older ones.

." L4 400

.'6L3 t

i..i..

200

/

/

""" L1

4. Discussion and conclusions

/

"0 /

(u (r~

//

0

,,

!

T

-

-

200 400 600 Simulated PAR (I~ mol m"2 s"~) Fig. 7. Upper, Layout of the DRC intercropping layout. Felled palms are represented by empty circles and remaining palms by filled circles. Grey strips indicate the areas used for intercropping experiments. Treatments L I-IA correspond to lighting gradients resulting from local stand thinning. Lower, Simulated and measured PAR values obtained for the different treatments (average daily PAR).

An original approach involving a coconut generator and specific radiative models was used to predict the radiative climate under coconut stands at different ages and densities. A good agreement was obtained between the simulated PAR and the in situ measurements and the small discrepancies may result from the variability existing under the stand. We can thus consider that both the simulated coconut mock-ups and the radiative models MIR and MUSC are valid. Actually, the work required to achieve such a study was heavy as compared to the work required by clas-

96

sical approaches which rely on a much more simple description of the vegetation; the knowledge of the leaf area index, for instance, can be sufficient to run the Monsi and Saeki's model. In return these simple models have a restricted field of application. Despite such models can be calibrated 2 in order to simulate the actual light transmission rate under a given stand, the extrapolation for other stands, e.g. with other densities, would be risky. The van Kraalingen's model, representing the crown of oil palm trees as a set of panels radially disposed within an hemisphere at the top of the trunk (van Kraalingen et al., 1989), is much more realistic. Nevertheless the validity of this model for different crown geometry (sometimes more sphe-

rical than hemispherical) is to be checked and its adaptation to coconut would, at least, need calibration. When using detailed mock-ups, no calibration is needed and the effect of any change in crown geometry can readily be assessed. Furthermore the radiative models MIR and MUSC are independent of the stand constitution; any planting pattern and density can be tested and the inter-trees variability is taken into account. Therefore the use of computerized coconut mock-ups ensures that the radiative simulations will remain valid for any density as long as the tree architecture is not deeply modified 3. This is quite important in so far as the radiative simulations will lead to recommendations for planting densities and agronomic practices. Measurements and simulations exhibited large differences between PAR transmission under coconut stands according to their age. Thus, practically, the management of intercropping must account the dynamic of growth of the coconuts and the search for long term optimal solutions must integrate the whole duration of the coconut stand life. One major result obtained by simulations is that the PAR transmission under coconuts at a given age is sensibly a linear function of the tree density, irrespective of their planting pattern. It is thus possible to adjust the tree density according to PAR requirement of intercrops. Because the light transmission is similar in triangular and square designs, the choice of a coconut planting design may be guided by practical considerations, e.g. ease in cultural practices like cross plowing in square designs. Further simulations are currently done to analyze the distribution of the transmitted PAR, i.e. the PAR quantity actually available for intercrops. A possible alternative to choosing a lower density for intercropping purposes can be the pruning of the coconuts. The simulations show that limiting the frond numbers to 18 in coconut stands at regular densities is quite effective to enhance the light transmission. Pruning seems a very flexible and cost-effective means to modify the light competition in a coconut based farming system. It can be used in existing stands to obtain the desired quantity of radiation suitable to a certain intercrop independent of palm density. If it

2For instance the extinction coefficient of the exponential Monsi-Saeki's model can be derived from radiative measurements.

3 Previous observations in a coconut density trial in C6te d'Ivoire supported this statement (Girard, 1992).

L4 ~

4

...

3

-,

L3

v

E

0o

1

i I

,

250

-

350

T

,

450

.---r

550

P A R ( i~ moles m -2 s -1)

500 "7,

.c

L3y

400

o)

~

"1o

300

L2L1~~_

200

n

L4

100

c •

250

T

-

350

T

r

'

~

450

i

550

P A R ( 11 moles m -2 s -1)

Fig. 9. Experimental response curves of corn and mungbean yield against the average daily PAR value (see Fig. 7 for the explanation of treatments L I-L4). Symbols represent the four intercropping campaigns of corn and mungbean.

97

could be ascertained that the practice had no long term detrimental effects on coconut yield then it could be adopted as one of the cultural management practices in an intercropping system. lntercrop yield prediction can be assessed through the simulation of the PAR transmission as long as PAR remains the more limiting factor. This statement seems to be valid for the DRC intercropping experiments because there is no important water deficit at Davao and the competition for nutrients is minimized through the fertilization. Despite competitions for

° ,oo f .C:

..~

300

"0

I

•

"7

._e

"t _ o 0

-o

"~., 200 C t~

¢~ ~

100 ~',==_ _................

85

-I

v . . . . . . . . .

105

.... Q

==.

,

125

,,

r-

145

165

Coconut density (trees ha -1)

6

A t~ ,.¢::

Q'....

!

I 4-~

N°.°.°

3

"7

400

m t-'

2

3O0 t

v

"o 1

.m,

200

t-

in 0

=--=._...

4

105

85

125

145

165

Coconut density (trees ha -1)

¢::3

100

85

,

,

105

125

•. . . . .

•

145

165

Coconut density (trees ha -1)

Fig. 11. Forecasted mungbean yields under 20 year (upper) and 40 year (lower) old coconuts at different densities. Same symbolsas in Fig. 9 with open symbols representing pruned stands.

Fig. 10. Forecasted corn yields under 20 year (upper) and 40 year

other resources than light can not be discarded, we can assume that predictions remain valid because the experimental corn and mungbean yield responses to the available PAR implicitly integrate these competitions. The generalization of the results presented herein requires further studies on coconut and intercrop materials. This is currently done within the frame of an EEC project 4. Several coconut materials have already been studied, one at different densities in C6te d'Ivoire, one in south Sumatra and three in the Vanuatu. Analysis are presently done in order to assess the minimal set of architectural parameters needed to describe the trees, e.g. number of fronds

(lower) old coconuts at different densities. Same symbolsas in Fig. 9, with open symbols representing pruned stands.

4 Contract TS3-CT92-0132: Coconut Based Farming Systems.

8 "7

m

4

•o

3

E o o

2

._~

.

85

.

.

.

.

~

105

. . . . . . . . . . .

r. . . . . . . . . . . . . .

125

r -

-

145

165

Coconut density (trees ha -1)

98

per tree, frond length, number of leaflets per frond and length of the longer leaflet (default values being taken for other parameters). On the other hand, several intercropping experiments are conducted both in the Vanuatu and the Philippines. In the latter country, the behavior of crops is also studied under artificial shading in order to get their response to the PAR available in the absence of any competition with coconuts. Several varieties are tested because the response to the quantity of PAR varies strongly: the most productive variety in high light situations may not always be the best under deep shade (Brnard et al., 1996).

References Anderson, M.C., 1966. Stand structure and light penetration. II. A theoretical analysis. J. Appl. Ecol., 3: 41-54. Aries, F., Pr6vot, L. and Monestiez, P., 1993. Geometrical canopy modeling in radiation simulation studies. In: C. Varlet-Grancher, R. Bonhomme and H. Sinoquet (Editors), Crop Structure and Light Microclimate, INRA, Versailles, pp. 159-173. B6nard, G., Margate, R.Z., Daniel, C., Eroy, M.N., 1996. Food intercrops under adult coconut palms. In: EEC-STD3-TS3CT92-0132, 3rd technical report, 62 pp. Brown, P.S. and Pandolfo, J.P., 1969. An equivalent obstacle model for the computation of radiation flux in obstructed layers. Agric. Meteorol., 6:407-42 I. Charles-Edwards, D.A. and Thorpe, M.R., 1976. Interception of diffused and direct beam radiation by a hedgerow apple orchard. Ann. Bot., 40:603-613. Chiapale, J.P., 1975. A numerical model for estimating the modification of heat budget introduced by hedges. In: D.A. de Vries and N.H. Afgan (Editors), Heat and Mass Transfer in the Biosphere. Part 1. Transfer Process in the Plant Environment. Proc. of Int. Centre for Heat and Mass Transfer 7th Int. Seminar, Dubrovnic Yugoslavia. Scripta book Co., Washington DC, pp. 457-466. Dauzat, J., 1994. Simulation des6changes radiatifs sur maquettes informatiques de Elaeis guineensis/Radiative transfer simulations on computer models of Elaeis guineensis. Ol6agineux, 49(3): 81-90. Dauzat, J., 1995. Coconut architecture modeling and simulation of the radiative climate under coconut. In: Proc: Seminar on European research working for coconut, Montpellier, France, Sept. 2-6, 1993, pp. 91-99.

Dauzat, J. and Eroy, M.N., 1995. Coconut architecture modeling and radiative climate simulation on computer models. In: EECSTD3-TS3-CT92-0132, 3rd technical report. 54 pp. Den Dulk, J.A., 1989. The interpretation of remote sensing, a feasibility study. Thesis, Wageningen. Dogniaux, R., 1973. Exposition :rnergrtique par ciel serein des parois odentres et inclinres. Donnres d'application pour la Belgique. Inst. Mrtro. Belgique. Ed: Misc., Srde B, 25:153 pp., 26: 86 pp. Girard, M.L., 1992. Climat radiatif sous cocoteraie et architecture des arbres. D.E.A. Report. Ecologie et Production Vrgrtale, option Agronomie. Paris VII, Institut National Agronomique, Paris XI. 28 pp. Goel, N.S., Knox, L.B. and Norman, J.M., 1991. From artificial life to real life: computer simulation of plant growth. Int. J. Gen. Syst., 18 (4): 291-319. Hallr, F., Oldeman, R.A.A. and Tomlinson, P.B., 1978. Tropical Trees and Forests, Springer-Verlag, Berlin, 109 pp. Kimes, D.S., 1984. Modeling the directional reflectance from complete homogeneous vegetation canopies with various leaf-orientation distributions. J. Opt. Soc. Am., 1: 725-737. Li, X. and Strahler, A.H., 1985. Geometrical-optical modeling of a conifer forest canopy. IEEEE Trans. Geosci. Remote Sens., GE23:705-721. Rapidel, B., 1995. Etude exprrimentale et simulation des transferts hydriques dans les plantes individuelles. Application au cafrier (Coffea arabica L.). Ph.D., Universit6 de Montpellier II Sciences et Techniques du Languedoc, 246 pp. Reffye (de), P., Edelin, C., Franqon, J., Jaeger, M. and Puech, C., 1988. Plant models faithful to botanical structure and development. Comput. Graph., 22: 151-158. Reffye (de), P., Houllier, F., Blaise, F., Barthrlrmy, D., Dauzat, J. and Auclair, D., 1995. A model simulating above- and belowground tree architecture with agroforestry applications. Agroforest. Syst., 30: 175-197. Riou, C., Valancogne, C. and Piere, P., 1989. Un modrle simple d'interception du rayonnement solaire par la vigne. Vrrification exprrimentale. Agronomie, 9: 441-450. Sinoquet H., 1989. Modrlisation de l'interception des rayonnements solaires dans une culture en rangs. I. Aspect throriques. Agronomie, 9: 125-135. Spitters, C.J.T., Toussaint, H.A.J.M. and Goudriaan, J., 1986. Separating the diffuse and direct component of global radiation and its implication for modelling canopy photosynthesis. Part I: components of incoming radiation. Agric. For. Meteorol., 38: 217-229. Van Kraalingen, D.W.G., Breure, C.J. and Spitters, C.J.T., 1989. Simulation of oil palm growth and yield. Agric. For. Meteorol., 46: 227-244.

© 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

99

Improving wheat simulation capabilities in Australia from a cropping systems perspective" water and nitrogen effects on spring wheat in a semi-arid environment Holger Meinke a'b'*, Graeme L. Hammer a, Herman van Keulen c'd, Rudy Rabbinge b, Brian A. Keating e aAgricultural Production Systems Research Unit, DPI/CSIRO/DNR, PO Box 102, Toowoomba, Qld 4350, Australia bDepartment of Theoretical Production Ecology, Wageningen Agricultural University, POB 430, 6700 AK Wageningen, The Netherlands CResearch Institute for Agrobiology and Soil Fertility (AB-DLO) PO Box 14, 6700 AA Wageningen, The Netherlands dAnimal Production Systems, Department of Animal Husbandry, Wageningen Agricultural University, Marijkeweg 40, 6709 PG, Wageningen, The Netherlands eAgricultural Production Systems Research Unit, CSIRO/DPI/DNR, 306 Carmody Road, St Lucia, QM 4063, Australia Accepted 13 April 1997

Abstract Systems approaches can help to evaluate and improve the agronomic and economic viability of nitrogen application in the frequently water-limited environments. This requires a sound understanding of crop physiological processes and well tested simulation models. Thus, this experiment on spring wheat aimed to better quantify water x nitrogen effects on wheat by deriving some key crop physiological parameters that have proven useful in simulating crop growth. For spring wheat grown in Northern Australia under four levels of nitrogen (0 to 360 kg N ha-~) and either entirely on stored soil moisture or under full irrigation, kernel yields ranged from 343 to 719 g m-2. Yield increases were strongly associated with increases in kernel number (9150-19950 kernels m-2), indicating the sensitivity of this parameter to water and N availability. Total water extraction under a rain shelter was 240 mm with a maximum extraction depth of 1.5 m. A substantial amount of mineral nitrogen available deep in the profile (below 0.9 m) was taken up by the crop. This was the source of nitrogen uptake observed after anthesis. Under dry conditions this late uptake accounted for approximately 50% of total nitrogen uptake and resulted in high (>2%) kernel nitrogen percentages even when no nitrogen was applied. Anthesis LAI values under sub-optimal water supply were reduced by 63% and under sub-optimal nitrogen supply by 50%. Radiation use efficiency (RUE) based on total incident short-wave radiation was 1.34 g MT t and did not differ among treatments. The conservative nature of RUE was the result of the crop reducing leaf area rather than leaf nitrogen content (which would have affected photosynthetic activity) under these moderate levels of nitrogen limitation. The transpiration efficiency coefficient was also conservative and averaged 4.7 Pa in the dry treatments. Kernel nitrogen percentage varied from 2.08 to 2.42%. The study provides a data set and a basis to consider ways to improve simulation capabilities of water and nitrogen effects on spring wheat. © 1997 Elsevier Science B.V.

Keywords: Spring wheat; Growth; Water; Nitrogen; Agricultural systems; Model

1. Introduction * Corresponding author. Tel: +61 76 881378; fax: +61 76 881193; e-mail: [email protected]

Reprinted from the European Journal of Agronomy 7 (1997) 75-88

Production systems modelling can be used to

100

answer questions at various levels of aggregation. However, modelling should not be seen as the panacea for all agricultural problems but rather as a convenient way of aggregating environmental interactions thus providing higher level data upon which decisions can be based. It integrates our knowledge of agricultural systems, allows generation of information useful to systems managers (e.g. What if? When? How often?) and highlights gaps in current understanding of the system. It is a means of making agricultural research more relevant to practice by adding value to existing knowledge and research efforts. Simulating the production system, the state of the system at any point in time is known, and alternative management options and their long-term impact on sustainability and productivity can be evaluated. Crop models form the basis of production systems models, but have rarely been developed specifically for inclusion into systems models. Mostly, emphasis has been on the prediction of yield rather than resource utilization (i.e. water and nitrogen). In light of the considerable research effort spent on existing models, better and more encompassing predictive performance of crop models and their components is more likely achieved by improving these models than by developing new ones. Such model performance evaluation requires data sets containing the necessary details for model initialisation. In north-eastern Australia, spring wheat is a major component of the dryland cropping system. Production varies strongly from season to season caused by extremely high rainfall variability in this region (Stone et al., 1996). In low rainfall years, crop yields are directly related to the amount of stored soil moisture, but in high rainfall years, nitrogen availability often limits production, unless fertilizer is applied. Simulation studies can be applied to analyse such a highly variable system and devise management strategies that result in overall better crop peformance, but also increase productivity of the whole cropping system (McCown et al., 1996). Thus, the objective of this series of papers was to develop a simulation capability for spring wheat that • •

is suitable for use in cropping systems models, is robust in its predictive ability across a wide

range of environmental conditions in Australia, and does not require parameters that are difficult to derive from experimental data. The first part of this series presents experimental data that provides a detailed data set for model testing. Subsequently, existing spring wheat simulation models are evaluated using this and other data sets from various agro-climatic regions (Meinke et al., 1997a). In the final part of this series, a new, integrated spring wheat model is introduced (Meinke et al., 1997b). It draws strongly on simulation approaches that have shown to perform well and that represent crop physiological processes in a simple, general way and are therefore more likely to give the right answers for the right reasons. Specifically, the objectives for this paper were: •

•

• •

to determine and quantify the effect of water and nitrogen limitations on wheat by using a physiological framework defining determinants of crop growth and yield to identify generic factors and concepts where possible, from a field experiment derive necessary coefficients to simulate the growth of wheat using a dynamic model of low to intermediate complexity. to compare the derived coefficients with values reported elsewhere and to generate a detailed data set suitable for use in comparative evaluation and improvement of existing wheat simulation approaches.

2. Materials and methods

2.1. Site specifications and agronomic details Spring wheat (cv. Hartog) was grown either under irrigation (irr, applied weekly based on evapo-transpiration estimates) or entirely on stored soil moisture (dry) during the winter of 1993 in Queensland, Australia (27°34'S, 151°52'E). Four levels of nitrogen, termed here as ON, 40N, 120N and 360N (in kg ha-l), were applied to a wheat crop grown on a uniform, alluvial, heavy cracking clay (Ug 5.24; Northcote, 1979) with 240 mm plant available water holding

101

capacity (PAWC) based on the maximum extraction depth of 1.5 m. To create a nitrogen responsive soil environment, three cover crops were grown on the site in succession prior to the experiment without N application. Rainfall was excluded from the dryland site using an automatic rain shelter covered with clear plastic. Each nitrogen treatment was replicated twice and plot sizes were 3.75 x 2.25 m in the irrigated and 2.75 x 2.25 m in the dryland area, with a minimum border size of 0.5 m; the areas for destructive samples measured two rows by 0.5 m. Rows had a north-south orientation with a row spacing of 0.25 m. Prior to sowing, the site was irrigated four times with a total amount of 235 mm to replenish soil water reserves. Spring wheat was sown on 24 June at a target population of 100 plants m -2. All plots received a basal fertilizer dressing immediately after sowing, containing trace elements as well as 15 kg ha -! P. Nitrogen fertilizer was broadcast on 6 July at rates of 5, 40 and 120 kg N ha -i to the ON, 40N and 120N treatments, respectively. A dose of 5 kg N ha -i was given to the control treatment (ON) to avoid poor establishment. The largest N application (360N) was split into three doses of 120 kg ha -l each, given on 6 July, 30 July and 9 September (27 days before anthesis). All plots were irrigated (25 mm) after the first nitrogen application. Subsequently, the dryland treatments received no further irrigation. Herbicide was applied at recommended rates to control broadleaved weeds. Soil samples to determine background nitrogen levels and volumetric soil water content were taken prior to sowing, at anthesis and after final harvest.

2.2. Weather data Environmental data were recorded electronically at 5 min intervals throughout the experiment and values integrated to daily data. Daily vapour pressure deficit (VPD, kPa), a measure of atmospheric evaporative demand, is commonly used to calculate crop transpiration efficiency (Sinclair et al., 1983). Tanner and Sinclair (1983) described a method to estimate average daily VPD from daily maximum (Tmax) and minimum temperatures (Tm~n).This method assumes that dew point temperature is always reached at Tmin and uses an empirical parameter, a, to calculate a weighted daily

average VPD (VPDav) from the difference between saturated vapour pressure (Svp) at Tmin and /'max, respectively"

VPDav = a(Svprm~ - Svprmi,)

(1)

The authors report a value of 0.75 but point out that the coefficient a may vary with season and environment. However, data to derive this coefficient are rarely available and generally the value reported by Tanner and Sinclair (1983) is assumed. From the temperature and humidity data, hourly values of VPD were calculated and weighted for the day-time period of crop transpiration by using hourly incident solar radiation. These values were then compared to those obtained from eq. (1).

2.3. Crop data Neutron probe access tubes were installed prior to sowing to a maximum depth of 1.6 rn and measurements were taken at weekly intervals. Sampling dates of the destructive harvests are given in Table 1. The third harvest (70 DAS) was omitted in the dry treatments due to space limitations under the shelter. Plants were partitioned into green leaf, stem and eventually dead leaf and spike. Leaf area was determined using an area meter (Delta-T Devices Ltd.). At anthesis, green leaf area was further segregated into flag leaf, and the rest as equal numbers into 'middle' leaves and 'bottom' leaves. Kernel yield (KY) was determined by threshing spikes taken at final harvest. Kernel number (KN) and kernel weight (KW) were measured by weighing 300 randomly selected seeds from each plot. All plant samples were dried for at least 72 h at 105°C before determining dry weight. Nitrogen content was determined for all samples (dried separately at 80°C) using Kjeldahl digests. Phenological development was recorded by monitoring ten plants and establishing the dates when 50% of plants had reached a particular stage.

2.4. Light interception Tube solarimeters (Delta-T Devices Ltd.) measuring incident total short-wave radiation were installed 25 days after sowing (DAS), including two reference tubes (one above the irrigated and one above the dryland crops). Soladmeters were placed on the soil sur-

102

Table 1 Calendar of events indicating harvest number and date. Event

Date

DAS

Sowing

24/06

0

First harvest Second harvest Third harvest

19/07 12/08 02/09

25 49 70

Fourth harvest Fifth harvest Sixth harvest Seventh harvest Seventh harvest

06/10 19/10 02/11 12111 16/11

104 117 131 141 145

Seventh harvest

19/11

148

Comments Full soil water profile established prior to sowing

Irrigated treatments only 'Anthesis' harvest

Dryland treatments only Irrigated ON, 40N and 120N treatments Irrigated 360N only

face perpendicular to the rows. Drought conditions prevailed throughout winter/spring of 1993 and the shelter was used only occasionally, mainly when irrigation was applied on the adjacent plots. Cumulative incident solar radiation differed by less than 8% between the two areas. Differences in incident radiation have been taken into account where appropriate. 2.5. Soil water extraction Soil water extraction data were analysed using a framework described in detail by Meinke et al. (1993) for sunflower and Robertson et al. (1993) for sorghum. By fitting continuous functions to measurements of soil water content, cumulative soil water extraction can be calculated for any period. Parameters derived in this way can be used as input into a soil water balance. The framework accounts for maximum plant available water content (MAWC) in each soil layer, the rate at which the soil water extraction front descends through the profile (EFV), the time it takes before the extraction front commences its decent at rate EFV (to) and the rate of water extraction in each soil layer (k0. MAWC is defined as the difference in volumetric soil water content between the lower limit (®L, lowest soil water content recorded in each layer, usually at final harvest) and the drained upper limit (19o). Thus, PAWC equals MAWC if the profile is fully charged at planting. Values for Ou and wet bulk densities were determined from an area

ponded for several weeks in the course of the experiment. With these parameters the time course of soil water extraction for each layer of a soil profile can then be fully described. 2.6. Data analysis Experimental data were analysed using a crop physiological framework that defines the determinants of crop growth and yield (Charles-Edwards et al., 1986). In such a framework, biomass accumulation is defined by either the product of intercepted radiation and its efficiency of use, or by the product of transpiration rate and transpiration efficiency. Kernel yield can be defined either as the product of total biomass and harvest index or as the product of kernel number and kernel size. The development of canopy leaf area is a major determinant in this framework as it controls light interception and transpiration. The influence of environmental factors, such as water and nitrogen availability, must be mediated via these key determinants of crop growth and yield. Data from the shelter and the irrigated area were pooled and statistically analysed as a combined analysis of variance with two factors, namely water and nitrogen. With respect to soil water characteristics, the experimental site was uniform as it varied little in the key parameters of upper and lower soil water content. Although the main effects of water could be tested against the residual term of the ANOVA (Payne, 1993), there is concern about the validity of this test and thus no formal statistical test of the main effect of water treatments has been done. The probabilities from the F-test for main effect of nitrogen (Pn), the water by nitrogen interactions (Pw,) and the associated standard error of differences for the interactions (SEd) were estimated.

3. Results The full data set, suitable for model testing is presented by Meinke (1996) and is available upon request. 3.1. Soil characteristics In all plots, background soil NO3 content increased

103

35

- e - Max Tamp

3O

- e - Min Tamp

m -2 day -~, but average daytime temperature remained relatively constant around 12-14°C until about DOY 260. Thereafter daily average temperatures increased rapidly to values around 20°C but radiation increased more slowly to peak values around 25 MJ m -2 day -I (Fig. l a,b). VPD increased from low values around 0.5 to a peak value of 3 kPa (Fig. lc). VPD estimates using Tanner and Sinclair's (1983) method compared well (R2= 0.84) with those calculated from hourly humidity and temperature data (Fig. 2). The environment can be described as initially mild and conducive to growth, but gradually becoming harsher with hot, dry conditions prevailing during kernel-filling.

2s

i| '°lO is

~

5 0

....

-5

170

~1~

190

210

230

(a)

. . . . . 250

270

290

310

330

DOY

3S

'u ~E 30 ~ 2s

- e - Rad_lrr o

Rad_Dry

~

',

o

]I is ~ lO k.

j

3.3. Crop data

s

o

~

0 170

,

,

,

,

,

l

,

190

210

230

250

270

290

310

DOY

(b) I

120

Rain & irrigation

-e--VPD

~' 100 E

~

2 A t a_

1 v

6o

t3 O. >

_~ 40 Ill a

Under irrigation, 50% anthesis occurred at 104 DAS, except in ON where it occurred at 99 DAS. All dry treatments reached 50% anthesis between 97 and 99 DAS.

330

2O o

0

. 170

j .! L IJ,l,__ 190

(c)

210

230

250

270

290

-1

310

DOY

Fig. 1. Weather conditions throughout the experiment: (a) maximum (closed symbols) and minimum (open symbols) temperature, (b) solar radiation in the irrigated (closed symbols) and dryland (open symbols) treatments and (c) VPD, rainfall and irrigation.

below 0.9 m (average; 0-0.9 m, 3.6; 0.9-1.6 m, 10.6 g N m-2). At anthesis, nitrogen had been extracted to a depth of 1.2-1.5 m in the dry treatment and either used or flushed below the maximum sampling depth under irrigation. Soil organic carbon content ranged from 2.24% at the surface to 0.65% at depth; wet bulk densities ranged from values around one in the top 0.8 m to 1.35 g cm -3 at depth.

3.3.1. Yield and yield components (Table 2) Irrigation increased mean kernel yield (KY) by 50% from 383 (dry) to 573 g m -2 (irr). Within the irrigated treatment, nitrogen application also increased KY, but high variability within the 40N treatment caused by substantial differences in background soil N at depth resulted in a Pwn value of 0.1. Irrigation reduced final harvest index (HI) from an average of 0.42-0.39. Kernel number (KN) increased with irrigation from a mean of 9800-14650 kernels m -2 and with nitrogen application from 10200 to 14950 kernels m -2. Nitrogen application increased KN under irrigation from 9450 (ON) to 19950 kernels m -2 (360N), but had no

i

....

1:1 line

....

R e g r e s s i o n line

3.0

o2O o....

,9,o. 2.0

o~

1.0 o

3.2. Weather Fig. l a-c provides an overview of the prevailing weather conditions during the experiment. Average daily solar radiation increased steadily during the experiment from values below 10 to around 25 MJ

Q o. >

..

..t*

~

o.o 0.0

o ~)

o

-o

R: • 0.84

,

,

,

,

,

,

0.5

1.0

1.5

2.0

2.5

3.0

3.5

A c t u a l weighted daily average V P D (kPa)

Fig. 2. Comparison between actual, weighted average vapour pressure deficit (VPD) and calculated VPD according to Tanner and Sinclair (1983).

c

0

P

Table 2 Yield and yield components at final harvest

DRY

Kernel yield (g m-2) Harvest index (g g-') Kernel number (m-2) Kernel weight (mg K-') Above ground dry matter (m-') Final tiller number (m-2) Kernel nitrogen percentage

IRR

P"

NO

N40

N120

N360

NO

N40

N120

N360

436 0.44 10931 39.9 995 360 2.08

367 0.4 1 9148 40.3 899 318 2.10

343 0.42 9172 37.3 817 238 2.07

387 0.42 9926 39.1 918 330 2.16

427 0.4 1 9448 45.2 1033 492 2.10

603 0.39 I484 1 40.5 1519 472 2.27

542 0.37 14263 38.0 1457 585 2.1 I

719 0.40 19950 36.0 I807 678 2.42

0.25 0.28 0.06 (0.01

0.10 0.46 0.32

PW"

sEd

0.10 0.67 0.02 <0.01 0.03 0.30 0.68

86 0.02 2014 I .2 185 98 0.16

Kernel yield, total above ground dry matter, kernel number, kernel weight, harvest index, final tiller number, and kernel nitrogen percentage are shown. Also shown are and the standard errors of treatment differences (SEd).Values presented in probabilities based on the F-test for the main effect of nitrogen (Po),water by nitrogen interactions (Pwn) bold indicate significance at P < 0.05.

105

2000

o*.to'8.,

1600 E

o

1200

13

OO O y ~ * ~ ' ~ B O ~ J o .

coo

=1

.

.

.

.

.

.

400

< 0

200

400

600

800

1000

1200

1400

Cumulative Intercepted radiation (MJ m "2)

Fig. 3. Correlation between cumulative intercepted total shortwave radiation (MJ m -2) and total above-ground biomass (DM, g m -2) across all treatments. The regression line was forced through the origin (y = 1.34x, R 2 = 0.90).

effect under dry conditions. Under irrigation kernel weight (KW) was reduced by nitrogen application. Final tiller number (FTN) differed among irrigation treatments, but not among nitrogen levels and ranged from an average of 312 tillers m -2 in the dry treatment to 678 in the 360N irr treatment.

3.3.2. Above-ground dry matter Total above-ground dry matter production (DM) differed among water treatments from anthesis to final harvest. Final values also differed for nitrogen levels under irrigation ranging from 1033 to 1807 g m -2.

pared to 1.7 in the dry treatments. There was also a significant effect of nitrogen on LAI. Under irrigation, anthesis LAI in 360N was more than double that of the ON treatment. Anthesis LAI in the dry treatment did not differ significantly across N Levels, although it was lower for the ON treatment. Specific leaf area (SLA, cm 2 g-l) in the dry treatment declined linearly from 208 cm 2 g-i at 25 DAS to 121 cm 2 g-i at 117 DAS with no significant N effects. Under irrigation pre-anthesis SLA increased from 185 (25 and 49 DAS) to 268 cm 2 g-t (70 DAS). Subsequently, values decreased for all nitrogen levels reaching minimum values of 131 in the NO and an average of 170 cm 2 g-i in all other nitrogen treatments at 131 DAS.

3.3.5. Nitrogen uptake and concentrations in plant components Total, above-ground plant nitrogen content continued to increase after anthesis in all treatments (Fig. 5). In most dry treatments, more than 50% of all nitrogen was taken up between 104 and 131 DAS. Under irrigation, N uptake rates were lower after anthesis, but even then the ON treatment accumulated 25% of its total N during kernel-filling. Only the 360 N irr had a similar plant N content at anthesis and at final harvest (22.0 and 24.4 g m -E, respectively). Dryland

-- ~ - - ON

8

=

40N

- -~- - 120 N

3.3.3. Radiation use efficiency Radiation use efficiency (RUE) was calculated for each treatment by fitting regressions of total plot dry matter against cumulative intercepted short wave radiation. Although there was scatter around the regression line, none of the slopes differed significantly and consequently neither nitrogen nor water levels affected RUE significantly. The overall value across treatments was 1.34 g MJ -! (Fig. 3). 3.3.4. Leaf area index Green leaf area index (LAI) at 49 DAS differed across water treatments. Mean LAI in the dry treatments (0.65) was almost double that under irrigation (0.36; Fig. 4). This was caused by higher soil temperatures in the dry treatment accelerating leaf area and crop development. At anthesis (104 DAS), LAI trends were reversed with respect to water treatments. In the irrigated treatments LAI values averaged 4.6 com-

6

--

0

25

50

75

lOO

3e0N

125

15o

125

1S0

DAS

(a) irrigated 6 6

°'°oQ.Qo

I

2 0 0

(b)

25

50

75

100

DAS

Fig. 4. Leaf area development in the dry and irrigated treatments. Vertical bars indicate pooled standard errors.

106

25

kernel N% varied from 2.10 to 2.42% (Table 2). The non-significant result with respect to N treatments was caused by the high variability within the 40N irr treatment where kernel N% varied from 1.98% in replicate one to 2.55% in replicate two, caused by differences in background N levels.

Dryland

"e 2o - -o. - ON i 15 = = 10

= 40N - , , . . ~,ON ---e---3

//'"F

_ J"J." .'

I: -

41 D

Z (

0

0

20

,-4

40

60

80

(a)

120

Irrigated

20 - -o- - ON

~,

.--ii,--120N

i=

10

_-

S

/ ;......... ~ . ,

"-'e-'- 360N

.i,, ° "

/..-~r/,~....

- • 1,-

,L

60

100

120

Z

0 (b)

20

40

80

3.4. Soil water extraction and transpiration efficiency

140

DAS 25

~E .o

100

140

DA$

Fig. 5. Cumulative nitrogen uptake in (a) the irrigated and (b) the dryland treatments. Vertical bars indicate + one standard error.

Green leaf nitrogen concentrations (leaf N%) were around 5% of dry matter for all treatments at 25 DAS and then declined almost linearly to values just above 2% in the dry treatment by 117 DAS. Under irrigation, the 360N treatment had a significantly higher leaf N% (3.5%) at anthesis. This difference was maintained until 131 DAS when the last green leaf was sampled in the irrigated plots and leaf N% values ranged from 1.7 (ON) to 2.7% (360N). Leaf nitrogen profiles determined at anthesis showed that N concentrations varied little among treatments but, as expected, were greater in the upper than in the lower part of the canopy. When green leaf area at anthesis was partitioned into flag leaves, middle leaves and bottom leaves, no treatment effects were found and values averaged 3.19, 2.86 and 2.2% N, respectively. Specific leaf nitrogen (SLN) at 25 DAS ranged from 2.06 to 2.85 g m -2 in the dry treatment and from 2.46 to 3.04 g m -2 under irrigation, respectively. By anthesis SLN values reached minimum values of 1.8 in the case of the dry treatment, 1.3 for ON to 120N irr and 1.6 g m -2 for 360N irr treatment. Kernel nitrogen concentration (kernel N%) did not differ significantly. Within the dry treatment kernel N% varied between 2.08 and 2.16%; under irrigation

Total water extraction in the dry treatment was not influenced by nitrogen treatments and averaged 240 mm (+26 mm). Water extraction was calculated using a framework presented in detail by Meinke et al. (1993). For these calculations we assumed that: (i) soil water in the top 0.1 m was lost to evaporation and (ii) the sphere of influence of the neutron source had a radius of 0.1 m (Fig. 6). Parameter values derived from neutron probe readings corresponded well with those determined from soil sampling at harvest (®L) and in the ponded area (Ou). Average rates of moisture extraction (k]) remained around 0.03 day -I to a depth of 0.8 m and then started to increase to a maximum value of 0.06 day -] at depth. EFV and to values were 1.96 and 21 DAS, respectively (Fig. 7). Daily, calculated rates of water extraction for the dry treatment across all nitrogen levels showed a rapid increase up to about 55 DAS (Fig. 8). At this time the crop became water limited and daily extraction rates declined from values above 4 to about 2.5 mm day -! around anthesis. Transpiration efficiencies (TE, g m -2 mm -]) under dry conditions, calculated at 117 DAS (early kernelfilling) and final harvest, were 4.3 and 3.9, respecVol Water Content (%) 0 o

20

40

60

80

i

|

i

I

2o 4o

5 6o v =

80

Q•

100 120 140

¢/

160

Fig. 6. Drained upper (open symbols) and lower limits (closed symbols) of soil water, averaged across the dry treatment. Horizontal bars indicate + one standard error.

107

Days After Sowing (DA8) o o 20

2o

40 i

i

60 _1

80

100

t

i

m

4o 80

period, Tanner and Sinclair's (1983) method using a value of 0.75 for their coefficient a (eqn (1)) can be used to estimate day-time average VPD in this region.

4.2. Yield and yield components

~100 120 140 160

Fig. 7. Start of soil water extraction in a soil layer (to) versus depth of the layer. The slope of the regression (R 2 = 0.91) represents the extraction front velocity, EFV (1.96 cm day-I), whereas the x-axis intercept represents commencement of extraction front descent, to (21 days); see text for details. Horizontal bars indicate + one standard error.

tively. When these values were corrected for VPD, the transpiration efficiency coefficient for above-ground biomass (TEe, g m -2 mm -j kPa, reduces to units of Pa) was 4.8 and 4.6, respectively, regardless of N treatment.

4. Discussion

4.1. Vapour pressure deficit (VPD) Crop transpiration efficiency (TE) can be derived by dividing the transpiration use efficiency coefficient (TEc), a conservative value for many species, by VPD (Tanner and Sinclair, 1983). Carberry and Bristow (1991) have shown the large impacts that errors in estimated VPD can have on crop simulations. Bristow and Carberry (1991) report that the assumption of minimum temperature equalling dew point temperature is not always valid, particularly under higher evaporative demand and in drier environments. Good agreement between actual and estimated VPD using Tanner and Sinclair's (1983) method was obtained in this experiment, but data were collected only during the winter/spring period, a time of generally low VPD (Fig. l c). It is likely that the accuracy of VPD estimates would decline on hot, dry summer days (Bristow and Carberry, 1991). Such a tendency is indicated by a slope < 1 of the regression in Fig. 2 corresponding to an under-estimation of high VPD values. However, the data suggest that at least for the winter/spring

Yield differences were mainly caused by differences in kernel number (KN). Fischer (1985) points out that much of the environmental yield variation occurring in wheat grown under optimal conditions at various locations is due to differences in KN. The variation could be explained by the amount of intercepted radiation in the 30-day period just prior to anthesis. In a later publication, Fischer (1993) reported a similar effect for wheat grown with different rates of applied nitrogen. At unchanged RUE, greater LAI and hence greater radiation interception lead to larger growth rates and consequently increased biomass production. Fig. 9 illustrates that intercepted radiation in this pre-anthesis period was linearly related to KN in the irrigated treatments. Under dry conditions there was no relationship. Dry treatments did not differ significantly in KN, KW or KY but differed in amounts of intercepted radiation around anthesis (Fig. 9). Woodruff (1983) reported a strong relationship between transpiration (rather than intercepted radiation) and KN for wheat grown in waterlimited environments. In our experiment, daily transpiration rates around anthesis were between 2.5 and 3 mm day -~ for all N levels in the dryland treatment (Fig. 8), explaining why they had similar KN values. The increase in KN with increased nitrogen application under irrigation was associated with a slight, but

;o E

5

3 ¢P

,-

|,

2

0

"10

,

.

50

70

.

. 90

. 110

. 130

150

Days after sowing (OAS)

Fig. 8. Time course of daily soil water extraction averaged for the dry treatment.

108

20000

~E

17000

14000 E ic 11000 E

t

8000 5000 200

250

300

350

400

450

500

Intercepted radiation (M J)

Fig. 9. Relation between cumulative intercepted short wave radiation for the period 30 days prior to anthesis and kernel number. Individual plot data for dry (open symbols) and irrigated (closed symbols)plots are shown. The regressionis for irrigated plots only (y = 71x - 16541; R2 = 0.71). significant linear reduction in kernel weight (KW, Fig. 10). Harvest index (HI) can be increased by either decreasing the proportional vegetative biomass or by increasing sink size relative to vegetative biomass. HI is often conservative for a cultivar but can vary with severity and timing of stress, particularly during time of yield formation (Cooper, 1980). Here, HI varied little (except for a slight reduction under irrigation) because water and nitrogen stresses were gradual in onset and not severe. Most of the response by the crop was mediated by effects on biomass accumulation rather than by its partitioning (Table 2). 4.3. RUE

Although nitrogen and water availability affected growth in many ways, RUE did not vary significantly among treatments. This is in contrast to Green (1987) who reports a quasi linear relation between nitrogen application and RUE for winter and spring wheat grown in the UK. Green's RUE value of 1.48 g MJ -l is approximately 10% larger than observed here and is likely a result of the non-linear relation between RUE and radiation flux density as affected by atmospheric transmission. Hammer and Wright (1994), in a theoretical analysis of RUE in peanut, report RUE increases of up to 0.4 g MJ -] as atmospheric transmission decreased from 0.75 (clear sky) to 0.35 (heavy cloud). This is comparable to the difference in conditions for wheat grown during the dry season in Northern Australia and wheat grown in the

UK. Sinclair et al. (1992) report similar effects for soybean and maize. Some of the smaller RUE values (0.94-1.34) reported by Siddique et al. (1989) were likely the result of more severe water stress limiting dry matter production in some of their experiments. This is supported by their small anthesis LAI. RUE is related to specific leaf nitrogen content (SLN). Sinclair and Horie (1989) suggested that rice and wheat have a similar relationship between SLN and RUE, with RUE reaching maximum values of > 1.3 g MJ -! at SLN values of > 1.6 g N m -2. However, data from the ON irr treatment suggest that even SLN values around 1.3 g m -2 did not reduce RUE. Because SLN remained high throughout the experiment and RUE did not vary with N treatment, the significantly smaller biomass and yield in the low N irrigated treatments must have been solely caused by a reduction in LAI and hence affected growth by reducing light interception and transpiration. 4.4. Leaf area

There is ample evidence that nitrogen shortage affects wheat biomass production first via effects on leaf expansion, leaf number and tillering, before net photosynthesis, and hence RUE, are affected (e.g. Green, 1987; Fischer, 1993). Strong effects of water deficiency on leaf expansion are also well known (e.g. Turner, 1986). Therefore, water and nitrogen deficiencies affect intercepted radiation and transpiration via leaf area development, canopy structure and leaf area duration. These sensitive crop physiological processes

5O Q

E 46 t

E 42

o

[ 38

. "

0

O •

O

• 34 E 3o

'

0

~

i

I

,

i

5000

10000

15000

20000

25000

Kernel number (m "2)

Fig. 10. Relation between kernel number (KN m-2) and kernel weight (KW, mg kernel-t). Closed symbols are for the irrigated and open symbols for the dry treatment. The indicated regression is for the irrigated treatment only (y=-0.0007x +49.5; R2 = 0.58).

109

are responsible for the frequently reported conservative nature of RUE and TEc (cf. Monteith, 1994) and demonstrate the importance of adequate LAI predictions in simulation models. After anthesis, amount and rate of leaf senescence was least in the ON irr treatment (Fig. 4). This was likely due to nitrogen available deep in the profile that became accessible to the crop just prior to anthesis. In essence, tillering and leaf expansion were reduced by low nitrogen levels during the early vegetative phase and when more nitrogen became available just prior to anthesis, leaf area could be maintained longer, as nitrogen demand could be met largely by uptake. The same effect was observed in studies using late nitrogen application (Fischer and Kohn, 1965). In the semi-arid subtropics, where large LAI often cannot be sustained due to low sub-soil moisture and insufficient rain, such an effect is desirable, because it increases the amount of photosynthate produced during kernel-filling, and hence, yield. This shows the value of having nitrogen available deep in the soil profile in this environment (Strong and Cooper, 1980). The same process is presumably largely responsible for the high green leaf nitrogen levels across all treatments at anthesis. In their spring wheat model, Sinclair and Amir (1992) used a critical value of SLN of 0.8 g N m -2 below which leaf area development was affected. In the present experiment, anthesis LAI at ON irr was only 43% of that for 360N irr, although SLN never fell below 1.3 g N m -2. Hence, SLN thresholds affecting leaf area dynamics in wheat need to be better quantified for use in simulation models. 4.5. Water extraction, T and TE in the dry treatment

Values describing the progression of the extraction front (Fig. 7) are in line with those reported elsewhere (Monteith, 1986; Thomas et al., 1995). Generally, values for EFV are considerably smaller for winter than for summer crops. This is related to: (i) lower atmospheric demand and hence less water extraction and (ii) lower crop growth rates due to lower temperatures and solar radiation during winter/spring. Our finding of an increase in the rate of water extraction (k0 below a depth of 0.8 m is in contrast to other work reported (Meinke et al., 1993; Robertson et al., 1993). This can be explained by examining

the time course of daily water extraction (Fig. 8): During the first 50 days, plant growth progressed at near potential rates. The steep increase in water extraction rates indicates the onset of the exponential growth phase and hence increasing demand. From Fig. 7, the actual root front velocity can be estimated by regressing tc values from layers 0.8 to 1.4 m against soil depth and forcing the regression through zero (1.5 cm day-J). These values represent the theoretical lower limits for EFV and to. This indicates that by 51 DAS the root system was well developed and roots were able to supply water from a total depth of 0.8 m which resulted in a maximum extraction rate of 5.2 mm day -I at that time. This marks the time when depth of the extraction front, which initially lagged behind the root front development by 21 days (to), equalled depth of the root front and can be regarded as a switch from a demand to a supply-limited situation. This implies that for the first 50 days, water extraction was determined by crop demand and hence proceeded below potential rates as determined by moisture availability and root characteristics, resulting in extraction rates (k0 for layers accessible during that time also below potential. Hence, kl values between 0.05 and 0.06 day -! , as found in the two deepest layers, are more likely to represent potential extraction rates than those from shallower soil layers. Meinke et al. (1993) suggested that differences in EFV found for sunflower grown on different soil types, but under similar environmental conditions, were proportional to differences in to (i.e., the lag period before the extraction front starts its descent). They argued that under similar levels of demand for soil water (similar environments), differences in soil water supply (i.e., soil type effects) caused differences in the supply/demand ratio and hence in EFV and to. Thomas et al. (1995), on the other hand, did not find such an association with soil type and attributed differences found in the lag phase to differences in air temperature and their effects on root and shoot growth. They investigated differences in crop water demand by varying the planting date for crops grown on the same soil. Both cases can be explained by considering the supply/demand ratio. Either decreasing supply or increasing demand reduced the supply/ demand ratio and resulted in lower EFV and a shorter lag phase. Further, when Thomas et al. (1995) com-

110

pared results for barley grown in different environments and on different soils they did not find an association between EFV and to as both demand and supply varied. The average value for TEc of 4.7 compares well with values of 4.5 Pa reported for other C3 species, confirming the conservative nature of TEc (Monteith, 1988). Thus, providing LAI is estimated correctly, total above-ground dry matter can be calculated with identical results using either RUE x intercepted radiation or the T x TEc. However, it still seems appropriate to switch between the two approaches as suggested by Chapman et al. (1993). Models that 'switch' between energy and water-limited situations fully utilize the convenient and physiologically sound concept of RUE (Monteith, 1994). At the same time they overcome the concern raised by DemetriadesShah et al. (1994) that growth calculated as a function of accumulated light interception and RUE can conceal the effect of other environmental factors.

4.6. Nitrogen in above-ground dry matter Although total nitrogen uptake differed markedly across treatments (Fig. 5), this was associated with corresponding changes in biomass. This explains why nitrogen percentages did not differ among treatments, except for 360N irrigated where presumably luxury N consumption occurred. The observed N uptake after anthesis is in contrast to work reported elsewhere (Gregory et al., 1981; Campbell et al., 1983) and to the assumptions in Sinclair and Amir's (1992) wheat model. The nitrogen deep in the profile, combined with an active and well established root system was the source of substantial nitrogen uptake after anthesis. Spiertz and Ellen (1978) found similar uptake patterns in their winter wheat experiment. It seems likely, particularly under semi-arid conditions, that potentially available N in the upper layers of the soil profile cannot be taken up by the plants due to soil dryness, while at depth there is often little N available. Conversely, in the more humid, European-type climates so much N is generally available during early crop growth that N requirements are largely satisfied by anthesis (cf. 360N irr). These conditions have probably contributed to the frequently made assumption that N uptake ceases at or shortly after anthesis. This is supported by data

from de Ruiter and Brooking (1994) who found a strong, negative correlation in pre- to post- anthesis nitrogen uptake in barley. Our results demonstrate how valuable N reserves deep in the profile can be during kernel-filling. The relatively high kernel N% in the ON irrigated treatment and the narrow range of kernel N% across all treatments further indicate that nitrogen shortage was not severe. When nitrogen limitation occurred, growth was moderated via changes in LAI while most other physiological processes were kept constant.

5. Conclusions In this study the crop physiological basis of water and nitrogen effects on spring wheat was examined by deriving relevant crop physiological parameters from a detailed data set covering a range of water and nitrogen levels. Using a crop physiological framework that considers the determinants of crop growth and yield and relating these findings to those found elsewhere should have improved the understanding and quantification of crop responses to water and nitrogen limitations. Specifically, derived parameter values such as RUE and TEc compared well with those reported from other climatic zones and showed little sensitivity to imposed water and N treatments. The significant treatment effects on crop growth and yield were largely a consequence of the high sensitivity of leaf area development to even moderate levels of water or nitrogen shortage. This study provides a data set and a basis to consider ways to improve simulation capabilities of water and nitrogen effects on spring wheat.

Acknowledgements We thank Mr S. Cawthray for his valuable technical support and excellent management of the experiment and Mr P. Poulton for the construction and maintenance of the meteorological station. The senior author is grateful to the Dutch Government for a visiting fellowship associated with this research and to the Queensland Government for support through their 'SARAS' scheme that assists staff in undertaking further studies.

111

Appendix A Abbreviations, their description and units in alphabetical order

Abbreviation

Description

Unit

Otj, OL

Upper and lower water content in each soil layer Days after sowing Total above-ground dry matter Day of the year Extraction front velocity Fraction of transpirable soil water Final fertile tiller number Harvest index Light extinction coefficient Kernel nitrogen concentration Rate of water extraction per soil layer Kernel number Kernel weight Kernel yield Leaf area index Green leaf nitrogen concentration Maximum plant available water content per soil layer Plant available water holding capacity of soil profile Probabilities for main effects of nitrogen and their interactions Radiation use efficiency Standard error of difference Specific green leaf area Specific green leaf nitrogen Saturation vapour pressure Time at which extraction front commences its descent at rate EFV Time of first water extraction in each soil layer Crop transpiration efficiency Transpiration use efficiency coefficient Maximum temperature Minimum temperature Vapour pressure deficit

mm mm-~

DAS DM DO Y EFV FTSW FTN HI k Kernel N% kl KN KW KY LAI Leaf N% MAWC PAWC P., Pwn

RUE SEa SLA SLN Svp to

tc TE TEc T~x

Train VPD

g m-2 cm day-~

g g-~ % day -t

g kernel -~ g m-2 m2 m-2 % mm mm

g MJ-I cm 2 g-~ g m-2 kPa DAS

DAS g m-2 mm -j g m -2 mm -I kPa °C °C kPa

References Bdstow, K.L. and Carberry, P.S., 1991. Minimum air temperature as a predictor of daily mean vapor density. Conference on Agricultural Meteorology, Extended Abstracts. 17-19 July 1991,

Australian Bureau of Meteorology, Melbourne, Australia, pp. 301-304. Campbell, C.A., Davidson, H.R. and McCaig, T.N., 1983. Disposition of nitrogen and soluble sugars in Manitou spring wheat as influenced by N fertilizer, temperature and duration and stage of moisture stress. Can. J. Plant Sci., 63: 73-90. Carberry, P.S. and Bristow, K.L., 1991. Sensitivity of crop yield prediction to errors in daily vapor pressure deficit. Conference on Agricultural Meteorology, Extended Abstracts. 17-19 July 1991, Australian Bureau of Meteorology, Melbourne, Australia, pp. 297-300. Chapman, S.C., Hammer, G.L. and Meinke, H., 1993. A sunflower simulation model: I. Model develgpment. Agron. J., 85: 725735. Charles-Edwards, D.A., Doley, D. and Rimmington, G.M., 1986. Modelling Plant Growth and Development. Academic Press, Sydney, Australia, 235 pp. Cooper, F.L., 1980. The effect of nitrogen fertilizer and irrigation frequency on a semi-dwarf wheat in south-east Australia. Aust. J. Exp. Agric. Anita. Husb., 20: 359-364. Demetriades-Shah, T.H., Fuchs, M., Kanemasu, E.T. and Flitcroft, I.D., 1994. Further discussions on the relationship between cumulated intercepted solar radiation and crop growth. Agric. Forest Meteorol., 68:231-242. de Ruiter, J.M. and Brooking, I.R., 1994. Nitrogen and dry matter partitioning of barley grown in a dryland environment. N.Z.J. Crop Hort. Sci., 22: 45-55. Fischer, R.A., 1985. Number of kernels in wheat crops and the influence of solar radiation and temperature. J. Agric. Sci. (Cambridge), 105: 447-461. Fischer, R.A., 1993. Irrigated spring wheat and timing and amount of nitrogen fertilizer. II. Physiology of grain yield response. Field Crops Res., 33: 57-80. Fischer, R.A. and Kohn, G.D., 1965. The relationship of grain yield to vegetative growth and post-flowering leaf area in the wheat crop under limited soil moisture. Aust. J. Agric. Res., 17: 281295. Green, C.F., 1987. Nitrogen nutrition and wheat growth in relation to absorbed solar radiation. Agric. Forest Meteorol., 41: 207-248. Gregory, P.J., Marshall, B. and Biscoe, P.V., 1981. Nutrient relations of winter wheat. 3. Nitrogen uptake, photosynthesis of flag leaves and translocation of nitrogen to grain. J. Agric. Sci. (Cambridge), 96: 539-547. Hammer, G.L. and Wright, G.C., 1994. A theoretical analysis of nitrogen and radiation effects on radiation use efficiency in peanuts. Aust. J. Agric. Res., 45: 575-589. McCown, R.L., Hammer, G.L., Hargreaves, J.N.G., Holzworth, D.P. and Freebairn, D.M., 1996. APSIM: a novel software system for model development, model testing, and simulation in agricultural research. Agric. Syst., 50: 255-271. Meinke, H., 1996. Improving wheat simulation capabilities in Australia from a cropping systems perspective. PhD thesis, Wageningen Agricultural University, CIP-Data Koninklijke Bibliotheek, Den Haag, 270 pp. Meinke, H., Hammer, G.L. and Want, P.J., 1993. Potential soil water extraction by sunflower on a range of soils. Field Crops Res., 32: 59-81.

112

Meinke, H., Rabbinge, R., Hammer, G.L. and Jamieson, P.D., van Keulen, H., 1997a. Improving wheat simulation capabilities in Australia from a cropping systems perspective. II. Testing simulation capabilities of wheat growth. Eur. J. Agron., 7: in press. Meinke, H., Hammer, G.L., van Keulen, H. and Rabbinge, R., 1997b. Improving wheat simulation capabilities in Australia from a cropping systems perspective. III. The integrated wheat model (I_WHEAT). Eur. J. Agron., 7: in press. Monteith, J.L., 1986. How do crops manipulate supply and demand? Philos. Trans. R. Soc. London Ser. A, 316: 245-259. Monteith, J.L., 1988. Does transpiration limit the growth of vegetation or vice versa? J. Hydrol., 100: 57-68. Monteith, J.L., 1994. Validity of the correlation between intercepted radiation and biomass. Agric. Forest Meteorol., 68: 213-220. Northcote, K.H., 1979. A Factual Key for the Recognition of Australian Soils, 4th edn. Rellin Tech. Publications, Adelaide, Australia, 123 pp. Payne, R.W. (Editor), 1993. GenstatTM 5 Release 3 Reference Manual. Clarendon Press, Oxford, 796 pp. Robertson, M.J., Fukai, S., Ludlow, M.M and Hammer, G.L., 1993. Water extraction by grain sorghum in a sub-humid environment. I. Analysis of the water extraction pattern. Field Crops Res., 33: 81-97. Siddique, K.H.M., Belford, R.K., Perry, M.W. and Tennant, D., 1989. Growth, development and light interception of old and modem wheat cultivars in a Mediterranean-type environment. Aust. J. Agric. Res., 40: 473-487. Sinclair, T.R. and Horie, T., 1989. Leaf nitrogen, photosynthesis and crop radiation use efficiency: a review. Crop Sci., 29: 9098.

Sinclair, T.R. and Amir, J., 1992. A model to assess nitrogen limitations on the growth and yield of spring wheat. Field Crops Res., 30: 63-78. Sinclair, T.R., Tanner, C.B. and Bennett, J.M., 1983. Water-use efficiency in crop production. BioScience, 34: 36-40. Sinclair, T.R., Shiraiwa, T. and Hammer G.L., 1992. Variation in crop radiation-use efficiency with increased diffuse radiation. Crop Sci., 32:128 l - 1284. Spiertz, J.H.J. and Ellen, J., 1978. Effects of nitrogen on crop development and grain growth of winter wheat in relation to assimilation and utilization of assimilates and nutrients. Neth. J. Agric. Sci., 26: 210-231. Stone, R.C., Hammer, G.L. and Marcussen, T., 1996. Prediction of global rainfall probabilities using phases of the Southern Oscillation Index. Nature, 384: 252-255. Strong, W.M. and Cooper, J.E., 1980. Recovery of nitrogen by wheat from various depths in cracking clay soil. Aust. J. Exp. Agric. Anita. Husb., 68: 82-87. Tanner, C.B. and Sinclair, T.R., 1983. Efficient water use in crop production: Research or re-search? In: H.M. Taylor, W.R. Jordan and T.R. Sinclair (Editors), Limitations to Efficient Water Use in Crop Production. ASA, Madison, WI, pp. 1-27. Thomas, Fukai, S. and Hammer, G.L., 1995. Growth and yield response of barley and chickpea to water stress under three environments in Southeast Queensland. If. Root growth and soil water extraction pattern. Aust. J. Agric. Res., 46: 35-48. Turner, N.C., 1986. Crop water deficits: a decade of progress. Adv. Agron. 39: 1-51. Woodruff, D.R., 1983. The effect of a common date of either anthesis or planting on the rate of development and grain yield of wheat. Aust. J. Agric. Res., 34: 13-22.

© 1997 ElsevierScience B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

113

Comparison of CropSyst performance for water management in southwestern France using submodels of different levels of complexity C.O. Stockle a'*, M. Cabelguenne b, P. Debaeke b aDepartment of Biological Systems Engineering, Washington State University, Pullman, WA 99164-6120, USA blNRA, Station d'Agronomie, BP 27, 31326 Castanet-Tolosan cedex, France Accepted 2 June 1997

Abstract A comparison of the performance of the CropSyst model when using evapotranspiration (ET) and soil water transport submodels of different levels of complexity was conducted. ET submodels included Penman-Monteith, Priestley Taylor, and a temperature-based submodel. Soil water transport submodels included a finite difference numerical method and a simple cascading method. Simulations of biomass production, yield, and water use of irrigated maize, soybean, and sorghum were compared with experimental data collected at Auzeville, southwestern France. Experimental data for growing seasons 1986, 1989, and 1990 (dry years) were used, including three irrigation levels (full, deficit, and no irrigation) each year. Calibration of a small number of crop parameters was conducted using the most complex submodels. Using these most complex submodels, simulations compared well with measurements, with the Wilmott index of agreement fluctuating from 0.956 to 0.997 (an index of 1.0 indicating perfect agreement). The model appeared suitable for applications over a wide range of water availability and stress conditions. Using simpler submodels, the performance of CropSyst tended to decrease, but generally not significantly. A simple cascading method for soil water transport (non-constrained drainage) appeared to be a valid alternative to more complex numerical methods. All ET submodels predicted similar seasonal ET, but differences in vapor pressure deficit estimation led to growth overprediction in some cases. Simpler submodels would have performed similarly to the most complex submodels if used during calibration. Successful applications of the CropSyst model appear feasible for sites with limited weather and soil data. © 1997 Elsevier Science B.V.

Keywords: Growth; Model; Stimulation; Complexity

1. Introduction The agricultural region around Toulouse, southwestern France, is characterized by a variable annual precipitation, fluctuating from 400 to 1000 mm, con-

* Corresponding author.

centrated during the winter months. The use of irrigation is required every season to grow summer crops, and often supplementary irrigation is needed to grow winter crops. Due to the erratic nature of crop water availability in this region, tools for strategic (longterm) and/or tactical (within season) analyses of water management options are needed. Crop growth simulation appears suitable for this purpose, but

Reprinted from the European Journal of Agronomy 7 (1997) 89-98

114 model corroboration is required before applications can be attempted. Some models have been evaluated in this region. SOYGRO (Wilkerson et al., 1983) was improved to predict growth and yield of several soybean cultivars using data collected by the Institut National de la Recherche Agronomique (INRA) at Auzeville, near Toulouse (Colson et al., 1995). This is, however, a single-crop model that cannot be used for water management of the range of crops grown in the region. EPIC (Sharpley and Williams, 1990), a soil erosion and generic crop model, was also evaluated at Auzeville (Cabelguenne et al., 1990). These authors concluded that EPIC was able to predict within the range of experimental error, except for wheat, but model improvements were required for better performance. Detailed studies of crop water extraction and response to drought suggested that EPIC's simulation of these processes could be improved (Cabelguenne et al., 1988). A modified version of this model, EPICPhase (Cabelguenne and Debaeke, 1995), has corrected some of these limitations. One important aspect of model applications for water management in a region is the availability of the required weather and soil input data. Models able to use submodels of varying degree of complexity so as to adjust to the available input data are desirable. CropSyst is such a model. CropSyst (Cropping Systems Simulation Model) represents an effort to simulate growth of single crops or crop rotations in response to management practices and provide estimates of environmental impact (Stockle and Nelson, 1997; Stockle et al., 1994). Crop growth-related processes are generally described with more detail in CropSyst than in EPIC. The model has been corroborated for wheat and maize at locations in the western USA (Stockle et al., 1994; Jara, 1995), and wheat in northern Syria (Pala et al., 1996). Evaluation of CropSyst for longterm rotations, including several crops at two Italian locations, was reported by Donatelli et al. (1996). For water management, CropSyst provides users with a choice of three submodels to predict evapotranspiration (ET). Soil water transport is modeled using either a cascading approach, where the soil is treated as a collection of water reservoirs filling by rainfall or irrigation, and emptying by ET and drainage (Leenhardt et al., 1995), or a finite difference numerical

solution to predict soil water fluxes (Campbell, 1985; Ross and Bristow, 1990). The choice of submodels affects the required amount of input data and computation time. However, its impact on the performance of the model has not been evaluated. The objective of this study was to compare the performance of CropSyst when simulating biomass production, yield, and water use of irrigated maize, soybean, and sorghum at Auzeville, southwestern France, using six combinations of three ET and two soil water transport submodels.

2. The CropSyst model CropSyst is a multiyear, multicrop, daily time-step dynamic simulation model designed as a tool to analyze growth and yield of single crops or crop rotations in response to weather, soil, and management. In addition, the model is able to predict soil erosion by water and the fate of chemicals in the soil (nitrogen and pesticides). Management options include irrigation, salinity management, nitrogen fertilization, chemical application, tillage operations, and residue management. Descriptions of crop growth and development, as well as the nitrogen and water submodels in CropSyst have been presented elsewhere (Stockle et al., 1994; Jara, 1995; Ferrer-Alegre, 1995; Stockle and Debaeke, 1997; Stockle and Nelson, 1997). Readers interested in details of the model and a manual can download a copy from the lnternet (www.bsyse.wsu.edu/cropsyst). Some details of the water balance pertinent to this study are given below. 2.1. ET submodels

Crop ET is obtained from reference ET multiplied by a crop coefficient (Doorenbos and Pruitt, 1977; Jensen et al., 1990). For the determination of reference ET, CropSyst offers three submodels: PenmanMonteith (PM), Priestley-Taylor (PT), and a simple temperature-based (TB) submodel. The calculation of reference ET in the PM and PT submodels follows the approach suggested by Jensen et al. (1990). The simple TB submodel is based on concepts by Campbell and Diaz (1988), Bristow and Campbell (1984), and Ndlovu (1994). The PM submodel requires maximum and mini-

ll5

mum temperature, maximum and minimum relative humidity, solar radiation, and wind speed, and is the preferred submodel when all input data are available (Allen et al., 1989; Jensen et al., 1990). The other submodels are simplifications of the PM submodel. The PT submodel is applied when humidity and wind speed data are not available. The PT version implemented in CropSyst allows the PT coefficient (which accounts for the lack of the aerodynamic and canopy resistance terms present in the PM submodel) to fluctuate with vapor pressure deficit (VPD) as suggested by Steiner et al. (1991). Mean daily VPD (VPD) in kPa is estimated from air temperature (Ndlovu (1994): VPD = C(e~m~- e~'mi") 1-a(e~m~ -e~mi.

(1)

where er,~ o and eTmin o a r e the saturation vapor pressure (kPa) at maximum and minimum temperature (°C), respectively, a is an aridity factor (0 for extreme humid and 0.1 for extreme arid conditions), and C is the ratio of the daily mean VPD to the daily maximum VPD. Using over a thousand daily data points, values of a and C for Auzeville were determined as 0.015 and 0.54, respectively. This mean daily VPD estimate is also used to calculate transpiration dependent biomass (BT) production following the model proposed by Tanner and Sinclair (1983). KnTT Br = ~ VPD

(2)

where Br is in kg/m 2 per day, KBr is the biomasstranspiration coefficient (kPa), and T is transpiration (kg/m 2 per day). The TB submodel uses the same implementation of the PT submodel, but with solar radiation (Rs) predicted from air temperature (Bristow and Campbell, 1984; Ndlovu, 1994). -B(AT)2~

fitted parameter (0.31 MJ/m 2 per day per °C for Auzeville). 2.2. Water transport in the soil profile

To distribute the water entering the soil, CropSyst offers two submodels: a cascading approach (C) and a finite difference numerical method (FD). Both methods require the same basic soil inputs (water content between-30 kPa, field capacity and-1500 kPa, wilting point). In addition, soil hydraulic properties required for the FD submodel are derived from texture (Campbell, 1985). In the simple cascading submodel, infiltrating water is passed on layer by layer down the soil profile as upper layers are refilled to field capacity. After the entire profile is at field capacity, any remaining water is considered deep percolation. Cascading infiltration is calculated only when rainfall or irrigation events occur. The numerical method divides the soil profile into elements separated by nodes, at which soil water potential, water content, and root fractions are defined. This approach solves simultaneously for water transport in the soil and crop water uptake. The change in water content in each node is given by the Richard's equation. Water content and the hydraulic conductivity are related to soil water potential at each node and the physical properties of the surrounding elements. A system of equations is constructed in finite difference form for each soil node. Appropriate boundary conditions are defined including irrigation, evaporation, and free drainage or a shallow water table. The system is solved numerically for each simulation day using the Kirchoff transform and the Newton-Raphson method (Campbell, 1985). The soil is allowed to be non-homogeneous by using the approach of Ross and Bristow (1990).

3. Materials and methods

(3) 3.1. FieM data

where Rs is in MJ/m 2 per day, R~o is the extraterrestrial solar radiation (MJ/m per day), Rso_30 is the extraterrestrial solar radiation 30 days before current day (MJ/m 2 per day), AT is the difference between maximum and minimum temperature (°C), and B is a

A long-term cropping systems experiment was conducted at the Auzeville experimental station of INRA, near Toulouse, France from 1983 to 1992. The main objective was to evaluate crop rotations at three input

116

Table 1 Crops and cultivars grown in the experimentused for the CropSyst validation study Crop

Input level Year 1986 Year 1989 Year 1990

Sorghum

Ill II I III II I III II I

Soybean Maize

Argence Argence Argence Weber Weber Kingsoy LG22 LG22 Brio42

Argence Argence Argence Alaric Alaric Kingsoy Volga Volga Eva

Argence Argence DK18 Alaric Alaric Kingsoy Volga Sabrina Eva

losses are negligible. In 1986, time progression data of above-ground biomass, green leaf area index, and ET throughout the growing season were available. In addition, the phenological stages of the crops were recorded this year. The soil information required by the model was available for all plots. Weather data were collected on site, including rainfall, maximum and minimum temperature, maximum and minimum relative humidity, wind speed, and sunshine hours. Solar radiation was estimated from sunshine hours (Lambert and Mougel, 1989).

3.2. Simulations levels. Input level I was unirrigated and received little fertilizer; level II received limited irrigation, restricted to the most sensitive growth phases, and moderate fertilization; and level III received full irrigation and a large amount of fertilizer. More details on the experiment can be found elsewhere (Cabelguenne et al., 1990; Debaeke et al., 1993). In this experiment, the main limiting factor was water, while fertilization was applied to satisfy crop requirements for each water input level. Three crops were included in this study: maize, sorghum, and soybean. Cultivars differed among treatments and years (Table 1), but they all belonged to the same maturity type. Three growing seasons were selected (1986, 1989, and 1990), which corresponded to dry years. This was done to maximize crop response to water deficit and to avoid water contribution to the root zone from an underlying water table, as it may occur in wet years. Final above-ground biomass, grain yield, and seasonal ET for 26 of the 27 plot 2 years of interest (3 years x 3 crops x 3 input levels) were used for comparison with 3 model simulations. B iomass and yield data were missing for sorghum in 1989, input level I. In addition, ET data for year 1990 were found unreliable and not used for model corroboration. Biomass and yield were determined from 5-m 2 samples collected at harvest in each plot. ET was approximated from a soil water balance based on weekly measurements of soil water content with a neutron probe, and records of precipitation and irrigation. This procedure may have some problems in properly accounting for soil evaporation, and assumes that the water table contribution to the soil water balance and percolation

Observed weather and soil data, initial soil water content, and irrigation calendars were input for the 26 plot-year simulations. Crop phenology observed in 1986 was the base for determining the thermal time requirement for each phenological stage. For sorghum, this did not represent a problem because mainly one cultivar was used during the 3 years (Table 1). For soybean, cultivars differed but had similar phenology (A. Bouniols, INRA at Auzeville, pers. commun.). For maize, cultivars used in 1986 were different to more modern cultivars used in subsequent years. Thermal time requirements to reach each phenological stage of the cultivar Volga was obtained from an independent source (P16net, 1995), and used for all other maize cultivars in 1989 and 1990. Crop parameters used as input for the simulations are shown in Table 2. Several of these parameters were input either as observed experimentally in this and other experiments at the same INRA station (e.g., maximum LAI for an unstressed crop) or as recommended in the CropSyst manual (Stockle and Nelson, 1997). A few parameters required calibration. The most important calibrated parameter was the biomass-transpiration coefficient (KBr). Typical values and the range of KBr variation are available (e.g., Tanner and Sinclair, 1983; Loomis and Connor, 1992), but the reported variation is large and the amount of data still limited so as to benefit from a calibration. The value of Ksr was adjusted to better approximate the progression of biomass accumulation (not the final biomass) observed for the input level III treatment of each crop in 1986. It must be noted that

117

KBT values in CropSyst are appropriate for use with daily (24 h) mean VPD values while those typically found in the literature are for use with daytime average VPD. Thus, KBT values obtained from calibration (Table 2) are about 30% smaller. The ET crop coefficient parameter required to estimate the ET ratio between the simulated crop at full development and the reference crop (Doorenbos and Pruitt, 1977) was adjusted to approximate the ET evolution (not the total seasonal ET) observed on the same plots. To capture the effect of water stress on phenology, leaf duration, and harvest index, parameters were set to approximate the available information from the non-irrigated plots in 1986. Model outputs of seasonal ET, above-ground biomass, and grain yield were compared with observed values. However, measured and simulated biomass for soybean were not compared because a sizable fraction of the above-ground biomass drops to the ground before harvest, which is not reproduced by

the model. The agreement between simulations and measurements was evaluated using regression analyses and statistical indices such as the root mean square error (RMSE), the ratio of RMSE over the observed average, and the Wilmott index of agreement (Wilmott, 1982)'. The latter is both a bounded (0 to l) and a relative measure of agreement, where a value of 1.0 implies perfect agreement between simulated and observed values.

4. Results and discussion Fig. 1 shows observed and predicted biomass, grain yield, and ET for simulations conducted using the most detailed ET and soil water transport submodels (PM and FD, respectively). A large range of response to water was obtained in the experiment, permitting to test the model for a wide range of conditions. For sorghum and maize biomass, simulated/observed

Table 2 Summary of crop parameters for CropSyst simulation Parameters Degree-days emergence (°C-d) Degree-days begin flowering (°C-d) Degree-days peak LAI(°C-d) Degree-days begin grain fi Iling (°C-d) Degree-days maturity (°C-d) Base temperature (°C) Cutoff temperature (°C) Phenologic sensitivity to water stress Maximum root depth (m) Maximum LAI Specific leaf area (mE/kg) Stem/leaf partition coefficient Leaf duration (°C-d) Leaf duration sensitivity to stress Solar radiation extinction coefficient ET crop coefficient Maximum water uptake rate (mm/day) Critical canopy water potential (kPa) Wilting canopy water potential (kPa) Biomass-transpiration coefficient (Pa)a Radiation-use efficiency (g/MJ) Maximum harvest index, HI HI sensitivity to stress during flowering HI sensitivity to stress during grain filling

Obs Obs Obs Obs Obs Man Man Cal Obs Obs Obs Obs Obs Cal Man Cal Man Man Man Cal Man Obs Cal Cal

Sorghum

Soybean

Maize (LG22)

Maize(Volga)

100 850 830 l l30 1450 8 23 2 1.7 8 22 2 1200 1.0 0.46 1.0 12 -1200 -1800 6.5 4 0.48 0.1 0.1

100 700 1200 930 1750 6 20 2 1.7 7 28 3 1200 1.0 0.46 1.0 12 -1000 -1500 3.5 2.5 0.3 0.4 0.4

100 850 875 1250 1850 6 23 3 1.8 7 22 3 1200 0.5 0.46 1.2 12 -1100 -1650 6.7 4 0.48 0.3 0.3

100 950 980 1200 2050 6 23 3 1.8 7 22 3 1200 0.5 0.46 1.2 12 -1100 -1650 7.0 4 0.48 0.3 0.3

Parameters were set as observed experimentally (Obs), extracted from the CropSyst manual (Man), or set by calibration (Cal). aThese coefficients are for use with daily (24-h) mean VPD values, as calculated in CropSyst.

118

data pairs are close to the 1" 1 line of perfect agreement with one significant outlier for maize. For sorghum and soybean yields, simulated/observed data pairs are close to the 1:1 line while results for maize include two significant outliers. ET results also show most data pairs close to the 1:1 line, with a trend to overprediction and a few outliers. Because detailed growth information was not available for 1989 and 1990, explanation for significant departures could not be found. The main problems are related to simulations for maize, a crop that was represented by several cultivars in the field experiment. Regression between observed and simulated values resulted in coefficients

~

600

p

40 r2 = 0.91 a = 2443 b = 0.86

© 30

of determination of around 0.9, regression slopes close to 1 (not different from 1.0 at 95% confidence) and intercepts close to 0 (not different from zero at 95% confidence). A statistical analysis of the results obtained is presented in Table 3. This analysis confirms the good performance of the model. In the case of sorghum, average simulated biomass and yield are close to observed values (overprediction of 4.0 and 6.0%, respectively), with RMSEs representing 6.8 and 11.8%, respectively, of the observed average. Wilmott indices of agreement are reasonably close to 1 (perfect agreement). ET results show an average model over-

r 2 0.88 a = 25.6 b = 0.98 n=17 I

i

:

~ /

/

/ /~l

t soo

r~

I / ~~[ ] . . . .

"/~

I

J

I

I._~7"

i

400

..~

~"

,

/

~

.a~e

l

I 200

0 10 20 30 40 Observed biomass (ton/ha)

o

-e

Sorghum'

/I ll/~/-" I

r2 = 0.89 a=947 O.~ / b = 0.94 oql ~ n -8 /~//~w []

/X'./

3 0

l

~

l

300

400

500

600

Observed ET (ram)

/'._ 0

3

6

-a4

12

Observed yield (ton/ha)

i

i !

m

,Id

elm

/

M, e

9

J

~3

,-9

,

/ [ ,~/" /~'./ •

.~2 o

b = 1.03

.

r2 = 0.91 a=284 b = 0.87

el

-,,=-78o

'Ill

~

] I

//7/4" - r2 _ 0.84

6 em

200

5

15

,IW

Sorghum Soybean Maize

[] •

0

~= 12

•

300

Sorghum

•

0 15

0

.............................Soybean i

i

i

l

1

2

3

4

5

O b s e r v e d yield (ton/ha)

Fig. 1. Comparison of observed and simulated values using the PM and FD submodels for crops grown in 1986, 1989, and 1990 at Auzeville, France. Upper left: biomass for sorghum (circle) and maize (square). Upper right: evapotranspiration (ET) for sorghum (circle), soybean (square), and maize (triangle). Lower left: grain yield for sorghum (circle) and maize (square). Lower right: grain yield for soybean.

119

prediction of 9.9%, a RMSE representing 14.4% of the average observed ET, and a low Wilmott index of agreement. Poor ET results are mainly due to one significant outlier out of five points (Fig. 1). For soybean, average yield was slightly underpredicted (less than 1%). RMSE represented 13.5% of the observed average, and the Wilmott index of agreement was close to 1.0. Simulated ET was 7.5% higher than measured, with RMSE amounting to 9.7% of the observed average ET and a high Wilmott index of agreement. B iomass results for maize also showed good agreement between simulated and measured values (3.6% average underprediction, RMSE representing 15.3% of the observed average, and a high Wilmott index of agreement). Results for grain yield showed a somewhat large RMSE (representing 21.3% of the observed average), mainly due to two significant outliers out of 9 points (Fig. 1). ET results showed excellent agreement between predicted and observed values, with a slight underprediction of the average (less than 1.0%), a low RMSE (representing 3.1% of the observed average ET), and an almost perfect Wilmott index of agreement. In terms of crop growth and yield as a function of water availability, the model appeared able to simulate the relative change of biomass, yield, and ET for a wide range of water stress conditions. Nevertheless, the model performance in the simulation of long-term rotations, including a wider range of crops, remains to be evaluated. The results discussed so far were obtained using the most complex submodels. Table 3 also includes the statistical performance of CropSyst when other submodels were used. The use of the PM/C combination produced only small changes in performance (usually lower) compared to PM/FD. An exception was the dramatic increase in the agreement found for ET in the case of sorghum, resulting from the better agreement for the outlying data point in Fig. 1. It appears that the cascading method can be used as a valid option to the numerical method under conditions of non-constrained drainage. The PT/FD combination resulted generally in a lower performance compared to PM/FD. Growth and yield were overpredicted with the PT submodel, particularly for sorghum. This was not the result of differences in ET prediction, which were similar for both the PM and PT submodels, but rather of differ-

ences in the determination of VPD, which affect directly the calculation of biomass production. The use of the PT submodel requires the daily estimation of VPD from temperature data, while the PM submodel calculates this quantity directly from measured humidity data. Furthermore, the PM/FD combination was used to calibrate the biomass transpiration coefficient (Table 2), a parameter controlling biomass accumulation as a function of transpiration and VPD. If the PT submodel was used for calibration, simulations using this submodel would likely improve to levels comparable with those reported for the PM submodel. Interestingly, simulations using the simple TB submodel in combination with the FD submodel resulted in a similar performance to that obtained with PM/FD. A consistently lower prediction of ET compensated for the VPD prediction problems found in the case of the PT submodel. Both the TB and the PT submodels share the same approach for VPD calculations. Performance of the combination using the simplest submodels (TB/C) was somewhat lower but not very different than that of PM/FD, despite substantial differences in required input data and computation time. CropSyst model applications to sites with limited weather data appear feasible, provided that a few years of solar radiation and humidity data are available to determine parameters for VPD and solar radiation estimation.

5. Conclusions

Using detailed ET (PM) and soil water transport (FD) submodels, CropSyst simulations of growth, yield, and water use of maize, soybean, and sorghum compared well with measurements over three growing seasons for crops grown under three levels of irrigation input at the AuzeviUe experiment station of INRA near Toulouse, France. The Wilmott index of agreement fluctuated from 0.956 to 0.997 (an index of 1.0 indicating perfect agreement), with most cases above 0.965 and one case (simulation of water use for sorghum) with a poor index of 0.786 due to one significant outlier out of five points. Particularly encouraging is the model's ability to simulate the relative change of biomass production, yield, and ET for a wide range of water stress conditions. However, the

120

Table 3 Summary of statistical results for the three crops included in the study and four simulation sets PM/FD

PM/C

PT/FD

PT/C

TB/FD

TB/C

8 16684 17358 1139 0.068 0.985 8 7601 8055 896 0.118 0.967 5 372 409 53.7 0.144 0.786

8 16684 17371 1221 0.073 0.981 8 7601 7919 797 0.105 0.975 5 372 387 26.8 0.072 0.934

8 16684 19023 2527 0.151 0.932 8 7601 8802 1509 0.199 0.917 5 372 411 54.7 0.147 0.783

8 16684 18967 2497 0.150 0.929 8 7601 8624 1324 0.174 0.938 5 372 388 27.8 0.075 0.93

8 16684 17196 1714 0.103 0.963 8 7601 8015 992 0.130 0.958 5 372 400 53.6 0.1 44 0.749

8 16684 17270 1751 0.105 0.958 8 7601 7937 859 0.113 0.969 5 372 378 30.8 0.082 0.896

Sorghum Biomass

Yield

ET

Number of data points Observed average (Oavg) (kg/ha) Predicted average (kg/ha) RMSE (kg/ha) RMSE/Oavg Wilmott index of agreement Number of data points Observed average (Oavg) (kg/ha) Predicted average (kg/ha) RMSE (kg/ha) RMSE/Oavg Wilmott index of agreement Number of data points Observed average (Oavg) (mm) Predicted average (mm) RMSE (mm) RMSE/Oavg Wilmott index of agreement

Soybean Biomass

Yield

ET

Number of data points Observed average (Oavg) (kg/ha) Predicted average (kg/ha) RMSE (kg/ha) RMSE / Oavg Wilmott index of agreement Number of data points Observed average (Oavg) (kg/ha) Predicted average (kg/ha) RMSE / (kg/ha) RMSE / Oavg Wilmott index of agreement Number of data points Observed average (Oavg) (mm) Predicted average (mm) RMSE (mm) RMSE / Oavg Wilmott index of agreement

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

9 2828 2804 381 0.135 0.970 6 412 443 42.0 0.102 0.956

9 2828 2897 434 0.153 0.957 6 412 434 35.3 0.085 0.968

9 2828 3061 465 0.164 0.959 6 412 440 38.4 0.093 0.962

9 2828 3167 543 0.189 0.941 6 412 430 31.3 0.076 0.974

9 2828 2845 391 0.138 0.968 6 412 438 35.9 0.087 0.966

9 2828 2961 475 0.168 0.946 6 412 426 27.5 0.067 0.978

Number of data points Observed average (Oavg) (kg/ha) Predicted average (kg/ha) RMSE (kg/ha) RMSE / Oavg Wilmott index of agreement Number of data points Observed average (Oavg) (kg/ha) Predicted average (kg/ha) RMSE (kg/ha) RMSE / Oavg Wilmott index of agreement

9 19038 18358 2921 0.153 0.966 9 8026 7847 1707 0.213 0.963

9 19038 18903 3234 0.170 0.954 9 8026 8023 1881 0.234 0.95

9 19038 19538 3308 0.174 0.957 9 8026 8362 1999 0.249 0.95

9 19038 20231 3756 0.197 0.940 9 8026 8660 2199 0.274 0.933

9 19038 18332 2981 0.157 0.963 9 8026 7939 1767 0.22 0.958

9 19038 18822 3197 0.168 0.954 9 8026 815 l 1919 0.239 0.946

Maize Biomass

Yield

121

Table 3 continued Summary of statistical results for the three crops included in the study and four simulation sets PM/FD ET

Number of data points Observed average (Oavg) (mm) Predicted average (mm) RMSE (mm) RMSE ! Oavg Wilmott index of agreement

6 416 414 13.0 0.031 0.997

PM/C 6 416 403 19.4 0.047 0.992

PT/FD 6 416 416 15.9 0.038 0.995

PT/C 6 416 405 20.3 0.049 0.991

TB/FD 6 416 408 24.8 0.060 0.987

TB/C 6 416 394 30.9 0.074 0.979

PM, Penman-Monteith ET submodel; PT, Priestley-Taylor submodei; TB, temperature-based model; C, cascading water transport; FD, finite difference water transport.

model performance in the simulation of long term rotations, including a wider range of crops, remains to be evaluated. The model performance resulting from the use of simpler submodels tended to decrease relative to that obtained using the more detailed PM/FD submodel combination. However, differences were minimal when the cascading method replaced the FD submodel. It appears that the cascading method can be used as a valid option to the numerical method under conditions of non-constrained drainage. Simpler ET submodels predicted similar seasonal ET values than the PM submodel, but differences in VPD estimation led to some growth and yield overprediction when using the PT submodel, particularly for sorghum. The PM/FD combination was used to calibrate the biomass-transpiration coefficient, a crop parameter that controls biomass accumulation as a function of transpiration and VPD. In this evaluation, simpler ET submodels would have produced better performance if used during the process of calibration. Successful applications of the CropSyst model appear feasible for sites with limited weather data, provided that a few years of solar radiation and humidity data are available to determine parameters for VPD and solar radiation estimation.

References Allen, R.G., Jensen, M.E., Wright, J.L. and Burman, R.D., 1989. Operational estimates of evapotranspiration. Agron. J., 8 l: 650662. Bdstow, K.L. and Campbell, G.S., 1984. On the relationship between incoming solar radiation and daily maximum and minimum temperature. Agric. For. Meteorol, 31: 159-166. Cabelguenne, M., Jones, C.A., Marty, J.R. and Quinones, H., 1988.

Contribution/t 1'6tude des rotations culturales: tentative d'utilisation d'un mod61e. Agronomie, 8: 549-556. Cabelguenne, M., Jones, C.A., Marry, J.R., Dyke, P.T. and Williams, J.R., 1990. Calibration and validation of EPIC for crop rotations in southern France. Agric. Systems, 33: 153-171. Cabelguenne, M. and Debaeke, P., 1995. Manuel d'utilisation du mod61e EWQTPR (EPIC-PHASE TEMPS REEL). Institut National de la Recherche Agronomique, Station d'Agronomie, Toulouse-Auzeville, France, 68 pp. Campbell, G.S. 1985. Soil physics with BASIC: transport models for soil-plant systems. Developments in soil science No. 14. Elsevier, New York, 150 pp. Campbell, G.S. and Diaz, R., 1988. Simplified soil-water balance models to predict crop transpiration. In: F.R. Bidinger and C. Johansen (Editors), Drought Research Priorities for the Dryland Tropics. ICRISAT, Patancheru, India, pp. 15-26. Colson, J., Bouniols, A. and Jones, J.W., 1995. Soybean reproductive development: adapting a model for European cultivars. Agron. J., 87: I 129- l 139. Debaeke, P., Hilaire, A., Levrault, F. and Puech, J., 1993. Evaluation of low-input cropping systems in southwestern France: some aspects of production variability. In: Proc. Int. Conf. on Integrated Arable Farming Systems, Nitra, SIovakia, pp. 73-80. Donatelli, M., Stockle, C.O., Ceotto, E. and Rinaldi, M., 1996. CropSyst validation for cropping systems at two locations of Northern and Southern Italy. Eur. J. Agron., in press. Doorenbos, J. and Pruitt, W.O., 1977. Crop water requirements. FAO Paper No 24 (revised), Rome, 144 pp. Ferrer-Alegre, F., 1995. A model for assessing crop response and water management in saline conditions. MS thesis, Washington State University, Pullman, 63 pp. Jara, J., 1995. Water use of corn: field experiment and simulation. PhD dissertation, Washington State University, Pullman, 107 pp. Jensen, M.E., Burman, R.D. and Allen, R.G., 1990. Evapotranspiration and irrigation water requirements. American Society of Civil Engineers, New York, 332 pp. Lambert, S. and Mougel, A., 1989. Gestion de l'eau dans les syst 6mes de culture ~ partir d'un mod61e agronomique. Mdmoire de travail de I'Ecole Nationale de la M6tdorologie, Toulouse, no. 261. Leenhardt, D., Voltz, M. and Ranbal, S., 1995. A survey of several agroclimatic soil water balance models with reference to their spatial application. Eur. J. Agron., 4: l-14.

122 Loomis, R.S. and Connor, D.J., 1992. Crop Ecology: Productivity and Management in Agricultural Systems. Cambridge University Press. 538 pp. Ndlovu, L.S., 1994. Weather data generation and its use in estimating evapotranspiration. PhD dissertation, Washington State University, Pullman, 143 pp. Pala, M., Stockle, C.O. and Harris, H.C., 1996. Simulation of durum wheat (Triticum durum) growth under differential water and nitrogen regimes in a Mediterranean type of environment using CropSyst. Agric. Sys., 51: 147-163. Pl6net, D., 1995. Fonctionnement des cultures de mai's sous contrainte azot6e. Doctoral thesis, Institut National Polytechnique de Lorraine, Nancy, France, 245 pp. Ross, P.J. and Bristow, K.L., 1990. Simulating water movement in layered and gradational soils using the Kirchhofftransform. Soil Sci. Soc. Am. J., 54: 1519-1524. Sharpley, A.N. and Williams, J.R., 1990. EPIC, Erosion/Productivity Impact Calculator: 1. Model Documentation. U.S. Department of Agriculture Technical Bulletin No. 1768, 235 pp. Steiner, J.L., Howell, T.A. and Scheneider, A.D., 1991. Lysimetric

evaluation of daily potential evapotranspiration models for grain sorghum. Agron. J., 83: 240-247. Stockle, C.O. and Debaeke, P., 1997. Modeling crop nitrogen requirements: a critical analysis. Eur. J. Agron., 7: 161-169. Stockle, C.O. and Nelson, R., 1997. Cropsyst User's Manual. Biological Systems Engineering Dept., Washington State University, Pullman, 186 pp. Stockle, C.O., Martin, S. and Campbell, G.S., 1994. CropSyst, a cropping systems model: water/nitrogen budgets and crop yield. Agdc. Sys., 46: 335-359. Tanner, C.B. and Sinclair, T.R., 1983. Efficient water use in crop production: research or research? In: H.M. Taylor, W.R. Jordan and T.R. Sinclair (Editors), Limitations to Efficient Water Use in Crop Production. ASA, Madison, WI, pp. 1-27. Wilkerson, G.G., Jones, J.W., Boote, K.J., Ingrain, K.T. and Mishoe, J.W., 1983. Modeling soybean growth for crop management. Trans. ASAE, 26: 63-73. Wilmott, C.J., 1982. Some comments on the evaluation of model performance. Bull. Am. Meteorol. Soc., 63: 1309-1313.

© 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

123

Root growth of three onion cultivars A.D. Bosch Serra *, M. Bonet Torrens, F. Domingo Oliv6, M.A. Melines Pages UdL- Dep. de Medi Ambient i Cibncies del Sbl, Rovira Roure 177, 25198 Lleida, Spain Abstract

Three onion (Ailium cepa L.) cultivars, Valenciana de Grano, Staro and Southport White Globe, were grown at a semi-arid site in northeast Spain in PVC tubes and in drip irrigated field plots. Shoot dry matter and root length were measured at different sampling dates. Under field conditions root length was measured at different soil depths. A linear regression model was developed to represent root growth: lnRL=a+b InSDW, (R2=0.95--0.96), where RL is total root length (ern per piano and SDW is total dry weight (g per plant). Valenciana de Grano shows a significantly different intercept (a) versus the other cultivars (P < 0.001); in this cultivar, from emergence to maturity, root length in relation to plant dry weight increases faster than in the other cultivars. From bulbing to maturity, 56 to 77% of root length has been recorded in the first 20 cm depth and less than 3% root length was deeper than 60 cm. No differences among cultivars were found on the final percentage of root length at different depths. The bulb dry matter yield ranged from 11 to 14 t.ha -1. The maximum root length densities (Lv) achieved from 0 to 20 cm depth were 8.1-9.1 cm.cm -a. These results on Lv values are much larger than those reported in literature, probably as a result of high-frequency fertigation. This pattern of root elongation and the root density and distribution have many practical implications for the design of irrigation and fertilization strategies, also for simulation models. Keywords: Allium cepa L.; Root development; Root distribution; Drip irrigation

1. Introduction

Onion (Allium cepa L.) is the main horticultural crop in the Pla d'Urgell (Catalonia) area. Catalonia is the fourth Spanish onion producer belt. The climate conditions (Bosch, 1993) do not allow early onion production as in other Spanish areas, and onions are cultivated mainly for conservation and late sale purposes. Valenciana de Grano (V. de Grano) is the most important cultivar in the area. Onions for dehydration appear as a complement to present onion production. The most limiting factor for yield is the availability of water and the water management system, mostly flood irrigation. In 1990 drip irrigation was introduced in the area and yields increased from 35 to up to 90 t ha -1 of fresh bulbs. Under these new conditions it is necessary to adjust * Corresponding author. Tel.: +34-73-702500; Fax" +34-73-238264; E-mail" angela.bosch@ macs.udl.es

agricultural practices; due to the importance of root functions, such as water and nutrient uptake, a study of the root system is needed. Some work has been done in Spain in order to improve cultural practices in cv. V. de Grano (Pardo, 1990; Suso, 1990), derived cultivars (Juan de et al., 1993), and in the evaluation of new cultivars (S.E.A., 1986). However work seeking a comparison between the general behaviour of cultivars, specifically in root growth and distribution of roots has lagged behind. Previous experiments (Greenwood, 1982; Moore et al., 1987; Birdsall and MacLeod, 1990) describing root growth and development were carried out using different onion cultivars and different management and climatic conditions, or under controlled conditions. Results obtained in these studies cannot be directly used in the Pla d'Urgell area because of the genetic variability in plant root systems and its interactions with environmental conditions. The aim of this paper is to quantify the develop-

124

ment of roots and to assess the differences in the rooting patterns of three onion cultivars, V. de Grano and two other cultivars suitable for dehydration purposes, Staro and Southport White Globe, under high-frequency irrigation. The data on root growth will be useful in order to design irrigation and fertilization strategies to optimise yields. They could also be used in models simulating onion crop growth.

2. Material and methods

2.1. Experiments Experiments were carried out in the Pla d'Urgell area; in the northeast of Spain during two seasons, 1991 and 1992. Geographical ¢oordenate of the study site are 41°38'N, 0°53'E and altitude 250 m. Average and accumulative monthly values of parameters which define the weather conditions prevailing during the study period are shown in Fig. 1. Reference evapotranspiration (ETr) was estimated by means of Penman's formula as modified by FAO. The monthly rainfall has been unusually high in March of 1991 and during May and June of 1992 (with a probability of occurrence less than 10%) if compared with average values in the area. The ETr during 1992 season has been 780 mm. In all experiments the same seed lots of onion ¢ultivars V. de Grano, Staro and Southport White Globe (S.W.G.) were used. These cultivars are usually sown from mid-January to mid-March, depending on soil moisture, for a bulb harvest in August. Experiments have been designed in order to avoid water and nutrient stresses that could affect root growth (Leskovar, 1995). To allow detailed measurements of the relation between root length and shoot dry weight, work has been done under two conditions; in tubes, during early stages and in the field under drip fertigation management. Root distribution was also assessed in the field. Results on root growth under controlled conditions should be used with caution when applied to field conditions (Zobel, 1995), however controlled conditions are very useful in the early stages of seedling growth due to lack of accuracy in root measurements (in the field) when root length is low.

For the first experiment (exp. 1), carried out during 1991, seeds were sown in PVC tubes and plants were maintained just before bulbing started; for the second experiment (exp. 2), carried out during 1992, seeds were sown in the field and lasted up to two weeks after maturity. The exp. 2 was located inside a large grower's field. The soil is classified as an Aqui¢ Xerofluvent, fine silty, mixed (calcareous), mesic (Soil Survey Staff, 1990). Soil structure was moderate subangular blocky, texture was silty clay loam from 0 to 35 cm and clay loam from 35 to 100 cm and organic matter content from 0 to 35 ¢m was 2% (w/w). The previous crop was wheat. Soil was ploughed during autumn prior to seeding. Exp. 1. This experiment was an outdoors complete randomized design with six replicates. Four onion seeds were sown on April 11th in each PVC tube (30 ¢m deep, 12 cm diameter) containing vermiculite and perlite to the top. The base of the tube was closed with a mesh. Tubes grouped in four were placed in single pots (20 cm deep, 34 cm diameter) for leachate control. Tubes were irrigated using Hoagland's solution (Hunter, 1979) three times a week. Pots were drained once a week and refilled with totally fresh solution. After emergence, the seedlings were thinned to one per tube while selecting for uniformity. Exp. 2. This experiment was a randomized complete block design with two replicates. Each plot had an area of 8.1 m 2. The row distance was 15 cm, and there were 9 rows in each plot, the O central rows were used for sampling purposes. Plants were directly drilled in a rather dry soil the 4 th March after the seedbed preparation. Seedling emergence was recorded thrice weekly from lengths of the second and fifth rows of each plot calculated to contain approximately one hundred emerged seedlings. After 50% emergence plants were thinned in order to achieve a density of 80 plants m -2. Plots were drip irrigated. Irrigation started a few days after drilling and 202 mm of water were applied in order to bring the full soil to field capacity. Soil water matric potential was mesured using tensiometers located at different depths (0-60 cm) and distances of the emitters. It was always higher than - 1 8 kPa in all places. The total amount of applied water during the season was 646 mm. Fertilizers were applied with irrigation water" 266 kg N/ha, 284 kg P/ha

125

and 463 kg K/ha. No herbicide was applied once the crop was established and weed control was done by hand.

2.2. Sampling and data acquisition Exp. 1. Six plants (shoot and roots) per cultivar were sampled on 18 May, 26 May, 1 June, 9 June and 17 June. At each sampling date the maximum bulb diameter and the minimum pseudostem diameter were measured and the total oven dry (65-70°C) weight of shoots were determined. Root samples were transferred into polythene bags and stored at 4"C until the roots were cleaned. Roots were cleaned with a jet of water and the length of the root system for each plant (RL) was measured directly with a ruler. Exp. 2. Plants were harvested and soil sampled for root measurements four times during the growing period on 20 June, 3 July, 21 July and 5 August. Soil was sampled using an auger of 8 cm diameter, which coincided with the distance between plants. Soil samples were taken beneath the plants at depths of 0-20 cm, 20-40 cm, 40-60 cm on 20 June and 3 July. Four plants per cultivar (two for each replicate) were sampled. In the last two samplings (21 July, 5 August) an additional sample at a depth of 60-80 cm was also taken; furthermore two soil samples were taken at mid-way points between the rows and at the same depths (0-20, 20-40, 40-60, 60-80 cm). These new added sampling points were between two plants which were also sampled. The root lenght measured between rows was related to the average shoot dry weight of the two adjacent plants. In the last two sampling dates six plants per cultivar (three for each replicate) were, thus, included in the sampling. Soil samples were transferred into polythene bags and stored at 4°C until the roots were extracted. The extraction of roots from soil matrix was done by washing with a sodium hexametaphosphate solution, and floating off the roots. Roots were separated using a 100-mesh screen and were stained with Congo Red. The root-length density (Lv in cm cm -3) under the area sampled was estimated for every soil sampiing depth using the linear intersection method of Newman (1966). In order to calculate the total root length per plant

at all sampling dates, two assumptions had to be made: a) root-length density, on these sampling dates, was constant in the total area occupied by the plant at the different depths measured, b) root length at depths lower than 60 cm can be neglected. Assumptions were based on Weaver and Brunner (1927) about the growth pattern of onion roots, (in a large plant, roots spread horizontally and then grown downwards, the bulk of the roots is in the upper 30 to 60 cm), as well as on field observations; particularly the short distance between plants in and between rows, and also, the optimum and uniform growth of onion plants under drip irrigation. Shoot dry weight (SDW) of sampled plants was determined as in exp. 1. Bulbing was characterized by the bulbing ratio (maximum bulb diameter/minimum pseudostem diameter). Onset of bulbing was dated when bulbing ratio exceeded 2.0 (Clark and Heath, 1962). In order to determine maturity date (80% foliar fall-over), pseudostems of fifty plants in the inner rows were assessed thrice weekly to see if they had become flaccid. Two weeks after maturity date, the bulbs of seventy-five plants were harvested and remains of the foliage was removed. Measured variables were bulb diameter, bulb fresh weight and oven bulb dry weight.

2.3. Statistical analysis Standard regression and variance analysis and comparison of means were carded out to assess root performance and differences between cultivars. The relationship between shoot dry weight and root length was calculated using data from the two experiments, assuming that the relationship was not modified during the experiments.

3. Results

3.1. Plant growth, bulb development and yield Exp. 1. Sampling dates correspond to 28, 36, 42, 50 and 58 days after 50% emergence. On the first sampling date (18 May) the mean number of leaves was two for all cultivars without significant differ-

126

30.0

~

35.0

25.0

30.0

~

Rs 1991

' 20.0

25.0

i

B

Rs 1992

20.0 ~

--0..- Tmax 1991

• ~,, 15.0

Q.

15.0 E 10.0

10.0 ~

5.0

5.0

0.0 AP

MY

JN

JY

AG

200.00 180.00 160.00 140.00

120.00

E

100.00

I.,.

~

-__ Tmax 1992 ---e-- Train 1992

0.0 MR

~

- 0 - - Train 1991

i

80.00 60.00 40.00 20.00 0.00

MR

AP

MY

JN

200.00 180.00 160.00 140.00 120.00 100.00 80.00 ~

r ' - " - t Err 1991

60.00 8_

- 4 - - Re 1992

Err 1992

--o-- Re 1991

40.00 20.00 0.00

JY

AG

Fig. 1. Environmental conditions, years 1991 and 1992. (Rs, solar radiation; Tmax, average monthly maximum temperature; Tmin, average monthly minimum temperature" ETr, total monthly reference evapotranspiration" Pre, total monthly rainfall).

ences. On the last sampling date (17 June), plants were not bulbing and there were no differences in bulbing ratio between cultivars (Table 1). Exp. 2. Sampling dates correspond to 79, 92, 110 and

125 d a y s a f t e r 50% e m e r g e n c e f o r cv. V. de

Grano;

82, 95, 113 a n d 128 d a y s a f t e r 50% e m e r -

gence for cv. Staro and 81, 94, 112 and 127 days after 50% emergence for cv. S.W.G. Staro and S.W.G. started bulbing between the first (20 J u n e ) a n d t h e s e c o n d (3 July) s a m p l i n g d a t e s a n d V. de G r a n o s t a r t e d o n 3 July. By t h e fourth s a m p l i n g d a t e (5 A u g u s t ) all c u l t i v a r s r e a c h e d m a t u r i t y .

Table 1 Bulb development in 1991 (exp. 1) and 1992 (exp. 2) growing seasons and bulb yield parameters in 1992 (exp. 2) growing season Sample date

Bulbing ratio (cm.cm -1)

Maturity date (80°,4 fall-over) Exp. 2** Fresh total bulb yield (g.m -2) Exp. 2 Dry matter in bulb (%) Exp. 2 Dry matter bulb yield (t.ha -1) Exp.2 Bulb diameter at harvest (cm) Exp. 2

09.06.91 17.06.91 07.07.91" 20.06.92 03.07.92 16.07.92"

Cultivar V. de Grano

Staro

S.W.G.

1.4 a 1.6 a 2.3 a 1.5 a 1.9 b 3.2 a 03.08.92 (123) 12.840 a 8.6 b 11.04 b 7.0 a

1.5 a 1.7 a 2.2 a 1.4 a 2.5 a 3.1 a 28.07.92 (120) 8.048 b 16.2 a 13.04 a 5.7 b

1.5 a 1.5 a 2.1 a 1.4 a 2.3 ab 3.1 a 29.07.92 (120) 8.040 b 17.4 a 13.99 a 5.7 b

Means with common letters are not significantly different by Duncan's Multiple Range Test at the 5% level. * At this date no root sample was taken. ** Numbers in brackets are days from 50% emergence.

127

Table 2 Uptake rate of N, P and K for V. de Grano between the two first sampling dates Period

Uptake rate (kg.ha-l.day -1)

Increment in shoot dry weight (kg.ha-l.day -1)

20.06-03.07

328.6

N(1)

pC2)

K(3)

8.4

1.3

6.6

(1) Unpublished results (Bosch A.D.) (2) Assuming 0.4% P in the whole plant (Ajakaiye and Greig, 1976). (3) Assuming 2.0% K in the whole plant (Zink, 1966).

3.2. Root length

Bulb yields were higher in V. de Grano while dry matter content was lower. The mean bulb diameter was acceptable for market (Table 1). All cultivars had a fast growth during the sampiing period, that means high rates of nutrient uptake as is showed for cultivar V. de Grano (Table 2).

In the field experiment, and for the last two sampiing dates, the shoot dry weight was not statistically different between plants, irrespective of their position in relation of the point sampled for root length

5OO A

400

000

•

g -.~ n,

lOO o

L

o**

---

0

I

I

I

I

!

J

~

P

l

5

10

15

20

25

30

35

40

45

50

8Dw (g per plant)

~4oo g .a I

t

O O lOO

000 o

l

o

5

10

,~,

15

I

1

i

I

I

I

20

25

30

35

40

45

40

45

50

8~W (g per plant)

500 A

400

oo

00

• •

•

g 100 a~

0 0

5

10

1

2o

3o

SOW (g per plant)

Fig. 2. Relationship between total root length (RL, m per plant) and shoot dry weight (SDW, g per plant) for three cultivars: (a) Valenciana de Grano, (b) Staro, (c) Southport White Globe. Data are from experiment 1 (SDW < 2) in tubes and experiment 2 (field).

128

LV (cm cm "a) 0 0 10 20 "~ 30

2 --'

-

I

'

4

6

I

I

.......

8

10

I

I

12 "

'

14

I"

4

(0.97;2.42;1.82;2.65) ;2.05)

.c 50

(0.15;0.24;0.84;1.08) -,-D-- D92

(0.28;0.14)

70

Dl10 D125

0 Ol A

,

2

4

~

,,.

6 ..

Lv (cm .cm-3) 8

I

,

-,t

10

12

I

I'

(0.91 ;1.62;2.09;2.60)

20

7080[

. - o - . Dg5 D11

(0.17;0.19)

~D12

. . . . . . . . .

Lv cm.cm "= 0 0 10 A20

2

4

)

!

"

6

8

I

!

-"

10

12

I

I

14

li 30

v 40

,C

(0.03;0.34;0.43:0.53)

a 60 70

(0.09;0.15)

90

--o-~1 - 0 - . D94 D112 .-)(-- D127

Fig. 3. Mean root length densities (cm cm -3) by interval depths and different days (D) from 50% emergence in field conditions of three onion cultivars" (a) Valenciana de Grano, (b) Staro and (c) Southport White Globe. Numbers in brackets are standard errors of Lv measured for each depth and sampling date ordered from the first to the last date.

measurement, plants sampled underneath or plants in rows around the sampling point. In addition, no statistical differences were observed for the root lengths measured. This fact corroborates the similitude of plants and the uniformity of root distribution. The total root length (RL, m per plant) plotted against the shoot dry weight (SDW, g per plant) for the different onion cultivars is shown in Fig. 2. Root lengths of the field experiment represented in Fig. 2 were taken from 0 to 60 cm depth. The relationship between the logarithmic values of these variables can be described as linear (Table 3)

being slightly improved by including a quadratic term (cOn (SDW)) 2) in a curvilinear regression. The standard error of the estimated values of ¢ is high, except for cv V. de Grano in exp. 2. Parameter values obtained by fitting the linear regression for the two experiments together are shown in Table 3. Comparing both straight lines, the slope b is not significantly different between cultivars. The intercept of V. de Grano is significantly different (P<0.001) from that of both Staro and S.W.G. The latter do not differ significantly mutually. The common slope (b) and intercept (a) for Staro and S.W.G. are 0.87 and 7.18 respectively.

129 Table 3 Models fitted between lnRL (cm per plant) and the InSDW (g per plant) for the different cultivas and experiments (1 in tubes before bulbing; 2 in the field, bulbing period) and parameter values and standard errors by fitting lnRL=a+b InSDW over the experimental period as a whole, (plant in tubes and in the field) for the different cultivars Cultivar

V. de Grano V. de Grano Staro Staro S.W.G S. W. G

Equation

R ~(residual d.f.)

linear* quadratic** linear quadratic linear quadratic

Linear equation. Exp. (1+2)

Exp. 1

Exp. 2

Exp. (1 +2)

a(s.e.)

b(s.e.)

0.89 (26) 0.90(25) 0.91(26) 0.91 (25) 0.87(27) 0.90(26)

0.82(14) 0.92(13) 0.89(17) 0.90(16) 0.65(15) 0.69( 14)

0.96(42) 0.97(41) 0.95(45) 0.98(44) 0.95(44) 0.97 (43 )

7.57(0.08) 7.14(0.08) 7.22(0.08) -

0.89(0.03) 0.86(0.03) 0.89(0.03) -

* linear: inRL = a+b InSDW ** quadratic: InRL = a+b InSDW+c(InSDW) 2

3.3. Root distribution with depth under field conditions The evolution of mean root length densities in field conditions is presented in Fig. 3. In less than fifteen days prior to bulbing, the mean Lv doubled (Staro, S.W.G.) or tripled (V. de Grano) in the first 20 cm depth. The maximum root length density is reached close to maturity (80% fall-over). Root densities tended to remain roughly constant on samples taken before and after onset of maturity.

The percentage of root length at different depths for every sampling date and cultivar is presented in Table 4. From bulbing to maturity, the soil is intensively explored by onion roots although less than 3% of root length was found deeper than 60 cm, as it was assumed during experimentation. In the first 20 cm, 54 to 77% of root length has been recorded. Layer 040 cm contains 85 to 90% of root length, depending on the cultivar. No differences in root distribution as a percentage in the different soil depths studied has been found between cultivars.

Table 4 Percentage of root length at different depths and at different sampling dates for different onion cultivars Sampling date*

Depth (crn)

20.06.92 (81)

00-20 20-40 40-60 00-20 20-40 40-60 00-20 20-40 40-60 60-80 00-20 20-40 40-60 60-80

03.07.92 (94)

21.07.92 (112)

05.08.92 (127)

Percentage of root length at different depths for each onion cultivar** V. de Grano

Staro

S.W.G

58 34 8 71 22 7 58 30 9 3 61 26 Il 2

67 (0.960) 25 (0.520) 8 (0.282) 72 (l.OlO) 20 (0.468) 8 (0.276) 69 (0.976) 20 (0.454) 9 (0.292) 2 (0.146) 77 (1.076) 15 (0.397) 6 (0.237) 2 (0.126)

56 (0.846) 34 (0.616) 10 (0.324) 62 (0.899) 31 (0.593) 7 (0.275) 62 (0.907) 30 (0.571) 6 (0.257) 2 (0.134) 66 (0.946) 25 (0.515) 7 (0.270) 2 (0.141)

(0.856) (0.623) (0.293) (1.004) (0.490) (0.257) (0.858) (0.574) (0.306) (0.180) (0.888) (0.534) (0.342) (0.136)

In the first and second sample date no soil sample was taken from 60 to 80 cm depth. * Figures in brackets are days from 50% emergence ** Figures in brackets are the data transformed to angles.

130

4. Discussion

4.1. Bulb yield Bulb dry matter yield obtained in our experimental field (Table 1) with an acceptable size for market standards, is higher than the highest recorded yields in experimental plots (Brewster, 1994). Probably yields could have been even slightly higher if bulbs would have been harvested four weeks after onset of maturity or later, because at harvest plants still had green leaves. Total bulb yield indicates very good, near optimum, growth for the cultivars used. The yield of cv. V. de Grano was doubled when compared with other experimental plots in Spain (Martin de Santa Olalla et al., 1994). In their study Recas, a V. de Grano derived cultivar was used and plots were irrigated by furrows to achieve optimum yield. The differences in yield between studies may be due to the use in their study of low planting density (25 plants m -2) and manual transplanting. These high yields also point to the potential of drip irrigation for water stress-sensitive plants like onion (Goltz et al., 1971; Millar et al., 1971).

4.2. Root length Results (Table 3) show that onion root length is related to shoot dry weight. The assumption that this relationship was not modified during the experiments can be corroborated for different reasons: a) the maximum difference in monthly average air temperature between the two years studied was 3°C, both in May and June, b) in both experiments (exp.1 and exp.2), root elongation was not restricted by soil water potential, c) although plants in tubes were irrigated using Hoagland's solution, this does not imply that onion root length would be different when compared with fertigation in field conditions, although Aung (1982) has shown that the number of primordia on the primary root of tomato seedlings increases by Hoagland's solution, and d) the last sample in the first experiment was taken prior to the period of rapid shoot growth before bulbing, when the volume of the tube was not limiting. Although the tedious work of measuring root lenght by a ruler could have reduced the precision of data,

it is difficult to be more accurate in field measurements by taking soil samples when root densities are low. The relationship between root length and shoot dry weight follows a linear trend when variables are expressed as logarithms. Howewer, it should be stressed that under distinct experimental conditions, (e.g. with water or nutrients being limiting), this relationship may not apply. Also, according to these results it is not necessary to include in the regression a term accounting for the rate of root decomposition, as was suggested by Greenwood et al. (1982). Several reasons can be provided to explain this. In the present work the last sample was taken not later than one week after onset of maturity, before the senescence period when roots die more rapidly than they are produced (Jones and Mann, 1963). Also, with direct drilling, as in our work, roots are not disturbed as much as in transplants, the system used by Greenwood et al. (1982). The value of the intercept (a) in the relationship InRL = a+b InSDW for cv. V. de Grano is higher than for the other cultivars, Staro and S.W.G. (Table 3), and shows the different strategy of this cultivar. V. de Grano initially invests clearly more energy in root growth than the other cultivars do; its root length increases in relation to shoot dry weight faster than the other cultivars. This is also true considering only exp.1 with low shoot dry weights ( < 2 g) or joining data from exp. 1 and exp. 2 during the bulbing period (first and second samplings). Under the present experimental conditions, where water and nutrients are not limiting, to have more roots in the same volume will normally not give an advantage. Futher work is necessary to assess if in less intensive systems, with lower plant densities, and under water stress conditions, the ability of V. de Grano to produce longer root system for a given shoot dry weight, especially in early stages, could give, at the end, a higher yield. According to DeMason (1990) onion has a homorhigic root system; roots are continuously produced in distinct rings at regular intervals in the onion stem and in older roots no lateral rebranching can be assumed (Weaver and Brunner, 1927; Melchior and Steudle, 1993). Thus, the increase in root length is controlled by the addition of new shoot-borne roots and by the elongation and branching of existing

131

roots. Our work shows that two weeks prior to bulbing, maximum root elongation occurs (Fig. 3), which could be considered a critical period. Root length increases quickly from 81 to 94 days after emergence and stabilizes during the last samplings. These results agree closely with those of Moore et al. (1987) working with cv. Brown Beauty in field conditions, irrigated at a - 4 0 kPa threshold. They found that the maximum root elongation rate occured 90 days after planting, while bulbing and root senescence started 96 and 115 days after planting, respectively. The initiation of the bulbing process itself occurred with an active root system (Table 1 and Fig. 3), as it was observed by Heath and Holdsworth (1948) who also stated that the emergence of new roots is suppressed at bulb development; thus the increase of root length at bulbing corresponds to the elongation of the existing roots. This root behaviour of our cultivars could also explain why in the work by Lis et al. (1967), drought applied at one-half or at 100% of the bulb's maximum weight caused a non-significant detriment to the organ's weight.

4.3. Root distribution with depth under field conditions The initial hypothesis that the rooting zone of onions is mostly restricted to about 60 cm depth is confirmed; only 2-3% of total root length was recorded below that depth in the studied cultivars. Moreover, 90% of the root system at bulbing is found in the top 40 cm of soil (Table 4), showing it is concentrated in a relatively small part of the profile. Greenwood et al. (1982), working on onions, found that ninety per cent of the root length was in the top 18 cm of soil throughout the season and the maximum rooting density in the top 15 cm was 1.0 cm.cm -3. Our maximum average values of root densities from 0 to 20 cm depth are between 8.1 and 9.1 cm.cm -3. Explanations to such large differences can be related to the bulb dry matter produced at harvest" 2.85 t ha -1 in Greenwood's experiment, indicating some limiting factors on growth, and between 11-14 t ha -~ in our experiment (Table 1). Butt (1968) stated that under conditions of limited photosynthesis (in his experiment by reducing light intensity), root was the organ most affected; that is,

a relatively greater fraction of assimilates is preferentially used in growth of the tops, while a smaller fraction is translocated to the roots. Also, the number of roots per plant was positively correlated with light intensity. Accordingly, if photosynthesis rate is increased, root growth can be favoured, probably in thickness but also in root number and length. Thus, if yield in terms of bulb dry matter increases four or five times, root density can increase in a higher proportion (Table 1 and Fig. 2). Comparing the maximum root density (Lv) in the top 20 cm of soil with data on cereals and other horticultural crops summarized by Barraclough (1989), the Lv of the present experiment is similar to cereals although, in a cereal crop, large differences in Lv, depending on growth conditions, may be found. As presented by Barraclough (1989), in irrigated winter wheat, as a consequence of 100 daydrought, Lv decreases from 12.2 cm.cm -3 to 7.3 cm.cm -3 for a shoot dry matter between 1063 g.m -2 and 1068 g.m -2, respectively, at the time of maximum root growth. However, onion roots are mostly concentrated on the top 40 cm, unlike cereals with a larger number of roots in deeper layers. Therefore, the calculations made for cereals about minimum root density in order to avoid limiting potential supply rates of nutrients to roots by diffusion may not apply for onions.

4.4. Implications for water and nutrient management From the point of view of water and nutrient management, we may thus describe drip irrigated onions as a shallow-rooting crop with 90% of roots confined to 40 cm depth, but with Lv values of 8-9 cm.cm -3 for the first 20 cm soil depth at high yield levels. Our observations of a rapid Lv increase prior to the initiation of first scales (Fig. 3) support the results of Lis et al. (1967) on cv. Valenciana transplants, who found that the beginning of bulb formation is a critical growth stage in relation to water supply. Drought occurring in this period causes a decrease in leaf number and bulb weight. Maintaining a continuous supply of water at this stage is essential in order to permit elongation of roots. Although under drip irrigation Lv and RL are

132

higher and the soil depth with high Lv are lower than the figures reported by Greenwood et al. (1982), drip irrigated onion crop has still a shallow rooting system, making compulsory a high irrigation frequency. This is because it needs a high soil water availability in a very limited soil depth, particularly during the most sensitive periods, like prior to initiation of first scales and bulbing. Nitrogen applications should be fitted to the above mentioned root pattern. The high root density would allow an immediate uptake and avoid leaching in the latest stages of growth, a period when a continous water supply is needed. In earlier stages root density is low and its pattern shallow (Fig. 3), thus a carefull splitting and timely application of nitrate fertilization will be needed. According to the rooting pattern the limited root system of young seedlings will require the placement of fertilizer close to seedling roots in order to enhance seedling growth, as has been noticed in earlier experiments (Brewster et al., 1991). The high uptake rates later in the season require potassium and phosphate to be readily available in the narrow rooting zone. Further research in this subject is needed for understanding onion growth and optimizing irrigation and fertilization scheduling. The total root length per unit of shoot dry matter (bulb+leaves) at the last sampling date, 5 August 1992, in our onion experiment ranges from 9 to 11 m.g -1. That implies, according to the data of Barraclough (1989), that onions of the present experiment have at least three times less root length per unit of dry matter produced than winter wheat. This shows that onion crop demands a higher level of soil fertility than cereals in order to sustain a given uptake rate. As can be seen in Table 2, uptake rates can be close to the maximum recorded in maize and higher than winter wheat (Barraclough, 1989). These results support the view of Greenwood et al. (1980 a, b) who stated that onions require a higher level of soil fertility than most other temperate vegetables to achieve maximum yield, and this is especially true for phosphorus and potassium in the last part of the growing period.

Acknowledgments The authors would like to thank J. Boixadera and D. Casanova for their comments on the manuscript and R. Batalla, J.R. Olarrieta and R. Teira for English corrections.

References Ajakaiye, M.B. and Greig, J.K., 1976. Response of "Sweet Spanish" onion to soil-applied zinc. J. Am. Soc. Hortic. Sci., 101: 592-596. Aung, L.H., 1982. Root initiation in tomato seedlings. J. Am. Soc. Hortic. Sci., 107: 1015-1018. Barraclough, P.B., 1989. Root growth and nutrient uptake by field crops under temperate conditions. Asp. Appl. Biol., 22: 227233. Birdsall, M. and MacLeod, R.D., 1990. Early growth of the root system in Allium cepa. Can. J. Bot., 68: 747-753. Bosch Serra, A.D., 1993. Clima. In: C. Herrero, J. Boixadera, R. Dan6s and J.M. Villar. Mapa de s61s de Catalunya 1:25000. Full n6m.: 360-1-2 (65-28) Bellvis. DGPIA and ICC, Barcelona, pp 42-44. Brewster, J.L., Rowse, H.R. and Bosch, A.D., 1991. The effects of sud-seed placement of liquid N and P fertilizer on the growth and development of bulb onions over a range of plant densities using primed and non-primed seed. J. Hortic. Sci., 66: 551-557. Brewster, J.L., 1994. Onions and other vegetable alliums. CAB International, Wallingford. 236 pp. Butt, A.M., 1968. Vegetative growth, morphogenesis and carbohydrate content of the onion plant as a function of light and temperature under field and controlled conditions. Mededelingen Landbouwhogeschool, Wageningen, No. 68-10. 211 pp. Clark, J.E. and Heath, O.V.S., 1962. Sudies in the physiology of the onion plant. V. An investigation into the growth substance content of bulbing onions. J. Exp. Bot., 13: 227-249. DeMason, D.A., 1990. Morphology and anatomy of Allium. In: H.D. Rabinowitch and J.L. Brewster (Editors), Onions and Allied Crops. Vol. 1. CRC Press, Boca Raton, pp 27-51. Goltz, S.M., Tanner, C.B., Millar, A.A. and Lang A.R.G., 1971. Water balance of a seed onion field. Agron. J., 63: 762-765. Greenwood, D.J., Cleaver, T.J., Turner, M.K., Hunt, J., Niendorf, K.B. and Loquens, S.M.H., 1980a. Comparison of the effects of potassium fertilizer on the yield, potassium content and quality of 22 different vegetable and arable crops. J. Agric. Sci., 95: 441-456. Greenwood, D.J., Cleaver, T.J., Turner, M.K., Hunt, J., Niendorf, K.B. and Loquens, S.M.H., 1980b. Comparison of the effects of phosphate fertilizer on the yield, phosphate content and quality of 22 different vegetable and arable crops. J. Agric. Sci., 95: 457-469. Greenwood, D.J., Gerwitz, A., Stone, D.A. and Barnes, A., 1982. Root development of vegetable crops. Plant Soil, 68: 75-96. Heath, O.V.S. and Holdsworth, M., 1948. Morphogenic factors as

133

exemplified by the onion plant. In: J.F. Danielli and R. Brown (Editors), Growth in relation to differentiation and morphogenesis. Symposia of the Society for Experimental Botany, II. Cambridge University Press, London, pp 326-350. Hunter, J., 1979. Hydroponics: A guide to soiless culture systems. UC Agric. Nat. Resour. Leaf., 2997. Jones, H.A. and Mann, L.K., 1963. Onions and their allies. Leonard Hill, London. 286 pp. Juan Valero de, J.A., Martin de Santa Olalla Mafias, F. and Fabeiro Cortrs, C., 1993. Efectos del riego sobre el crecimiento y los rendimientos, cuantitativo y cualitativo, de un cultivo de cebolla (Allium cepa L.). Riegos Drenajes, 73: 20-28. Leskovar, D.I. and Stoffella, P.J., 1995. Vegetable seedling root systems: morphology, development and importance. HortScience, 30:1153-1159. Lis, B.R., de Ponce, I., Covagnaro, J.B. and Tizio, R.M., 1967. Studies of water requirements of horticultural crops: II. Influence of drought at different growth stages of onion. Agron. J., 59: 573-576. Martin de Santa Olalla, F., de Juan Valero, J.A. and Fabeiro Cortrs, C., 1994. Growth and production of onion crop (Allium cepa L.) under different irrigation schedulings. Eur. J. Agron., 3: 85-92. Melchior, W. and Steudle, E., 1993. Water transport in onion (,4Ilium cepa L.) roots. Plant Physiol., 101: 1301-1315. Millar, A.A., Gardner, W.R. and Goltz, S.M., 1971. Internal water

status and water transport in seed onion plants. Agron. J., 63: 779-784. Moore, F.D., Wallner, S.J., Ells, J.E., Richwine, P.A., Bosley, D.B. and McSay, A.E., 1987. Timing of onion irrigations. CSU Exp. Stn. Tech. Bull. 26 pp. Newman, E.I., 1966. A method of estimating the total lenght of root in a sample. J. Appl. Ecol., 3: 139-145. Pardo Iglesias, A., 1990. La competencia de las malas hierbas con el cultivo de la cebolla (,,lllium cepa L.) en siembra directa: Predicci6n de prrdidas y escarda quimica. UPM-ETSIA, Tesis doctoral, Madrid. 157 pp. S.E.A., 1986. Assaigs de varietats de ceba (per deshidrataci6). DARP. Barcelona.19 pp. Soil Survey Staff, 1990. Keys to Soil Taxonomy. Tech. Mort. 6. SMSS, Blacksburg. 754 pp. Suso Martinez de Bujo, M.L., 1990. Descripci6n del crecimiento y establecimiento de una escala decimal de estados fenol6gicos de la cebolla (Allium cepa L.) de siembra de primavera. UPMETSIA, Tesis doctoral, Madrid. 214 pp. Weaver, J.E. and Brunner, W.E., 1927. Root development of vegetable crops. Mc Graw-Hill, New York. 351 pp. Zink, F.W., 1966. Studies on the growth rate and nutrient absorption of onion. Hilgardia, 37: 203-218. Zobel, R.W., 1995. Genetic and environmental aspects of roots and seedling stress. HortScience, 30: 1189-1192.

This Page Intentionally Left Blank

1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van Ittersum and S.C. van de Geijn (Editors)

135

Interspecific variability of plant water status and leaf morphogenesis in temperate forage grasses under summer water deficit J.-L. Durand a'*, F. Gastal a, S. Etchebest a, A.-C. Bonnet a, M. Ghesqui6re b aStation d'Ecophysiologie des Plantes Fourragb.res, INRA, 86 600 Lusignan, France bStation d'Am~lioration des Plantes Fourragi~res, INRA, 86 600 Lusignan, France

Accepted 1 April 1997

Abstract Leaf water status and elongation rate (LER) of five forage grasses (Lolium perenne, Lolium multiflorum, Festuca arundinacea, F.a. var. glauscescens a wild species related to F.a., and a genotype derived from a hybrid between F.a. var. glauscescens and L.m. called L4F4) were compared during two summer periods in the field in 1994 and 1995. Variations in predawn leaf water potential (ffd) indicated differences between genotypes in terms of water availability. This was consistent with neutron probe measurements which showed that L. multiflorum had a much shallower rooting system than L. perenne and Festuca arundinacea. Noon leaf water potentials (fin) of L. multiflorum remained at relatively high values despite that species' disadvantage in terms of rooting depth. The L4F4 hybrid exhibited higher ffd, a greater depth of water extraction than L. multiflorum and higher ¢/n than in F.a. var. glauscescens. The response of LER to ¢/d in the hybrid was similar to that of the Festuca and Lolium parents. By contrast with 1994, in 1995, even at high ffd, LER of rainfed Lolium perenne plants was only approximately 44% of the irrigated plants. To a lesser extent a significant year effect could also be observed in the other species. The role of the partitioning of leaf elongation between day and night in determining these differences is discussed. © 1997 Elsevier Science B.V. Keywords: Leaf water potential; Leaf elongation rate; Italian ryegrass; Perennial ryegrass; Tall fescue; lnterspecific hybridization

I. Introduction Forage production in Europe is systematically below potential during summer due to water deficit. The main effect is a reduction of leaf area, leading to decreased radiation interception (Onillon et al., 1995). Leaf elongation controls canopy leaf area expansion in frequently defoliated grasses (Nelson et al., 1977). *Corresponding author. Tel.: +33 5 49556094; fax: +33 5 49556068; e-mail: [email protected]

Improving grassland perennity without losing too much productivity is necessary. The present work focuses on growth potential under mild water deficit. Growth could be enhanced either by improving the efficiency of water acquisition or by decreasing the sensitivity of the plants to low water status. Recent efforts in plant breeding have focused on combining favourable traits of Lolium multiflorum (high feeding value and productivity) and Festuca arundinacea (drought resistance) by inter-genera hybridization (Humphreys and Thomas, 1993; Ghesqui~re et al.,

Reprinted from the European Journal of Agronomy 7 (1997) 99-107

136

1996). Comparison of the sensitivity of grasses to water deficit in the field is difficult because water acquisition interferes with dessication tolerance. To compare genotypes in different sites or different climatical situations, direct measurements of plant water status are a prerequisite for understanding which traits were modified by breeding. Predawn leaf water potential is an indication of the soil water availability (Slatyer, 1967). Leaf elongation rate (LER) responds strongly to decreasing ~kd(Barker et al., 1989; Durand et al., 1995; Puliga et al., 1996). Furthermore, water deficits often combine with nutrients limitations under natural conditions (Garwood and Williams, 1967). Nitrogen limitation was shown to have large effects on leaf elongation (Gastal et al., 1992; Onillon et al., 1995). The aim of the work reported here was (1) to assess a methodology for comparison of genotypes under field conditions, and (2) to compare the Lolium and Festuca parents with the hybrid with respect to water status and leaf elongation.

2. Material and methods The two parents studied were Lolium multiflorum L. (cultivar Adret) and Festuca arundinacea var. glauscescens Boiss., a wild genotype from the Alpes (Ghesqui6re et al., 1996). The artificial hybrid population (called L4F4) was obtained at Lusignan as described by Ghesqui~re et al. (1996). Two other related species were studied as agronomical references" L. perenne (cultivar Callan) and F. arundinacea Schreb. (cultivar Clarine). The latter was included in the study because for this cultivar the depth of water extraction, its response of LER to temperature, leaf water potential and nitrogen status were known in our conditions. Four swards (2.5 x 5 m 2) of each genotype were sown in Lusignan (mid west France) in May 1993 and 1994 on a homogenous plot. Soil characteristics were described in Ducloux and Chesseron (1988). The experimental units were set in a hierarchical design with one irrigated bloc and one rainfed. Each of them was split in two sub-blocs in which the five genotypes were randomly distributed. Irrigation was applied once a week to balance potential evapotranspiration. The same sowing density (approximately 1200 seeds/m 2) was used for all genotypes. P

and K fertilizer was applied at non-limiting rates once a year (in February) and N fertilizer was applied after each cut (120 kg/N per ha). The study was conducted the year after sowing during the summer regrowth, following defoliation to 5 cm height. The latter was carried out on 13 June in 1994 and 19 June in 1995. Climatic data (air temperature and humidity at 2 m, incident global irradiance and precipitation) were measured at an automatic weather station located at about 200 m from the plots. Using a pressure chamber, predawn and noon water potentials (fro, and ~k,, respectively) were measured on six fully expanded leaves per genotype every three or four days following the recommendations of Turner (1981). In irrigated plots, only l~d was measured. In 1995, neutron probe measurements were made in rainfed plots to determine the vertical distribution of water uptake from the soil. One tube per rainfed plot was used (two tubes per genotype). Leaf length was measured on 30 labelled tillers per genotype every 3-4 days and the mean leaf elongation rate per tiller over the 3 - 4 days period (LER) was computed. Above ground dry matter was measured by harvesting an area of 1.5 m 2 at the end of regrowth (20 and 27 July in 1994 and 1995, respectively). Total nitrogen content of the above ground dry matter was determined in ground sub-samples using an elemental analyser (Carlo Erba NA 1500). A nitrogen nutrition index (NNI) was computed following the method defined by Lemaire et al. (1989) and Gastal et al. (1992). In tall fescue growth is optimum at NNI = 1 and close to nil at NNI = 0.2 (Gastal et al., 1992). In 1995, tiller density in each plot was counted in two areas of 0.375 mE/plot. Statistical analyses were performed using the GLM and REG procedures of S AS software.

3. Results In irrigated plots, the predawn water potential, ~kd, was always higher than -0.2 MPa and averaged -0.15 MPa. In rainfed plots, l~d varied according to rainfall and year (Fig. 1a,b). The first l0 days after cutting, ~d averaged -0.5 MPa in 1994 and -0.7 MPa in 1995 (Fig. l a,b). All genotypes exhibited the same overall trends in ffd. However, significant differences appeared under the most severe drought conditions. Fes-

137 tuca arundinacea and L4F4 had the highest ~kd throughout the experiment. Lolium perenne had a lower ffd than the other genotypes on days 18, 25 and 32 in 1994, on days 7, I l, and 32 in 1995. ~bd of Lolium multiflorum was lower than ~kdof F.a. in 1995 (on days 15 and 25), but not in 1994. The sub-species F.a var. glauscescens showed significantly lower ~d than F.a on day 9 in 1994 and on days 4, 11 and 15 in 1995. In rainfed swards, leaf water potential at noon (4'n) varied between-1 and-2.5 MPa (Fig. l c,d). In contrast to ~kd,~knwas on average higher in 1995 than in 1994. Genotypic differences were also observed. Lolium multiflorum had the highest midday leaf water potential (see Fig. l d days 22 and 29 in 1995, for example). Festuca species tended to have lower

values than Lolium species and Festuca arundinacea var. glauscescens had the lowest fin values. There were clear variations in soil water extraction (Fig. 2). Between day 2 and day 15 after defoliation, both Festuca species and Lolium perenne had a similar distribution of water extraction. Although the quantities of water evapo-transpired in the plots differed, extraction reached a soil depth of 180 cm. By contrast, Lolium multiflorum hardly extracted water at more than 80 cm depth (Fig. 2b). The hybrid L4F4 extracted water down to an intermediate depth of approximately 120 cm. In irrigated and rain fed treatments, yields were higher in 1994 than in 1995 for all genotypes (Table 1). Under irrigated conditions, the relatively low yield

Days after cutting 0

5

10

15

20

Days after c u t t i n g

25

30

35

0

5

10

15

20

25

30

35

o

O.

, ~ ..o,5

-0,5

+o.=_

i h

=0 302.

20 -~ I

l

0!

-1 b

|

0

5 i ¸

10

15

20

I

I

I

25 j'

"'I

....

30

35

I

i

0 0

"0,5

-0.5 A -1.0 " M a. Z -1.S ¢

) " -2.0 -

o~

5

10

15

20

I

I

I

I

25 ' .....

I

'"

30

35

•

I

-

,~, -1.0 -

~

-1.5 -2.0

-2.5 -

-2.5 -3.0

.

Days after c u t t i n g

Days after c u t t i n g 0

.

~C

Fig. 1. Rainfall ((a,b) indicated with bars) and leaf water potential in rainfed plots of different grass genotypes after cutting in 1994 (a,c) and 1995 (b,d). (a,b) Predawn measurements, (c,d) noon measurements. Q, Lolium perenne; B, Lolium multiforum; ~7, Festuca arundinacea var. glauscescens; A, Festuca arundinacea; <>, L4F4. Error bars indicate 1 SE. The horizontal broken line in (a) and (b) indicates the mean value of predawn leaf water potential in irrigated plots. Each point represents the mean of five replicates.

138

I

I

I

I

0 ~ !,

30

30

60

6o

A

E

v0

J

i

A

E

90

O. @

a

i

Q. 0

120

0

150

9o

120

150

• L. multlflorum: 19 m m

• L. perenne: 53 m m W F. arundlnacea: 65 m m

@ L4F4:35 m m

vF. a. up. glauscescenl 45 m m

180

180 ' "" 0 1 o' % 2 o' % 3o% ' 4o% Percentage of water consumption

;

lO'%

30'

,,

Percentage of water consumption

Fig. 2. Percentage of total water extraction between 2 and 15 days after cutting in 1995 in grass swards of different genotypes. The absolute water extraction is given in mm. Each data point represents the mean of two replicates. o f L4F4 as c o m p a r e d to o t h e r g e n o t y p e s was associated to a l o w tiller density (only visually e s t i m a t e d

years in irrigated plots ( T a b l e 1). In 1995 h o w e v e r

Lolium perenne had an N N I o f 1, indicating that it

in 1994, and T a b l e 1 for 1995) b e c a u s e o f a p o o r

w a s not limited by nitrogen supply. N N I in the rainfed

e m e r g e n c y after s o w i n g . E a c h year, d r o u g h t had a

plots w a s not significantly l o w e r than in the control

d r a m a t i c effect on yield but the relative effect o f

plots.

soil w a t e r deficit w a s s t r o n g e r in 1995. Lolium per-

T i l l e r density greatly varied b e t w e e n g e n o t y p e s

enne and Festuca arundinacea var. glauscescens

(Table 1). T h e effect o f w a t e r deficit on tiller density

a p p e a r e d m o r e sensitive to w a t e r deficit than the

at the end o f r e g r o w t h w a s significant for Lolium mul-

others. H o w e v e r the particularly l o w yield m e a s u r e d in the latter g e n o t y p e in 1995 w a s partly e x p l a i n e d by its p r o s t r a t e d form, a large fraction o f the h e r b a g e b e i n g u n d e r the 5 c m cutting height. This m a d e the h a r v e s t p r o c e d u r e less efficient for that species. T h e v a l u e o f n i t r o g e n nutrition index a v e r a g e d o v e r all g e n o t y p e s w a s a p p r o x i m a t e l y 0.85 in both

tiflorum, and LnF 4 but not for the other g e n o t y p e s . H o w e v e r , the r e d u c t i o n o f tiller density i n d u c e d by d r o u g h t w a s less in L4F4. U n d e r well w a t e r e d conditions, L E R varied f r o m about 10 mrn/day in F.a var. glauscescens up to a m a x i m u m o f 40 m m / d a y in Lolium multiflorum (Fig. 3). Festuca arundinacea, Lolium perenne and LnF4 e x h i b i t e d very similar

Table 1 Dry matter yield (DM in g/m2), nitrogen nutrition index (NNI; Lemaire et al., 1989), and tiller density (in tillers/m 2) at 35 days after defoliation

Lolium multiflorum

L4F4

Lolium perenne

Festuca arundinacea

Festucaarundinacea ssp. glauscescens

(L.m. x F.a. ssp. g.)

1994 Irrigated Rain fed 1995 Irrigated Rain fed

DM yield

NNI

Tiller density

DM NNI yield

Tiller density

DM NNI yield

Tiller density

DM NNI yield

Tiller density

DM NNI yield

Tiller density

279 a 84!) 196a 27b

0.8a 0.7 a 0.8 a 0.7 a

2833a 1013b

225a 0.9 a 50b 0.9a 161a 0.9 a 18b 0.9a

_ _ 1500a 907 b

334a 0.9 a 145b 0.9 a 311 a I a 15b 0.9a

8406a 6540 a

326a 0.9 a 145b 0.9a 235a 0.9a 41 b 0.9 a

1660a 1673a

336a 0.9 a 67b 0.8 a 237 a 0.9a 7b 0.8 a

2253 a 1873b

Means of two measurements for DM and NNI, four measurements for tiller densities. a'bFor each genotype, and each year, data followed by the same letter are not significantly different (P < 0.05).

139

LER values. For all the genotypes except Lolium multiflorum, the values of LER measured in 1995 were similar to those measured in 1994. This was consistent with the variations of air temperature in both years (Fig. 3) from the beginning (15°C) to the end of the measurement periods (25°C). In 1994 in rainfed conditions F.a var. glauscescens, transiently regained the LER of the control plants. The maximum LER of other genotypes grown under water deficit was 20-80% of the irrigated plots, depending on the genotype. The reduction was more pronounced in 1995. Averaged over both years, Lolium multiglorum and L4F4 had a greater LER than Lolium perenne and Festuca arundinanceae while F.a. var. glauscescens had the lowest LER (values + SEM were 16.0 + 1.2, 14.12 + 1.1, 10.6 + 0.8, 9.6 + 1.0 and 7.0 + 0.6 mm/day, respectively).

4. Discussion

Values of ~d were generally lower in 1995 than in 1994 and this was consistent with estimates of soil water content. There was less rainfall during the experimental period in 1995 than in 1994. A simple soil water balance (Leenhardt et al., 1995) was computed for Festuca arundinacea. Assuming that the water reserve at field capacity was 200 mm (Lemaire and Denoix, 1987), the calculation indicated that approximately 30% of the reserve was used at the time of defoliation in 1994, whereas 50% of the reserve was consumed in 1995. Festuca species and Lolium perenne extracted water to a greater depth than did Lolium multiflorum in agreement with Garwood et al. (1979). The hybrid L4F4w a s intermediate, indicating that the rate and/or duration of root elongation was increased in comparison to the Lolium parent. In these two genotypes in particular, it must be noticed that due to the low tiller densities and presumably low leaf area index, a significant fraction of the water lost in the top 10 cm horizon might have evaporated, directly from the soil increasing the contribution of that region of soil. Actually, the yield values in rainfed plots suggested that leaf area index was probably less than one in all genotypes. Within that range, water consumption is approximately linearly related to LAI, so relatively small differences in LAI have a large impact on

water consumption. Final yields did not show the dynamic of herbage growth. From our visual observations, it was clear that many leaves had died earlier under the rain fed conditions especially in Lolium perenne and Lolium multiflorum. This caused an unknown fraction of the gross productivity to be lost before harvest. This is the reason why no estimate of water use efficiency was computed using these data. Indeed, over the whole growth period, or for periods of time, water consumption in Lolium perenne could have been higher than in Fesmca arundinacea. This was suggested by leaf ff measurements in Lolium perenne under rainfed conditions at 0600 h by Garwood et al. (1979). in our study, fin was often lower in both Festuca species than in the other species. Due to the higher number of tillers in Lolium perenne and because single leaf area is smaller in that species (data not shown), even at a high rate of canopy evapotranspiration, due to shading and partitioning of water, transpiration per leaf was likely to be reduced. L4F4 and Lolium multiflorum also had relatively high ~k, compared to Festuca. However, the canopy structure they formed was much more open. In such conditions, evaporative demand per leaf increases. High ~n tended therefore to indicate an efficient stomatal control. The values of LER obtained in our study were comparable to those published earlier by Norris (1982). It is essential to take into consideration the fact that under our conditions reproductive tillers were always present in Lolium multiflorum and L4F4. Reproductive growth is correlated with higher LER in Lolium perenne (Peacock, 1975) and F. arundinacea (Gastal et al., 1992). Within the temperature range observed, LER variations measured in Festuca arundinacea were consistent with the model of Gastal et al. (1992). However, the decline of LER after 25 and 28 days after cutting in 1994 and 1995 respectively was not consistent with the trend in air temperature. Since the NNI values measured on the herbage harvested at the end of the growth period were in the range of 0.8-0.9, the decrease in LER was probably the result of nitrogen deficiency (Gastal et al. 1992). L. perenne which had an NNI of 1 in 1995 did not exhibit the same decline of LER. Higher vapour pressure deficit at the end of the experiments (data not shown) might also have accounted for lower LER (Gale et al., 1970; Bunce, 1978). The rank in

140

40-

40 -

F.arundinaceae

30

3O

20

20 m, 10

I

~.~

~-

rr

t~'~ oo

~m

10

"~1

. , , , l ~ , , , I , , , W l , ~ , ; ' l

0 40

-

r~nmlwVVWl'~,,,l,,,,

0

35

| 2o

25 !

N lO ~*,vv

0

I ~,,

10

2s .- / . .

, v I , v , ~ 1 v w w ' I

20

30

"i.-.

15

E r •~ l ~

20 "" " ~

~::

,,...,"",

10

20

....

30

days after defoliation

0

I'W~'l

10

l " " la i ".i~ I ww ~ ' ~ 1 '

20

~''1

30

i

40

40

3o

~10| .j

v ....

v'''

2ot

"~

~15

0

5

40

" ~

40

I

I

10

I

0 10 20 30 10 20 30 4O F.a. ssp. glauscescens 45 - L. multiflorum

30

"0

L. perenne

I,

O.g ~'!~1:/"b ~"t~

] Ol

....

0

~ ....

10

, ....

20

, ....

30

,

40

days after defolation

Fig. 3. Mean leaf elongation rate per tiller (n = 30) and daily mean air temperatures. O, e, 1994; F-I, I , 1995. Open symbols, rain fed; closed symbols, irrigated. The vertical lines on the right of each graph indicate the maximum standard errors throughout the experiment.

LER of genotypes was generally retained in unirrigated conditions. To compare the genotypic sensitivity to limiting water availability under rainfed conditions, the ratio between LER in rainfed plots and LER in irrigated plots was computed. It was then plotted versus the average ~kd between two leaf length measurements. Mean ~kd was computed using measurements made every 3 to 4 days, and also interpolated values. There were three patterns of response: (1) In Festuca arundinacea, the response of LER to ~kd was highly significant and consistent in both years (Fig. 4a); (2) by contrast, there was no significant relationship between ffd and LER in Lolium perenne (Fig. 4b). In that species, LER was approximately 60% and 40% of the control values in 1994 and 1995, respectively. Barker et al. (1989) showed that Lolium perenne exhibited a small but significant rate of leaf elongation even at low ~'d. Van Loo (1992) showed that Lolium perenne had a LER near 60% of control at ~'d = --1.3

MPa. (3) The hybrid L4F4 and its parents exhibited a similar, significant response to decreasing ffd (Fig. 4a,b): breeding did not alter the sensitivity of elongation to soil drought. But as for Lolium perenne, at high ffd rainfed plants had a relative LER lower in 1995 than in 1994. The comparison between the five genotypes made possible by our experiment gave a general overview of interspecific differences. It is clear however that intraspecific differences of eco-physiological responses are large in grass species (Thomas, 1986) and could account for some of the differences observed. However, the similarity of the response of relative LER to ~'d for all five genotypes studied but Lolium perenne likely holds true for most of the cultivars of Lolium multiflorum, Festuca arundinacea and the hybrids. In Lolium perenne, Van Loo (1992) and Norris (1982) found the same kind of relatively high LER at low ~bd in different cultivars. This indicates that the differences found in our study also are likely characteristic of the species. Beside the genotypic classification made possible by this analysis, it is possible to infer some hypotheses on the mechanisms behind such results. In 1995, the initial soil water content and initial ~kd were lower, whereas ~kn were higher. This indicated that despite transient high ~d recovery, the low soil water availability had a lasting effect on LER in all species but Festuca arundinacea. The fact that there was a single response curve for F. arundinacea showed that LER depended more closely on l~d in that species. This suggested that the elongation probably occured preferentially during the night, as was shown under other conditions (Durand et al., 1995). Since the relationship depended on the year for the other species, three hypotheses could be proposed. (1) Leaf water potential varies during the day and the same l~d could correspond to lower ~kn in 1995 but, again, this was not confirmed by our measurements (Fig. l c,d). (2) Plant water status at dawn depends on the moistest part of the soil explored by roots. It has been established that roots growing in drier parts of the soil can emit chemical messages altering LER (Zhang and Davies, 1990). ABA concentration in the growth zone and LER were uncorrelated in barley (Dodd and Davies, 1996) but contrary results were found by Maheshwari et al. (1996) in that same species. Puliga et al. (1996) showed some genotypic variations in a significant negative relationship between ABA concentration in

141

the mature leaves and LER. (3) Drought induces a shrinkage of the leaf growth zone of grasses (Spollen and Nelson, 1994; Durand et al., 1995). Following rewatering, the recovery of the maximum length of the growth zone is not immediate (Durand et al., 1995). The growth zone of plants was hence probably shorter in 1995 than in 1994, due to the more pronounced initial soil water deficit. Furthermore, Thomas et al. (1995) have shown that the drought-induced shrinkage of the elongation zone was more pro-

nounced in Lolium multiflorum than in Festuca arundinacea. Hypotheses (2) and (3) are compatible.

5. C o n c l u s i o n

A complete analysis of drought effects on yield was not the main purpose of that study. It clearly appeared that yield and LER are not simply linked. Indeed, the effects of water deficit on leaf senescence rate, flower-

1.0 A

- 1 ira

r"

AA

o.a m

e

:O

o.e3

-

t,,,,,

ill w4,

w

- 0.6~"

v o

o

m ml

-

0.4

mm

........................

0.2 o0

/

-

-0.8

-0.6

~ I

....

I

-0.4

-0.2

O

0.2 m

o

o

0

....

0

o

9.

I

I

0

-.8

-0.6

Mean Yd (MPa)

I

0

I .....

-0.4

-0.2

0

Mean ¥ d (MPa) 1.0

1.0

I, v

d

m :O

o.e

tin 0.8 :D

g

pB~

<

X."

0.6

~

-

0.6

........ ¢ ........ "~'~'" v v n

v

a_ o 0 0.2 =

.......--'""

0 T.~

-

0.4D Q

Q. o - 0.2 °

rl

0 !

i

-0.6

"'1

. . . . . . .

-0.4

Mean ¥d (MPa)

I

-0.2

.

.

.

.

0

.

•

o

V V<~... [:] 0....... "~- ' ¢ d b ~

0.4 ~

-0.8

o o o

O.4 o

•............

(I)

/

,9o

)

~

0.e

om .

I

-0.8

-0.6

'

I'

'

.0.4

I

.0.2

' '

-

0

0

Mean Yd (MPa)

Fig. 4. Ratio of leaf elongation rate in rainfed plots to elongation rate in irrigated plots as a function of the mean predawn leaf water potential. (a) Festuca arundinacea (A, &); (b) Lolium perenne (O, e ) ; (c) Lolium multiflorum (El, I ) ; Festuca arundinacea vat glauscescens (~7, V), and hybrid L4F4 ((>, • ) in 1994, (d) same as in (c) but 1995. (a-d) Closed symbols, 1994; open symbols, 1995. The lines are drawn from linear regression of Y on X.

142

ing and tillering are as important in some of these species as its effect on LER. The method presented in that study allowed the ranking of the genotypes in terms of sensitivity of LER to soil water deficit. Actually, ffd could probably be used in mixed canopies and natural grasslands where estimates of soil water availability for different species is very difficult if not impossible. The five genotypes of temperate grasses differed in their ability to extract water, to control transpiration and their response to leaf water status. Some favourable traits associated with these functions appeared to be assembled in the phenotype of LaF4. This optimistic conclusion must be qualified by the fact that the lower leaf area resulting from flowering and sensitivity to water deficit partly explained the maintenance of a favourable plant water status. Some important questions remain regarding leaf water status, dependant daily variations of LER. Mechanistic studies on the structure and physiology of the growth zone of leaves are needed to explain the lasting effects of soil water deficit on LER and the contrasted response between Lolium perenne and the other species. Comparative studies under controlled conditions, where temperature and ¢'d would be kept at constant values could bring only to firm conclusions.

Acknowledgements The present work was supported by the R6gion Poitou-Charentes (XI th contrat de plan).

References Barker, D.J., Chu, A.C.P. and Korte C.J., 1989. Ryegrass herbage yield components and their response to water deficit stress. In: XIV International Grassland Congress, The French Grassland Society, Association Franqaise pour la Production Fourrag~re. Versailles, pp. 503-504. Bunce, J.A., 1978. Effects of water stress on leaf expansion, net photosynthesis, and vegetative growth of soybeans and cotto. Can. J. Bot., 56: 1492-1498. Dodd, I.C. and Davies, W.J., 1996. The relationship between leaf growth and ABA accumulation in the grass leaf elongation zone. Plant Cell Environ., 19: 1047-1056. Ducloux, J. and Chesseron C., 1988. Les terres rouges ~ ch~taigners de l'ouest de la France (contribution/t l'6tude de leur gen6se). Ann. Soc. Sci. Nat. Charente Maritime, 7: 853-868.

Durand, J.L., Onillon, B., Rademacher, I. and Schnyder, H., 1995. Drought effects on cellular and spatial parameters of leaf growth in tall fescue. J. Exp. Bot., 46:1147-1155. Gale, J.R., Naaman, A. and Poljakoff-Mayber, A., 1970. Growth of Striplex halimus L. in sodium chloride salinated culture solutions as affected by humidity of the air. Aust. J. Biol. Sci., 12: 947952. Garwood, E.A. and Williams, T.E., 1967. Growth, water use and nutrient uptake from subsoil by grass swards. J. Agric. Sci. (Camb.), 69: 125-130. Garwood, E.A., Tyson, K.C. and Sinclair, J., 1979. Use of water by six grass species. 2. Root distribution and use of soil water. J. Agric. Sci. (Camb.), 93: 25-35. Gastal, F., B :61anger, G. and Lemaire, G., 1992. A model of the leaf extension rate of tall fescue in response to nitrogen and temperature. Ann. Bot., 70: 437-422. Ghesqui~re, M., Emile, J.-C., Jadas-H6card, J., Mousset, C., Traineau, R. and Poisson, C., 1996. First in vivo assessment of feeding value of Festulolium hybrids derived from Festuca arundinacea var. glauscenscens and selection for palatability. Plant Breeding, 115: 238-244. Humphreys, M. and Thomas, H., 1993. Improved drought resistance in introgression lines derived from Lolium multiglorum x Festuca arundinacea hybrids. Plant Breeding, 111: 155-161. Leenhardt, D., Voltz, M. and Rambal, S., 1995. A survey of several agroclimatic soil water balance models with reference to their spatial application. Eur. J. Agron., 4: 1-14. Lemaire, G. and Denoix, A., 1987. Croissance estivale en mati :6re s6che de peuplements de ft~tuque 61ev6e (Festuca arundinacea Schreb.) et de dactyle (Dactylis glomerata U) dans l'ouest de la France. II. Interaction entre les niveaux d'alimentation hydrique et de nutrition azotEe. Agronomie, 7: 381-389. Lemaire, G., Gastal, F. and Salette, J., 1989. Analysis of the effect of N nutrition on dry matter yield of a sward by reference to potential yield and optimum N content. In: XIV International Grassland Congress, The French Grassland Society, Association Franqaise pour la Production Fourrag6re. Versailles, pp. 179180. Maheshwari, M., Munns, R. and Chandler, P.M., 1996. The effect of water stress on leaf elongation, ABA accumulation and XET gene expression in barley (Hordeum vulgare L.). 2nd International Crop Science Congress, New Delhi, Abstracts, 137 PP. Nelson, C.J., Asay, K.H. and Sleper, D.A., 1977. Mechanisms of canopy development of tall fescue genotypes. Crop Sci., 17: 449-452. Norris I.B., 1982. Soil moisture and growth of contrasting varieties of Lolium, Dactylis and Festuca species. Grass For. Sci., 37: 273-283. Onillon, B., Durand, J.-L., Gastal, F. and Tournebize R., 1995. Drought effects on growth and carbon partitioning in a tall fescue sward grown at two rates of nitrogen fertilization. Eur. J. Agron., 4: 91-99. Peacock, J.M., 1975. Temperature and leaf growth in Loliurn perenne. III. Factors affecting seasonal differences. J. Appl. Ecol., 12: 685-697.

143

Puliga, S., Vazzana, C. and Davies, W.J., 1996. Control of crop leaf growth by chemical and hydraulic influences. J. Exp. Bot., 47: 529-537. Slatyer, R.O., 1967. Plant Water Relationships. Academic Press, London, 366 pp. Spollen, W.G. and Nelson, C.J., 1994. Response of fructan to water deficit in growing leaves of tall fescue. Plant Physiol., 106: 329336. Thomas, H., 1986. Water use characteristics of Dactylis glomerata L., Lolium perenne L., and Lolium multiflorum Lam. plants. Ann. Bot., 57:211-223. Thomas, H., James, A.R. and Humphreys, M.W., 1995. Leaf

growth and water relations in Lolium and Festuca. Proceedings of Interdrought 96, Montpellier, VIII, 21 pp. Turner, N.C., 1981. Techniques and experimental approaches for the measurements of plant water status. Plant Soil, 58: 339366. Van Loo, E.N., 1992. Tillering, leaf expansion and growth of plants of two cultivars of perennial ryegrass grown using hydroponics at two water potentials. Ann. Bot., 70:511-518. Zhang, J. and Davies, W.J., 1990. Changes in concentration of ABA in xylem sap as a function of soil water status can account for changes in leaf conductance and growth. Plant Cell Environ., 13: 277-285.

This Page Intentionally Left Blank

1997 ElsevierScience B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

145

Evaluation of sunflower (Helianthus annuus, L.) genotypes differing in early vigour using a simulation model F. Agtiera a'*, F.J. Villalobos a'b, F. Orgaz a alnstituto de Agricultura Sostenible, Consejo Superior de Investigaciones Cientificas, Apartado 4084, C6rdoba, Spain bDepartamento de Agronomfa, Universidad de C6rdoba, Apartado 3048, C6rdoba 14080, Spain Accepted 13 June 1997

Abstract High early vigour (plant dry matter in the early development stages) in sunflower may be a desirable character under waterlimited environments as it may contribute to higher transpiration efficiency and reduced soil evaporation. However a high early vigour causes a more rapid use of soil water, which may threaten crop water supply during seed filling. This is also influenced by the seasonal pattern of rainfall and genotype season length. The objective of this research was to simulate the performance of cultivars of sunflower differing in season length and early vigour under a Mediterranean climate. A simulation model of the sunflower crop, OILCROP-SUN, was modified and used to find the optimum combination of early vigour and season length of the genotypes for different environments. Field experiments were carried out during 1992, 1993 and 1994 in Cordoba, Spain (38°N). Sunflower populations with similar genetic background but with differences in early vigour were used to study the association of this trait with other characters and with the genetic parameters required to run the model. Changes in early vigour were simulated by modifying leaf growth rate in the model. The field experiments showed that high early vigour, as measured by stem volume 425°Cd after emergence, is not related to reduced root growth and is not associated with season length. Simulations showed that the highest seed yield is obtained using genotypes with high early vigour, provided that their season length is long enough for the growing conditions. © 1997 Elsevier Science B.V. Keywords: Sunflower; Early vigour; Simulation model

I. Introduction

B = E T x WUE

(1)

Hence: Passioura (1977) proposed that dry matter accumulation (B) of crops could be analyzed as a function of evapotranspiration (ET) and water-use efficiency (WUE):

*Corresponding author. Departamento de lngenierfa Rural, Escuela Polit6cnica, Campus Universitario, Almerfa 04120, Spain. E-mail: [email protected]

Y = E T x WUE x HI

(2)

where Y is seed yield and HI is the harvest index. This model has become a framework to examine ways to improve crop yields in water-limited environments (Cooper et al., 1983; Ludlow and Muchow, 1990; Sadras and Villalobos, 1994). Under such conditions high early vigour (EV) may be a desirable character as it may contribute to higher transpiration

Reprinted from the European Journal of Agronomy 7 (1997) 109-118

146

efficiency (TE) and reduced evaporation (E) from the soil surface, increasing WUE and hence, dry matter accumulation (Loss and Siddique, 1994). However, high EV causes a more rapid use of soil water, threatening crop water supply during seed filling. The threat also depends on the seasonal pattern of rainfall (amount and distribution) and cultivar season length. Thus HI may be reduced for high EV, long-season genotypes, especially in dry years. Therefore, to define the ideotype (Donald, 1968) for a specific environment we need to find the optimum combination of EV and genotype season length, knowing the environmental characteristics (principally seasonal rainfall amount and distribution), and knowing if EV is related to other characters. Evaluation of EV in cereals has been carried out by different methods. In most studies visual scores of plant aspect were used (Turner and Nicolas, 1987; Boukerrou and Rasmusson, 1990; Ceccarelli et al., 1991; Regan et al., 1992). In other studies, EV was related to ground cover assessed from photographs (Whan et al., 1991; Regan et al., 1992), or via measurements of crop reflectance for specific wavelengths (Elliot and Regan, 1993). In sunflower, Boujghagh (1994) evaluated EV using a visual score. Crop simulation models have been used for different purposes (Loomis et al., 1979). One important .~ 40~. 30. ~ 20w ~ 10I11 F0 -30 "7>,

80-

"10

A

E E 60-

E -20

.--I ,.,..I

z

z O m

40-

~ 20-

-10 o I

1

J

F M A M

1

I

I

I

I

I

J

J

A

I

I

I

I

S O N D

MONTH OF YEAR

Fig. 1. Mean maximum (k) and minimum (A) temperatures, monthly rainfall (O) and total solar radiation (0) at Cordoba, Spain (38°N), averaged for 27 years of data (1966-1992).

aspect has been to study the effects of a character on yield, provided that the model considers interactions with other characters and with the environment (Austin, 1993). For example, Austin (1982) used a wheat simulation model to study the effects of leaf orientation on yield. Fereres et al. (1993) and Sadras and Villalobos (1994) used a sunflower simulation model to study the effects of genotype season length on yield. However, in these last two studies the association of season length with other characters was not included. Our objectives were to study the association of EV with other characters in sunflower, and to find the optimum combination of EV and season length for different environments using a simulation model.

2. Materials and methods

2.1. Field experiments All field experiments were carried out at the Agricultural Research Center of Cordoba, Spain (38°N, 4°W). The soil is a deep sandy-loam, classified as Typic Xerofluvent. Cordoba has a Mediterraneantype climate with high temperatures in the summer and most rainfall concentrated between late autumn and spring. Fig. 1 shows the maximum and minimum temperature, rainfall and solar radiation averaged over 27 years. In all field experiments fertilizer was applied before planting at rates of 55 kg/ha of N as NH4, 55 kg/ha of P205 and 55 kg/ha of K20. An additional 50 kg/ha of N as NO3 was applied before anthesis. Weeds were controlled with a presowing application of trifluralin and by hand during the experiment. Several furrow irrigations were applied during the season to avoid crop water stress. The B-FSS-88 sunflower population (FernandezMartinez et al., 1990) was used. This population has a broad genetic base showing high plant-to-plant variability in season length, height and other morphological traits. On January 29th 1992, B-FSS-88 seeds were sown and grown under non-limiting conditions. A sample of 32 plants was harvested 425°Cd after emergence (base temperature (Tb) = 4°C; Villalobos and Ritchie, 1992). On this date, stem volume, calculated from stem diameter and stem height, the number

147 of leaves, total plant leaf area (equal to green leaf area at this estage), and above ground dry matter (leaves, stem and petioles, separately) were determined for each plant by drying at 60°C until constant weight. EV was assumed as above ground plant dry matter 425°Cd after emergence. On the same date stem diameter and stem height were measured in 557 plants grown in a plot divided into subplots according to the grid method (Gardner, 1961) to reduce the effect of environmental variation. At flowering, 10% of higher and 10% of lower EV plants from every subplot were selected using the criteria of stem volume and both groups were recombined independently to obtain two new populations. These new populations, HV from high EV plants and LV from low EV plants, together with starting population, B-FSS-88, were sown on 10 February 1993, 14 February and 14 April 1994 under nonlimiting conditions. Stem volume at approximately 430°Cd after emergence and date of first anthesis were recorded for every plant. A total of 2822 plants in 1993, 1383 at the first sowing date of 1994 and 863 at the second sowing date of 1994 were evaluated. Populations were divided into groups of different season length (emergence to first anthesis period): three groups in 1993, four for the first sowing date in 1994 and three groups for the second sowing date of 1994. A sample of each group was harvested at physiological maturity to determine numbers of seed per head.

2.2. Pot experiments Seeds of LV and HV were sown in 1993 (26 May) and in 1994 (14 January) in 301 pots placed in the field and filled with a sandy-loam soil. Plants were harvested 670°Cd (Tb =4°C) in 1993 and 527°Cd (Tb = 4°C) in 1994 after emergence and the above ground and root dry matter were determined.

2.3. Model development The analysis was performed using a modified version of OILCROP-SUN (Villalobos et al., 1996), a model of the development, growth and yield of the sunflower crop which takes account of soil water and nitrogen balances. The model uses a minimum set of weather and soil data and five cultivar-specific genetic

parameters: three related to the development and two related to yield. The duration of the emergence-first anthesis period (tl, days) is calculated as: tl=

+ [(15-HxP2] +

[

~.-~, 112J

(3)

where P1 (thermal time from emergence to end of juvenile phase) and P2 are two genetic parameters. P2 represents the amount (in days/h) that development is slowed when crop is grown in photoperiod shorter than the optimum, which is considered to be 15 h. H is the photoperiod (hours) at the end of the juvenile phase, TT~ is the average daily thermal time (Tb = 4°C) in the emergence-end of the juvenile phase, and Tr2 is the average daily thermal time from the start of floral initiation to first anthesis. The duration of the first anthesis to physiological maturity period (t2, days) is determined as follows:

t2 =P5/17"

(4)

where P5 is another genetic parameter and TT is the average daily thermal time of this period. Seed number was estimated from the maximum possible number of seeds per head (genetic parameter, G2). The genetic parameter G3 represents the potential kernel growth rate during the linear kernel filling phase (mg/day). Both G2 and G3 should be measured in plants grown under optimum conditions and low plant population density. Modifications of OILCROP-SUN in order to simulate sunflower genotypes differing in EV used data from the field and pot experiments. Above ground dry matter and root dry matter of both pot experiments were positively and significantly (P < 0.01) correlated (Fig. 2). Linear regressions were y = 0.65x and y = 0.99x (x = above ground dry matter, g/plant; y = root dry matter, g/plant) for the 1993 and 1994 experiments, respectively. Relationships between above ground plant dry matter and stem diameter, stem height, leaf number, leaf area and stem volume in the 1992 field experiment were highly significant in all cases (Table !). Stem volume was used to estimate early vigour as it is a rapidly measured index, which is advantageous when many plants have to be screened. Linear regressions between season length and stem volume in the 1993 and 1994 field experiments

148

et al. (1993), who found a negative correlation between EV and specific leaf area in barley. Data in Table 1 show that EV was related both to leaf area and leaf number 425°Cd after emergence, which means that high EV was associated with high leaf growth rate.

8

7-

a)

O

65-

Oo

0

4c

N 2~lIx 0 w

-t

t

I

I

I

1

2.4. Modifications in the model

i

0 1 2 3 4 5 6 7 8

>aIx: 6I-- 5 O O 4IX 321-

b)

d

0

I

012:34

•

'

~i 6 -~ 8

ABOVE GROUND DRY MATTER (g.plant "1)

Fig. 2. Relationships between above ground and root dry matter for the 1993 (a) and 1994 (b) pot experiments. Each point represents data from an individual plant. Fitted linear regressions were y = 0.65x (R = 0.68**, 1993) and y = 0.99x (R = 0.81"*, 1994). **P < 0.0l. showed negative, significant (P < 0.01) correlation coefficients and broad variability for both characters. However, regression coefficients were very low (--0.018 day/cm 3 in 1993,-0.057 day/cm 3 in first sowing of 1994 and --0.043 day/cm 3 in second sowing of 1994), and for every value of stem volume a broad range of season length was observed. Differences in stem volume were not related to differences in specific leaf area measured at the same date (R = 0.23, P = 0.20). Therefore the higher leaf area of high EV plants was not due to thinner leaves. These results differ from those obtained by Richards

According to the results of our experiments, differences in EV were included in the model by multiplying leaf dry matter accumulation rate and leaf expansion rate by a coefficient (LC). The value of this coefficient was determined using the 10% of plants with the highest stem volume and the 10% of plants with the lowest stem volume in the 1992 field experiment. Variances were equal between the subgroups (P < 0.05). The average stem volume of each sub-group was calculated. The corresponding dry matter was estimated using the linear regression fitted between stem volume and above ground dry matter (Table 1). Thereafter, the model was modified (as described below) testing different LC's to obtain in the simulations equal high and low estimated dry matter 425°Cd after emergence. The coefficients obtained were 0.4 for low EV and 1.6 for high EV. This modification allowed simulating genotypes differing in EV, without a substantial change in dry matter partitioning between shoot and root. Therefore no more changes in the model were required. The range from minimum (0.4) to maximum LC (1.6) was divided into seven classes to simulate seven EV levels with LC's of 0.4, 0.6, 0.8, 1.0, 1.2, 1.4 and 1.6. To estimate the genetic parameters the data from the experiments and several assumptions described

Table 1 Linear regression of above ground dry matter (y, g/plant) on several morphological indexes determined 425°Cd after emergence in the 1992 field experiment (y = a + b x x) Index (x)

a

b

R2

Stem diameter (mm) Stem height (cm) Leaf number Leaf area (cm2) Stem volume (cm 3)

-5.98 (1.100)** -0.49 (0.079)** -5.51 (1.928)** -0.08 (0.009)** 1.77 (0.378)**

1.04 (0.098)** 0.34 (0.059)** 0.69 (0.120)** 0.01 (0.003)** 0.20 (0.018)**

0.79** 0.52** 0.52** 0.96** 0.80**

SEs are shown in parentheses. **P < 0.01.

149

below were made. Groups of season length made in the populations of field experiments of 1993 and 1994 showed a wide range in the emergenceanthesis period (from 71 to 93 days). Three of them were taken to represent short, medium and long season length, with 74, 84 and 93 days, respectively. With these values and assuming P2 equal to 2 days/ h, which is close to that of most of the genotypes calibrated previously, P1 values were 200, 275 and 350°Cd for short, medium and long season length respectively. For all season length groups, P5 was considered equal to the average of previously calibrated genotypes (P5 = 650°Cd). The parameter G2 was estimated using the positive and significant (P < 0.05) linear regression fitted between seed number per head and stem volume for every selected season length group of the field experiments. G3 was estimated for each genotype by the significant (P < 0.05) linear regression fitted between G2 and G3 for genotypes calibrated previously: G3 = 3.07-4.33 x 10 -4 x G2

(5)

With this relationship the model may seem overparameterized. However this equation gives only a estimation of G3, which is useful in this work. When the model is used to simulate seed yield of a particular cultivar, its G3 parameter is necessary because the model is very sensitive to this parameter. Considering these results, simulations were run choosing 21 genotypes (seven levels of EV and three different season lengths). The parameters of all these genotypes are given in Table 1. We have assumed the same sensitivity to photoperiod and the same seed filling period for all genotypes because we have not calibrated plants differing in EV. We have data of the other genetic parameters, which have been considered.

1 March) and two different soils depths (water holding capacities of 165 and 300 mm respectively).

3. R e s u l t s

Average first anthesis dates in the simulations were 15 April, 5 May and 22 May for short, medium and long season lengths, respectively, at the first sowing date, and 16 May, 31 May and 13 June at the second sowing date.

3.1. Seed yield Fig. 3 shows the relationships between average seasonal seed yield and average EV for all cases studied. Short season cultivars showed a positive yield response to increased EV for all conditions. For shortseason cultivars lowest yields were similar in all cases (ca. 500 kg/ha) but the highest yields varied with soil depths (ca. 1300 kg/ha in deep soil and ca. 1000 kg/ha in shallow soil). A similar response of yield to variation in EV was found for medium season cultivars except for the case of shallow soil and late sowing date, which did not show any response. For this type of cultivar a different range of yield was observed for each environment. With early sowing on a deep soil, simulated yield ranged from 1250 to 2900 kg/ha. For late sowing date on a deep soil, the lowest yield was "

3000

25002 ~ 2000,_. 1500- / ¢'" 1000-

~ D ..i

l--m

t

'

tu 30004 >Q 2 5 0 0 qJ "' 2000-~ w

I

o.........

" Si

I

..... "" b) deep, 1/3~/

, .

,

i

i

'i'

i

1

i

i

----i i 'i I I

....

c) shallow, 111

1500--I

~/- m ....

2.5. Simulations ' 2

I

I

e

a)deep.11

,,1,

m

/

° .."°"

10001"! / s . ..... s

The simulations were performed for 20 years of daily weather data of Cordoba (Spain). We assumed a wheat-sunflower rotation, 7.1 sunflower plants/m 2 and rainfed conditions, on a medium texture soil. Combination of three season lengths and seven levels of EV were tested on two sowing dates (1 January and

]I

....... s

500....... 0

J

i 4

i 6

i 8

,

i i 101214

t

2 4 6 8 101214

EARLYVIGOUR(g.plant"1) Fig. 3. Relationships between average seed yield and average EV of sunflowerat Cordoba (Spain) in the simulations. Deep soil (a,b), shallow soil (c,d). Sowing date Ill (a,c) and 1/3 (b,d). Dotted line, short season genotypes; dashed line, medium season genotypcs; solid line, long season genotypes.

150

2.4

E

2.0

~

m

1.6

; ...........

'

1.2 . ...." a) p, 1/1~ 0.8 , m m , I l

'"~ 2.4 2.0

I

i ....;" i

c) shallow, 1/1 I

d) shallow, 113!

//~m

I

1.6

1.2 0.8

//

...

S

f_-....'..S;...... ' 1

....................

, i i l i I'" i' 'l i i I 2 4 6 8 10 12 14 2 4 6 8 101214 EARLY VlGOUR (g.plant -1) i

Fig. 4. Relationships between average WUE (dry matter at maturity/ET from emergence to maturity) and average EV of sunflower crop at Cordoba (Spain) in the simulations. Deep soil (a,b), shallow soil (c,d). Sowing date 1/1 (a,c) and 1/3 (b,d). Dotted line, short season genotypes; dashed line, medium season genotypes; solid line, long season genotypes.

similar to that of the early sowing date, but the upper yield was lower at 2500 kg/ha. On the shallow soil, at the first sowing date, yield ranged from 750 to 1500 kg/ha. At the second sowing date yield was constant (ca. 500 kg/ha). The response of yield to EV was more complex in the case of long season cultivars. For undroughted conditions, as is the case of deep soil and first sowing date, an intermediate EV resulted in a yield similar to a high EV (2600 kg/ha). For the other cases the response of yield to EV was very small. On the deep soil at the second sowing date yield was ca. 1250 kg/ha. On the shallow soil, yield was ca. 500 and ca. 250 kg/ha for the first and the second sowing dates, respectively.

3.2. Water-use efficiency Fig. 4 shows the relationships between average WUE in terms of above ground dry matter at maturity, and average EV. In all cases, for a given value of EV, longest season genotypes had the highest value of WUE, and for a given genotype season length, the WUE increased with EV. Furthermore, the early sowing date showed a higher WUE than the late sowing date. On the deep soil and at first sowing date, WUE ranges were 1.00-1.40, 1.40-2.00 and 1.70-2.30 g/ dm 3 for short, medium and long season length genotypes, respectively. These values were smaller at the

second sowing date (0.90-1.20, 1.20-1.80 and 1.602.00 g/dm 3 for short, medium and long season genotypes, respectively). On the shallow soil the ranges in WUE decreased from the first (0.90-1.30, 1.20-1.80 and 1.40-2.20 g/dm 3) to the second sowing date (0.80-1.20, 1.20-1.55 and 1.20-1.60 g/dm 3) for short, medium and long season length genotypes. To investigate the behaviour of WUE, the relationships between the transpiration efficiency (TE; dry matter at maturity/transpiration from sowing to maturity) and EV, and the relationships between the transpiration/ET (T/ET) ratio and EV were obtained. Fig. 5 shows that high TE was related to a high EV, except for short season length genotypes, which showed a small response to increased EV. Taking into account the equation given by Tanner and Sinclair (1983), which relates dry matter production (B) and transpiration (T): B - K x T/VPD (where K is a coefficient depending of species and environmental conditions, and VPD is the vapour pressure deficit), is easy to understand that early planting date increased TE in all the cases studied due to the lower VPD values during its growing period. Ranges of TE were similar in both soil depths for a given combination of sowing date and genotype season length. For short season genotypes at the first sowing date, the range in TE was ca. 2.80-2.50 g/dm 3, while at the second sowing date it showed a constant value of ,~ 3.6 'E 3.2 ~

¢0

~ >oz w iT', II i,i z 0

b) deep, 1/3

t

2.8 2.4

/"-ms

2.0 i

a) deep, 111~ i w i i !

36 3.2-. ~ , - - - m

;

I

I

I

I

" i t

==. 2.8-. // n ":L.. . . . . . . . . s (n 2.4..... z ~; 2.0- c)shallow, 1/1

2

I

1'o1'214

d) shallow, 113 ,,

_tl-- 1 I

i

i

s l

i

I

2 4 6 8 101214

EARLY VIGOUR (g-plant "1) Fig. 5. Relationships between average TE (dry matter at maturity/T from emergence to maturity) and average EV of sunflower crop at Cordoba (Spain) in the simulations. Deep soil (a,b), shallow soil (c,d). Sowing date l/l (a,c) and 1/3 (b,d). Dotted line, short season genotypes; dashed line, medium season genotypes; solid line, long season genotypes.

151

o.7-1

,I

°.fi'.. ..........'

/

,

./

0.4

-"" I": a) deeP' 1/11 0.3, i i i r i I

0.7 ~

0.3

i

b) deep, ~ i

I

I

, 1'I

'"C) shallOW' 1i

o.6o.~: 0.4£

/ I

...."" ......s/

.., .Y" ...... ..r""~

I

soil (ca. 0.30 to ca. 0.35 for both sowing dates), whereas in the shallow soil this response was very small at the first sowing date (ca. 0.24 for all values of EV) and negative at the second sowing date (0.260.12). Long season genotypes showed a negative response of HI to EV on the deep soil (0.32-0.26 and 0.30-0.20 at the first and second sowing dates, respectively), and a very small response in the shallow soil (0.14-0.12 and 0.10-0.08 at the first and second sowing dates, respectively).

........s

, , l , 2 4 6 8 101214

I

,

I

,

,

,

I

2 4 6 8 101214

4. Discussion

EARLY VIGOUR (g.plant -1) Fig. 6. Relationships between average T/ET ratio and average EV of sunflower crop at Cordoba (Spain) in the simulations. Deep soil (a,b), shallow soil (c,d). Sowing date 1/1 (a,c) and I/3 (b,d). Dotted line, short season genotypes; dashed line, medium season genotypes; solid line, long season genotypes.

ca. 2.20 g/dm 3. Medium season length genotypes showed ranges of ca. 2.40-3.10 and ca. 2.00-2.60 g/dm 3 at the first and second sowing dates, respectively. Long season genotypes TE's ranges were ca. 2.60-3.40 and ca. 2.20-2.65 g/dm 3 at the first and second sowing dates, respectively. The T/ET ratio (Fig. 6) showed a positive response to increased EV and genotype season length, indicating that high EV contributes to reduced soil evaporation, although the impact decreased as season length and thus, seasonal ET increased. Medium and late genotypes showed similar values in all cases, ranging from ca. 0.50 to ca. 0.67. Short season genotypes presented a wider range in T/ET ratios with values from ca. 0.32 to ca. 0.57. For all genotypes the T/ET ratio was higher in the second than in the first sowing (Fig. 6). 3.3. Harvest index

Relationships between average HI and average EV for all the conditions studied are shown in Fig. 7. Short season genotypes had a positive response of HI to increased EV for all the cases, except for the shallow soil and late sowing date, where the response was very small. On the deep soil, HI ranges were 0.25-0.35 and 0.28-0.36 at the first and second sowing dates, respectively. Medium season length genotypes showed a positive response of HI to EV for deep

The fact that above ground dry matter was positively correlated with root dry matter means that selection for high EV will not be associated with indirect selection for a poor rooting system. Differences in these relationships between the two pot experiments may be caused by differences in the harvest dates. An increase in the above ground dry matter/root dry matter ratio related with the development of the sunflower plant has been observed by Trapani et al. (1992). Regression coefficients in the linear regressions between season length and stem volume were very low and for every value of stem volume a broad range of season length was observed. Therefore, the shortening of season length produced by the increase 0.36 0.32~

/--.-'~.--" m I

m/'--.~--~.7 .....

o.24-t "-

0.20-] 0.16-1 0.12d 0.08~

i

I

a) deep, 1111 i I I I

0.28-1 .............s 0.24-1 ...."~ . . . . 0.20-1 Z 0.16-]

I

i

b) deep, 113I I I I I

I

!

0"08/ , i i i 'i J 2 4 6 8 101214

-!

"---x

m I

I J, i :--1 ! r 1 2 4 6 8 101214

EARLY VIGOUR (g.plant "1) Fig. 7. Relationships between average HI and average EV of sunflower at Cordoba (Spain) in the simulations. Deep soil (a,b), shallow soil (c,d). Sowing date 1/l (a,c) and 1/3 (b,d). Dotted line, short season genotypes; dashed line, medium season genotypes; solid line, long season genotypes.

152

of EV was very small and a selection for EV should not be related to a reduction in season length. These results are not in agreement with those of Gimenez and Fereres (1986), who found than EV was inversely correlated with season length, although they used sunflower genotypes with a narrow genetic base. A similar result to Gimenez and Fereres (1986) was found for Ceccarelli et al. (1991) with different oat genotypes. However, Regan et al. (1992) did not find a similar relation in wheat. In all the cases in the simulations, maximum yield corresponded to highest EV, independently of season length. The range of WUE values in these experiments (0.80-2.3 g/dm 3) was wider than that found by Soriano et al. (1994) for sunflower under different irrigation treatments: 1.77-2.85 g/dm 3, or by Orgaz et al. (1990): 1.16-1.36 g/dm 3, and Gimeno et al. (1989): 1.4-2.5 g/dm 3, at winter and spring sowing dates. The increase in WUE in response to EV and season length was due to a combined improvement of TE and the T/ET ratio. The higher TE of high EV plants is related to the ability to grow faster under low temperatures when VPD is low (Tanner and Sinclair, 1983), as is the case in early sowing dates. Higher VPD following late sowing dates reduces differences among plants differing in EV. Higher TE values in short season genotypes with low EV is not evident when expressed in terms of glucose instead of dry matter (data not shown). High dry matter early in plant development improves the T/ET ratio as a larger ground area is covered by the crop, hindering the passage of radiation to the soil and decreasing E from the soil surface when there is a high probability of the soil being wet by rainfall. Gregory et al. (1984) and Cooper et al. (1987) observed, in cereals, an increase in yield associated with a faster early growth and reduced evaporation from the soil surface. For a given EV, lower T/ET values of short season genotypes may be due to the seasonal change in this ratio which is low at the beginning of the growing season and increases thereafter. Thus, short season genotypes have a shorter time of high T/ET values and the final value is lower than that of long season genotypes. Similar effects to those observed with high EV may be achieved with crop management techniques. For example with early sowing dates, crops will grow under low VPD and TE will increase, although this

early sowing date is limited by slow germination rates at low temperatures, which may cause the incidence of diseases (Gimeno et al., 1989). The T/ET ratio may be improved by increasing fertilizer application, which will have a positive effect on crop growth, or by increasing planting density (Anderson, 1992). With all these techniques, adequate season length cultivar should be selected to avoid water stress during seed filling due to a quick use of water in the vegetative period. Under good conditions of water supply (deep soil, early planting) our analysis shows large differences in yield between short (1500 kg/ha) and medium or long season cultivars (2600-2900 kg/ha). This is partly the result of assuming the same planting density for all the genotypes. Villalobos et al. (1994) have shown that short season cultivars under irrigation respond to planting density up to 12 plants/ m 2. Thus, the interaction between planting density and season length may also play an important role in the performance of sunflower under the studied conditions. Results of simulations show that it is important to take into account the combination of season length of the genotype and EV to ensure water supply to the seed during the filling period. Reductions in yield as a response to increased EV were related to reductions in HI. Moreover, in the case of short season genotypes, yield is determined by water supply (soil depth), being little affected by sowing date. Decreases in sunflower HI due to water stress during the seed filling period have been observed by Gimeno et al. (1989) and Orgaz et al. (1990), and in simulations by Sadras and Villalobos (1994). They found a decrease in HI on shallow soils when genotype season length increased but no response on deep soils, which is in agreement with our results. Our analysis indicates that the optimum strategy to obtain maximum seed yield for the conditions of this study would be to use medium season length cultivars with high EV on a deep soil. Similar cultivars would be the optimum strategy on a shallow soil at the first sowing date. A different rotation to the wheat-sunflower one assumed in the simulations would change this optimum strategy. For example, a sunflower-sunflower rotation would cause a higher water use than a sunflower-wheat rotation, and short season genotypes would be the optimum choice more frequently.

153

5. Conclusions High EV in sunflower populations is not related to a low dry matter accumulation in the root. Moreover, the negative association between EV and season length does not imply substantial differences between plants with high and low EV. We can find plants with high EV and a determined season length. Thus, selection for high EV in a sunflower breeding program to improve seed yield will not be correlated with indirect selection for a poor rooting system or a short season length. Using a modified version of O I L C R O P - S U N to analyze the effect of EV on sunflower yield under different situations we concluded that high EV is a positive character. Higher yield was related to higher EV for every environment studied as a result of increases in WUE, the T/ET ratio and TE. The optim u m strategy would be to use the genotype with a season length adequate for a given environment and with high EV. In the cases studied, m e d i u m season length genotypes produce the m a x i m u m yield, except for a late sowing date on shallow soil, when short genotypes performed the best. This analysis may be extended to determine the o p t i m u m sunflower season length at any environment taking into account soil and weather data. The physiological bases of differences in EV remain u n k n o w n and deserve further research. Future work should also explore the effect of rotations and fertilizer m a n a g e m e n t on the performance of sunflower genotypes differing in EV and season length. The analysis may be performed by linking the simulation model with a Geographic Information System to define the ideotypes for different areas within a region.

Acknowledgements The authors are grateful to nez for his c o m m e n t s on the held a pre-doctoral fellowship de Investigaciones Cientfficas

J.M. Fernandez-Martimanuscript. F. Agtiera from Consejo Superior (Spain).

References Anderson, W.K., 1992. Increasing grain yield and water use of

wheat in a rainfed Mediterranean type environment. Aust. J. Agric. Res., 43:1 - 17. Austin, R.B., 1982. Crop characteristics and the potential yield of wheat. J. Agr. Sci., 98: 447-453. Austin, R.B., 1993. Augmenting yield-based selection. In: M.D. Hayward, N.O. Bosemark and I. Romagosa (Editors), Plant Breeding: Principles and prospects. Chapman and Hall, London, pp. 391-405. Boujghagh, M., 1994. Varibilite genetique des cultivars de tournesol en semis d'hiver dans la region du sais-fes. Helia, 20: 6780. Boukerrou, L. and Rasmusson, D.D., 1990. Breeding for high biomass yield in spring barley. Crop Sci., 30: 31-35. Ceccarelli, S., Acevedo, E. and Grando, S., 1991. Breeding for yield stability in unpredictable environments: single traits, interaction between traits, and architecture of genotypes. Euphytica, 56: 169-185. Cooper, P.J.M., Keatinge, J.D.H. and Gughes, G., 1983. Crop evapotranspiration -a technique for calculation of its components by field measurements. Field Crops Res., 7: 299-312. Cooper, P.J.M., Gregory, P.J., Keatinge, J.D.H. and Brown, S.C., 1987. Effects of fertilizer, variety and location on barley production under rainfed conditions in Northern Syria. II. Soil water dynamics and crop water use. Field Crops Res., 16: 67-84. Donald, C.M., 1968. The breeding of crop ideotypes. Euphytica, 17: 385-403. Elliot, G.A. and Regan, K.L., 1993. Use of reflectance measurements to estimate early cereals biomass production on sandplain soils. Aust. J. Exp. Agric., 33: 179-183. Fereres, E., Orgaz, F. and Villalobos, F.J., 1993. Water use efficiency in sustainable agricultural systems. In: D.R. Buxton, R. Shibles, R.A. Forsberg, B.L. Blad, K.H. Asay, G.M. Paulsen and R.F. Wilson (Editors), International Crop Science. Crop Science Society of America, Inc., Madison, WI, pp. 83-89. Fernandez-Martinez, J.M., Dominguez, J., Gimenez, C. and Fereres, E., 1990. Registration of three sunflower high-oil nonrestorer germplasm populations. Crop Sci., 30: 965. Gardner, C.O., 1961. An evaluation of effects of mass selection and seed irradiation with thermal neutrons on yield of corn. Crop Sci., l: 241-245. Gimenez, C. and Fereres, E., 1986. Genetic variability in sunflower cultivars under drought. II. Growth and water relations. Aust. J. Agric. Res., 37: 583-597. Gimeno, V., Fernandez-Martinez, J.M. and Fereres, E., 1989. Winter planting as a mean of drought escape in sunflower. Field Crops Res., 22: 307-316. Gregory, P.J., Shepherd, K.D. and Cooper, P.J.M., 1984. Effects of fertilizer on root growth and water use of barley in Northern Syria. J. Agric. Sci. Camb., 103: 429-438. Loomis, R.S., Rabbinge, R. and Ng, E., 1979. Explanatory models in crop physiology. Annu. Rev. Plant Physiol., 30: 339-367. Loss, S.P. and Siddique, K.H.M., 1994. Morphological and physiological traits associated with wheat increases in Mediterranean environments. Adv. Agron., 52: 229-276. Ludlow, M.M. and Muchow, R.C., 1990. Acritical evaluation of traits for improving crop yields in water-limited environments. Adv. Agron., 43: 107-153.

154

Orgaz, F., Gimenez, C. and Fereres, E., 1990. Efficiency of water use in winter plantings of sunflower in a Mediterranean climate. In: A. Scaife (Editor), Proceedings of the First Congress of the European Society of Agronomy, Paris, France, pp. 13-14. Passioura, J.B., 1977. Grain yield, harvest index and water-use of wheat. J. Aust. Inst. Agric. Sci., 43: 117-120. Regan, K.L., Siddique, K.H.M., Turner, N.C. and Whan, B.R., 1992. Potential for increasing early vigour and total biomass in spring wheat. II. Characteristics associated with early vigour. Aust. J. Agric. Res., 43: 541-553. Richards, R.A., Lopez-Castafieda, C., Gomez-Macpherson, H. and Condon, A.G., 1993. Improving the efficiency of water use by plant breeding and molecular biology. Irrig. Sci., 14: 93104. Sadras, V.O. and Villalobos, F.J., 1994. Physiological characteristics related to yield improvement in sunflower (Helianthus annuus, L.). In: G.A. Slafer (Editor), Genetic Improvement of Field Crops. Marcel Dekker, New York, pp. 287-320. Soriano, A., Villalobos, F.J. and Fereres, E., 1994. Response of sunflower grain yield to water stress applied under different phenological stages. In: M. Borin and M. Sattin (Editors), Proceedings of the Third Congress of the European Society of Agronomy, Abano-Padova, Italy, pp. 246-247. Tanner, C.B. and Sinclair, T.R., 1983. Efficient water use in crop

production: research or re-search?. In: H.M. Taylor, W.R. Jordan and T.R. Sinclair (Editors), Limitations to Efficient Water Use in Crop Production, Am. Soc. Agron., Madison, Wl. pp. 1-27. Trapani, N., Hall, A.J., Sadras, V.O. and Vilella, F., 1992. Ontogenic changes in radiation use efficiency of sunflower (Helianthus annuus, L.) crops. Field Crops Res., 29: 301-316. Turner, N.C. and Nicolas, M.E., 1987. Drought resistance of wheat for light-textured soils in a Mediterranean climate. In: J.P. Srivastava, E. Porceddu, E. Acevedo, and S. Varma (Editors), Drought Tolerance in Winter Cereals. Wiley, New York, pp. 203-216. Villalobos, F.J. and Ritchie, J.T., 1992. The effect of the temperature on leaf emergence rates of sunflower genotypes. Field Crops Res., 29: 37-46. Villalobos, F.J., Sadras, V.O., Soriano, A. and Fereres, E., 1994. Planting density effects on dry matter partitioning and productivity of sunflower genotypes. Field Crops Res., 36: 1-11. Villalobos, F.J., Hall, A.J., Ritchie, J.T. and Orgaz, F., 1996. OILCROP-SUN: a development, growth, and yield model of the sunflower crop. Agron. J., 88: 403-415. Whan, B.R., Carlton, G.P. and Anderson, W.K., 1991. Potential for increasing early vigour and total biomass in spring wheat. I. Identification of genetics improvements. Aust. J. Agric. Res., 42:347-361.

© 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

155

Options for breeding for greater maize yields in the tropics A. Elings*, J.W. White, G.O. Edmeades The International Centerfor Maize and Wheat Improvement (CIMMYT), Lisboa 27, Apdo. Postal 6-641, 00660 Mexico, D.F, Mexico

Accepted 13 April 1997

Abstract

Options for breeding for greater maize yields in the tropics were quantitatively examined with a crop growth simulation model that was tested against field data of five genotypes in four environments. Simulations indicate that at high production levels, grain filling of maize is sink-limited, and that increasing the number of kernels per m2 through larger primary ears, prolificacy or greater plant densities, will lead to increased grain yields. On a theoretical basis, it is concluded that larger primary ears lead to greater grain yields at all plant growth rates, and that increased prolificacy leads to greater grain yields only if plant growth rate exceeds a threshold. Under nitrogen limited growing conditions, selecting for genotypes that extract more nitrogen from soils, and for delayed leaf senescence, show promise for increasing yields. For crop growth limited by moisture availability around flowering, continued selection for improved kernel set leads to greater grain yields. © 1997 Elsevier Science B.V. Keywords: Model application" Nitrogen availability; Prolificacy; Sink-source relationships; Tropical maize; Water availability; Zea mays L

1. Introduction

Average grain yields of maize (Zea mays L.), which is grown extensively in the tropical and sub-tropical environments of the developing world, have been rising since 1960; from 1.5 to 3.1 t ha -~ in West Asia and North Africa; from 1.2 to 2.8 t ha -1 in South, East and Southeast Asia, and from 1.2 to 2.0 t ha -~ in Latin America (CIMMYT, 1990). Major reasons for increased grain yields are the greater use of fertilizer and improved germplasm. Yield increase in subSaharan Africa was relatively low, viz. from 0.9 to 1.2 t ha -~. Reasons are the limited use of fertilizer *Corresponding author. Tel.: +52 5 7269091; fax: +52 5 7267559; e-mail [email protected]

and improved germplasm, and intercropping, which may have disguised greater increases in land productivity (CIMMYT, 1990). These averages hide wide variation; for example, average grain yields in marginal areas of Central America are 0.3-0.7 t ha -~ (Brizuela et al., 1993). Average experimental yields in CIMMYT's international networks varied from 4.4 t ha -~ for early tropical germplasm to 6.5 t ha -~ for late subtropical germplasm (CIMMYT, 1994), and at CIMMYT's own experimental fields 14.0 t ha -I has been obtained. There exists, therefore, a wide gap between actual and experimental yields. As demand for food will continue to rise (Rosegrant et al., 1995), actual yield levels must be increased. Increases in grain yield have generally been realized by combined adoption of new cultivars and crop management

Reprinted from the European Journal of Agronomy 7 (1997) 119-132

156

methods. Whereas crop management concentrates on influencing the environmental conditions, breeding tries to modify the crop phenotype in response to the environment. To benefit optimally from improved environmental conditions, one or more matching cultivars are required, which will result in increased resource use efficiency (de Wit, 1992), expressed as kg grain per unit input, be it solar radiation, nutrients, or water. Five broad levels of agricultural production can be distinguished (Penning de Vries and van Laar, 1982). At the potential production level, growth occurs under conditions of ample supply of water and nutrients, and is determined by radiation and temperatures in interaction with the morpho-physiological make-up of the plant. Low water, soil nitrogen and perhaps phosphorus and zinc availability can limit growth for at least part of the growing season, and define production levels two, three and four, respectively. Together, they determine the attainable production level, found on around 95% of the tropical maize growing area (Edmeades et al., 1997). Actual production may be further reduced if growth is limited by the presence of weeds, pests, diseases or pollutants. Breeding for increased grain yields in tropical maize in favorable environments has so far concentrated on reduction of plant height, resulting in increased harvest index, improvement of disease and insect resistance, and reduced lodging (Dowswell et al., 1996). For environments with water and nitrogen deficits, breeding has increased partitioning to the ear, which is associated with number of ears per plant and the anthesis-silking interval (ASI), and which has resulted in increased number of grains per m E (Edmeades et al., 1997). For environments with nitrogen deficit, breeding has also reduced senescence rate (Lafitte and Edmeades, 1994a). In this paper, we quantify increases in grain yield of tropical maizes that theoretically can be expected through alteration of certain crop characteristics. The suggested selection criteria are not new, but we attempt, nonetheless, to make explicit possible selection gains. The study does not try to be exhaustive and concentrates on breeding prospects at three production levels, viz. at a high production level (cf. Boote and Tollenaar, 1994) and under nitrogen and waterlimited growing conditions, each illustrated by the performance of five cultivars.

2. Materials and methods 2.1. Field evaluations

Five maize cultivars adapted to the lowland tropics were evaluated to quantify growth characteristics. Across 8328 BN C6 is a yellow dent, late open-pollinated variety (OPV), that was selected under conditions of low soil nitrogen availability and gives relatively high grain yields under these conditions. CML247 × CML254 is a white dent, late hybrid, that is currently one of the highest yielding lowland tropical materials available from CIMMYT. Pool 16 C20 is a white dent, early maturing open-pollinated population. La Posta Sequfa C4 is a white dent, late OPV, that was selected under conditions of limited water availability. PR 8330 is a white dent, early OPV. Evaluations were conducted at CIMMYT's experiment stations near Poza Rica (PR) and Tlaltizap~in (TL). PR is located in a lowland tropical climate at 60 m altitude near the Gulf of Mexico (20°32'N, 97°26'W), and TL is located in a mid-altitude tropical climate at 940 m altitude in the central mountains of Mexico (18°4 I'N, 99°08'W). Soils are classified as sandy loam Tropofluvent (Entisol) at PR and as clay Pellustert (Vertisol) at TL. Evaluations took place during the rainy summer season of 1995 at PR and TL and during the dry winter season of 1996 at TL. Smallest, average and largest values of minimum and maximum temperatures and incoming solar radiation are given in Table 1. Treatments at PR differed primarily in soil nitrogen availability, which resulted in low, medium and high levels of leaf nitrogen concentration. The low and medium nitrogen treatments were conducted on fields that had not received chemical nitrogen for 9 and 6 years, respectively, and the high nitrogen treatment received 75 kg N ha -~ at sowing and 125 kg N ha -1 at 1 month after sowing. The 1995 experiment at TL received 150 kg N ha -! at sowing and 50 kg N ha -l at anthesis. Experiments at TL during the dry season of 1996 received 150 kg N ha -l at sowing and were conducted at three levels of moisture availability. At the 'low moisture' level irrigations were halted about 4 weeks before anthesis and resumed about mid-way through grain filling; at the 'medium moisture' level, irrigations were halted about 1 week before anthesis and not resumed; and at the 'adequate moisture' level,

157

efficiency during vegetative growth (RUEveg, in g MJ -I) were calculated as total above ground biomass at final harvest and anthesis, respectively, divided by amounts of intercepted PAR from emergence to harvest, and from emergence to anthesis, respectively. To account for variation in RUEveg due to leaf nitrogen content, RUEveg can be related to the amount of leaf nitrogen (NL; kg N ha -~ ground surface area) (Sinclair and Horie, 1989; ten Berge et al., 1994). For PR, this relation was described by fitting

irrigation was applied as needed to ensure near-optimal growth, based on long-term field experience. Genotypes were sown in three replications in randomized complete blocks, using eight-row plots of 10.5 m length at a density of 5.33 plants m -2. Row width was 0.75 m, and plant spacing within rows was 0.25 m. Eight sub-plots of ten plants each were harvested at intervals of 2-3 weeks, and after physiological maturity, a sub-plot of 30 plants was harvested. Dry weights of stems (including leaf sheaths), green leaf tissue, dead leaf tissue, panicles, ear husk leaves, ear cob and kernels, and leaf nitrogen concentration (g N g-i leaf tissue; micro-Kjeldahl) were determined at each harvest. Ten plants in the center of the sub-plot for final harvest were marked soon after emergence, and length and width of each fully expanded leaf were measured, and senesced leaves were identified, 1 day before or after a harvest. Areas of individual green leaves were calculated as length x width x 0.75 (Montgomery, 1911), from which total green leaf area (LAI; ha ha -l) was derived. Solar radiation above and below the canopy was measured on bright days every 2-3 weeks (Sunfleck Ceptometer, Decagon Devices, Inc.). Dates at which 50% of the 30 plants in the sub-plot for final harvest produced pollen (male flowering) and at which 50% of the plants showed exerted silks (female flowering) were recorded. ASI (d) was defined as the difference between these two flowering dates. Physiological maturity was defined as the date at which a black layer had formed at the base of the kernels. Intercepted photosynthetically active radiation (PARi, MJ m -2) was calculated with the crop growth model that is described below, using observed values of LAI as input. The model calculates total PAR as 50% of daily total solar irradiation. Seasonal radiation use efficiency (RUE~eas, in g MJ -l) and radiation use

RUEveg = RUEmax × (1 - e (-~xNL))

(1)

in which RUEmax is the maximum radiation use efficiency at high amounts of leaf nitrogen, and ~ is a coefficient. Because RUEveg in TL was substantially lower than in PR, TL data were not included in the regression analysis.

2.2. Simulation model The theoretical consequences for biomass and grain yield of altered crop characteristics were assessed with a maize crop growth simulation model that was based on SUCROS87 (Spitters et al., 1989). Total daily rate of canopy CO2 assimilation is calculated from the daily incoming radiation, temperature and LAI by integrating instantaneous CO2 assimilation. The light extinction coefficient (k) is an input parameter that can be varied. The rate of leaf photosynthesis at light saturation is strongly related to leaf nitrogen concentration (van Keulen and Seligman, 1987). Maintenance and growth respiration requirements are calculated on the basis of weights and chemical composition of plant organs (Penning de Vries et al., 1989). After subtraction of respiration requirements from gross assimilation, net daily growth rate is obtained. The dry matter produced is partitioned

Table 1 Smallest, average and largest values of minimumand maximumdaily temperatures,and daily solar irradiation, for the 1995 wet and 1996dry cycles at CIMMYT's experimental stations near Poza Rica (PR) and Tlaltizaplin (TL) Station

PR TL

Cycle

1995 wet 1995 wet 1996 dry

Minimum daily temperature (°C)

Maximum daily temperature (°C)

Daily radiation (MJ m-! day-I)

Min.

Ave.

Max.

Min.

Ave.

Max.

Min.

Ave.

Max.

17.8 7.0 3.0

22.4 17.0 I 1.3

25.0 20.0 24.0

26.0 27.0 18.0

32.9 30.6 32.2

36.0 33.4 38.2

7. l 14.9 3.4

20.0 27.7 25.9

28.8 34.1 36.8

158

Table 2 Total above-ground dry matter (DM, kg ha-I), grain yield (GY, kg ha-I), maximum leaf area index (LAImax, ha ha-l), leaf area index at maturity (LAImat, ha ha-~), kernels per plant (KPP), anthesis-silking interval (ASI, d), and the maximum content of leaf N during the crop cycle (NL.... ; kg N ha -~ ground surface area) of five cultivars during the 1995 wet and 1996 dry cycles at CIMMYT's experimental stations near Poza Rica (PR) and Tlaltizap~in (TL) Cycle

Growing condition

Character

La Posta Sequfa C4

Across 8328 BN C6

PR 8330

TL95 dry

Med. N

DM GY LAImax LAImat ASI KPP NL.max

12651 ~ 5286 a 3.74 0.27 0 391 43.10

11950 a 5108 a 2.95 0.40 0 388 44.12

10167 b 4610 a 2.51 0.54 0.3 361 37.86

Pool 16 C20

CML254 x CML247

Average

9654 b 4650 ~ 2.13 0.33 1.0 346 32.84

15387 6474 4.11 0.77 1.3 480 53.31

11955 5226 3.09 0.46 0.5 393 42.25

Levels of water availability TL96 dry

Low

DM GY LAImax LAlmat KPP ASI

7781 1995 3.68 0.91 205 1.7

-

-

4875 709 2.59 0.24 77 11.0

7226 781 4.17 1.54 66 20.3

6627 1162 3.48 0.90 116 11.0

TL96 dry

Intermediate

DM GY LAlma~ LAlmat KPP ASI

11603 ab 2928 ab 3.68 2.73 228 0.3

9104 b': 1822 a 3.67 1.90 155 4.3

7477 c 1734 a 2.66 1.79 142 3.0

7898 c 2446 ac 2.54 0.16 197 1.0

12508 a 3761 ~: 4.12 3.09 347 4.0

9718 2538 3.33 1.93 214 2.5

TL96 dry

Adequate

DM GY LAlmax LAImat KPP ASI

19294 a 6743 4.70 0.95 446 -0.7

-

-

14881 6650 3.06 0.01 436 0.3

20343 a 6544 5.46 1.99 429 0.7

18173 6646 4.41 0.98 437 0.1

6643 1485 2.60 0.61 1.3 150 17.41

6744 1920 2.41 0.65 2.7 184 21.94

6041 2013 2.04 0.36 2.7 182 25.31

5226 1690 2.11 0.22 3.7 159 25.18

6966 2197 2.83 0.39 1.0 240 25.45

6324 1861 2.40 0.45 2.3 183 23.06

8407 c 3535 c

8637 c 4520 ac

Levels of soil nitrogen availability PR95 wet

Low

DM GY LAImax LAImat ASI KPP

NL.max PR95 wet

Medium

DM GY LAImax LAImat ASI KPP NL....

10751 a 4849 ab 3.21 1.90 -0.7 362 39.85

12246 ab 5110 ab 3.10 1.87 0 367 39.01

2.31 0.73 1.0 266 27.08

2.25 0.74 0.7 342 28.31

13021 b 5837 b 3.63 2.08 0 410 37.13

10612 4770 2.90 1.46 0.2 349 34.28

159

Table 2 continued Cycle

Growing condition

Character

La Posta Sequfa C4

Across 8328 BN C6

PR 8330

Pool 16 C20

CML254 x CML247

Average

PR95 wet

High

DM GY LAImax LAI,,~t ASI KPP NL,max

13206 a 6044 a 3.53 2.32 -1.0 417 61.67

13474 a 5913 a 3.81 2.23 0 418 63.36

10355 b 4081 3.02 1.06 0 310 51.02

10559 b 5226 a 2.97 0.45 -0.7 371 57.68

16221 7787 4.22 2.65 -1.0 490 72.86

12771 5810 3.51 1.74 --0.5 401 61.32

Data from the same cycle with different superscripted letters differ at P < 0.05 (given only for weights).

among the various plant organs. Phenological development is tracked as a function of ambient daily average temperature. Before canopy closure, the leaf area increment is calculated from the daily average temperature, as carbohydrate production is assumed not limiting to leaf expansion. After canopy closure, the increase in leaf area is obtained from the increase in leaf weight. Integration of daily growth rates of the organs and leaf area results in dry weight increment during the growing season. Alternatively, leaf area can be made model input. The model contains an option to compute daily growth rate on the basis of leaf area index, PAR~, leaf nitrogen content, and RUE. Determination of kernels per plant (KPP) is based on work by Edmeades and Daynard (1979) and Tollenaar et al. (1992), who related KPP to plant growth rate around flowering. They indicated that at rates higher than about 6.5 g CH20 plant -1 day-l, a second ear forms in semi-prolific cultivars, and KPP is described by two intersecting hyperbolas. Important characters are the plant growth rate at which kernels begin forming on a second ear (Xi,t) and the number of kernels on the primary ear if kernels begin forming on a second ear (Yi,t). The full equations are given in Tollenaar et al. (1992). The parameters Xint and Yi,t are model inputs that can be varied, which permits simulation of semi-prolific cultivars. Maximum grain growth rate is computed from KPP and a grain filling rate of 8.5 mg kernel -~ day -l during the linear phase (derived from data in Lafitte and Edmeades, 1995). Potential grain growth rate is equal to the daily amount of dry matter available from vegetative tissue for grain growth. Actual grain growth rate is the minimum of potential and maximum grain growth rates.

Although water deficits at all times affect canopy development, they are especially critical around male and female flowering, when the ear has a relatively weak sink capacity (Schussler and Westgate, 1995), and low carbohydrate availability causes reduced ear growth and kernel abortion (Bassetti and Westgate, 1993), resulting in grain yield reduction. Reduced growth rate at flowering also causes increase of ASI, which is therefore a character associated with KPP. Bolafios and Edmeades (1993b) described the relation between ASI and KPP on the basis of well-watered and droughted experiments as: KPP =

e7.08-0.82 x (ASI+ 1.1)0.5

(2)

For a given value of ASI, the fractional reduction in kernels per plant can be estimated. Since a mechanistic approach was not possible, ASI was used as input in simulations, as a reflection of a short-term growth limitation at flowering which had no negative consequences for later crop growth.

3. Results 3.1. Field evaluations

Total above-ground biomass and grain yield of the five evaluated cultivars are presented in Table 2. Average biomass increased from 6627 to 18173 kg ha-l with increasing moisture availability (TL 1996) and from 6324 to 12771 kg ha -l with increasing soil nitrogen availability (PR 1995). Average grain yields increased from 1162 to 6646 kg ha-l, and from 1861 to 5810 kg ha -l, respectively. The late hybrid CML254 x CML247 gave greatest biomass and grain yields,

160

Table 3 Seasonal radiation use efficiency (RUE~as) and radiation use efficiency during vegetative growth (RUEveg) for four experiments conducted at Poza Rica (PR) and Tlaltizapdn (TL) in 1995 Cultivar

TL95

PR95 Low N

La Posta Sequfa C4 Across 8328 BN C6 PR 8330 Pool 16 C20 CML254 x CML247 Average

Medium N

High N

RUE~eas

RUEveg

RUEseas

RUE,,eg

RUE ....

RUEveg

RUEseas

RUEveg

1.52 1.56 1.58 1.63 1.60 1.58

1.84 2.13 1.92 1.99 2.16 2.01

1.61 1.66 1.58 1.45 1.45 1.55

2.38 2.64 2.84 2.60 2.09 2.51

1.99 2.33 1.95 2.00 1.99 2.05

3.06 2.83 2.78 2.47 2.53 2.73

2.37 2.38 2.10 2.24 2.25 2.27

3.48 3.13 3.10 3.28 3.26 3.25

except under water deficit, where the drought-adapted OPV La Posta Sequfa C4 gave greatest grain yield. The two early OPVs PR 8330 and Pool 16 C20 gave lowest biomass and grain yields. Total biomass and grain yields were associated with maximum LAI during the crop cycle (LAImax) and LAI at maturity (LmImat) (Table 2). ASI remained below 1 day at adequate soil moisture availability, and increased to 4.3 days for Across 8328 BN C6 at intermediate water availability, and to 20.3 days for CML254 ×CML247 at low water availability. ASI of the drought tolerant cultivar La Posta Sequfa was only 1.7 days at low water availability. ASI was zero or less at high nitrogen availability, and increased to average values of 0.2 and 2.3 days at medium and low nitrogen availability. Grain yield and KPP were linearly related: yield = -860 + 16.14 x KPP (r 2 = 0.97) for all experiments, and yield = -675 + 15.95 x KPP (r 2= 0.96) for TL 1995 only. Values of RUEseas at PR and TL in 1995 varied between 1.45 and 2.38 g MJ -~ (Table 3). RUEveg varied from 1.84 to 3.48 g MJ -~ among cultivars and environments and increased with increasing soil nitrogen availability. Fitting eqn (1) for experiments in PR in 1995 resulted in estimates of RUEmax of 3.26 g MJ -~, and of e of 0.1172 (Fig. 1). With the exception of the hybrid which had a relatively low RUEveg at low NL, there were few differences among cultivars. The light extinction coefficient (k) was determined by regressing fraction intercepted solar radiation (SR0 on LAI. Combined data of the five cultivars at PR and TL in 1995 could be approximated by

SR i =0.92 x (1 - e (-kxLAl))

(3)

with k = 0.53 (Fig. 2). The light extinction coefficient showed little variation among cultivars, environments and development stages. Variation in leaf nitrogen concentration among the five cultivars was low. Leaf nitrogen concentration of the cultivars in the high N treatment gradually decreased from 0.039 g g-~ at early vegetative stage to about 0.025 g g-I at maturity (Fig. 3). Leaf nitrogen concentration in the low N treatment decreased from 0.034 g g-l at the early vegetative stage to 0.015 g g-~ at flowering and to 0.011 g g-n at maturity. Cultivars in the medium N treatment and PR and in TL were characterized by intermediate levels of leaf nitrogen con3.5

00 •

3.0

"&

n

x

n

-'-' 2.5 x

2.0

~n

It i I I ! I

• La Posta Sequia C4 • Across 8328 BN C6 • PR8330 @ Pool 16C20 RUE,,~ = = CML254xCML247 3.26 X (1-e ('°1172xNL)) !""-'regression

1.5 1.0 0.5 0.0

0

20

40

60

80

leafN content(kg ha"t) Fig. I. Radiation use efficiency during the vegetative phase (RUEveg) in relationto leafN content (Nt.,kg N ha -t ground surface area) at Poza Rica (PR) and Tlaltizap~in(TL) for the 1995 wet summer season and five cultivars. The regression uses PR data only.

161

SR, = 0.92 x (1-e (°s3= uu))

0.9 0.8 0.7 ~" 0.6 =

0.5

~

0.4

•

La Posta Sequia C4

El Across BN C6

0.3

•

0.2

@ Pool 16 C20 X

0.1

PR 8330

CML247 x CML254

"--"regression

0

1

2

3

4

5

LA! (-) Fig. 2. Fraction intercepted solar irradiation (SR0 in relation to green leaf area index (LAI, ha ha -I) of five tropical cuitivars grown at three soil nitrogen levels at Poza Rica and one nitrogen level at Tlaltizapdn during the 1995 wet summer season.

centration. The maximum NL during the crop cycle (NL,max) varied from an average of 23.06 kg ha -i at low soil nitrogen availability to 61.32 kg ha-I at high soil nitrogen availability. Average NL,max at TL was 42.25 kg ha-I (Table 4). A large green leaf area was maintained after flowering at high production levels, which resulted in high potential grain filling rates. For example, simulation of grain filling processes of La Posta Sequfa C4 at high N availability in PR in 1995 shows that the potential grain growth rate was 250-300 kg ha-~ day-Z (Fig. 4). As the maximum grain growth rate was about 170 kg ha-~ day-~, the result was a sinklimited grain fill.

opment is fragmentary. Instead of rejecting the simulation model as a tool, a better approach incorporates that knowledge in data interpretation. Therefore, although simulations of crop and grain growth need improvement, the model was considered suitable to quantify expected changes in grain yield as a consequence of changes in crop characteristics. Focus was placed on genotypic and phenotypic variation, and less on the environmental variation that certainly exists over time and place in the lowland tropics. Therefore, simulations were carried out for all five cultivars, but only for the environmental conditions under which they had been evaluated.

3.3. Options to increase grain yield at high production levels 3.3.1. Radiation use efficiency A literature review by Kiniry et al. (1989) reported RUEveg ranging from 2.1 to 4.5 g MJ -l, with an average RUEveg of 3.5 g MJ -I. RUE in our own experiments varied from 1•84 to 3.48 gv~j-l among cultivars and environments, and increased with increasing soil nitrogen availability (Table 3). Crop growth at PR in 1995 was simulated using leaf nitrogen concentrations observed in the high N treatment during that season and using a high RUEveg of 4.5 g MJ -~. To reflect aging effects, RUE was assumed to decline linearly with accumulating thermal time units to 50% of its pre-anthesis value mid-way between anthesis and physiological maturity. Simulated biomass and grain 0.040 i~

PR, low N

1

3.2. Model evaluation The model is still in its developmental phase and was only calibrated and tested for the 1995 experiments, since leaf nitrogen data were not yet available for 1996, and a water balance routine was not yet in place. Total above-ground dry matter production greater than 10 t ha-1 was about 20% over-estimated, and grain yield was with about 800 kg ha-! over-estimated at low levels of soil nitrogen availability (Fig. 5). However, rankings of observed and simulated total biomass and grain yield within an experiment were similar. Crop growth simulation models are never fully accurate because our knowledge of the physiological processes underlying crop growth and devel-

17 im

I--ll--PR, medium N

0.035

}--~PR., hish N ~TL

M @

0.030 uM

u 0.025

8 M

~ o.o2o ~ o.o15 0.010 20

30

40

50

60

70

80

90

100

time (days after sowing) Fig. 3. Average leaf nitrogen concentrations of five tropical cultivars grown at three soil nitrogen levels at Poza Rica (PR) and at Tlaltizapdn (TL) during the 1995 wet summer season.

162

Table 4 Simulated increases in total above-ground dry matter production (DM, kg ha -I) and gr ain yield (GY, kg ha -!) of five cultivars during the 1995 wet summer cycle under various growing conditions at CIMMYT's experimental stations near Poza Rica (PR) and Tlaltizap~in (TL) in 1995 Cycle

Growing condition

La Posta Sequfa C4

Across PR 8330 8328 BN C6

Pool 16 C20

CML254 x CML247

Average

DM GY

17411 7921

15479 8272

16003 8289

12910 5244

15831 8628

15527 7671

DM DM DM DM DM DM DM DM DM DM DM DM

15450 15105 13703 5776 5666 5212 11140 10905 9995 12746 12512 11497

15750 15355 13793 6384 6253 5722 11409 11163 10230 13475 13236 12239

12398 12088 10831 5629 5514 5047 7921 7744 7084 11085 10853 9930

11807 11473 10168 5191 5081 4641 9355 9121 8248 11859 11581 10532

20526 20081 18388 7049 6927 6420 11658 11474 10679 17564 17269 16125

15186 14820 13377 6006 5888 5408 10297 10081 9247 13346 13090 12065

GY

7589

9064

7215

8044

13569

9096

Yint = 300, 400, 500, 600 kernels ear -! PR95 wet High N KPPm~, 417 300 GY 400 GY 500 GY 600 GY

418 4097 5302 6385 7134

310 4258 5622 7001 8271

371 3853 5056 6285 7056

533 4220 5552 6818 7878

4781 6331 7804 9232

4242 5581 6859 7914

Xint = 6.49 and 6; 5.5, 5 g CH20 day -t PR95 wet High N 6.49, 6 GY 5.5 GY 5 GY TL95 wet Medium N 6.49, 6 GY 5.5 GY 5 GY

5521 5521 5521 5616 5616 9349

5879 8753 8753 5401 5401 10054

4016 4016 7103 3793 3793 3793

5050 5050 5050 4556 4556 4556

8203 8203 8203 7305 13349 13349

Plant density = 5.33, 6.67, 8 plants m -i PR95 wet High N 5.33 DM GY 6.67 DM GY 8 DM GY

12837 5521 13970 6901 14586 7711

13807 5879 14817 8944 15835 9771

11360 4016 12461 6684 13582 7675

12208 5050 13277 8071 14415 9011

17555 8203 19115 13225 20107 14547

13553 5734 14728 8765 15759 9743

8191 3875 8608 3880 8990

8274 3130 8997 3133 9028

7918 3099 8290 3158 8643

8394 4227 8856 4240 9289

8331 3251 8750 3559 9139

RUEveg = 4.5 g MJ -! PR95 wet High N

k = 0.53, 0.5, 0.4 TL95 dry Medium N

PR95 wet Low N

PR95 wet Medium N

PR95 wet High N

Source-limited grain fill PR95 wet High N

0.53 0.5 0.4 0.53 0.5 0.4 0.53 0.5 0.4 0.53 0.5 0.4

Leaf N concentration = 0%, 5%, 10% and 25% above observed PR95 wet Low N 0% DM 8880 GY 3274 5% DM 9327 GY 3386 10% DM 9747

163

Table 4 (continued)

La Posta Sequfa C4

Across PR 8330 8328 BN C6

3473 10869 3638

3957 10011 4058

3202 9951 3292

Post-anthesis lower limit to LAI = O, 0.5, 1, 1.5 ha ha -t PR95 wet Low N 0 DM 5775 GY 2254 0.5 DM 5775 GY 2254 1.0 DM 5850 GY 2275 1.5 DM 6141 GY 2490

6383 2926 6382 2926 2522 2962 6918 3235

5632 2229 5667 2264 2866 2408 6228 2706

Cycle

Growing condition

25%

GY DM GY

Pool 16 C20

CML254 x CML247

Average

3092 9535 3190

4281 10457 4507

3601 10165 3737

5207 2004 5325 2103 5746 2433 6358 2966

7048 3948 7048 3948 7162 3985 7455 4188

6009 2672 6040 2699 6229 2813 6620 3117

Simulation data result from: increased radiation use efficiency during the vegetative phase (RUEveg) to 4.5 g MJ-j" from a reduced light extinction coefficient (k), representing more erect leaves; a source-limited grain fill; an increased maximum number of kernels on the primary ear (Y~t); a reduced daily plant growth rate at which a second ear starts forming (X~,t); increased plant density; increased leaf nitrogen concentration; and increased green LAI after silking. The number of kernels per plant observed at PR under high N (KPPma~) is given in case of variation in Xint.

yield were on the average 2756 kg ha-~ and 1861 kg ha-] , respectively, greater than the observed averages (Tables 1, and 4).

3.3.2. Light interception Our five cultivars were characterized by a relatively 300

~250

high k of 0.53. Much tropical maize germplasm is characterized by prostrate leaves; however, some new germplasm carries more erect leaves as a result of selection. We simulated growth for crops each characterized by leaf areas as observed in the field experiments, and by k-values of 0.5, 0.45 and 0.4, respectively. For the same leaf area, a better light penetration in the canopy reduced simulated biomass and grain yield at all levels of leaf nitrogen concen20000

• •

U

~150 15000

.m

U

.g

,~ 5o

"~ lO000

0

]

~ 40

50

e # ~ '/

60

70

80

90

I00

.j

5ooo

m

time (days after sewing) Fig. 4. Simulated daily amount of carbohydrates available from photosynthesis and translocation of stem reserves for grain growth (kg ha-j day-J; 0, O), and simulated maximum daily grain growth rate determined by number of kernels per ha and maximum grain growth rate (kg ha-~ day-J; II, I-I), for La Posta Sequfa C4 at high (O, O) and low (1:], O) soil nitrogen availability at Poza Rica during the 1995 wet summer season.

I

0

,

5000

t

I

I

10000

15000

20000

observed weight (kg ha"t) Fig. 5. Observed and simulated above-ground biomass ( • ) and grain weights (0) for the five cultivars evaluated during the 1995 wet summer season at Tlaltizap:tn and at Poza Rica for three levels of leaf nitrogen.

164

100 = PR, high N 1 •-II--PR, medium N I

90 80

--&-- PI~ low N

70 60 .~

50

.~ ..~

4o

~

2o

30

0

5

10

15

20

ASI (d)

Fig. 6. Average simulated relative grain yield reduction for the five cultivars evaluated during the 1995 wet summer season at Tlaltizap~in (TL) and at Poza Rica (PR) for three levels of leaf nitrogen content in relation to the anthesis-silking interval (ASI).

tration (Table 4). Simple computations of light interception profiles and photosynthesis rates in canopy layers of I ha ha -l showed that a reduction in k results in greater canopy photosynthesis only if the LAI increases to values of 5 - 6 (depending on the amount of radiation) and greater (data not given; see also Duncan, 1971). At lower LAIs, the effects of lower amounts of intercepted radiation are not sufficiently off-set by increases in radiation use efficiency.

3.3.3. Sink capacity Analysis of ear growth showed that at high production levels, grain filling is sink-limited (Fig. 4) and therefore, a greater KPP should increase grain yields. Simulated grain yields determined only by the supply of carbohydrates from photosynthesis and translocation of stem reserves, and using observed leaf areas and leaf nitrogen concentrations, vary under high N conditions from 7215 kg ha -~ for PR 8330 to 13569 kg ha -~ for the hybrid (Table 4). One way to select for greater KPP is to increase the maximum number of kernels on the primary ear. Table 4 gives relative increases in simulated grain yield at high N availability in PR in 1995 for the five cultivars as a consequence of increasing the maximum number of kernels on the primary ear from 300 to 600 kernels per ear. Xint was maintained at 6.5 CH20 plant -l day -l. On average, grain yields increased in three steps from 4242 to 7914 kg ha -~. Alternatively, prolificacy can be

selected for through reduction of the growth rate at which a second ear is formed. Simulated reduction of X~ntto 5.5 g CH20 plant -I day -l resulted in the formation of an additional ear, more KPP, and greater sink capacity for the hybrid in TL and Across 8328 BN C6 in PR. Simulated reduction of Xi,t to 5.0 g CH20 plant -l day -~ resulted in an additional ear for PR 8330 in PR and La Posta Sequfa C4 and Across 8328 BN C6 in TL. These are also the cases in which highest simulated growth rates around flowering were realized. Simulated grain yield increased if an additional ear was simulated (Table 4). The number of kernels per m 2, and thus sink capacity, can also be increased through increased plant density. Increasing in plant density from 5.53 to 6.67 and 8 plants m -2, at high leaf nitrogen concentrations and leaf areas as observed, resulted in simulated average biomass increases of 9 and 16%, respectively, and in large simulated average grain yield increases of 53 and 70%, respectively (Table 4).

3.4. Options to increase grain yield under limited nitrogen availability 3.4.1. Nitrogen uptake The effects of increased nitrogen uptake were simulated by assuming increases of 5, 10 and 25% in leaf nitrogen concentration under low nitrogen conditions in PR in 1995. This resulted in average simulated biomass increases of 5.0, 9.7, and 22.0%, respectively, and in average simulated grain yield increases of 1.2, 2.3 and 6.1%, respectively (Table 4). Predicted grain yield increases under high nitrogen conditions were lower than 1%, as grain filling is not a sourcelimited process under these conditions. 3.4.2. Leaf senescence LAI in field experiments started to decrease under low nitrogen conditions around flowering, reaching very low values at maturity, and in some cases, all leaf material had senesced by the end of the growing season. Extended leaf longevity, resulting in greater late-season LAI will be particularly effective if the flow of carbohydrates during the linear phase of the grain filling period increases as a result of this. The effect of a greater leaf longevity was simulated by making observed LAI model input, and by restricting its decline to 0.5, 1.0 and 1.5 ha ha -l after anthesis.

165

LAI took observed values, but after anthesis was not allowed to drop below the defined minimum values. In simulation studies of the low nitrogen treatment in PR in 1995, if the observed decline of LAI after anthesis was limited by a minimum value of 0.5 ha ha -l , then grain yields increased by 4.9% from 2004 to 2103 kg ha -l for Pool 16 C20 and 1.6% from 2229 to 2264 kg ha -I for PR 8330 (Table 4). These were the two cultivars with lowest LAIs throughout the growing season, and the only two cultivars with LAIs below 0.5 ha ha -~ at maturity (Table 4). For a lower limit of LAI of 1.0 ha ha -I, average grain yield increase was 5.2%, and for a lower limit of LAI of 1.5 ha ha -~, average increase was 16.6%.

3.5. Options to increase grain yield under limited water availability 3.5.1. Sink capacity of the young ear The maximum observed ASI in TL in 1996 was 20.3 days, which resulted in a yield reduction of 88%. ASIs up to 20 days were used as input in simulations, as a reflection of a short-term growth limitation at flowering which had no negative consequences for later crop growth. Simulations indicated that the relative decreases in grain yield were similar among the nitrogen levels in the PR and TL environments (Fig. 6). This is a direct consequence of the similar fractional declines in KPP in all environments and the limitations that the low numbers of KPP set on grain filling. Simulated grain yield decrease started at ASIs of 1-3 days, and reached values of about 90% at an ASI of 20 days.

4. Discussion

Leaf nitrogen contents in TL in 1995 were slightly below those of the medium nitrogen level treatment in Poza Rica in 1995, which suggests that the demand for nitrogen was not met at TL. This does not explain, however, the low value of RUEveg at TL after correction for leaf nitrogen content (Fig. 1). Some other growth limiting factor that we have not observed may have limited crop growth rate. Further experiments may indicate whether, and why, growth at this location is in general less efficient than at PR. Simulated improvements in RUE at high levels of

leaf nitrogen concentration lead to greater biomass and grain yield (Table 4). An observed RUEveg of 3.25 g MJ -~ corresponded with a biomass production of 12771 kg ha -l, and a simulated RUE~eg of 4.5 g MJ -j corresponded with an average biomass production of 15527 kg ha -I. This implies that every improvement in RUEveg of 0.1 g MJ -I corresponds to an increase in biomass production of 220 kg ha -~, and at a harvest index of 0.5, an increase in grain yield of I l0 kg ha -l. Examples of increased RUE through breeding are known. Four cycles of selection in Across 8328 BN showed increased biomass production without differences in nitrogen uptake (Lafitte and Edmeades, 1994b), and therefore a greater RUE. Another attempt at CIMMYT to select for RUE (Chapman and Edmeades, 1995) was less successful. Although differences in RUE explained differences in biomass production, selection was hampered by the few differences among cultivars, and the large genotype .'.x environment interaction. Richards (1996) mentions the contrast between wheat and barley: most increase in wheat yield has come from an increase in harvest index, whereas for barley, harvest index and biomass have equally contributed to improved yields. Tollenaar (1989) reported on increased dry matter accumulation of newer maize hybrids, and speculated that this could be partly attributed to increased tolerance to plant density stress, which would result in greater RUE. It appears therefore that at high production levels, at which moisture and nutrients are not growth limiting, direct selection for RUE offers some promise. Grain fill at high leaf nitrogen concentrations was limited by the supply of assimilates from current photosynthesis and translocation from the stem reserves (Fig. 4). Therefore, increasing the number of kernels per m 2 should increase grain yield. This can be achieved through increase of the number of kernels per plant, or through increased plant density. KPP can be increased through selection for more kernels on the primary ear before a second ear is formed (Yi,t), or through breeding for reduction of the plant growth rate at which a second ear is formed (Xi,t). As increased Yinttheoretically leads to an increased number of kernels on the primary ear at any growth rate below Xi,t, simulated increases in Yint caused greater grain yields for all cultivars, which were characterized

166 by different growth rates during sink development. Reduced Xint, however, only leads to more KPP if plant growth rate exceeds Xint, which was not always the case in simulations. Therefore, theoretically, breeding for larger primary ears appears preferable. However, larger ears may not be the best practical option for obtaining greater grain yields. Good husk cover may be difficult to preserve, and bird and insect injury may increase. Prolificacy may in that case be a better option, especially if ears are harvested by hand. Either crop growth rate during sink formation must be increased, or Xint must be reduced. Breeding for increased crop growth rate is difficult, although breeding for increased remobilization of stem reserves may be an option. Development of semi-prolific hybrids at CIMMYT (D. Beck, pers. commun.) suggests that breeding for a lower value of gin t is possible. In any case, breeding has to guard against stem lodging, which may increase through greater ear weight and reduced stalk strength if more stem reserves are translocated. Simulated increases in plant density from 5.53 to 6.67 and 8 plants m -2 resulted in large average grain yield increases of 53 and 70%, respectively (Table 4). As green LAI was held constant, this is not explained by better light interception, but by formation of more primary ears on more plants with the same amount of intercepted energy. However, this approach requires tolerance to high plant density (lodging, barrenness) and sufficiently high soil nitrogen availability to sustain a minimum plant growth rate during sink development. Reduced plant growth rate and ear abortion may be the consequence if nitrogen becomes growth limiting, which is often the case in tropical maize farming. The simulation model that was used did not account for such plant variation, and therefore may have over-estimated yield increases. Simulation of more erect leaves only results in increased crop growth at LAIs greater than 5-6, which confirms existing views (Duncan, 1971). At lower LAIs, the effects of reduced interception of radiation are not sufficiently off-set by increases in RUE. However, such high LAIs are not common in tropical maize, and, if achieved through increased plant density, typically result in lodging and barrenness. The latter is caused by inter-plant competition, which reduces the amount of assimilates per plant available for early kernel development. An effect

that the model does not account for, however, is the possible important role of the ear leaf as a preferential carbohydrate supplier to the developing ear (Edmeades et al., 1979). A better lighting of ear leaves therefore probably helps to reduce barrenness. Water or nitrogen deficit may change leaf angle and shape (e.g., leaf rolling), which affects light interception. The grain filling process was source-limited under low nitrogen experimental conditions (Fig. 4). This implies that increasing sink capacity is unlikely to result in greater grain yields. A possible avenue is to select for cultivars with a root architecture that extracts more nitrogen from the soils, presuming that this will result in higher leaf nitrogen concentrations. Our simulation studies suggested that a 1.2% increase in grain yield would result for each 5% increase in leaf nitrogen concentration. Greater nitrogen uptake, however, has been difficult to select for (Lafitte and Edmeades, 1994b). eqn (1) implies that at low values of NL more canopy nitrogen leads to a greater RUE and growth rate. This may result in a greater amount of carbohydrates available for grain filling. Although all maize cultivars had accumulated about 15 kg leaf N ha -~ ground surface area after about 4-5 weeks after sowing in all experiments, the maximum amounts of NL in green leaf tissue varied under low nitrogen conditions from 17.4 kg ha -~ for La Posta Sequfa C4 to 25.4 kg ha -1 for CML254 x CML247. We have no experimental data to explain these cultivar differences. Possibly, different root architecture plays a role in capture of nitrogen early in the season before it is lost by leaching or denitrification. Extended leaf longevity under low nitrogen conditions is particularly effective if the LAI after silking decreases and grain filling becomes source-limited, as is illustrated by the simulation studies. Breeding for delayed senescence is possible (Lafitte and Edmeades, 1994b). However, this will be at the cost of nitrogen availability for grain filling, which may reduce grain yields. Better assessments of the effects of delayed senescence require a model that incorporates a crop nitrogen balance, including within-plant nitrogen translocations. Our field data were insufficient to analyze in detail grain filling under water limited growing conditions (TL in 1996). However, the close relation between grain yield and kernels per plant (see also Bolafios

167

and Edmeades, 1993a) indicates that grain filling after mid-season drought stress is sink-limited. Increasing plant density to increase the number of ears and kernels m -2 is a risky strategy under water limited growing conditions, as the increased evaporative demand in combination with possible prolonged drought can increase end-season water deficit. Therefore, breeding preferably concentrates on increasing the number of ears (usually below one ear per plant) and kernels per plant, which can be achieved by selecting in managed drought environments for increased grain yield, reduced ASI, and increased number of ears per plant. For a given value of ASI, the fractional reduction in kernels per plant can be estimated. Simulated grain yield decrease started at ASIs of 1-3 days, and reached values of about 90% at an ASI of 20 days. The relationship described by Bolafios and Edmeades (1993b) between ASI and grain yield implies a 96% decrease in yield for an ASI of 20 days. CML254 x CML247 had at low water availability an ASI of 20.3 days, which was associated with a yield reduction of 88%. However, genetic and environmental variation in ASI is wide, which makes forecasting its behavior in a particular environment difficult. There are, to our knowledge, no data available on the relationship between photosynthesis or crop growth rate around flowering and the length of the ASI. Quantification of these relations would enable development of explanatory modules that, on the basis of photosynthesis or crop growth rate, determine both the potential number of kernels per plant and the fractional reduction due to water deficiency.

Acknowledgements We thank Srs. Lazaro Castafieda Tolentino at PR and Juan Carlos Bahena at TL, and their staff, for collecting all field data, and Drs. J. Bolafios, M. Reynolds and S. Pandey for reviewing drafts.

References Bassetti, P. and Westgate, M.E., 1993. Water deficit affects receptivity of maize silks. Crop Sci., 33: 279-282. Ten Berge, H.F.M., M.C.S. Wopereis, Riethoven J.J.M. and Drenth, H., 1994. Description of the ORYZA-0 modules. In: H. Drenth, H.F.M. ten Berge and J.J.M. Riethoven (Editors),

ORYZA Simulation Modules for Potential and Nitrogen Limited Rice Production. SARP Research Proceedings, AB-DLO, Wageningen, TPE-WAU, Wageningen, IRRI, Los Bafios, pp. 7-42. Bolafios, J. and Edmeades, G.O., 1993a. Eight cycles of selection for drought tolerance in lowland tropical maize. I. Responses in grain yield, biomass, and radiation utilization. Field Crops Res., 31: 233-252. Bolafios, J. and Edmeades, G.O., 1993b. Eight cycles of selection for drought tolerance in lowland tropical maize. II. Responses in reproductive behavior. Field Crops Res., 3 l: 253-268. Boote, K.J. and Tollenaar, M., 1994. Modeling genetic yield potential. In: K.J. Boote, J.M. Bennett, T.R. Sinclair and G.M. Paulsen (Editors), Physiology and Determination of Crop Yield. ASAJ CSSA/SSSA, Madison, WI, pp. 533-565. Brizuela, L., Zea, J.L., Aguiluz, A., Dubon, T. and Bolafios, J., 1993. Selecci6n para tolerancia a sequfa en tuxpefio C6 x BS19 a tray,s niveles de estr~s en Centro Am6rica. In: J. Bolafios, G. Safn, R. Urbina and H. Barreto (Editors), Programa Regional de Mafz para Centro Am6rica y Caribe: Sintesis de Resultados Experimentales 1992. CIMMYT-PRM, Guatamala, pp. 63-\ 66. Chapman, S.C. and Edmeades, G.O., 1995. Radiation use efficiency of lines in a tropical maize population. Paper presented at Australian Agronomy Conf. 1995. CIMMYT, 1990. 1989/90 CIMMYT World Maize Facts and Trends: Realizing the Potential of Maize in Sub-Saharan Africa. CIMMYT, Mexico, 71 pp. CIMMYT, 1994. International Maize Testing Programs: 1992 Final Report. CIMMYT, Mexico, 362 pp. de Wit, C.T., 1992. Resource use efficiency in agriculture. In: P.S. Teng and F.W.T. Penning de Vries (Editors), Systems Approaches for Agricultural Development. Elsevier, Amsterdam, pp. 125-151. Dowswell, C.R., Paliwal, R.L. and Cantrell, R.P., 1996. Maize in the Third World. Westview Press, 268 pp. Duncan, W.G., 1971, Leaf angle, leaf area and canopy photosynthesis. Crop Sci., I l: 482-485. Edmeades, G.O. and Daynard, T.B., 1979. The relationship between final yield and photosynthesis at flowering in individual maize plants. Can. J. Plant Sci., 59: 585-601. Edmeades, G.O., Daynard, T.B. and Fairey, N.A., 1979. Influence of plant density on the distribution of 14C-labelled assimilate in maize at flowering. Can. J. Plant Sci., 59: 577-584. Edmeades, G.O., B~inziger, M., Elings, A., Chapman, S.C. and Ribaut, J.-M., 1997. Recent advances in breeding for drought tolerance in maize. In: M.J. Kropff, P.S. Teng, P.K. Aggarwal, J. Bouma, B.A.M. Bouman, J.W. Jones and H.H. van Laar (Editors), Applications of Systems Approaches at the Field Level, Vol. 2, Proc. 2nd Int. Symp. Systems Approaches for Agricultural Development,6-8 December, 1995, IRRI, Los Bafios, Philippines, pp. 63-78. Kiniry, J.R., Jones, C.A., O'Toole, J.C., Blanchet, R., Cabelguenne, M. and Spanel, D.A., 1989. Radiation-use efficiency prior to grain filling for five grain-crop species. Field Crops Res., 20: 51-64. Lafitte,H.R. and Edmeades. G.O., 1994a. Improvement for toler-

168

ance to low soil nitrogen in tropical maize I. Selection criteria. Field Crops Res., 39: 1-14. Lafitte, H.R. and Edmeades, G.O., 1994b. Improvement for tolerance to low soil nitrogen in tropical maize II. Grain yield, biomass production, and N accumulation. Field Crops Res.. 39: 1525. Lafitte, H.R. and Edmeades, G.O., 1995. Stress tolerance in tropical maize is linked to constitutive changes in ear growth characteristics. Crop Sci., 35: 820-826. Montgomery, E.G., 1911. Correlation Studies in Corn. 24th Nebraska Agric. Exp. Stn. Report, Lincoln, NE, pp. 108-159. Penning de Vries, F.W.T. and van Laar, H.H., 1982. Simulation of Plant Growth and Crop Production. Simulation Monographs, PUDOC, Wageningen, The Netherlands, 308 pp. Penning de Vries, F.W.T., Jansen, D.M., Berge, H.F.M. ten, and Bakema, A., 1989. Simulation of Ecophysiological Processes of Growth in Several Annual Crops. Simulation Monographs, PUDOC, Wageningen, The Netherlands, 271 pp. Richards, D.A., 1996. Increasing the yield potential of wheat: manipulating sources and sinks. In: M.P. Reynolds, S. Rajaram and A. McNab (Editors), Increasing Yield Potential in Wheat: Breaking the Barriers. CIMMYT, Mexico, pp. 134-149. Rosegrant, M.W., Agcaoli-Sombilla, M. and Perez, N.D., 1995. Global Food Projections to 2020: Implications for Investment.

Food, Agriculture and the Environment Discussion Paper 5, IFPRI, Washington, DC, 54 pp. Schussler, R.H. and Westgate, M.E., 1995. Assimilate flux determines kernel set at low water potential in maize. Crop Sci., 35: 1074-1080. Sinclair, T.R. and Horie, T., 1989. Leaf nitrogen, photosynthesis and crop radiation use efficiency: a review. Crop Sci., 29: 9098. Spitters, C.J.T., van Keulen, H. and van Kraalingen, D.W.G., 1989. A simple and universal crop growth simulator: SUCROS87. In: R. Rabbinge, S.A. Ward and H.H. van Laar (Editors), Simulation and Systems Management in Crop Protection. Simulation Monographs, Pudoc, Wageningen, pp. 147-181. Tollenaar, M., 1989. Physiological basis of genetic improvement of maize hybrids in Ontario from 1959 to 1988. Crop Sci., 31: I 19124. Tollenaar, M., Dwyer, L.M. and Stewart, D.W., 1992. Ear and kernel formation in maize hybrids representing three decades of grain yield improvement in Ontario. Crop Sci., 32: 432-438. van Keulen, H. and Seligman, N.G., 1987. Simulation of Water Use, Nitrogen Nutrition and Growth of a Spring Wheat Crop. Simulation Monographs, PUDOC, Wageningen, The Netherlands, 310 pp.

Section 4 MANAGING RESOURCE USE Nitrogen budgets of three experimental and two commercial dairy farms in the Netherlands J.J. Neeteson and d. Hassink. ....................................................................................................................... 171 Resource use at the cropping system level P. C. Struik and F. Bonciarelli ...................................................................................................................... 179 Reprinted from the European Journal of Agronomy 7 (1997) 133-143 The efficient use of solar radiation, water and nitrogen in arable farming: matching supply and demand of genotypes ,4.J. Haverkort, H. van Keulen and M.I. Minguez .......................................................................................... 191 Soil-plant nitrogen dynamics: what concepts are required? E.,4. Stockdale, J.L. Gaunt and J. Vos......................................................................................................... 201 Reprinted from the European Journal of,4gronomy 7 (1997) 145-159 Modeling crop nitrogen requirements: a critical analysis C. O. Stockle and P. Debaeke ....................................................................................................................... 217 Reprinted from the European Journal of Agronomy 7 (1997) 161-169 Maize production in a grass mulch system - seasonal patterns of indicators of the nitrogen status of maize B. Feil, S. V. Garibay, H.U. ,4mmon and P. Stamp ........................................................................................ 227 Reprinted from the European Journal of Agronomy 7 (1997) 171-179 Nitrogen transformations after the spreading of pig slurry on bare soil and ryegrass lSN-labelled ammonium T. Morvan, Ph. Leterme, G.G. Arsene and B. Mary ...................................................................................... 237 Reprinted from the European Journal of Agronomy 7 (1997) 181-188 Size and density fractionation of soil organic matter and the physical capacity of soils to protect organic matter J. Hassink, A.P. Whitmore and J. Kub6t ....................................................................................................... 245 Reprinted from the European Journal of Agronomy 7 (1997) 189-199

Characterization of dissolved organic carbon in cleared forest soils converted to maize cultivation L. Delprat, P. Chassin, M. Linbres and C. ,lambert ....................................................................................... 257 Reprinted from the European Journal of Agronomy 7 (1997) 201-210 Analysis of impact of farming practices on dynamics of soil organic matter in northern China H.S. Yang and B.H. Janssen .......................................................................................................................... 267 Reprinted from the European Journal of Agronomy 7 (1997) 211-219 Agronomic measures for better utilization of soil and fertilizer phosphates K. Mengel .................................................................................................................................................... 277 Reprinted from the European Journal of Agronomy 7 (1997) 221-233

This Page Intentionally Left Blank

(~ 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

171

Nitrogen budgets of three experimental and two commercial dairy farms in the Netherlands J.J. Neeteson *, J. Hassink Research Institute for Agrobiology and Soil Fertility (AB-DLO), P.O. Box 129, NL-9750 A C Haren, The Netherlands Abstract

Results of recent Dutch research on the quantification of nitrogen (N) budgets of grazed grassland fields and dairy farms are reviewed to obtain a better quantitative insight into the contribution of the various processes in the N cycle to N losses on dairy farms. The results are also used to investigate the feasibility of possible future regulations on maximum permissible values of the N surplus on dairy farms. The N surplus of grassland fields was assumed to equal the difference between N input through fertilizer, urine and dung, and atmospheric deposition, and the N output through harvested grass. Under experimental conditions the N surplus of grassland fields was found to be high, ranging from about 200 to almost 700 kg N ha -1 yr-1. With few exceptions it was not possible to explain the surplus entirely by measured N losses and N accumulation in soil organic matter. The N surplus on the complete-farm scale, i.e. the difference between the annual input of N to the farm and the annual output from the farm through agricultural products, did not exceed 200 kg N ha -1 yr -1 on a farm where a major effort is put into matching economic and environmental demands. At commercial farms managed according to "good agricultural practice" the value of the complete-farm N surplus is higher, about 250 kg N ha -1 yr-~. There is a strong positive linear relationship between the N fertilizer application rate and the complete-farm N surplus. It is concluded that the N surplus on dairy farms can be reduced considerably by a combination of measures such as reducing N fertilizer inputs and applying limited grazing. It is technically possible to meet the most severe standards proposed for the N surplus on dairy farms.

Keywords: Nitrogen budget; Nitrogen surplus; Nitrogen loss; Dairy farms; Nitrogen management; Regulations

1. Introduction

While producing agricultural goods farmers have to meet both economic and environmental requirements. Obviously, farming should be profitable to the farmer himself. Society, however, demands that agricultural production takes places in a sustainable manner, which implies that no severe ecological damage occurs. Nitrogen (N) may have a harmful effect on the environment through nitrate leaching or runoff, nitrous oxide emission and ammonia vol-

* Corresponding author. AB-DLO, P.O. Box 129, 9750 AC Haren, The Netherlands. Telephone: +31-505337204, fax: +31-505337291.

atilization. As far as N is concerned, farm management should be directed to minimizing N losses. In other words, the difference between the total N input into the farm by fertilizers, manures, fodder, concentrates, seeds, atmospheric deposition and symbiotic N fixation and the total N output by agricultural products should be as small as possible. Future Dutch regulations on N management will be based on this difference, the N surplus. The Dutch government proposes to set maximum permissible values for the N surplus on arable land and grassland. The values proposed for 1998 are 175 and 300 kg N ha -1 yr -1 for arable land and grassland, respectively. It is intended to decrease the values stepwise to 100 kg N ha -1 yr -1 for arable land and 180 kg N ha -1 yr -1 for grassland in the year 2008.

172

This paper reviews results of recent Dutch research on the quantification of N budgets of grazed grassland fields and dairy farms to obtain a better quantitative insight into the contribution of the various processes in the N cycle to nitrogen losses on dairy farms and to investigate the technical feasibility of the criteria proposed for the N surplus. The paper focuses on dairy farms because about twothirds of the total Dutch farmland is used as grassland by dairy farmers. Moreover, since the total amount of nitrogen cycling annually on dairy farms is among the highest in agricultural production systems, a relatively small improvement in the efficiency of the system may have a great impact on decreasing N losses on the national scale. The research presented has been conducted by the Research Institute for Agrobiology and Soil Fertility (AB-DLO) in Wageningen and Haren, the Experimental Station for Cattle, Sheep and Horse Husbandry (PR) in Lelystad, the Department of Agronomy of Wageningen Agricultural University (WAU), the Winand Staring Centre for Integrated Land, Soil and Water Research (SC-DLO) in Wageningen, the Nutrient Management Institute (NMI) in Wageningen, and the Centre for Agriculture and Environment (CLM) in Utrecht.

was executed in two blocks of four paddocks of each 1-1.5 ha. Four annual rates of N fertilizer were applied: 250, 400, 550 and 700 kg N ha -1. The paddocks were stocked with spring-calving Friesian dairy cows according to a put and take continuous grazing system. Stocking rates were adjusted regularly to keep a target sward height of 6 cm. Averaged over the three-year experimental period the number of cow grazing days was about 600. Further experimental details are described by Deenen (1994). 2.2. De Meenthoeve

The experiment was carried out on a sand soil in Achterberg (province of Gelderland). The soil had been under grass for a great number of years and in 1981 the old sward was reseeded. In 1986 and 1987 a rotational grazing experiment was conducted with young steers on paddocks of 0.2 ha. Averaged over the experimental period the number of cow grazing days was about 800. Four annual rates of N fertilizer were applied" 250, 400, 550 and 700 kg N ha -1. Further experimental details are described by Deenen (1994). 2.3. De Marke

The materials and methods used to obtain the resuits discussed in this paper are only briefly presented here. Full details can be found in the original publications to which reference is made. The data presented originate from medium-term field experiments at the experimental dairy farms De Minderhoudhoeve, De Meenthoeve, De Marke and from monitoring the commercial dairy farms Kloosterboer and Achterkamp.

Detailed measurements were performed at a wet and at a dry grazed grassland site at the experimental dairy farm De Marke on a sand soil in Hengelo (province of Gelderland). The experiment was conducted during the period 1992-1994. The average annual N fertilizer application rate was 115 and 151 kg N ha -1 at the wet and dry site, respectively. The average annual N input with slurry, exclusive of dung and urine from the grazing herd, amounted to 196 and 171 kg N ha -1 to the wet and dry site, respectively. Further experimental details are described by Hack-Ten Broeke et al. (1996).

2.1. De Minderhoudhoeve

2.4. Kloosterboer

The experiment was carried out on a well drained young sedimentary calcareous silty loam soil in Swifterbant (province of Flevoland). The soil had been under grass for more than 20 years. In August 1985 the sward was reseeded with perennial ryegrass. In 1986, 1987 and 1988 a continous grazing experiment

The commercial dairy farm of Mr. Kloosterboer is located on a sand soil in Laren (province of Gelderland). In 1993/1994 the farm consisted of 21.8 ha grassland and 11.2 ha arable land used for silage maize porduction. Annual nitrogen inputs to and outputs from the farm were monitored during six

2. Materials and methods

173

years. The N bookkeeping took place from 1 May 1988 to 30 April 1994. The stocking rate was 1.8 milking cows per ha exclusive of accompanying young cattle. Annual milk production reached about 7500 kg per cow. Further farm and farm management characteristics are given by Den Boer et al. (1996).

2.5. Achterkamp The commercial dairy farm of Mr. Achterkamp is located on a clay soil in Oosterhout (province of Noord Brabant). In 1994/1995 the farm consisted of 40.7 ha grassland and 16.2 ha arable land used for silage maize production. Annual N inputs to and outputs from the farm were monitored during five years. The nitrogen bookkeeping took place from 1 May 1990 to 30 April 1995. The stocking rate was 1.3 milking cows per ha exclusive of accompanying young cattle. In 1995 the annual milk production reached about 7500 kg per cow. Further information on farm characteristics and on farm management is given by Den Boer et al. (1996). 2.6. Methods used to quantify the contribution of the

various processes in the N cycle Nitrate leaching at De Meenthoeve and De Marke

was determined with porous ceramic cups, whereas at De Minderhoudhoeve drain water was analysed. Ammonia volatilization was measured at De Minderhoudhoeve with the micrometeorological mass balance method (Bussink, 1994). At De Meenthoeve and at De Marke it was estimated on the basis of assumptions made by Biewinga et al. (1992). Denitrification was determined at De Marke in undisturbed soil samples which were incubated with acetylene (Ryden et al., 1987). Acetylene blocks the transformation of N20 to N2 and the N20 production is then a measure of the rate of denitrication. At De Minderhoudhoeve and at De Meenthoeve denitrification was estimated from the difference between soil mineral N in autumn and spring minus the amount of nitrate leached during winter. Changes in the amount of N in soil organic matter at De Minderhoudhoeve and De Meenthoeve were quantified by determining total soil organic N and the bulk density in each spring and autumn in the top 10 cm of soil and the 10-25 cm soil layer (Hassink and Neeteson, 1991). Five replicates were taken per layer, each replicate being a combination of 10 cores. At De Marke, it was quantified by determining the amount of N in the active fractions of the soil organic matter pool and the bulk density in each spring and autumn in the top 10 cm of soil and the 10-25 cm soil layer (Hassink, 1996).

Table 1 N budget (input/output, kg N ha -1 yr -1) of grazed grassland fields at De Minderhoudhoeve (loam soil) and De Meenthoeve (sand soil). The values given are average values of three (De Minderhoudhoeve) or two years (De Meenthoeve). After Hack-Ten Broeke et al. (1996) and Hassink and Neeteson (1991). Entry

N fertilizer application rate (kg N ha -1 yr -1) De Minderhoudhoeve

Fertilizer Urine Dung Atmospheric deposition Total input Output through gross grazed and mown Surplus Ammonia volatilization Denitrification Nitrate leaching Accumulation in SOM a Total losses and accumulation Unaccounted for aSOM = Soil organic matter

De Meenthoeve

250

400

550

700

250

400

550

700

251 165 70 40 526 330 196 19 8 11 250 288 -92

398 234 80 40 752 424 328 25 22 31 250 328 0

552 281 86 40 959 491 468 30 35 44 250 359 109

694 297 87 40 1118 515 603 31 44 53 250 308 295

268 279 85 40 672 405 267 47 88 59 0 196 71

406 323 86 40 855 449 406 53 96 75 0 224 182

517 343 84 40 984 466 518 56 115 134 0 305 213

672 353 80 40 1145 472 673 56 2 !6 163 0 435 238

174 Table 2 N budget (input/output, kg N ha -1 yr -1) at sites of grazed grassland at De Marke (sand soil). After Hassink et al. (1996). Entry

Dry site Wet site

Fertilizer Slurry, dung and urine Atmospheric deposition Symbiotic N fixation Total input Output through grass grazed and mown Surplus Ammonia volatilization Denitrification Nitrate leaching Accumulation in soil organic matter Accumulation in stubble and roots Total loss and accumulation Unaccounted for

151 307 50 34 542 278 264 14 11 83 101 55 264 0

115 282 50 19 466 308 158 11 27 22 73 24 157 1

The quantification of the N content in fertilizer, urine, dung, harvested grass, milk and meat and the estimation of the amount of N deposited from atmosphere are described by Deenen (1994).

3. Results

3.1. N budgets of dairy farms." the field scale In Table 1 N budgets of grazed grassland fields with various levels of N fertilizer application are presented. The budgets pertain to experimental fields at De Minderhoudhoeve (loam soil) and De Meenthoeve (sand soil). The N surplus of the fields was assumed to equal the difference between N input through fertilizer, urine and dung, and atmospheric

Table 3 Complete-farm N budgets (input/output, kg N ha -1 yr -1) of the dairy farm De Marke (sand soil). Each recorded year runs from 1 May to 30 April. After De Vries (1995) and Van Keulen et al. (1995). Entry

Farm year

37 82 53 0 49 12 5 238 65 11 21 97 141

0 123 74 0 49 12 0 258 6 68 0 74 184

deposition, and the N output through grass harvested (grazed+mown). The N surplus appeared to be high, ranging from about 200 to almost 700 kg N ha -1 yr -1. The surplus increased with increasing levels of N fertilizer. Independent of the level of N fertilizer the surplus at De Meenthoeve was about 70 kg N ha -1 higher than it was at De Minderhoudhoeve. This was due to the higher stocking rate at De Meenthoeve causing a higher N input through urine which was only partially matched by a higher N output through grass harvested. Since ammonia volatilization, nitrate leaching, and N accumulation in soil organic matter were also quantified at the fields studied, it is possible to investigate to which extent the calculated N surplus )lus (kg N/ha.w)

4OO

800 II

/

60O 400

94/95

Roughage from elsewhere Concentrates Fertilizer Manures from elsewhere Atmospheric deposition Symbiotic fixation Miscelaneous Total input Meat Milk Miscelaneous Output through produce Surplus

N sur

N sur flus (kg N/ha.W)

93/94

~lb

Y

I

300

I

2OO 100

200

0 N fertilizer application rate (kg N/ha.w)

Fig. 1. Relationship between N fertilizer application rate to grazed grassland fields and N surplus at the field scale.

100

200

300

400

N fertilizer application rate (kg N/ha.w)

Fig. 2. Relationship between N fertilizer application rate and N surplus of dairy farms.

175 Table 4 Complete-farm N budgets (input/output, kg N ha -1 yr -1) of the dairy farm Achterkamp (clay soil). Each recorded year runs from 1 May to 30 April (Den Boer et al., 1996). Entry

Farm year

Roughage from elsewhere Concentrates Fertilizer Manures from elsewhere Atmospheric deposition Symbiotic fixation Total input Meat Milk

Miscelaneous Output through produce Surplus

90/91

91/92

92/93

93/94

94/95

0 45 237 33 46 4 365 9 45 45 99 266

6 44 178 0 46 4 278 10 55 -24 41 237

30 92 222 0 46 4 394 9 55 38 102 292

12 66 239 0 46 4 367 10 52 24 86 281

1 68 187 0 46 4

could be explained in terms of losses and accumulation. The data presented in Table 1 show that averaged over the four N fertilizer application rates 80% of the surplus at De Minderhoudhoeve could be ascribed to losses and accumulation. At De Meenthoeve, however, only 62% of the surplus could be explained. The values of N accumulation at De Minderhoudhoeve in Table 1 are similar for all fertilizer treatments, because differences among treatments were non-significant, notwithstanding a lower N accumulation at the lower N fertilizer application rates. When making N budgets it is seldomly possible to take full account of the contribution of all N flows.

306 11 54 14 79 227

Detailed measurements at two sites with contrasting soil water status at the experimental farm De Marke, however, showed that it is possible to set up closed N budgets of grazed grassland fields by taking explicitly account of N accumulation in soil organic matter (Table 2). The calculated N surplus could thus be entirely explained by measured N losses and accumulation. In Fig. 1 the N surpluses given in Tables 1 and 2 are plotted against the respective N fertilizer application rates. Fig. 1 shows that there is a strong positive linear relationship. The N surplus at the field scale thus is largely determined by the level of N fertilizer application.

Table 5 Complete-farm N budgets (input/output, kg N ha -1 yr -1) of the dairy farm Kloosterboer (sand soil). Each recorded year runs from 1 May to 30 April (Den Boer et al., 1996). Entry

Roughage from elsewhere Concentrates Fertilizer Manures from elsewhere Atmospheric deposition Symbiotic fixation Total input Meat Milk

Miscdaneous Output through produce Surplus

Farm year 88/89

89/90

90/91

91/92

92/93

93/94

0 72 305 34 52 5 468 14 64 4

0 82 201 17 52 4 356 8 72 5

1 84 196 0 52 4 337 12 69 - 1 80 257

0 90 156 0 52 4 302 12 70 - 20 62 240

29 134 146 0 52 4 365 13 70 17 100 265

0 94 182 0 52 4 332 14 67 12 93 239

82

85

386

271

176

3.2. Nitrogen budgets of dairy farms: the complete-farm scale Table 3 presents the complete-farm N budgets of the experimental farm De Marke. At this farm the objective is to fulfil both economic and environmental objectives. Farm management aims at increasing the use of internal N flows within the farm in order to reduce demand for N inputs to the farm in the form of roughage and fertilizers. By doing so, the N surplus at the whole-farm scale, i.e. the difference between the annual input of N to the farm and the annual output from the farm through agricultural products, did not exceed 200 kg N ha -1 yr -1 (Table 3). On commercial farms managed according to "good agricultural practice" the value of the complete-farm N surplus is higher, about 250 kg N ha -1 yr -1 (Tables 4 and 5). Tables 4 and 5 also show that there was little between-year and between-farm variation in the N surplus. The relationship between N fertilizer application rate and the complete-farm N surplus was established from the data presented in Table 3, Tables 4 and 5. The strong positive linear relationship (Fig. 2) suggests that the N fertilizer application rate has a major impact on the N surplus.

4. Discussion

4.1. N budgets The N budgets of grazed grassland fields (field scale) and dairy farms (complete-farm scale) presented in this paper show that annual N inputs are high, several hundreds kg N ha -1. Annual N output in agricultural products, however, is low. At the field scale it was about 50% of the total annual N input (Tables 1 and 2) and at the farm scale only about 25% (Table 3, Tables 4 and 5). The low N efficiency of dairy farming systems is well-known (Van Der Meer and Van Uum-Van Lohuyzen, 1986). With few exceptions (Table 2) the difference between total N input and N output through products, the N surplus, calculated at the field scale could not be explained entirely by the measured N losses and accumulation into soil organic matter. This can be due to insufficient account that was taken of soil and

crop heterogeneity while measuring the contribution of the various processes and interpreting the data obtained. The values assessed may differ widely from the "real" values when errors are made in sampiing procedures and statistical analyses. It is also possible that relevant processes through which N losses occur have not been quantified at all, e.g. N2-emission after nitrification of ammonium (Pel et al., 1997). The data presented in Tables 1 and 2 show that nitrate leaching and in some instances denitrification are major loss mechanisms of N at grazed grassland fields. It should be noted that the values for denitrification at De Meenthoeve given in Table 1 seem to be unrealistically high. This is probably due to the indirect manner the values have been determined, i.e., from changes in soil mineral N which were corrected for nitrate leaching (see Materials and Methods). By doing so, the values for denitrification are likely to be overestimated, since all loss mechanisms during winter with the exception of nitrate leaching are then ascribed to denitrification.

4.2. N surplus The complete-farm N surpluses presented in this paper are less than 200 kg N ha -1 yr -1 when a major effort is made to reach environmental goals (Table 3) and about 250 kg N ha -1 yr -1 when good agricultural practice is followed (Tables 4 and 5). These values are considerable lower than the values of about 600 and 400 kg N ha -~ yr -~ reported by Van Keulen et al. (1995) for current intensively and extensively managed dairy farms, respectively. In 1990/1991 the average N surplus of 2099 Dutch commercial dairy farms calculated on the basis of general assumptions and some farm data available, appeared to be 419 kg N ha -1 (Bronwasser, 1992). The values calculated ranged from 159 to 618 kg N ha -1 yr -1, depending on total N application rate per ha, milk production per ha and stocking rate (Bronwasser, 1992). Other studies suggest that production intensity and/or N application rate can not be used as simple indicators for the N surplus, since there were large differences in the calculated N surplus among farms with similar production intensity and/ or N application levels (Anonymous, 1994; Daatselaar, 1989; Daatselaar et al., 1990). However, the results presented in this paper, which

177

were derived from actual measurements rather than assumptions, indicate that the N surplus largely depends on the N fertilizer application rate, both on the field scale (Fig. 1) and on the farm scale (Fig. 2).

4.3. Strategies to reduce the N surplus It is possible to reach a low N surplus at the farm level for various farming types (Anonymous, 1994). Farms with a high production intensity should make better use of the N in their internal flows of nutrients and feed. External inputs by fertilizers and feed produced elsewhere can then be lowered. Farms with a low production intensity by using no or little N fertilizer, generally reach a low N surplus. The results presented in Fig. 2 suggest that reducing N fertilizer application rates may have a large effect on lowering the N surplus. This is confirmed by model calculations of Van Der Putten and Vellinga (1996) who found a reduction of the N surplus of more than 60 kg N ha -1 yr -~ when 100 kg fertlizer N ha -1 yr -t less was applied than when the current recommendation was followed. The nitrogen surplus could be further reduced with 50 kg N ha -1 yr -1 when limited grazing instead of rotational grazing was applied together with reduced N fertilizer application rates. Adjustment of farm management can considerably reduce the N surplus. The best strategy is to take several measures simultaneously such as lower application rates of N fertilizer, limited grazing, lower stocking rates combined with increased per-cow production, and injection of slurries into the soil. Most of these measures have little effect on farmers' profitability, except limited grazing which requires extra costs for fodder ensilage and slurry collection and application (Mandersloot and Van Scheppingen, 1994).

4.4. Feasibility of governmental norms for the N surplus The Dutch governement intends to set up regulations to limit the N surplus on dairy farms. Maximum permissible values proposed range from 300 kg N ha -1 yr -1 in 1998 to 180 kg N ha -1 yr -1 in 2008. Although it may not be scientifically justified, it was a political decision to exclude symbiotic N fixation

and atmospheric deposition from the calculation of the N surplus used in the proposed legislation. The N surplus obtained at commercial farms with "good agricultural practice" (Tables 4 and 5) then reaches values of about 200 kg N ha -1 yr -1. The N surplus obtained at the experimental farm De Marke with a major effort to match both economic and environmental objectives (Table 3) then amounts to about 100 kg N ha -1 yr -1. This suggests that it is technically possible to reach the governmental norms for the N surplus. With "good agricultural practice" the most stringent criterion can almost be reached. From the results obtained at De Marke it can be expected that even the most severe criteria will not be exceeded when farm management moves somewhat from "good agricultural practice" towards management which takes more account of environmental objectives.

References Anonymous, 1994. Many equations and many unkowns. An analysis of farming types and differences in input-output relationships in Dutch dairy farming (in Dutch). NRLO Report 94/1. National Council for Agricultural Research, The Hague, the Netherlands, 112 pp. Biewinga, E.E., Aarts, H.F.M. and Donker, R.A., 1992. Dairy farming at severe environmental norms. Farm and research plan of the Experimental Farm for Dairy Farming and Environment (in Dutch). Report 1. De Marke, Hengelo, the Netherlands, 238 pp. Bronwasser, K., (Ed.), 1992. DELAR: An analysis of key figures in dairy farming in 1990-1991 (in Dutch). Publication 27. Informatie en Kennis Centrum Veehouderij, afdeling Rundvee-, Schapen- en Paardenhouderij. Lelystad, the Netherlands, 98 PP. Bussink, D.W., 1994. Relationships between ammonia volatilization and nitrogen fertilizer application rate, intake and excretion of herbage nitrogen by cattle on grazed swards. Fertil. Res., 38: ll-121. Daatselaar, C.H.G., 1989. An analysis of differences in nutrient budgets among dairy farms (in Dutch). Publication 3.144. Agricultural Economics Research Institute (LEI-DLO), The Hague, the Netherlands, 34 pp. Daatselaar, C.H.G., De Hoop, D.W., Prins, H. and Zaalmink, B.W., 1990. An analysis of nutrient use efficiency at dairy farms. Research Report 61. Agricultural Economics Research Institute (LEI-DLO), The Hague, the Netherlands, 98 pp. Deenen, P.A.J.G., 1994. Nitrogen use efficiency in intensive grassland farming. Thesis. Agricultural University, Wageningen, the Netherlands, 140 pp. Den Boer, D.J., Van Middelkoop, J.C. and Bussink, D.W., 1996.

178

Minimising nutrient losses in dairy farming (in Dutch). Nutrient Management Institute (NMI), Wageningen, the Netherlands, 105 pp. De "Cries, C.K., 1995. Grassland and fodder husbandry at the farm scale (in Dutch). In: H.F.M. Aarts (Editor), Weide- en Voederbouw op De Marke: op Zoek naar de Balans tussen Produktie en Emissie. Report 12. De Marke, Hengelo, the Netherlands, pp 7%89. Hack-Ten Broeke, M.J.D., Van Der Putten, AM.L, Corr6, W.J. and Hassink, J., 1996. Nitrogen losses to the environment (in Dutch). In: J.W.G.M. Loonen and W.E.M. Bach-De Wit (Editors), Stikstof in Beeld. Onderzoek inzake de mest- en ammoniakproblematiek in de veehouderij 20. Agricultural Research Department (DLO), Wageningen, the Netherlands, pp 78-98. Hassink, J., 1996. Nitrogen in stuble, microbial biomass, roots and active organic matter fractions (in Dutch). In: M.J.D. HackTen Broeke and H.F.M. Aarts (Editors), Integrale Monitoring van Stikstofstromen in Bodem en Gewas. Report 14. Experimental Station for Cattle, Sheep and Horse Husbandry (PR), Lelystad, the Netherlands, pp 55-63. Hassink, J. and Neeteson, J.J., 1991. Effect of grassland management on the amounts of soil organic N and C. Neth. J. Agric. Sci., 39: 225-236. Hassink, J., Aarts, H.F.M., Corr6, W.J, and Hack-Ten Broeke, M.J.D., 1996. Internal nitrogen flows in the soil-crop system at six observation sites (in Dutch). In: M.J.D. Hack-Ten Broeke and H.F.M. Aarts (Editors), Integrale Monitoring van Stikstofstromen in Bodem en Gewas. Report 14. Experimental Station for Cattle, Sheep and Horse Husbandry (PR), Lelystad, the Netherlands, pp 93-105. Mandersloot, F. and Van Scheppingen, A.T.J., 1994. Manure and ammonia issues at the farm scale and sector scale (in Dutch).

In: M.H.A. De Haan and N.W.M. Ogink (Editors), Naar Veehouderij en Milieu in Balans. Onderzoek inzake de mest- en ammoniakproblematiek in de veehouderij 19. Agricultural Research Department (DLO), Wageningen, the Netherlands, pp 125-146. Pel, R., Oldenhuis, R., Brand, W., Vos, A., Gottschal, J.C. and Zwart, K.B., 1997. Combined methanotrophic nitrificationdenitrification under micro-aerobic and thermophilic conditions in a model composting-system: a 15N and laC tracer study. J. Appl. Environ. Microbiol. (in press). Ryden, J.C., Skinner, J.H. and Nixon, D.J., 1987. A soil core incubation system for the field measurement of denitrification using acetylene-inhibition. Soil Biol. Biochem., 19: 753-757. Van Der Meer, H.G. and Van Uum-Van Lohuyzen, M.G., 1986. The relationship between inputs and outputs of nitrogen in intensive grassland systems. In: H.G. Van Der Meer, J.C. Ryden and G.C. Ennik (Editors), Nitrogen Fluxes in Intensive Grassland Systems. Martinus Nijhoff Publishers, Dordrecht, the Netherlands, pp 1-18. Van Der Putten, A.H.J. and Vellinga, Th.V., 1996. The effect of grassland management on the use of applied nitrogen. In: J.W.G.M. Loonen and W.E.M. Bach-De Wit (Editors), Stikstof in Beeld. Onderzoek inzake de mest- en ammoniakproblematiek in de veehouderij 20. Agricultural Research Department (DLO), Wageningen, the Netherlands, pp 36-59. Van Keulen, H., Aarts, H.F.M., Hermans, C. and De Wit, J., 1995. Prospects of Diary Farming and Environment (in Dutch). In: A.J. Haverkort and P.A. Van Der Werff (Editors), Hoe Ecologisch Kan de Landbouw Worden? AB-DLO Thema's 3. Research Institute for Agrobiology and Soil Fertility, Wageningen, the Netherlands, pp 137-144.

t~ 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

179

Resource use at the cropping system level P.C. Struik a'*, F. Bonciarelli b aDepartment of Agronomy, Wageningen Agricultural University, Haarweg 333, 6709 RZ Wageningen, The Netherlands bIstituto di Agronomia Generale e Coltivazione Erbacee, Universita' di Perugia, Perugia, Italy

Accepted 11 July 1997

Abstract

This paper illustrates the basic ideas of good crop rotations, adequate crop husbandry and high resource-use efficiencies and some relevant ecological approaches. The use of special crops to prevent the need of high inputs of crop protectants or to reduce losses of nutrients at the level of the cropping system deserves special attention in research. Examples are given for the ecological control of soil-borne fungi, parasitic weeds, nitrogen loss and other sustainable techniques to increase the resourceuse efficiency at the cropping system level. © 1997 Elsevier Science B.V. Keywords: Cropping system; Crop rotation; Nutrient balance; Resource-use efficiency; Sustainability; Biological control; Leguminous crop; Ley crop; Nematodes; Nutrient catch crop; Residual nutrients; Soil-borne fungi; Trap crop; Weed

1. Introduction

1.1. Success leads to a new approach

About 200 years ago science started to influence agricultural practice. This resulted in a high technological level of agriculture in western Europe, associated with large inputs and high yields per hectare. The approach was so successful that, finally, it induced a co-evolution of new views on management of the 'green space', landscape, environment, resources and nature. These new views result in an agriculture which contrasts strongly with the original struggle to produce as much as 'nature would allow farmers'. Nowadays, agricultural technology and

sciences may serve other goals than 'merely' efficient food and feed production. There is an increasing pressure from governments to strive for additional goals such as maintenance of characteristic (semi-)natural and traditional, cultural landscapes and give more space and chance to 'natural' components in the agro-ecosystems. So, the hierarchy of objectives in agro-ecosystems has changed dramatically over the past few decades. This requires new heuristics for scientists and farmers. 1.2. The new approach is a balance

From an ecological point of view, sustainable agriculture should focus on a balance between the following goals:

* Corresponding author. Fax: +31 317 484575; e-mail: [email protected] Reprinted from the European Journal of Agronomy 7 (1997) 133-143

maximisation of the use of beneficial natural processes in the cropping system (such as

180

•

•

• •

•

nitrogen fixation and antagonistic or synergistic relationships between (micro)organisms). maximisation of the recycling of certain elements (nutrients, carbon, etc.). However, it must be taken into account that (almost by definition) agro-ecosystems are not fully closed systems for certain components. optimal use of internal resources, the carrying capacity of the agro-ecosystem, and the genetic potential of plants and animals. optimal management and use of variation within the system. restricted use of external resources, especially as far as they are unfriendly to nature, the environment, the user of the resource or the consumer of the end product. If external resources are needed, they should be used at maximum use efficiency (in terms of output per input; Van Ittersum and Rabbinge, 1997) and with a minimum of emission to the environment (in terms of loss per unit area per unit of time). durable use of critical resources (either internal or external), which are easily lost or damaged, such as soil and water, or which are short in the long-term, such as phosphate, energy and land.

The basis of sustainable agriculture is a good crop rotation, adequate soil and water management, and proper husbandry of the different crops in the rotation. Agronomically, farmers should aim at the minimum input of each production resource required to allow maximum utilisation of all other resources. Consequently, above a certain minimum, higher inputs of yield-increasing factors (such as water and nutrients) result in higher yields per unit area and are associated with higher efficiencies (expressed as output per unit of input) of other resources (De Wit, 1992; Rabbinge et al., 1994), but at the same time might cause large residues or emissions per unit area (Nijland et al., 1997).

1.3. Balancing requires information on a longer time scale Many processes relevant to resource-use efficiency (RUE) are so slow or long-lasting that they also have

effects at the time scale of an entire rotation. Examples are the behaviour of organic matter in the soil, the changes in the suppressiveness of a soil for a certain disease or the changes in the weed seed bank. This paper focuses on these processes. We will concentrate on crop rotation first and especially illustrate the (management of) effects on yield-reducing factors (pests, diseases, weeds) and the (management of) effects on yield-increasing factors at the cropping system level. As an example of the latter we will describe the management of nutrient cycling at the cropping system level. Thereafter, we will briefly discuss nutrient management at the crop level (the links of the rotation). Finally, the paper illustrates potential uses of crops to improve the sustainability.

2. Crop rotation Crop rotation is a more or less fixed pattern in the succession of crops on a certain field. Component crop species, frequency of each crop and the crop sequence all affect the yielding ability of the entire rotation and of each individual crop. Moreover, a certain crop rotation also means a more or less fixed pattern of management and inputs, and thus changes in the soil and water resources with time. RUEs at the crop rotation level are therefore not only determined by short-term efficiencies of component crops but also by long-term processes influenced by tillage, the different crops in the rotation and their management. These processes are reflected in the physical soil fertility (organic-matter content, water-holding capacity, etc.), the chemical soil fertility (pH, availability of nutrients, presence of heavy metals, etc.), and the biological soil fertility (presence of useful microorganisms, soil-borne pests, diseases, weeds, etc.). The carry-over effects of the crops might not always be very homogeneous over the entire field. Growing a crop may often mean increasing variability (Almekinders et al., 1995). The physical fertility is affected by each crop, the type and timing of cropping practices in each crop, and the measures taken during fallow periods to improve the physical fertility or to control the weed population. Differences in effects of crops may arise from differences in duration of the canopy cover, rooting patterns, amount of effective organic matter left after harvest, other effects on soil structure, etc. To

181

some extent, these characteristics can be manipulated. For physical soil fertility the hydrology of the site is extremely important as well as the short-term and long-term effects of management practices aiming at optimal soil and water use and soil and water conservation. Chemical soil fertility is affected by fertiliser application; the effects of crops on nutrient fixation and mobilisation, mineralisation and losses of nutrients; the effects of crops on salinisation or pH; the effects of crops on distribution or concentrations of chemical pollutants; the amount and quality of crop residues; and the rate of degradation of crop residues. The most obvious effects of crop rotation are usually found through its effect on the biological soil fertility. We will elaborate on that below, but in general one can state that the higher the frequency of crops sensitive to the same soil-borne diseases or other biological stresses (pests, weeds), the higher the need for crop protectants to control them. In contrast, the higher the diversity of crops with different positive effects on beneficial organisms, the lower the need for crop protectants. While setting out general rules for a good crop rotation one should consider at least: •

the effects of the preceding crops on physical, chemical and biological fertility of the soil;

• •

the sensitivity of the following crops to these effects; the cumulation of these effects over cropping systems.

These effects can be modelled to some extent using models such as CropSyst (Stockle and Nelson, 1994; Van Evert and Campbell, 1994) and others. In general, alternating crops with contrasting effects on the physical, chemical and biological soil fertility is usually advisable. For example, crops with a strong negative effect on the amount of effective organic matter should be alternated with crops that enhance the content of effective organic matter. Vereijken (1995) formulated some semi-quantitative rules for a good rotation. The Agronomy Institute of the University of Perugia (Italy) is co.nducting a long-term rotation experiment on the effects of crop residues since 1972, which may serve as an example for long-term effects of cropping system management on the physical, chemical and biological soil fertility. In this experiment winter wheat is grown in continuous cropping or in several rotations characterised by different wheat/ maize ratios (0.50, 0.67, 0.75, 0.80, 0.83). The crop residues are either removed or buried (with an addition of 1 kg N/100 kg of dry matter). Twenty years after the initiation of the experiment, several soil char-

Table 1 Effect of burying or removing crop residues on some physical, chemical and biological properties of the soil in a long-term rotation trial in progress in Perugia (Italy) since 1972 Crop residues

Stable aggregates (%) Atterberg index Proctor max bulk density (t m-3) Field capacity (% by weight) Organic C (%) Humified organic C (%) Total N (%) Microbial biomass (ng g-I h-i) FDA-hydrolase activity (ng g-i h-i) Dehydrogenase activity (ng TPF g-i 24 h-I) Catalase activity (mg 02 g-i h-i) Mites (#/unit soil) Collembola (#/unit soil)

Significancea

Removed

Buried

38.0 14.6 1.72 22.7 0.812 0.265 0.107 155 29.9 163 420 14.1 0.6

42.3 ]6.7 !.69 23.2 0.944 0.320 0.118 196 37.6 192 360 25.9 3.2

* * * * ** ** ** ** ** ** ** ** **

* means significant at 0.01 < P < 0.05; ** means significant at P < 0.01 (Source: Perucci et al., 1997 and unpublished data Istituto di Agronomia, Perugia)

182

acteristics were assessed. Small but significant improvements of many soil characteristics became apparent as a consequence of burying crop residues (Table 1).

3. Yield-reducing factors relating to crop rotation A crop rotation, which has been maintained for a large number of cycles, has created a certain longterm balance between soil organisms. Even with this balance, yields may be considerably lower than potentially possible, at least too low to make optimal use of the other resources. Nevertheless, when this balance is disturbed, for example by applying a certain chemical killing part of the organisms, then the population density of the non-target organisms may also change, either resulting in a positive, a negative or no effect on yield (Table 2). Table 2 shows that a higher cropping frequency of potato strongly increased the infections by soil-borne pathogens such as Verticillium dahliae (causing the early dying syndrome) and Rhizoctonia solani (causing stem canker and black scurf) in potato. In the case of Verticillium, also the cropping sequence influenced the disease patterns. Maize was a better preceding crop than sugar beet in the 1:2 rotation of potato.

Potato after sugar beet was equally infected as potato after potato. The use of nematicides greatly influenced the percentage of infested stems in all rotations and for both diseases. For the Verticillium infection, controlling the nematodes reduced the disease. There are several explanations for that, one being the fact that fewer nematodes means fewer gates for the fungus to enter the roots. For Rhizoctonia, the effect of the nematicides was opposite: control of nematodes increased the infection with Rhizoctonia, most likely because the nematicides killed the mycophagous nematodes, collemboles and other mesofauna, thus stimulating the build up of the population of the fungus, especially after maize. In contrast, the infection by Colletotrichum (another soil-borne fungus) was hardly affected by potato frequency, crop sequence or by application of nematicides. These effects, but also the effects of the nematodes and the nematicides themselves are reflected in the tuber yields (Table 2). The maximum yield loss by continuous cropping of potato was 20%. This yield loss is considerable, considering the fact that it occurred in the absence of damage caused by cyst nematodes. These results suggest that: 1. Synergistic and antagonistic effects occur in crop-

Table 2 Percentage of potato stems infected with Verticillium dahliae (averaged over 1983-1986), percentage of potato plants infected with Rhizoctonia solani (averaged over 1981-1986), percentage of potato stems infected with Colletotrichum coccodes (averaged over 1983-1986), and tuber dry matter yield (g/m2; averaged over 1981-1986) in four rotations in control plots and plots treated with nematicides (Scholte, 1989) Verticillium *( % )

Rhizoctonia a ( % )

Colletotrichum b (%)

Tuber yield~ (g/m 2)

Nematicide

No

Yes

No

Yes

No

Yes

No

Yes

Rotation P MP SP MSBBP Average**

49 39 50 21 40

34 20 38 13 26

48 22 23 9 26

62 41 32 14 37

35 29 33 28 31

32 30 36 27 31

99 131 118 152 125

122 144 154 167 147

P, potato; M, maize; S, sugar beet; B, barley. aContinuous cropping significantly higher than both 1:2 rotations, which were significantly higher than the 1:5 rotation. bNo significant rotation effects on infection by Colletotrichum. CContinuous cropping significantly lower than other rotations in almost all cases; the 1:5 rotation significantly higher than SP without nematicide and than MP with nematicide. All assessments in all years were done ca. 74 days after planting. Compiled from different chapters of Scholte (1989) based on the same longterm field experiment on a sandy soil in the Netherlands. *Rotation effect significant at P < 0.01, P + SP being significantly different from MP and MP being significantly higher than MSBBP. **Nematicide effect significant at P < 0.001 for Verticillium, Rhizoctonia, and tuber yield, but not significant for Colletotrichum.

183

ping systems. This might lead to an ecological approach of crop protection, which would be much better and durable than trying to introduce one antagonist into a complex soil-plant-microorganism system. 2. It is possible to influence such synergistic or antagonistic effects by cultural practice. 3. The level of the other inputs must be adapted to these effects and to the potential of the rotation, even though additional resources might increase the crop vigour thereby increasing the resistance against the rotational diseases. In several crops (potato, sugar beet, cereals) relevant interactions between effects of nematodes or fungi and water or nutrient supply have been shown (Darwinkel, 1980; Haverkort et al., 1989; Smit, 1996). Usually, they are consistent with the general ideas about RUE already expressed in the introduction. The mechanisms, however, are sometimes very complex and go far beyond the simple fact that a resource might make the crop less sensitive to biotic stress. In a long-term rotation experiment in progress in Perugia (Italy) since 1972, a strong effect of narrow rotation of the wheat crop was observed. The wheat yields were highest when wheat was grown after maize, whereas yields were reduced when wheat was continuously grown on the same plot. The main factor explaining the yield reduction was the so-called 'take-all syndrome' (Fusarium, Geumannomyces). This syndrome is not present every year when wheat Table 3 Take-all damage (affected stems: 0--9) on a 6-year rotation maizewheat- wheat-wheat- wheat- whe at

1st wheat b 2nd wheat 3rd wheat 4th wheat 5th wheat Continuous wheat (since 1972) R 2 in regression 'rain yield/take-all estimates'

1993

1994

1996 a

0.3 1.8 3.0 3.9 2.0 1.9

1.5 2.3 5.0 4.1 3. I 2.5

0.3 2.8 2.5 4.4 3.9 2.5

0.75

0.73

0.67

aln 1995 no damage was observed. bWheat following maize. (Source: Data from lstituto di Agronomia, Perugia.)

Table 4 Take-all attack (%) on the third cycle of wheat after different crops or fallow Previous crops (3 years before)

Lucerne Tall fescue Mixture (lucerne + tall fescue) Fallow

N-fertilisation on wheat (kg ha -I) NO

NI00

22 15 16 40

1I I 6 11

(Source: Bianchi and Bonciarelli, 1980.)

is grown, but is frequent. Its occurrence depends on the season. Table 3 shows that in the six-cycle rotation, maize followed by 5 years of wheat, damage was found in the last 4 years, except for 1995. Damage was lowest when wheat was grown immediately after maize. When wheat was cropped continuously for 5 years or more, there was a so-called 'decline' effect: the disease became less severe, because of a new balance between pathogens and their antagonists. Twothirds to three-quarters of the variation in wheat yield was accounted for by the 'take-all damage'. A late effect of rotation was described by Bianchi and Bonciarelli (1980) when wheat was repeatedly grown for 3 years on the same field, following either different crops or fallow (Table 4). During the third cycle, wheat was affected more by the take-aU syndrome when three cycles had been grown after lucerne ley or fallow than after tall fescue or a mixture of lucerne and tall fescue. Nitrogen fertilisation reduced the disease. Several studies have been carded out in Italy to investigate the effect of crop rotation on weed flora dynamics. After six cycles of a wheat-maize rotation, the total number of weed seeds in the top soil layer and the actual weed infestation were substantially reduced compared with continuous cropping of maize (Table 5). Non-chemical ways to reduce these rotational problems are the use of resistant or tolerant cultivars, organic amendments, treating or removing crop residues, adaptation of the cropping techniques thus avoiding the problem (for example by delaying sowing time) or reducing the problem by increasing the vigour of the crops, special soil tillage techniques, biological control through introducing or stimulating

184

Table 5 Number of weed seeds recorded in the top 0.15 m-soil layer and actual weed flora (in unweeded plots) after 6 years of two different rotation systems (Covarelli and Tei, 1988) Rotation

m-m-m-m-m-m w-m-w-m-w-m

Number of seeds m-"

Weeded

Unweeded

24 500 19300

55 800 i 8 920

Actual weed flora no. of weeds m-" 422 161

m, Maize; w, wheat.

antagonists, farm hygiene and growing of special crops (see below). An example of such non-chemical strategies is illustrated in Table 6. In very specific cases the build up of the population of survival structures can be reduced by haulm treatments. Scholte et al. (1996) summarised some results obtained at the Department of Agronomy, Wageningen, on the effect of removing the plant debris on the number of microsclerotia of Verticillium in the soil (Table 6). The effects visible in March 1994 were caused by treatments in 1991 and 1992: removing plant debris reduced the Verticillium inoculum of either isolate tested considerably.

4. Nutrient management strategies at the cropping system level Differences among crops and their cultivars in recommended (economically optimal) applications and nitrogen use efficiency are large; residual N is therefore very variable (for overview, see Neeteson, 1994). Residual N from commercial fertiliser, nitrogen fixation, mineralisation, deposition or organic manure will be lost or will have after-effects later in the rotation. Examples of the influence of the residual effect of N fertilisation on yield of succeeding crops are given for a long-term experiment in Italy in which wheat was grown for 3 years after different preceding crops. Table 7 shows that input of nitrogen in a 4-year ley crop of tall fescue resulted in a significant increase in the yield of the ley crop and of the succeeding unfertilised wheat crop both in the first and in the second year after the four-year ley. In the third year the effect

became negligible. Effects of nitrogen during the 4 ley-crop years were absent when lucerne was grown, both on the yield of the leguminous ley crop itself and on its after-effects. However, wheat yields were higher in the first year after lucerne than in later years. Table 8 shows grain yields of wheat grown consecutively for 2 years on the same field, as affected by two N-fertilisation levels imposed in four different, previously grown annual crops. There was an interaction between N-level and previous crop. The effect of N-fertilisation in the previous crop in improving soil fertility and thus wheat yield became already much smaller in the second year. The effect of red clover on soil fertililty, however, was still visible, at least at NO. In other experiments, the residual effect of N could only be demonstrated if excessive amounts of N were supplied to the preceding crop (Table 9). The residual N proved very sensitive to leaching on this clay-loamy soil when the winter period was wetter than usual. At the cropping system level the efficiency of nitrogen is determined by the level of input, the form and timing of input, the efficiencies of utilisation by the different component crops and the degree to which N remaining in the soil or in crop residues can be kept within the boundaries of the cropping system and can be utilised by later crops. Soil structure affects nutrient use efficiency (Van Ittersum and Rabbinge, 1997), partly by its direct effect on attainable yield, partly indirectly by its effect on root density of the crop. Efficiency is optimal when the following aims are met: 1. Maximum use of the nutrients supplied by adjustTable 6 Effects of removal of plant debris on the number of microsclerotia of Verticillium dahliae in the soil in March 1994 Isolate

PI PI FI FI

Crop sequence 1991

i 992

1993

No. cfu g-i

P PR F FR

P PR P PR

PR PR PR PR

126 51*** i 99 28***

Isolates: PI, potato isolate; FI, field bean isolate. Crop sequences: F, field bean; P, potato; R, removal of plant debris, cfu, Colony forming units. ***Significantly different from the control at P < 0.001). After Scholte et al. (1996).

185

ing the supply to the demand, by synchronisation of supply and demand, and by synlocalisation (the nutrient is available where it can be taken up). Nutrient uptake needs to be predictable to achieve this. Greenwood et al. (1990) showed that there is a close relation between above-ground dry weight in the crop and the above-ground nitrogen concentration, which is very robust over a wide range of conditions and species. Under non-limiting conditions the nitrogen demand can therefore be derived from the expected production curve. Similar relationships for other nutrients are more difficult to assess: the variation not accounted for in the case of P and K is usually much higher than in the case of N (Greenwood et al., 1980). 2. Optimal use of crop residues for the increase of soil fertility, for example by maintaining the proper C:N ratio in the soil to allow optimal rates of breakdown of organic matter. 3. Maximum reduction of emission during the periods between the main crops, e.g. by growing nutrient catch crops (see later) or by incorporating straw. 4. Proper soil tillage; soil tillage may increase the efficiency of nutrient supply, water and other resources. Microvariability in plant and soil characteristics and their interactions are crucial for a proper management of nutrients (e.g. Van Noordwijk and Wadman, 1992). Variation must be taken into account when determining agronomically optimal rates of fertiliser with minimum ecological damage. Part of the variability in for example availability of water or of nutri-

ents may persist, increase in time and interfere with other aspects of crop management. Managing variation is therefore crucial for sustainable resource management at the cropping system level (Almekinders et al., 1995). A final example of the effects of crop rotation and nitrogen supply on the long-term balances of N, but also other nutrients, i.e. P and K is given in Table 10. Continuous cropping of potato increased the need for N, P and K compared to the 1:2 rotation. Additional N increased the nitrogen surplus similarly for the two crop rotations; N was the only nutrient present in excessive amounts.

5. Nutrient management strategies per link in the crop rotation The relation between nitrogen uptake and yield is fairy fixed (see also above). For nitrogen use efficiency (defined as output per unit input) the two following crop types can be distinguished: crops without change in N-recovery (for definition see Vos, 1996b) with an increase in N-supply until the agronomically optimal level (i.e. the maximum level at which the Nrecovery is still at its best) and crops with a decrease in N-recovery with an increase in N-supply (Vos, 1996b). Beyond the agronomically optimal supply the recovery decreases with an increase in supply for both types. The type of fertiliser is relevant to the magnitude of and variation in the losses. In all cases nitrogen residues are unavoidable. The dynamics of N availability cannot be accurately pre-

Table 7 Direct effect of N-fertilisation during the ley crop on the yield of lucerne and fescue ieys and the effect of residual N observed in the first, second and third years of the following continuous wheat. Wheat crops were fertilised with 0 N N-fertilisation (kg ha -t)

DM yield (t ha -I in 4 years)

Grain yield (t ha -j) of the following continuous wheat I st year after

0 75 150 300 600

2nd year after

3rd year after

L

TF

L

TF

L

TF

L

TF

46 46 45 48 49

14 21 31 42 47

5.5 5.6 5.6 5.5 5.8

1.9 1.9 2.1 2.7 5.0

6.0 5.6 5.8 5.8 6.1

4.2 4.0 4.4 4.6 5.4

3.2 3.1 3.2 3.2 3.4

3.0 2.8 3.1 3.2 3.3

L, lucerne; TF, tall fescue. (Elaborated from Bianchi and Bonciarelli, 1980.)

186 Table 8

Table 10

Residual effect of previous crops and their N fertilisation

Nitrogen, phosphorus and potassium balances (in kg ha-~.year) in two rotations

Previous crops

Red clover NO N300 Italian rye grass NO N300 Maize NO N300 Wheat NO N300 Fallow NO N300

1st wheat (NO)

2nd wheat (NO)

1969/70

1970/71

1971/72

4.0 4.5

4.9 5.4

2.9 3.0

3.4 3.7

2.1 2.9

2.2 3.9

2.6 2.7

2.9 3.0

2.0 4.3

3.1 5.9

2.2 2.7

2.6 2.9

2.0 2.7

1.2a 1.8a

2.1 2.2

2.4 2.7

3.5 5.1

3.9 4.8

2.2 3.3

2.4 2.7

(NO, 0 kg ha-I; N300, 300 kg ha-~) on the grain yield (t ha-I) observed in the first and second years of wheat. aHeavy attack of take-all (80%) vs. negligible attacks (2-5%) following other crops than wheat. (Elaborated from Bonciarelli, 1972 and Bonciarelli and Bianchi, 1980.) dicted, not in time and not in amount; similarly crop growth and amounts of N in crop residues are still unpredictable to a large extent. Crop residues will affect the soil fertility. Depending on their C:N

Table 9 Residual N taken up after different rates of N fertilisation in preceding crops (kg ha-I) N to preceding cropsa (kg ha-t)

50 100 200 (economic rate) 400 800

Balances (kg ha-l.year)

1972/73

Nitrogen taken up by following wheat (kg ha-I) Rainy yearb

Dry yearc

0 2 7 18 36

2 12 33 80 143

aAverages of four crops: silage maize, forage maize as catch crop, sudan grass and Italian rye grass. b1968--69:635 mm rain in period October-March (152 mm more than average). ¢1969-70:323 mm rain in period October-March (160 mm less than average). (Elaborated from Bonciarelli and Monotti, 1973.)

RINI R IN2 R2NI R2N2

N

P

K

52 90 38 78

1 0 -2 -5

1 9 -35 -27

R l, Alternating a cereal and a potato crop; R2, continuous cropping of potato and at two nitrogen levels: N1, -15% recommended quantity; N2, +15% recommended quantity. Data are averages over 4 years and over two levels of input of organic matter on a sandy soil containing peat. Recommended quantity for cereal: I l0 kg ha-~, and for potato: 200 kg ha-~. Method of calculation and data are derived from Vos (1996a). ratio, soil characteristics, tillage and cropping practices and weather, the proportions of N lost or carried over to the next growing season vary considerably. The emissions can be reduced, albeit not to zero. If nitrogen emissions are kept extremely low, usually the chemical soil fertility is reduced in the long term. This may not be true for situations in which nitrogen catch crops are grown or for other nutrients.

6. Use of special crops to improve sustainability Growing of legumes (improving nitrogen and phosphorus availability); green manure crops (physical, chemical and biological soil fertility); lure, trap and killing crops (biological control or suppression); 'wake-up crops' (inducing suicidal germination or hatching); cover crops (preventing soil erosion); and nutrient catch crops (keeping nutrients available for subsequent crops) can help to improve the sustainability of the cropping system, i.e. to maintain the natural resources and the carrying capacity of the cropping system. An example of using green manure crops to increase the population of the mycophagous soil fauna, thus suppressing a soil-borne fungal disease is illustrated in Table 11. Growing oats as a green manure crop reduced the proportion of potato stems affected by Rhizoctonia stem canker, probably by stimulating the nematodes and perhaps partly by shifts in the ratios of collemboles species.

187

Table 11 Effects of the green manure crop oats on the relative number of collemboles and nematodes in the soil and on the Rhizoctonia stem canker disease index on potato in the following growing season in two field experiments Relative number of collemboles

Control Oats

1992

1994

Relative number of nematodes 1994

100 127

100 123

100 1043"

Disease index (0-100) 1992

1994

26 10"

67 51*

Field experiments lasted two years. In year 1, the soil was infested with R. solani by growing a potato crop from seed potatoes with black scurf. In the autumn of year 1 either no crop or oats was grown. In year 2, potato was planted to test the effects of the green manure. *Means significantly different from control. After Scholte et al. (1996). Even parasitic weeds can be reduced by growing trap crops that produce germination stimulants but cannot be infected by the parasite, and thus induce suicidal germination. The example is from continuous cropping of tobacco in southern India, but similar examples can be found for faba bean in the Mediterranean area. The yields of broomrape were reduced by the trap crop because the part of the seed bank that was not dormant, was lured by growing the trap crop before the tobacco was planted and could be infected (Table 12). Part of the yield advantage in tobacco may also have been caused by the green manuring effect of the trap crop. The use of nitrogen catch crops is relatively well documented. An extreme case where nitrogen catch crops are grown to prevent high N uptake by the commercial crop is found in Italy. In Umbria flue cured tobacco is an important crop on irrigated alluvial Entisols. The rather high fertility of these soils is not suitable to produce the low nicotine content which is required in that type of tobacco. To remove the excessive nitrogen from the soil before planting the tobacco, the common practice is to grow a catch crop

Table 13 Growth and N uptake by an oat catch crop before growing tobacco

Table 12 Effects of trap crops on relative above ground yield (%) of the parasitic weed broomrape (Orobanche cernua) and relative economic yield of tobacco. Data derived from Dhanapal and Struik (1996) Trap crop Sunhemp (Crotalaria juncea L.) Redgram (Cajanus cajan L.) Millsp.) Sunflower (Helianthus annuus L.) Fallow

of forage oats sown in fall (October) and mown and removed in April-May, just before transplanting the tobacco. Table 13 gives some unpublished results of a series of samplings and analyses of both oat biomass production and nitrogen uptake. More often nutrient catch crops are grown to prevent nitrate leaching. An example, showing the potential for the Netherlands, is presented in Table 14, based on data of Vos (1996b). The performance of nutrient catch crops strongly depends on the sowing time, because temperature and light intensity decline rapidly in autumn, whereas the chances of excessive water and killing night frosts increase. For all special crops grown to improve sustainability one rule is important: they have to fit in the sequence of main crops and should not interfere with necessary soil tillage. They even may facilitate soil tillage by reducing the soil water content in early spring. Especially their response to light and temperature in dependence of sowing date and their effects on water availability need further research to optimise their use.

Broomrape Tobacco 17

33 51 100

173 157 129 100

Sampling date (1996)

DM (t ha-I)

N uptake (kg ha-j)

March

1.0 1.7 2.4 4.4 5.7 7.6 8.0

26 39 53 68 84 86 86

April

May

1 15 29 12 19 26 3

Unpublished data, Istituto di Agronomia, Perugia.

188

Table 14 Average benefit of nutrient catch crops (average of different species) in a cropping system Previous crop

Reduction in N-loss from the system (g m-2)

Oats Spring wheat Potato Sugar beet

2.6 3.9 3.1 0.4

Catch crops were sown as soon as possible after the harvest of the main crops. Losses were for the period autumn year X until spring year X + 1. Based on Vos (1996b).

7. Some final remarks Tools are strongly needed to allow analytical studies on the effects of cropping system management on • • •

the productivity of each crop in the rotation; the environmental risks; and the stability, resilience and durability of the cropping system.

Investigations into options to maintain a short rotation of a crop with low self-tolerance by making use of non-chemical strategies to avoid yield-reducing conditions are also required. Ecological approaches may aim at reducing rate of multiplication of the pathogen, stimulating its antagonists, or both. In practice, variation in RUE is strongly influenced by differences in 'farming styles' among farmers, even under similar environmental conditions and financial returns (Van der Ploeg, 1990). Apparently, not only knowledge, but also skill and motivation to apply it, are relevant.

References Almekinders, C.J.M., Fresco, L.O. and Struik, P.C., 1995. The need to study and manage variation in agro-ecosystems. Neth. J. Agric. Sci., 43: 127-142. Bianchi, A.A. and Bonciarelli, F., 1980. Effect r6siduei de fertilisation azot6e appliqu6e ~t des prairies pures et mixtes de luzerne et de f6tuque 6lev6e. S6minaire Agrimed 'M6thodologie d'l~tude des Syst~mes de Culture', Toulouse, 7-9 May 1980. Bonciarelli, F., 1972. Trials on crop sequences and N fertilisation. Rivista Agron., 1972, VI(1): 44-48. Bonciarelli, F. and Bianchi, A.A., 1980. Evaluation de la fertilitd6

r6siduelle de diff6rentes cultures soumises h diff6rents niveaux de fertilisation azotEe. S6minaire Agrimed 'M6thodologie d'l~tude des Syst~mes de Culture, Toulouse, 7-9 May 1980. Bonciarelli, F. and Monotti, M., 1973. Residual effect on wheat of different rates of nitrogen applied to annual forage grasses. Rivista di Agron., VII, (2-3): 150-158. Covarelli, G. and Tei, F., 1988. Effect de la rotation culturaie sur la flore adventice du ma'fs. 8~me Colloque International sur la BioIogie, l'Ecologie et la Systematique des Mauvaises Herbes, Dijon, pp. 477-484. Darwinkel, A., 1980. Grain production of winter wheat in relation to nitrogen and diseases. I. Relationship between nitrogen dressing and yellow rust infection. Zeitsch. Acker. Pflanzen., 149: 299-308. De Wit, C.T., 1992. Resource use efficiency in agriculture. Agric. Systems, 40:125-15 I. Dhanapai, G.N. and Struik, P.C., 1996. Broomrape control in a cropping system containing bidi tobacco. J. Agron. Crop Sci., 177: 225-236. Greenwood, D.J., Barnes, A., Liu, K., Hunt, J., Cleaver, T.J. and Loquens, S.H.M., 1980. Relationships between the critical concentrations of nitrogen, phosphorus and potassium in 17 different vegetable crops and duration of growth. J Sci. Food Agric., 31: 1343-1353. Greenwood, D.J., Lemaire, G., Gosse, G., Cruz, P., Draycott, A.A. and Neeteson, J.J., 1990. Decline in percentage of N of C3 and C4 crops with increasing plant mass. Ann. Bot., 66: 425-436. Haverkort, A.J., Vos, J., Groenwold, J. and Hoekstra, O., 1989. Crop characteristics and yield reduction of potato due to biotic stress in short rotations. In: J. Vos, C.D. van Loon and G.J. Bollen (Editors), Effects of Crop Rotation on Potato Production in the Temperate Zones. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 273-290. Neeteson, J.J., 1994. Residual soil nitrate after application of nitrogen fertilizers to crops. In: D.C. Adriano, A.K. lskandar and I.P. Murraka (Editors), Advances in Environmental Sciences. Contamination of Groundwaters. Science Reviews Ltd., Northwood, pp. 347-365. Nijland, G.O., Schouls, J. and Oomen, G.J.M., 1997. The Relation Between Nutrient Application, Nutrient Uptake, Production and Nutrient Residues. Wageningen Agricultural University Papers 97-3, Wageningen (in press). Perucci, P., Bonciarelli, U., Santilocchi, R. and Bianchi, A.A., 1997. Effect of rotation, nitrogen fertilization and management of crop residues on some chemical, microbiological and biochemical properties of soil. Biol. Fertil. Soils (in press). Rabbinge, R., Diepen, C.A. van, Dijsselbloem, J., Koning, G.J.H. de, Latesteijn, H.C. van, Woltjer, E.J. and Zijl, J. van, 1994. Ground for choices: a scenario study on perspectives for rural areas in the European Community. In: L.O. Fresco, L. Stroosnijder, J. Bouma and H. van Keulen (Editors), The Future of the Land; Mobilizing and Integrating Knowledge for Land Use Options. Wiley, Chichester, UK, pp. 95-121. Scholte, K., 1989. Effects of Crop Rotation on the Incidence of Soil-Borne Pathogens and the Consequences For Potato Production. Doctoral Thesis, Wageningen Agricultural University, Wageningen, The Netherlands, 143 pp.

189

Scholte, K., Mol, L. and Lootsma, M., 1996. Control of Verticillium dahliae and Rhizoctonia solani by cultural practices. In: Abstracts of Conference Papers, Posters and Demonstrations, 13th Triennial Conference of the European Association for Potato Research, Veldhoven, The Netherlands, 14-19 July i 996, pp. 134-135. Smit, A.B., 1996. PIEteR: a Field Specific Bio-economic Production Model for Decision Support in Sugar Beet Growing. Doctoral Thesis, Wageningen Agricultural University, Wageningen, The Netherlands, 201 p. Stockle, C. and Nelson, R., 1994. CropSyst User's Manual. BSE Dep, WSU, Pullman, USA, 186 p. Van der PIoeg, J.D., 1990. Heterogeneity and styles of farming. In: J.D. van der PIoeg (Editor), Labor, Markets and Agricultural Production. Westview Press, Boulder, pp. 1-36. Van Evert, F.K. and Campbell, G.S., 1994. CropSyst: a collection of object-oriented simulation models of agricultural systems. Agron. J., 86:325-331. Van Ittersum, M.K. and Rabbinge, R., 1997. Concepts in production ecology for analysis and quantification of agricultural inputoutput combinations. Field Crops Res., 52: 197-208.

Van Noordwijk, M. and Wadman, W.P., 1992. Effects of spatial variability of nitrogen supply on environmentally acceptable nitrogen fertilizer application rates to arable crops. Neth. J. Agric. Sci., 40: 51-72. Vereijken, P., 1995. Designing and testing prototypes. 2nd Progress Report of EC Concerted Action AIR3-CT920755, Wageningen, The Netherlands, 90 pp. Vos, J., 1996a. Input and offtake of nitrogen, phosphorus and potassium in cropping systems with potato as a main crop and sugar beet and spring wheat as subsidiary crops. Eur. J. Agron., 5:105-114. Vos, J., 1996b. Nitrogen cycle related to crop production in cool and wet climates. In: R. Samulesen, B. Solsheim, K. Pithan and E. Watten-Melvaer (Editors), Crop development for the cool and wet regions of Europe. Nitrogen supply and fixation of crops for cool and wet climates. Proceedings COST 814 Workshop, Troms~, September 7-9 1995. Office for Official Publications of the EC, Luxembourg, Luxembourg, pp. 3-14.

This Page Intentionally Left Blank

(~ 1997 ElsevierScience B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

191

The efficient use of solar radiation, water and nitrogen in arable farming" matching supply and demand of genotypes A.J.

Haverkort ~,*, H. van Keulen ~, M.I. Minguez b

a DLO-Research Institute for Agrobiology and Soil Fertility (AB-DLO), P.O. Box 14, 6700 AA Wageningen, The Netherlands b Depto Produccion Vegetal: Fitotecnia. E.T.S. lngenieros Agronomos, Universidad Politecnica de Madrid, Ciudad Universitaria, 28040 Madrid, Spain Abstract

The length of the growing season of a crop depends firstly on the suitable period of the year when temperatures allow growth of a particular crop. The amount of intercepted solar radiation then determines the potential dry matter production. The length of the season, however, may be reduced by lack of water. The resulting yielding ability determines the nitrogen requirement of the crop. Water and nitrogen uptake may be increased through management strategies that increase water use and improve water use efficiency in semi-arid environments. This paper shows the implications of the length of the available growing season and amount of available water for the desired genotypic characteristics of a species. Subsequently, the amount of nitrogen needed to reach the highest yields provided best use is made of solar radiation and available water is discussed in quantitative terms. Finally, recent research developments are discussed such as how the optimal temporal distribution of nitrogen application can be derived from periodical assessment of its availability in the soil and more importantly from observations and calculations of amounts of nitrogen already taken up and expected nitrogen uptake till harvest. Keywords: Resource use efficiency; Solar radiation; Water use efficiency; Nitrogen use efficiency; Ideotyping; Plant breeding

1. Introduction

Consumers, producers and scientists are aware that agricultural production should optimize the use of natural resources and minimize emissions to the environment. Extraction of ground and surface water and its pollution with nitrate pose serious risks to the natural environment. To obtain potential yields (all minerals and water needed is supplied and no (a)biotic factors present) in a given environment, resources, notably water and nutrients should be applied at rates which are both economically and environmentally unacceptably high. Attainable yield (all inputs supplied at levels that are economically justified) by farmers, may be unsustainable at the long term. To realise the food and export require* Corresponding author.

ments for Europe on the whole, groundwater is threatened because of contamination with nitrogen and because of depletion of this resource. This is a serious risk for natural ecosystems. A growing proportion of food, following consumer demands is produced without chemical fertilizers and biocides but such attainable organic yields are lower than current yields and their resource-use efficiency is amenable to debate (de Wit, 1992). Yields can be expressed in terms of resource availability (water (W), solar radiation (R) and nitrogen (N)) and resource-use efficiency (E), i.e. grams of dry matter produced per unit resource used by the crop, e.g. per g water (WUE), per joule intercepted solar radiation (RUE) or per g of nitrogen taken up by the crop (NUE). Fresh crop yield (Y) is then expressed as Y = R x R UE x H I / D M C

(1)

192

Y = W × WUE × HI/DMC

(2)

Y = N X NUE X HI/DMC

(3)

where HI is the harvest index and DMC is the dry matter content of the harvested produce. The greater the availability of a resource for single factor experiments, the lower its efficiency. When more inputs are increased simultaneously, efficiency of both may increase. De Wit (1992) stated that a production factor which is in minimum supply contributes more to production, the closer other factors are to their optimum. Strategic research should be into the identification of the minimum of each production factor needed to allow maximum utilization of all other resources. Temperature, solar radiation and crop species or cultivar are the main growth defining factors determining the lengths of the available growing season and actual growth cycle and hence the potential yields. Actual yields are mainly limited by the availability of water and nutrients, especially nitrogen and by pests, diseases and weeds. Advances have been made in maximizing the availability of resources and increasing resource use efficiencies through selection of crop species (C4/C3plants) and breeding for adaptation to adverse conditions such as drought and high temperatures. Crop management practices, however, matching crop cycles with periods of low evaporative demand and concentration of limiting resources such as strategic irrigation and application of fallow (Loomis and Connor, 1992) and organic farming techniques may have great impact as well. The most efficient use of water and nitrogen is realized by supplying them in crop management aimed at yields that are close to the potential. Then other (a)biotic constraints are reduced as much as possible and provided that the financial rate of return of each unit of water and nitrogen added is still positive. Environmental (water conservation and contamination) constraints and market demands (for products from organic farming), however, often require supplies below the economic rate. The aim of this paper is to highlight and integrate recently developed promising quantitative approaches in optimizing light, water and nitrogen

use efficiencies and to discuss how these approaches may be made operational in arable farming. Firstly we illustrate that the length of the growing season of any crop is determined by its temperature requirement that then determines during which part of the year the crop can be grown. The intercepted solar radiation during that part of the year determines the potential amount of dry matter that can be produced. Genotypes able of reaching such potential yields should have a length of the growth cycle matching that of the growing season. Secondly it is shown that attainable yields under water limited (rainfed) conditions are lower and genotypes (or species) should have a shorter length of a growth cycle than under irrigated conditions. The trade off between lower yields due to reduced solar radiation in parts of the year but increased yields due to improved water use will be illustrated. Finally, the nitrogen uptake of the crop depends on the yields as dictated by light interception and water use. It will be shown that water should not be limiting if optimal use of nitrogen is required. Split dose application based on soil and newly developed crop observation techniques may offer the best scope for improved used of this resource.

2. Radiation interception and radiation use efficiency

Breeding creates genotypes with the highest economic yield, with a specific quality for a specific environment. The specific environment is mainly determined by temperature. In environments with mean monthly temperatures below 5°C or above 25°C, for instance, potato normally is not grown commercially. In subtropical regions rice is often grown during the warmest part of the year and potato (or wheat) during the cooler part of the year. This is illustrated in Fig. 1. In the Mediterranean area the growing season is limited in spring by low temperatures at planting and by high temperatures and drought towards harvest. In northern Europe low temperatures limit the length of the growing season in spring (late frosts in March, April and May but also at the end of the season temperatures are low again (early frosts in September and October). Beside temperature, other factors may determine

193

T, R, DL

•

ooe •

qbo e e

I

•

°ee e oQ

ot

•

l

•

I,

'

;potato,'-; ! e

time

I

rice------), |

0

Fig. 1. Schematical representation of the growing seasons of potato and rice related to the temperature and radiation regimes in a subtropical area. T =temperature (drawn line), R = solar radiation and DL = daylength (broken line).

the available length of the growing season. In tropical highlands, for instance, temperatures are suited for potato production throughout the year. Rainfall at the equator, however, is in two main rainy seasons six months apart, necessitating two crops per year. Theoretically this would restrict the length of the growing season to 6 months for ware potato crops as part of this crop is used as seed for the next crop because no separate seed potato production system

exists at the farmers' level. But practically this period has to be reduced by 2 to 3 months because the tubers that are harvested, start to sprout again after two to three months only. Ideal genotype (ideotypes) should then have a length of the growing cycle of 100 days. Longer cycles would lead to an imbalance of the growing and seed rest periods. In temperate climates rainfall often determines the workability of the soil and hence planting or sowing and the harvest period. Another important factor that may limit the length of the growing season and consequently the length of the growth cycle is related to market prices. Prices are often higher at the beginning of the harvest period of crops that are harvested fresh, necessitating a reduction of the length of the season. Ideotypes have a length of the growth cycle characterized by a green leaf area that maximizes interception of solar radiation during the available growing season to accumulate as much dry matter as possible. Earlier genotypes, too early divert dry matter to the harvestable parts (grains or tubers) so that not sufficient assimilates are available for the foliage that then senesces and dies. Genotypes that are too late still have full ground cover with green leaves at the end of the available growing season which is indicative of an unfavourable distribution of dry matter to the foliage and to the harvestable parts Yield (g/m2)

1SO0 Ytolal

GROUND COVER 100%

,

, ,,

.,,

,,

tulm¢

1.s to 2.s)

0%

, •

PI

Emer. Full cover

,.,

,.,

_

i

Senem:ence

1000 MJ/m2 INTERCEPTEO RADIATION

Fig. 2. Schematical representation of tuber production in potato based on ground cover (left) going from 0 (between planting (Pl) and emergence (Emer) to 100% until the onset of senescence. The slope of total (Ytotal) and tuber (Ytuber) dry matter production in the relationship between yield in g per m 2 and intercepted solar radiation in MJ per m 2 is the light use efficiency (E).

194

of the crop. To identify ideotypes with the desired length of the growth cycle first the length of the available growing season is determined as it is restricted by adverse growing conditions or market demands. Secondly, an assesment is made of the yield determining factors (temperature, radiation, daylength and cultivar) that cannot be changed nor influenced by the farmer once the crop is planted, with emphasis on the influence of such factors on the length of the growth cycle. A simple model describing growth and development of crops is based on light interception, utilization of light to produce dry matter, allocation of dry matter to the harvestable parts and of the percentage of water in the harvestable parts. Haverkort and Kooman (1997) descibed the use of crop growth modeling in breeding for genotypes with the aid of such a LINTUL model" Light Interception and Utilization of Light, based on the principle that the amount of light that is intercepted by the crop is converted into crop dry matter through a conservative light use efficiency (Equation 1). This principle is illustrated in Fig. 2 for the potato crop. The dry matter distribution pattern as influenced by temperature and photoperiod between foliage and harvestable produce determines the length of the growing season and is used to genetically match the length of the growth cycle with that of the growing season.

3. Water uptake and water use efficiency

Even in a region where it rains relatively abundantly, such as in north western Europe, lack of water is one of the factors most limiting growth and quality of crops (Haverkort and Goudriaan, 1994). The precipitation deficit during the growing season from April through September, in the Netherlands for instance, usually exceeds 100 mm. Many soils do not have sufficient water storage capacity within the rooted zone to cover this deficit if crops are not irrigated. Irrigation is often not possible because of its high costs or because of legal restrictions associated with its extraction. In southern Europe similar observations apply during the actual growing seasons in winter and spring of annual crops such as potatoes and wheat. These crops are harvested be-

fore the highest temperatures and levels of evaporative demand are reached. The efficient use of the available amount of water is of increasing concern to maintain production and quality. A dry spell at the beginning of the growing season leads to retardation of emergence and early growth. A short transient drought period in the course of the growing season, however, may only slightly reduce growth but it may strongly affect crop development and quality of the produce. A terminal drought that intensifies in the course of the growing season (Mediterranean) and long dry spells in the second half of the growing season (temperate) have the greatest influence on yield. These kind of droughts are more frequent than early drought as crop transpiration increases and water pools from the winter period become exhausted. Crops then form fewer new leaves and accelerate leaf shedding leading to a premature senescence In a series of experiments on light sandy soils with four potato cultivars which were either irrigated or not, periodic harvests were carried out to determine the value of the yield components given in Equation 1. The water stress occurred after tuber formation between mid June and mid August i.e. a long period of transient drought. The results are shown in Table 1. The losses due to drought were highest for the relatively early cultivar Darwina (-45%) and smallest for the latest cultivar Elles (-20%). Yield losses were mainly due to a reduction of intercepted radiation because of earlier senesecence. The light use efficiency was affected less than intercepted radiation. The cultivar that suffered little losses of intercepted radiation showed a stronger reduction in light use emciency (Elles -10%) than cv. Darwina (-1%) which reacted mainly to water stress by leaf shedding. About 75% of the total amount of dry matter produced by the crop, is found in the Table 1 Relative values of yield components (Equation l) of rainfed versus irrigated (=100%) plots for four potato cultivars of increased lateness (Haverkort et al., 1992) Cultivar

Y

R×

RUE ×

HI/

DMC

Desiree Darwina Mentor Elles

77 55 73 80

88 62 87 93

99 99 97 90

94 94 97 95

105 105 111 101

195

tubers by the time of crop senescence (HI = 0.75). Drought reduced the harvest index only slightly and no differences among cultivars were found. The tolerance of the cultivar Elles for drought lies in its abundant formation of foliage associated with its lateness. Whereas drought reduced its light use efficiency, the cultivar effectively overcame relatively long periods of absence of precipitation. Making use of the principle shown here Haverkort and Goudriaan (1994) quantitatively demonstrated through modeling that early cultivars of potato obtain higher yields than late cultivars in Mediterranean conditions where it does not rain in summer whereas late cultivars obtain the highest yields in temperate climates with a modest precipitation deficit in the middle of the growing season. For production of fodder, producers, beside quality, may choose from crops depending on water availability and timing of the harvest. Table 2 shows the water-use efficiency of several fodder crops. Maize seems to be a very efficient user of water expected on the basis of its assimilatory (C4) pathway. The low efficiencies of lucerne are associated with reduced dry matter accumulation at the cost of fixing nitrogen. Grass has a low water use-efficiency and is further hampered by an imperfect stomatal regulation and a low harvest index as grass partitions a relatively great amount of dry matter to its roots. Fodder beet has two advantages over many other crops" it has a low transpiration coefficient and a high harvest index. There must be other reasons than resource use efficiency why the bulk of fodder in Europe does not consist of fodder beet but rather of grass. These are related to the advantageous regular supply, protein quality and storability of grass compared to beet. Table 2 shows the water-use efficiency of maize to be about 6 g harvestable crop dry matter per litre of water used. It is, however, important to supply the crop with water at a crucial moment during its development. For potato this is during tuber formation. Haverkort et al. (1990) showed that following a short dry period at tuber initiation, yields were not much decreased but the number of tubers was decreased considerably exerting a strong influence on the quality (size grading) of the produce. Sink capacity in potato apparently did not limit growth in the range considered. For maize the situation is dif-

Table 2 Water-use efficiencies (g dry matter harvestable produce per litre of water used by the crop) of some fodder crops (after Aarts et al. 1996) Crop ryegrass lucerne silage maize fodder beets

optimal water suply droughtedconditions 1994

1995

1994

1996

2.86 2.16 6.02 4.55

2.99 1.42 6.29 4.44

2.78 2.10 5.49 4.55

2.69 1.66 6.33 4.72

ferent. When water is not sufficiently available during cob initiation, cob density is low resulting in limited sink capacity that through its feedback to assimilation leads to a reduction of growth far more than proportional to reduced water supply (Artlipp et al., 1995). When a particular resource is in short supply, its efficiency increases. The radiation use efficiency (Equation 1) and the water use efficiency (Equation 2) increase when light respectively water are in minimum supply. So a high resource efficiency is associated with the short supply of the resource. Therefore the best strategy for a high yield is to increase the availability of the resource. Increasing water use under rainfed conditions consequently is better achieved by optimizing water uptake than the water use efficiency. In Mediterranean conditions this can be achieved by matching the crop cycle with the rainy season which has a lower evapotranspirative demand, and by allowing a high proportion of water to be spent in transpiration with minimum losses to evaporation, drainage and runoff. This has led to earlier sowing dates, certain management practices and to selection of cold-resistant resistant cultivars. The effects of these strategies on yield, evapotranspiration (ET) and water use efficiency (WUE = Yield/ ET) are shown for field-grown sunflower in southern Spain (Gimeno et al., 1989) and for simulated rainfed faba beans in central Spain (Diaz-Ambrona et al., 1996) (Table 3). Sunflower can be sown early in winter in warmer regions, while in the case of colder areas in central Spain, sowing dates should be earlier in order to allow for crop establishment when soils are still warm, go through the winter in a cold resistant growth stage, and rapidly cover the soil in April. Under terminal water deficits or terminal drought,

196

Table 3 Yield, evapotranspiration (ET) and water use efficiency (WUE expressed as yield in kg per ha per mm rain fallen during the growing season) of winter- and spring-sown sunflower and faba beans in southern Europe under rainfed conditions Crop

Sowing date

Yield (kg ha -1)

Seasonal ET (ram)

WUE (kg ha -1 mm -1)

Sunflower

December 15 March 15 October 31 November 30 February 1

3050 640 3491 2408 1025

481 372 272 269 259

6.34 1.72 12.80 8.95 3.96

Faba beans

to apply the nitrogen at adequate quantities when the crop needs it. The efficiency of the use of an external resource is highest when all other resources are applied at their optimal level. Therefore it is essential that water does not limit growth if optimal use of nitrogen is required. Seligman and van Keulen (1981) described a model PAPRAN (Production of Arid Pastures limited by Rainfal And Nitrogen) including a quantitative description of the water and nitrogen balance in the soil, their availability to the crop (a mixture of annuals) and their effect on crop growth. The model was validated with data collected between 1971 and 1981 in Israel. Results of simulation studies, in which the interactions between water and nitrogen availability were examined, showed that the response to these two factors in terms of dry matter production could not be described simply by Liebig's 'law of the

an earlier sowing date may increase water use during the vegetative growth in detriment to the grain filling period. It would be necessary to match the length of the phenological stages to the pattern of water availability, especially under terminal drought.

4. Nitrogen uptake and nitrogen use efficiency Nitrogen fertilization is of importance for a number of reasons. As yield is much related to the amount of nitrogen available to the crop, and to avoid the risk of sub-optimal fertilization, farmers tend to over supply the crop. Excess nitrogen (from previous seasons, from mineralization and from application) may lead to excess nitrogen remaining in the soil at harvest, susceptible to leaching in the winter. These tendencies make it increasingly important

• 0 kg • 30kg. o 60kg o 90kg 6 1 2 0 kg 150 kg

primary production ( kg ha -I )

6000 -

N ha "1 . .... . . ... .. .

A

m e a n annual r a i n f a l l , 132mm •264mm primary production o 396mm ( kg ha " ! ) • 52 7 m m

6000

2 4 kg h a " m m "

/

sooo -

i

/ /

~ /

: ~

sooo

o

///logo 4000

-

~

~

4000

r-"---" ///.~,//s

3000"

2000 1000

0

3000

~..-"'~'~'--"

2000~r/

// 100

1000 ! ~ • ~ . _ o . _ _ _ . ,

200

300 400 500 600

m e a n annual r a i n f a l l

(ram)

[

0

..

i

30

i _

60

nitrogen

-_ I

90

,

I

120

• i

150

opplicOtaon ( kg hO"1 )

Fig. 3. Simulated response of mean primary production over a 21-year period to mean annual rainfall at different levels of nitrogen nutrition (A) and to nitrogen nutrition at different mean annual rainfall levels (B). In both cases soil depth is 1.5 m.

197

minimum' (Fig. 3). The maximum efficiency of rainfall utilization (kg dry matter ha -1 mm -~ rainfall) was 24 at high nitrogen availability, and declined to about 10 in the situation were nitrogen was the main limiting factor. Maximum nitrogen use efficiency (kg dry matter/kg N) was 40, at high rainfall, which is close to the biological maximum for C3species, whereas at low rainfall (hence where water is the main limiting factor) the value was only 5. In the intermediate range of nitrogen and water availabilities, the two factors appeared to be limiting at different times during the growing season, and the effect shows an interaction between the two factors. However, in all cases it is clear that the efficiency of utilization of one resource increases when the supply of another resource is closer to its optimum level. Crop nitrogen content increases with crop age and the amount of dry matter accumulated although its concentration decreases. The required (by the crop) nitrogen supply from the soil (from fertilizer and from mineralization), at any moment, depends on the amount of nitrogen already present in the crop, the expected amount still to be taken up until harvest, and the availability of mineral (mainly nitrate from mineralization of organic matter) nitrogen in the soil. Lack of information on the crop nitrogen content when sampling the soil is a bottleneck in the development of nitrogen fertilization expert systems. Apart from massive destructive sampling, presently no alternative methods are available to assess the crop nitrogen content. Reflectance and gas exchange characteristics may yield information on single leaves but research needs to find an answer to the question LAI

Brussels sprouts

/

rZ=O'913

4+

=

how to non-destructively determine the crop nitrogen status. Study of crop nitrogen uptake using crop ecological principles and methodologies, systems analysis and modeling in relation to genetic and environmental conditions is needed to implement expert systems aimed at optimizing nitrogen fertilization. Split dose application of nitrogen is becoming increasingly popular in crop production. Part of the total expected nitrogen requirement of the crop is then applied before or at planting and another part is applied later during the crop cycle. An advantage of split application, compared to one single application before planting is that this practice reduces the environmental risk of losses through leaching, volatilization and immobilization and it offers the financial possibility to reduce the total amount to be applied. Reduced nitrogen application aimed at adjusting application rates to crop needs reduces emission of nitrate and ammonia to the environment and enhances the environmental friendliness of crop production. Basic research is needed to develop methodologies for the assessment of crop nitrogen contents at the same moment that the soil is sampled for the amount of mineral nitrogen still present in the soil. Based on these two data a more sound advice for supplemental nitrogen fertilization can be given than based on soil mineral nitrogen only. The central hypothesis is the general applicability of the relationship between crop nitrogen content and its leaf area index (LAD. The leaf area index of the crop at the moment of sampling seems to be related to the total amount of LAI

" f

|

°y"

Leeks r2=0"975

4

•

I i Ig

2

2

0

" 0

0 100

200

N-uptake (kg/ha)

300

0

100

200

300

N-uptake (kg/ha)

Fig. 4. Relationship between leaf' area index (LAD and total amount of nitogcn taken up by the crop in leeks (Booij et al., 1996. Different symbols represent different N-fertilizer levels.)

198

The suitable temperature range for a crop The suitable period for crop, potential yield calculated from intercepted solar radiation Water limited length of the available growing season Attainable total dry matter yield and harvest index Ideotype identification with length of the cycle matching that of the season Nitrogen requirement of the crop calculated from water limited yields Split nitrogen application depending on availability in soil (sampling) and crop (sensing) Fig. 5. nitrogen taken up by the crop (Fig. 4). This recent finding may offer elegant possibilities to non-destructively asses the amount of nitrogen present in the crop. The expected total uptake leads to the amount that is still required. The crop requirement should be present in the soil or be applied. Two methods of LAI (thus crop nitrogen content) assessment are presently under investigation. The first is through modelling as it is known that the relative leaf extension rate as a function of temperature has a conservative value. Emergence date and the initial amount of leaf area at emergence are then crucial data. A second way is through non-destructive measuring with the aid of infrared reflectance, The second method probably will approach the LAI-value better than the first method.

5. Conclusions

In this paper it is illustrated that to optimally use solar radiation, water and nitrogen in a given environment, the following conditions must be met. The genotype used should have a length of a growth cycle (from emergence to senescence) that matches the length of the growing season. The season firstly is determined (limited) by temperature and secondly it may be further reduced when water is a limiting factor in (partly) rainfed conditions. Once the season has been defined, water limited yields can be calculated and the proper degree of cultivar lateness can be defined. Subsequently the crop nitrogen need is

calculated and the nitrogen supply at any moment is synchronized with future need until harvest. The procedure is schematically represented in Fig. 5. The approach described in this paper, i.e. matching of supply and demand of water and nitrogen of genotypes that have a length of a growth cycle that matches that of the growing season, is inspired by and based on research papers and current research. It is expected that it will assist in the need in modern agriculture to substitute inputs by knowledge such as to increase the efficient use of scarce or potentially environmentally harmful substances.

References

Aarts, H.F.M., Grashoff, C. and Smid, H.G., 1996. Evaluation of perennial ryegrass, lucerne, silage maize and fodder beets under drought. Grassland and land use systemsG. Parente, J. Frame and S. Orsi (Eds.). Proceedings of the 16th meeting of the European Grassland Federation: 363-366. Artlipp, T.S., Madison, J.T. and Settler, T.L., 1995. Water deficit in developing endosperm of maize; cell division and nuclear DNA edoreduplication. Plant Cell Environ., 18, 1034-1040. Booij, R, Kreuzer, A.D.H., Smit, A.L. and van der Weft, A.K., 1996. Effect of nitrogen availability on light interception, dry matter production and nitrogen uptake of Brussels sprouts and leeks. Neth. J. Agric. Sci., 44: 3-19. De Wit, C.T., 1992. Resource use efficiency in agriculture. Agric. Syst., 40: 125-151. Diaz-Ambrona, C.H., Conde, J.R., Hoyos, P. and Minguez, M.I., 1997. Simulation of water consumption in a cereal-legume rotation in a Mediterranean environment. International Congress of Agricultural Engineering. Madrid, September 1996. (in press.)

199 Gimeno, V., Fern~indez-Martinex, J.M. and Fereres, E., 1989. Winter planting as a means of drought escape in sunflower. Field Crops Res, 22: 307-316. Haverkort, A.J., Boerma, M., Velema, R. and Van de Waart, M., 1992. The influence of drought and cyst nematodes on potato growth. 4. Effects on crop growth under field conditions of four cultivars differing in tolerance. Neth. J. Plant Pathol., 98: 179-191. Haverkort, A.J. and Goudriaan, J., 1994. Perspectives of improved tolerance of drought in crops. Asp. Appl. Biol., 38: 79-92. Haverkort, A.J. and Kooman, P.L., 1997. The use of systems analysis and modelling of growth and development in potato ideotyping under conditions affecting yields. Euphytica (in press).

Haverkort, A.J., van de Waart, M. and Bodlaender, K.B.A., 1990. The effect of early drought stress on numbers of tubers and stolons of potato in controlled and field conditions. Potato Res., 33: 89-96. Loomis, R.S. and Connor, D.J., 1992. Crop Ecology: productivity and management in agricultural systems. Cambridge university press, Cambridge, 538 pp. Seligman, N.G. and van Keulen, H., 1981. PAPRAN: A simulation model of annual pasture production limited by rainfall and nitrogen. In: M.J. Frissel and J.A. van Veen, Eds. Simulation of nitrogen behaviour of soil-plant systems. Pudoc, Wageningen pp. 99-121.

This Page Intentionally Left Blank

© 1997 ElsevierScience B. It. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geij'n (Editors)

201

Soil-plant nitrogen dynamics" what concepts are required? E.A. Stockdale a'*, J.L. Gaunt a, J. V o s b aSoil Science Department, IACR-Rothamsted, Harpenden, Herts., AL5 2JQ, UK bDepartment of Agronomy, WageningenAgricultural University, P.O. Box 341, 6700 AH Wageningen, The Netherlands

Accepted 2 June 1997

Abstract

Soil-plant N dynamics lie at the heart of some of the questions being asked of researchers by farmers, environmentalists and policy makers. Our aim in this paper is to highlight areas in which research is needed to address these questions. Although we have a general understanding of many processes, fundamental understanding of the processes of the soil-plant system is not complete. Improved understanding of crop and soil processes should lead to the continuous improvement of simulation models, which are able to integrate the complex effects of management and environmental factors. However, farmers cannot wait for the achievement of perfect models, but need researchers to put their current knowledge to use. We suggest that for both crops and soils, diagnostic measurements be used in conjunction with the best of, or a combination of, current models. Development work should be carded out with all possible speed to draw suitable models and diagnostics together. We stress the importance of conducting research to understand and improve N efficiency at a range of scales and indicate the need for the involvement of related disciplines, such as statistics, to allow the development of robust guidelines and methodologies for up- and down-scaling measurements and models. Above all, it is essential that our understanding of the processes of soil-plant dynamics continues to underpin the development of strategies for dynamic optimisation and improve simulation models that are used in fertiliser recommendation systems. © 1997 Elsevier Science B.V. Keywords: Simulation models; Diagnostics; Dynamic optimisation; Scaling

1. Introduction

Research into soil-plant nitrogen (N) dynamics has been carried out from the very beginnings of agricultural investigation (Russell, 1966). The soil-plant N cycle is composed of a large number of complex and interacting processes which transform and transport nitrogen in, out and throughout the soil-plant system (Fig. l). Within the disciplines of both plant and soil science, countless experiments have been performed * Corresponding author. Tel.: +44 1582 763133, x 2665; fax: +44 1582 760981; e-mail: [email protected]

in laboratory and field to elucidate the mechanisms of N cycle processes and their interactions with one another and the environment, as examination of the literature rapidly shows. Recently mathematical modelling has begun to integrate our understanding of the soil-plant N cycle and the soil, plant, environmental and management factors which govern it. Models draw together current knowledge and hypotheses about biological systems, their subsystems and interactions and are able to integrate scattered data into a coherent whole (Jenkinson, 1990). However, the complexity of the cycle and the large number of interacting factors which control it means that, even for

Reprinted from the European Journal of Agronomy 7 (1997) 145-159

202

temperate agricultural soils, our models do not closely approach reality (de Willigen, 1991). Soil-plant N dynamics lie at the heart of some of the questions being asked of researchers by farmers, environmentalists and policy makers. Farmers seek to apply economic optimum rates of fertiliser, considering the costs of application and the effect on crop quality as well as yield (Neeteson and Wadman, 1987; Vos, 1995). Farmers therefore seek answers to questions about components of their system, e.g., the amount of N released for a following crop when a ley is ploughed or the fertiliser value of their manure. Environmental concerns are focused on nitrogen losses from soils which may pollute the environment. Leaching is the major route by which nitrate enters ground and surface waters, while denitrification and nitrification are significant sources of N20, an important greenhouse gas (Royal Commission on Environmental Pollution, 1996). Improved efficiency of N use at a field and farm scale, both increasing crop yield and quality and reducing losses, is dependent upon dynamic optimisation to match supply of N and the N requirements of the crop at a field scale. This optimisation requires measurement and prediction of soil N supply, crop uptake and their variability (Vos and Marshall, 1994). Policy makers seeking to improve resource use in agriculture or reduce emissions of a pollutant on a national or regional scale usually express their questions to researchers at a catchment or landscape scale. Although their questions may be similar to those of farmers or environmentalists, at such scales a diverse range of managed and natural ecosystems are included and optimisation must include a consideration of the balance and interaction between these ecosystems, as well as their individual efficiencies (Tamm, 1991). Soil-plant N cycle processes assume changing levels of importance as scale changes and different measures are required to influence the N flows (Vos, 1996). Aggregation is used to simplify the pools measured or estimated (Robertson, 1982) but the methodologies available for up-scaling plot and field measurements to a catchment or regional scale are primitive. Our aim in this paper is not to provide a comprehensive review of the literature on soil-plant N dynamics, but to highlight areas in which research is needed to address the questions of farmers, environ-

mentalists and policy makers. Our question is: What concepts are required to complete our understanding of soil-plant N dynamics and which principles can be applied practically to improve management of the N cycle now? We put forward some promising research concepts in four main areas: soil N supply; crop N uptake, N losses from the soil-plant system and the application of data and models at a range of scales.

2. Soil N supply Recently emphasis has been placed on the measurement of inorganic N (usually nitrate) in soil before planting or at a specific time during the crop growing season to assess soil N supply (the 'Nmin' approach). In climates without extensive overwinter leaching, the use of a pre-planting test for soil nitrate is widely used; in the western United States before 1982, 16 states were using such a test (Keeney, 1982). A presidedress nitrate test has been developed for maize (Magdoff et al., 1984), which allows a period of nitrate accumulation under field conditions before sampling. Similar approaches are used in many countries in continental Europe and these were summarised by Mengel (1991). Snapshot measurements of soil inorganic N can only give a partial prediction of soil N supply since the bulk of soil N is found in organic forms. In a long term monitoring scheme, the average amount of mineral N measured in topsoil during autumn in the UK was only 76 kg ha -z compared to 7000 kg N ha -~ found in the soil organic matter (Shepherd et al., 1996). After comparing 14 models of the soil-plant cycle, de Willigen (1991) concluded that the biological processes controlling N supply were not well simulated, reflecting our incomplete understanding of the processes of mineralisation, immobilisation and their controls. Mineralisation is the process by which ammonium is released by soil micro-organisms as they utilise soil organic materials as an energy source, while immobilisation of ammonium and nitrate by micro-organisms is determined by the demands of protein synthesis. These processes reflect the properties of the substrate being mineralised and its interaction with the environment. The balance between them largely controls N

203

supply from soil. Measurement of mineralisation has until recently been limited to determination of net rates, which are the integration of a number of soil N processes. Relationships with environmental and management factors are difficult if not impossible to determine, since the factors act on other interacting processes as well as directly on mineralisation and immobilisation (Jarvis et al., 1996). Measurement of gross mineralisation rates by applying 15N and considedng isotope dilution theory (Kirkham and Bartholomew, 1954) enables mineralisation to be distinguished from immobilisation (e.g., Wessel and Tietema, 1992). Gross mineralisation rates may be able to be further resolved into the supply from a number of organic matter pools (Barraclough, 1995). The application of this technique under laboratory and field conditions provides the opportunity to establish a clearer mechanistic understanding of how the composition of soil organic matter and plant residues influence nitrogen mineralisation. Where pool dilution experiments are combined with direct tracer methods, determining the fate of N as well as rates of mineralisation, then with careful interpretation, we will be able to rapidly increase our understanding of soilplant N dynamics. Soil organic matter is not homogenous and is composed of a continuum of materials stabilised against mineralisation to varying degrees (Skjemstad et al., 1988). Chemical fractionation techniques have been used to define organic matter structures and have shown the presence of a wide range of functional groups (Stevenson, 1982). However, the fractions obtained by destructive chemical techniques have not been clearly related to soil N supply. Laboratory incubations of soil under various conditions have been used widely to indicate the potential of soils to supply N. These biological incubations and the quicker and more precise chemical extractions are believed to release available N pools preferentially. New methods continue to be developed and both approaches have been reviewed at length (Harmsen and van Schreven, 1955; Keeney, 1982; Jarvis et al., 1996). The limitations of N availability indices have been recognised; they were not expected to integrate the numerous inter-related soil, plant, environment and management factors which control N release and plant growth (Bremner, 1965), but simply to p r o vide extra information for assessment of soil N sup-

ply. N availability indices have not been used widely as part of fertiliser recommendation systems (Keeney, 1982). Today chemical and biological indices are thought to provide only a relative indication of N availability among soils differing in management (Bundy and Meisinger, 1994) and as such are used with other indicators to assess soil quality (FrancoVizcafno, 1997). Applying solid state J3C-NMR spectroscopy to whole soils (Wilson, 1987) indicates, in some cases, that carbohydrate and aliphatic compounds are more labile than other classes such as aromatics (Kinesch et al., 1995). However, Randall et al. (1995) concluded that for whole soils the proportions of different chemical species remain remarkably constant in response to different long-term management practices. The importance of the physical location of organic matter within the soil matrix, as well as its chemical composition, in influencing resistance to decomposition has been acknowledged widely (Oades et al., 1988). Soil organic matter physically separated into sand, silt and clay size fractions has been shown to decline in soils at different rates (Dalai and Meyer, 1986). Densiometric techniques have been developed to separate labile organic matter, usually after a partial or complete disruption of soil aggregates (Christensen, 1992). These labile fractions have been correlated to both soil microbial biomass and nitrogen mineralisation (Janzen et al., 1992; Hassink, 1992). The division of soil organic matter into pools of material, which behave similarly, is at the heart of many simulation models of soil carbon and nitrogen dynamics. However, a fundamental, and widely recognised deficiency of current models is that these pools are assumed to have biological significance, but are impossible to measure (Christensen, 1996; Jarvis et al., 1996). The Slow and Passive pools of organic matter of the Century model (Parton et al., 1987) and the Resistant, Decomposable and Inert organic matter pools of the Rothamsted Carbon Turnover Model (Jenkinson, 1990), are examples of these. Physical fractions obtained after destruction of soil structure prior to sieving at 53 #m (Cambardella and Elliot, 1992) have been related to the protected organic matter (POM) fraction of the Century model (Parton et al., 1987). The development of non-invasive and physical methods of dividing soil organic matter into pools may allow functional pools in models to be defined

204

and measured. However, the relationship of physical fractions to mineralisation still needs to be more clearly established. It is important to realise that the division of organic matter into a number of pools represents a simplification of the continuum of soil organic matter, both in terms of chemical characteristics and physical location, which exists in soils. The achievement of measurable model pools is not therefore the ultimate aim. Relationships between soil structure, management and N turnover are included rarely and inadequately in models (Elliott et al., 1996) and improved prediction of N supply from soil organic matter and crop residues will depend on understanding and quantifying all these effects. Improved understanding and prediction of mineralisation and immobilisation alone will not enable us to predict soil N supply, since mineralisation is only one of the processes of the soil N cycle. Models of the soil N cycle (e.g., ANIMO, Rijtema and Kroes, 1991; DAISY, Hansen et al., 1991; SUNDIAL, Smith et al., 1996) explicitly incorporate descriptions of the processes of soil N turnover (Fig. 2) and enable the effects of seasonal variations and timing of manage-

ment practices to be simulated. The integration of an improved understanding of mineralisation into a complete model of the soil N cycle is necessary to improve predictions of soil N supply. Using rapid and accurate field measurements of soil mineral N, along with estimates of mineralisation and denitrification, fertiliser inputs to grassland have been reduced by 30% whilst maintaining production levels (Titchen and Scholefield, 1994). Such a combination of field measurement and predictive modelling seems to be the way forward to achieve practical prediction of soil N supply for farmers. However, models are not yet widely used in fertiliser recommendation systems and it is unclear whether field measurements of soil inorganic N, N availability indices or organic matter pool sizes could be used to tune the models to specific field situations or adjust model recommendations during the growing season.

3. Crop N uptake Opportunities exist to modify the physiological

SOIL

PLANT

v

v

f Reduced •~,organic N Senesence//Glutamine /. synthetase

Volatilization Fertilizer Dry and wet Absorption deposition

1 1

//Nitrite reductase ~' / (dark and light)

l

~

Nitratereductase

~(dark

Absorption \

and light)

,

Denitrification

T lN;ol

Fertilizer Dry and wet deposition

t

Nitrification

1

] M,,,'~mmobilizatioJ ~ineraization

[!iiJ}iiiiiiii}~Up.,take

Volatilization Root/soil interface

OrganicN

1

V Leaching

Fig. 1. Simplified nitrogen cycle showing pools and processes in plant and soil.

205 efficiency of plants' use of NO3 (Harper, 1994). Nitrate reductase has a key role in nitrate metabolism and may be a point of metabolic limitation (Eichelberger et al., 1989; Campbell, 1990). Research is also being carded out to elucidate nitrate transport mechanisms (McClure et al., 1990) and to identify and clone nitrate uptake genes (Tsay et al., 1993). The gene encoding high affinity nitrate uptake in higher plants has not yet been identified. This paper does not address this research in detail. However, increased plant uptake efficiency of NO3 could significantly affect N losses as well as crop growth and yield. Efficient use of nitrogen in crop production requires an understanding of the relationship between nitrogen application, soil N supply, nitrogen uptake and that between uptake rates and growth rates. The so-called 'three quadrant diagrams' (e.g., de Wit, 1992) have been used to analyse crop nitrogen response retrospectively. The first quadrant is a plot of crop yield versus N applied, i.e., the overall or 'agronomic response'. In quadrants 2 and 3 the relationship between nitrogen uptake and nitrogen applied, determined primarily by soil processes, and the relationship between nitrogen uptake and yield are presented. Such plots are relatively easily obtained but are much more enlightening than the simple standard response curve, presenting hypotheses about the relationships between nitrogen application, soil N supply, nitrogen uptake and yields

Atmosphere I I Fertiliser

I

IOrganic Manure

Biomass Ammonium

Debris

Nitrate

Stubble & Straw Crop

Denitrificationl I Leachina I IHarvestl I Volatilisation

Fig. 2. Processes included in the Rothamsted Nitrogen Turnover model, a central part of SUNDIAL, from inputs through transformations to outputs. Nitrogen turnover is considered as a set of simple transformationprocesses (arrows)between discrete N compartments (boxes), where mineralisationis only one of the complex links.

with different application timings or placements (Black, 1993). A major limitation to three quadrant diagrams is that they do not provide any information about the dynamics of crop or soil processes. The relationship between N uptake rate and growth rate is described by the physiological efficiency of N use for a crop. Ingestad and Agren (1992) demonstrated that during exponential growth, the relative growth rate is proportional to the relative N uptake rate when this is constant. Nitrogen concentration in the plant is then also stable and controlled by the ratio between the relative uptake and relative growth rates. This approach allows the determination of physiological plant responses to N applications at a constant 'relative addition rate'. Although such an approach is important for increasing our understanding of the controls on plant growth, it is difficult to apply in the field. Linear rather than exponential growth occurs following canopy closure, nitrogen supply is depleted continually during growth or is topped up at arbitrary intervals and internal redistribution of N within the plant affects the simple relationship between uptake and growth. Based on numerous experiments, Greenwood et al. (1986, 1990) described a fairly fixed relation between crop mass and the critical N concentration required to ensure maximal growth. C3 and C4 species show different curves (Greenwood et al., 1990). Independent analyses showed that the 'Greenwood dilution curves' hold approximately in wheat (Justes et al., 1994) and potato (Vos, 1995). The concept of critical nitrogen concentration could therefore be used as a diagnostic tool against which to compare the nitrogen status of a crop (Justes et al., 1994). Total N requirements to meet yield targets without N deficiency can be calculated for the whole season or for growth periods. In practice, stresses other than nitrogen may reduce plant production and therefore modify the function describing the critical nitrogen concentration; this seems to invalidate the concept physiologically. Therefore the critical N concentration may be better related to development stage than crop mass. Unfortunately this would mean that critical (and minimum and maximum) concentrations for each of the development stages of crops would need to be determined. The required soil N supply for any period depends not only on crop N demand, but also on the crop uptake efficiency. A measure of that efficiency is the

206

apparent nitrogen fertiliser recovery, which is defined by: (N uptake of fertilised crop - N uptake of unfertilised crop)x Fertiliser applied. Apparent nitrogen recoveries usually range from 0.4 to 0.7 with 0.8 as an upper limit (Greenwood and Draycott, 1989). Recoveries are low where conditions favour losses of nitrate from the soil by denitrification or leaching and where carbon is available net microbial immobilisation will occur at least temporarily. Root exploitation of the soil volume is never 100% effective and this will also cause recoveries to be less than unity. However, the factors and processes determining the apparent recovery and its maximum value are not entirely understood. The efficiency of uptake is rarely accounted for in fertiliser recommendations systems, except implicitly. Computer simulation models are used as tools to integrate knowledge and explore the behaviour of the soil-plant system. Dynamic simulation models have been developed for a number of crops, e.g., CERES for maize, wheat and other cereals (Jones and Kiniry, 1996); ORYZA for rice (ten Berge et al., 1994) and CROPSYST suitable for a range of crops (Stockle and Nelson, 1996). However, as yet, models are rarely used to guide practical fertiliser decisions. Most models are able to adequately simulate the crop N uptake in response to variable N supply through the season. However, the effects of variable N supply on developmental processes, particularly changes in leaf area index, are not well understood and cannot be modelled using the concept of critical N concentrations (Greenwood et al., 1990). The distribution of nitrogen between component plant parts, such as leaves, stems and storage organs, as well as the final harvest index, is relatively constant irrespective of the average nitrogen concentration of the plant (e.g., Biemond and Vos, 1992). A fairly fixed distribution of nitrogen is therefore used in many empirical models of crop growth. However, in dynamic models the internal nitrogen distribution needs to be explained as a result of the underlying processes, e.g., turnover of proteins and sink-activity of competing organs. Plant N can be crudely divided into nitrate and reduced nitrogen pools, where the latter can be subdivided into metabolically active nitrogen and inactive fractions. The inactive fraction consists of N in storage proteins (seeds, tubers, etc.) and nitrogen contained in structural elements of

leaves, stems, roots and storage organs. As development proceeds, a larger amount of nitrogen is allocated at comparatively low concentrations in nonactive pools and this causes the decline in the critical nitrogen concentration with increase in crop mass (Caloin and Yu, 1984). Practical agriculture cannot wait until all details of responses of crops to nitrogen are quantified. Fertilisation guidelines and recommendation systems developed empirically have been in use for decades and are constantly being modified (e.g., Anon, 1994). Usually a fertiliser recommendation is given only once in the season, before or just after planting. Such a recommendation is necessarily inexact, since both the soil N supply and crop N requirement are not known in advance and depend on the future weather pattern. Therefore, wherever possible, the growing season should be divided into shorter periods for which recommendations are made, but which create opportunities to adjust and top up applications of fertiliser during the course of the season. Such 'dynamic optimisation' approaches are already in use in high value crops and where fertiliser applications are routinely split, e.g., the pre-sidedress nitrate test for maize (Magdoff et al., 1984). At the beginning of each of the growing periods, a decision is taken on the nitrogen nutrition, using quantitative data on the nitrogen status of crop and soil and estimates of the requirement of the crop and the supply from net mineralisation. Precise, robust and accurate field methods are required to collect the data needed for decision support. Comparison of tissue N concentrations with critical N concentrations is impractical as a diagnostic tool, not least due to the time involved in conducting analyses for total tissue N. Nitrate concentration in plant tissues can be measured using field diagnostic kits (Nitsch and Varis, 1991) and has been suggested as a useful diagnostic indicator of N status (Barraclough, 1993; Elliott et al., 1993). Tissue testing has not been widely adopted by growers, since critical nitrate levels can vary greatly from site to site (Beringer and Hass, 1979) and can change rapidly with time (Vos and Bom, 1993). Tissue nitrate testing is also relatively time consuming while the seemingly simple operations require good analytical skills. Field monitoring of crop N status is also possible

207

using hand-held chlorophyll meters (Fox et al., 1994; Peltonen et al., 1995) and near infra-red reflectometry (Young et al., 1993). Chlorophyll concentrations have been related to tissue N concentrations in different crops, e.g., rice (Peng et al., 1993) and potato (Vos and Bom, 1993). Future research should establish guidelines to help interpret chlorophyll values and to convert these into estimates of N status and future crop N requirement. Most fertiliser response trials have been carded out with single dose treatments but dynamic optimisation of N will require an understanding of the ability of crop to capture and utilise N under varied N management. The pattern of N uptake in potato crops can be drastically influenced by the dose and timing of split applications (Vos, 1995). For wheat, delaying N fertiliser applications beyond Feekes growth stage 7 can lower the potential yield response (Lutcher and Mahler, 1988). However, for most crops there is little information about which development stage N applications will continue to influence crop growth and quality of harvested components and in what form N should be applied to achieve maximal effect.

4. N losses

The focus of environmental interest is loss processes. Matching soil N supply and crop uptake minimises the amount of mineral N in the soil solution at any time and thereby losses. Thus the research concepts outlined in previous sections will also address environmental concerns to reduce N losses. The main rate limiting process controlling the availability and loss of the mobile nitrate ion may be nitrification, the process by which ammonium is converted to nitrate, rather than mineralisation. Despite a reasonable knowledge of the ecology of the bacteria involved (Prosser, 1986), nitrification remains a poorly defined process in many soils. In temperate tilled agricultural soils, nitrification rates are usually limited by mineralisation (Harmsen and van Schreven, 1955). However, in grassland soils significant quantities of ammonium may accumulate where swards are grazed or farm wastes are applied (Jarvis and Barraclough, 1991). The balance between mineralisation and nitrification is changed under changing environmental conditions and management

(Willison et al., 1997). While the factors controlling nitrification are known at a microsite scale, more work needs to be done to understand the controlling factors operating at larger scales. The use of ~SN dilution techniques to measure gross nitrification rates in conjunction with other processes may be key in this area. The root of the problem of nitrate leaching is 'untimely nitrate', i.e., nitrate which remains in the soil after crop uptake has ceased and therefore is vulnerable to loss in drainage. Some leaching loss of N seems inevitable however efficiently N is taken up by the crop, as plant processes have higher threshold temperatures for activity than mineralisation processes occurring in the soil (Vos, 1992). Even where dynamic optimisation is used, situations arise where large amounts of nitrate accumulate in the soil profile, for example, where fertiliser recovery is unexpectedly low due to drought or disease or following crops which leave large amounts of rapidly mineralised residues. In such situations, agricultural rotations may be modified to include catch or cover crops, which are able to absorb residual N and reduce losses of N by leaching (Martinez and Guiraud, 1990; Jensen, 1991). While such practices are usually effective in their year of application, some work has shown that the N is mineralised in subsequent years, and the leaching loss is delayed, rather than prevented entirely. Nitrous oxide is produced in soils by the microbially mediated processes of nitrification and denitrification; the capacity of soils to produce N20 increases as N availability increases. There have been a number of studies of N20 fluxes from soils, and the range of emission factors is considerable, generally emissions from natural ecosystems and uncultivated lands range between 0.1 and 9.1 kg NEO-N ha-! y-i. Emission factors from agricultural lands are more variable and generally higher (Bouwman, 1990). In agronomic terms the losses of N as N20 are probably negligible, however soils are a large source of N20 which contributes approximately 5% to radiative forcing. There is considerable uncertainty associated with estimates of N20 losses from soils because the factors controlling denitrification and nitrification are complex and still relatively poorly understood and emissions are temporally and spatially variable, typical coefficients of variation are in excess of 100%. Volatilisation of ammonia is also a possible route

208 5. Scaling

for N loss and ammonia deposition can cause pollution in natural ecosystems. The factors controlling NH3 emissions from soils and short and long range transport of NH3 between ecosystems are not well understood. However, injection or rapid incorporation of ammonium or urea fertilisers and manure, slurry or sewage sludge which reduce losses of NH3 significantly are amongst management practices recommended to UK farmers (Anon, 1992) and prescribed to Dutch farmers.

-'././///..(~_~

At different scales, processes assume changing levels of importance and different measures are required to influence the flows of input, output and losses (Vos, 1996). The driving variables for soil and crop processes are independent at a given scale, but as scale increases in time or space these may interact and new independent driving variables must be defined (Elliott and Paustian, 1996). As the tem-

,e

FIELD - water, nitrification, organic matter decomposition

, t i o ~ ~~.. ~ . . ' - ' ~ matter, physical disruption

.'-'... ~. ~~ .~..~..""

.

ORGANISM - oxygen, nitrate, carbon Fig. 3. Soil N cycle processes can be studied at different scales. However, the driving variables for crop and soil processes may change; this is illustrated for denitrification (adapted from Groffman et al., 1988; Jarvis et al., 1996).

209

poral and spatial scale of investigation increases, the primary factors controlling processes at the cellular level are affected by many physical and biological factors. At larger scales these secondary factors become increasingly significant and the process can be simplified and expressed as a function of the secondary variables (Groffman et al., 1988). This is illustrated for the process of denitrification in Fig. 3. As geographic size increases, a range of ecosystems may be included within system boundaries and fluxes between ecosystems and processes occurring at depth in soil may become important. Up-scaling research from pots or small plots to field, farm or landscape scales means that by necessity fewer measurements are possible in either time or space, and the uncertainty of estimates is increased. It is usually simply not practical to divide a field into a multitude of small plots and make replicate measurements on each, never mind at the landscape scale. An understanding of the soil-plant processes studied and their controlling factors allows us to define the system at an appropriate scale with regard to the goal of the study and to qualify the sub-systems processes and pools according to their importance to the system, e.g., discarding measurements of water erosion in flat or gently sloping areas. It is also crucial that research is seen within a hierarchical framework, where the research scale, can be related to scales above and below. N budgets have been calculated at a range of scales from global (Jenkinson, 1990; Isermann, 1993) national (Vos, 1996 (Netherlands); Royal Society, 1983 (UK)), catchment (Roberts, 1987; Lord, 1992), farm (Barry et al., 1993; Granstedt, 1992) field (Vinten et al., 1992) to small plot (Jenkinson and Parry, 1989; van Faassen and Lebbink, 1994). The underlying assumption of a nutrient budget is that of mass balance, i.e., N inputs to the system minus N outputs equal the change in N storage within the system. Although the nature and amount of inputs and outputs vary among farming systems, regions and even between fields, the mass balance concept provides a framework that can be applied systematically across a diversity of systems and scales (Committee on Long Range Soil and Water Conservation, 1993). The detail with which budgets are compiled varies; Stinner et al. (1984) included inputs, storage and processing by insects and micro-organisms in their budget. Farm budgets may be simple farm gate balances (Nguyen

et al., 1995) or include details of nutrient transfers between fields and within the farming system (Guiking et al., 1994). Careful definition of the system boundaries and compartments is essential (Meisinger and Randall, 1991). The construction of budgets at increasing scales necessitates aggregation of small compartments, where the degree of simplification is a function of lack of information or understanding of certain processes and the time and effort available for fine resolution, and it is made at the expense of both detailed insight and precision (Robertson, 1982). Where much has to be estimated rather than calculated, there is a danger of producing very speculative analyses (Dodgshon, 1994). However, budgets are often constructed with unjustified certainty, when specific values are assigned to fluxes or pools for which a wide range of values would be as accurate, and with inadequate documentation so that they are unverifiable (Robertson, 1982). N budgets may allow key processes or sites of high potential loss to be identified rapidly (Barry et al., 1993), allow the assessment of the efficiencies of nutrient use in the system to be determined (Karlovsky, 1981; Fowler et al., 1993) or add to our understanding of the importance of the processes and their interactions. It is relatively easy to compile N budgets at a range of scales, so that at its most detailed a regional budget is simply a compilation of the individual catchment budgets for the important systems of a region (Robertson, 1982). However, simple input/output balances can never deal adequately with the complexity of the N flows in the soil-plant system and a fundamental understanding of soil and plant processes is also required. In theory, N flows determined within a nested hierarchical classification similar to that used for the compilation of budgets (Lanyon and Beegle, 1989) can be brought together at any level to match information requirements. However, it is very difficult to use small-scale process measurements to estimate fluxes at field and landscape scales, e.g., for N20 fluxes (Tiedje et al., 1989). The simplest approach to up-scaling is to use the mean of a simple randomised sample to estimate the mean of the whole area or time period; this can be coupled with determination of the standard deviation to give a 95% confidence level, rather than a single value for the estimate of the mean. The size of the sample is important in attaining a representative mean

210

for the property (Goovaerts and Chiang, 1993). Where the area can be divided into a number of sub-units, e.g., soil series or land uses, stratified simple random samples will give a more representative estimate of the mean or confidence interval for the whole area, without necessarily increasing the total number of samples taken (Barnett, 1991). The cost and time necessary for numerous systematic measurements may hinder the use of these simple approaches; in this case measurement of an easily determined concomitant variable or combination of such variables along with a regression approach may provide useful estimates (e.g., Parton et al., 1987). This approach can be used to determine the probability level associated with the prediction, since the range of variation in any parameter is as important as the estimated mean value (Arrouays et al., 1995). Models can also be used to generate the required process or pool size estimates from other easily obtained data; Post et al. (1996) modelled global C dynamics by dividing the earth into half-degree cells and deriving the input values needed to run a carbon turnover model by using a model of net primary production which needed only simple inputs. Geostatistical algorithms, such as kriging, can also be used to estimate values for soil properties where sufficient data are available to calculate the variogram (McBratney et al., 1982; Voltz and Webster, 1990). Where other data are also known, e.g., soil mapping, co-located indicator co-kriging or full indicator co-kriging have been shown to improve the definition of cobalt and copper deficient areas (Goovaerts and Joumel, 1995). There is increased interest in such methods not just for up-scaling but also to allow the spatial distribution of soil properties to be mapped within fields in response to the increased use of yield-mapping (Stafford et al., 1996). However, our understanding of the factors causing spatial variability in crop yield and their interactions is relatively poor at the field scale (Parkin, 1993). The impacts of spatial variability in soils on plant growth and losses is not well understood and the temporal persistence of spatial patterns, e.g., caused by previous grazing is not known. The technology necessary to implement spatial management within fields, precision farming, is far in advance of the basic scientific knowledge necessary to underpin management recommendations. There is increased demand from policy makers that

models be linked, tested under a range of environments and applied on larger scales, both in space and time. Up- and down-scaling of models is not easy; Leffelaar (1990) itemised the criteria to be considered when determining an appropriate scale at which to perform simulations. Changing the scale of a model may result in changes in the scope of the model, the heterogeneity of input values and the data requirements of the model (Smith, 1995). There is need for closer involvement of related disciplines, such as statistics, to allow the development of robust methodologies and guidelines which enable us to scale up and scale down N cycle processes and models (Gaunt et al., 1997), so that the data collected for N cycle processes at the cellular, small plot or catchment scales for example can be applied at a range of scales.

6. Conclusions

Improved N use efficiency is a common goal of the farming community and environmental lobby, albeit for different reasons. Empirical relationships have been established between crop N uptake and plant growth and to describe the partitioning of N within crops. Our understanding of the dynamics of these processes and our ability to simulate them is increasing, but as yet remains incomplete. Similarly the factors controlling many soil processes, e.g., mineralisation, nitrification, denitrification are complex and relatively poorly understood. There are many areas in which fundamental understanding of the processes of the soil-plant system needs to be increased. N mineralisation is a key soil process; we would like to highlight the importance of integrative studies using ~SN both as a tracer within the plant-soil system and to quantify the effects of the chemical and physical protection of organic matter on gross transformations. Increased understanding of the metabolic limitations and genetic basis of N use and uptake in crops will also lead to improvement of the physiological efficiency of N use. Improved understanding of crop and soil processes at a fundamental level should lead to the development of continuously improving simulation models, which are able to integrate the complex effects of management and environmental factors. We see the develop-

211

ment of models which contain measurable pools of soil organic matter as a key step forward. Models of crop growth, the soil N cycle and plant-soil models have been developed. However, these are little used in current fertiliser and farm management recommendation systems. Farmers cannot wait for our understanding of plant-soil dynamics to be perfect or for the achievement of perfect models, but need researchers to put their current knowledge to use. New diagnostic approaches for field crops such as the use of near infrared reflectance and chlorophyll meters show promise but research is needed to clearly establish the relationships between the indicators and plant N uptake. There is also the possibility that measurements of soil mineral N or some N availability indices might be used as soil diagnostics. We suggest that for both crops and soils, diagnostic measurements be used in conjunction with the best of, or a combination of, current models. Diagnostics could be used to increase the field-specific nature of recommendations or to adjust model recommendations during the growing season. This would enable a greater use of dynamic optimisation strategies in the field. At present such a vision seems a distant possibility, but development work should be carried out with all possible speed to draw suitable models and diagnostics together. Fine tuning fertiliser recommendations using dynamic optimisation strategies also requires research to establish the crops ability to capture and utilise N under varied management regimes. Research leading to improvement in N efficiency, as described above, will also impact directly on N losses from agricultural systems. However, we stress the importance of conducting research to understand and improve N efficiency at a range of scales to address the needs not only of farmers, but also environmentalists and policy makers. Careful definitions of the system under study are important and an awareness of the relationship of the system under study to the level above and below is crucial. The methodologies used for up- and down-scaling process rates and pool sizes are relatively primitive. The demand that models should also be readily up- and down-scaled has led to the need for the involvement of related disciplines, such as statistics, to allow the development of robust guidelines and methodologies. The mass balance approach is a powerful tool to compare

and evaluate systems with regard to their efficiency or potential for loss. However, we need to take care when presenting budgets; the compilation of a budget is only a first step and rarely indicates understanding of any system. We also need an increased understanding of the importance of spatial variability on crop N uptake and loss processes, if we are to provide a sound scientific basis for the development of precision farming. Perhaps most important of all is that research is not carried out in isolation and that our increasing understanding of the fundamental processes of soil-plant dynamics continues to underpin the development of strategies for dynamic optimisation, to allow the identification of and improve simulation models that are used as fertiliser recommendation systems. In this way we will be able to provide answers to the questions of farmers, environmentalists and policy makers.

Acknowledgements For all their help, advice and hard work: David S Powlson, Toby Willison, Pete Barraclough. IACR receives grant-aided support from the UK Biotechnology and Biological Sciences Research Council and the authors also acknowledge funding from the UK Ministry of Agriculture, Fisheries and Food.

References Anon. 1992.Code of Good Agricultural Practice for the Protection of Air. HMSO, London, 74 pp. Anon. 1994.Fertiliser Recommendations(sixth ed.). MAFFReference Book 209. HMSO, London, 193 pp. Arrouays, D., Vion, I. and Kicin, J.L., 1995. Spatial analysis and modelling of topsoil carbon storage in temperate forest humic loamy soils of France. Soil Sci., 159:191-198. Barnett, V., 1991.SampleSurvey.Principles and Methods. Edward Arnold, London, 105 pp. Barraclough, D., 1995. ZSNisotope dilution techniques to study soil nitrogen transformations and plant uptake. Fertilizer Res., 42: 185-192. Barraclough, P.B., 1993. Nutrient storage pool concentrations in plants as diagnostic indicators of nutrient sufficiency. Plant Soil, 155/156: 175-178. Barry, D.A.J., Goorahoo, D. and Goss, M.J., 1993. Estimation of nitrate concentrations in groundwater using a whole farm budget. J. Environ. Qual., 22: 767-775. Ten Berge, H.F.M., Wopereis, M.C.S., Riethoven, J.J.M., Thiya-

212

garajan, T.M. and Sivasamy, R., 1994. The Oryza-0 model applied to optimize nitrogen use in rice. In: H.F.M. ten Berge, M.C.S. Wopereis and J.C. Shin (Editors), Nitrogen Economy of Irrigated Rice: Field and Simulation Studies. SARP Research Proceedings, April 1994, pp. 235-253. Beringer, V.H. and Hass, G., 1979. Brauchbarkeit der Pflanzenanalyse zur Bemessung sp~iter N-gaben zu Winterweizen. Landwirtsch. Forsch., 32: 384-394. Biemond, H. and Vos, J., 1992. Effects of nitrogen on the development and growth of the potato plant. 2. The partitioning of dry matter, nitrogen and nitrate. Ann. Bot., 70: 37-45. Black, C.A., 1993. Soil Fertility Evaluation and Control. Lewis, Boca Raton, FL, 746 pp. Bouwman, A.F., 1990. Exchange of greenhouse gases between terrestrial ecosystems and the atmosphere. In: A.F. Bouwman (Editor), Soils and the Greenhouse Effect. Wiley, Chichester, pp. 61-127. Bremner, 1965. Nitrogen availability indices. In" C.A. Black et al. (Editors), Methods of Soil Analysis. Part 2. Agronomy Series 9. ASA, Madison, WI, pp. 1324-1345. Bundy, L.G. and Meisinger, J.J., 1994. Nitrogen availability indices. In: R.W. Weaver et al. (Editors). Methods of Soil Analysis. Part 2. SSSA Book Series 5. SSSA, Madison, WI, pp. 951984. Caloin, M. and Yu, O., 1984. Analysis of the time course of change in nitrogen content in Dactylis glomerata L. using a model of plant growth. Ann. Bot., 54: 69-76. Cambardella, C.A. and Elliot, E.T., 1992. Particulate soil organic matter changes across a grassland cultivation sequence. Soil Sci. Soc. Am. J., 56: 777-783. Campbell, W.H., 1990. Purification, characterization and immunochemistry of higher plant nitrate reductase. In: Y.P. Abrol (Editor), Nitrogen in Higher Plants. Research Studies Press, Taunton, pp. 65-91. Christensen, B.T., 1992. Physical fractionation of soil and organic matter in primary particle size and density separates. Adv. Soil Sci., 20: 1-90. Christensen, B.T., 1996. Matching measurable soil organic matter fractions with conceptual pools in simulation models of carbon turnover: revision of model structure. In: D.S. Powlson, P. Smith and J.U. Smith (Editors), Evaluation of Soil Organic Matter Models Using Existing Long-Term Datasets. Series 1 Global Environmental Change, NATO ASI Series, Vol. 38, pp. 143-161. Committee on Long Range Soil and Water Conservation (CLSWC), 1993. Nitrogen and phosphorus mass balances: methods and interpretation. In: Soil and Water Quality: An Agenda for Agriculture. National Research Council National Academy Press Washington, DC, pp. 431-477. Dalai, R.C. and Meyer, R.J., 1986. Long-term trends in fertility of soils under continuous cultivation and cereal cropping in Southern Queensland. III. Distribution and kinetics of soil organic carbon in particle-size fractions. Aust. J. Soil Res., 24: 293-300. Dodgshon, R.A., 1994. Budgeting for survival: nutrient flow and traditional Highland farming. In: S. Foster and T.C. Smout, The History of Soils and Field Systems. Scottish Cultural Press, Aberdeen, pp. 83-93. Eichelberger, K.D., Lamber, R.J., Below, F.E. and Hageman, R.H.,

1989. Divergent phenotypic recurrent selection for nitrate reductase activity in maize. II Efficient use of fertiliser nitrogen. Crop Sci., 29:1398-1402. Elliott, D.E., Pelham, S.D. and Reuter, D.J., 1993. Synchronising diagnosis and correction of nitrogen deficiency in barley grown in semi-arid environments. Plant Soil, 155/156: 363-366. Elliott, E.T. and Paustian, K., 1996. Why site networks? In: D.S. Powlson, P. Smith and J.U. Smith (Editors), Evaluation of Soil Organic Matter Models Using Existing Long-Term Datasets. Series l Global Environmental Change, NATO AS! Series, Vol. 38, pp. 27-36. Elliott, E.T., Paustian, K and Frey, S.D. 1996. Modeling the measurable or measuring the modelable: a hierarchical approach to isolating meaningful soil organic matter fractions. In: D.S. Powlson, P. Smith and J.U. Smith (Editors), Evaluation of Soil Organic Matter Models Using Existing Long-Term Datasets. Series l Global Environmental Change, NATO ASI Series, Vol. 38, pp. 161-180. Fowler, S.C., Watson, C.W. and Wilman, D., 1993. N, P, and K on organic farms: herbage and cereal production, purchases and sales. J. Agric. Sci. (Cambridge), 120: 353-360. Fox, R.H., Piekielek, W.P. and MacNeal, K.M., 1994. Using a chlorophyll meter to predict nitrogen fertilizer needs of winter wheat. Commun. Soil Sci. Plant Anal., 25:17 l - 18 I. Franco-Vizcafno, E., 1997. Comparative soil quality in maize rotations with high or low residue diversity. Biol. Fertil. Soils, 24: 32-38. Gaunt, J.L., Riley, J., Stein, A. and Penning de Vries, F.W.T., 1997. Requirements for effective modelling strategies. Agric. Systems, 54: 153-168. Goovaerts, P. and Chiang, C.N., 1993. Temporal persistence of spatial patterns for mineralizable nitrogen and selected soil properties. Soil Sci. Soc. Am. J., 57: 372-381. Goovaerts, P. and Journel, A.G., 1995. Integrating soil map information in modelling the spatial variation of continuous soil properties. Eur. J. Soil Sci., 46: 397-414. Granstedt, A., 1992. Case studies on the flow and supply of nitrogen in alternative farming in Sweden I Skilleby Farm 1981-87. Biol. Agric. Hortic., 9: 15-63. Greenwood, D.J. and Draycott, A., 1989. Quantitative relationships for growth and N content of different vegetable crops grown with and without ample fertilizer-N on the same soil. Fertilizer Res., 18: 175-188. Greenwood, D.J., Neeteson, J.J., and Draycott, A., 1986. Quantitative relationships for the dependence of growth rate of arable crops on their nitrogen content, dry weight and aerial environment. Plant Soil, 91: 281-301. Greenwood, D.J., Lemaire, G., Gosse, G., Cruz, P., Draycott, A. and Neeteson, J.J., 1990. Decline in percentage N of C3 and C4 crops with increasing plant mass. Ann. Bot., 74: 397407. Groffman, P.M., Tiedje, J.M., Robertson, G.P. and Christensen, S., 1988. Denitrification at different spatial and temporal scales: proximal and distal controls. In: J.R. Wilson (Editor), Advances in Nitrogen Cycling in Agricultural Ecosystems. Proceedings of a Symposium, Brisbane, Australia. CAB International, Wallingford, pp. 174-192.

213

Guiking, F.C.T., Jansen, D.M. and Fresco, L.O., 1994. The use of simplified nutrient balances at farm level to determine boundary conditions for sustainable production. In: J.K. Syers and D.L. Rimmer (Editors), Soil Science and Sustainable Land Management in the Tropics. Proceedings of a British Society of Soil Science Meeting, Newcastle, 1993. CAB International, Wallingford, pp. 248-257. Hansen, S., Jensen, H.E., Nielsen, N.E. and S vendsen, H., 1991. Simulation of the nitrogen dynamics and biomass production in winter wheat using the Danish simulation model DAISY. Fertilizer Res., 27: 245-259. Harmsen, G.W. and van Schreven, D.A., 1955. Mineralization of organic nitrogen in soil. Adv. Agron., 7: 299-398. Harper, J.E., 1994. Nitrogen metabolism. In: K.J. Boote, J.M. Bennett, T.R. Sinclair and G.M. Paulsen (Editors), Physiology and Determination of Crop Yield. ASA, CSSA, SSSA. Madison, WI, pp. 285-302. Hassink, J., 1992. Effects of soil texture and structure on carbon and nitrogen mineralization in grassland soils. Biol. Fertil. Soils, 14: 126-134. Ingestad, T. and ,~gren, G.I., 1992. Theories and methods on plant nutrition and growth. Physiol. Plant., 84: 177-184. Isermann, K., 1993. Territorial, continental and global aspects of C, N, P and S emissions from agricultural ecosystems. In: R. Wollast, F.T. Mackenzie and L. Chou (Editors), Interactions of C, N, P and S Biogeochemical Cycles and Global Change. SpringerVerlag Berlin, pp. 79-121. Janzen, H., Campbell, C., Brandt, S., Lafond, G. and TownleySmith, L. 1992. Light-fraction organic matter in soils from long term crop rotations. Soil Sci. Soc. Am. J., 56: 17991806. Jarvis, S.C. and Barraclough, D., 1991. Variation in mineral nitrogen content under grazed grassland swards. Plant Soil, 138: 177-188. Jarvis, S.C., Stockdale, E.A., Shepherd, M.A. and Powlson, D.S., 1996. Nitrogen mineralization in temperate agricultural soils: processes and measurement. Adv. Agron., 57: 188-237. Jenkinson, D.S. and Parry, L.C., 1989. The nitrogen cycle in the Broadbalk wheat experiment: a model for the turnover of nitrogen through the soil microbial biomass. Soil Biol. Biochem., 21: 535-541. Jenkinson, D.S., 1990. The turnover of organic carbon and nitrogen in soil. Philos. Trans. R. Soc., London Ser. B, 329: 361-368. Jensen, E.S., 1991. Nitrogen accumulation and residual effects of nitrogen catch crops. Acta Agric. Scand., 41: 333-344. Jones, C.A. and Kiniry, J.R. 1996. Ceres-Maize: A Simulation Model of Maize Growth and Development. Texas A and M University Press, College Station, TX. Justes, E., Mary, B., Beynard, J.-M., Machet, J.-M. and ThelierHuche, L., 1994. Determination of a critical nitrogen dilution curve for winter wheat crops. Ann. Bot., 74: 397-407. Karlovsky, J., 1981. Cycling of nutrients and their utilisation by plants in agricultural ecosystems. Agro-Ecosystems, 7: 127144. Keeney, D.R., 1982. Nitrogen Availability Indices. In: A.L. Page, R.H. Miller and D.R. Keeney (Editors), Methods of Soil Analysis. Part 2. ASA, Madison, WI, pp. 711-734.

Kinesch, P., Powlson, D.S. and Randall, E.W., 1995. t3C-NMR studies of SOM in whole soils. Part 2 A case study of some Rothamsted soils. Eur. J. Soil Sci., 46: 139-146. Kirkham, D. and Bartholomew, W.V., 1954. Equations for following nutrient transformations in soil, utilizing tracer data. Soil Sci. Soc. Am. proc., 19:189-192. Lanyon, L.E. and Beegle, D.B., 1989. The role of on farm nutrient balance assessments in an integrated approach to nutrient management. J. Soil Water Conserv., 44: 164-168. Leffelaar, P.A., 1990. On scale problems in modelling: an example from soil ecology. In: R. Rabbinge, J. Goudriaan, H. van Keulen, F.W.T. Penning deVries and H.H. van Laar (Editors), Theoretical Production Ecology: Reflections and Prospects. Simulation Monographs No. 34, Pudoc, Wageningen. Lord, E.I., 1992. Modelling of nitrate leaching: nitrate sensitive areas. Aspects Appl. Biol. (Nitrate and Farming Systems), 30: 19-28. Lutcher, L.K. and Mahler, R.L., 1988. Sources and timing of spring topdress nitrogen on winter wheat in Idaho. Agron. J., 80: 648654. McBratney, A.B., Webster, R., McLaren, R.G. and Spiers, R.B., 1982. Regional variation of extractable copper and cobalt in the topsoil of south-east Scotland. Agronomie, 2: 969-982. McCiure, P.R., Kochian, L.V., Spanswick, R.M. and Schaff, J.E., 1990. Evidence for co-transport of nitrate and protons in maize roots. I. Effects of nitrate on the membrane potential. Plant Physiol., 93:28 i-289. Magdoff, F.R., Ross, D. and Amadon, J. 1984. A soil test for nitrogen availability to com. Soil Sci. Soc. Am. J., 48: 13011304. Martinez, J. and Guiraud, G., 1990. A lysimeter study of the effects of a rye-grass catch crop, during a winter wheat/maize rotation, on nitrate leaching and the following crop. 3. Soil Sci., 41: 516. Meisinger, J.J. and Randall, G.W., 1991. Estimating nitrogen budgets for soil-crop systems. In: R.F. Follett, D.R. Keeney and R.M. Cruse (Editors), Managing Nitrogen for Groundwater Quality and Farm Profitability. SSSA, Madison WI, pp. 85124. Mengel, K., 1991. Available nitrogen in soils and its determination by the 'Nmin' method and by electroultrafiltration (EUF). Fertilizer Res., 28: 251-262. Neeteson, J.J. and W adman, W.P., 1987. Assessment of economically optimum application rates of fertilizer N on the basis of response curves. Fertilizer Res., 12: 37-52. Nguyen, M.L., Haynes, R.J. and Goh, K.M., 1995. Nutrient budgets and status in three pairs of conventional and alternative mixed cropping farms in Canterbury, New Zealand. Agric. Ecosystems Environ., 52: 149-162. Nitsch, A. and E. Vails, 1991. Nitrate estimates using the Nitracheck Test for precise N-fertilization during plant growth and, after harvest, for quality testing of potato tubers. Potato Res., 34: 95-105. Oades, J.M., Waters, A.G., Vassallo, A.M., Wilson, M.A. and Jones, G.P., 1988. Influence of management on the composition of organic matter in a red-brown earth as shown by ~3C nuclear magnetic resonance. Aust. J. Soil Res., 26: 289-299.

214

Parkin, T.B., 1993. Spatial variability of microbial processes in soil - a review. J. Environ. Qual., 22:409-417. Parton, W.J., Schimel, D.S., Cole, C.V. and Ojima, D.S., 1987. Analysis of factors controlling soil organic matter levels in Great Plains grasslands. Soil Sci. Soc. Am. J., 51: 11731179. Peltonen, J., Virtanen, A. and Haggren, E., 1995. Using a chlorophyll meter to optimise nitrogen fertiliser application for intensively managed small-grain cereals. J. Agron. Crop Sci., 174: 309- 318. Peng, S., Garcia, F.V., Laza, R.C. and Cassman, K.G., 1993. Adjustment for specific leaf weight improves chlorophyll meter's estimate of rice leaf nitrogen concentration. Agron. J., 85: 987-990. Post, W.M., King, A.W. and Wullschleger, S.D., 1996. Soil organic matter models and global estimates of soil carbon. In: D.S. Powlson, P. Smith and J.U. Smith (Editors), Evaluation of Soil Organic Matter Models using Existing Long-Term Datasets. Series 1 Global Environmental Change. NATO ASI Series, Voi. 38, pp. 201-224. Prosser, J., 1986. Nitrification. Special Publication of Society of General Microbiology. IRL Press, Oxford, 217 pp. Randall, E.W., Mahieu, N., Powlson, D.S. and Christensen, B.T., 1995. Fertilization effects on organic matter in physically fractionated soils as studied by ~3C-NMR: results from two long term field experiments. Eur. J. Soil Sci., 46: 557-565. Rijtema, P.E. and Kroes, J.G., 1991. Some results of nitrogen simulations with the model ANIMO. Fertilizer Res., 27: 189198. Roberts, G., 1987. Nitrogen inputs and outputs in a small agricultural catchment in the Eastern part of the United Kingdom. Soil Use Mgmt., 3: 148-154. Robertson, G.P., 1982. Regional nitrogen budgets: approaches and problems. Plant Soil, 67: 73-79. Royal Commission on Environmental Pollution, 1996. Sustainable Use of Soil, 19th Report. HMSO, London, 260 pp. Royal Society, 1983. The Nitrogen Cycle of the United Kingdom. Report of a Royal Society Study Group. The Royal Society, London, 576 pp. Russell, E.W. 1966. A History of Agricultural Science in Great Britain. Allen and Unwin, London, 493 pp. Shepherd, M.A., Stockdale, E.A., Powlson, D.S. and Jarvis, S.C., 1996. The influence of organic N mineralization on the management of agricultural systems in the UK. Soil Use Mgmt., 12: 7685. Skjemstad, J.O., Vallis, I. and Myers, R.J.K., 1988. Decomposition of soil organic nitrogen. In: J.R. Wilson (Editor), Advances in Nitrogen Cycling in Agricultural Ecosystems. Proceedings of a Symposium, Brisbane, Australia. CAB International, Wallingford, pp. 145-155. Smith, J.U., 1995. Models and scale: up- and down-scaling. In: A. Stein, F.W.T. Penning de Vries and P.J. Schotman (Editors), Models in Action. AB-DLO, Haren, pp. 25-41. Smith, J.U., Bradbury, N.J. and Addiscott, T.M., 1996. SUNDIAL: a PC based system for simulating nitrogen dynamics in arable land. Agron. J., 88: 38-43. Stafford, J.V., Ambler, B., Lark, R.M. and Catt, J., 1996. Mapping

and interpreting the yield variation in cereal crops. Comput. Electron. Agric., 14: 101-119. Stevenson, F.J., 1982. Organic forms of nitrogen. In: F.J. Stevenson (Editor), Nitrogen in Agricultural Soils. ASA, CSSA, SSSA, Madison, WI, pp. 67-122. Stinner, B.R., Crossley, Jr., D.A., Odum, E.P. and Todd, R.L., 1984. Nutrient budgets and internal cycling of N, P, K, Ca and Mg in conventional tillage, no-tillage and old-field ecosystems in the Georgia piedmont, Ecology, 65: 354-369. Stockle, C.O. and Nelson, R. 1996. CROPSYST. Biological Systems Engineering Department. Washington State University, Pullman, WA. Tamm, C.O., 1991. Nitrogen in Terrestrial Ecosystems. Questions of Productivity, Vegetational Changes and Ecosystem Stability. Ecological Studies Vol. 81, Springer-Vedag. Berlin. 115 pp. Tiedje, J.M., Simkins, S. and Groffman, P.M., 1989. Perspectives on measurement of denitrification in the field including recommended protocols for acetylene-based methods. Plant Soil, 115: 261-284. Titchen, N.M. and Scholefield, D., 1994. Tactical fertilizer application on commercial dairy farms. Institute of Grassland and Environmental Research Annual Report 1993, 72 pp. Tsay, Y.F., Schroeder, J.I., Feldman, K.A. and Crawford, N.M., 1993. The herbicide sensitivity gene CHL1 of Arabidopsis encodes a nitrate inducible nitrate transporter. Cell, 72: 705713. van Faassen, H.G. and Lebbink, G., 1994. Organic matter and nitrogen dynamics in conventional versus integrated arable farming. Agric. Ecosystems Environ., 51: 209-226. Vinten, A.J.A., Vivian, B.J. and Howard, R.S., 1992. The effect of nitrogen fertilizer on the nitrogen cycle of two upland arable soils of contrasting textures. Proc. Fert. Soc., 329: 1-48. Voltz, M. and Webster, R., 1990. A comparison of kriging, cubic splines and classification for predicting soil properties from sample information. J. Soil Sci., 41: 473-490. Vos, J., 1992. Growth and nitrate accumulation of catch crops. In: E. Francois, K Pithan and N. Bartiaux-Till (Editors), Nitrogen Cycling in Cool and Wet Regions of Europe - COST 814. Proceedings of a Workshop at Gembloux, Belgium. Commission of the European Communities, Brussels, pp. 103109. Vos, J., 1995. Nitrogen and the growth of potato crops. In: A.J. Haverkort and D.K.L. MacKerron (Editors), Potato Ecology and Modelling of Crops under Conditions Limiting Growth. Current Issues in Production Ecology, Vol. 3. Kluwer, Dordrecht, pp. 115-128. Vos, J., 1996. Nitrogen cycle related to crop production in cool and wet climates. In: R. Samuelsen, B. Solsheim, K. Pithan and E. Watten Melvaer (Editors), Nitrogen Cycling in Cool and Wet Regions of Europe - COST 814. Proceedings of a Workshop held at Tromso, Norway, 1995. European Commission, Brussels, pp. 3-14. Vos, J. and Born, M., 1993. Hand-held chlorophyll meter: a promising tool to assess the nitrogen status of potato foliage. Potato Res., 36: 301-308. Vos, J. and Marshall, B., 1994. Nitrogen and potato production: strategies to reduce nitrogen leaching. In: Proceedings of the

215

12th Triennial Conference of the European Association for Potato Research, Pads, France, 1993 (INRA, France), pp. 101-110. Wessel, W.W. and Tietema, A., 1992. Calculating gross N transformation rates of ~SNpool dilution experiments with acid forest litter: analytical and numerical approaches. Soil Biol. Biochem., 24:931-942. de Willigen, P., 1991. Nitrogen turnover in the soil-crop system; comparison of fourteen models. Fertilizer Res., 27: 141149. Willison, T.W., Tlustos, P., Murphy, D.V., Baker, J.C., Goulding,

K.W.T. and Powlson, D.S., 1997. Short-term effects of nitrogen on methane oxidation in soils. Eur. J. Soil Sci., in press. Wilson, M.A., 1987. NMR Techniques and Applications in Geochemistry and Soil Chemistry. Pergamon Press, Oxford. de Wit, C.T., 1992. Resource use efficiency in agriculture. Agric. Systems, 40:125-15 I. Young, M.W., Davies, H.V. and MacKerron, D.K.L 1993. Comparison of techniques for nitrogen analysis in potato crops. In: M.A.C. Fragoso and Beusichem M.L. van (Editors), Optimization of Plant Nutrition. Developments in Plant and Soil Sciences, Vol. 53. Kluwer, Dordrecht, pp. 7-11.

This Page Intentionally Left Blank

t~ 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geij'n (Editors)

217

Modeling crop nitrogen requirements" a critical analysis C.O. Stockle a'*, P. Debaeke b aBiological Systems Engineering Department, Washington State University, Pullman, WA 99164-6120, USA blNRA, Station d'Agronomie, BP 27, 31326 Castanet Tolosan cedex, France Accepted 2 June 1997

Abstract Four approaches to simulate N requirements of wheat were testedusing data collected at the Auzeville experiment station of INRA, near Toulouse, France, from an experiment providing a wide range of soil N available for crop uptake. In these approaches, crop N requirements are expressed in terms of characteristic plant N concentration curves (maximum, critical, and minimum), representing expected concentration for a given crop N status throughout the growing season. Modeling approaches were evaluated for their ability to discriminate between N-limited and non-limited wheat plots as well as to properly represent the upper and lower limits of observed plant N concentrations. Best results were obtained using the growth dilution concept to represent the characteristic curves, while others based on temperature sums, growth stages, or fraction of the growth cycle were less satisfactory. Simulation of crop growth and N uptake based on N requirements estimated using the growth dilution concept resulted in a relationship between biomass at harvest and N uptake that correctly described an upper boundary for all observed data points. However, simulated and observed crop N uptake on a plot by plot basis resulted in low agreement. This was attributed to uncertainty in the measurement of initial soil N and crop N uptake, and the effect of other growth reducing factors (e.g. diseases) and possibly physical and/or chemical restrictions to field N uptake normally not accounted for by crop growth models. © 1997 Elsevier Science B.V. Keywords: Growth; Model; Simulation; Wheat

I. Introduction Simulation models are increasingly used for the assessment of crop productivity and the impact on the environment that may result from given combinations of weather, soil, crop characteristics, and water and N management. For this purpose, the proper simulation of crop N requirements is important, both in terms of amount and distribution throughout the growing season. Several approaches have been proposed to simulate * Corresponding author. Tel.: +l 509 3353564; fax: +l 509 3352722; e-mail: [email protected]

crop N requirements. We selected four that are representative and able to model wheat N requirements, corresponding to those included in the following crop growth models: AFRCWHEAT2 (Porter, 1993), Daisy (Hansen et al., 1991), EPIC (Williams et al., 1989), and CropSyst (Stockle and Nelson, 1996). In these approaches, crop N requirements are expressed in terms of characteristic plant N concentration curves, which represent the expected concentration for a given crop N status throughout the growing season. The most complete approaches define three such curves: a maximum (Nmax), a critical (N¢,t), and a minimum (Nmi,) plant N concentration. Plant growth is not limited by N if plant concentration

Reprinted from the European Journal of Agronomy 7 (1997) 161-169

218

is at or above Ncrit, while Nmax establishes the maximum crop N uptake. Below Ncnt, plant growth is reduced, stopping completely when plant N concentration reaches Nmin. It should be noticed that these definitions do not consider the quality of the harvested crop. Plant N concentration is not constant but decreases with time, and so do the three characteristic curves. To describe this process, models express these curves as a function of crop growth stage (AFRCWHEAT2), the fraction of the growth cycle (EPIC), or thermal time (Daisy). On the other hand, research has shown that Nc~it decreases with increasing shoot biomass according to an allometric equation (Salette and Lemaire, 1981; Greenwood et al., 1990), usually referred to as the growth dilution law. This concept has been tested with field data and shown able to discriminate between well-supplied and N-deficient crops (e.g. Justes et al., 1994; P1Enet, 1995). Furthermore, single allometric equations for C3 and Ca crops have been

6 5 ~4

proposed (Greenwood et al., 1990). Similar functional relationships with biomass may be used for Nmax and Nmi, (Justes et al., 1994). An implementation based on this concept was introduced to the CropSyst model. It must be noted, however, that adequate simulation of crop N requirements may not guarantee the ability of models to simulate crop growth in response to soil available N and N management, which is the ultimate objective in the application of these models. In most models, N uptake depends on crop N requirements (as needed to maintain Nmax) but also on the attainable N uptake as given by N concentration, moisture, and root distribution in the soil profile (Jones and Kiniry, 1986; Williams et al., 1989; Hansen et al., 1991; Porter, 1993; Stockle et al., 1994). The attainable N uptake may or may not satisfy crop N requirements. The objectives of this study were (1) to test four approaches to model crop N requirements using data collected for winter wheat at Auzeville, southern France, during the growing season of 1993, and (2) 6

k•

AFRCWHEAT2

5

"'. o

~

EPIC

~4

Z 3

o°°°

i,,1

Z3

~

.............

1

o 0

,

v

,

,

20

40

60

80

o 100

0.2

o.o

Growth stages

0.4

0.6

0.8

1.0

Fraction of the cycle

DAISY MODEL

5

~4

CROPSYST

~-~--~

~4

;,.

z 3 N

z3

f= " r ' ~ O

--~2

"

0

500

•

n •

•°-.j~.

o

1000

1500

Degree Days

2000

2500

0

•

oDD o

o

....

3

......

6

9

12

Aboveground biomass (Mg/ha)

Fig. 1. Comparisonof four models for estimating characteristic plant N concentration curves with data from N-limited and non-limitedplots of winter wheat cultivar Soisson grown at Auzeville, France in 1992-1993. (Symbols: open square, N limited; closed square, N non-limited; Lines: dashed, Nman;solid, Nc,it;dotted, Nmin).

219

compare observed and simulated biomass production and crop N uptake associated with crop N requirements, estimated using the best of the four approaches from objective 1, under a wide range of available nitrogen.

curves are defined. A generic exponential function of the fraction of the growing season (Fgs) is used to calculate Nc~it. The value of Fgs is determined as the ratio of the current temperature sum to the total temperature sum at maturity. For wheat, the function is Ncrit - 100 (0.01 + 0.05 exp(-2.67 Fgs)). Nmi, is calculated as half of No,it.

2. Approaches to model crop N requirements 2.4. CropSyst A description of four approaches to determine the characteristic concentration curves used to model crop N requirements throughout a growing season is given below. Some of these include a separate set of curves for shoots and roots. For this study, only shoot concentrations are of interest.

2.1. AFRCWHEA~ In this wheat model, Nmax and Nmin curves are defined. Nmax is set at 4.5% from emergence to initiation of the terminal spikelet, from which point it falls to a value of 0.5% by the end of grain filling. Nmin starts at 2.5%, increases slightly until the double ridge stage, and then falls to 0.25% by the end of grain filling. The details of the change of Nmax and Nmin as a function of developmental stage are given graphically by Porter (1993), and are reproduced here in Fig. 1 using the phenological scale proposed by Zadoks et al. (1974).

2.2. Daisy In this wheat model, Nmax, Ncrit , and Nmi, curves are defined. Characteristic N concentrations are given as a function of air temperature sum (base 0°C) from emergence. All curves have a constant concentration up to a temperature sum of 100°C - days, with values of 5%, 3%, and 2% for Nmax, Nc~it, and Nmin, respectively. After a temperature sum of 1100°C- days, these values are constant at values of 1.2%, 1.0%, and 0.7%, respectively. All concentrations decay exponentially for temperature sums between these two boundaries. These three curves are given graphically by Hansen et al. (1991), and reproduced in Fig. 1.

2.3. EPIC In this genetic crop growth model, Neat and

Nmi n

Nmax, Ncrit, and Nmin curves are defined in this genetic crop growth model. Plant N concentration (N%), in percent, is assumed to be related with biomass accumulation (B) as follows: N% = a B -b, where a and b are fitted parameters (Salette and Lemaire, 1981; Greenwood et al., 1990; Justes et al., 1994; Pl6net, 1995). The value of Nmax during early growth is required as input parameter. Then, the characteristic N concentration curves are given by: Nmax= min (Nmax, amax B-°'35), Ncrit = min (0.7 Nmax, acrit B-°'35), and Nmin = rain (0.4 Nmax, aminB-0"35), where B is biomass for an unstressed crop (t/ha), amax= Nmax/ (2-0"35), acrit= 0.7 Nmax/(1.5-°'35), and ami n "" 0.4 Nmax/(O.5-°'35). This implementation, based on data from the references given above, is a generalization that works well for both C3 and C4 species. For wheat, a value of 5.0% was used for Nmaxduring early growth (see Fig. 1). The growth dilution law seems to work well up to flowering (Justes et al., 1994). After this point, the CropSyst implementation reduces linearly the three characteristic curves so as to meet specified (input parameters) values of Nmax, Ncrit, and Nmi, at maturity.

3. Methods Experimental data collected for winter wheat during 1993 growing season at the INRA station in Auzeville, southwestern France, were used to compare the four modeling approaches (Debaeke et al., 1996). The soil was a deep silty-clay loam, with an organic matter content of 1.6% (0-30 cm). The cultivar Soissons was grown in 16 plots (8 preceding crops x2 input levels). Preceding crops were faba bean, maize, pea, rapeseed, sorghum, soybean, sunflower, and wheat. Input levels differed by sowing date (high-input sown on 30 November 1992 and low-input on 16 December

220

1992) and N fertilization. N fertilizer rate was calculated using the French balance sheet method (R6my, 1981), adapted for southern France, using yield goals of 5 (low-input, N1) and 8 t/ha (high-input, N3). Briefly, this method calculates the N fertilizer dose as a function of soil availability and crop requirements for a yield objective, and includes a correction factor to increase the N dose when a limitation to N uptake by soil structure or soil moisture is expected. No supplementary irrigation was required in 1993. Each wheat plot was divided into two sections: (1) a central area (410 m2), receiving N fertilizer according to the yield objective (Nx treatment); (2) an unfertilized lateral area (60 m2), which was kept free of N fertilizer to assess the soil N contribution (NO treatment), yielding a total of 32 treatment combinations. Mineral N was determined before N fertilizer supply, in December 1992, from 5 soil cores per plot taken to a depth of 1.20 m in 0.2-m increments. Above-ground dry matter and N concentration were measured at tillering, stem elongation, shooting, anthesis and maturity. At each sampling date, the plants in five 0.25-m 2 quadrats were collected for biomass and N determination. At maturity, dry matter and N concentration were measured separately for grain, culm and chaff using the Kjeldahl method. Growth stages were monitored regularly and calculations of thermal time were done from emergence. Data were analyzed to discriminate between N-limited and non-limited plots using the method described by Justes et al. (1994). As a clarification, this method of discrimination is not based on the growth dilution model for Ncrit given above, and its use does not affect the predictive ability of this model. The Ncrit curves calculated using the four modeling approaches introduced above were evaluated for their ability to separate N-limited and non-limited plots throughout the growing season. Also Nmax and Nmi, curves for each approach were evaluated for their ability to define the upper and lower limits of observed plant N concentrations. To examine the biomass production and crop N uptake associated with crop N requirements, crop growth simulations using the best approach for estimating crop N requirements (objective 1) were performed for the 32 treatment combinations and compared to the experimental observations. Modeling of N uptake and associated crop growth is concep-

tually similar in most crop growth models, including the four introduced above. The implementation in the CropSyst model was used for this evaluation. Details of concepts and equations used to model crop growth as affected by water and nitrogen availability are given elsewhere (Stockle et al., 1994; Stockle and Nelson, 1996). Table 1 shows the crop input parameters used to implement simulations of winter wheat growth. These parameters are typical for the wheat cultivar Soissons grown in Auzeville. Phenology was adjusted as observed in the experimental plots. A few calibrated crop parameters (Table 1) were set based on plots not included in the evaluation reported herein.

4. Results and discussion

4.1. Evaluation of four approaches to model crop nitrogen requirements Fig. 1 compares the performance of the four approaches evaluated. The lower curve (Nmin) in the Table 1 Summary of crop parameters for CropSyst simulations Parameters Degree - days emergence (°C - days) D e g r e e - days begin flowering ( ° C - days) Degree - days peak LAI (°C - days) D e g r e e - days begin grain filling (°C - days) Degree - days maturity (°C - days) Base temperature (°C) Cutoff temperature (°C) Maximum root depth (m) Maximum LAI Specific leaf area (mE/kg) Stem/leaf partition coefficient Leaf duration (°C - days) Solar radiation extinction coefficient ET crop coefficient Maximum water uptake rate (mm/day) Critical canopy water potential (kPa) Wilting canopy water potential (kPa) Biomass-transpiration coefficient (Pa) Radiation-use efficiency (g/MJ) Maximum harvest index, HI

Obs Obs Obs Obs Obs Man Man Obs Obs Obs Obs Obs Man Cal Man Man Man Cal Man Obs

150 1600 1550 1780 2350 0 25 1.5 7 22 2.2 1200 0.46 1.2 10 -1300 -2000 4.5 3 0.45

Parameters were set as observed experimentally (Obs), extracted from the CropSyst manual (Man), or set by calibration (Cal).

221

AFRCWHEAT2 model effectively includes the observed minimum concentrations, with a few data points lying below the curve. However, the upper limit curve (Nmax) has problems up to stage 35 (midshooting), and again at about stage 90 (maturity), with many data points (even for N-limited conditions) above the Nmaxcurve. This will result in underprediction of the N requirement of wheat, thus affecting the simulation of crop growth in relation to available N. The performance of the approach in the EPIC model also presents problems. The Nm~,curve appears adequate, but the Ncnt curve has problems separating N-limited from non-limited data points, particularly before a fraction of the cycle of 0.4. Because crop N stress is simulated once plant N concentrations are below Ncr~t,this will lead to misrepresentation of stress during early growth (before completion of 40% of the growing cycle), a critical period of wheat growth and development. In addition, because a curve for Nmaxis not specified, crop N requirements are underpredicted, which may lead to improper simulation of crop growth in relation to available N, and will misrepresent the potential for N extraction. The implementation in the Daisy model, which includes three characteristic curves, has the worst performance of all approaches tested. Throughout most of the growing season, described in terms of temperature sums, most data points from plots not limited by N and a significant amount of N-limited data points are above the Nmax curve. Under these conditions, wheat N requirements will be severely underpredicted. The Ncnt curve does not discriminate between N-limited and non-limited data points. The Nmincurve, however, represents well the minimum concentrations observed. Given this performance, this approach should have significant limitations in predicting crop N requirements and crop response to N availability for these Auzeville data. The method based on growth dilution (CropSyst) is able to describe well the three characteristic curves that define crop N requirements and crop response to N availability. The Nerit curve discriminated well between N-limited and non-limited data points. The Nmax and Nmin curves effectively served as envelope curves to define the maximum and minimum limits for most observed data points with few exceptions. It seems that methods based on the growth dilution concept (Greenwood et al., 1990) should be preferred to

simulate crop N requirements. This approach has also the advantage of being easier to implement across different crop species and cultivars than the crop specific relations of some of the other models, and it is likely to be more transferable among growth environments.

4.2. Simulating nitrogen uptake and crop growth in relation to crop N requirements After establishing the good performance of the growth dilution concept for the simulation of crop N requirements, an analysis of the nitrogen uptake and crop growth associated with such requirements was performed. Fig. 2 shows a comparison of measured (symbols) and simulated (lines) evolution of biomass and plant N concentration for selected plots. Plot labels are composed of a plot identifier (first two characters) and a N treatment identifier (last two characters). As introduced above, NO and N3 correspond to no- and high-nitrogen treatments, respectively. For plots F8-N3 and F8-N0, following pea, there was an excellent agreement between observed and simulated values (good soil structure, large amount of initial N). The same was the case for F6-N3, after early-harvested sorghum. For plots F6-N0 and C5-N3, the agreement was good for plant N concentration, and also for biomass, except for the last (C5N3) or two last (F6-N0) observations. In the case of C5-N3, the plant N concentration was overpredicted for the measurement previous to the last, but the agreement between simulated and observed biomass was good. This resulted in a slight overprediction of simulated N uptake and a final simulated biomass larger than observed. A similar situation was found for plot F6-NO, somewhat enhanced by simultaneous overprediction of biomass and plant N concentration before the last observation. For plot C l-N0, although measured and simulated plant N concentrations agreed well, simulated crop N uptake was insufficient to support the observed biomass production. Fig. 3 shows the comparison between observed and simulated aboveground N uptake for all the treatment combinations. The linear regression between simulated and observed N uptake has an interception of zero and slope of 0.98, very close to the 1:1 line of perfect agreement, but with a weak correlation

222 coefficient (r = 0.822). The agreement between simulated and observed N uptake is low for reliable model applications. One source of uncertainty is the varia-

bility of the five samples used to determine field N uptake. The average coefficient of variation was 15.4%, with a minimum of 5.3% and a maximum of

25 20

6

20

10 gt~

•i

s 0

Plot FS-N3 0

,

,

,

,

|

0

50

100

150

200

250

25

25

20

20

,

,

!

!

i

0

50

100

150

200

,0 250

"'

6

ot F6-N0 ,lw

3~

2Z Plot F6-N3 ,

,

,

,

!

0

50

100

150

200

i

0

ZS0

!

|

0

50

-

|

|

,

100

150

200

'

6

25 20

O

250

5~

20

Plot C1-N0

0 IN

,ha

Plot C5-N3 O

.m

3e,

•

.

!

$

O 2

-

Z m

0 0

50

100

150

200

Days after emergence

250

!

!

i

i

,

0

50

100

150

200

0

250

Days after emergence

Fig. 2. Comparison of observed (symbols) and simulated (lines) evolution of biomass and plant N concentration for selected plots (triangle, plant N concentration; circle, biomass; error bars represent + 1SD). The first two characters in the plot labels correspond to plot identification, and the last two characters to N treatment (see text).

223

300

~

250

0

•

•

o 200 .,~

15o Z

00

•-

50

I

I

I

I

I

I

0

50

100

150

200

250

300

Observed N U p t a k e (kg/ha) Fig. 3. Comparison between observed and simulated N uptake for 32 N treatment combinations on winter wheat cultivar Soisson grown at Auzeville, France in 1992-1993.

29.1%. Plotting the 32 means versus all the samples resulted in a linear regression explaining 88.5% of the observed variability. Other sources of variability exist, as discussed below. Fig. 4 compares observed and simulated N uptake as a function of soil N availability. Available soil N was determined as the sum of the initial soil N to a depth of 120 cm, plus fertilization and mineralization during the growing season. Initial soil N fluctuated from 50 to 215 kg/ha, and fertilization from 0 to 200 kg/ha. Soil N mineralization was estimated as 40 kg/ha from simulations, a magnitude in agreement with the N balance sheet method introduced above. Thus, a wide range of available soil N is represented in the 32 treatment combinations included in this study, fluctuating from 90 to 370 kg/ha. For this wide range, the simulated N uptake is strongly linearly correlated with soil N availability (r - 0.998), as expected for a system where no constraints to uptake and growth other than available soil N are assumed (soil moisture and rooting depth were not limiting factors). For the measured N uptake, the correlation is weaker (r - 0.84), with large variability around the regression line. For example, for a soil N availability of about 325 kg/ha, observed N uptake fluctuated from 146 to 257 kg/ha.

Other agronomic factors may have affected crop growth in some of the experimental plots (e.g., rust infection under low-input management) reducing N uptake in relation to N available. Sampling size (5 cores/470-m 2 plots) to determine initial soil N content, a quantity with significant spatial variability, may have been insufficient to obtain adequate measurements. The differences in uptake for similar N availability levels, however, appear high to be explained by this kind of reasoning only. It is possible that physical and/or chemical restrictions to N uptake in the field, other than those normally accounted for by crop growth models, may have also played a role. For example, soil structure could have been affected by wet soil conditions during harvest of some of the preceding crops as well as during sowing of the winter wheat. Fig. 5 shows observed and simulated biomass as a function of the corresponding observed or simulated aboveground N uptake. Simulated data points describe a N uptake/biomass production function that is an upper envelop for the observed data points, the latter showing large variability. This indicates that the characteristic plant N concentration curves were effective in defining crop N requirements and limitations to growth associated with plant N concentration. 31111

m

o o

~2500

~

200-

oo Z

o

1500

0 0

I

o o 0

o O

O~o ~

o

o

~

i~i tt

O~'

100-

0

0

II

~

o

~

"

0

50-

0

•

'

so

o

I"

i

I

I

100

1so

200

2so

.... I

300

I

3so

400

Observed Soil N Availability (kg/ha) Fig. 4. Observed (open circles) and simulated (closed circles) aboveground N uptake as a function of the observed soil N availability (initial soil N + N fertilization + net mineralization).

224

20000-

0 ~IIP~ e 0

• ¢I1

o • o @ d3

o~o~

o

o

OOo o

12000

o~ o 0 •~

i~0 o

8000

4000

0

I

I

I

I

I

50

100

150

200

250

Aboveground

300

N Uptake (kg/ha)

Fig. 5. Observed (open circles) and simulated (closed circles) biomass at harvest as a function of corresponding observed and simulated aboveground N uptake.

Data points falling below the simulated envelop must have resulted from growth reducing factors not accounted for by the model, which would also explain the differences between simulation and measurements shown in Fig. 2. No pattern of relationship between these data points and preceding crop or N input level was found.

5. C o n c l u s i o n s

The use of the growth dilution concept provided a good base to estimate characteristic plant N concentration curves (maximum, critical, and minimum concentrations) throughout the growth cycle. Using the implementation of this concept in the CropSyst model, the simulated critical curve discriminated well between N-limited and non-limited experimental wheat plots growing under a wide range of soil N availability. The simulated maximum and minimum curves effectively served as envelope to define the upper and lower N concentration limits for most observed data. Other approaches, based on empirical relationships with the fraction of the growth cycle, growth stages, or temperature sums, were less satisfactory.

The simulated relationship between biomass production and N uptake correctly described an upper boundary for all observed data points. Points falling below the curve represented limitation to growth other than nitrogen. Despite adequate simulation of crop N requirements, simulated and observed crop N uptake on a plot by plot basis resulted in low agreement. Simulated crop N uptake was highly correlated with soil N availability, while the correlation for the observed data was weaker. Uncertainty in the measurement of initial soil N and crop N uptake, other growth reducing factors (e.g., diseases) not accounted for by the model, and possibly physical and/or chemical restrictions to field N uptake not normally included in crop growth models may have limited the ability to simulate crop growth in response to N availability.

References

Debaeke, P., Aussenac, T., Fabre, J.L., Hilaire, A., Pujoi, B. and Thuries, L. 1996. Grain nitrogen content of winter bread wheat (Triticum aestivum L.) as related to crop managementand to the previous crop. Eur. J. Agron., 5: 273-286. Greenwood, D.J., Lemaire, G., Gosse, G., Cruz, P., Draycott, A. and Neeteson, J.J. 1990. Decline in percentage N of C3 and C4 crops with increasing plant mass. Ann. Bot., 66: 425-436. Hansen, S., Jensen, H.E., Nielsen, N.E. and Svendsen, H. 1991. Simulation of nitrogen dynamics and biomass production in winter wheat using the Danish simulation model Daisy. Fert. Res., 27: 245-259. Jones, C.A. and Kiniry, J.R. 1986. CERES-Maize, a Simulation Model of Maize Growth and Development. Texas A and M University Press, Texas, TX, 193 pp. Justes, E., Mary, B., Meynard, J.M., Machet, J.M. and ThelierHuch6, L. 1994. Determination of a critical nitrogen dilution curve for winter wheat crops. Ann. Bot., 74: 397-407. Pl6net, D. 1995. Fonctionnement des cultures de mai's sous contrainte azot6e. Doctoral Thesis, Acad6mie de Nancy-Metz, lnstitut National Polytechnique de Lorraine, France, 247 pp. Porter, J.R. 1993. AFRCWHEAT2: a model of the growth and development of wheat incorporating responses to water and nitrogen. Eur. J. Agron., 2: 69-82. R6my, J.C. 1981. Etat actuel et perspectives de la mise en oeuvre des techniques de pr6vision de la fumure azot6e. C. R. Acad. Agric. Fr., 67: 859-874. Salette, J. and Lemaire, G. 1981. Sur la variation de ia teneur en azote des gramin6es fourrag~respendant leur croissance: formulation d'une loi de dilution. C. R. Acad. Sci. Paris Ser. III, 292: 875-878. Stockle, C.O., Martin, S. and Campbell, G.S. 1994. CropSyst, a

225

cropping systems model: water/nitrogen budgets and crop yield. Agric. Syst., 46: 335-359. Stockle, C.O. and Nelson, R. 1996. CropSyst User's Manual. Biological Systems Engineering Department, Washington State University, Pullman, WA, 186 pp. Williams, J.R., Jones, C.A., Kiniry, J.R. and Spanel, D.A.

1989. The EPIC crop growth model. Trans. ASAE, 32: 497511. Zadoks, J.C., Chang, T.T. and Konzak, T.T. 1974. A decimal code for the growth stages of cereals. Weed Res., 14: 415421.

This Page Intentionally Left Blank

© 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

227

Maize production in a grass mulch system - seasonal patterns of indicators of the nitrogen status of maize B. Feil a'*, S.V. Garibay b, H.U. Ammon b, P. Stamp a alnstitute of Plant Sciences, ETHZ, Universitdltstrasse2, CH-8092 Zurich, Switzerland bEidgenOssische Forschungsanstaltfiir Agrari~kologie und Landbau, Reckenholzstrasse 161, CH-8046 Ziirich, Switzerland Accepted 13 April 1997

Abstract In hilly regions with high precipitation, planting maize (Zea mays L.) into living mulches of winter cover crops may alleviate some of the problems (erosion, runoff of agrochemicals, nitrate leaching) associated with conventional maize cropping. Nitrogen was found to be more yield-limiting for maize planted into a grass sod than for traditionally grown maize. The aim of the present study was to follow the seasonal patterns of some indicators of the N status (whole-shoot concentrations of N and nitrate as well as leaf greenness, as measured by the SPAD 502 chlorophyll meter) of silage maize in three cropping systems. The experiments were conducted in the midlands of Switzerland, where annual precipitation is high (>1000 mm), for 3 years. The cropping systems were PLOUGH (maize sown into the bare autumn-ploughed soil), GRASS/ HERB (maize planted into a living, subsequently herbicidally killed Italian ryegrass [Lolium multiflorum Lam.] sod), and GRASS/MECH (similar to GRASS/HERB, but the growth of the ryegrass sod was mechanically regulated). Planting was done with a strip tillage seeder. There were two N treatments: 110 and 250 kg N/ha (inclusive of the mineral N content of the soil from 0 to 90 cm depth). With 110 kg N/ha, the following differences between the cultural systems were found: (i) as early as the 3rd leaf stage, PLOUGH maize exhibited higher whole-shoot concentrations of nitrate than sod-planted maize; (ii) between the 3rd and 6th leaf stages, chlorophyll meter measurements revealed that the uppermost unfolded leaves of PLOUGH maize were markedly greener than those of maize on the mulch seeding plots; (iii) from the 6th leaf stage onwards, PLOUGH maize showed a higher N concentration than did maize grown in grass swards. Under 250 kg N/ha, the differences between the cultivation methods were much less pronounced. It is concluded that efforts to optimize the environmentally friendly living mulch systems should focus on reducing the competition between maize and the cover crop for N. © 1997 Elsevier Science B.V.

Keywords: Living mulch; Maize; N status; SPAD 502; Strip tillage

1. Introduction The Swiss midlands are hilly and the annual precipitation is high (>1000 mm). Most of the area devoted to maize production is moldboard ploughed *Corresponding author. Tel.: +41 1 6324746; fax: +41 1 6321143; e-mail: [email protected]

in autumn; the fields remain unplanted in winter, and the maize is sown into a finely pulverized seedbed. Sloping cropland treated in this manner is susceptible to erosion. Further environmental problems associated with conventional maize cropping practices are surface runoff of herbicides (Rtlttimann et al., 1995) and plant nutrients (Prasuhn and Braun, 1994) and nitrate leaching into the ground water.

Reprinted from the European Journal of Agronomy 7 (1997) 171-179

228

Research conducted at various locations in the Swiss midlands and in the Jura range demonstrated that sowing maize in winter cover crop residues, killed by frost or herbicides, in conjunction with minimum tillage is a very effective means of controlling soil erosion and runoff of agrochemicals (Rtittimann et al., 1995). Under these cropping systems, however, maize produced lower silage yields than under plough tillage (Rtiegg et al., 1997). Another approach to alleviate the environmental problems associated with conventional maize production is to plant the maize into a living winter cover crop sward (Box et al., 1980; Echtenkamp and Moomaw, 1989; Klocke et al., 1989). A number of advantages of maize production in living mulches of Italian ryegrass have been reported. In contrast to non-winterhardy cover crops, Italian ryegrass provides a considerable hay or haylage yield in spring (Garibay et al., 1997). After the harvest of maize, the cover crop sward can be used for autumn pasture. It was found that mechanically stunted grass strips between the maize rows harbored many predatory insects and spiders; the maize was markedly less infested by maize smut (Ustilago maydis), aphids (Rhopalosiphon maidis), and European corn borers (Ostrinia nubilalis) (Bigler et al., 1995a,b). J~iggi et al. (1995) observed that the earth worm (Lumbricus terrestris) biomass was significantly higher than under plough tillage. Planting maize into living mulches of Italian ryegrass reduces the need for herbicides and may help to prevent the development of herbicide-resistant weed populations (Ammon et al., 1995). After the maize harvest, the grass strips regrow and remove mineral N from the soil, thus reducing the hazards of soil erosion and nitrate leaching during the cool season. Garibay et al. (1997) reported that maize planted into living mulches of Italian ryegrass has high N requirements to reach maximum yield, thus suggesting that cover crop and main crop compete for N. It is unclear when differences in N status between sodplanted maize and maize grown conventionally occur. Tissue concentrations of N (Cerrato and Blackmer, 1991; Binford et al., 1992) and nitrate (Iversen et al., 1985; McClenahan and Killorn, 1988) as well as leaf chlorophyll contents (Dwyer et al., 1991; Blackmer and Schepers, 1994) have been proposed as indicators of the N status of maize. To determine the periods during which differences in N status between

sod-planted and plough tilled maize are detectable, we monitored the seasonal patterns of the concentrations of N and nitrate in the tops and the chlorophyll contents of the uppermost fully expanded leaves.

2. Materials and methods

2.1. Experimental site and climatic conditions Field experiments were conducted in 1990/91, 1991/92, and 1992/93 on a farm near Zurich (47030 ' N, 8030 ' E, 424 m asl). The soil was an Eutric Cambisol (FAO classification) with 480 g/kg sand, 360 g/ kg silt, 160 g/kg clay, and 21 g/kg organic matter in the top soil (0-30 cm). Cropping system precipitation (precipitation during the 14 month period from August to September) was 1004 mm, 1990/91; 779 mm, 1991/92; and 1099 mm, 1992/93. During the growth cycle of maize, precipitation totaled 268 mm, 1991; 303 mm, 1992; and 489 mm, 1993; the rainfall distribution during the maize growing season in the various years is given in Table 1. The mean daily temperature during the growth cycle of maize ranged from 16.6°C in 1993 to 18.0°C in 1991. The levels of soil P and K (Garibay et al., 1997) were sufficient for optimum plant growth. The experimental fields had been planted previously with winter wheat in 1990 and 1992 or winter rye in 1991.

2.2. The maize cropping systems The cover crop stands (Italian ryegrass) were established in the late summer of the year preceding the planting of maize. Three maize cropping systems Table 1 Monthly rainfall (mm) during the maize growing seasons and 40 year average Year Month

1991

May 101 June 174 July 56 August 17 September 68

1992

1993

40 year average

22 112 86 74 41

80 89 182 92 89

90 113 108 120 80

229

were 75 cm apart. In the GRASS/HERB and GRASS/ MECH systems, only the tilled strips were sprayed with 9 l/ha Primafit A ® (190 ml/l metolachlor plus 95 ml/l atrazine plus 95 ml/l pendimethalin) for weed control. In contrast, in the PLOUGH system, the entire area was treated with 9 l/ha Primafit A ®. The grass strips between the maize rows in the GRASS/HERB system were killed by applying 30 g/ha of the herbicide Titus ® (25 g/kg rimsulfuron) at the 1st and 2nd leaf stages of maize. Grass growth on the GRASS/MECH plots was suppressed with a one row 80 kg heavy mulching machine at the 1st, 3rd, and 6th leaf stages of the maize crop. More information on the cropping systems is given in Garibay et al. (1997).

were compared: (i) PLOUGH (the ryegrass stands were broken up with a moldboard plough in autumn/ winter; the maize was sown into the bare soil), (ii) GRASS/HERB (the maize was planted into a living ryegrass sod; the grass strips between the maize rows were killed by post-emergence applications of a herbicide), and (iii) GRASS/MECH (the maize was sown into a living ryegrass sod whose growth was suppressed mechanically). Planting was done with a one-pass minimum strip tillage seeder which operated at a depth of 15 cm and tilled 30 cm wide bands of soil. 2.3. Cover crop and maize management

Italian ryegrass (Lolium multiflorum Lam. cv. Lipo) was drilled at a rate of 20 kg seed/ha in August on the entire experimental area. Thirty kilograms N/ha (ammonium nitrate) was broadcast approximately 1 month after sowing. The ryegrass was clipped and removed from the plots in October. One-third of the plots was moldboard ploughed in autumn/winter. The grass on the remaining plots was clipped at ground level and removed a second time in spring, shortly before sowing of maize. Maize (Zea mays L. cv. Atlet; FAO 250) was planted at 100000 seeds/ha in May. The final plant densities were 9.6 (PLOUGH; GRASS/HERB), and 9.4 (GRASS/MECH) plants/m 2. The seeding depth was about 5 cm. The maize rows

2.4. Fertilization

There were two N treatments (referred to subsequently as N110 and N250): 110 kg N/ha (mineral N content of the soil plus fertilizer N applied at sowing) and 250 kg N/ha (N110 plus fertilizer N applied at the 4th and 6th leaf stages of maize) (Table 2). Rates of 40 kg (1991), 26 kg (1992), and 41 kg (1993) N/ha were applied as ammonium nitrate to all plots with the planter; the fertilizer was mixed with the soil and evenly distributed in the tilled strips by the rotary hoe. The remainder of the N was applied by hand immediately after planting; the N fertilizer

Table 2 Mineral N content of the soil (0-90 cm depth) shortly before maize planting, and rates and timings of N application PLOUGH i 991

NIIO Maize planting c N250 Maize planting c 4th leaf stage d 6th leaf stage d

GRASS a 1992

1993

1991

1992

1993

Soil mineral N (kg N/ha) b 70 84 Fertilization (kg N/ha)

69

23

23

I1

40

26

41

87

87

99

40 70 70

26 70 70

41 70 70

87 70 70

87 70 70

99 70 70

aGRASS/MECH and GRASS/HERB. bMeasured according to Wehrmann and Scharpf (1979), modified by Walther (1983). CUnder PLOUGH, the N fertilizer was incorporated in the tilled strips; the GRASS plots were treated in the same manner, but the remaining N fertilizer was placed right in the maize rows. dplaced right in the maize rows.

230

was placed right in the maize rows and not incorporated.

2.5. Plant sampling, plant analyses, and SPAD measurements Maize was harvested when 50% of the plants on the PLOUGH plots had reached the 3rd, 6th, and 9th leaf stages (fully expanded leaves), at 50% pollen shedding, and at silage maturity ( - 320 g dry matter/kg fresh weight). On average, these sampling dates corresponded to 204, 364, 510, 648, and 1176°Cd after maize planting (8°C base temperature). Ryegrass was sampled on the same dates as the maize and, in addition, at the 1st leaf stage of maize. The grass was cut at ground level. The sampling area was 3 m 2 except at silage maturity when it was 21 m 2. Aliquots of the maize and grass samples were dried at 65°C for 72 h and ground to pass through a 0.75-mm screen. Plant material collected in 1991 was digested with hot sulfuric acid at 150°C and afterwards at 420°C for 90 min; the analysis for ammonium was done with an autoanalyzer (Autoanalyzer II, Technicon Industrial Systems). The concentration of N in the remaining samples was assessed with a micro-Kjeldahl system (Kjeltec Auto 1030, Tecator). For the determination of nitrate, 50 mg plant material was incubated with 0.5 ml of 80% ethanol for 10 min at 60°C. After adding 5 ml of bidistilled water, samples were placed in a shaking-bath at 60°C for 50 min. The tubes were centrifuged at 3000 rev./min for 10 min; the supernatant Table 3 Dry matter yields (t/ha) of silage maize under three cropping systems and two N treatments. Data are means across three cropping years Cropping system

N treatment

Yield

PLOUGH GRASS/HERB GRASS/MECH PLOUGH GRASS/HERB GRASS/MECH

NIl0 NI I0 Nll0 N250 N250 N250

16.3 11.2 7.6 17.8 16.6 15.1 0.5

SE F-test Cropping system N treatment Cropping system x N treatment Significant at *P = 0.05 and **P = 0.01.

was analyzed for nitrate with an autoanalyzer (Evolution II, Alliance Instruments). To estimate the leaf chlorophyll content, the SPAD 502 chlorophyll meter from Minolta was used. Meter readings were taken on ten representative plants from each of the two center rows on the uppermost fully expanded leaf, midway between the base and tip and the leaf margin and midrib of the lamina.

2.6. Experimental design and statistical analysis The experiment was laid out as a randomized complete block design with four replicates. In the analysis of variance, years were treated as random effects (Gomez and Gomez, 1984).

3. Results

3.1. Maize silage yield The analysis of variance indicated a significant (P = 0.05) year x cropping system x N rate interaction, but the ranking of the cropping systems was similar in the various cropping years (Garibay et al., 1997). This paper, therefore, presents yield means across the years only. With N ll0, the GRASS/ MECH system produced only 47% and the GRASS/ HERB system only 69% of the total aboveground dry matter of maize produced under PLOUGH (Table 3). As compared with the N 110 treatment, the N250 treatment resulted in yield gains for all cropping systems, but the increment was smallest for the PLOUGH system. Nevertheless, PLOUGH was still the most productive cropping method; maize grown under GRASS/HERB and GRASS/MECH produced 95% and 85% of the dry matter produced under PLOUGH. The cropping system x N rate effect on dry matter yield was highly significant (P = 0.01).

3.2. Seasonal trends of some indicators of the N status of maize The N concentration in the tops was initially similar under all maize production systems (Fig. l a). From the 6th leaf stage onwards, however, the N concentration was higher in PLOUGH maize than in sodplanted maize. Irrespective of N level, GRASS/

231

0

--,

!

,

I

'

I

50

'

I

'

8.0

Ill

•

s.o

b

t

-

a 40

.|

4.0

30

2.0 Z

10 0.0 0

t.l

0

.

l

3OO 6 0 o

,

l

9oo

,

I

1200

Growing degree days (*C d)

0

300 600 900 1200

Growing degree days (*C d)

Fig. 1. Concentrations of N (a) and nitrate (b) in the tops of silage maize under three cropping systems, two N treatments, and five developmental stages (3rd, 6th, and 9th leaf stages, pollen shedding, silage maturity). Data are means across two (nitrate concentration at the 3rd leaf stage) or three cropping years. Bars are SE.

HERB maize tended to have a higher N concentration than maize under GRASS/MECH, but the differences were smaller with N250 than with N110. At the 3rd leaf stage, the N contents of the maize tops were highest under PLOUGH and lowest under GRASS/HERB (Table 4). From the 6th leaf stage onwards, however, GRASS/HERB maize contained more N than did GRASS/MECH maize; PLOUGH maize had still the highest N content. The N treatment did not affect the ranking of the cropping systems, but the differences among the cropping systems were more pronounced under N 110 than under N250.

At the 3rd and 6th leaf stages, nitrate levels were by far the highest for PLOUGH maize (Fig. l b). With N II0, the nitrate concentrations decreased rapidly, independent of cropping system. From the 9th leaf stage onwards, maize on the GRASS/HERB and GRASS/MECH plots contained only trace amounts of nitrate, whereas in PLOUGH maize, small quantities of nitrate were still detected at pollen shedding. Under N250, maize in the PLOUGH system had the highest nitrate concentration throughout the cropping period. While differences between GRASS/HERB and GRASS/MECH maize were still small at the 3rd leaf stage, nitrate concentrations tended to be higher for GRASS/HERB maize than for GRASS/MECH maize at the 6th and 9th leaf stages. In interpreting the changes in SPAD readings over time (Fig. 2), it must be taken into account that the measurements were made on different leaves. Maize development was only slightly affected by the cropping systems; it is possible, therefore, to compare the cropping methods on a given observation date. With N110, PLOUGH maize had the greenest leaves from the third (1991 and 1993) or from the first measurement (1992) onwards. In 1991 and 1993, differences between the mulch seeding systems occurred when the plants had reached or just passed the 6th leaf stage, while in 1992, the curves started to diverge only after pollen shedding. From these developmental stages onwards, the SPAD readings were slightly (1992) or clearly (1991 and 1993) higher under the

Table 4 Nitrogen content of the tops of silage maize (kghaa) under three cropping systems, two N treatments, and five stages of development. Data are means across three cropping years Cropping system

N treatment

3rd leaf stage

6th leaf stage

9th leaf stage

Pollen shedding

Silage maturity

PLOUGH GRASS/HERB GRASS/MECH PLOUGH GRASS/HERB GRASS/MECH SE F-test Cropping system N treatment Cropping system x N treatment

N110 N110 N110 N250 N250 N250

4.3 3.6 4.1 4.1 3.7 3.9 0.2

35.2 22.9 18.9 38.0 24.7 22.7 0.9

89.5 47.9 29.1 107.8 73.7 61.1 1.5

110.9 56.7 36.7 156.0 123.7 108.5 3.4

157.9 84.4 49.3 206.4 174.7 158.0 5.9

ns

*

*

**

**

ns

ns ns

(*) *

** *

* *

ns

Significant at (*)P = 0.10, *P = 0.05 and **P = 0.01. ns not significant.

232

•

I

'

I

Nl10 ~ ~

50

'

i

•

i

'

-?

i

'

I

N250 ~~~

'

i

'

competed with the maize crop after the first herbicide application.

-

I

1-

~

40

•

1991

~ 3o

4. Discussion

or) 20

_

-~. , •

I i

, •

I I

, '

I i

, '

,'

1 1

Ii

,•

Ii

', ~

I

,

II

5O w

~ 40

1992

O.

50 c

GRASS/HERB GRASS/MECH

"'0-_

20

~

~ -- N250~~

~I~LO ~,

-

-

40

1993

n 20 , 300

600

I 900

I 1200

,

I

,

300

Growing degree days

I 600

(*C

,

I 900

,

I 1200

d)

Fig. 2. Seasonal trends in chlorophyll meter readings taken on the uppermost fully expanded leaves of silage maize under three cropping systems and two N treatments. Arrows indicate the 3rd, 6th, and 9th leaf stages, pollen shedding, and silage maturity. GRASS/HERB system than under the GRASS/MECH system. With N250, the variations between the maize production systems were, generally, fairly small and inconsistent over time.

3.3. Nitrogen contents of the grass strips Immediately before the grass strips were mulched, i.e. at the 1st, 3rd, and 6th leaf stages of the maize, the N contents of the grass strips in the GRASS/MECH system varied from about 20 to 30 kg/ha (Table 5). After the last mowing, the grass swards recovered slowly, and the grass N contents remained almost constant (N110) or even decreased (N250) during grain filling. At the 1st leaf stage, i.e. prior to the first herbicide application in the GRASS/HERB system, grass N contents were the same in the GRASS/ MECH and GRASS/HERB systems. Even though the herbicide treatments were carded out between the first and second samplings, the grass in the GRASS/HERB system accumulated an additional 8 kg N/ha (mean of N I I 0 and N250) between the 1st and 3rd leaf stages (data not shown), indicating that the cover crop still

The present data demonstrate that differences in the N status of PLOUGH maize and sod-planted maize already occur at very early stages of maize development, even though more fertilizer N (on average 56 kg/ha) was invested in the GRASS/HERB and GRASS/MECH systems to offset the low mineral N contents of the soil immediately prior to maize sowing (Table 2). Furthermore, it is noteworthy that the extra N applied to the GRASS/HERB and GRASS/MECH plots was placed in the maize rows, suggesting that the spatial distribution of N was more favorable for maize in the mulch seeding systems than for PLOUGH maize. Nevertheless, as early as the 3rd leaf stage, low nitrate concentrations indicated the beginning N deficiency of sod-planted maize (Fig. l b). It was not until the 6th leaf stage that differences in the wholeplant N concentration between the cropping systems became detectable (Fig. I a). Even though concentrations of N tend to increase with increasing availability of N (Binford et al., 1992), the tissue N concentration is not a perfect indicator of the N supply to the plants. First, the tissue test for N has limited ability to detect excesses of available N in the soil (Binford et al., 1992). Second, it may be that differences in the concentration of N merely reflect differences in the developmental stage of the plants because the concentration of N declines with increasing age, i.e. increasing plant Table 5 Nitrogen content of the aboveground biomass of the grass strips (kg/ha) in the GRASS/MECH system under two N treatments. Corrections were made to account for the fact that the tilled strips were not grassed over. Data are means across 3 years Developmental stage

N 110

N250

SE

F-test~

Ist leaf stage 3rd leaf stage 6th leaf stage 9th leaf stage Pollen shedding Silage maturity

i 7.5 20.3 27.3 9.8 17.1 18.1

20.0 20.7 30.0 ! 2.6 19.4 9.4

1.1 0.9 2.2 1.1 1.9 4.8

ns ns ns ns ns ns

ans not significant.

233

mass (Greenwood et al., 1990). Third, under severe N deficiency, improving the availability of N to the crop may cause slight decreases in the tissue N concentration. This phenomenon is usually explained by nutrient dilution brought about by the higher production of plant dry matter (Wikstr6m, 1994). Consequently, the observation that, from the 6th stage onwards, PLOUGH maize showed the highest whole-plant concentration of N does not prove that more N was available to PLOUGH maize than to sod-planted maize. However, PLOUGH maize also had the highest N shoot content (Table 4), indicating that less N was available to sod-planted maize than to PLOUGH maize. The observation that, at the 3rd leaf stage, the tops of GRASS/HERB maize contained less N than those of GRASS/MECH maize may be traced to herbicide stress in the GRASS/HERB system. Before the 6th leaf stage was reached, plants treated with N 110 showed another symptom of N stress: the uppermost fully expanded leaves of sod-planted maize were slightly paler than those of conventionally produced maize (Fig. 2). Low SPAD readings are indicative of low concentrations of chlorophyll (Dwyer et al., 1991) and N (Schepers et al., 1992; Blackmer et al., 1994; Dwyer et al., 1995). In fact, from the 6th leaf stage onwards, the whole-shoot concentration of N was lower for the plants in the mulch seeding systems than for those on the PLOUGH plots (Fig. l a). Thus, the SPAD instrument detected a reduced N supply to the plants at a relatively early stage. The determination of the plant N status with a chlorophyll meter is simple, fast, and non-destructive, suggesting that the SPAD 502 instrument is a useful tool for testing tissues and may aid in the management of N fertilizer. However, the meter's costs may be prohibitive to many farmers. Furthermore, at least in the present study, the SPAD technique was often unable to detect the differences in N status between the cropping systems when the N supply was high. This may be due to the fact that SPAD readings reach a plateau at high leaf N concentrations (Dwyer et al., 1995). It is possible that the SPAD method provides better results if the measurements are made on older leaves (Piekielek and Fox, 1992), i.e. leaves with relatively low N concentrations. While tissue testing may be very helpful in detecting cropping system effects on the N status of maize in a given environment, for a given cultivar, and at a given stage

of development, the applicability of tissue analyses in the management of N fertilizer is limited by the lack of generally valid critical values (Cerrato and Blackmet, 1991). Binford et al. (1992) concluded that a tissue test based on concentrations of N in young plants is not a viable alternative to soil nitrate tests. Unfortunately, soil testing for mineral N in maize cropping systems which use living mulches seems to be impractical for two reasons. In living mulch systems, N fertilizer placement in or near the maize rows is a prerequisite for attaining acceptable maize yields (Garibay et al., 1997). Banding of N fertilizer leads to an extremely uneven distribution of mineral N in the soil, and the results of soil tests for mineral N, therefore, are strongly dependent on the sampling location (Garibay, 1996). Furthermore, it is unclear to which extent mineral N dectected in the soil is available to the main crop. Various factors may have contributed to the relatively poor N status of sod-planted maize. First, it is possible that more soil N was mineralized on the autumn-ploughed land than on the strip-tilled plots where roughly two-thirds of the topsoil remained undisturbed (Dowdell and Cannell, 1975; Powlson, 1980). Second, it is likely that some N was microbially immobilized in the tilled strips. Data presented by Jensen (1991) indicate that the C/N ratio in the roots of well developed Italian ryegrass stands is fairly high (60-70). Roots were reported to make up more than 60% of the total dry matter of the residues (stubble plus roots) of Italian ryegrass (Renius et al., 1992), suggesting that the average C/N ratio in the grass residues is above the critical value for net N immobilization. The third and likely the most important factor is that the grass strips removed N from the soil (Table 5). The grass N content was almost unaffected by the level of N supply. This can be attributed, in part, to the fact that the N fertilizer was placed in the maize rows. Furthermore, due to its larger canopy (data not shown), maize under N250 was more competitive with the grass strips than it was under N110. The stagnation (N110) of and decline (N250) in grass N after pollen shedding is probably due to the fact that the grass was strongly shaded by the fully developed maize canopy during the second half of the growing season. In the GRASS/MECH system, at least 80 kg mineral N/ha was taken up by the grass (summed

234

grass N contents on the mulching dates plus maximum grass N content later in the season) and converted to organic N (Table 5). The bulk of grass N returned to the soil surface after mulching. It is assumed that some grass N was mineralized and soon reassimilated by the regrowing grass strips, while the rest remained organically bound. There is no indication in Fig. l a,b and Fig. 2 that the decaying grass residues provided significant amounts of N to the maize crop late in the season. In evaluating the ecological value of the mulch seeding systems tested, it must be considered that N bound in the grass debris may be mineralized after tillage operations, thus increasing the nitrate leaching hazard during periods when the soil is uncropped or when the N uptake capacity of the crop is limited. It is concluded that efforts to optimize the environmentally sound living mulch systems should focus on reducing the competition between maize and the cover crop for N. An earlier and/or a more effective suppression of the cover crop sod would help to improve the availability of N to the main crop. Another approach is the use of legumes instead of Italian ryegrass as a cover crop. Legumes have relatively low C/N ratios (McKenney et al., 1993) suggesting that, in the tilled strips, N would be released rather than immobilized. Legume cover crops are less vigorous than Italian ryegrass and, therefore, are weaker competitors for N. Furthermore, as compared to Italian ryegrass, a more rapid decomposition of the debris in the mechanically stunted cover crop strips and, thus, a faster release of N may be expected. A negative aspect of leguminous winter cover crops is, however, that they provide only low dry matter yields in autumn and spring.

Acknowledgements This project was funded by the Swiss Federal Government and carried out within the scope of the European program COST 814.

References Ammon, H.-U., Scherrer, C. and Mayor, J.-P., 1995. Unkrautentwicklung und Bodenbedeckung. Agrarforsch., 2: 369-372.

Bigler, F., Waldburger, M. and Frei, G., 1995a. Vier Maisanbauverfahren 1990 bis 1993: Krankheiten und Schadlinge. Agrarforsch., 2: 380-382. Bigler, F., Waldburger, M. and Frei, G. 1995b. Vier Maisanbauverfahren 1990 bis 1993: lnsekten und Spinnen als NUtzlinge. Agrarforsch., 2: 383-386. Binford, G.D., Blackmer, A.M. and Cerrato, M.E., 1992. Nitrogen concentration of young corn plants as an indicator of nitrogen availability. Agron. J., 84, 219-223. Blackmer, T.M. and Schepers, J.S., 1994. Techniques for monitoring crop nitrogen status in corn. Commun. Soil Sci. Plant Anal., 25. 1791-1800. Blackmer, T.M., Schepers, J.S. and Varvel, G.E., 1994. Light reflectance compared with other nitrogen stress measurements in corn leaves. Agron. J., 86: 934-938. Box, Jr., J.E., Wilkinson, S.R., Dawson, R.N. and Kozachyn, J., 1980. Soil water effects on no-till corn production in strip and completely killed mulches. Agron. J., 72: 797-802. Cerrato, M.E. and Blackmer, A.M., 1991. Relationships between leaf nitrogen concentrations and the nitrogen status of corn. J. Prod. Agric., 4: 525-531. Dowdell, R.J. and Cannell, R.Q., 1975. Effect of ploughing and direct drilling on soil nitrate content. J. Soil Sci., 26: 53-61. Dwyer, L.M., Tollenaar, M. and Houwing, L., 1991. A nondestructive method to monitor leaf greenness in corn. Can. J. Plant Sci., 7 I: 505-509. Dwyer, L.M., Anderson, A.M., Ma, B.L., Stewart, D.W., Tollenaar, M. and Gregorich, E., 1995. Quantifying the nonlinearity in chlorophyll meter response to corn leaf nitrogen concentration. Can. J. Plant Sci., 75:179-182. Echtenkamp, G.W. and Moomaw, R.S., 1989. No-till corn production in a living mulch system. Weed Technol., 3: 261-266. Garibay, S.V., 1996. Maize Production in Living Mulches in a Humid Temperate Climate. Ph D thesis, ETH Zurich, Switzerland. Garibay, S.V., Stamp, P., Ammon, H.-U. and Feil, B., 1997. Yield and quality components of silage maize in killed and live cover crop sods. Eur. J. Agron., 6:179-190. Gomez, K.A. and Gomez, A.A., 1984. Statistical procedures for agricultural research. 2nd edn., John Wiley and Sons, New York. Greenwood, D.J., Lemaire, G., Gosse, G., Cruz, P., Draycott, A. and Neeteson, J.J., 1990. Decline in percentage N of C3 and C4 crops with increasing plant mass. Ann. Bot., 66: 425-436. Iversen, K.V., Fox, R.H. and Piekielek, W.P., 1985. The relationships of nitrate concentrations in young corn stalks to soil nitrogen availability and grain yields. Agron. J., 77: 927-932. J~iggi, W., Oberholzer, H.-R. and Waldburger, M., 1995. Vier Maisanbauverfahren 1990 bis 1993: Auswirkungen auf das Bodenleben. Agrarforsch., 2: 361-364. Jensen, E.S., 1991. Nitrogen accumulation and residual effects of nitrogen catch crops. Acta Agric. Scand., 41: 333-344. Klocke, N.L., Nichols, J.T., Grabouski, P.H. and Todd, R., 1989. Intercropping corn in perennial cool-season grass on irrigated sandy soil. J. Prod. Agric., 2: 42-46. McClenahan, E.J. and Killorn, R., 1988. Relationship between basal corn stem nitrate N content at V6 growth stage and grain yield. J. Prod. Agric., 1: 322-326.

235

McKenney, D.J., Wang, S.W., Drury, C.F. and Findlay, W.l., 1993. Denitrification and mineralization in soil amended with legume, grass, and corn residues. Soil Sci. Soc. Am. J., 57: 1013-1020. Piekielek, W.P. and Fox, R.H., 1992. Use of a chlorophyll meter to predict sidedress nitrogen requirements for maize. Agron. J., 84: 59-65. Powlson, D.S., 1980. Effect of cultivation on the mineralization of nitrogen in the soil. Plant Soil, 57: 151-153. Prasuhn, V. and Braun, M., 1994. Absch~itzung der Phosphor- und Stickstoffverluste aus diffusen Quellen in die Gewiisser des Kantons Bern. Schriftenreihe der FAC Liebefeld 17. EidgenGssische Forschungsanstalt fur Agrikulturchemie und Umwelthygiene, CH-3097 Liebefeld-Bern. Renius, W., Liitke Entrup, E. and Liitke Entrup, N., 1992. Zwischenfruchtbau zur Futtergewinnung und Grtindiingung. 3rd Edition. DLG-Vedag, Frankfurt (Main). Riiegg, W.T., Richner, W., Stamp, P., and Feil, B., 1997. Growth

and productivity of minimum tillage maize planted in winter cover crop residues. Eur. J. Agron. (in press). Riittimann, M., Schaub, D., Prasuhn, V. and Rtiegg, W., 1995. Measurement of runoff and soil erosion on regularly cultivated fields in Switzerland - some critical considerations. Catena, 25: 127-139. Schepers, J.S., Francis, D.D., Vigil, M. and Below, F.E., 1992. Comparison of corn leaf nitrogen concentration and chlorophyll meter readings. Commun. Soil Sci. Plant Anal., 23:2173-2187. Walther, U., 1983. Die heutigen Bodenuntersuchungsmethoden im Dienste einer optimalen Pflanzenern~rung. Die Griine, 7: 315. Wehrmann, J. and Scharpf, H.C., 1979. Der Mineralstickstoffgehait des Bodens als Ma/~stab fur den Stickstoffdtingerbedarf (NminMethode). Plant Soil, 52: 109-126. WikstrGm, F., 1994. A theoretical explanation of the Piper-Steenbjerg effect. Plant, Cell Environ., 17: 1053-1060.

This Page Intentionally Left Blank

© 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

237

Nitrogen transformations after the spreading of pig slurry on bare soil and ryegrass using 15N-labelled ammonium T. Morvan a'*, Ph. Leterme a, G.G. Arsene b, B. Mary c aUnit~ INRA d'Agronomie, ENSAR, 65 rue de Saint-Brieuc, F35042 Rennes, France bUSAMVB Timisoara, Colea Arodulini, 19, Timisoara 1900, Romania CUnit~ 1NRA d'Agronomie, rue Fernand Christ, BP 101, F02004 Laon Cedex, France

Accepted 13 April 1997

Abstract A short field experiment (27 days) was carded out in summer 1995, to study the effect of an actively growing grass sward on nitrogen transformations of a pig slurry. The ammonium fraction of the slurry was labelled with (~5NH4)2SO4. The slurry was spread manually on microplots in mid-June, at the rate of 3 l/m2, on a cut ryegrass sward, and compared with bare soil. Absorption of 15N-labelled NH4 by the grass occurred very rapidly, attaining 41% after 13 days and showing no further change at 27 days. The gaseous losses, mainly through volatilization of ammonia, were considerable. ~SN recovery in soil and plant material on day 27 was 42.5% (+1.2) on the bare soil, versus 57.4% (+3.1) on the ryegrass. The grass sward significantly reduced: (i) volatilization, as shown by the difference of 14.9% in ~SNrecovery, on the 6th day; (ii) immobilization, which was 25% (+2.2) on day 27 on bare soil and 16.4% (-1-2.9) in the presence of ryegrass. 15N-labelled inorganic nitrogen was completely depleted beneath the ryegrass, 27 days after application, whereas ammonium was depleted and the nitrate was equal to 16.4% (-I-1.6) of the applied NH4 on the bare soil. It is clearly apparent that the ammonia from the slurry is more efficiently used when applied to an actively growing sward, rather than to bare soil, even though a significant portion of the plant is involved in internal recycling. © 1997 Elsevier Science B.V. Keywords: Slurry; 15N; Grassland; Volatilization; Immobilization

1. Introduction Slurries, because of their high ammonium content, provide nitrogen which is quickly available for crops. The availability of the ammonium fraction is determined by gaseous loss and microbial immobilization. The first occurs mainly through the volatilization of

*Corresponding author. Tel.: +33 99 287231; fax: +33 99 287230; e-mail: [email protected]

ammonia, and is highly variable, (20-70% of total ammonium nitrogen (TAN) applied) (Lauer et al., 1976; Beauchamp et al., 1982; Pain et al., 1989; G6nermont, 1996), whereas microbial immobilization represents 15-35% of the TAN (Morvan et al., 1996). Slurry incorporation slightly reduces ammonia volatilization, but is not always possible, especially as it may lead to plant injury, for example after late winter applications on wheat or rapeseed. Furthermore, high nitrogen utilization efficiencies have been obtained for slurry ammonium

Reprinted from the European Journal o f Agronomy 7 (1997) 181-188

238

after surface spreading of diluted pig slurries in actively growing wheat and were correlated with low levels of volatilization. A reduction in volatilization brought about by living plants has been reported and might be expected from: (i) absorption by plant leaves of ammonia volatilized from the underlying soil, as reported by Denmead et al., 1976; (ii) absorption of ammonium through the roots and (iii) microclimatic effects due to the canopy (Faurie and Bardin, 1979). Contradictory results have however been reported by other authors (Thompson et al., 1990), which could be explained by the greater surface area resulting from slurry retention on the leaves. What effect do plants have upon nitrogen immobilization? We might expect high rates of microbial immobilization under a grass sward, because of the large amounts of root exudate. Short-duration experiments, involving the measurement of actual rates of NH4 uptake by plants and microbes, have shown that microbial immobilization may be five times higher than plant uptake (Jackson et al., 1989). Ledgard et al. (1989), studying the partitioning of ~SN-labelled ammonium applied to grass-clover pasture confirmed that microbial immobilization was a significant component of 15N balance, but also observed that more fertilizer was immobilized when plant growth was slow, due to lower temperatures. Immobilization is known to depend on the amount and persistence of ammonia in the soil. Thus, we may suppose that an actively growing plant able to rapidly absorb significant amounts of inorganic nitrogen, will reduce microbial immobilization. The few studies describing the effect of plants on nitrogen transformations after slurry addition sometimes give contradictory conclusions about volatilization, and rarely provide a complete description of soil-plant behaviour and competition, either because only one process was studied, or because the time scale of the experiment was too short, or too long. The aim of the present work was to obtain a better understanding of the effect of an actively growing plant on nitrogen transformations of a pig slurry ammonia pool, using labelled ammonium. A short field experiment was therefore carried out from midJune to mid-July, in order to ensure that climatic conditions were favourable to plant growth.

Table 1 Physical and chemical properties of the soil, and slurry composition Soil properties Particle size distribution (%) Clay Silt Sand Total N (%) pH KCI Bulk density of the soil layers 0-10 cm 10-20 cm Slurry composition N-NH4 (g/l) (after addition of ammonium sulfate) Total N (Kjeldhal) (g/l) pH Dry matter (%)

14.4 72.5 13.1 0.13 6.2 1.53 1.50 4.02 6.12 7.36 1.4

2. Materials and methods

2.1. Site and design The experiment was conducted at Le Rheu Experimental Station (INRA), in western France, on a loamy soil. Some of the chemical and physical properties of this soil are summarized in Table 1. The fate of the ammonia fraction of a pig slurry was studied using 15N-labelled NH4. Two treatments were compared: surface spreading on a ryegrass sward, and on bare soil. Daily temperatures were relatively high, varying from 15°C to 25°C (the mean air temperature was 20.3°C over the 27 days), and were favourable both to plant growth and to ammonia volatilization and nitrogen biotransformations, such as nitrification, mineralization and immobilization. The amount of rainfall and change in soil moisture in the soil surface layer are shown in Fig. 1; 2 x 10 mm were applied by irrigation, the day before the start of the experiment, and on day 3, to prevent the soil from drying out. The soil moisture therefore varied from 60 to 100% of the field capacity over the first 13 days and was favourable for microbial activity and plant growth. No significant differences were noticed between the two treatments during this period. The plots were 4.70 m x 2.30 m and laid out in a

239

0,25

30

~'25-

0,2

g 820

is

0

E

I

', ', ) ', : : : ', : : I : I'1 : I ;l:m,,-,,-,-: 1 3 5 7 9 11 13 15 17 19 21 23

, I ;

Time

I

i ', , 0

25

27

(days)

I m Rainfall + Irrigation - - Soil moisture BS* (%) 1 --,- Soil moisture R* (%)

nium at the rate of 188 g/ha), followed by removal of the aerial residues, just before spreading. The characteristics of the soil surface (structure, bulk density) and soil moisture, controlling initial infiltration of the slurry, and consequently the intensity of ammonia volatilization, were therefore similar in each treatment. Chemical weed control probably had no effect on the biological activity of the soil as the kinetics and level of immobilization of ZSN-labelled nitrogen were similar to those observed after a comparable slurry application to a soil that was not subjected to chemical weed control (unpublished data).

Fig. 1. Rainfall plus irrigation, and change in soil moisture on the bare soil (BS) and under ryegrass (R).

2.2. Soil and plant sampling

randomized block design with three replications. Each plot was divided into five microplots of 0.60 m 2, corresponding to the sampling area for a given date of measurement. Microplots were separated from each other by a 30-cm discard area and from the boundaries by 50-cm guard strips. The ammonium fraction of a pig slurry, (composition in Table 1), was enriched with 15N using a solution of (15NH4)2SO4 10% atom excess, which was added and thoroughly mixed to the slurry. The initial atom excess of the ammonium pool (1.089%) was determined just after spreading, by sampling the soil surface. The enriched slurry was applied at the rate of 3 l/m 2 on June 20th 1995, between 1500 h and 1645 h, in warm, sunny conditions. The amounts of ammonium and organic nitrogen applied were equal to 121 and 63 kg N/ha respectively. A watering can with a distribution bar was used to obtain as even a distribution as possible, and the plots were divided into microplots each receiving the same quantity of slurry. No run-off was observed after spreading, despite the high soil compaction (bulk density of the 0-10 cm soil layer: 1.53 g/cm3). The slurry was applied: (i) to a 2-year-old cut ryegrass (Lolium perenne) sward, which had not been fertilized with nitrogen since the date of sowing. The nitrogen content in plant material harvested at a height of 0.5 cm was therefore very low (1,12%) at the beginning of the trial. The grass sward was cut one week before the experiment, and was 10-12 cm high on the day of application; (ii) to bare soil, obtained by chemical destruction of the grass (glifosate ammo-

The microplots were sampled 1, 3, 6, 13 and 27 days after slurry spreading. The aerial parts were cut just above the soil, on a square area of 0.25 m 2. The plant material was washed free of soil and slurry, the volume of washing water was measured, and the plants and water sampled for inorganic nitrogen and 15N analysis. Soil samples were taken from the 0-10 cm and 1020 cm depths, to determine root biomass, organic and inorganic soil nitrogen. A 60 mm diameter probe was used for root sampling, and a 20 mm diameter probe for soil sampling. For soil nitrogen analysis, samples were obtained from each microplot by mixing 27 cores from each soil layer, and passing the whole sample through an 8-mm mesh sieve. Soil samples for roots were obtained by mixing six cores taken with the 60-mm diameter probe. The roots were separated from the soil by washing on a 2-mm mesh sieve; herbage residues were removed by hand. Above-ground parts and roots were dried at 60°C, finely ground to powder and put into tin containers for total nitrogen content determinations and ~5N atom excess analysis.

2.3. Analytical procedures Inorganic nitrogen in the soil was determined in a KC1 extract (600 ml 1 M KCI/300 g fresh soil, shaken for 30 min, then filtered through a Whatmann 42 filter), using the fractionated steam distillation with MgO for ammonium and Dewarda's alloy for nitrate analysis (Drouineau and Gouny, 1947). Organic plus clay-fixed nitrogen and ~SN was measured on a sample

240

of moist soil, after removal of the inorganic nitrogen, as described by Recous et al., 1988; the sample was dried at 60°C, and finely ground. Atom excess was determined on subsamples of plant, soil, and dried solutions resulting from the steam distillations, using a total combustion technique linked to a VG SIRA 9 mass spectrometer (Recous et al., 1988).

20

I

15

Z

_=

® 10 JD M Z u~

¥

Bare soil

s

0 3. R e s u l t s

I

Ryegrass]

5

10

15

20

25

30

35

Time (days)

3.1. Inorganic nitrogen dynamics Inorganic nitrogen occurred mainly in the surface soil layer (0-10 cm). The amounts of 15NHn-labelled ammonium observed in the 10-20 cm soil layer did not exceed 0.5% of the applied ammonium. Small amounts of 15N-labelled nitrate, not exceeding 1.3% of the applied nitrogen, were measured in the lower layer of the bare soil, on days 13 and 27. The fate of the ammonium nitrogen is shown in Fig. 2: the curves represent the change in the amount of ammonium in the 0 - 2 0 cm soil layer, and ammonia deposition on the leaves in the ryegrass treatment. The amount of ammonium deposited on the plant leaves was relatively large on day 1: this suggests that the initial deposition was probably considerable, if comparable with the great decrease of the ammonium pool in the soil observed during the first day. Ammonium amounts decreased sharply during the

Fig. 3. The proportions of the ~SN-labelled NH4 present as ~SNlabelled NO3 in the 0-20 cm soil layer. (Vertical bars indicate the standard deviation of the three replicates; SD not shown are smaller than symbol size.) first 6 days. The kinetics of 15N-labelled NH4 were similar in each treatment; the plant tended to stimulate the depletion of ammonium, particularly between days 6 and 13. The change in 15N-labelled NO3 in the soil is shown in Fig. 3. Some nitrification of the slurry ammonium had occurred by the end of the experiment, because the ammonium pool had been depleted by that time, in both treatments. ~SN-labelled NO3 content steadily increased until the end of the experiment in the bare soil, attaining only 16% of the nitrogen applied, but it was completely depleted beneath the sward. We did not observe a true latent period at the beginning of the trial as is often reported (Le Pham et al., 1984).

3.2. Dry matter and nitrogen absorption dynamics ,-=- Bare soil i--~- Ryegrass ! OepositJon

80 ~ z

~

4o 20 0 0

5

10

15

20

25

30

Time (days)

Fig. 2. The proportions of 15N-labelled NH4 in the 0-20 cm soil layer, and deposition on ryegrassleaves, after application of 121 kg N-NH4/ha from a pig slurry. (Vertical bars indicate the standard deviation of the three replicates; SD not shown are smaller than symbol size.)

The pattern of 15N absorption by the whole plant is shown in Fig. 4; nitrogen absorption occurred rapidly: the nitrogen utilization efficiency had already reached 11.5% the day after spreading and had attained 41% by day 13 and remained stable till day 27. Because the soil moisture, temperature and soil ammonia content were similar in both treatments (Figs. 1 and 2) during the first 6 days, it can be assumed that nitrification occurred at the same rate in both treatments, during this period. It then follows from the comparison of 15N-labelled nitrate dynamics and ~SN inorganic nitrogen absorption that absorption was mainly due to the uptake of ammonium or ammonia. The two routes of assimilation, through the roots and leaves, were probably efficient (see Section 4),

241

iod corresponding to the amount of nitrogen taken up by the grass.

35 30 25 z

it)

3.4. 15N balance

2O 15 It

10 5

Roots_J

0 ..

i

I

0

5

10

I

t

15

20

25

30

Time (days)

Fig. 4. The proportions of the ISN-labelled NH4 present in the above-ground parts and roots following the pig slurry application. (Vertical bars indicate the standard deviation of the three replicates; SD not shown are smaller than symbol size.)

and equal quantities can be found in aerial plant parts and roots on days 1 and 3. Plant dry matter and nitrogen content increased simultaneously during the experiment. The aboveground parts of the ryegrass rose from 4.1 to 6.8 t dry matter/ha, representing a mean growth rate of 0.1 t dry matter/ha per day. The nitrogen content of the aerial parts rose from 1.12 to 1.74%. The proportion of tSN-labelled nitrogen measured in the whole plant was 15.8% of the amount of JSN-labelled NH4 applied, on day 1, and reached a maximum value of 33.4% on day 13.

3.3. Immobilization The pattern of immobilization is shown in Fig. 5. In both treatments, ~5N immobilization rose sharply during the first 3 days; on bare soil, it exceeded the value in the ryegrass with 6% as early as the first day, and continued to increase, whereas it remained steady in the ryegrass sward. The difference observed on day 1 might be related to the significant deposition of ammonia and organic carbon from the slurry on the leaves (ammonia deposition representing 7% of the total applied ammonium, on day l) (Fig. 2), which thus reduced the amount of ammonium and organic carbon available in the soil. On the other hand, the increased margin between days 6 and 13 can be explained by the competitive effect of the plants which absorbed the inorganic ~5N-labelled nitrogen, the depletion of the ammonium pool during this per-

Fig. 6 shows the change in non-recovery of the ~SN, corresponding to gaseous losses. Denitrification was probably low during this experiment, in view of the climatic conditions and soil moisture. Ammonia volatilization probably accounts for most of the gaseous losses, all the more so as the kinetics of the ~SN nonrecovery exhibit a typical pattern of volatilization, as described by Sommer and Olesen (1991) and by Jarvis and Pain (1990). In fact, cumulative volatilization can generally be described by an asymptotic exponential curve, and is well fitted to the following equation: I.'-" Vmax ( l - e -kt) (Moal, 1995). The optimal values of the parameters of this equation were calculated for each treatment, using the Nlin procedure in the SAS package (SAS, 1988). The losses were the same on day l, in both treatments, attaining 40% of the applied nitrogen. This suggests that there was no significant microclimatic effect of the canopy, or that differences compensated. Volatilization was significant on the bare soil between days l and 3 (+l 9% according to the adjusted curve), but almost stopped after day 1 on the ryegrass treatment (+2% according to the adjusted curve). This halting of volatilization in the grass sward was unexpected, as (i) the amounts of ammoniacal nitrogen measured in the surface layer were still high on days l, 3 and 6, and were only a few kg N/ha less than the amounts measured on bare soil, (ii) volatilization after 35 u

30

I

•

i- OlV, o 15

.

I

1

14- a m ~ , I

-

-.0-

Ryegrass

0 0

5

10

15

20

25

30

Time (days) Fig. 5. Proportions of 15N-labelled NH4 immobilized in the two treatments. (Vertical bars indicate the standard deviation of the three replicates.)

242

80

60 z

¥

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

I

40

20

0 0

5

10

15

20

25

30

Time (days) [ =

Bare soil

:

Ryegrass ...... Bare soil adjust*

Ryegrass adjust* j

Fig. 6. Proportions of 15N-labelled NH4 not recovered, corresponding to gaseous losses, during the days following pig slurry application. Values were adjusted using the equation: y = a (1- e -k') (Vertical bars indicate the standard deviation of the three replicates.)

slurry spreading generally takes place over a period of 6 days (Jarvis and Pain, 1990; Moal, 1995) or even longer (Grnermont, 1996). The difference in ~SN balance between the two treatments attained a maximum from the 6th day onwards, and remained steady until the end of the experiment. The high rate of volatilization was the result of the climatic conditions, high ammonium content of the slurry, and surface spreading.

4. Discussion

The difference of 14.5% in nitrogen recovery shows that the grass sward significantly reduced volatilization. Similar results were obtained by Whitehead and Raistrick (1992) who used a direct method to quantify ammonia volatilization after spreading livestock urine; they observed a difference of 16% between cumulative volatilization over 8 days, on ryegrass, compared with bare soil. We observed, in agreement with these authors, that the effect of the canopy on volatilization was negligible during the first day, but substantial during the next 2-3 days. This effect could also be due to the rapid growth and great increase in nitrogen content during the 13 days following spreading. The low dry matter content of the slurry might also have reduced retention of the slurry on the canopy, despite the significant deposition of ammonium observed 24 h after the application. An

application of cattle slurry would probably have led to much higher gaseous losses from the canopy (Thompson et al., 1990), because of the higher dry matter content of cattle slurry. We also observed that volatilization halted after day 1 on the grass sward, despite the presence of significant amounts of ammoniacal nitrogen in the surface soil layer, between days 1 and 6. It is therefore almost certain that significant amounts of ammonia were given off at the soil surface in this treatment, the flux being slightly lower than that produced on bare soil, due to the lower temperatures of the soil surface (shade of sward) and slightly lower amounts of ammonium under grassland, on days 1 and 3. The total halting of net volatilization in the ryegrass treatment can only therefore be explained by the plant's absorption of the ammonia given off at the soil surface. It is consistent with the findings of Denmead et al. (1976) that ammonium concentration measured under a grass-clover pasture was greatest near the ground surface and exponentially decreased with height above the ground. The sink effect of the grass sward towards ammonium would in our experimental conditions thus be due both to absorption by the roots and direct assimilation of ammonia by the leaves (Faurie and Bardin, 1979, Jarvis and Pain, 1990). This is also in good agreement with the results of Porter et al. (1972) and Hutchinson et al. (1972) who showed that plant leaves from different species can absorb significant amounts of ammonia from the air. Lockyer and Whitehead (1986) and Whitehead and Lockyer (1987), measuring the uptake of gaseous ammonia by the leaves of Italian ryegrass exposed in chambers to different contents of ammonia in the air, observed that the amount of ammonia absorbed increased linearly with ammonia air concentration. In our experiment, the very low initial inorganic content of the soil probably contributed to the stimulation of ammonia absorption by the canopy, but contents comparable to those that we observed are frequently measured under grassland, outside the periods of fertilizer application; the situation under study was therefore not exceptional for this criterion. We also observed that the rate of immobilization was lower beneath the grass sward (Fig. 5). The driving force of immobilization is the amount of decomposable carbon (Recous et al., 1990). In our case, the carbon that could be assimilated by microorganisms

243

came (i) from the slurry which was supplied in equal quantities in both treatments, (ii) from dying leaves and roots and from root exudates in the case of ryegrass, and dead roots in the bare soil treatment, and (iii) from native organic matter. Mineral nitrogen is often a limiting factor of decomposition, which explains the very high stimulation of immobilization in the days immediately following an application, even without the addition of organic carbon; in this case, however, the measured levels of immobilization remain moderate, in the order of several mg N/kg soil, and are much greater following the addition of a carbon substrate (Recous et al., 1990). In this experiment, several factors could explain the differences between the two treatments: (i) the amounts of ammoniacal nitrogen measured at the soil surface during the first 6 days were not limiting for immobilization; the lower level of immobilization under ryegrass was therefore due to a lower availability of C-substrate, attributable in part to the deposition of organic matter from the slurry on the sward and the presence in bare soil of dead roots that had not yet been decomposed at the time of slurry application; (ii) the halting of immobilization of ~SN labelled nitrogen after the 6th day in the ryegrass treatment (whereas this continued on the bare soil) coincided in contrast with the total disappearance of mineral nitrogen under this treatment, due to plant absorption. Immobilization over 27 days accounted for 25% of the ~SN labelled nitrogen applied to the bare soil, and 16% in the ryegrass treatment. These rates of immobilization are relatively low, given that: (i) high levels of immobilization are expected under grassland (attaining 40-60% of 15Nrecovery, according to Jackson et al. (1989)) and that; (ii) soluble C was provided by the slurry. According to earlier findings (Ledgard et al., 1989; Guiraud et al., 1992), this low rate of immobilization might be due to low persistence of the ammonium pool, in our experimental conditions. Thus, this experiment shows that on a short time scale an actively growing grass sward can significantly modify partitioning of the ammonium pool following the application of a pig slurry. The amounts of 15N-labelled NH4 absorbed by the plants, attaining at the end of the experiment 41% of nitrogen applied for the whole plant, and 29% for the aerial parts, should be compared with the 16.4% of inorganic nitrogen available on the bare soil.

A study of the factors determining the infiltration and deposition of the slurry and the foliar assimilation of gaseous ammonia is required, as these factors govern the amounts of nitrogen volatilized, and thus have a considerable effect on nitrogen efficiency.

Acknowledgements We thank B. Blaise, Y. Fauvel for valuable assistance in the field and laboratory, R. Aubr6e for assistance in the field sampling, and O. Delfosse for 15N analysis.

References Beauchamp, E.G., Kidd, G.E. and Thurtell, G., 1982. Ammonia volatilization from liquid dairy cattle manure in the field. Can. J. Soil Sci., 62: ! 1-19. Denmead, O.T., Freney, J.R. and Simpson, J.R., 1976. A closed ammonia cycle within a plant canopy. Soil Biol. Biochem., 8: 161-164. Drouineau, G. and Gouny, P., 1947. Contribution /~ r6tude du dosage de I'azote nitrique par la m6thode Devarda. Ann. Agron., 17: 154-164. Faurie, G. and Bardin, R., 1979. La volatilisation de l'ammoniac. II. Influence des facteurs climatiques et du couvert v :6g~tal. Ann. Agron., 30:401-414. G6nermont, S., 1996. Mod61isation de la Volatilisation d'Ammoniac apr~s l~pandage de Lisier sur Parcelle Agricole. Thesis, University Paul Sabatier, Toulouse. Guiraud, G., Marol, C. and Fardeau, J.C., 1992. Balance and immobilization of (15NH4)2SO4 in a soil after the addition of Didin as a nitrification inhibitor. Biol. Fert. Soils, 14: 23-29. Hutchinson, G.L., Millington, R.J. and Peters, D.B., 1972. Atmospheric ammonia: absorption by plant leaves. Science, 175:771772. Jackson, L.E., Schimel, J.P. and Firestone, M.K., 1989. Short-term partitioning of ammonium and nitrate between plants and microbes in an annual grassland. Soil Biol. Biochem., 21: 409-415. Jarvis, S.C. and Pain, B.F., 1990. Ammonia volatilisation from agricultural land. Proc. Fert. Soc., 298: 3-35. Lauer, D.A., Bouldin, D.R. and Klausner, S.D., 1976. Ammonia volatilization from dairy manure spread on the soil surface. J. Environ. Qual., 5: 134-141. Ledgard, S.F., Brier, G.J. and Sarathchandra, S.U., 1989. Plant uptake and microbial immobilization of 15N-labelled ammonium applied to grass-clover pasture - Influence of simulated winter temperature and time of application. Soil Biol. Biochem., 21: 667-670. Le Pham, M., Lambert, R. and Laudelout, H., 1984. Estimation de ia valeur fertilisante azot :6e du lisier par simulation num6rique. Agronomie, 4: 63-74.

244

Lockyer, D.R. and Whitehead, D.C., 1986. The uptake of gaseous ammonia by the leaves of Italian ryegrass. J. Exp. Bot., 37: 919927. Moal, J.F., 1995. Volatilisation de l' Azote Ammoniacal des Lisiers apr~.s }~pandage: Quantification et l~tude des Facteurs d'Influence. Cemagref Dicova 229 pp. Morvan, T., Leterme, P. and Mary, B., 1996. Quantification par le marquage isotopique 15N des flux d'azote cons6cutifs ~ un 6pandage d'automne de lisier de porc sur triticale. Agronomie, 16: 541-552. Pain, B.F., Phillips, V.R., Clarkson, C.R. and Klarenbeek, J.V., 1989. Loss of nitrogen through ammonia volatilization during and following the application of pig or cattle slurry to grassland. J. Sci. Food Agric., 47: 1-12. Porter, L.K., Viers, F.G. and Hutchinson, G.L., 1972. Air containing nitrogen-15 ammonia: foliar absorption by corn seedlings. Science, 175: 759-761. Recous, S., Fresneau, C., Faurie, G. and Mary, B., 1988. The fate of labelled ~SN urea and ammonium nitrate applied to a winter wheat crop. Plant Soil, 112: 205-214. Recous, S., Mary, B. and Faurie, G., 1990. Microbial immobiliza-

tion of ammonium and nitrate in cultivated soils. Soil Biol. Biochem., 7: 913-922. SAS, 1988. SAS/STAT~ User's Guide, Release 6.03 Edition. SAS Institute Inc. Cary, NC, 1028 pp. Sommer, S.G. and Olesen, J.E., 199 I. Effects of dry matter content and temperature on ammonia loss from surface-applied cattle slurry. J Environ. Qual. 20: 679-683. Thompson, R.B., Pain, B.F. and Lockyer, D.R., 1990. Ammonia volatilization from cattle slurry following surface application to grassland. I. Influence of mechanical separation, changes in chemical composition during volatilization and the presence of the grass sward. Plant Soil, 125:109-117. Whitehead, D.C. and Lockyer, D.R., 1987. The influence of the concentration of gaseous ammonia on its uptake by the leaves of Italian ryegrass, with and without an adequate supply of nitrogen to the roots. J. Exp. Bot., 38: 818-827. Whitehead, D.C. and Raistrick, N., 1992. Effects of plant material on ammonia volatilization from simulated livestock urine applied to soil. Biol. Fert. Soils, 13: 92-95.

~) 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van Ittersum and S.C. van de Geijn (Editors)

245

Size and density fractionation of soil organic matter and the physical capacity of soils to protect organic matter Jan Hassink a'*, Andrew P. Whitmore, Jaromir Kubfit b aResearch Institutefor Agrobiology and Soil Fertility (AB-DLO), P.O. Box 129, 9750 AC Haren, The Netherlands bResearch Institute of Crop Production, Drnovskd 507, 16106 Praha 6, Ruzyne, Czech Republic Accepted 16 May 1997

Abstract Soil organic matter (SOM) has important chemical, physical and biological functions in the soil. It is difficult to predict the dynamics of SOM because it is very heterogeneous and because its behaviour is affected by soil texture. In this study we used a new size and density fractionation to isolate SOM fractions that differ in stability and we estimated the amount of SOM that can be preserved in different soils. An investigation was carded out into (1) how fast size and density fractions of soil organic matter respond to changes in C input, (2) whether the capacity of soils to preserve C by its association with clay and silt particles is limited and related to soil texture and (3) whether the long term dynamics of soil C can be described with a simple model that makes the assumption that the net rate of decomposition of soil C does not simply depend on soil texture, but on the degree to which the protective capacity of the soil is already occupied. Light and intermediate fractions of the macroorganic matter (> 150/zm) respond much faster to changes in C input than smaller size fractions. This shows that the light and intermediate macroorganic matter fractions can be used as early indicators of effects of soil management on changes in SOM. There was a close positive relationship between the proportion of particles <20 #m in a soil and the amount of C associated with this fraction in the top 10 cm of grassland soils. Arable sandy soils, which contained less C than corresponding grassland soils, had the same amounts of C associated with the fraction <20 #m, indicating that the amount of C that can become associated with this fraction had reached a maximum. The observed relationship: C in fraction <20 #m (g/kg soil)= 6.9 + 0.29x % particles <20 ~m can be used as a first estimation for the capacity of a soil to preserve C. The amount of C in macroorganic matter is controlled by soil management, while the amount of C protected by clay and silt particles is controlled mainly by soil texture. The simulations of the changes in C in soil without input of C or with additions of lucerne or chaff were excellent in both sandy and clay soils. The build-up of C in soils receiving farmyard manure (FYM) was not simulated so well. © 1997 Elsevier Science B.V.

Keywords: Soil organic matter; Size and density fractionation; Physical protection; Capacity; Modelling

1. Introduction Organic matter is a key component of soil which affects its physical, chemical and biological proper*Corresponding author. Tel.: +31 50 5337320; fax: +31 50 5337291; e-mail: [email protected]

ties. Organic matter improves soil structure, increases the water holding capacity and promotes biological transformations such as N-mineralization. Maintenance of sufficiently high levels of organic matter is a prerequisite for sustainable, high, production levels of crops. Agricultural management practices influence the amount of organic matter in soil and cause

Reprinted from the European Journal of Agronomy 7 (1997) 189-199

246

changes in the rate of soil organic matter turnover. It often takes years, however, before changes in agricultural management lead to detectable changes in the quantity and quality of soil organic matter (SOM). This makes it difficult to evaluate the effect of different management strategies on SOM dynamics. Several models have been constructed to predict organic matter dynamics. SOM is very heterogeneous and is composed of a series of pools from very active to passive (Schimel et al., 1985). To account for this, models of organic matter dynamics generally include compartments with a rapid tumover rate and a slower turnover rate. A major problem related to predicting SOM dynamics is that, except for the microbial biomass, the different compartments can not be determined directly by chemical or physical fractionation procedures (Paustian et al., 1992). Successful development of techniques for direct measurement of pool sizes would represent a major step towards appropriate verification of models (Bonde et al., 1992). Size and density fractionation show promise for physically dividing SOM into pools differing in compesition and biological function (Christensen, 1992). Size fractionation is based on the observation that sand-size organic matter (macroorganic matter; > 150/~m) is often more labile than organic C in the clay and silt size fractions (Tiessen and Stewart, 1983; Feller et al., 1991). Density fractionation is based on the observation that during humification parts of SOM become more associated with mineral particles and thus occur in particles of higher density (Barrios et al., 1996). Meijboom et al. (1995) have developed a new and simple density fractionation procedure using silica suspensions and recovered three density fractions in the macroorganic matter: the light fraction consisting of recognizable plant residues, the intermediate fraction of partly humified material and the heavy fraction of amorphous organic material. It was shown that the decomposition rates decrease in the order light, intermediate and heavy macroorganic matter, while the decay rates of C in the clay and silt size fractions were lowest (Hassink, 1995a). Size and density fractionation might enable us to identify labile fractions that respond much faster to changes in organic matter input than total SOM and can serve as sensitive indicators of changes in the SOM content (Janzen et al., 1992; Barrios et al., 1996). It has been recognized that soil texture and soil

structure have a predominant effect on organic matter decomposition. It is generally accepted that there is more physical protection in fine-textured soils than in coarse-textured soils (Jenkinson, 1988; Van Veen and Kuikman, 1990). As a consequence fine-textured soils have higher organic C contents than coarse textured soils when supplied with similar input of material (Jenkinson, 1988; Hassink, 1994). Considerable published evidence indicates that one of the principal factors responsible for physical protection of organic matter in soils is its ability to associate with clay and silt particles (Theng, 1979). Little is known, however, about the capacity of soils to protect organic matter physically. Hassink et al. (1997) suggested that the physical capacity of a soil to preserve organic matter is limited. Hassink (1996) defined the protective capacity as the maximum amount of C that can be associated with clay and silt particles in the soil. He suggested that the degree of saturation of the protective capacity of a soil would affect the preservation of newly added carbon in residues and not soil texture per se. Less of the applied C would be preserved in the soil when all protective sites were occupied than when sites were available to stabilize organic C. This explains why the preservation of applied C is directly related to soil texture when C is applied to fine-textured soils with low organic matter contents (e.g. Amato and Ladd, 1992) and why there is no correlation between soil texture and preservation of applied C in soil when fine-textured soils with higher organic matter contents are used (e.g. Gregorich et al., 1991). Based on the assumptions that (1) the preservation of applied C is controlled by the degree of saturation of the clay and silt size fractions with SOM (instead of soil texture per se) and (2) that the protection of SOM can be described kinetically in the same way as adsorption and desorption, a simple simulation model was developed that predicts the long term dynamics of SOM (Hassink and Whitmore, 1997). Short term changes in SOM may best be studied by focusing on the labile organic matter fractions, while long term dynamics of SOM are probably affected mainly by the capacity of a soil to protect organic C physically. The objectives of the present study are: (1) to determine how fast size and density fractions of SOM respond to changes in input of organic inputs and to identify measurable indicators most sensitive to changes in SOM; (2) to estimate the amount of C that

247

can be protected physically by association with clay and silt particles (<20 #m); and (3) to test whether a model that explicitly describes physical protection as a function of the capacity of clay and silt particles to hold organic matter, can simulate the long term dynamics of C in sandy and clay soils in the Netherlands that received either no organic C or annual applications of different organic materials and a clay-loam soil in the Czech Republic that received either no organic C or annual applications of manure.

2. Materials and methods

2.1. Experimental sites To determine how fast size and density fractions of SOM respond to changes in input of organic inputs and to identify indicators most sensitive to changes in SOM (objective 1), we defined and fractionated SOM from sand and clay soils that had received no input and samples of the same soils that had received annual applications of different organic materials. The soils were located in Haren, in the Northern part of the Netherlands. Characteristics of the soils at the begin-

ning of the experiment (1961) are given in Table 1. The soils have been kept bare since 1961 and received (1) no organic C and N or (2) 10 t C per ha/year by the application of either lucerne (2.5% N, 43.6% C), wheat chaff (0.8% N, 42.5% C) or farmyard manure (FYM; 2.9% N, 42.7% C). The residues were applied at the beginning of June and mixed through the top 25 cm of the soil. Soil samples of the top 25 cm were taken at 0.5, 3, 8, 15 and 25 years after the start of the experiment. To estimate the amount of C that can be protected physically by association with clay and silt particles (objective 2), we sampled the top 10 cm of grassland soils which had been under grass for at least 30 years and determined C in size and density fractions. Grassland soils generally have higher C contents than arable soils, because they receive more organic C each year and are not tilled (Jenkinson, 1988; Lugo and Brown, 1993). The grasslands were grazed by dairy cattle and received 400-500 kg fertilizer-N ha/year. Characteristics of the grassland soils are presented in Table 1. In the sandy soil of Tynaarlo we compared the distribution of C between organic matter fractions in soil from a grassland field with soil from an adjacent arable field which had been under a 4 year rotation of winter wheat, sugarbeet, barley and ware potatoes. In the

Table 1 Characteristics of the top 25 cm of the sandy and clay soils at the beginning of the long term experiment in 1961 (objective 1) and the top 10 cm of the grassland soils used for objective 2 Location

% particles

pH (K-CI)

C in size fraction (g/kg) <20 #m

20-150 #m

Total Soil C (g/kg)

< 2 #m

< 2 0 #m

> 150 #m

ND ND

17.0 65.6

7.4 7.0

7.2 15.2

2.7 6.6

0.4 0.4

10. l 22.2

3.0 24.1 25.8 51.1 1.9 8.4 45.8 2.6 1.0

5.8 36.5 42.6 76.0 2.8 13.3 65.9 5.0 3.5

4.9 4.8 5.8 5.4 5.0 5.0 7.1 4.4 5.4

9.7 21.2 19.9 31.0 1.4 11.3 20.3 13.0 3.6

5.8 a 11.9a 14.3 a 25.2 a 14.4 a 34.7 a 7.1 22.2 a 3.3

7.9 8.8 5.7 4.5 3.9 7.7 4.6 5.9 6.7

23.4 41.9 39.9 60.7 19.8 53.7 30.2 41. l 14.2

Objective 1 Haren Sand Clay

Objective 2 Achterberg Burum Zaltbommel 1 Zaltbommel 2 Heino Finsterwolde I Finsterwolde 2 Tynaarlo Cranendonck

ND, Not determined. aCalculated as total soil C -

C in fractions < 2 0 and > 150 #m.

248

sandy soil of Cranendonck we compared the distribution of C among organic matter fractions in a 30 year old grassland field with a field that had been under maize for at least 25 years. We also compared the distribution of organic matter in the top layer (0-20 cm) with deeper soil layers (30-40 and 60-80 cm deep) at this one location. To test our model that explicitly describes physical protection (objective 3), we simulated the long term dynamics of soil C in the sand and the clay soil receiving lucerne, chaff or FYM residues annually since 1961 (see above) and a clay loam from Ruzyne in the Czech Republic receiving 0, 80 or 160 t/ha FYM annually.

referred to as particle size fraction 20-150 ~tm. The suspension passing the 20 ~m sieve was placed at 4°C till it had sedimented (usually after 24 h) and the clear solution was sucked off; this fraction will be referred to as particle size fraction <20 ~m. It was assumed that the particle size fractions 20-150 #m and <20 ttm were derived primarily from microaggregates as microaggregates are resistant to wet-sieving (Tisdall and Oades, 1982). All fractions were dried at 40°C and analyzed for total C and N. In most of the grassland soils (indicated in Table 1), the amount of C in the 20-150 #m fraction was not determined, but calculated as the difference between total soil C and C in the fractions <20/~m and > 150/~m.

2.2. Size and density fractionation of soil organic matter

2.3. Modelling

Dried soil samples taken from the long term experiment on the sand and clay soil that had been stored were rewetted before fractionation. From the other soils field-moist samples were used for fractionation. We determined five fractions: the light, intermediate and heavy fractions of the macroorganic matter fraction (> 150 #m) and the size fractions 20-150/~m and <20 #m. For the density fractionation of the macroorganic matter, samples of 250 g were washed through two sieves (top sieve, mesh size 250 /~m; bottom sieve, 150 ttm). The soil was pushed through the top sieve, till the water passing the sieve became clear. In this way all macroaggregates were destroyed. The mineral fraction was discarded by decantation. After combining the organic fractions from both sieves, it was further fractionated in silica suspensions with a density of 1.13 and 1.37 g/cm 3 as described by Meijboom et al. (1995). The macroorganic matter was separated into three fractions: a light fraction (density <1.13 g/cm3); an intermediate fraction (density between 1.13 and 1.37 g/cm 3) and a heavy fraction (density > 1.37 g/cm3). To isolate the finer fractions, samples of 50 g were washed on three sieves (top sieve, mesh size 250 ttm; second sieve, 150 /zm; bottom sieve, 20 /zm). The suspension passing the bottom sieve of 20/~m was collected in a bucket. The soil was pushed through the top sieve again (destroying the macroaggregates), till the water passing the sieve became clear. The material accumulating on the 20 ttm sieve will be

Hassink and Whitmore (1997) have suggested a model for the dynamics of the physical protection of organic matter in soil based on a mechanism analogous to adsorption and desorption. Two SOM pools are distinguished: Protected (PoM) and Non-protected (NoM) organic matter. NoM is attacked by soil microorganisms relatively easily, whereas PoM cannot be decomposed unless it is first released by desorption. Assuming that the capacity of soil to protect organic matter (protective capacity; X) has a maximum value that cannot be exceeded, it is likely that the rate at which NoM becomes protected will slow down as this maximum is approached. The rate of desorption of POM on the other hand, is independent of X. Hassink and Whitmore (1997) found a good relationship between the clay content of a soil and X (maximum amount of SOM that can be protected by the soil). Based on this relationship we estimated X for three different soils: the sand and the clay soil receiving lucerne, chaff or FYM residues annually since 1961 (see above under objective 1) and a clay loam from Ruzyne in the Czech Republic. Composted manure was applied annually to this last soil at the rate of 0, 80 or 160 t/ha containing 0, 8 or 16 t/ha carbon (Nov~k and Apfelthaler, 1971). Additions stopped after 31 years of the experiment and the decrease in soil C was determined for another 5 years. Manure was added to the top 10 cm (80 t/ha) or top 20 cm (160 t/ha) of the experiment and at least in the first year therefore the concentrations of added organic matter were the same. However, because manure

249

Table 2 Amount of C in the light, intermediate and heavy macroorganicmatter fractions and the fractions 20-150 ttm and < 20/~m in the top 25 cm of the bare sandy and clay soil (Haren) that received no organic C and N 0.5, 3, 8, 15 and 25 years after the start of the experiment(g/kg soil) Fraction

Time (years) 0.5

3

8

15

25

Sandy soil Light Intermediate Heavy 20-150 tzm <20 #m

0.06 0.06 0.24 2.70 7.23

0.05 0.08 0.19 3.20 6.30

0.02 0.04 0.14 2.85 6.11

0.01 0.02 0.12 3.18 5.16

0.01 0.04 0.17 3.89 5.12

0.08 0.10 0.35 6.63 15.22

0.07 0.06 0.30 6.02 14.02

0.02 0.04 0.20 5.32 13.23

0.02 0.02 0.23 5.78 11.26

0.04 0.06 0.28 6.38 10.23

Clay soil Light Intermediate Heavy 20-150 #m <20 ttm

was added to the same depth of soil each year, the mass of soil in this depth decreased as the bulk density decreased, and the effective concentration of manure added (per mass of soil) increased each year. Both soils were sampled to 20 cm depth and this means that the apparent concentration of organic matter in the soil receiving 160 t/ha increased more quickly than in the soil receiving 80 t/ha. Corrections for the change in bulk density were made using a relationship derived by Whitmore et al. (1992). Lucerne, chaff and manure contained respectively 8%, 11.2% and 29% lignin (Haan de, 1977); Whitmore and Matus (1996) give a formula for reducing the rate of decomposition of crop residues in relation to their lignin content. The manure added to the Ruzyne experiment was assumed to contain 29% lignin also, but because it was composted we further assumed that 5% would be chemically resistant to attack and not decompose during the course of the experiment (e.g. Jenkinson et al., 1988).

2.4. Chemical analysis Total C in soil, in macroorganic matter fractions and in the particle size fraction 20-150 #m was defined as dichromate-oxidizable C according to Kurmies (Mebius, 1960). Total C in the particle size fraction < 2 0 #m was determined with a CHN autoanalyzer (Carlo Erba NA 1500).

2.5. Statistical analysis The relationships between the percentage of soil particles < 2 0 #m and total soil C and C associated with the particle size fractions < 2 0 #m, 2 0 - 1 5 0 / z m and > 150 #m were analyzed with correlation and regression techniques (Genstat, 1987).

3. Results The identification of soil organic matter fractions that are sensitive to changes in input of organic materials (sandy and clay soil of objective 1) At all sampling times, more than 90% of all soil carbon was present in the fractions 20-150 #m and < 2 0 #m in the treatments where no organic material was applied to the soil (Table 2tablehere>). The amount of C in the heavy macroorganic matter fraction and the 20-150 #m fractions did not change significantly with time, while C in the other fractions decreased with time. The relative changes in the amounts of C in size and density fractions due to the application of chaff, lucerne and FYM are expressed as the amounts of C in the size and density fractions in the chaff, lucerne and FYM treatments, divided by the corresponding amounts in the treatments where no organic inputs were applied 0.5, 3, 8, 15 and 25 years after the start of the treatments. There were no

250

L

o18 ~16

-=- I

-o- H

~

20-150/Jm --=.- < 20/Jm

"514 13L .c: 12

O~o 8

0

0

5

10

15

20

25

The relative changes were larger for the FYM treatment than for the lucerne and chaff treatments. The application of FYM increased the amount of C in the heavy macroorganic matter fraction relatively more than the application of lucerne or chaff (Figs. 1 and 2).

3.1. Relationship between soil texture and C in organic matter fractions in grassland soils (soils of objective 2)

Time (years) Fig. I. The amounts of C in the light, intermediate and heavy macroorganic matter fractions and the 20--150 ~tm and <20/~m fractions in the chaff treatment divided by the corresponding amounts of C in the fractions in the no input treatment 0.5, 3, 8, 15 and 25 years after the start of the experiment. Average of the sandy and clay soil.

differences in the relative changes of C in the fractions between the sand and clay soils and between the lucerne and chaff treatments. The amounts of C in each of the fractions derived from the chaff and FYM treatments divided by the corresponding amounts in the treatments receiving no input were averaged for the sand and clay soils (Figs. 1 and 2). Generally, when organic materials were added to the clay and sand soil pronounced changes took place quickly in the amounts of C in the light and intermediate fractions of the macroorganic matter pool. The heavy fraction of the macroorganic matter pool responded more slowly, while changes in the fractions 20-150 ~m and the <20 ~tm were relatively slow and small (Fig. 1 and 2). Within a few years of application of organic materials, changes in the light and intermediate fractions of the macroorganic matter fraction could be determined easily. The relative increases in the light and intermediate fractions of the macroorganic matter pool reached a maximum within 8-15 years after the start of input of organic material. At this maximum, the amounts of C in the light and intermediate macroorganic matter fractions were 6-23 times greater in the treatments where organic material was applied than in the treatment where there was no input of organic material. The amount of C in the heavy macroorganic matter fraction and the fractions 20-150 ttm and the <20 ttm fraction continued to increase relative to the nil input treatments. After 25 years, the relative increases ranged from 4 to 19 for the heavy macroorganic matter fraction and from 1.3 to 4.4 for the 20-150/~m and the <20/~m fractions.

We investigated the relationships between soil texture and organic C in different organic matter fractions. For the fraction > 150 #m, we present the sum only of the light, intermediate and heavy fractions, because the density fractions did not give any additional information with respect to the capacity of soils to protect C physically. There was a highly significant correlation (r = 0.91) between the clay and silt content of grassland soils and the amount of C associated with this fraction (Table 1; Fig. 3; C associated with the fraction <20 ~tm = 6.9 + 0.29x % particles <20/~m). The amounts of C in the fractions 20-150 ~tm and > 150 ttm did not correlate significantly with the clay and silt fraction and varied considerably between soils with similar clay and silt content (Table 1). C in size fractions in soils with similar textures but differing in organic matter input (Tynaarlo and Cranendonck soil of objective 2 and the sandy and clay soil of objective 1) In spite of the fact that the arable field from Tynaarlo, the maize field from Cranendonck and the 25 o

L

+

I

--e- H

~

20-150/Jm

-a~

< 20 pm

20

u. 5 I

5

10 15 Time (years)

20

25

Fig. 2. The amounts of C in the light, intermediate and heavy macroorganic matter fractions and the 20-150 #m and <20 ttm fractions in the FYM treatment divided by the corresponding amounts of C in the fractions in the no input treatment 0.5, 3, 8, 15 and 25 years after the start of the experiment. Average of the sandy and clay soil.

251

Table 3 Amounts of C in different size fractions and total soil C in the top 10 cm of the grassland and arable soil in Tynaarlo and the top 20 cm and the soil layers at 30-40 and 60-80 cm depth in the grassland and maize field in Cranendonck (g/kg) Location

C in fraction (treatment)

Total soil C

<20 #m

20-150 #m

> 150 #m

11.9 I 1.8

9.6 5.4

11.6 2.8

36.3 23.8

3.6 3.4 3.4 3.6 3.5 3.1

4.7 4.0 3.6 3.1 2.1 1.7

6.7 3.5 2.8 2.6 1.2 0.9

15.9 12. l 9.7 8.8 7. l 7.2

Tynaarlo Grassland Arable

Cranendonck 0-20 cm grass 30-40 cm grass 60-80 cm grass 0-20 cm maize 30-40 cm maize 60-80 cm maize

fractions was generally close to the total amount of soil C (Table 3). The application of chaff, lucerne and FYM to the sand and clay soil for 25 years (long term experiment of objective l) resulted in a considerable increase in total soil C in comparison with the treatment where n o organic residues were applied (Table 4). In the sandy soil, the increase was concentrated in the 20-150 #m fraction, while in the clay soil most of the increase took place in the < 2 0 #m fraction (Table 4). In the clay soil, the sum of C in the fractions was close to the total amount of soil C; in the sandy soil, the sum of C in the fractions was 40% higher than total soil C for the no input treatment, while in the other treatments the sum of C in the fractions was again close to total soil C (Table 4). No explanation can be offered for the high recovery in the sand soil without input.

3.2. Modelling (sandy and clay soil of objective 1 and Czech soil at Ruzyne) deeper layers of the grassland and maize field from Cranendonck had much lower total amounts of soil C than the corresponding top layers of the grassland fields, the amounts of C associated with the clay and silt fraction were not less (Table 3). The amounts of C in the fractions 20-150 #m and > 150 #m were less in the arable field from Tynaarlo and the maize field from Cranendonck than in the corresponding grassland sites, and decreased with increasing depth at Cranendonck (Table 3). The sum of C in the different

The simulations of the increase in total soil C in soil with addition of lucerne or chaff, or the decrease in C in soil in the treatments without addition were excellent (Figs. 4 and 5) in both the sand and clay soils. The simulated build-up of organic matter in the soils receiving FYM also agreed well with the measurements during the first 13 years. During the second half of the experiment, however, a sharp increase in soil C was measured, which could not be simulated

Table 4 Amounts of C in different size fractions after 25 years of no input or annual applications of chaff, lucerne and FYM (g/kg) and the percentage of increase in soil C (in comparison with the no input treatment) that is associated with the fraction < 20 #m for the sand and clay soil in Haren <20 tLm

20-150 •m

> 150 #m

Increase in fraction < 20 t~m (% of total increase)

Total soil C

5.1 7.5 8.5 12.4

3.9 9.4 10.1 21.5

0.2 1.6 1. I 5.5

25.5 32.0 24.1

6.6 16.8 16.5 42.5

10.2 15.4 16.5 26.0

6.4 9.7 10.0 19.2

0.4 1.6 0.9 4.3

53.2 60.8 48.6

14.7 25.9 25.0 52.7

Sandy soil No input Chaff Lucerne FYM

Clay soil No input Chaff Lucerne FYM

252

(Figs. 4 and 5). In all treatments receiving additions, an odd decline in soil C during the last 3 years of the experiment was measured. The build-up and decline of organic C at Ruzyne was simulated well (Fig. 6) and it is particularly interesting to see that simulating the change in bulk density allowed us to reproduce the extra increase in concentration of organic C (in g per C/kg soil) in the plot receiving 160 t/ha 20/cm each year despite the fact that the addition on this plot was nominally the same as that on the plot receiving 80 t/ ha 10/cm.

•

•

0

c

~1

. - ••

........... "'.2.'.2.......................... .... ~ ........ -i.-..-..-.-.-..-.~...............~.....o•

II,."-..~"............A ~'.'~-':m .......• ........•

10

o

0

i

1

50

100

I

I

I

I

150

200

250

300

350

Time (months)

4. Discussion The first objective was to test how fast different size and density fractions respond to input of organic material. We found that the light and intermediate fractions of the macroorganic matter fraction were much more sensitive to input of organic residues than the other fractions. This is in line with the results of studies in other agricultural systems (Janzen et al., 1992; Barrios et al., 1996). In a previous study Hassink (1995b) found that the C:N ratio as well as the amounts of the light and intermediate macroorganic matter fractions were affected more by the long term differences in residue input than other fractions. The light and intermediate fractions of the macroorganic matter fraction can be used as sensitive indicators of changes in SOM. This enables us to detect effects of

0u}

-r

o~ v

o

35

Grasslands

~

A r a b l e Tyn, Cr

u

FYM l o n g - t e r m

~ 3o

~ 2s o

~ 20 V

._~ 15

~_ ~o .c_ 0 0 0

i

i

i

l

20

40 Particles < 20 p m (%)

60

80

Fig. 3. Relationship between C in the particle size fraction <20 #m (clay and silt in g/kg soil) and the percentage of soil particles <20 #m in the top 10 cm of grassland soils, the top 10 cm of an arable field in Tynaarlo, the top 20 cm of a maize field in Cranendonck and the top 25 cm of the FYM treatments of the long term experiment in the sandy and clay soil.

Fig. 4. The dynamics of organic carbon in the sandy soil with no input (&), and annual inputs of chaff (O), lucerne (m) and FYM (O) during a 25 year period. Lines are simulations, points are measured values.

agricultural practices on SOM dynamics within a much shorter period. The second objective was to test whether the amount of C that can become associated with clay and silt particles is limited and to quantify the relationship between soil texture and the maximum capacity of a soil for C to be associated with clay and silt particles. We observed a close positive correlation between the percentage of soil particles <20 ttm and the amount of C associated with these particles. We found that although the top layers of the sandy grassland soils in Tynaarlo and Cranendonck contained more C than the corresponding arable sites and the deeper soil layers, the amounts of C associated with the clay and silt fraction were not different. The increases in soil C were only observed in the larger size fractions. This suggests that the amount of C that can become associated with the clay and silt fraction had reached a maximum in these sandy soils. The hypothesis that soils have a limited capacity to protect C by their association with clay and silt particles is also confirmed by the observation that the relationship between soil texture and the amount of clay and silt associated C in the sandy and clay soil that had received FYM for 25 years was the same as for the grassland soils (Fig. 3). The observation that total soil C did not increase any further with FYM application (both in the sandy and clay soil; Figs. 4 and 5) also suggests that the protective capacity of the soils was saturated after 25 years of FYM application. We assume that the observed relationship between the percentage of soil particles <20/~m and the amount

253

•

•

•

Q

/

-Q

• .,."e •II " m I

•

................. o© . . . . . . . . . . . . . ii"......... r ..... . ii ......

111. . . . . . . . . . . . . . . . . .

i'..'..'.. . . . . . . . . . . . . . . . . . . . . . . " . . . . . . . . .

.L

10 / 0

I 50

I 100

I 150

I 200

Time

(months)

="

l 250

•

A--'&

I 300

Fig. 5. The dynamics of organic carbon in the clay soil with no input (A), and annual inputs of chaff ((3), lucerne ( 1 ) and FYM (O) during a 25 year period. Lines are simulations, points are measured values.

of C associated with this fraction can be used as a first estimation for the capacity of a soil to preserve C. Not only soil texture but also clay type may affect the capacity of soils to protect organic C physically. Specific surfaces of clays vary from 2-4 m2/g for quartz (Wilding et al., 1977), 6-39 m2/g for kaolinite (Dixon, 1977), 50-100 m2/g for illite to 800 m2/g for smectite and vermiculite (Robert and Chenu, 1992). Soil dominated by clays with a high specific surface area are expected to adsorp more humic substances than soils dominated by clays with a low specific surface area (Tate and Theng, 1980). Although the dominant clay minerals in Dutch sandy soils are quartz and kaolinite (clay minerals with a low specific surface) and the dominant clay minerals in fine-textured Dutch soils are illite and smectite (clay minerals with a higher specific surface) (Favejee, 1949; Marel, 1949; Breeuwsma, 1990), the clay and silt particles in the sandy soils had higher C contents than the clay and silt particles in fine-textured soils (Hassink et al., 1997). In most cases a decrease in pH favours the bonding of organics to clay particles (Varadachari et al., 1994). This might explain why the amount of clay and silt associated C was greater for the Tynaado soil (pH 4.4) than for the Cranendonck soil (pH 5.4; Table 3). The results suggest that exposure of clay particles to free organic matter and aggregation plays an important role in determining the actual protective capacity (X) in soil. Moreover, most clays with a low specific surface area have a relatively high ratio of external to internal surface area, whereas for clays with a high specific surface area the opposite is true (Robert and

Chenu, 1992). Possibly, many organic compounds do not penetrate the interlayer space and adsorb only on the external surfaces (Tate and Theng, 1980). The third objective was to test whether a simple model that explicitly describes physical protection as a function of the capacity of clay and silt particles to hold organic matter could simulate the dynamics of soil C in different soils with different inputs assuming that the protective capacity (X) was directly related to soil texture as described by Hassink and Whitmore (1997). Except for the second half of the FYM experiment in the sand and clay soil, the results of the model were in line with the experimental values. The unexplained increase in soil C in the FYM experiment in the sand and clay soil might be due to the presence of chemically resistant C in FYM that accumulated during the course of the experiment, or due to the fact that with the application of FYM, soil particles are applied to the soil, possibly leading to a gradual increase in the protective capacity of the soil. In the coarse sandy soils of Tynaarlo and Cranendonck, increases in soil C under grassland were concentrated in the sand size fraction; in the heavier textured sandy soil of the long term experiment, the accumulation of C in the chaff, lucerne and FYM treatments mostly took place in the 20-150 #m fraction in contrast to the treatment with no application, while in the clay soil of the long term experiment the accumulation of C was concentrated in the <20 #m fraction. The percentage of C accumulating in the <20 /~m fraction was less in the FYM treatment than in the chaff and lucerne treatments (Table 4).

.-.50

• • ~ . , , , - •

•'~

,o

• ...

30

,I/

C

7

"-

/ . . O

,,~20

•............:.._

,.-.;

•.

,.,'"'""

•

" ............

• •

•

".,,... •

"""...... • "

• •

........• "~"

. . . . . . . . . . . . . . . . o.O.o.t.o ~ °.o.~.o.~O.O.°o.%O o _ o , • • • •

10

" I ........

0 0

i IO0

i 2OO

I 3OO

I 400

5OO

Time (months) Fig. 6. The dynamics of organic carbon in the clay-loam soil in the Czech Republic receiving 0 (O), 80 (n) or 160 (A) t/ha FYM annually for 31 years and no addition during a subsequent 5 years. Lines are simulations, points are measured values.

254 This may be due to the fact that the < 2 0 #m fraction in the FYM treatments has reached its capacity. The results strengthen the assumption that the degree of saturation of the clay and silt fraction determines the physical protection of applied C and that with this description of physical protection we are able to simulate the long term dynamics of soil C under different conditions. We conclude that the light and intermediate fractions of the macroorganic matter are most sensitive to changes in C input and that the amount of macroorganic matter depends on soil management and is independent of soil texture, while the amount of clay and silt associated C is strongly affected by soil texture and only affected by management as long as the protective capacity of this fraction is not yet saturated. Our results are in agreement with results of Quiroga et al. (1996) who found for Argentinian soils that the amount of C associated with particles < 5 0 #m was only affected by soil texture and not by soil management, while coarse organic matter was strongly affected by soil management, but not by soil texture. We can conclude that the light and intermediate fractions of the macroorganic matter pool can be used as early indicators of effects of changes in soil management on the amount and quality of SOM. The capacity of soils to preserve SOM can be estimated from the clay and silt content of the soil. The observation that the capacity of soils to preserve SOM is limited has important consequences for the estimation of the long term behaviour of SOM and the estimation of the amounts of SOM that can be stored in soils in a sustainable way.

References Amato, M. and Ladd, J.N., 1992. Decomposition of 14C-labelled glucose and legume materials in soil: properties influencing the accumulation of organic residue C and microbial biomass C. Soil Biol. Biochem., 24: 455-464. Barrios, E., Buresh, R.J. and Sprent, J.I., 1996. Organic matter in soil particle size and density fractions from maize and legume cropping systems. Soil Biol. Biochem., 28: 185-193. Bonde, T.A., Christensen, B.T. and Cerri, C.C., 1992. Dynamics of soil organic matter as reflected by natural 13C abundance in particle size fractions of forested and cultivated oxisols. Soil Biol. Biochem., 24: 275-277. Breeuwsma, A., 1990. Mineralogical composition of Dutch soils. In: W.P. Locher and H. de Bakker (Editors), Soil Science of the

Netherlands. Malmberg, Den Bosch, The Netherlands, pp. 103107 (in Dutch). Christensen, B.T., 1992. Physical fractionation of soil and organic matter in primary particle size and density separates. Adv. Soil Sci., 20: 1-89. Dixon, J.B., 1977. Kaolinite and serpentine group minerals. In: R.C. Dinauer (Editor), Minerals in Soil Environments. SSSA, Madison, WI, pp. 357-403. Favejee, J.Ch.L., 1949.The mineralogical composition of the clay fraction of Dutch soils. Landbouwk. Tijdschr., 61: 167-171 (in Dutch). Feller, C., Fritsch, E., Poss, R. and Valentin, C., 1991. Effet de la texture sur le stockage et la dynamique des matieres organiques dans quelques soil ferrugineux et ferrallitiques. Cah. O.R.S.T.O.M. Ser. Pedol., 26: 25-36. Genstat Manual, 1987. A General Statistical Program. Clarendon, Oxford. Gregorich, E.G., Voroney, R.P. and Kachanoski, R.G., 1991. Turnover of carbon through the microbial biomass in soils with different textures. Soil Biol. Biochem., 23: 799-805. de Haan, S., 1977. Humus, its formation, its relation with the mineral part of the soil, and its significance for soil productivity. In: Soil Organic Matter Studies, Vol. 1. International Atomic Energy Agency, Vienna. pp. 21-30. Hassink, J., 1994. Effects of soil texture and grassland management on soil organic C and N and rates of C and N mineralization. Soil Biol. Biochem., 26: 1221-1231. Hassink, J., 1995a. Decomposition rate constants of size and density fractions of soil organic matter. Soil Sci. Soc. Am. J., 59: 1631-1635. Hassink, J., 1995b. Density fractions of macroorganic matter and microbial biomass as predictors of C and N mineralization. Soil Biol. Biochem., 27: 1099-1108. Hassink, J., 1996. Preservation of plant residues in soils differing in unsaturated protective capacity. Soil Sci. Soc. Am. J., 60: 487491. Hassink, J. and Whitmore, A.P., 1997. A model of the physical protection of organic matter in soils. Soil Sci. Soc. Am. J., 61: 131-139. Hassink, J., Matus, F.J., Chenu, C. and Dalenberg, J.W., 1997. Interactions between soil biota, soil organic matter and soil structure. Adv. Agroecol., in press. Janzen, H.H., Campbell, C.A., Brandt, S.A., Lafond, G.P. and Townley-Smith, L., 1992. Light-fraction organic matter in soils from long term crop rotations. Soil Sci. Soc. Am. J., 56: 1799-1806. Jenkinson, D.S., 1988. Soil organic matter and its dynamics. In: A. Wild (Editor), Russel's Soil Conditions and Plant Growth, 1l th edition. Longman, New York, pp. 564-607. Jenkinson, D.S., Hart, P.B.S., Rayner, J.H. and Parry, L.C., 1988. Modelling the turnover of organic matter in long-term experiments at Rothamsted. Intecol Bull., 15: 1-8. Lugo, A.E. and Brown, S., 1993. Management of tropical soils as sinks or sources of atmospheric carbon. Plant Soil, 149: 2741. van der Marel, H.W., 1949. Mineralogical composition of a heath podzol profile. Soil Sci., 67: 193-207.

255

Mebius, L.J., 1960. A rapid method for the determination of organic carbon in soil. Anal. Chim. Acta, 22: 120-124. Meijboom, F.W., Hassink, J. and Van Noordwijk, M., 1995. Density fractionation of soil macroorganic matter using silica suspensions. Soil Biol. Biochem., 27:1109-1111. Nov~, B. and Apfelthaler, R., 197 I. Effect of manuring and fertilization on carbon and nitrogen transformations in soil. In: Studies about Humus. Transactions of the International Symposium Humus et Planta V. Prague, pp. 73-76. Paustian, K., Parton, W.J. and Persson, J., 1992. Modelling soil organic matter in organic-amended and nitrogen fertilized long-term plots. Soil Sci. Soc. Am. J., 56: 476-488. Quiroga, A.R., Buschaiazzo, D.E. and Peinemann, N., 1996. Soil organic matter particle size fractions in soils of the semiarid argentianian Pampas. Soil Sci., 161: 104-107. Robert, M. and Chenu, C., 1992. Interactions between soil minerals and microorganisms. In: G. Stotzky and J.M. Bollag (Editors), Soil Biochemistry, Vol. 7. Marcel Dekker, New York, pp. 307404. Schimel, D.S., Coleman, D.C. and Horton, K.A., 1985. Soil organic matter dynamics in paired rangeland and cropland toposequences in North Dakota. Geoderma, 36: 201-214. Tate, K.R. and Theng, B.K.G., 1980. Organic matter and its interactions with inorganic soil constituents. In: G.K.G. Theng (Editor), Soil with a Variable Charge. N. Z. Soc. Soil Sci., Lower Hutt, New Zealand, pp. 225-249.

Theng, B.K.G., 1979. Formation and Properties of Clay-Polymer Complexes. Elsevier, Amsterdam. Tiessen, H. and Stewart, J.W.B., 1983. Particle-size fractions and their use in studies of soil organic matter. II. Cultivation effects on organic matter composition in size. Soil Sci. Soc. Am. J., 47: 509-514. Tisdall, J.M. and Oades, J.M., 1982. Organic matter and water stable aggregates in soils. J. Soil Sci., 13: 141-163. Varadachari, Ch., Mondal A.H., Nayak, D.C. and Ghosh, K., 1994. Clay-humus complexation: effect of pH and the nature of bonding. Soil Biol. Biochem., 26:1145-1149. Van Veen, J.A. and Kuikman, P.J., 1990. Soil structural aspects of decomposition of organic matter by microorganisms. Biogeochemistry, 11: 213-233. Whitmore, A.P. and Matus, F.J., 1996. The decomposition of wheat and clover residues in soil measurements and modelling. Proc. 8th Nitrogen Workshop on Soils, Ghent. Whitmore, A.P., Bradbury, N.J. and Johnson, P.A., 1992. The potential contribution of ploughed grassland to nitrate leaching. Agric. Ecosys. Environ., 39: 221-233. Wilding, L.P., Smeck, N.E. and Drees, L.R., 1977. Silica in soils: quartz, crostobalite, tridymite and opal. In: D.C. Dinauer (Editor), Minerals in Soil Environments. SSSA, Madison, Wl, pp. 471-552.

This Page Intentionally Left Blank

© 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geo'n (Editors)

257

Characterization of dissolved organic carbon in cleared forest soils converted to maize cultivation L. Delprat a'*, P. Chassin a, M. Lin6res a, C. Jambert b alNRA, Department of Agronomy, BP81, 33 883 Villenave, France bUniversity of Paul Sabatier, Laboratoire d'A6rologie, 118 route de Narbonne, 31 062 Toulouse, France Accepted 9 April 1997

Abstract

The impact of cultivation on the nature of dissolved organic carbon (DOC) has been studied in order to understand differences in denitrification rates observed in situ. The quantitative aspect of DOC was studied by means of porous-cup vacuum samplers and water extraction. Qualitative characterization was achieved using 8~3C analysis and tangential ultrafiltration with nominal cut-offs at 10000 and 100000 Da. During the first stage, cultivation intensified mineralization of the initial organic matter. This led to an increase of 2- to 5-fold in the quantity of DOC, with an important spatial variability. This phenomenon generated mainly medium molecules (MM) and large molecules (LM). These molecules were supposed to be fulvic and humic acids, and large colloids or humic metal complexes. During the second stage, once the original organic matter is stabilized, DOC concentrations decreased with time of cultivation. Both soil organic carbon and DOC were enriched in carbon originating from maize with time of cultivation. DOC was mainly composed of small molecules (SM) and MM. These molecules might be organic or amino acids, polysaccharides, or fulvic acids. From a methodological point of view, it is noteworthy that DOC characteristics determined following extraction by means of porous cups differed from those for water extraction. © 1997 Elsevier Science B.V. Keywords: Dissolved organic carbon; Extraction methods; Cultivation impact; t~13C Maize

1. Introduction

In Southwest France, most cleared forest lands have been converted into continuous maize cropping. The loss of nitrogen upon conversion to intensive maize cultivation has been shown to be significant (Jambert, 1995): N2 and N20 fluxes reached 45 kg N/ha per year after 7 years of cultivation and 8 kg N/ha per year after

*Corresponding author. Tel.: +33 5 56843042; fax: +33 5 56843054 e-mail: lineres@b°rdeaux'inra'fr

3 years. N20 fluxes represent 33 and 25%, respectively. Under forest, there are no fluxes even after nitrate addition. These emissions mainly occurred during the summer period after fertilization input. In autumn, CH4 emissions have been measured. While N20 production occurred near the surface (around 20 cm), CH4 production was observed between 50 and 80 cm. N20 is mainly produced during microbiological denitrification and to a lesser extent during nitrification (Firestone and Davidson, 1989). Biological denitrification is controlled by the availability of mineral

Reprinted from the European Journal of Agronomy 7 (1997) 201-210

258

nitrogen, oxygen and organic substrate. In our case, the limiting factor seems to be the nature of the organic substrate, given that nitrates are in excess and the type of soil is similar. The correlation between denitrification rate and the nature of organic matter has been widely studied. This correlation is improved when instead of the total organic carbon of soil, the easily mineralizable carbon or the soluble organic matter are considered (Burford and Bremner, 1975; Germon et al., 1983; Bijay-Singh et al., 1988; Beauchamp et al., 1989; Katz et al., 1985; McCarty and Bremner, 1992). A correlation between water soluble carbon and easily mineralizable carbon has also been shown (Burford and Bremner, 1975; McCarty and Bremner, 1992). On the other hand, the addition of plant residues (maize, wheat straw) or water extracts of these residues increased the denitrification potential of soils (McCarty and Bremner, 1992). In order to understand differences in denitrification rates observed in the field, we have studied the dissolved organic carbon (DOC) extracted in situ with porous cups or by water extraction. We hypothesized that cultivation modifies the nature of DOC and that the soluble organic carbon originating from maize preferentially contributes to denitrification. The obTable 1 Chemical characteristics of soils for each treatment: total organic carbon (Anne method) and nitrogen (Kjeldahl method), pH and cation exchange capacity (CEC)

TREAT0 0-25 25-50 50--75 TREAT92 0-25 25-50 50-75 TREAT88 0-25 25-50 50-75 TREAT69 0-25 25-50 50-75

Organic C (g/kg)

Organic N (g/kg)

pHH2o

CEC (cmol/kg)

44.4 20.8 6.9

1.46 0.85 0.38

4.01 4.27 4.40

4.77 1.39 0.76

28.3 23.2 12.2

1.02 0.85 0.50

4.78 5.13 4.89

3.19 3.24 0.89

22.1 11.7 7.2

0.97 0.56 0.37

5.38 5.26 5.21

3.55 1.20 0.94

18.2 16.4 5.8

0.91 0.87 0.34

5.81 5.81 5.70

2.15 2.48 1.64

jectives of this study were: (1) to identify the quantitative and qualitative modifications of DOC after clear cutting of a pine forest, and (2) to analyse the evolution of the nature of DOC after many years of cultivation.

2. Materials and methods

2.1. Site and soil description The field site is located in the 'landes de Gascogne', in Southwest France. This area (1.2 106 ha) is characterized by sandy acid soil with high quantity of organic matter: it belongs to the podzosol family. Precipitation averages 700 mm/year, and annual temperature 12°C. Four sites were studied: a pine forest (Pinus pinaster Ait.) (TREAT0) and three fields of maize (Zea mays L.). The first has been cultivated since 1969 (TREAT69), the second since 1988 (TREAT88) and the latter since 1992 (TREAT92). Soils of each treatment were characterized (Table 1).

2.2. Isolation of dissolved organic carbon DOC was extracted by means of two different methods: porous cup vacuum samplers and a water extraction on fresh soil. Porous cup vacuum samplers were installed at 20, 40 and 60 cm depth (6 repetitions/depth). The soil solution was collected applying 0.6 bars vacuum during 24 h. Soil samples for water extraction were taken at three different depths: 0-25, 25-50 and 50-75 cm (6 replicates/depth). Extraction was conducted using 50 g equivalent dry soil and 100 ml of distilled water. The soil/water suspensions were shaken for 30 min and centrifuged at 10 000 × g for 10 min. Both solutions were filtered at 0.45/zm. DOC was quantified with a Dohrmann DC 180. The organic carbon was oxidized with potassium persulfate under UV activation, and the emitted CO2 detected by infrared spectroscopy. The experimental schedule of sample analysis is reported in Table 2.

2.3. Characterization of dissolved organic carbon Molecular weight fractionation was performed by tangential ultrafiltration (Amicon TCF10) with nominal cut-offs at 10 000 and 100 000 Da. Concentra-

259

Table 2 Experimental schedule

DOC porous cup extraction

DOC water extraction

TREAT0

TREAT92

TREAT88

February, March and April 1993

February, March, April and August 1993 March and August 1994 March and August 1995 March 1995

February, March and April 1993

March 1995 March 1995

March 1995 March 1995

TREAT69

March 1996 March 1996

tions, pH and volumes used were adjusted to the same values for all samples. Carbon isotope ratios of soil organic matter and DOC were measured on a system composed of an elemental CHN autoanalyzer (Carlo Erba NA 1500) coupled to a mass spectrometer (Finnigan Mat Delta S). Carbon 13 natural abundance is expressed in tSI3C (%o) by reference to the international standard PDB using NBS 19 (precision 0.3%0):

2.4. Statistical analysis

813C(%o)

3.1. Quantitative variation of dissolved organic carbon

- -

(Rsample/Rstandard -- 1) × 1000

Data were tested for homogeneity of group variances using Bartlett's test. They were square root transformed if necessary. The data were analysed by analysis of variance (Dagnelie, 1973).

3. R e s u l t s

R - 13C / ] 2 C

3.1.1. Porous cup extraction The organic matter originating from forest has a 8o equal to-27.6°/oo whereas the maize organic matter has a c5] equal to -12%o. After a given time of cultivation, measurement of ~3C abundance in ~5units provides the means to calculate the fraction of organic matter from each source (Balesdent, 1991).

The pattern of DOC concentrations extracted with porous cups over a short term period (February, March and April 1993) is shown in Table 3. DOC concentrations of TREAT0 and 88 fluctuated in the same range ( 1 6 - 7 4 mg/l). Vertical variation of concentrations was not significant (*P < 0.05) on these

Table 3 Descriptive statistics for dissolved organic carbon concentrations extracted by means of porous-cup vacuum samplers, over a 3 month period (February, March and April 1993) Mean (mg/l) TREAT0 20 cm 40cm 60cm TREAT92 20 cm 40 cm 60 cm TREAT88 20 cm 40 cm 60 cm

SE (mg/l)

CV (%)

Median (mg/I)

Minimum (mg/i)

Maximum (mg/l)

9 11 11

44 36 34

7 14 8

17 39 25

43 36 35

32 21 20

54 74 51

12 13 17

200 56 62

93 10 17

47 18 27

180 56 59

68 39 35

342 74 99

15 11 14

46 42 29

11 10 11

23 24 39

44 44 27

29 28 16

69 58 55

SE, standard error; CV, coefficient of variation.

260 Table 4 Arithmetic mean ( + SE) of dissolved organic carbon extracted by means of porous-cup vacuum samplers, and soil water contents on TREAT92 March 1993 20 cm DOC (mg/l) H20 (%) 40 cm DOC (mg/l) H20 (%) 60 cm DOC (mg/1) H20 (%)

March 1994

August 1994

March 1995

August 1995

180 + 97 -

84 + 36 19

102 + 38 14

79 + 25 16

145:1:97 10

56 + 5

90 + 15 17

101 + 55 10

52 + 7 14

170 + 120 11

54 + 6

100 + 32 10

76 + 7 9

68 + 22 11

70 + 36 9

-

-

two treatments. T e m p o r a l variation over this short term period was not significant (*P < 0.05). T h e s e concentration values were in the range observed in other studies for porous cup extraction: i.e. 2 0 - 5 0 mg/l at 20 or 30 c m depth and 1 0 - 2 0 mg/l at 60, 90 or 200 c m (Vance and David, 1989; H u g h e s et al., 1990; G u g g e n b e r g e r and Zech, 1993; W e s e m a e l and Verstraten, 1993). M c L a u g h l i n et al. (1994) reported average values of 32 + 16 mg/l at 30 c m depth and 18 + 12 mg/l at 2 m with no temporal trends over a 4 m o n t h period. After the first year of cultivation, D O C concentrations on T R E A T 9 2 were higher than those o b s e r v e d on T R E A T 0 and 88, especially at 20 c m

depth (Table 3). There was a 2- to 6-fold increase c o m p a r e d to T R E A T 0 , with higher spatial variability. L o n g term evolution (from M a r c h 1993 to A u g u s t 1995) of D O C concentrations on T R E A T 9 2 was measured (Table 4). All depths were significantly perturbed after 2 years of cultivation: m e a n concentrations fluctuated b e t w e e n 50 and 180 mg/l with 1 4 - 7 0 % variability. Concentrations and variabilities were higher in August than in March when biological activity is important. Hughes et al. (1990) also pointed out a 3-fold increase of D O C concentrations in soil solution after afforestation. D O C concentrations on T R E A T 6 9 were c o m p a r e d

Table 5 Descriptive statistics for dissolved organic carbon extracted by means of porous-cup vacuum samplers

TREAT0a 20 cm 40 cm 60 cm TREAT92a 20 cm 40 cm 60 cm TREAT88a 20 cm 40 cm 60 cm TREAT69b 20 cm 40 cm 60 cm

Maximum

Mean (mg/1)

SE (mg/l)

CV (%)

Median (mg/1)

Minimum (mg/l)

5 5 5

43 43 63

20 9 14

46 20 23

42 39 66

17 34 40

56 53 77

7 6 9

79 52 68

25 7 22

32 13 33

86 50 63

38 45 36

104 61 111

8

30

6

21

30

21

38

6 6 6

36 16 19

7 5 12

20 34 64

37 13 15

24 11 7

43 25 35

a, March 1995; b, March 1996; SE, standard error; CV, coefficient of variation.

(m~l)

261

Table 6 Descriptive statistics for dissolved organic carbon extracted from fresh soil with distilled water

TREAT0a 0-25 cm 25-50 cm 50-75 cm TREAT92a 0-25 cm 25-50 cm 50-75 cm TREAT88a 0--25 cm 25-50 cm 50-75 cm TREAT69b 0-25 cm 25-50 cm 50-75 cm

Mean (mg/I)

SE (mg/I)

CV (%)

Median (mg/1)

Minimum (mg/l)

Maximum (mg/l)

6 6 6

59 34 65

9 10 41

15 29 64

65 33 51

47 19 25

7I 49 117

6 6 6

62 81 94

19 28 27

31 35 28

57 89 91

41 42 60

96 109 137

6 6 6

52 71 81

4 14 16

8 19 19

52 71 85

46 47 57

56 86 95

6 6 6

16 13 19

5 3 8

30 21 43

15 13 14

9 I0 11

22 17 30

a, March 1995; b, March 1996; SE, standard error; CV, coefficient of variation.

to DOC concentrations on TREAT92 and 88 (Table 5). Experiments were done in a different year for this treatment, but in the same period of the year. Our results and other studies have shown that seasonal variation of DOC was significant whereas short term variation was not (Grieve, 1990; Hughes et al., 1990; McLaughlin et al., 1994; Liechty et al., 1995). After 26 years of maize cultivation, mean DOC concentrations were two to three times less than the ones on TREAT92, all depths included. They were two times less at 40 cm depth compared to TREAT88.

3.1.2. Water extraction DOC concentrations with water extraction are shown in Table 6. Mean DOC concentrations on TREAT0 varied between 34 and 65 mg/l all depths included. DOC concentrations on TREAT92 were significantly (*P < 0.05) higher than those on TREAT0 in the 25-50 cm layer. DOC concentrations on TREAT92 and 88 varied in the same range with no statistical difference (*P < 0.05). Although concentrations on TREAT92 had shown higher mean values and spatial variability than TREAT88. Mean DOC concentrations on TREAT69 were 4- to 6-fold less than those on TREAT92. These concentrations were higher than the ones commonly observed in other studies with water extraction (Burford and Bremner,

1975; Bijay-Singh et al., 1988). Nevertheless, comparison was difficult because of the differences between extraction parameters (soil pretreatment, soil water ratio, contact time, speed and time of centrifugation). McCarty and Bremner (1992) found DOC concentrations ranging between 13 and 21 mg/ I on four cultivated soils with similar extraction parameters. These were equivalent to the ones we had measured on TREAT69, but much less than those on TREAT92 and 88. We have calculated quantities of DOC per kilogram of soil for both extraction methods. The concentrations of the soil solution extracted by means of porous cups were multiplied by soil humidity. The hypothesis is that the extracted solution (80-90% of the soil water content with the applied vacuum in this type of soil) is representative of all the water in the soil. Quantities of DOC per kilogram of soil with water extraction were 8- to 10-fold higher than DOC extracted by means of porous cups.

3.2. Qualitative variation of dissolved organic carbon 3.2.1. Molecular weight fractionation Three molecular weight families of DOC were characterized: small molecules <10 000 Da (SM), medium molecules 10 000-100 000 Da (MM), and

262

Trea~

a

DOC (rag !4 )

0

TreatO b

DOE (mg !~) 0

10 20 30 40 50 60 70

10 20 30 4( 50 60 70

204 <

I00<>

100< |

I

>I

.__. . . . _d . . . . . . . . m

~

B

4O <

! 100<

m

•

W!

100<> :::: ..... . ,..,

>1

6O

I 1~<

I

<

m

m

100<>

/

>I

(

Fig. 1. Distribution in molecular weight families of DOC on TREAT0. (a) Porous-cup vacuum sampler; (b) water extraction.

large molecules > 100000 Da (LM). This was done for each treatment with both extraction methods. The comparison of the two extraction methods for TREAT0 showed that the distribution of DOC in molecular weight families was completely different (Fig. 1). DOC extracted with porous cups (TREAT0a) displayed a dominance of MM at all depths (4254%). Proportions of SM and LM varied between 14 and 30%. DOC extracted with water (TREAT0b) showed a dominance of LM (56-60%), and a minor fraction of SM (12-18%). In considering the porous cup extraction on TREAT92, 88, and 69 (Fig. 2), DOC on TREAT92 exhibited a dominance of MM (54-66%), and a minor fraction of LM (8-15%) for all depths, TREAT88 had a dominance of SM (55%), and minor fraction of LM (9%) at 40 cm depth, and TREAT69 had an equal proportion of SM and MM (44-50%), except at 20 cm where MM predominated (52%). Molecular distribution of DOC on TREAT0 was in accordance with the results of Homann and Grigal (1992). They found that 60% of DOC from a forest soil had a molecular weight over to 14 kDa as determined by dialysis. Distributions of molecular weight with water extraction are reported in Fig. 3. All treatments exhib-

ited a dominance of LM (44-70%). Their proportions and quantities increased with increasing depth and with decreasing time of cultivation. SM were in minor fraction (5-22%) for all treatments. According to Thurman (1985), SM was comprised of fulvic acids and small molecules such as fatty acids, amino acids, carbohydrates and hydrophilic acids. MM included polysaccharides, humic and some fulvic acids, whereas LM were mainly humic metal complexes and colloids. Our results globally showed a high proportion of molecules > 10 kDa, which is a consequence of the soil properties, in that it has a weak sorption capacity and high concentrations of iron or aluminium which can form complexes. The comparison of extraction methods showed that water extracts a much higher proportion of LM. These molecules are potentially soluble molecules but they do not circulate intensively in the profile.

3.2.2. Carbon isotope ratios Carbon isotope ratio measurements were made on soil organic matter and on DOC with both extraction methods. Carbon isotope ratios of soil organic matter under maize cultivation showed an increase in ~5~3C

263

a

Treat92

a

Treat88

DOE (rag l's) 0 10 20 30 40 50 60 70

D O E (rag I't)

b

Treat69

DOC (mg i4) 0 10 20 30 40 50 60 70

0 10 20 30 40 50 60 70

'_-

20 a n

<10

100,o10

100o10

>100

>100 I l l

,lOan

[! 40

ml
.....

40¢m . ~ <10

m

100o10

100o10 . ~

I~I0

>100

>I00

60 a n

60 e m . ~

<10

I0 ~

100o10

>1oo (kD)

'2,J

(kD)

Fig. 2. Distribution in molecular weight families of DOC extracted by means of porous-cup vacuum samplers. (a) March 1995; (b) March 1996.

with time of cultivation (Table 7). The proportion of carbon originating from maize was calculated; it represented 3, 7 and 12% at 20 cm for TREAT92, Treat88 a

Treat92 a

DOE (mg P) 0

0 20 cat

m

100o10 ||

Treat69 b I0 20 30 40 50 60 70 I

0 20 ¢m

<10

<10

100<>10

100<>10

>100

>100

40 em

40 a n

<10

<10

I00o10

I00<>I0

>I00

III • .-: :-::::.: ...........

60(311

l-Hi

DOC (mg I-')

D O C (ragI")

I0 20 30 40 50 60 70

20 a n

88 and 69, respectively. This organic carbon originating from maize showed a decrease of 3 and 4% with depth for TREAT88 and 69, respectively. These

100 (kD)

i

L,

li

!

I l

II

clnll

<10 100o10 >ioo

,

|

>100 60

I0 20 30 40 50 60 70

j

II

ii

;

'

i

!

)

I

!

,

I i i

~iiiiiiii

(kD)

Fig. 3. Distribution in molecular weight families of DOC extracted from fresh soil with distilled water. (a) March 1995; (b) March 1996.

264

Table 7 Carbon isotope ratios of soil organic matter and dissolved organic carbon as a function of depth and treatments

~13C(%0) TREAT0a

TREAT92a

TREAT88a

TREAT69b

-27.6 -28.1 -28.0

-27.2 -27.2 -27.0

-26.5 -26.6 -27.0

-25.8 -26.3 -26.3

extracted with porous cups cm -27.2 cm -27.6 cm -27.3

-26.4 -22.0 -26.6

-20.9 -

-22.2 -23.9 -23.3

-25.4 -26.2 -26.7

-24.4 -25.7 -26.6

-24.2 -25.5

Soil organic carbon 0-25 cm 25-50 cm 50-75 cm DOC 20 40 60

DOC extracted with HzO 0-25 cm 25-50 cm 50-75 cm

-27.0 -27.2 -27.4

a, March 1995; b, March 1996.

values were much smaller than the ones encountered on maize fields with equivalent cultivation time, but different soil nature. In a silty soil, Puget et al. (1995) found that after 6 and 23 years of maize cultivation the soil organic matter had 20 and 45%, respectively of carbon originating from maize. Whereas Arrouays et al. (1995) showed after the same time a 10 and 20% enrichment in acid humic loamy soils. The organic matter can not form stable aggregates in a sandy soil. There is no protection effect, and thus a higher rate of mineralization occurs (Puget et al., 1995). Mineralization rate can be six times higher under maize fields than under forest soils (Arrouays et al., 1995). Carbon isotope ratios of DOC were greater than those measured on soil organic matter for all treatments and with both extraction methods (Table 7). Moreover, the carbon isotope ratios of DOC extracted with porous cups were higher than those from water extraction, except at 20 cm depth on TREAT92. The proportion of carbon originating from maize increased with increasing time of cultivation, as measured with water extraction: 11-5% on TREAT92, 17-5% on TREAT88 and 19-12% on TREAT69 with increasing depth for all treatments. DOC originating from maize, as determined with porous cup extraction, was higher at 40 cm depth on TREAT92 and 88 (35 and 42%, respectively) than on TREAT69

(23%). This can be explain by an accumulation of maize residues which occurred around 30 cm depth as a consequence of ploughing. TREAT92 and 88 were ploughed whereas TREAT69 was not and residues were at the soil surface. Porous cup extractions provide measurements of the dissolution of organic matter under unperturbed conditions which occurred naturally in the soil, these can reflect very brief and ponctual events. Whereas water extractions do not reflect such events because of the extraction protocol which is much more 'aggressive' (higher water/soil ratio and higher surface contact than in natural conditions) and which gives the mean of measured parameters on the overall layer.

4. Discussion

4.1. Evolution of dissolved organic carbon after clear cutting Porous cup extraction on TREAT92 showed an increase in DOC concentration and variability which appeared at 20 cm depth after the first year of cultivation, and at 40 and 60 cm after the second year (Tables 3 and 4). Concentrations reached exceptionally high values (50-200 mg/l) which represented a 3- to 5-fold increase of the DOC concentration compared to

265

TREAT0. The molecular size distribution of DOC on TREAT92 showed higher proportions of MM and LM compared to DOC on TREAT0 (Figs. 2 and 3). The carbon isotope ratio measurements pointed out an endogenous origin of DOC on TREAT92, except at 20 cm depth as determined with porous cup extraction (Table 7). In the first stage, cultivation practices intensified the mineralization rate of the initial organic matter. Mobilized DOC might be highly degraded plant material (lignin, cellulose and hemicellulose degradation products) and microbially synthesized products (carbohydrates, amino acids). This result is in accordance to Arrouays et al. (1995) who found that initial soil organic carbon can be separated in two parts: a very labile pool (40%) mineralized during the first years of cultivation (half time life 1.6 years) and a more refractory one (half time life 43.3 years). This was an indirect effect resulting from physical and chemical disturbance of soil due to cultivation practices (liming, ploughing, irrigation, fertilization) which activated microbial development. Biological activity is known to control DOC production by (1) oxidative degradation of plant-derived compounds and (2) release of microbial metabolites (Hughes et al., 1990; Lemaitre et al., 1995; Guggenberger and Zech, 1993; Guggenberger, 1994). At the same time, heterotrophic microorganisms controlled DOC consumption, since DOC can be the substrate for denitrifyers or anaerobic microbes (Katz and al., 1985; Paul and Beauchamp, 1989; McCarty and Bremner, 1992).

4.2. Evolution of dissolved organic carbon after many years of cultivation After 7 years of cultivation, DOC concentrations on TREAT88 were equivalent or less than the ones on TREAT92 (Table 3). TREAT69, after 26 years of maize cultivation, showed much less DOC concentrations than TREAT92 (Table 5). Molecular size distributions of DOC on TREAT88 and 69 exhibited a preponderance of SM and MM as measured with porous cup extraction (Fig. 2). The proportion of LM extracted with water decreased with increasing time of cultivation (Fig. 3). According to Thurman (1985), these molecular size fractionations indicate an increase of fulvic acids and smaller molecules (fatty acids, amino acids, carbohydrates and hydrophilic

acids) and a decline of humic acids and large humic metal complexes. The DOC had a higher proportion of carbon originating from maize as cultivation time increased (Table 7). Once the original organic matter is stabilized, the DOC originates mainly from maize and occurs in the fraction < 10 000 Da. This is a direct effect of cultivation practices by modification of DOC source. The input of organic matter originating from maize represents in total 15 t/ha per year dry matter (Lubet and Juste, 1985; Plenet, 1995), and represents an important source of easily mineralizable carbon. The fraction of soil organic matter coming from maize increases during the first decade of cultivation and reaches a plateau thereafter (Arrouays et al., 1995).

5. Conclusions Factors which control DOC concentrations in soil solution are complex because of their diversity, intensity and heterogeneity in and between treatments. However, these results readily point out the disturbance generated by cultivation practices. During the first stage, cultivation intensifies mineralization of the initial organic matter, leading to an increase of 2- to 5fold in the quantity of DOC, with an important variability. This phenomenon generated mainly MM and LM. During the second stage, once the original organic matter is stabilized, DOC concentrations decreased with increasing time of cultivation. Both soil organic carbon and DOC were enriched in carbon originating from maize with increasing time of cultivation, and DOC was mainly composed of SM and MM. From a methodological point of view, it is clear that the two methods of extraction compared influence the DOC characteristics recorded.

Acknowledgements This study was carried out as part of the Agriges program and in collaboration with university of Paul Sabatier in Toulouse, France. The authors wish to express their gratitude to Professors L. Labroue and R. Delmas for their helpful assistance and comments, and to Aquitaine region and INRA for their financial contributions.

266

References Arrouays, D., Balesdent, J., Mariotti, A. and Girardin, C., 1995. Modelling organic carbon turnover in cleared temperate forest soils converted to maize cropping by using ~3C natural abundance measurements. Plant Soil, 173: 191-196. Balesdent, J., 1991. Estimation du renouvellement du carbone des sols par mesure isotopique ~3C, precision, risque de biais. Cah. Orstom, sEr. PEdol. XXVI, 4:315-326. Beauchamp, E.G., Trevors, J.T. and Paul, J.W., 1989. Carbon sources for bacterial denitrification. Adv. Soil Sci., 10: 113142. Bijay-Singh, Ryden, J.C. and Whitehead, D.C., 1988. Some relationships between denitrification potential and fractions of organic carbon in air-dried and field-moist soils. Soil Biol. Biochem., 20 (5): 737-741. Burford, J.R. and Bremner, J.M., 1975. Relationships between the denitrification capacities of soils and total, water-soluble and readily decomposable soil organic matter. Soil Biol. Biochem., 7: 389-394. Dagnelie, P., 1973. ThEorie et Mrthodes Statistiques. Vol. 1. Les Presses Agronomiques de Gembloux ed., Gembloux, Belgique, 378 pp. Firestone, M.K. and Davidson, E.A., 1989. Microbiological basis of NO and N20 production and consumption in soil. Trace gases between terrestrial ecosystems and the atmosphere. Andrea and Schimel (Editors), Chichester, pp. 7-21. Germon, J.C., Pinochet, X. and Catroux, G., 1983. Relations between the parameters characterizing the kinetics of potential denitrifying activity and the various forms of soil carbon. 3rd Coll. Int. Ecol. Microb., East Lansing, Michigan, Wl. Grieve, I.C., 1990. Seasonal, hydrological, and land management factors controlling dissolved organic carbon concentrations in the loch fleet catchments, southwest Scotland. Hydrol. Processes, 4:231-239. Guggenberger, G., 1994. Acidification effects on dissolved organic matter mobility in spruce forest ecosystems. Environ. Int., 20 (1): 31-41. Guggenberger, G. and Zech, W., 1993. Dissolved organic carbon control in acid forest soils of the Fichtelgebirge (Germany) as revealed by distribution patterns and structural composition analyses. Geoderma, 59:109-129. Homann, P.S. and Grigal, D.F., 1992. Molecular weight distribution of soluble organics from laboratory-manipulated surface soils. Soil Sci. Soc. Am. J., 56: 1305-1310.

Hughes, S., Reynolds, B. and Roberts, J.D., 1990. The influence of land management on concentrations of dissolved organic carbon and its effects on the mobilization of aluminium and iron in podzol soils in Mid-Wales. Soil Use Manage., 6 (3): 137-145. Jambert, C., 1995. Emissions de composes azotEs dans I'atmosph/~re par les agrosyst~mes fertilisEs: ma'fsiculture dans les landes de Gascogne. PhD Thesis, Paul Sabatier University, Toulouse, France, 110 pp. Katz, R., Hagin, J. and Kurtz, L.T., 1985. Participation of soluble and oxidizable soil organic compounds in denitrification. Biol. Fert. Soils, 1:209-213. Lemaitre, A., Tavant, Y., Chaussod, R. and Andreux, F., 1995. Characterization of microbial components and metabolites isolated from a humic calcic soil. Eur. J. Soil Biol., 31 (3): 127-133. Liechty, H.O., Kuuseoks, E. and Mroz, G.D., 1995. Dissolved organic carbon in northern hardwood stands with differing acidic inputs and temperature regimes. J. Environ. Qual., 24: 927-933. Lubet, E. and Juste, C., 1985. CinEtique de la production de mati~re s~che et prEl~vements d'EIEments nutritifs par une culture irriguEe de mai's ~ haute potentialitE de rendement. Agronomie, 5 (3): 239-250. McCarty, G.W. and Bremner, J.M., 1992. Availability of organic carbon for denitrification of nitrate in subsoils. Biol. Fertil. Soils, 14: 219-222. McLaughlin, J.W., Lewin, J.C., Reed, D.D., Trettin, C.C., Jurgensen, M.F. and Gale, M.R., 1994. Soil factors related to dissolved organic carbon concentrations in a black spruce swamp, Michigan. Soil Sci., 158 (6): 454-462. Paul, J.W. and Beauchamp, E.G., 1989. Effect of carbon constituents in manure on denitrification in soil. Can. J. Soil Sci., 69: 49-61. Plenet, D., 1995. Fonctionnement des cultures de ma'fs sous contrainte azotEe. DEtermination et application d'un indice de nutrition. Th~se INPL, Nancy, France, 247 pp. Puget, P., Chenu, C. and Balesdent, J., 1995. Total and young organic matter distributions in aggregates of silty cultivated soils. Eur. J. Soil Sci., 46: 449-459. Thurman, E.M., 1985. Organic Geochemistry of Natural Waters. Nijhoff/Junk, Dordrecht, The Netherlands. Vance, G.F. and David, M.B., 1989. Effect of acid treatment on dissolved organic carbon retention by a spodic horizon. Soil Sci. Soc. Am. J., 53: 1242-1247. Wesemael, B.V. and Verstraten, J.M., 1993. Organic acids in a moder type humus profile under a Mediterranean oak forest. Geoderma, 59: 75-88.

t~ 1997 ElsevierScience B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

267

Analysis of impact of farming practices on dynamics of soil organic matter in northern China H.S. Yang, B.H. Janssen* Department of Soil Science and Plant Nutrition, WageningenAgricultural University, P.O. Box 8005, 6700 EC Wageningen, The Netherlands

Accepted 1 April 1997

Abstract

The objectives of this study are to predict the long-term soil organic matter (SOM) dynamics in arable land in northern China, and to suggest measures for the maintenance and improvement of SOM content. With a newly developed model calibrated to relevant data collected from this region, the course of SOM content and SOM quality was calculated over a time span of 50 years. Two levels of annual grain yields were considered: 'mean yields' of 7500 kg/ha and 'high yields' of 15 000 kg/ha, and three scenarios for organic inputs into the soil: (1) roots and stubble; (2), as (1) plus one-third of straw produced in situ; (3) as (2) plus farmyard manure (FYM) made with another one-third of the straw produced in situ. It was predicted that with mean yields SOM contents can be maintained at 10, 13 and 15 g/kg under the three scenarios, and with high yields at 15, 19 and 21 g/kg. The present SOM contents are often below 10 g/kg. Depending on whether present SOM contents are below or above the steady-state levels, they will increase or decrease, respectively. The fraction of annual mineralization of SOM will increase with the magnitudes of annual organic inputs, indicating improvement in SOM quality. Hence, in most arable fields in northern China, both quantity and quality of SOM will rise upon increases in annual crop production, or when a large portion of straw is returned to the soil, directly or indirectly via FYM. © 1997 Elsevier Science B.V. Keywords: Arable farming; Crop residues; Farmyard manure; Northern China; Roots and stubble; Soil organic matter

1. Introduction

Northern China, primarily the up-land of HuangHuai-Hai Plain, is the most important region for grain production (mainly wheat and maize) in China. In the 1980s, the production of wheat and maize in this region accounted for 48 and 40% of the national production, respectively (Liu and Mu, 1988; Chinese Academy of Agricultural Sciences, 1989). Two crops per year, winter wheat followed * Corresponding author. Tel.: +31 317 483842/482339; fax: +31 317 483766; e-mail: [email protected]

by summer maize, is the dominating cropping system. Before the 1980s, soil fertility in this region was maintained mainly by use of organic fertilizers such as farmyard manure (FYM), compost, green manure, straw and organic wastes. At present, soil organic matter (SOM) content is below 10 g/kg in most soils, and more than half of the arable land has a medium to low fertility according to the local standard (Cao et al., 1986; Wang et al., 1988; Zhao, 1989; Portch and Jin, 1995). Being the most important organic resource in the region, straw is shared among domestic fuel, animal fodder and organic fertilizers (Liu and Mu, 1988).

Reprinted from the European Journal of Agronomy 7 (1997) 211-219

268

Since the early 1980s, some changes have taken place in agriculture in China: (1) the scale of farming has been reduced from production teams to farmers' households with unchanged public ownership of the land; (2) chemical fertilizers, especially for nitrogen, have become widely available and therefore the main source of crop nutrients; (3) annual crop yields have increased considerably; (4) labor costs have risen significantly, and economic returns are now playing a very important role in farmers' decision making. Meanwhile, proper farming machinery is still lacking and hand-work is still dominating in the field. These changes have brought about changes in organic inputs to soils. Returning of straw into the soil and use of FYM are vanishing practices at many places. The planting area of green manure crops is shrinking, and other forms of organic fertilizers, such as municipal wastes, are becoming less attractive in arable farming (Zhou, 1989; Yang, 1990; Portch and Jin, 1995). As a result, the dynamics of SOM becomes uncertain, particularly in the long run. In field trials contrasting results have been reported. When roots and stubble were the only inputs, SOM contents could decrease, increase, or remain unchanged (Zhang et al., 1985; Wang et al., 1986a; Cheng, 1987; Zhao et al., 1987, 1990, 1991; Jiang et al., 1990; Zeng et al., 1992). A certain level of SOM content is a basic requirement for sustainable production of high crop yields. According to Pieri (1989), there is a serious risk of physical soil degradation when SOM content is less than 0.05 times (clay + silt)content, both expressed in mass fractions. Many experimental results have indicated that with the present often low SOM contents it is difficult to substantially increase yields (Jiang et al., 1991; Zhou, 1991). This implies that in the major part of the arable land SOM contents should be improved rather than maintained if the growing demands for grain production are to be met. Therefore, analysis and prediction of long-term SOM dynamics is crucial for the design and evaluation of farming strategies directed to sustainable grain production. The objectives of this study are to predict dynamics of both quantity and quality of SOM over 50 years based on data collected from the study area, and to suggest measures to be taken for the maintenance and improvement of SOM content.

2. Materials and methods

The calculation of SOM dynamics was based on an equation proposed by Yang (1996): Yt=Y0 x e x p [ - R 9 x 0c x t ) I-s]

(1)

in which t is time; Y0 and Yt are the substrate C quantities at t = 0 and t, respectively; f is a temperature correction factor; R9 (dimension ts- ~) is the average relative mineralization rate between t = 0 and t = 1 at the temperature of 9°C; and S (dimensionless) denotes the speed of 'aging' or the decrease in decomposability of the substrate. This equation proved valid under diverse environmental conditions and for existing SOM, as well as for various types of organic materials that are commonly applied in arable farming, such as plant materials and animal manure (Yang, 1996). Experimental data on C remaining in soil were collected for SOM, FYM, green manure, and straw and roots of wheat and maize (Kortleven, 1963; Kolenbrander, 1974; Jenkinson, 1977; Jenkinson and Rayner, 1977; Wang et al., 1984, 1989a,b; Green Manuring Group, 1985; Zhang et al., 1986; Xiong et al., 1987; Xu et al., 1993). Of the total 48 cases with 301 observations, three cases of SOM and one case of FYM refer to places outside northern China. The one case of FYM is from western Europe (Kolenbrander, 1974) and it was used because it refers to the situation existing before the present intensive animal production in western Europe, and hence may be comparable to the FYM produced in traditional farmers' households in northern China. Since only two sets of data on SOM from northern China were available, three other sets of data were used as well, one from The Netherlands (Kortleven, 1963) and two from England (Jenkinson, 1977; Jenkinson and Rayner, 1977). From the collected data, the two main model parameters, R9 and S, in the equation above were derived by non-linear regression, applying Statgraphics 7.1 (STSC Inc., 1986). The values of f in the equation were obtained by averaging the values of f corresponding to monthly mean temperatures, as suggested by Yang (1996). Data on organic inputs from roots, stubble and straw were derived from grain yields. Experimental results from Anderson (1988) and Tong et al. (1988) indicate that the partitioning of dry matter among

269

Table 1 Dry matter quantities (kg/ha) of various crop parts at mean and high yield levels Yield levelGrain

Mean High

7500 15000

Straw

Roots

Total

M o w n Stubble

9150 13360

6900 10020

2250 3240

2780 3660

roots, straw and grain of wheat is not much affected by changes in growing conditions in the case of wheat, but in the case of maize it is. Consequently, the average mass ratios of grain: straw: roots (in the top 20 cm) for maize should be adjusted for different crop yields. For annual grain yields (wheat plus maize) of 7500 kg/ha, the average ratios are 1: 1.22:0.37 (Zhang and Wang, 1984; Wang et al., 1986b; Li et al., 1990). Interpolation and extrapolation of experimental results from Tong et al., (1988) indicate that the annual biomass production of straw and roots rises by only 46% and 32%, respectively, when the summed annual grain yields of maize and wheat increase from 7500 to 15000 kg/ha. At both yield levels, a part of the total production of straw was allocated to stubble. According to data by Wang et al. (1988), annually 2250 kg/ha of stubble remains in the field, when the grain yield of wheat plus maize is 7500 kg/ha. This means that straw is subdivided into stubble and mown straw in the ratio of 1"3. The quantities of dry matter and carbon in the various crop components at the two yield levels are given in Table 1. For stubble and mown straw added to the soil, the same values of R9 and S were taken. Three management scenarios for organic inputs into the soil were formulated: (1) only roots and stubble; (2) as (1) plus return of one third of the straw produced in situ; (3) as (2) plus FYM made with another one-third of the straw. Scenario 1 represents the very common practice in this region. Scenario 2 was derived from the estimates by Liu and Mu (1988) that about one third of the present straw production can be used for returning. Scenario 3 represents maximum organic inputs in arable farming, assuming that one-third of straw is consumed as domestic fuel. Two levels of annual grain yields, being summed yields of wheat and maize, are considered: 'mean yields' of 7500 kg/ha and 'high yields' of 15000

kg/ha. The former is the average yield in this region in the early 1990s (Ministry of Agriculture of China, 1992), and the latter represents the current maximum yield that has been realized in some areas (Chinese Academy of Agricultural Sciences, 1989). Three initial SOM contents are regarded, namely 5, 10 and 20 g/kg, representing low, medium to high, and very high levels in this region (Soil and Fertilizer Institute of CAAS, 1986; Zhao, 1989). The mean annual temperature ranges from 12°C (in Beijing) to 14°C (in Zhengzhou) (Liu and Mu, 1988; Meteorological Institute of CAAS, 1994). Therefore, 13°C is taken as the average value representing this region, and the corresponding value o f f in eqn (1) is 1.7. All calculations were confined to the top 20 cm of soil. Soil bulk density was set at 1.3 g/cm, C content of root, stubble and straw at 450 g/kg, and that of SOM at 580 g/kg. Following the suggestion by Zhang et al. (1985), it was assumed that after the conversion of straw into FYM, still 56% is recovered in FYM of the C originally in the straw that is supplied to animals as fodder and bedding.

3. Results Table 2 shows the values of the two model parameters, R9 and S, for the four materials. The value of R9 for FYM is about equal to that of roots. Usually FYM is found more resistant to mineralization than roots, but in China, FYM still contains a considerable portion of straw. Which may explain why FYM is relatively easily mineralized. The predicted changes in SOM content under the three scenarios are shown in Fig. 1 for the three initial SOM contents and the two yield levels. Finally a Table 2 Values for R9 (per year) and S and their standard errors (SE), as derived by non-linear regression with Eq. 1 from the indicated number of observations found in literature R9

Straw 1.11 Roots 0 . 8 0 FYM 0.81 SOM 0.057

SE

S

SE

adj. R2 Number

0.022 0.026 0.023 0.0079

0.66 0.67 0.50 0.46

0.017 0.022 0.029 0.042

0.84 0.89 0.88 0.91

100 36 67 22

270

SOM,

g kg-1

25 i

scenario-1

scenario-2

scenario-3

i

20

15 ,,. ,,,

-,,

.,-

s s

10

,o,

,

o

i

o

!

2s

so

o

,

i,

i

25

,

50

o

e

,

t

2s

so

years

Fig. 1. Dynamics of SOM for three scenarios. Initial SOM contents are 5, 10 and 20 g/kg. Mean yields ~ , steady-state SOM content will be reached for each scenario-yield combination (Table 3), irrespective of initial SOM contents. For Scenario 1, in which roots and stubble are the only organic inputs, SOM content can be maintained at 10 and 15 g/kg at mean yields and high yields, respectively. These are about two-thirds of the values obtained in Scenario 3, in which FYM and straw are applied. The prediction for Scenario 3 could not be validated since no relevant experimental results were available. For Scenarios 1 and 2, the model predictions agree reasonably well with the experimental results from Cheng (1987), 2-year trial; Zhao et al. (1987), 5-year trial; Jiang et al. (1990), 5-year trial; and from Zhou (1991), 12-year trial. The last one is shown in Fig. 2. To reach the steady state, SOM contents may increase or decrease, depending on whether the present contents are below or above the steady-state levels (Fig. 1). The absolute annual change in SOM content is related to the difference between the present and the final SOM contents, and hence it decreases over time, sharply in the beginning and slowly at a later stage, as shown in Fig. 3. The fraction of SOM mineralized per year (FM) can be considered as a measure of SOM quality, higher values indicating higher quality. Newly formed SOM has a higher FM than existing (old) SOM, and hence SOM quality improves with increasing proportions of

High yields m .

newly formed SOM in total SOM. Such situations are created by increasing annual organic inputs, and hence the quality rises from Scenario 1 to Scenario 3, and is lower for mean yields than for high yields (Fig. 4). For a given scenario-yield combination, FM is higher for low than for high initial SOM contents. Fig. 4 also shows that FM may change over time. Three patterns of change can be distinguished. The first one is an increase of FM followed by a decrease; it is seen when SOM content increases over time (compare Fig. 1). The second pattern is that FM remains constant; it is found when SOM content remains unchanged over time. The third pattern is a decrease of FM followed by an increase; it is found when SOM content decreases over time (compare Fig. 1). Since SOM contents will approach specific steadystate levels, the values of FM should also approach

Table 3 SOM contents (SOM, g/kg) and fractions of SOM that are mineralized annually (FM, %) in steady state, for mean and high yields, and three scenarios of organic matter management Yield level Mean High

Scenario 1

Scenario 2

Scenario 3

SOM

FM

SOM

FM

SOM

FM

10 15

4.8 4.8

13 19

5.2 5.2

15 21

5.6 5.6

271

S O M , g kg-1

N+straw

N

14

14 ,,

12

12 10

10 ,

,

l

0

,

t

4

'

*

l

8

-

12

:

0

:

*

'!

4

years

'

* .

.

.

.

8

I

12

years

Fig. 2. Observed (points) and model calculated (lines) SOM contents for a field trial from Zhou (1991). N is the treatment with use of N fertilizer, and N + straw is with N fertilizer plus returns of straw.

certain specific steady-state levels, which are independent of initial SOM contents (Table 3). They differ among the three scenarios because roots, stubble and FYM differ in values of R9 and S, but they hardly differ (not visible in the first decimal, Table 3) for yield levels because the distribution of organic matter inputs is practically the same for both yield levels.

4. Discussion and conclusions

Accumulation of SOM is the result of input and output. The output is difficult to manage, as mineralization rate is largely determined by environmental

factors such as temperature and soil properties. Temperature was standardized in our calculations. For soil properties, no adjustment could be made because appropriate data for the calibration of model parameters were lacking. Fortunately, in the HuangHuai-Hai Plain, the variation in soil texture is restricted to the range between sandy loams and loamy clays, and the variation in pH to values between 6 and 8 (Liu and Hseung, 1990). Therefore we may consider the model outcomes as representative for most soils in the region. Inputs such as FYM and straw can be controlled, but roots and stubble are more difficult to control. When no extra organic input is given (Scenario l),

annual change of SOM, g kg "1

0.6 ~

Scenario-2

~nad~.l

Q

Scenario-3 H, 5 ~kg SOM 41-

tt

M, 5 ~kg SOM

0.3

H, I0 g/kg S O M 4.

M, 10 g/kg SOM

-)---~"---,--~ ~r

[

H, 20 g/kg S O M dL-

M, 20 ~kg SOM -0.3

,,

0

, .....

25

I

50

0

I

I

25

50

I

0

25

'

I

5o

years Fig. 3. Annual change of SOM content in three scenarios, with three initial SOM contents and two yield levels. M, mean yields, H, high yields.

272

fraction mineralized annually, % 12

Scenario-3

Scenario-2

Scenario-1

10

(

""......."~ r~'

-

I

. " ..........

. . . . -fl-. . . . 1 3 ' ....... -13

~

M, 5 g/kg SOM

......-(~. . . . .

-G

.....-;..

H, 5 g/kg SOM 41-

-0

H, 10 g/kg SOM

t

4-

M, 10 g/kg SOM H. 20 g/kg SOM M, 20 g/kg SOM

!'

'

I

0

.~

25

I

50

. . . . .

0

I

25

,

., I

-

50

!

0

"

-I

25

50

years Fig. 4. Fraction of SOM mineralized/year in three scenarios with three initial SOM contents and two yield levels. M, mean yields, H, high yields. the steady-state SOM content is determined by the annual return of roots and stubble which are closely related to crop yield (Tong et al., 1988), and therefore not better manageable than yield itself. Once the yield level is known, the quantity of roots is fixed, and so is the steady-state SOM content. The value of 10 g/kg (Table 3 and Fig. 1) at mean yields reflects the current situation in northern China. When annual grain yields

increase to 15 000 kg/ha, the steady-state SOM content would rise to 15 g/kg. Depending on whether the present value is above or below 10 (or 15) g/kg, SOM content will decrease or increase. The contradictory results of decreasing, increasing, or unchanging SOM contents that have been observed in field trials, as mentioned above, may be related to differences in initial SOM contents and yield levels.

indirect use, % 100

80

100

•

yield

,•mean ',

',

60

',

20

lscenario-1

I,

',

',

' 2

I-

',\

',\

40

80

,\

',

',

\,

\"

~

\ ~

"

[scenario-2 le lscenario-3

60 ~,

40

20

40

direct use, %

80

100

" ~

'~

\

", ,

-, ~.

0

'\

',~

'20e 22

20 ,

'18 60

~

',

O

0

"ld

.

.l ~,

20

mm

\.,.

', ', ~

40

', ~,

s

60

\ ~

\ I

80

100

direct use, %

Fig. 5. Steady-state SOM content (g/kg) as indicated by the figures alongside the isolines, for mean yields and high yields. Positions of the three scenarios are among all possible combinations of direct and indirect use of straw (via FYM). Total use is less than or equal to 100%.

273

Scenario 2 has immediate practical significance in view of the estimates by Liu and Mu (1988) that about one-third of the present straw production can be returned. That would eventually result in SOM contents of 13 and 19 g/kg at mean and high yields, respectively. Scenario 3, in which all straw, except one third that is used as domestic fuel, is returned directly, or indirectly via FYM, may be seen as the situation of maximum organic inputs in this region. The values of 15 and 21 g/kg represent the upper limits of SOM contents in the long run when annual yields are 7500 and 15 000 kg/ha, respectively. Theoretically SOM contents higher than 15 and 21 g/kg for the two yield levels may be obtained. Fig. 5 is a map of isolines for steady-state SOM contents calculated for all possible combinations of direct and indirect use of straw. If all straw is directly worked in, SOM contents would be about 19 and 26 g/kg for mean and high yields, respectively, and if all straw is returned via FYM, the final values would be about 15 and 21 g/kg. That would take away the use of straw as fodder or fuel, which is not a realistic scenario at present. So, according to the model calculations, direct application of straw is more effective than indirect application via FYM. Scenarios 2 and 3 may be considered as two examples of possible combinations of direct and indirect use of straw (Fig. 5). Since the annual changes in SOM content decrease with time (Fig. 3), the steady-state SOM contents (Table 3) will be established only in the long run. It implies that linear extrapolations of changes in SOM contents found in trials lasting only a few years, as reported, e.g. in Chen and Wang (1987), will likely result in an overestimation of actual changes. A sufficiently high SOM content is needed for high and sustainable crop yields, and conversely, high yields are needed to improve SOM content. Although Scenarios 2 and 3 are not pointless, in many practical situations only a supplemental contribution can be expected from FYM and straw. In the past it has not been different. Yang (1996) suggests that roots must have been the most important source for SOM in this region. Increasing biomass production of roots and stubble by increasing yields can be achieved by, among others, the use of chemical fertilizers. The strategy of improving SOM content by means of chemical fertilizers, the so-called 'obtaining SOM by chemical means' approach, has been suggested in

the last decade (Cao et al., 1986; Liu and Mu, 1988; Zhou, 1989; Zeng et al., 1992). Along with changes in quantity, there are changes in quality of SOM, in terms of annual mineralization fraction, due to the change in SOM composition. Similar effects have been found by, e.g. Van Dijk (1981), and Janssen (1984). Therefore, the annual release of nutrients from SOM will increase more than proportionally compared to the increase in SOM content. Farmers should be encouraged to return crop residues to the soil, directly or indirectly, in order to increase SOM contents above the base levels defined by roots and stubble. On the other hand, it should be realized that significant improvements will always require long-term efforts. In addition, farmers should be advised to raise the quality of FYM. At present, soil forms the largest portion of FYM. Less use of soil during FYM making would result in a better quality, and this would effectively reduce the costs of transport and spreading in the field. Summarizing, in most arable fields in northern China, both the quantity and the quality of SOM will rise upon increases in annual crop production, e.g. as a result of the use of chemical fertilizers. Similar positive effects on SOM will be brought about by returning (a part of) the straw, either directly by leaving it in the field or indirectly via FYM. Theoretically, the direct return is more efficient for SOM content and quality, but it may be less attractive for farmers because it takes away the use of straw as fodder.

References

Anderson, E.L., 1988. Tillage and N fertilization effects on maize root growth and root:shoot ratio. Plant Soil, 108: 245-251. Cao, Z., Li, Z., Ling, Y., Li, A. and Xu, Z., 1986.Soil fertility and rational fertilization in Huang-Huai-Hai Plain. Soil Sci. China, 18:289-295 (in Chinese). Chen, L. and Wang, J., 1987. Influence of green manure on soil organic matter. Chinese J. Soil Sci., 19:270-273 (in Chinese). Cheng, S., 1987. Accumulation and decompositionof soil organic matter in Loess Soil. Acta Univ. Septentrionali Occident. Agriculturae, 15:62-68 (in Chinese). Chinese Academy of Agricultural Sciences, 1989. Harnessing of Huang-Huai-Hai Plain and Agricultural Exploitation. Agricultural Science and Technology Press of China, Beijing (in Chinese). Green Manuring Group, 1985. Study on effective conditions for

274

accumulation of soil organic matter by green manures. Annual Report on Research, Soil and Fertilizer Institute of CAAS, pp. 73-75 (in Chinese). Janssen, B.H., 1984. A simple method for calculating decomposition and accumulation of 'young' soil organic matter. Plant Soil, 76: 297-304. Jenkinson, D.S., 1977. Studies on the decomposition of plant material in soil. V. The effects of plant cover and soil type on the loss of carbon from 14C labelled ryegrass decomposing under field conditions. J. Soil Sci., 28: 424-434. Jenkinson, D.S. and Rayner, J.H., 1977. The turnover of soil organic matter in some of the Rothamsted classical experiments. Soil Sci., 123: 298-305. Jiang, R., Li, Z. and Li, D., 1990. Studies on the role of chemical and organic fertilizers in promoting the fertility of Yellow FluvoAquic Soils. Acta Pedol. Sinica, 27:179-185 (in Chinese). Jiang, R., Zhang, H., Li, Z., Zhang, M., Li, D. and Liu, M., 1991. Effect of incorporation of chemical and organic fertilizers on soil fertility. In: X. Zhang (Editor), Development in Soil Fertility Research. Science and Technology Publisher of China, pp 162-168 (in Chinese). Kolenbrander, G.J., 1974. Efficiency of organic matter in increasing soil organic matter content. Transaction of the 10th International Congress of Soil Science, Moscow, Vol. 2, pp. 129-136. Kortleven, J., 1963. Quantitative aspects of humus accumulation and decomposition. Landbk. Onderz. 69, 1, Pudoc, Wageningen (in Dutch) 109 pp. Li, W. and Hseung, Y., 1990. The comprehensive improvement of the soils in the Huang-Huai-Hai plain. In: Institute of Soil Science, Academia Sinica, Soils of China. Science Press, Beijing, pp. 724-733. Li, B., Huang, S. and Dong, P., 1990. Theory and Technology for High Yield of Maize in Huang-Huai-Hai Plain. Scientific Bookand-Journal Press of China, Beijing (in Chinese). Liu, X. and Mu, Z., 1988. Cultivation System in China. Agricultural Press of China, Beijing (in Chinese), p. 529. Meteorological Institute of CAAS., 1994. Meteorological Data Base. Ministry of Agriculture of China., 1992. Agricultural Statistics of 1991 of China. Agricultural Press of China, Beijing (in Chinese). Pied, C., 1989. Fertilit6 des terres de Savanes: bilan de trente ans de recherche et de d6veloppement agricoles au sud du Sahara. Paris [etc.]: Minister/~ de la Cooperation et du D6veloppement, pp. 329-331. Portch, S. and Jin, J.-Y., 1995. Organic manure and biofertilizer use in China. Fert. News, 40: 79-83. Soil and Fertilizer Institute of CAAS, 1986. Zoning of Chemical Fertilizers of China. Agricultural Science and Technology Press of China, Beijing (in Chinese). STSC Inc., 1986. Statgraphics User's Guide. USA. Tong, G., Zhang, J., Xu, X. and Tang, Y., 1988. A study on biomass production of roots, stem and leaf of some major crops. Chinese J. Soil Sci., 19(3): 115-117 (in Chinese). Van Dijk, H., 1981. Some notes on the importance of mineralization and immobilization of nitrogen in making fertilizer recommendations. In: I.N.R.A. Reims (Editor), Colloque HumusAzote, Commissions I and II, pp. 151-160.

Wang, Z., Gu, B., Jia, S. and Zhang, Y., 1986. A study on the effect of incorporation of organic and inorganic fertilizers on improvement of soil fertility and crop yield. In: The Office of Integrated Harnessing and Exploration of Nanpi Area in Hei-Long-Gang Region, Hebei Province, Collection of Papers on Integrated Harnessing and Exploration of Nanpi County, Hebei Province, pp. 141-146 (in Chinese). Wang, W., Zhang, J., Wang, W., Cai, D. and Zhang, M., 1986. A study on soil organic matter balance and measures of its improvement in Huang-Huai-Hai region. In: Soil and Fertilizer Institute of CAAS, Annual Report on Research, Soil and Fertilizer Institute of CAAS, pp. 38-42 (in Chinese). Wang, Z., Li, X., Gu, B. and Jia, S., 1984. Report of effects on soil fertility of returning different organic materials into soil. Internal report of Soil and Fertilizer Institute, Hebei Province (in Chinese), p. 5. Wang, W., Wang, W., Cai, D. and Zhang, M., 1989. A study on decomposition rate of soil organic matter in Beijing farmland. Chinese J. Soil Sci., 20(5): 224-225 (in Chinese). Wang, W., Wang, W., Zhang, J., Cai, D. and Zhang, M., 1989. Decomposition of crop residues in soils of Beijing. Chinese J. Soil Sci., 20(3): 113-115 (in Chinese). Wang, W., Zhang, J., Wang, W., Cai, D. and Zhang, M., 1988. A study on organic matter balance in farmland in Huang-Huai-Hai Plain. Sci. Agricuit. Sinica, 21:19-26 (in Chinese). Xiong, S., Cheng, C. and Guo, M., 1987. A preliminary study on the effect of ploughing Sesbania cannabina pers on fertility of a sandy loam soil near Beijing. Acta Agricult. Universit Pekinensis, 13:335-342 (in Chinese). Xu, X., Zhang, J., Wang, J. and Tang, Y., 1993. Studies on decomposition rate and remaining fraction of various organic materials and their influence on humus composition and optical properties. Chinese J. Soil Sci., 24:53-56 (in Chinese). Yang, S., 1990. The causes for declining soil fertility and comprehensive strategies for improvement. In: Association Division of Chinese Science and Technology Association (Editor), Studies on Land Degradation and Prevention, pp. 321-324 (in Chinese). Yang, H.S., 1996. Modelling Organic Matter Mineralization and Exploring Options for Organic Matter Management in Arable Farming in Northern China. Ph.D. thesis, Wageningen Agricultural University, The Netherlands., 159 pp. Zeng, M., Jin, W., Yao, Y. and Yang, Y., 1992. Positive effect of coordinated application of organic and inorganic fertilizers in long-term experiments at one site. Soil Fert.,l: I - 6 (in Chinese). Zhang, J., Cai, D. and Wang, W., 1985. A stage summary on the experiment of returning crop residues into Concretion Black Soil. Annual Report on Research, Soil and Fertilizer Institute of CAAS, pp. 69-70 (in Chinese). Zhang, H., Liu, M. and Zhang, J., 1986. A study on decomposition of organic materials in dryland soils. Soil Fert., 4 : 7 - 1 1 (in Chinese). Zhang, J. and Wang, W., 1984. The Effect of roots and stubble on improvement of soil fertility. Chinese J. Soil Sci., 15:63-64 (in Chinese). Zhao, Q., 1989. Soil resources and their utilization in China. In: E. Maitby and T. Woilersen (Editors), Soils and Their Management - a Sino-European Perspective. Elsevier, London, pp. 3-18.

275

Zhao, S., Yan, H. and Lin, S., 1987. Effect of the methods of soil fertility improvement on crop yield and soil fertility. Shandong Agricultural Sci., 6:10-13 (in Chinese). Zhou, M., 1989. Discussion on organic fertilizer and balanced fertilization. Chinese J. Soil Sci., 20:145-146 (in Chinese). Zhou, G., 1991. Effect of long-term application of organic fertilizer on soil fertility in Black Loess Soil. In: X. Zhang (Editor),

Development in Soil Fertility Research. Science and Technology Publisher of China, pp.. 169-175 (in Chinese). Zhou, M., Lu, C. and Du, G., 1990. Potential degradation of soil fertility and corresponding strategies in Concretion Black Soil in northern Anhui Province. In: Association Division of Chinese Association of Science and Technology (Editor), Studies on Land Degradation and Prevention, pp. 336-340 (in Chinese).

This Page Intentionally Left Blank

© 1997 ElsevierScience B.V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

277

Agronomic measures for better utilization of soil and fertilizer phosphates Konrad Mengel* Institute of Plant Nutrition Justus Liebig University, Siidanlage 6, D 39390 Giessen, Germany Accepted 16 June 1997

Abstract

Global known phosphate deposits are a finite resource which will run out in about four centuries at the present consumption rate. Since about 90% of the phosphate mined is used for fertilizer, soil and fertilizer phosphate should be efficiently used. Various agronomic measures are discussed relevant for saving phosphate and avoiding losses. Phosphate fertilizer rates should be adjusted to measured requirements for phosphate using soil tests. Particularly in areas with high livestock intensities soils frequently are much enriched in available phosphate and do not need further phosphate application whether in organic or in inorganic form. Excessively high levels of available soil phosphate, much higher than required for optimum crop production increase the hazard of phosphate loss by wind and water erosion and even leaching. Loss of plant available phosphate in soils occurs by phosphate fixation which is especially strong in acid mineral soils. Such losses can be dramatically reduced by liming soils to a pH of 6-7. In tropical areas where lime frequently is not available row placement of phosphate fertilizer is recommended. Oxisols with a very low pH liming, however, may promote phosphate fixation due to the formation of phosphate adsorbing Al complexes. Biological assimilation of phosphate may prevent inorganic phosphate from fixation by soil particles. Organic anions produced during the decomposition of organic matter in soils as well as the excretion of anions by plant roots depress phosphate adsorption by competing with phosphate for binding sites at the adsorbing surface. Hence farming systems and rotations which bring much organic matter into soils contribute to a better use of soil and fertilizer phosphate. Mycorrhization of plant roots with appropriate fungi ecotypes may essentially improve the exploitation of soil phosphates. The choice of the appropriate phosphate fertilizer type is crucial for its efficient use. This applies particularly for apatitic fertilizers of which the availability is poor in weakly acid to neutral and calcareous soils. © 1997 Elsevier Science B.V. Keywords: Phosphate availability; Phosphate fertilizer; Livestock; Farm yard manure; Phosphate reserves; Phosphate fixation; Ca phosphates; pH; Liming; Mycorrhiza; Cropping systems

1. Introduction Phosphate deposits are finite resources. According to Sheldon (1982) known deposits of phosphate rock will last about 400 years at current rates of exploitation. Werner (1982) distinguishes between * Tel.: +49 641 9939161; fax: +49 641 9939199.

three categories of phosphate resources as shown in Table 1. Reserves are phosphate deposits which under the prevailing economic and technological conditions are worth mining. Phosphate resources comprise all known global phosphate deposits including those which under the present conditions cannot be mined for economic and technological reasons. Technological reasons are mainly the contamination of phosphate

Reprinted from the European Journal of Agronomy 7 (1997) 221-233

278

rock with Fe, A1, and/or Mg which disturb the processing, or logistic reasons such as the far remoteness of deposits. High Cd concentrations in phosphate rock is today also a reason for European countries not to use it as fertilizer source (Baechle and Wolstein, 1984). Speculative resources (line 3 in Table 1) are phosphates not yet discovered and the existence of which is based on geological hypothesis. This latter category also comprises deposits at great depth and deposits with low P concentrations. Since this category is very hypothetical, it should not be considered as a realistic phosphate reserve. The longevity of phosphate deposits shown in Table 1 is based on the assumption that the phosphate consumption rates per year still increase during the last two decades of the 20th century and then remain constant from the beginning of the year 2000. According to the FAO Fertilizer Yearbook the phosphate consumption rates increased until 1988/89 with a peak consumption of 38 x 10 6 t/ year and then declined with a minimum consumption of 29 x 106 t in 1994 followed by an increasing tendency. From this trend it is clear that the longevities of the reserves and the reserves plus resources are very short as compared with the history of mankind and therefore mining and consumption of phosphates should be handled with much care and any waste of this resource should be avoided. About 90% of the phosphate mined is used for the production of fertilizers (Werner, 1982). The fertility of European soils being exhausted of available phosphate in the last century by cropping without compensating for phosphates removed from the soil by crops, was much improved by phosphate fertilizer application at the end of the last century and in the first half of the 20th century (Boulaine, 1992). In developing countries there are still large areas of agricultural land with insufficient available phosphate and hence require phosphate fertilization particularly under the pressure of an increasing world population. This precarious situation demands very careful and economic use of phosphates. The flow of phosphate goes from the deposits to agricultural land and from here partially into the crops which may be eaten by humans and animals. Phosphate in crops consumed by farm animals is largely recycled with farmyard manure or slurry to the soils. Phosphate in plant parts or in animals and in animal products exported from the farm is lost for

the farm and in many cases also as potential sources of fertilizer phosphate. From the phosphate harvested in crops a high proportion is discharged into public waste systems and not returned to agricultural land. In Europe about 25% of the phosphate excreted by man is used as fertilizer (Winteringham, 1992). In developing countries the proportion of phosphate recycled to agricultural land with human excrements will decline with the increasing proportion of population living in primitive urban societies where recycling of phosphate in wastes is hardly possible. A considerable amount of available plant phosphate is also lost in agricutural land by a permanent transformation of soil phosphates into stable forms which are not available to plant roots. Even phosphate leaching into deeper soil layers not accessible to roots may occur particularly in organic soils (Munk, 1972). Deforestation and overgrazing leads to wind and water erosion and therefore also to a loss of phosphates bound to fine organic and inorganic soil particles. A substantial amount of phosphate in eroded particles and also in urban wastes finally flows into the ocean from where it cannot be recovered (Isermann, 1990). Saving phosphate is also a question of an efficient use of soil and fertilizer phosphate by farmers. In this paper, pertinent agronomic measures for improving phosphate efficiency are discussed. These measures are: fertilizing phosphate according to soil tests for available phosphate, providing an optimum soil pH for phosphate availability, using appropriate phosphate fertilizer types and practicing rotations and farming systems with crop species capable of mobilizing fixed or less soluble soil phosphates.

2. Phosphate fertilizing according to P soil tests The level of available soil phosphate should meet Table I Phosphate reserves, resources and longevity (Wemer, 1982) P reserves and resources

109 t

Longevity, years

Reserves Reserves + resources Reserves + resources + hypothetical resources

35 130 1130

85 360 3400

279

the demand of crops but should not be much higher than the optimum since otherwise major losses of available phosphate may occur. In this context the term available means phosphate which is accessible to and can be taken up by plant roots. Loss by run off of available soil phosphate with soil particles (Sharpley, 1993) will be higher when soils are richer in available phosphate. The same is true for the hazard of leaching of phosphate out of the rooting zone which may occur in organic soils (Munk, 1972) and in sandy soils particularly if overloaded with fertilizer phosphate (Isermann, 1990; Mozaffari and Sims, 1996; Peters and Basta, 1996). Consequently levels of available phosphate in soils which are above the optimum requirement for crop growth lead to a dissipation of phosphate and hence should be avoided. In the last five decades a remarkable number of P soil test methods has been developed which apply to various soil types (Hesse, 1971). In the following, particular investigations were carded out in Germany in which the 'DL method' was used. DL denotes double lactate since soils are extracted with a Ca lactate solution brought to a pH of 3.6 by the addition of HC1 (Egner, 1955; Hoffmann, 1991). Numerous earlier (Schwerdt and Jessen, 1961) and more recent field trials (Brtine and Heyn, 1984) with arable crops have shown that the DL-method is a reliable soil test precisely indicating the available soil phosphate. Only in cases in which soils were fertilized with rock phosphates the DL extract yields data which are higher than the actual available soil phosphate (Werner, 1969). For such soils the 'CAL method' is recommended. Although reliable soil tests for available phosphate are at disposition regular P soil testing covers only a limited percentage of agricultural land. In Germany with about 17 x l 0 6 ha agricultural land about 600000 P soil tests are made per year which represent only a small percentage of agricultural land (information from VDLUFA, Association of the German Research Stations). It is supposed that many soils are enriched in available phosphate and crops do not respond to further P fertilization as was reported by Arnold and Shepherd (1990) quoted after Bhogal et al., 1996) for UK. The same is true for areas with intensive livestock husbandry (Leinweber, 1996; Mozaffari and Sims, 1996; Peters and Basta, 1996). Fertilizing these soils wastes phosphate. Leinweber et al. (1994), taking representative soil samples from an

area in north Germany with intensive livestock husbandry, found low to medium DL-P concentrations in forest soils while far more samples from arable land and grassland showed high to extremely high DL-P concentrations. The frequency distribution of DL-P of Leinweber's investigation is shown in Fig. 1 for the various cropping systems. About 95% of the samples from grassland, arable land and special cultures, mainly raspberries and asparagus, had DL-P concentrations which were higher than the level above which crops do not respond to P fertilizer application. In these soils, heavily treated with slurries, phosphate is not only enriched in the top layer but also in deeper soil layers up to 1 m (Wemer et al., 1988). Since crops also feed from phosphate in deeper layers of the rootn= 116

15

Grasstand

'0 1 IIIHO 20.E

15n = 85

c

g

10-

Special cultures

5 ,i,.a

¢1 ¢J IZ

0 2015-

n = 689

Arable land

105

~l::ll~l=a B R a . . . . . .

0 !

0

-

,

i

i

200

i

600 800 m g P / k g soil

400

Fig. I. Frequencyof distribution of available phosphatein Nortbem

Germany in an area with intensive livestockhusbandry(Leinweber et al., 1994). Available phosphate measured by the extraction with a lactate solution (DL method, Egner, 1955). A concentration of 100 mg DL soluble P/kg soil may be considered as sufficient for arable crops. Specialcultures are mainlyraspberries and asparagus.

280 ing zone this deeper located available phosphate must be taken into consideration when assessing the quantity of available soil phosphate. The problem of phosphate surplus in agricultural land is comprehensively discussed by Isermann (1993). Spiertz (1991) reported that on average, milking farms in the Netherlands had an excess of 30 kg P/ha per year which lead to an enormous accumulation of P in soils. To reduce the P fertilization in such farms is not easy since the P is imported into farms with the concentrates required for feeding the high livestock rates and farmers need this high intensity for making their living because their own acreage for forage production is not sufficient. Under environmental aspects and the aspect that phosphate is a finite resource, a solution must be found for these farms.

3. Phosphate fixation

The term phosphate fixation means the transformation of plant available phosphate in soils into a nonavailable form. Two major processes may be involved in this transformation: the formation of less soluble Ca phosphates from water soluble phosphates and the adsorption of phosphate to the surfaces of soil particles. The latter process is the most important. In this context the term phosphate fixation includes phosphate occlusion brought about by adsorption by Fe lil oxides and oxyhydroxides. Formation of less soluble Ca phosphates follows the sequence: Ca dihydrogen phosphate > Ca monohydrogen phosphate > Ca octophosphate > apatite, from which Ca dihydrogen phosphate is most soluble and apatite sparingly soluble in water (Olsen et al., 1977; Sposito, 1989). The reaction sequence to less soluble phosphates is promoted by high pH and high Ca 2+ concentrations in the soil solution whereas high H + and low Ca 2+concentrations have an inverse effect. It is doubtful whether even under favorable conditions such as in calcareous soils with high pH and high Ca 2÷ concentrations in the soil solution the crystalline apatite is formed. At least this process of crystalline Ca phosphate formation proceeds at low rates (Parfitt, 1978). According to Olsen et al. (1977) it is the octophosphate which accumulates in soils with higher soil pH. The solubility of octophosphate is high enough for optimum plant supply (Olsen et al., 1977; Sposito,

1989). Therefore, the formation of less soluble Ca phosphates does not represent a major process in phosphate fixation. Uptake and metabolization of inorganic phosphate by microorganisms means a transient reduction of plant-available phosphate. Assimilation of inorganic phosphate is paralleled by the formation of inorganic phosphates from organic phosphate which partially originates from dead microbial biomass. Particularly in the rhizosphere there is a high turnover of organic phosphates into inorganic phosphate (Helal and Sauerbeck, 1984) mediated by the relatively high phosphatase concentrations near the root surface (Tarafdar and Jungk, 1987; Helal and Dressier, 1989). The most important process of phosphate fixation is represented by the specific adsorption of phosphate to soil particles such as sesquioxides, clay minerals, allophanes, calcite as well as AI and Fe humate complexes. This so called chemi-adsorption occurs by ligand exchange in which the OH-on the adsorbing surface is exchanged by a phosphate anion (Fig. 2). In the first step the phosphate is bound only with one bond to the surface (mononuclear bond) in the following step a second anion equivalent of the phosphate is bound to the surface (binuclear bond) the latter being much more stable than the mononuclear bond (Parfitt, 1978). The adsorption process is promoted by low 0

0

I

I

Fe-OH 0 I Fe-OH

-0

0

" HO

P / \

OH-

OH

'~

~

I

O

0 Fel - 0 \

.pO

0 I

mononuctear bond

0

i

Fe-- 0

0

0 P I / \OH Fe-OH OH I

O

I

Fe-O

P /

\OH

0 binuctearbond

OH"

~,1,

~ ": ~ ~

a

Fel - - 0 \ 0

/i 0 P\OH

I

Fe-OH I

0

Fig. 2. Principle of phosphate adsorption onto an adsorbingsurface. (1) Ligand exchange between the OH- of the surface and the phosphate. (2) The mononuclear bound phosphate is deprotonated. (3) The deprotonated phosphate exchanges with another OH- of the surface and a binuclear bond is formed. The reaction sequence is reversible and phosphate desorption is driven by OH-.

281

OH-concentrations (Fig. 2) which means that adsorption is particularly strong at low soil pH. This relationship is of utmost agronomic importance and is discussed in detail in the remaining part of this section. Barekzai and Mengel (1985) investigated the influence of the contact time between soil and phosphate fertilizer (superphosphate) on its P availability for Lolium perenne grown in pots. From the ten soils tested only the results of the two extreme soils are shown (Fig. 3), an acid brown earth (7% clay, DLP = 9 mg P/kg soil, in KCI solution, pH 4.6) and a subsoil from a rendzina (67% CaCO3, DL-P = 0.8 mg P/kg soil, pH 7.6). According to these characteristics the first soil should favor phosphate adsorption and the latter the formation of less soluble Ca phosphates. Both soils were very low in available phosphate. In the acid soil the contact time had a highly significant impact on the phosphate uptake of the grass. Phosphate fertilizer, given 6 months before seeding, yielded a significantly lower recovery than phosphate Papplication 6months [~J, 3rnonths[~ l.O

before seeding, ~

at seeding calcareous

acid soil

soil

30

a,b 7

a // // 0

ca. 20 E 10

-,

/. C m

-

//

;"

..

/

" J

1. c u t

~

_

r/ l

~

,,'/

..--.

f

J

m

~'-

J

~

f j

~

,,-I

.

2. c u t

I. cut

2. cut

Fig. 3. Uptake of fertilizer phosphate by Lolium perenne from an acid and a calcareous soil as related to the time of P fertilizer application: 6 months before seeding, 3 months before seeding, at seeding. Grass was cut two times (Barekzai and Mengel, 1985). 'a' denotes a significant difference at the 0.1% level between phosphate application at seeding and 6 months before; 'b' a significant difference at the 5% level between phosphate application at seeding and 3 months before seeding; 'c' a significant difference at the 5% level between phosphate application 3 months before seeding and 6 months before seeding.

fertilizer applied just before seeding. Also a 3 months contact time still had a significantly negative effect on the efficiency of the P fertilizer. This pattern found in the first cut of the grass was also evident in the second cut. In the calcareous soil the fertilizer/soil contact time had no influence on the P uptake of the grass. Obviously there was no major formation of unavailable Ca phosphates during a period of 6 months otherwise the rates of phosphate uptake by the grass should have declined with an increase in the soil/fertilizer contact time. This statement is in line with results of Olsen et al. (1977) who found that above pH 6 it is the solubility of the octocalcium phosphate which controls the phosphate availability, and added phosphates to such soils have a very high coefficient of recovery. The fast adsorption of fertilizer phosphate in acid soils was also found by Mozaffari and Sims (1996). In the acid soil, laboratory experiments of Barekzai and Mengel (1985) showed a strong phosphate adsorption (Fig. 4). In the treatment with zero contact time the adsorption curve was much flatter than the curves obtained after a contact time of 6 and 12 weeks. At the zero contact time the highest P rate resulted in a P concentration of 32 mg P/l; the same P application rate gave only a P concentration in the soil solution of about 2 mg P/l after a contact time of 6 weeks. This demonstrates the enormous reduction in P availability most likely due to specific adsorption. Further, it could be shown in laboratory experiments that phosphate availability in this acid soil was substantially increased by the incorporation of CaO into the soil (Barekzai and Mengel, 1985). From this experiment it is clear that the efficiency of soil and fertilizer phosphates depends highly on soil pH. This probably is true for all mineral soils with a potential for phosphate adsorption. The relevance of soil pH for the efficiency of phosphate fertilizer is supported by field trials. According to Werner and Wichmann (1972) the recovery of phosphate by crop uptake was much higher on neutral and calcareous soils than on acid soils. Sturm and Isermann (1978) in evaluating the phosphate recovery in long-term field experiments also found that soil pH was of high importance for P recovery. The recovery of fertilizer P was calculated from the P uptake of crops and the change in available soil phosphate since an increase in available soil P means a corresponding increase in the recovery of fertilizer P and

282

t

0

... 400 lll/6weeks

weeks

', 100

g

'

8

'

¢6

'2).

'

3'2

~

Soil solution, mg P/L Fig. 4. Phosphate buffer curve of an acid soil. Directly after P application (0 weeks), 6 weeks and 12 weeks after P application (Barekzai and Mengel, 1985). vice versa a decrease of available P a reduction of P recovery. Available soil phosphate was determined by the DL- method. In the field trials quoted by Sturm and Isermann (1978) no rock phosphate was applied and therefore the data in Table 2 give a good indication of the fertilizer phosphate recovery. The most interesting results of this investigation are shown in Table 2 which clearly demonstrate the enormous impact of soil pH on phosphate use efficiency and from which the conclusion is drawn that much phosphate can be saved if soil pH is appropriate. A high recovery of phosphate on Luvisols with a neutral to alkaline pH was also found by Jungk et al. (1993). Humates may be involved in phosphate adsorption which is particularly true for large areas of representative arable soils derived from loess. According to investigations of Gerke and Hermann (1992) and Gerke et al. (1995) Fe and AI may be adsorbed by carboxylic groups of humic acids and then adsorb

0 II

Table 2 Percentage recovery of fertilizer phosphate in long-term field trials on representative agricultural soils in relation to the lime status of soils. Recovery = P uptake of the crop + change in DL-P in the soil (Sturm and Isermann 1978) Lime status Arable soils, Arable soils, Arable soils, Arable soils, Arable soils, Grassland

phosphate as shown in Fig. 5. In the experiments of Gerke et al. (1995) phosphate adsorption was somewhat higher at pH 6.2 than 5.2. This surprising pH effect presumably is due to a higher deprotonation of humate carboxylic groups at the higher pH which may promote the adsorption of Fe hydroxides. Gerke and Hermann (1992) suggest that these P-Fe-humate complexes play a role in the turnover of fertilizer phosphate particularly in Luvisols derived from loess. Whether this kind of phosphate complex has a stronger impact on phosphate availability than that associated with sequioxides is not yet clarified. Phosphate adsorption is a particular problem in highly weathered soils of the tropics (Oxisols and Ultisols) because of their high phosphate adsorption potential. For phosphate melioration they require high P fertilizer rates in the range of 170 kg P/ha (Haynes, 1984). Most of these soils are acid and require liming which does not improve phosphate availability in all cases. Liming may induce polymerization of A1 cation species which because of their high positive charge are strong phosphate adsorbers (Haynes, 1984). According to Hauter (1983) the decrease of phosphate availability due to liming of Oxisols is associated with their very low pH (3.7-4.4) while at a soil pH of 5.5 liming had a beneficial effect on phosphate availability. Sims and Ellis (1983) reported that lim-

% Recovery very well supplied with lime well supplied with lime moderately supplied with lime poorly supplied with lime poorly supplied dry locations

80 70 65 60 50 80

C

0 F¢

N

P /

Fig. 5. Adsorption of phosphate to an humate Fe complex. The Fem is adsorbed onto the humate with a covalent bond and a coordinate bond (modified after Gerke and Hermann, 1992).

283

ing an Ultisol increased the available soil P and enhanced P uptake by oats considerably. In order to save fertilizer phosphate on these strongly phosphate fixing soils band placement of fertilizers is recommended (Wemer and Scherer, 1995). As shown earlier (Fig. 2) adsorption is an exchange of ligands and with an increase in soil pH the OHconcentration increases, OH-competing with phosphates for binding sites at the adsorbing surface. Organic anions may also compete with phosphate for binding sites (Parfitt, 1978). Fox et al. (1990) found a number of organic anions capable of replacing phosphates from adsorbing surfaces. Citrate seems to be a very potent competitor for adsorbed phosphate (Gerke, 1994). Under anaerobic conditions soluble phosphate increased in the soil solution (Welp et al., 1983) mainly due to the reduction of complex bound Fem associated with the release of soluble phosphate. As

Thomas slag

o

•~

R°ck,Ph°sph"

20

oOo

E

~¢"'..-"~

~

w,thout P

I

/,ram Distance from the

root surface

Fig. 6. Water soluble phosphate originating from various phosphate fertilizers in the rhizosphere of young rape. The phosphate fertilizer had been dressed in a 10 years lasting field trial. The horizontal lines designate the level of water soluble phosphate in the bulk soil. PARP, partially acidulated rock phosphate (Steffens, 1987). Thomas slag is a non-crystalline, non-water soluble phosphate fertilizer which gradually dissolves in soils and therefore is well available to plant roots. Rock phosphate is a crystalline phosphate fertilizer (apatite), non-soluble in water which is dissolved in acid soils. PARP, partially acidulated rock phosphate which consists of about to 50% of rock phosphate and 50% of water soluble phosphate.

was shown by Sah and Mikkelsen (1986) occluded phosphates may be solubilized under anaerobic conditions because of the reduction of Fe m to Fe 2÷ (the Roman superscript indicating an Fe complex, the Arabic superscript a dissolved Fe ion). The process is of particular importance for flooded rice soils.

4. Organic phosphate and mycorrhiza in soil cropping systems In arable land the concentration of organic phosphate is in the order of 50% of total phosphate in the upper soil layer and in grassland soils the proportion of organic phosphate may be even higher (Sharpley, 1985). A substantial part of organic phosphate, up to 100 kg P/ha, may be fixed in microbial biomass (Brookes et al., 1984). Phosphate thus immobilized may easily be mineralized and hence become available for crops. Sharpley (1985) reported that there is a seasonal variation in available organic soil phosphate decreasing in spring with crop growth and increasing in late autumn and winter. This pattern was particularly distinct in soils not treated with inorganic phosphate fertilizer showing that the plants drew phosphate from this organic pool. In calcareous soils, the phosphate of the soil solution is mainly present in organic form (Dalai, 1977) and therefore in these soils phosphate transport to plant roots is mainly brought about by organic phosphates which may be easily mineralized in the plant rhizosphere enriched with phosphatases (Tarafdar and Claassen, 1988; Dou and Steffens, 1993). About half of the organic phosphates in soils is present as myo-inositol-phosphates from which the inositol-hexaphosphate is adsorbed to sequioxides similar as inorganic phosphate (Dalai, 1977). The adsorption of inositol-hexaphosphate is relatively strong since the molecule has six phosphate groups which may be bound to soil particles. Diminishing the numbers of phosphate groups bound to inositol decreases the possibility of phosphate adsorption and thus improves phosphate availability (Evans, 1985). Myo-inositol-2-monophosphate is virtually not adsorbed (Evans, 1985), and quite mobile in soils (Dou and Steffens, 1993). In addition inositolhexaphosphate can also form rather insoluble salts with Ca 2÷and Mg 2÷, a process which may affect phosphate availability. Other organic phosphates such as

284

phospholipids and nucleotide phosphates do not accumulate in soils as they are easily mineralized (Dalal, 1977; Tarafdar and Claassen, 1988). Hence the large pool of organic soil phosphate is potentially available for plants. This is also true for the non-soluble organic phosphate from which a great part is present in the form of microbial biomass and which will be mineralized after the death of microorganisms. Therefore, in contrast to the fixed inorganic phosphate the immobilized organic phosphate represents a potential pool of available phosphate and measures which promote the formation of organic phosphate in soils and may restrict the fixation of inorganic phosphate and thus contribute to an efficient use of soil phosphates. Sources of organic phosphates in soils are plant residues, green manure, microbial biomass, and farm yard manure (FYM). For this reason cropping systems have a distinct impact on the content of organic phosphates in soils as well as on the assimilation of inorganic phosphates by fungi and bacteria and the mineralization of organic phosphates by phosphatases. Oberson et al., 1993, 1996; reported that regular application of FYM to soil increased the organic phosphate content. This effect may be due to organic phosphate present in the FYM but also to inorganic phosphate being assimilated by soil microorganisms after FYM application. The latter especially raised the ATP concentration which, according to the authors means an increase in microbial biomass. The impact of FYM on the concentration of ATP in the upper soil layer is shown in Table 3, from the work of Oberson et al. (1993). It is evident that in all treatments receiving FYM the ATP concentration was significantly increased which means that FYM had a beneficial effect on microbial biomass Table 3 Effect of FYM on the ATP concentration in soils. Soil samples taken at ear emergence of winter wheat (Oberson et al., 1993) P fertilizer kg P/ha

No FYM 80% FYM + 20% mineral P 40% FYM + 60% mineral P 100% mineral P

Rate of P appl. kg P/ha per year

ATP ~tg/kg soil*

28 31

843 a 1217c 1160 bc

47

1006abc

46

945 ab

development and hence on the storage of potentially available phosphate. Parallel with the increase of microbial biomass the acid phosphatase activity was increased by FYM application which means that also the enzyme activity rendering organic phosphate into a form directly taken up by plant roots was promoted. The positive effect of FYM on the efficient use of soil and fertilizer phosphate availability is enhanced by organic anions produced during the decomposition of organic matter. They compete with inorganic phosphate for adsorption sites and thus reduce the fixation of phosphate (Werner and Scherer, 1995). In addition FYM may improve soil structure and favor root growth and thus the exploitation of soil phosphates by roots (Keita and Steffens, 1989). Farms producing FYM frequently also grow arable forage crops such as red clover and alfalfa which not only contribute to the nitrogen status of soils by symbiotic N2 fixation but they also may exploit fixed soil phosphate by the excretion of root exudates as was shown for red clover excreting citrate which mobilizes adsorbed soil phosphate (Gerke, 1994). Rotations with diverse crop species generally will contribute to a better exploitation of soil phosphates. Mycorrhization of plant roots may considerably improve the accessibility of soil phosphate to plants mainly by increasing the contact surface between the soil matrix and the mycorrhized plant root. This is particularly true for leguminous species (Barea and Acon-Aguilar, 1983). The problem with mycorrhiza exploiting soil phosphate for the host plant is the high specificity between the host plant and the endomycorrhizal fungi (Lioi and Giovannetti, 1987; Diederichs, 1991). Inoculation of soils with the appropriate fungi still meets with difficulty (Hall, 1987). If the fungi/root symbiosis is efficient remarkable crop yield increases may be obtained due to a better exploitation of soil phosphate (Hall, 1984).

5. Phosphate fertilizer types Phosphate fertilizer types differ in their solubility with the most important difference between amorphous and crystalline forms. The latter comprises the rock phosphates and partially acidulated rock phosphate (PARP) which still contains a portion which is crystalline and represents apatite. The solu-

285

bility of fluoro-apatite is given by the following equation: Cas(PO4)3F + 4H ÷ ~

5Ca 2÷ + 3HPO~- + HF

From the equation it is evident that high Ca 2÷ and phosphate concentrations hamper, and increasing H ÷ concentrations in the soil solution promote the dissolution of apatites. Solubility of apatites (rock phosphates) depends also on the degree of isomorphic substitution of PO 3- by COl- and is higher the more phosphate is substituted by carbonate (Anderson et al., 1985). Generally in acid soils pH < 5.0 (measured in CaCI2 solution) the efficiency of rock phosphate is as high as that of acidulated phosphate (Mengel, 1986). In soils with a higher pH, however, the efficiency is poorer and may be even nil. In such cases the application of rock phosphate means a waste of the phosphate resource. Steffens (1994) investigated a number of representative arable soils for their availability of phosphate originating from various fertilizer types. Phosphate release rates obtained by repeated extraction of soils with electro-ultrafiltration (EUF) followed the Elovich equation and reflected well the phosphate availability of various fertilizer types for crops. The agreement of the released phosphate with the Elovich equation means that the P release rates declined with the number of extractions. Highest release rates were obtained from basic slag (Thomas phosphate) and superphosphate and lowest rates from rock phosphate. This pattern of phosphate fertilizer solubility was also found in the rhizosphere of rape as shown in Fig. 6 (Steffens, 1987) from which can be seen that partially acidulated phosphate took an intermediate position. This means that mainly the water soluble portion contributed to P solubility. For this reason also the application of partially acidulated rock phosphates on soils with a poor solubility for apatite means a waste of phosphate (Resseler and Wemer, 1989). The pretention that apatitic phosphate will render soluble in soils by time is only correct if soils are acid (Renno and Steffens, 1985). Such soils, however, if not organic soils, should be limed in order to improve the availability of adsorbed phosphate as discussed above. If lime is not available rock phosphates may be an alternative choice. The poor performance of apatitic phosphate found in representative arable soils in Europe (Mengel, 1986) is in

agreement with experiences made on laterite soils in Western Australia (Bolland et al., 1988; Bolland and Gilkes, 1990). Also in these field trials the direct and residual effect of apatitic fertilizer was poor as compared with superphosphate. Only on the humic sandy podsols in south Western Australia with annual rainfall > 800 mm Bolland (1996) found a superiority of apatitic fertilizers as compared with superphosphate. Under these particular conditions superphosphate may be leached out from the top soil layer. Proton excretion and mycorrhizal colonization of plant roots may contribute to the solubilization of apatite. According to Hauter and Steffens (1985) the high proton excretion of red clover roots, typically for symbiotically living leguminous species (Mengel, 1994), contributed to the dissolution of rock phosphate. An interesting effect of mycorrhizal infection was found by Steffens (1992). A farmer having applied rock phosphate for years on a calcareous soil finally ended in a severe phosphate deficiency of sugar beets. Field trials carried out on this soil with different phosphate fertilizer types gave the results shown in Table 4. It is evident that the response of sugar beets to rock phosphate application was nil in contrast to superphosphate which produced a remarkable yield increase. In sunflowers, however, the effect of rock phosphate was as high as the effect of superphosphate; in wheat the high rate of rock phosphate was as good as the low rate of superphosphate. Sunflower roots were well colonized with mycorrhiza, sugar beet roots were not. This example shows the beneficial effect of mycorrhiza on the acquisition of rock phosphate. The decision for a farmer to apply Table 4 Effect of superphosphate and hyperphos (phosphate rock) on the beet yield and grain yield of variouscrops. The calcareous soil had a pH of 7.4 (after Steffens, 1992) Phosphate, applied

Sugar beet, Sunflowers, Wheat, 1987 1988 1989 Yield in t/ha

No phosphate Hyperphosphate low rate Superphosphate low rate Hyperphosphate high rate Superphosphate high rate Least significant difference, 5%

32.8 32.3 46.5 32.3 50.0 7.9

3.77 4.35 4.19 4.18 4.39 0.31

6.59 6.81 7.33 7.55 7.51 0.51

286

rock phosphate depends therefore also on the crop species and on the species in the rotation. If the rotation comprises a sugar beet crop the level of available soil phosphate should be maintained by applying acidulated phosphates or amorphous forms of phosphate such as sinterphosphate (CaNa phosphate produced in soda production) or Thomas phosphate. In many cases 50% of an amorphous phosphate will give the same yield as 100% of the apatitic phosphate as shown in Table 4. Hence from the viewpoint of saving the phosphate resource 50% of the amorphous phosphate is the better choice.

6. Perspectives As shown above (Table 4) mycorrhizal symbiosis with plant roots represents an interesting mechanism for the efficient use of soil phosphates. This potential will be only used if the available soil phosphate level is not too high. The high specificity in the relationship between fungus species/host plant species (Hall, 1987; Lioi and Giovannetti, 1987; Diederichs, 1991) and the problem of inoculating soils with the proper fungus species still represents severe obstacles for the use of mycorrhiza in practical farming and need further scientific and technical efforts. Excretion of organic anions and protons by plant roots under the conditions of insufficient phosphate supply is a further mechanism which merits attention. Proton excretion may help to solubilize apatitic phosphates (Hoffland et al., 1990), and the excretion of organic anions to desorb adsorbed phosphate (Hedley et al., 1982). The latter authors were the first who found that Brassica napus was capable of responding to an insufficient phosphate supply by an enhanced excretion of H+. Protons alone, however, would rather depress the availability of adsorbed phosphate than mobilize it. In addition to the proton excretion, rape roots also excrete organic anions, especially malate which may desorb adsorbed soil phosphate (Hoffland et al., 1989). This response of Brassica napus to low phosphate supply was not found with Lolium multiflorum (Ruiz, 1992). Zhyu et al. (1990) found that rice plants excrete citrate under the conditions of insufficient phosphate supply and that Japonica species were more efficient in citrate excretion than Indica species. Ae et al. (1990) reported that pigeon peas

have no particular potential to exploit Ca phosphates but they are capable of excreting a tartrate derivate (piscidic acid) which mobilizes adsorbed soil phosphate. Of particular interest are the proteoid roots of Lupinus albus capable of excreting large amounts of citrate (Gardner and Parbery, 1982) which mobilize insoluble soil phosphate the effect being due to the citrate and not to the release of H+ (Gardner et al., 1983). According to Dinkelacker et al. (1989) citrate excreted by proteoid roots of Lupinus albus may also solubilize Ca phosphate by chelating the Ca 2+of insoluble Ca phosphate. Red clover is also a potential species in excreting citrate by roots (Gerke, 1994) the quantities being released were in the same range as citrate excreted by proteoid roots of Lupinus albus (Gerke et al., 1994).

7. Conclusions A more efficient use of soil and fertilizer phosphates demands agronomic and scientific efforts. From the agronomic measures such as the selection of the appropriate phosphate fertilizer type, adjusting fertilizer rates to soil tests and liming soils to an optimum pH level may be easily implemented by European farmers since lime and various phosphate fertilizer types are available. In areas with excessively high levels of available soil phosphate due to intensive livestock farming, cropping systems should be developed with a closer integration of crop and animal production so that phosphates excreted by farm animals are efficiently used for crop production. Such farming systems should comprise a broader diversity of crop species in the rotation including forage crops which are particularly efficient in exploiting soil phosphates by mycorrhiza and/or excretion of organic anions by roots. The implementation of such farming systems substituting the intensive livestock production needs not only agronomic but particularly political and economical measures. Scientific and technical efforts are required for selecting appropriate endomyccorhizal fungi ecotypes and practicable soil inoculation techniques. The physiological mechanism by which plants respond to an insufficient phosphate supply such as the secretion of organic anions by plant roots needs elucidation.

287

References

soil as influenced by dissolved organic compounds. Soil Sci., 140:251-255.

Ae, N., Arihara, J., Okada, K., Yoshihara, R. and Johansen, C., 1990. Uptake mechanism of iron-associated phosphorus in pigeon pea growing on Indian Alfisols and its significance to phosphorus availability in cropping systems. In: M. Koshino (Editor), Transactions, 14. Intern. Congr. Soil Science Vol. II. Kyoto, pp. 164-169. Anderson, D.L., Kussow, W.R. and Corey, R.B., 1985. Phosphate rock dissolution in soil: indications from plant growth studies. Soil Sci. Soc. Am. J., 49: 918-925. Baechle, H.T. and Wolstein, F., 1984. Cadmium components in mineral fertilizers. The Fertilizer Soc. Proc. No 226, London. Barea, J.M. and Acon-Aguilar, C., 1983. Mycorrhizas and their significance in nodulating nitrogen-fixing plants. Adv. Agron., 36: 1-54. Barekzai, A. and Mengel, K., 1985. Alterung von wassedSslichem Diingerphosphat bei verschiedenen Bodentypen. Z. Pflanzenerniihr. Bodenk., 148: 365-378. Bhogal, A., Young, D.S., Ralph, R., Bradley, S. and Craigon, J., 1996. Modelling the residual effect of phosphate in the Ropsley (UK) field trial 1978-1990. Fert. Res., 44: 27-36. Bolland, M.D.A., 1996. Effectiveness of Ecophos compared with single and coastal superphosphate. Fert. Res., 45: 37-49. Bolland, M.D.A. and Gilkes, R.J., 1990. Rock phosphates are not effective fertilizers in Western Australia soils: a review of one hundred years of research. Fert. Res., 22: 79-95. Bolland, M.D.A., Gilkes, R.J. and Allen, D.G., 1988. The residual value of superphosphate and rock phosphates for lateritic soils and its evaluation using three soil phosphate tests. Fert. Res., 15: 253-280. Boulaine, J., 1992. Histoire de r Agronomic en France. TEC-DOC, Pads, London, New York, pp. 337-346. Brookes, P.C., Powlson, D.S. and Jenkinson, D.S., 1984. Phosphorus in the soil microbial biomass. Soil Biol. Biochem., 16: 169-175. Brtine, H. and Heyn, J., 1984. P-Dtingeempehlung for die Praxis. Die Wirkung verschiedener Phosphat-Diinger-Formen in mehrjiihrigen Feldversuchen. DLG-Mitteil., 2: 80-83. Dalai, R.C., 1977. Soil organic phosphorus. Adv. Agron., 29: 83117. Diederichs, C., 1991. Influence of different P sources on the efficiency of several tropical endomycorrhizal fungi in promoting the growth of Zea mays L. Fert. Res., 30: 39-46. Dinkelacker, B., RiSrnheld, V. and Marschner, H., 1989. Citric acid excretion and precipitation of calcium citrate in the rhizosphere of white lupin (Lupinus albus L.). Plant Cell Environ., 12: 285292. Dou, H. and Steffens, D., 1993. Mobilitiit und Pflanzenverfiigbarkeit von Phosphor aus organischen und anorganischen Formen in der Rhizosph~e von Lolium perenne. Z. Pflanzenern~r. Bodenk., 156: 279-285. Egner, H., 1955. Neue Beitr~ige zur chemischen Bodenuntersuchung unter besonderer Bertlcksichtigung der Laktatmethode. Landwirtsch. Forsch. Sonderheft, 6: 28-32. Evans, A., 1985. The adsorption of inorganic phosphate by a sandy

Fox, T.R., Comerford, N.B. and McFee, W.W., 1990. Phosphorus and aluminium release from a spodic horizon mediated by organic acids. Soil Sci. Soc. Am. J., 54: 1763-1767. Gardner, W.K. and Parbery, D.G., 1982. The acquisition of phosphorus by Lupinus albus L. I. Some characteristics of the soil/ root interface. Plant Soil, 68: 19-32. Gardner, W.K., Barber, D.A. and Parbery, D.G., 1983. The acquisition of phosphorus by Lupinus albus L. III. The probable mechanism by which phosphorus in the soil/root interface is enhanced. Plant Soil, 70: 107-124. Gerke, J., 1994. Kinetics of soil phosphate desorption as affected by citric acid. Z. Pflanzenern~r. Bodenk., 157: 17-22. Gerke, J. and Hermann, R., 1992. Adsorption of orthophosphate to humic-Fe-complexes and to amorphous Fe-oxide. Z. Pflanzenern~ihr. Bodenk., 155: 233-236. Gerke, J., Meyer, U. and RSmer, W., 1995. Phosphate, Fe and Mn uptake of N2-fixing red clover and ryegrass from an oxisol as affected by P and model humic substances application. Z. Pflanzenerniihr. Bodenk., 158, 261-268. Gerke, J., Rtimer, W. and Jungk, A., 1994. The excretion of citric and malic acid by proteoid roots of Lupinus albus L.; effects on soil solution concentrations of phosphate, iron and aluminium in the proteoid rhizosphere in samples of an oxisol and a luvisol. Z. Pflanzenemiihr. Bodenk., 157, 289-294. Hall, J.R., 1984. Field trials assessing the effect of inoculating agricultural soils with endomycorrhizal fungi. J. Agric. Sci., 102, 725-73 I. Hall, J.R., 1987. A review of VA mycorrhizal growth responses in pastures. Angew. Botanik, 61: 127-134. Hauter, R., 1983. Phosphatmobilisierung in Abh~gigkeit vom pH des Bodens unter besonderer Beriicksichtigung der Rhizosph~e. Ph.D. Thesis FB Ern~rungswissenschaften, Justus-Liebig-Universit~it, Giessen. Hauter, R. and Steffens, D., 1985. Einfluss einer mineralischen und symbiontischen Stickstofferniihrung auf Protonenabgabe der Wurzein, Phosphataufnahme und Wurzelentwicklung yon Rotklee. Z. Pflanzenemiihr. Bodenk., 148, 633-646. Haynes, R.J., 1984. Lime and phosphate in the soil-plant system. Adv. Agron., 37, 249-315. Hedley, M.J., White, R.E. and Nye, P.H., 1982. Plant-induced changes in the rhizosphere of rape (Brassica napus Var. Emerald) seedlings. III. Changes in L value, soil phosphate fractions and phosphatase activity. New Phytol., 91: 45-56. Helal, H.M. and Dressier, A., 1989. Mobilization and turnover of soil phosphorus in the rhizosphere. Z. Pflanzern~r. Bodenk., 152: 175-180. Helal, H.M. and Sauerbeck, S.B., 1984. Influence of roots on C and P metabolism in soil. Plant Soil, 76: 175-182. Hesse, P.R., 1971. A Textbook of Soil Chemical Analysis. John Murray, London. Hoffland, E., Findenegg, G.R. and Nelemans, J.A., 1989. Solubilization of rock phosphate by rape. II Local root exudation of organic acids as a response to P-starvation. Plant Soil, 113: 161-165. Hoffland, E., Findenegg, G.R., Leffelaar, P.A. and Nelemas, J.A.,

288

1990. Use of a simulation model to quantity the amount of phosphate released from rock phosphate by rape. In: M. Koshino (Editor), Transactions Vol. II. Intern. Congr. Soils Sci. Kyoto, pp 170-175. Hoffmann, G., 1991. Methodenbuch, B and I, Die Untersuchung von Brden. VDLUFA-Verlag, Darmstadt, 1991. lsermann, K., 1990. Share of agriculture in nitrogen and phosphorus emissions into the surface waters of Western Europe against the background of their eutrophication. Fert. Res., 26: 253-269. Isermann, K., 1993. Territorial, continental, and global aspects of C, N, P and S emissions from agricultural ecosystems. In: R. Wollart, F.T. Mackenzie and L. Chou (Editors) Interactions of C, N, P and S Biochemical Cycles and Global Change. SpringerVerlag, Berlin, Heidelberg, pp 79-121. Jungk, A., Claassen, N., Schulz, V. and Wendt, L, 1993. Pflanzenverfiigbarkeit der Phosphatvorr~ite ackerbaulich genutzter Btiden. Z. Pflanzenernt.ihr. Bodenk., 156: 397-406. Keita, S. and Steffens, D., 1989 Einfluss des Bodengeftiges auf Wurzelwachstum und Phosphataufnahme von Sommerweizen. Z. Pflanzenemahr. Bodenk., 152, 345-351. Leinweber, P., 1996. Phosphorus fractions in soils from an area with high density of livestock population. Z. Pflanzenern~ihr. Bodenk., 159:251-256. Leinweber, P., Geyer-Wedell, K. and Jordan, E., 1994. Phosphorgehalte von Brden in einem Landkreis mit hoher Konzentration des Viehbestandes. Z. Pflanzenern~ihr. Bodenk., 157, 383-385. Lioi, L. and Giovannetti, M., 1987. Variable effectivity of three vesicular-arbuscular mycorrhizal endophytes in Hedysarum coronarum and Medicago sativa. Biol. Fertil. Soils, 4: 193-197. Mengel, K., 1986. Umsatz im Boden und Ertragswirkung rohphosphathaltiger Dtingemittel. Z. Pflanzenern/ihr. Bodenk., 149: 674-690. Mengel, K., 1994. Symbiotic dinitrogen fixation - its dependence on plant nutrition and its ecophysiological impact. Z. Pflanzenernahr. Bodenk., 157, 233-241. Mozaffari, M. and Sims, J.T., 1996. Phosphorus transformations in poultry litter-amended soils of the Atlantic Coastal Plain. J. Environ. Qual., 1357-1365. Munk, H., 1972. Zur vertikalen Wanderung mineralischer Phosphors~iure bei starker PhosphatdiJngung. Landwirtsch. Forsch., 27/I SH: 192-199. Oberson, A., Besson, J.M., Maire, N. and Sticher, H., 1996. Microbiological processes in soil organic phosphorus transformations in conventional and biological cropping systems. Biol. Fertil. Soils, 21: 138-148. Oberson, A., Fardeau, J.C., Besson, J.M. and Sticher, H., 1993. Soil phosphorus dynamics in cropping systems managed according to conventional and biological agricultural methods. Biol. Fertil. Soils, 16:111-117. Olsen, A.R., Bowman, R.A. and Watanabe, F.S., 1977. Behaviour of phosphorus in the soil and interactions with other nutrients. In: J.R. Ansiaux, J.O. Comas, R. Gervy, D.A. Lewis and L.J. Carpentier (Editors), Phosphorus in Agriculture. I.S.M.A. Ltd., Paris, pp. 31-46. Parfitt, R.L., 1978, Anion adsorption by soils and soil materials. Adv. Agron., 30: 1-50.

Peters, J.M. and Basta, N.T., 1996. Reduction of excessive bioavailable phosphorus in soils by using municipal and industrial wastes. J. Environ. Qual., 25: 1236-1241. Renno, J.M. and Steffens, D., 1985. Einfluss von Dtingungszeitpunkt und Phosphatform auf Alterung und Verftigbarkeit von Phosphat. VDLUFA-Schriftenreihe, 16:171-178. Resseler, H. and Werner, W., 1989. Properties of untreated rock residues in partially acidulated phosphate rocks affecting their reactivity. Fert. Res., 20: 135-142. Ruiz, L., 1992. Mobilisation du phosphore des apatites dans la rhizosphere. Role de l'excrrtion de protons par les racines. Ph.D. Thesis, Universit6 des Sciences et Techniques du Languedoc, Montpellier. Sah, R.N. and Mikkelsen, D.S., 1986. Effects of anaerobic decomposition of organic matter on sorption and transformations of phosphate in drained soils: Effects on amorphous iron content and phosphate transformation. Soil Sci., 142: 346-351. Schwerdt, K. and Jessen, W., 1961. Die Hrhe des Ertragszuwachses bei P-Steigerungsversuchen in Abh~ingigkeit vom Laktat-Wert und der Stellung in der Fruchtfolge bei Getreide. Landwirtsch. Forsch., 14: 148-159. Sharpley, A.N., 1985. Phosphorus cycling in unfertilized and fertilized agricultural soils. Soil Sci. Soc. Am. J., 49: 905-911. Sharpley, A.N., 1993. Assessing phosphorus bioavailability in agricultural soils and runoff. Fert. Res., 36: 259-272. Sheldon, R.P., 1982. Phosphat - der unentbehrliche Rohstoff. Spektrum der Wissenschaft 8, 16-23. Sims, J.T. and Ellis, B.G., 1983. Adsorption and availability of phosphorus following the application of limestone to an acid, aluminous soil. Soil Sci. Soc. Am. J., 47: 888-893. Spiertz, J.H.J., 1991. Integrated agriculture in the Netherlands. In: F. Doazan (Editor) Quelles Fertilisations Demain? COMIFER, Strasbourg, pp. 52-63. Sposito, G., 1989. The Chemistry of Soils. Oxford University Press, New York, Oxford, pp. 93-96. Steffens, D., 1987. Einfluss langj~ihriger Dtingung mit verschiedenen Phosphatdtingerformen auf die Phosphatverftigbarkeit in der Rhizosph~ire von Raps. Z. Pflanzenern~ihr. Bodenk., 150: 75-80. Steffens, D., 1992. Ertragswirksamkeit und Umsatz yon apatitischen PhosphatdiJngemittein im Vergleich zu vollaufgeschlossenen Phosphatformen im Boden unter besonderer Berticksichtigung von Standort und pflanzenphysiologischen Faktoren. Habilitation Thesis FB Agrarwissenschaften. JustusLiebig-Universit~it, Giessen. Steffens, D., 1994, Phosphorus release kinetics and extractable phosphorus after long-term fertilization. Soil Sci. Soc. Am. J., 58, 1702-1708. Sturm, H. and lsermann, K., 1978. Oberlegungen zur langfristigen Ausnutzung von Mineraldtinger-Phosphat auf AckerbSden. Landw. Forsch. Sonderh., 35: 180-192. Kongressband. Tarafdar, J.C. and Claassen, N., 1988. Organic phosphorus compounds as a source of P for higher plants through phosphatases produced by plant roots and microorganisms. Biol. Fert. Soils, 5: 308-312. Tarafdar, J.C. and Jungk, A., 1987. Phosphatase activity in the rhizosphere and its relation to the depletion of soil organic phosphorus. Biol. Fert. Soils, 3: 199-204.

289

Welp, G., Herms, U. and Brtimmer, G. 1983. Einfluss von Bodenreaktion, Redoxbedingungen und organischer Substanz auf die Phosphatgehalte der Bodenl6sung. Z. Pflanzenem~ihr. Diing. Bodenk., 146: 38-52. Werner, W., 1969. Kennzeichnung des pflanzenverfUgbaren Phosphates nach mehrj~ihdger Diingung mit verschiedenen Phosphaten. Z. Pflanzenern~r. Bodenk., 122: 19-32. Werner, W., 1982. Rohphosphate - zukiinflige Verfiigbarkeit als Rohstoff fiir Diingemittel. Chem. Industrie, 3: I-4. Werner, W. and Scherer, H.W., 1995. Quantity/intensity relation and phosphorus availability in south Brasilian latosols as affected by form and placement of phosphorus and farmyard manure. In: Date et al. (Editors) Plant Soil Interactions at Low pH. Kluwer, Dordrecht, pp 129-133. Werner, W. and Wichmann, H., 1972. Untersuchungen zur Pflan-

zenverfiigbarkeit des durch langjiihrige Phosphatdtingung angereichenen Phosphates. Z. Pflanzenem~ihr. Bodenk., 133: 4-17. Werner, W., Fdtsch, F. and Scherer, H.W., 1988. Einfluss langj~ihdger GUllediingung auf den N~ihrstoffhaushalt des Bodens. 2. Mitteilung: Bindung und L/~slichkeitskriteden der Bodenphosphate. Z. Pflanzenem~r. Bodenk., 15 l: 63-68. Winteringham, F.P.W., 1992. Biogeochemical cycling of phosphorus. In: A. Cottenie (Editor), Phosphorus Life and Environment - from Research to Application. World Phosphate Institute, Casablanca, Morocco, pp. 325-336. Zhyu, L., Weiming, S. and Xiaohui, F., 1990. In: M. Koshino (Editor), The Rhizosphere Effects of Phosphorus and Iron in soils. Transactions, 14. Intern. Congr. Soil Sci., Vol. II, Kyoto, pp. 147-152.

This Page Intentionally Left Blank

Section 5 DESIGNING FARMING SYSTEMS A methodical way of prototying integrated and ecological arable farming systems (I/EAFS) in interaction with pilot farms P. Vereijken .................................................................................................................................................

293

Reprinted from the European Journal of Agronomy 7 (1997) 235-250 The Log~rden project: development of an ecological and an integrated arable farming system C.A. Helander ..............................................................................................................................................

309

Integrated crop protection and environment exposure to pesticides: methodes to reduce use and impact of pesticides in arable farming F. G. Wijnands .............................................................................................................................................. Reprinted from the European Journal o f Agronomy 7 (1997) 251-260

319

Use of agro-ecological indicators for the evaluation of farming systems C. Bockstaller, P. Girardin and H.M.G. van der Werf. ................................................................................

329

Reprinted from the European Journal of Agronomy 7 (1997) 261-270 Model-based explorations to support development of substainable farming systems: case studies from France and the Netherlands I'V.A.H. Rossing, J.M. Meynard and M.K. van Ittersum ...............................................................................

339

Reprinted from the European Journal of Agronomy 7 (1997) 271-283 Learning for substainable agriculture B.M. Somers .................................................................................................................................................

353

This Page Intentionally Left Blank

© 1997 ElsevierScience B. E All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

293

A methodical way of prototyping integrated and ecological arable farming systems (I/EAFS) in interaction with pilot farms P. Vereijken* Department of Crop WeedScience, Research institutefor Agrobiology and Soil Fertility, P.O. Box 14, 6700AA Wageningen, The Netherlands

Accepted 13 June 1997

Abstract A methodical way of prototyping integrated and ecological arable farming systems (I/EAFS) in interaction with pilot farms is presented. It concerns a comprehensive and consistent approach of 5 steps. Step l is establishing a hierarchy of objectives considering the shortcomings of current farming systems in the region. Step 2 is transforming the objectives in a set of multiobjective parameters, to quantify them and establishing a set of multi-objective farming methods to achieve them. Step 3 is designing a theoretical prototype by linking parameters to farming methods and designing the methods in this context until they are ready for initial testing. Step 4 is laying out the prototype on at least l 0 pilot farms in appropriate variants and testing and improving the prototype (variants) until the objectives, as quantified in the set of parameters, have been achieved (after repeated layout). Step 5 is disseminating the prototype (variants) to other farms with gradual shift in supervision from researchers to extensionists. This methodical method is being elaborated and tested by a European network of more than 20 research teams, sponsored by the European Union (AIR-concerted action). The teams express their achievements in a consistent set of 6 parts of an identity card of their prototype. The 6 parts of the EAFS-prototype of the author's team are presented to illustrate the methodical approach. Part 6 presents the state of the art. It shows that the desired results are progressively being achieved, which may be considered as the best proof of the effectiveness of prototyping. © 1997 Elsevier Science B.V. Keywords: Sustainable farming systems; Crop rotation; Nutrient management; Nature and landscape r

1. Introduction The European Union (EU) is facing an agricultural crisis with two major symptoms: deterioration of rural income and employment and deterioration of environment, nature and landscape. The basic mechanism is a continuous intensification causing surplus production and price fall on the one hand and ecological deterioration on the other hand. Therefore, a crucial ques* Tel.: +31 0317 475970; fax: +31 0317 475952.

tion for the Common Agricultural Policy (CAP) is to alleviate the symptoms of intensification on the short term and to find a sustainable solution on the long term. In the early 1990s, various EU countries started promoting integrated farming systems to alleviate the agricultural crisis, when drastic reductions in inputs of pesticides and fertilisers were achieved with initial prototypes on experimental farms. Subsequently, in 1993 the EU decided to sponsor a network of research teams on prototyping integrated arable farming systems (IAFS). The setting up of the network should be

Reprinted from the European Journal of Agronomy 7 (1997) 235-250

294

combined with development and standardisation of the methods of prototyping in a concerted action within the third EU framework programme for agricultural research. The objective of this paper is to explain and illustrate the methods of prototyping as developed by the research network. Most research teams joined the network to develop IAFS prototypes feasible for the main group of farms. This group must try to be competitive on the world market, based on high and efficient production, and this gives only limited scope for pursuing non-marketable objectives such as environment, and nature/landscape. Therefore, a more consistent integration is needed for such long term objectives. Consequently, many research teams also or exclusively develop an IAFS for the long term, albeit that this IAFS is as yet only feasible for pilot groups of farms. Contrary to shortterm IAFS, these long-term IAFS place income/profit subordinate to environment, and rely on ecologically aware consumers willing to pay premium prices for food products with high added value and a credible label. In the short-term IAFS, chemical crop protection is minimised to the benefit of the environment (integrated crop protection). In the long-term IAFS, chemical crop protection is fully replaced by a package of non-chemical measures, to achieve ambitious objectives in environment, nature/landscape and quality and sustainability of food supply. So, long-term IAFS are based more on ecological awareness and knowledge than short-term IAFS. Therefore, our prototypes of long-term IAFS are simply called EAFS (ecological arable farming systems), and short-term IAFS are referred to as IAFS. Organic systems can be considered to be a forerunner of EAFS, but they have no quantified objectives in environment and nature/landscape and as a result, they need to be considerably improved to become acceptable to the majority of consumers. Nevertheless, organic farming has a strategic significance to Europe because it is the first example of the market model of shared responsibility of consumers and producers for the rural areas. Therefore, many research teams are collaborating with a pilot group of organic farms which have primarily been selected for their willingness to achieve more than is required by current minimal guidelines of the EU organic label. Selected on a set of general and specific criteria, 22

research teams from 14 EU and 3 associated countries have been brought together into the network, since the start in 1993 (Fig. 1). Together they invest more than 30 scientist years per annum in prototyping. This paper focuses on a methodical approach of 5 steps, developed within the network as a common frame of reference for prototyping I/EAFS. The consecutive steps are presented and illustrated by the state of the art of the author's own project on EAFS with a group of pilot farms (NL 2), started in 1991.

2. Methodical prototyping of I/EAFS (5 steps) Building on initial experience with an experimental farm at Nagele (Vereijken, 1992) and the input of the research leaders from the network, prototyping of I/ EAFS has been elaborated in a methodical way of 5 formal steps (Vereijken, 1994, 1995, 1996); (Outline 1, next page). The outcome of these 5 steps is expressed in parts of an identity card for the prototype to facilitate the cooperation within the team and the exchange with the other teams in the network. In the following sections the 5 steps are explained in more detail and illustrated by the various parts of the identity card of our prototype EAFS for the central clay region in The Netherlands (NL 2).

.,.jv

UK2 )

@~E ~.9?NL2

1~ UKj~:~/~ c OB 1

D!

/O OF2F

Concerted Action AIR 3-CT920755 O ongoingIIEAFSprojects in 1995 Q Drojectsin preparation

Fig. 1. European network of research teams prototyping I/EAFS.

295

Outline 1. Methodical way of designing, testing, improving and disseminating prototypes of integrated and ecological (arable)farming systems (I/EAFS) 1. Hierarchy of objectives: making a hierarchy in 6 general objectives, subdivided into 20 specific objectives as a base for a prototype in which the strategic shortcomings of current farming systems are replenished (Part 1 of the identity card of a prototype). 2. Parameters and methods: transforming the major specific objectives (10) into multi-objective parameters to quantify them, establishing the multiobjective farming methods needed to achieve the quantified objectives (Part 2 of the identity card).

3. Design of theoretical prototype and methods: designing a theoretical prototype by linking parameters to farming methods (Part 3 of the identity card), designing methods in this context until they are ready for initial testing (multifunctional crop rotation as major method and Part 4 of the identity card). 4. Layout of prototype to test and improve: laying the prototype out on an experimental farm or on pilot farms in an agro-ecologically appropriate way (Part 5 of the identity card), testing and improving the prototype in general and the method in particular until (after repeated laying out) the objectives, as quantified in the set of parameters, have been achieved (Part 6 of the identity card). 5. Dissemination: Disseminating the prototype by pilot groups (<15 farmers), regional networks (15-50 farmers) and eventually by national networks (regional networks interlinked) with gradual shift in supervision from researchers to extensionists.

The procedure is simple: in the first round the general objectives are rated from 6 to 1 in descending order of importance. In the second round the specific objectives within each general objective are rated from 3 to 1 in descending order of importance (in food supply by 3, 2, 1, 0, 0 because there are 5 specific objectives, not 3). By this procedure the author's team has drawn up the hierarchy of objectives as step 1 in pilot project NL 2 (Fig. 2). It clearly shows we want to prototype an EAFS, building forth on organic farming as a forerunner and improving it on 3 strategic shortcomings: nutrient management, care of nature and landscape and quality production. This hierarchy of objectives should not be considered as just the vision of our prototyping team. Although we proposed it, the group of pilot farms has taken it over after ample discussions during several study meetings. In our experience, the hierarchy of objectives is a simple and effective instrument to achieve consensus between researchers and farmers. It could also be a good instrument to achieve consensus if more parties were involved, such as organisations of consumers or environmental groups. In that case a useful procedure would be to first let the various parTable l General and specific social values and interests (not in order of importance) involved in agriculturea General

Specific

Food supply

General Abiotic environment

Quantity Quality Stability Sustainability Accessibility Nature/ landscape

2.1. Hierarchy of objectives (step 1) Employment Table 1 presents 6 general values or interests involved in agriculture, each subdivided into 3 or 5 specific values or interests. The first step for prototypists of farming systems is to establish a hierarchy of objectives within this framework, taking into account the shortcomings of farming systems in the region and the targeted contribution the prototype should deliver to improve the situation in the short term (IAFS) or the long term (EAFS).

Specific

Farm level Regional level National level Health/ well-being Basic income/profit Farm level Regional level National level "~Simplifiedfrom Vereijken (1992).

Soil Water Air

Flora Fauna Landscape

Farm animals Rural people Urban people

296

ABIOTIC ENVIRONMENT

ties draw up their own hierarchy of objectives. Secondly, the various hierarchies should be highlighted and critically examined. Thirdly, a common hierarchy of objectives should be negotiated, based on a thorough weighting of the various arguments and sealed by a memorandum of mutual understanding or rather an agreement of cooperation and mutual support.

.............

Water

-

-

q

J

U

Air FOOD SUPPLY

t

_ ..+-~..-.~.,~:..:~.... ~

Quality

:.~= --... - _,--_-~

. . . .

Sustainability

EJ

-]

.........

Quantity

o

j

.....

Stability

Accessibility NATUR FJLANDSCAPE

Flora

[]

Landscape Fauna BASIC INCOME/PROFIT

2.2. Parameters and methods (step 2)

~]

.... ~.~ =-:-~'

E ,

t'"l

,-- []

Farm level R e g i o n a l level N a t i o n a l level

-----~ i

HEALTH/WELL-BEING Rural p e o p l e

]

Farm animals Urban people EMPLOYMENT

D

!

l

E; ~

[]

Farm level R e g i o n a l level

.., ...............

N a t i o n a l level

. . . . . 0

i

I

!

I

1

2

3

4

i

5

6

Rating

Fig. 2. Hierarchy of objectives in EAFS prototyping in Flevoland (NL 2) as an example of Part 1 of a prototype's identity card in the I/EAFS-network (squares, average of 9 European EAFS prototypes, rating explained in text).

In Flevoland (NL 2), the abiotic environment is the main objective, ahead of nature/landscape and food supply. Although pesticides have been abandoned, the abiotic environment remains of primary concern since soil fertility in EAFS is chiefly maintained by recycling organic waste, especially manure. Because organic fertilisers generally contain nutrients in ratios which do not correspond with the crop needs, accumulation and eventually leaching of certain nutrients can only be avoided by sophisticated nutrient management focusing on agronomically desired and ecologically acceptable nutrient reserves in the soil. Nature/landscape is the second main objective, since current organic farming has no explicit guidelines and technology for this increasingly scarce commodity. An ecological infrastructure will overcome this shortcoming and stimulate ecologically aware consumers to switch to ecological products. In Flevoland, development of an ecological infrastructure will focus on vegetation of the ditch sides, attractive to man and animals. Food supply is the third main objective, with the focus on an optimum balance of quantity and quality, as an indispensable basis for basic income/profit and health/well-being. This balance, called quality production, requires new and sophisticated technology, including a multifunctional crop rotation as a major substitute for external inputs, notably pesticides.

Having put the objectives into a hierarchy, prototypists need to transform them into a suitable set of parameters to quantify them. Subsequently, the quantified objectives are used as the desired results at the evaluation of the prototypes. Prototypes are tested and improved until the results achieved match the desired results. Given the overwhelming number of parameters available, there are two major reasons for not using a large set. Firstly, using a large set is time-consuming and expensive. Secondly, doing so does not ensure that the objectives are integrated which is crucial because the objectives may conflict in many ways. Consequently, prototypists must first identify a limited set of multi-objective parameters, to ensure that the objectives are integrated sufficiently. Additionally, they must establish a set of specific parameters for those objectives that are not or only insufficiently integrated by the set of multi-objective parameters. To develop I/EAFS prototypes in which potentially conflicting objectives are sufficiently integrated, prototypists need a suitable set of farming methods and techniques. Current methods and techniques mostly serve one or two of the set of objectives and harm the others. Chemical crop protection is a clear example. Therefore, it should first be looked for integrating methods and techniques which bridge the gaps between conflicting objectives and are not harmful to the others. Additionally, specific methods may be established aimed at major specific objectives that are insufficiently covered by the set of integrating methods. In this way, the author's team has quantified the objectives and established the methods as step 2 in pilot project NL 2 (Table 2). In the first column of Table 2 the top 10 specific objectives are listed, drawn up from the hierarchy of objectives (Fig. 2) by multiplying the ratings of the specific objectives by the

297 Table 2 Parameters and methods in EAFS prototyping in Flevoland (NL 2) as an example of Part 2 of a prototype's identity card in the I/EAFSnetwork Top 10 objectives

Top I0 objectives quantified in multi-objective parameters (defined in Outline 2)

Top 10 objectives achieved by multi-objective farming methods (defined in Outline 2)

I. Abiotic environment-soil

1.1 EEP-soil = 0 1.2 20 < 30a P A B > 1 if P A R < 20 PAB < 1 if PAR > 30 b 1.3x < ya KAB> I ifKAR<x KAB< I ifKAR>yb 1.4 NAR (0-100 cm) < 70 kg ha -I 2.1 Ell > 5% farm area 2.2 PSD > 50 El -I of a farm 2.3 PSDN > 20 El section -j (100 m) 3.1 QPI > 0.9 crop-i 3.2 SRI < 9 4.1 EEP-water = 0 4.2 NDW < 11.2 mg 1-j (EU-norm) See 1 5.1 FDI > 10 flowers m -I (Apr-Oct) See 2 5.2 BSD > ? 6.1NS>0 6.2 HHW < 25 h haSee 3 See 3 See 1-6 See 1-6 10.1 EEP-air = 0 See I Total parameters: 12 EU, 4 local

I. I - 1.4 MCR

2. Nature/landscape-flora

3. Food supply-quality 4. Abiotic environment-water

5. Nature/landscape-landscape

6. Basic income/profit-farm level

7. Food supply-quantity 8. Health/well-being-urban people 9. Basic income/profit-reg, level 10. Abiotic environment-air

1.2-1.4 ENM

2. EIM (target species sowing included) See 1 See I

See 2 (bird habitats included) 6. I FSO See See See See

1 and 2 1 and 2 I and 2 1, 2 and 6

See 1 Total methods: 4 EU, 0 local

aPw and K counts are the usual parameters of available reserves of P and K in the Netherlands. For K the optimum range depends on clay and organic matter contents and varies from farm to farm. blf actual PAR or KAR is in optimum range, PAB or KAB = 1.

ratings of the general objectives which they belong to. In the second column of Table 2, the top 10 specific objectives have been transformed into and quantified by a set of 16 parameters, of which 12 are on a shortlist of the I/EAFS network and 4 have a local status. In the third column of Table 2 the 4 farming methods are listed, needed to achieve the top 10 specific objectives as transformed into the set of 16 parameters. The 16 parameters and 4 methods are briefly defined in Outline 2.

Outline 2. Brief definitions of the 16 parameters and 4 methods used to quantify and achieve the top 10

specific objectives for the prototype EAFS in NL 2 (as listed in Table 2) (A) Parameters 1.1. Environment exposure to pesticides-soil (EEP-soil) = active ingredients (kg ha -j) x 50% degradation time (days). 1.2. P available reserves ( P A R ) - Pw count in N L - mg l -I P20~ in the cultivated soil layer, 1"60 extracted with water. P annual balance (PAB) = P input/P output. 1.3. K available reserves (KAR)= K-count in NL = mg K20 in 100 g air-dry soil from

298

the cultivated layer, l" l0 extracted with 0.1 N HCI. K annual balance (KAB) = K input / K output. 1.4. N available reserves (NAR) = kg ha-i Nmin in the soil layer 0-100 cm at the start of the period of precipitation surplus, e.g., N leaching. 2.1. Ecological infrastructure index (EII) - % of farm area managed as a network of linear and non-linear habitats and corridors for wild flora and fauna, including buffer strips. 2.2. Plant (target) species diversity (PSD) = number of species/EI of a farm, with conspicuous flowers by colour and/or shape, attractive for fauna and recreationists. 2.3. Plant (target) species distribution (PSDN)= mean number of target species/100 m of EI. 3.1. Quality production index (QPI) crop product-z _ quality index (QI) × production index (PI) crop product -l = (achieved price kg-l/top quality kg -I) x (on market kg ha-l/on field kg ha -l) crop product -i (0 <_ QPI <_ 1). 3.2. Soil compaction risk index (SRI) crop-1 = soil compaction index (SI) x risk index (RI) crop-l = extend of top soil compaction (025 cm) crop -I by harvest at full water saturation × probability of full water saturation of top soil at date of harvest (0<_SRI< 1). 4.1. Environment exposure to pesticides-water (EEP-water) - EEP-soil x mobility (mobility = Kom-1 and Kom = partition coefficient of the pesticide to dry matter and water fractions of the soil/organic matter fraction of the soil). 4.2. N drainage water (NDW)= mg 1-I Nmin in drainage water, averaged on the period of precipitation surplus. 5.1. Flower density index (FDI) = mean number of flowers/m per month of EI infrastructure. 5.2. Bird species diversity ( B S D ) - number of migratory and sedentary bird species. 6.1. Net surplus (NS) = total returns - all costs, including an equal payment of all labour hours. 6.2. Hours hand weeding (HHW)= mean number of hours ha-l in hand weeding.

10.1. Environment exposure to pesticides-air (EEP-air)- active ingredients (kg ha -l) x vapour pressure (Pa at 20-25°C). (B) Methods 1.1.-1.4. Multifunctional crop rotation (MCR) = a farming method with such alternation of crops (in time and space) that their vitality and quality production can be put safe with a minimum of remaining measures or inputs. 1.2.- 1.4. Ecological nutrient management (ENM) = a farming method with such tuning of input to output of nutrients, that soil reserves fit in ranges, which are agronomically desired and ecologically acceptable. 2. Ecological infrastructure management (ELM) = such layout and management of a network of landscape elements, that it is accessible and liveable to wild flora and fauna and attractive to urban and rural recreationists. 6.1. Farm structure optimisation (FSO) = a mostly indispensable method to render an agro-ecologically optimal prototype also economically optimal, by establishing the amounts of land, labour and capital goods, which are minimally needed to achieve the desired net surplus.

2.3. Design of theoretical prototype and methods (step 3) Most of the methods of the European shortlist have to be designed or redesigned, because they are nonexistent or not ready for use. However, they cannot be designed independently from each other and in arbitrary order, because they should be multi-objective and should achieve the set of objectives quantified by the set of parameters within a consistent farming system and by mutual support. Consequently, in step (3), we first establish major and minor links between the methods and the parameters they should help to achieve in a theoretical prototype before proceeding with designing the methods in their appropriate context. In this way we have designed a theoretical prototype and the methods in this context as step (3) in pilot project NL 2 (Fig. 3). This theoretical prototype

299

NS

~SDN,FD~

~

i (EEP=0)]

~ ~NAR,

BSD

Q

P/KAR NOW

farmingmethods

II. order ol d~i~l,.g)

,: i ,'+'r"'+"+"+

----'-'-

2

m a j o r links

ISRI,HHWI

Fig. 3. Theoretical prototype for EAFS in Flevoland (NL 2) as an example of Part 3 of a prototype's identity card in the I/EAFSnetwork (see Outline 2 for explanation of parameters and methods).

shows the major and minor methods needed to achieve the desired results for each objective, e.g., for each parameter. Vice versa, it also shows which

In Flevoland, the major 10 objectives as quantified in 16 parameters are achieved by 4 multi-objective methods and made ready for use in the order that follows. (1) Muitifunctional crop rotation (MCR) is the major method to achieve desired results in quality production indices (QPI product-I) without using pesticides (EEP = 0), net surplus (NS), hours hand weeding (HHW) and soil compaction risk index (SRI). It is also a method supporting P and K annual balance (P/KAB), P and K available reserves (P/KAR), N available reserves (NAR), N drainage water (NDW) and bird species diversity (BSD). (2) Ecological nutrient management (ENM)is the major method to achieve desired results in P and K annual balances, P and K available reserves, N available reserves and N drainage water. It is also a method supporting quality production indices (without using pesticides) and net surplus. (3) Ecological infrastucture management (ELM) is the major method to achieve desired results in ecological infrastructure (El), bird species diversity, plant species diversity (PSD) and local parameters of flora: Plant species distribution (PSDN) and flower density index (FDI). It is also a method supporting quality production indices and net surplus. (4) Farm structure optimisation is the finalising method to achieve the desired result in net surplus, if the current amounts of land, labour or capital goods of a pilot farm fail to do so with the agronomically and ecologically optimised prototype EAFS. Currently, it is not clear if FSO is needed. Therefore, it is not developed yet.

parameters are supported by a method and thus indicates the potential impact of a method. Consequently, the theoretical prototype defines the context and the order of designing the methods. In all theoretical prototypes of the I/EAFS-network, multifunctional crop rotation (MCR) plays a central role as a major method to achieve desired results in the multiobjective parameters of soil fertility and environment (SRI, EEP, P/KAR, etc.), as well as in the quality production indices (QPIs product-') and the major parameters of economic and labour efficiency (NS and HHW). Consequently, M C R should be designed primarily to provide for a well-balanced 'team' of crops requiring a minimum of inputs that are polluting and/or based on fossil energy (nutrients, pesticides, machinery, fuel) to maintain soil fertility and crop vitality as a basis for quality production (Outline

3). Outline 3. Procedure of designing a multifunctional crop rotation (MCR) for I/EAFS (A) Identifying and characterising potential crops for your region or farm: •

•

making a list of crops (set-aside included) in diminishing order of marketability and profitability (>6 crops for IAFS and >8 crops for EAFS); characterising the crops in their potential role in the M C R in biological, physical and chemical terms, as is done in Table 3.

(B) Drawing up an M C R based on (1) and simultaneously fulfilling a multi-functional set of demands: • •

•

filling the first rotation block with crop no. 1.; filling subsequent blocks while preserving biological soil fertility by limiting the share per crop species to _<0.25 in IAFS and _<0.167 in EAFS and the share per crop group to _<0.50 in IAFS and _<0.33 in EAFS; filling subsequent blocks, while preserving physical soil fertility by consistently scheduling a crop with a high rating of soil cover (erosion-susceptible soils) or effect on soil structure (compaction-susceptible soils) after a crop with a low rating, overall the M C R

300

•

•

•

resulting in a soil cover >-1 in IAFS and = 0 in EAFS and a soil structure >-1 in IAFS and >0 in EAFS (ratings explained below Table 3); filling subsequent blocks while conserving chemical soil fertility by consistently scheduling a crop with a high rating of N transfer before a crop with a high rating of N need and a crop with a low N transfer before a crop with a low N need, overall the MCR resulting in an N need <3 in IAFS and <2 in EAFS; filling single blocks by 2 or 3 crops with corresponding characteristics, if needed for reasons of limited labour capacity or limited market demand; ensuring crop successions are feasible in terms of harvest time, crop residues and volunteers from preceding crops.

Being the central method, and also the first to be designed, MCR is an appropriate Part 4 of the identity card, after theoretical prototype as Part 3. Tables 3 and 4 present the MCR of one of the 10 pilot farms in NL 2 as an example of Part 4 of the identity card of an EAFS prototype. Table 3 first presents the selection of the most profitable crops eligible for the MCR of the pilot farm in question, with their major characteristics concerning biological, physical and chemical soil fertility. Subsequently, Table 4 pre-

sents the MCR which optimally complies with the multifunctional set of demands.

2.4. Layout of prototype to test and improve (step 4) Step (4) implies testing and improving the prototype until the objectives as quantified in the set of parameters have been achieved. Because it is the most laborious and expensive step, requiring at least a full rotation of the prototype on each field ( 4 - 6 years for IAFS-EAFS), it is crucial that all preceding steps have been followed with the greatest accuracy. Therefore, it is useful to take a critical retrospective view before proceeding to step (4): •

•

does the hierarchy of objectives really cover the shortcomings of conventional arable farming (IAFS) or organic farming (EAFS) in the target region (not too low ratings for 'new' objectives such as nature and too high ratings for 'old' objectives such as basic income/ profit to ensure that one is really innovating and not just slightly ahead of the main group of farmers) (step 1)? have the objectives really been transformed in the appropriate set of multi-objective parameters (not too few but certainly not too many parameters?) and has each objective been quantified appropriately (not more but certainly not less ambitious than needed)

Table 3 Multifunctional crop rotation for EAFS in Flevoland (NL 2) as an example of Part 4 of a prototype's identity card in the I/EAFS-network: selection of crops by pilot farm 6 (crops in order of profitability)* Crop

Biological

Physical (ratings)

Chemical (N ratings)

no.

Species 1 Carrot 2 Potato 3 Onion 4 Celeriac 5 Sugar beet 6 Pea, bean 7 Wheat 8 Oats 9 Barley 10 Grassclover Mean of crop selection

Group"

Coverb

Rootingc

Compaction d

Structurec+d

Uptakee

Transfer f

Umbel. Solan. Lil. Umbel. Chen. Leg. Cer. Oats Cer. Leg.

-2 -2 --4 -2 -2 -2 -2 -2 -2 0 -2.0

1 1 1 1 1 2 3 3 3 3 1.9

--4 -2 -2 -4 -4 -1 -1 -1 -1 -1 -2.1

-3 -1 -1 -3 -3 1 2 2 2 2 -0.2

4 5 4 4 5 0 4 3 3 2 3.4

I 2 1 1 1 2 1 1 2 2 1.4

*For footnotes a-f, see Table 4.

301 Table 4 Multifunctional crop rotation for EAFS in Flevoland (NL 2) as an example of Part 4 of a prototype's identity card in the I/EAFS-network: multifunctional crop rotation of pilot farm 6 Block

Crop

no.

no.

I 1/5 II 6 III 2 IV 10 V 3/4 VI 7 VII VIII Mean of crop rotation

Biological

Physical (ratings)

Chemical (N ratings)

Species

Group a

Cover b

Structure c+d

Uptake e

Transfer f

Need s

Carrot/sugar beet Pea, bean Potato Grass clover Onion/celeriac Wheat

Umbel./chen. Leg. Solan. Grass/leg. Lil./umbel. Cer.

-2•-2 -2 -2 0 -4/-2 -2

-3•-3 I -1 2 -1/-3 2

4/5 0 4 2 4/4 4

1/1 2 2 2 1/1 1

3/4 -1 2 0 2/2 3

Share species -~ < 0.167

Share group -j < 0.25

-1.8

-0.2

3.2

1.5

1.6

aGenetically and phytopathologically related groups, such as cereals, legumes, crucifers and chenopodes, composites, umbellifers, iiliaceae. All subsequent blocks of perennial crops are counted as 1 block.

bNo cover in autumn and winter = -4, no cover in autumn or winter = -2, all others = 0 (green manure crops included). CCereals, grasses and lucerne = 3, root, bulb and tuber crops = 1, all others = 2 (green manure crops included). dCompaction by mowing in summer =-1 and autumn =-2, lifting in summer = - 2 and in autumn = -4. eN uptake by crop from soil reserves: legumes = 0. All other crops: 25-50 kg ha-I = !, 50-100 kg ha-I = 2, 100-150 kg ha-I = 3, 150-200 kg ha-~ = 4, etc. (N uptake = N product + N crop residues). fN transfer is the expected net contribution of N to subsequent crop, based on N residues in the soil after harvest, N mineralisation from crop residues and N losses by leaching and denitrification. In this rating, the effect of green manure crops should be included. N transfer < 50 kg ha-I = 1, 50-100 kg ha-I = 2, 100-150 kg ha-I = 3. gN need (block x) = N uptake (block x) - N transfer (block x - 1). N need is net N input to be provided by manure or N fertiliser.

and has the appropriate set of methods been established (not too many single-objective and too few multi-objective methods) (step

• •

2)? should the theoretical prototype be redesigned to link up with possible changes in the first two steps (step 3)? Testing a prototype means laying it out on an experimental farm or on a group of pilot farms and ascertaining if the results achieved correspond with the desired results. If all the methods of the theoretical prototype have been designed, an initial layout is not very complicated in the case of an experimental farm, providing a possible supervising committee and the farm m a n a g e r think it acceptable and manageable. However, much more time is generally needed to come to a first layout for pilot farms (Outline 4).

Outline 4. Preparations to come to a first layout of a theoretical prototype on pilot farms 1. Forming a pilot group

•

generating interest by articles in agricultural periodicals or by public meetings; inviting potential pilot farmers to attend study meetings; selecting pilot farmers according to general criteria such as being full-timers on farms of sufficient size, having appropriate production activities, being located in the region, having a particular soil type etc., but also according to agro-ecological criteria such as field adjacency and field size.

2. Making a variant of the prototype for each pilot farm, in interaction with the farmer • • • •

variant of multifunctional crop rotation (in time and space); variant of integrated or ecological nutrient management; variant of ecological infrastructure management; etc.

The basic task of I/EAFS designers, to replace phy-

302

V

pea I

s. wheat

s. wheat

III

poppy carrot

IV celeriac

II

10 EAFS-pilot farms:

onion

(lowest-highest) farm area 23 - 53 ha field adjacency 1 mean field size 4 - 8 ha mean field length/width 3:1 crop rotation blocks 6- 7 adjacency of subsequent blocks 0 - 0.33 share of cereals 0.16- 0.33 ecological infrastructure 0.04 - 0.06

Vl

ware potato

pilot farm 6: I-VI Crop rotation blocks 1995 - - Ecological Infrastructure <

300 m

Fig. 4. Layout of EAFS pilot farms in Flevoland (NL 2) as an example of Part 5 of a prototype's identity card in the I/EAFS-network. sico-chemical methods by biological methods and techniques, requires an appropriate concept:

I/EAFS is an agro-ecological whole consisting of a'team' of steadily interacting and rotating crops, plus their accompanying (beneficial or harmful) flora and fauna. The designer's task can thus be specified: design a rotation with a maximum of positive interactions and a minimum of negative interactions between the crops. These interactions strongly influence physical, chemical and biological fertility of the soil and consequently vitality and quality production of the crops. However, a multifunctional crop rotation (MCR) cannot cope with semi-soilborne and airborne harmful species. Therefore, an agro-ecologically optimum layout of I/EAFS should meet additional criteria

(Outline 5). In line with these criteria, we have designed an appropriate agro-ecological layout for any EAFS variant on the 10 pilot farms in NL 2 (Fig. 4).

Outline 5. Criteria for an agro-ecological layout of I/ EAFS 1. Field adjacency = 1. All fields of a farming system should be adjacent to each other, to obtain an agro-ecological whole as a prerequisite for an agro-ecological identity. 2. Field size > 1 ha. To obtain a prototype farming system with sufficient agro-ecological identity, the fields as sub-units have to be of a minimum size. 3. Field length/width < 4. Round or square fields contribute optimally to the agro-ecological identity of a farming system. Therefore, a maximum is

303

required based on 4 (IAFS) or 6 (EAFS) rotation blocks, at least (crop rotation in time). 5. Adjacency of subsequent blocks = 0. Harmful semi-soilborne species are to be prevented from following their host crop by a crop rotation without any adjacency of subsequent blocks to ensure

to be set to the length/width ratio of fields, to limit the loss in identity. 4. Crop rotation blocks > 4 (IAFS) or >6 (EAFS). The shorter the crop rotation, the greater the biotic stress on the crops and the need for external inputs to control that stress. Therefore, crop rotation is

PAR

OPI wheat

NDW

onion Relative shortfall of achieved (a) to desired (d) results*

PAB

NAR

PSD

I N R ~

.........

1992 1996

EEP i

f

decreased

PSDN /

increased

f J J .

SRI

.

.

.

.

.

.

.

.

remained n o t yet tested

KAB

/ /

*relative shortfall = (a-d)/d

NS

QPI potato HHW FDI

KAR QPI carrot

Parameters (in order of increasing shortfalls of achieved to desired results)

Desired results

Achieved results

EEP = Exposure Environment to Pesticides INR -, Infrastructure for Nature and Recreation NAR = N Available Reserves NDW = N Drain Water PAR = P Available Reserves QPl • Quality Production Index (wheat) QPI = Quality Production Index (onion) PAB - P Annual Balance PSD • Plant Spedes Diversity PSDN = Plant Spedes Distribution KAB - K Annual Balance QPI = Quality Production Index (potato) KAR = K Available Reserves QPI = Quality Production Index (carrot) FOI - Rower Density Index HHW = Hours Hand Weeding NS - Net Surplus SRI • Soilcompaction RiseIndex

0 (air, water, soil) • 0.05 < 70 kg/ha (0-10Ocm) < 11.2 NO3-N mg/I 20 < Pw-count < 30 • 0.9 (average) • 0.9 (average) 0.B < PAB < 1.2 • 50 species/INR > 20 species/INR-section (100 m) 0.6 < KAB < 0.8 • 0.9 (average) 14 < K-count < 20 • 0.9 (average) • 10 flowers/m/month INR < 500 hours/farm • 0 guilders/ha ?

0 0.052 60 9.91 24.7 0.95 0.89 1.24 42 16 1.21 0.68 25.7 0.49 2.6 1306 ? ?

,

.

. . . . . . . . . . . . . . .

Main causes of shortfall 1996

MCR, ENM MCR, ENM ENM slow response slow response ENM MCR, ENM ENM MCR, ENM slow response MCR

Methods to be improved in: Ready Accept- Manage- Effectfor use ability ability iveness

ENM ENM

MCR MCR ENM

ENM ENM

MCR ENM

ENM

MCR MCR

Fig. 5. State of the art for EAFS in Flevoland (NL 2) 1992-1995, as an example of Part 6 of a prototype's identity card in the I/EAFS-network (the prototype is all-round if achieved results match the desired results).

304

crops are not just moved to an adjacent field from year to year (crop rotation in space). 6. Share of cereals < 0.5 (IAFS) or <0.3 (EAFS). The larger the share of cereals in rotation, the greater the biotic stress and the need for external inputs for this crop group, the largest in European arable farming. Therefore, the crop rotation should have a maximum of 0.5 (IAFS) or 0.3 (EAFS) of cereals. 7. Ecological infrastructure > 5% of I/EAFS area. To bridge the gap between 2 growing seasons, airborne and semi-soilborne beneficials need an appropriate ecological infrastructure of at least 5% of the farm area. By laying out a prototype, it can be tested. By testing it will appear to what extent the desired results for any parameter have been achieved. If a shortfall appears between achieved and desired results, the prototype should be improved in the parameter in question, by adjusting the major or minor methods involved according to the theoretical prototype. The state of the art in step 4 for EAFS in NL 2 clearly shows which of the 16 parameters still have to be improved before the prototype is 'all round' (Fig. 5). It also proves that our prototyping is effective, considering the clear progress from 1992 to 1996. Improving a prototype is a matter of relating the shortfalls between achieved and desired results to the methods and improving them in a targeted way. Such shortfalls between achieved and desired results may arise from one or more of the following 4 causes: the method(s) in question is not ready for use, or not manageable by the farmer, or not acceptable to the farmer or not effective. In positive terms, step 4 (testing and improving) has been finalised if the prototype in general and the methods in particular fulfil these 4 consecutive criteria. Consequently, improving the prototype implies the following procedure (Outline 6).

Outline 6. Procedure to improve prototypes of I/ EAFS 1. Establishing which parameters have shortfalls between achieved and desired results. 2. Establishing from the theoretical prototype which methods are involved.

3. Establishing which criteria are not yet fulfilled by these methods: • • • •

ready for use; manageable by the farmers; acceptable to the farmers; effective.

4. Establishing targeted improvements to meet the successive criteria. 5. Laying out and retesting. The 4 criteria will have already received much attention before the prototype is laid out for the first time, especially in the case of testing and improving a prototype on pilot farms. Commercial farmers want to be sure a prototype is feasible and all its methods are safe? Nevertheless, on-farm testing will certainly bring to light various shortcomings of individual methods, which should be improved based on the set of 4 criteria. One major reason why a method may not appear ready for use, is unexpected occurrence of factors which interfere to such an extent that the method needs to be revised to include these factors and their effects. As a result, methods will gradually evolve from simple and subjective to comprehensive and objective.

Examples • •

management factors such as choice of crops and varieties, machines, fertilisers, pesticides; agro-ecological factors such as pests, diseases, weeds, and physical and chemical soil status.

Even if ready for use, a method may still not appear to be manageable to the farmers, for several reasons.

Examples • • •

planning or operations too complicated; too laborious to fit into the labour film; too specific to be carried out with the usual machinery.

Even if ready for use and manageable, a method may still not appear to be acceptable to the farmers, for several reasons.

305

Examples • •

too high costs and/or too few benefits, at least in the short term; too little confidence in utility and/or effectiveness.

Even if ready for use, manageable and acceptable, a method may still not appear to be effective to achieve the desired result in a certain parameter. This conclusion may be premature, in case of parameters with a slow response. Apart from this, the major reason why a method indeed may not be effective is that the theoretical prototype is too simple or distorted considering the method and the parameter in question.

Examples the method needs support by another method; the method has only a minor influence, another method should be established as the major method. Because most parameters are under control of more than one method, and many parameters have a slow response, effectiveness is the most difficult and also the most time-consuming of all 4 criteria to establish. Testing and improving a prototype will take at least 4 years for I/EAFS and 6 years for EAFS, corresponding with one run of the prototype as a complete crop rotation on each field, before reliable responses of abiota (soil, groundwater) and biota (crops, flora and fauna) are obtained. The effectiveness of the methods and the overall prototype can only be established on the basis of these reliable responses in terms of the multi-objective parameters. Theoretically, the number of years needed for step 4 would be the sum of the years needed to fulfil the first 3 criteria and the years needed to fulfil the fourth criterion. In practice, however, biota and abiota start developing a response from the first year the prototype is laid out, provided the prototype is well designed and will not change dramatically in subsequent years. As a result, the adaptation of abiota and biota will mostly occur simultaneously with the testing and improving by farmers and researchers, so step 4 could be done in a minimum of 4-6 years. However, this does not imply that all parameters will have

achieved a steady state. For example, it may take decades before possible excessive reserves of soil P have been diminished or depleted organic matter reserves have been replenished to desired ranges. Nevertheless, if the shortfalls between achieved and desired results are incontrovertably decreasing from year to year, you may speak about reliable responses proving the effectiveness of the prototype. As a result, the final step of dissemination can be envisaged with confidence.

2.5. Dissemination (step 5) If the first 4 steps of prototyping have been made on a single experimental farm, the prototype cannot just be handed over to the extension service for wide-scale dissemination! It is because such a prototype does not cover region-specific ranges in soil, climate and management conditions, which are crucial for its manageability, acceptability and effectiveness. Therefore, prototyping on an experimental farm always needs a follow-up with pilot farms to elaborate a range of variants of the prototype. Consequently, prototyping in interaction with pilot farms saves a lot of time and money and is always preferable. In addition, a group of capable and motivated farmers provides an indispensable technological and social base for an innovation project, which should include dissemination throughout the region. For this purpose we have developed a model of interactive prototyping with pilot farms (Fig. 6). Since it has appeared to work quite satisfactorily in our EAFS project, we have proposed it as a standard to the teams in the I/EAFSnetwork. In the case of interactive prototyping with 10-15 pilot farms, step 4 can be finalised with 10-15 variants of the prototype covering the regional ranges of soil, climate and management. This provides for an excellent base for wider dissemination throughout the region. The initial group of pilot farms can act as demonstration farms and the farmers can be involved in training and guiding farmers willing to convert. To disseminate the prototype in wider circles, regional extension services should be trained to participate and gradually take over the innovation project. The interaction model (Fig. 6) can be maintained to convert groups of farms in a programme of at least 4 years.

306

PILOT FARMS

various farms situations

RESEARCH TEAM

)

(~

theoretical prototype

Fig. 6. Interactive prototyping: designing, testing and improving a prototype by interaction of pilot farms and research team. 3. Discussion

All over the world, agriculture is still being intensified, causing destabilisation of agro-ecosystems and environmental pollution. In developing countries, it is understandable for various reasons, especially in those countries where food production can hardly keep pace with population increase. In industrialised countries, it is absurd when one considers the growing surpluses of agricultural products, the decreasing income and employment in most rural areas and the growing concern of the consumers about the quality of their food. Fortunately, there is also a growing awareness that these immense problems cannot be solved one by one on an ad-hoc basis, but that a more comprehensive and sustainable approach of agriculture is needed. As a result, several new approaches have been proposed, such as sustainable (Allen and van Dusen, 1988; Edwards et al., 1990), integrated (Vereijken and Royle, 1989) and alternative agriculture (National

Research Council of USA, 1989). However, their use is limited because they are hardly defined in measurable terms, elaborated into concrete farming systems and tested for feasibility. Therefore, the current methodical prototyping has been developed to enable agronomists to design, test and improve more sustainable farming systems in interaction with pilot farms. The methodical prototyping of I/EAFS presented here has its roots in two global organisations. The concept of IAFS is based on the work of the crop protectionists, assembled in the International Organisation for Biological and Integrated Control (IOBC). Initially, most working groups of IOBC aimed at the control of single pest species. However, this brought about various problems, such as insufficient cost effectiveness and the emergence of new pests. Therefore, they developed a wider scope and aimed at integrated protection of crops. During the last decade, the scope has further been widened to IAFS, comprising

307

the entire crop rotation and also considering soil cultivation and fertilisation (Anonymous, 1977; Vereijken et al., 1986; El Titi et al., 1993). The concept of EAFS has been developed by the teams in the I~AFS network, searching for a more consistent and sustainable elaboration of IAFS. They have been inspired by the concept of organic farming, as defined by the standards and guidelines of the International Federation of Organic Agriculture Movements (IFOAM) (Geier, 1991). The great advantage of the organic concept is that it offers a market model of shared responsibility by producers and consumers for a sustainable and multifunctional management of the rural areas as agro-ecosystems. It is expressed by the principle of premium prices for the added ecological value of the products under label. This provides for the necessary economic base for the consistent and sustainable elaboration of IAFS, to be called EAFS. However, EAFS should go further than is demanded by the IFOAM guidelines for organic farming in sustainable and multifunctional management of the environment, nature/landscape and health/well-being of people and farm animals. The methodical prototyping of I/EAFS presented here starts at the stage where most fanning systems research stops, and that is the stage of analysis and diagnosis (Gibbon, 1994). However, the I~AFS teams of the EU network, strong in agronomy and ecology, may improve their start by building on a more comprehensive and profound rural and farming systems analysis from research teams, strong in sociology and economy such as those led by Van der Ploeg (1995) and Sevilla Guzman and ISEC Team (1994). The methodical prototyping of I/ EAFS presented here is still provisionally elaborated considering the interaction with pilot farmers in general and the last step (5) of dissemination in particular. In this respect, the I/EAFS teams could also benefit from the expertise developed by teams, such as those led by R61ing (1994). With this enlargement and reinforcement of their capacity, the teams of the I/EAFS network have excellent chances to succeed where up to now most farming systems researchers failed (Gibbon, 1994). Their work comes further than the step of diagnosis and analysis, and includes the subsequent steps of design, layout for testing and improving, and eventually dissemination. Initial results are encouraging. Most teams are pro-

gressively achieving the desired results, although the effectiveness of prototyping can still be improved in various ways (Vereijken, 1996). Nevertheless, the clear progress we achieved in our EAFS prototype for Flevoland (Fig. 5) may be considered as an example of the effectiveness of the prototyping in the I/ EAFS network. The entire methodical approach to prototyping F EAFS will be available in a manual at the end of the current EU concerted action. References Allen, P. and van Dusen, D. (Editors), 1988. Global perspectives on agroecology and sustainable agricultural systems. Proceedings of the Sixth International Conference of International Federation Organic Agriculture Movements. Agroecology Program, University of California, Santa Cruz, 721 pp. Anonymous, 1977. An approach towards integrated agricultural production through integrated plant protection. IOBC/WPRS Bulletin no. 4, 163 pp. Edwards, C.A., Lal, R., Madden, P., Miller, R.H. and House, G. (Editors), 1990. Sustainable Agricultural Systems. Soil and Water Conservation Society, Iowa, 696 pp. El Titi, A., Boiler, E.F. and Gendrier, J.P., 1993. Integrated production, principles and technical guidelines. Publication of the Commission: IP-Guidelines and Endorsement. IOBC/WPRS Bulletin no. 16, 96 pp. ISBN 92-9067-048-0. Geier, B. (Editor), 1991. IFOAM Basic Standards of Organic Agriculture and Food Processing, 20 pp. Oecozentrum Imsbach, D66696 Tholey-Theley, Germany. Gibbon, D., 1994. Farming systems research/extension: background concepts, experience and networking. In: J.B. Dent and M.J. McGregor (Editors), Rural and Farming Systems Analysis. European Perspectives. Proceedings of the First European Convention on Farming Systems Research and Extension, Edinburgh 1993, pp. 3-19. AB International. ISBN 0851989144. National Research Council of USA, 1989. Alternative agriculture. Report of the Committee on the Role of Alternative Farming Methods in Modern Production Agriculture. NRC, Washington, DC, 448 pp. van der Ploeg, J.D., 1995. From structural development to structural involution: impact of new development in Dutch agriculture. In: J.D. van der Ploeg and G. van Dijk (Editors), Beyond Modernization, the Impact of Endogenous Rural Development. Van Gorcum, Assen, The Netherlands, pp. 109-147. R61ing, N., 1994. Interaction between extension services and farmer decision making: new issues and sustainable farming. In: J.B. Dent and M.J. McGregor (Editors), Rural and Farming Systems Analysis. European perspectives. Proceedings of the first European Convention on Farming Systems Research and Extension, Edinburgh, 1993. AB International, pp. 280-291 (ISBN 0851989144). Sevilla Guzman, E. and ISEC Team, 1994. The role of farming

308 systems research/extension in guiding low input systems towards sustainability: an agro-ecological approach for Andalusia. In: J.B. Dent and M.J. McGregor (Editors), Rural and Farming Systems Analysis. European Perspectives. Proceedings of the First European Convention on Farming Systems Research and Extension, Edinburgh, 1993, AB International, pp. 305-319 (ISBN 0851989144). Vereijken, P. and Royle, D.J. (Editors), 1989. Current Status of Integrated Arable Farming Systems Research in Western Europe. IOBC/WPRS Bulletin 1989/XII/5, Wageningen, 76 pp. Vereijken, P., 1992. A methodic way to more sustainable farming systems, Neth. J. Agric. Sci., 40: 209-223. Vereijken, P., 1994. Designing prototypes. Progress Report 1 of the Research Network on Integrated and Ecological Arable Farming

Systems for EU and Associated Countries. AB-DLO, Wageningen, 90 pp. Vereijken, P., 1995. Designing and testing prototypes. Progress Report 2 of the Research Network on Integrated and Ecological Arable Farming Systems for EU and Associated Countries, ABDLO, Wageningen, 90 pp. Vereijken, P., 1996. Testing and improving prototypes. Progress Report 3 of the Research Network on Integrated and Ecological Arable Farming Systems for EU and Associated Countries. ABDLO, Wageningen, 69 pp. Vereijken, P., C.A. Edwards, A El Titi, A. Fougeroux and M. Way, 1986. Report of the Study Group: Management of Farming Systems for Integrated Control. IOBC/WPRS Bulletin no. 9 (ISBN 92-9057-001-0).

© 1997 ElsevierScience B.V. ,411rights reserved Perspectives for ,4gronomy - ,4dopting Ecological Principles and Managing Resource Use M.K. van Ittersum and S.C. van de Geijn (Editors)

309

The Log irden project" development of an ecological and an integrated arable farming system C.A. Helander * Rural Economy and Agricultural Society of Skaraborg, P.O. Box 124, S-532 22 Skara, Sweden Abstract

On the research-farm Loghrden in south-west Sweden a farming system project is carried out according to the methodology for farming systems research elaborated by a European research network in an EU Concerted action. The main emphasis is on development of an Ecological Arable Farming System (EAFS) and of an Integrated Arable Farming System OAFS). The design of the project does not give priority to comparisons between the different systems, but a Conventional Arable Farming System (CAFS) is also used in the project as a reference. The aim of the project is to achieve a long-term persistent, sustainable and productive food supply in combination with a minimum of negative impacts on the abiotic environment. In the Ecological Arable Farming System (EAFS) there were lower yield levels than expected, probably mainly due to nitrogen deficiency during especially the early part of the growing season. The economic result in EAFS, as net surplus, was dramatically improved in 1995 compared to previous years. This is due to the new situation when Sweden joined the EU, with high arable area payments, together with a large demand for ecological products on the market that gave higher prices of these products. In the Integrated Arable Farming System OAFS) the total use of pesticides was reduced by 70% during 1995, but the use of herbicides was above the desired level of 50% compared with the Conventional Arable Farming System (CAFS). The average yields in IAFS were similar to those in the conventional system. Despite this, the economic result for IAFS was lower, the reason being higher machinery costs, extra costs for seeds for undersowing (white clover), and more expensive weed control in the undersown crops. Also the yields of triticale (not grown in the other systems) were low due to winter damage. Keywords: Arable farming; Ecological; Integrated farming systems; Sustainable agriculture

1. Introduction

The predominant goal for agriculture has been that of high production but is now gradually changing for other objectives. Overproduction in combination with an increased awareness of negative impacts of conventional agriculture on environment, nature and the landscape, in addition to a questioning of the long-term sustainability of the present production systems, raised the need for new production methods (Ebbersten, 1990). For instance, the use of fertilisers and pesticides has increased substantially * Tel: +46 511 13160; Fax: +46 511 186 31; E-mail: [email protected]

during the last 30 years in most countries, to a great extent without corresponding increases in yields. This means that the utilisation efficiency has decreased simultaneously with increasing inputs. Such aspects partly require a new research methodology as traditional research in agriculture has generally been concentrated on one or two production factors at a time, without consideration of system effects. The aim of the project described in this paper is to consider at the farming system as a whole, the main objective being to achieve a long-term persistent, sustainable and productive food supply with a minimum of negative impacts on the abiotic environment. Additional objectives are the minimum input of external energy by means of maximum use of

310

bioenergy produced on the farm (fuel from rapeseed) and to minimise the use of other external inputs, such as nitrogen fertilisers. New objectives, such as quality of the abiotic environment, landscape and nature values, agronomic sustainability and animal welfare, are integrated with traditional agricultural goals. For this, a research-farm, Logfirden, in the southwest of Sweden was chosen. The farming system project at Logfirden is connected to a working group within IOBC (International Organization of Biological and Integrated Control of Noxious Animals and Plants) (El Titi et al., 1993) and also to a European research network in an EU Concerted action (Vereijken, 1994; 1995; 1997). This European network includes the project leaders from most of the ongoing farming system research projects in western Europe, such as the Nagele-project in the Netherlands (Vereijken, 1992), the former Lautenbach-project in Germany (El Titi, 1990) and the LIFE project in England (Jordan and Hutcheon, 1994).

Experimental F a r m ( S-1 )

Log~den

[-]

EAFS (22.0ha)

D

IAFS (2S.0ha)

I

CAFS ( 12.0 ha)

I-VIIICrop rotation blocks a-b

Rotations with 75-50% cereals

Ecological infi'astruetum .)

N

It

(n b)

2. Materials and methods

A large-scale farming system research project started in 1991 at the Logfirden research farm, Grfistorp, (58*20' N, 12"38' E) Sweden. The total area for the experiment covers 60 ha of arable land, the size of each field being between 2.5 and 4.0 ha, see map of the farm, Fig. 1. The soil on the farm is a fertile, very heavy clay soil (40-50% clay), with an organic matter content between 2 and 3%. The soil structure is rather poor, due to compaction with heavy machinery and a spring-cereal dominated crop rotation.

~-- 300m

Fig. 1. Map of Logfirden showing the design of the farming system research project.

The weed seed bank is rather small, with Matricaria inodora and Stellaria media as dominating species. The average annual rainfall is about 600 mm. The main emphasis is on development of an Ecological Arable Farming System (EAFS) and an Integrated Arable Farming System (IAFS). The design of the project does not give priority to comparisons

Table 1 The Multifunctional Crop Rotations (MCR) at Logfirden. The rotation in the ecological system was changed from 1996, but this table shows the rotation used in '91-'95 Year

Conventional

Ecological

Integrateda)

Integrated b)

1 2 3 4 5 6 7 8

peas w-wheat oats w-wheat s-rape w-wheat oats w-wheat

peas w-wheat fieldbeans oat vetch w-wheat set-aside rye

peas w-wheat (undersown) oats w-wheat s-rape w-wheat (undersown) oats triticale

peas w-wheat (undersown) set-aside (grass/lucerne) set-aside (grass/lucerne) w-rape w-wheat (undersown) oats triticale

311

Eli

NS, EE QPI~ ~ (IEP 8~, ......

SCI SR SSC

Q

farmingmethods 0.arckr6fdm~m) parameters

PAR, PAB

major links

NAR OMAB

Fig. 2. Theoretical EAFS prototype of Log~rden. For explanation of codes, see text.

PI

Ell

NS, EE

QPI ~uc~ 1

I o--L

PAB PAR

SR

ss_.g__c

Q

farmingmethocb .o

parameters majorlinks

OMAB NAR

Fig. 3. Theoretical IAFS prototype of Log~rden. For explanation of codes, see text.

312 Table 2 Average yields during 1993-95 for crops grown in two or all three of the farming systems Crop

W-wheat Oats S-rape

Integrated

Conventional

Ecological

kg ha -1

kg ha -1

E/C

kg ha -1

I/C

6940 6450 2070

42001) 36801)

0.612) 0.582 )

7080 6350 2350

1.02 0.98 1.15

l) Only results from 1993-94. 2) Average for conventional 1993-94, compared to the same years for ecological.

between the different systems. The Ecological Arable Farming System (EAFS) implies that no chemical fertilisers or agrochemicals are being used. The yields are expected to be lower, 70-80% of conventional farming, and have to be compensated through higher prices. The Integrated Arable Farming System (IAFS) is a system that emphasises reduction of inputs. The yields are expected to be slightly lower than in a conventional system, but the economic result for the farmer is not expected to be reduced. In both these systems the methods described below are being used to develop a long-term persistent, sustainable and productive farming system. In IAFS the yearly planning and the decisions in field are made by the project leader together with the farm manager. The Conventional Arable Farming System (CAFS) is used as a reference and reflects a common type of farming in the region. In CAFS the yearly planning and the decisions in field are made by a local agricultural advisor together with the farm manager. The project at Log~trden follows the methodology for farming systems research elaborated by the European research network on integrated and ecological arable farming systems (Nilsson, 1994; Vereijken, 1994; 1995; 1997). The methods used follow a European shortlist (Vereijken, 1994). They are used in the following order: 1.

2.

Multifunctional Crop Rotation (MCR): the major method to preserve soil fertility in biological, physical and chemical terms and to sustain quality production with a minimum of inputs (pesticides, fertilisers, support energy and labour). Integrated/Ecological Nutrient Management (INM/ENM): supports MCR by maintaining agronomically desired and ecologically accept-

3.

4.

5.

6.

able nutrient reserves in the soil and contributes, together with MSC (see below), to maintain an appropriate content of organic matter. Minimum Soil Cultivation (MSC), only in IAFS: supports MCR by incorporating crop residues, controlling weeds and restoring physical soil fertility, while maintaining sufficient soil cover as a basis for avoiding nutrient losses, shelter for natural enemies and for landscape/nature values. Ecological Infrastructure Management (EIM): supports MCR by providing airborne and semi-soilborne beneficials a place to overwinter and recover and disperse in spring. In addition, EIM should achieve different nature and landscape objectives. Integrated Crop Protection (ICP)~ only in IAFS" supports MCR and EIM by selectively controlling remaining harmful species with minimal exposure to the environment of pesticides. Farm Structure Optimisation (FSO): the method to make a farming system economically optimal by determining the minimum amounts of land, labour and capital needed.

The standardised design of these methods is described by Vereijken (1995; 1997), where also the methods and parameters used at the Log~rden project are described. The three different parts of the farm, Ecological (EAFS), Integrated (IAFS) and Conventional (CAFS), have different crop rotations (see Table l) using the Multifunctional Crop Rotation (MCR) concept (Vereijken, 1994). The IAFS has been divided into two different rotations, results presented in this paper being from rotation b. The integrated system is based on non-ploughing tillage practices. In

313

Table 3 Soil Cover Index (SCI). SCI is the extent to which the fields on a farm is covered by crops overall the year. Desired result: SCI > 0.8 Year

Conventional (CAFS)

Ecological (EAFS)

Integrated (IAFS)

1993 1994 1995

0.67 0.54 0.79

0.69 0.61 0.76

0.89 0.80 0.81

the ecological and conventional system ploughing is carried out almost every year. The systems are evaluated using 11 (EAFS) or 12 (IAFS) multiobjective parameters from the European shortlist (Vereijken, 1994). Furthermore, some local parameters are operated. All agronomic data recorded are processed by means of a yearly analysis program to quantify the parameters that are used in the evaluation. The different parameters are linked to one or more of the methods used for development of the Ecological Arable Farming System (EAFS) and the Integrated Arable Farming System (IAFS). Figs. 2 and 3 presents the theoretical prototypes for EAFS and IAFS showing the major and minor methods to be followed to achieve the desired result for each parameter. Results from the following 6 multi-objective parameters are presented in this paper" 1. Soil Cover Index (SCI) 2. Nitrogen-Available Reserves (NAR) 3. Nitrogen Utilisation (local) 4. Energy Efficiency (EE) 5. Pesticide Index (PI), (only in IAFS) 6. Net Surplus (NS)

Desired result: 80% soil cover Desired result: < 30 kg N ha -1 Desired result: > 70% efficiency

parameters, but the results from these analyses are still under evaluation.

3. Results

The results are presented below in terms of parameters and related methods. The project is still, after five years, in a development stage and therefore the results only demonstrate the progress so far. It is not possible to make a direct comparison between the different systems due to the layout of the experiment. The average yields of crops grown in all three systems (winterwheat and oats) and for springrape grown in CAFS and IAFS are presented in Table 2. The average yields of "ecological" crops and of "integrated" crops are compared with the yield of the "conventional" crops as relative numbers, E/C and I/C. These numbers can only be seen as an indication of the level of difference. After a few more years (at least a full crop rotation of eight years) the systems should have reached some kind of equilibrium and then the average yields from the different systems should become more reliable.

Desired result: > 6 (output/input) Desired result: < 50% of CAFS Desired result: > 0 SEK

Five other parameters are also used, but results are not presented in this paper. These parameters are: Phosphor-Available Reserves (PAR), Phosphor-Annual Balance (PAn), Organic Matter-Annual Balance (OMAB), Quality Production Index (QPI) and Environmental Exposure to Pesticides (EEP). Two more parameters that are planned to be used are Soil Respiration (SR) and Soil Structure and Compaction (SSC). An extensive analytical programme is carried out every year as a basis for these

3.1. Soil Cover Index (SCI) The desired result for this parameter is to have crops covering the ground for more than 80% of the year. A high SCI is very important for reducing the risks of nutrient leaching and soil compaction. It is also important for building up the organic matter and for a high biological activity in the topsoil. The IAFS achieved the target each year but not the EAFS nor the CAFS although the EAFS had marginally higher values than the CAFS (Table 3).

3.2. Nitrogen-Available Reserves (NAR) NAR is the major parameter used to evaluate the

314

Table 4 Nitrogen-Available Reserves (NAR) in kg N ha -1 (0-90 cm), after the indicated crop, at start of the leaching period. Desired level: 30 kg N ha -1 Year

Conventional

Ecological

NminSoil

Crop

NminSoil

Crop

NminSoil

Crop

1992 1993 1994 1995

40 58 64 35

w-wheat s-rape oats w-wheat

76 51 42 41

green manure rye field beans w-wheat

56 31 43 37

s-rape w-wheat oats triticale

Integrated and Ecological Nutrient Management (I/ ENM). The goal is to achieve a NminSoil level below 30 kg N ha -1 (in the 0-90 cm soil layer) at the start of the leaching period (Nov-Dec). This level is probably difficult, but possible to reach. So far the values measured in all three systems have been above the desired level, see Table 4.

Integrated

utilisation of nitrogen was achieved in the ecological system with no chemical fertilisers, followed by the conventional system. A nitrogen balance sheet for 1995 is also presented in Table 5, where an average for all inputs and outputs in each system is calculated.

3.4. Energy Efficiency (EE) 3.3. Nitrogen utilisation A local parameter used in 1995 to evaluate the Integrated/Ecological Nutrient Management (I/ ENM) is a Nitrogen Utilisation calculation. The nitrogen uptake in harvested plant products was divided by all inputs of nitrogen, i.e. fertilisers, manure and nitrogen from nitrogen-fixing organisms. The results are shown in Table 5. The best

EE is a parameter which is planned to be used for the evaluation of the energy efficiency in the farming system. The method of calculating energy efficiency is still under discussion. Calculations presented in Table 6 are made by means of a computer model where the total energy content in the harvested plant products (in kWh/ha) is divided by the total input of energy (in kWh/ha) into the cropping system. The

Table 5 Nitrogen balance sheet and Nitrogen Utilisation in 1995. Nitrogen balances (kg N/ha) are calculated for inputs and outputs as an average for all harvested crops in each system. Nitrogen Utilisation is calculated as nitrogen in harvested plant products divided by all inputs of nitrogen Nitrogen balance sheet Inputs: N-fertilisers Seeds N-fixating crops 1) Deposition from air Manure 2) Total input Outputs: Crops Straw (off field)

Total output Nitrogen Utilisation

Conventional

Ecological

Integrated

132 4 14 12

5 9 14 15

76 3 14 -

162

43

93

108 14 122 75%

33 1 34 79%

50 6 56 60%

1) In 1995 the pea-crop was completely damaged due to very wet conditions in May and June, no pea-crop was harvested. 2) The conventional system were given 40 ton/ha of slurry in 3 of the 4 fields. This is more than a normal supply for a single year.

315 Table 6 Energy Efficiency (EE) for 1995. EE is calculated by dividing the energy in harvested plant products by the total energy input into the cropping system. Desired level: EE > 6.0

Average for all harvested crops Best crop

Conventional

Ecological

Integrated

4.7 6.0 (w-wheat)

1.9 5.6 (s-wheat)

3.6 7.4 (w-wheat)

total input includes energy used to produce machinery, diesel, electricity, fuel oil, nitrogen, pesticides and seed all calculated in kWh/ha. The calculations in Table 6 are fully based on the actual use on Loghrden experimental farm in 1995. The target value is EE > 6 (output of energy divided by input of energy) as average for all harvested crops. This level was not reached in any of the systems in 1995.

3.5. Ecological Infrastructure Index (Eli) An Ecological Infrastructure Management (EIM) is very important for creating improved biodiversity which provides useful support for a sustainable farming system. Eli is the main parameter for measuring the level of ecological infrastructure. At Log~rden, 6% of the arable land in both the ecological and the integrated system has been used for ecological infrastructures such as hedges, green belts and pathways. In the conventional system no part of the arable land is used for ecological infrastructure.

3.6. Pesticide Index ( PI) A Pesticide Index (PI) is used as a parameter to evaluate the need for pesticide treatments in the In-

tegrated Arable Farming System (IAFS) compared to the Conventional Arable Farming System (CAFS). No pesticides are used in the ecological system and therefore this index is not used in EAFS. The calculations presented in Table 7 are based on the use of pesticides in the integrated system compared to the use in the conventional system. The actually used dosage and number of treatments for each type of pesticide (insecticides, fungicides and herbicides) is recalculated to an average for the whole crop rotation in each system (for instance: one fungicide treatment with half the recommended dose on all fields will give the PI=0.5 for fungicides). The total index calculated for Log~rden is well below the desired level of 50% for the use of insecticides and fungicides. However, the use of herbicides is above the desired level.

3. 7. Net Surplus (NS) NS is a parameter for evaluation of the economic efficiency of the whole farming system. Table 8 shows the NS for all three systems during 1993-95. NS is calculated for each year's conditions. It is important to notice that Sweden joined EU in 1995. This resulted in completely different conditions, with for instance, an arable payment and an extra payment for ecological farming.

Table 7 Pesticide Index (PI) for 1995. PI is based on the actually used dosage and number of treatments for each type of pesticide in each system compared to recommended dose for each pesticide. The use in the integrated system is divided by the use in the conventional system. Desired level for Pl IAFS/CAFS: < 0.5 Type of pesticide

Pl for CAFS

PI for IAFS

PI for IAFS/CAFS

Insecticides Fungicides Herbicides Mean of all pesticides

1.18 0.82 0.82 0.94

0.09 0.11 0.60 0.26

0.08 0.14 0.72 0.30

316

Table 8 Net Surplus (NS) in SEK ha -1. NS is gross revenues minus all costs, including payments for all labour hours and for the land used. Desired result: NS > 0 Year

Conventional

Ecological

Integrated

1993 1994 19951)

130 -280 880

-2660 -2840 980

- 160 -2140 -40

1) EU/CAP

chanical weed control was quite successful. The number of weeds was low and did not cause any major problems. The economic result in EAFS, as net surplus, was dramatically improved in 1995 due to the new situation when Sweden joined the EU with high arable area payments. A large demand for ecological products on the market gave high prices of these products.

4.2. Nitrogen management 4. Discussion

The project is still, after five years, in a development stage and therefore the results only demonstrate the progress so far. It is very important to realise that some of the positive effects, for instance a higher nitrogen mineralisation capacity, which are expected from a change into an ecological or integrated farming system, are still to come.

4.1. Ecological Arable Farming System (EAFS) In the Ecological Arable Farming System the yields were lower than expected, probably caused by nitrogen deficiency during especially the early part of the growing season. The low level of plant-available nitrogen was, at least partly, related to a very compact soil and thereby low microbial activity. Compact soil may also have caused losses of nitrogen due to denitrification under wet conditions (as during June of 1995). The results of the Energy Efficiency (EE) calculations very clearly show the importance of crop yield levels. The yields in 1995 were very poor for many crops, especially in the ecological system but also in the integrated. Table 6 shows that the desired level of EE >6.0 can be achieved for some individual crops, but that the target level is not reached for the average of the whole crop rotation. The result for Soil Cover Index (SCI) in the EAFS has not reached the desired level (Table 3). This is one of the reasons for making a change in this rotation: vetch and winterwheat have been substituted by set-aside (an undersown green manure crop) and winterrape. Weeds, insects and diseases constituted a smaller problem than expected. The use of me-

A low level of available nitrogen before the winter period will reduce the risk for losses during this period when the crops take up no or very little nitrogen. So far, the NAR values measured in all three systems have been above the desired level. The best possibilities to reduce NAR include a higher precision in the application of nitrogen and more restricted soil tillage after harvest, and especially not to stimulate nitrogen mineralisation in the autumn by means of early soil tillage. Another factor of importance, to be better used, is to bind nitrogen organically during winter, by means of winter crops.

4.3. Integrated Arable Farming System (IAFS) The general use of external inputs (pesticides and chemical fertilisers) in conventional farming in Sweden is very low compared to that in many other western European countries. This means that there is not much scope for reduction of, for instance, pesticide use. In spite of that, the total use of pesticides was reduced by 70% during 1995 in the Integrated Arable Farming System (Table 7). The use of herbicides is still above the desired level of 50% compared to the conventional farming system. To improve this, more mechanical weed control will be carried out. Mechanical weed control proved to be quite successful in the ecological system since the number of weeds was low and did not cause any major problems. The reduction in the use of nitrogen fertilisers and in the use of fuel for soil preparation also has been substantial. The average yields in IAFS were similar to those in the conventional system. Despite this, the economic results for IAFS were lower, the reason being higher machinery costs, extra costs for seeds for undersowing (white clover), and

317

more expensive weed control in the undersown crops. Also the yields of triticale (not grown in the other systems) were low due to winter damage. To improve the economic result for IAFS we are looking into better alternatives for weed control in the undersown crops and also more winter-hardy varieties of triticale. The full effect from the undersown white clover being a higher nitrogen mineralisation capacity is hopefully still to come. The economic results for IAFS in other comparable projects such as the Nagele project and the LIFE project show positive results compared to conventional systems. The fact that we in general have a low use of external inputs in conventional farming in Sweden is perhaps one reason for not getting that positive results in the Log~rden project, when we compare the integrated with the conventional system.

Acknowledgments The arable farming system research project at Logarden has been financed by the Swedish Board of Agriculture and the Agricultural Society of Skaraborg.

References Ebbersten, S., 1990. Lantbruksvetenskap - en omvirldsanalys inf6r 2000-talet med sarskild h/insyn till agronom-, hortonom- och landskapsarkitektutbildningarna. SLU/F6rvaltning nr 16, Sveriges Lantbruksuniversitet, Uppsala. El Tiff, A., 1990. Farming System Research at Lautenbach, Germany. Schweiz. Landwirtsch. Forsch., 29(4): 23%247. El Tiff, A., Boiler, E.F. and Gendrier, J.P., 1993. Integrated Production, Principles and Technical Guidelines. IOBC/WPRS Bulletin OILB/SROP, Vol. 16 (1), 97 p. Jordan, V.W.L. and Hutcheon, J.A., 1994. Economic viability of less-intensive farming systems designed to meet current and future policy requirements: 5 year summary of the LIFE project. Asp. Appl. Biol., 40: 61-68. Nilsson, C., 1994. Integrated farming systems research at Alnarp. Proceedings NJF symposium 'Integrated systems in agriculture', 1-3 December 1993 in Norway: pp. 65-70. Vereijken, P., 1992. A methodic way to more sustainable farming systems. Neth. J. Agric. Sci., 40: 209-223. Vereijken, P., 1994. 1. Designing Prototypes, Progress Reports of Research Network on Integrated and Ecological Arable Farming Systems for EU and associated countries. AB-DLO, Wageningen, 87 p. Vereijken, P., 1995. 2. Designing and Testing Prototypes, Progress Reports of Research Network on Integrated and Ecological Arable Farming Systems for EU and associated countries. AB-DLO, Wageningen, 90 p. Vereijken, P., 1997. A methodical way of prototyping integrated and ecological arable farming system (I/EAFS) in interaction with pilst farms. In: M.K. van lttersum and S.C. van de Geijn (eds.), Proceedings of the 4th ESA Congress, Elsevier, Amsterdam, the Netherlands, pp.

This Page Intentionally Left Blank

© 1997 Elsevier Science B. II". All rights reserved Perspectives for Agronomy- Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

319

Integrated crop protection and environment exposure to pesticides: methods to reduce use and impact of pesticides in arable farming F.G. Wijnands* Applied Researchfor Arable Farming and Field Production of Vegetables, P.O. Box 430, NL 8200 AK, Lelystad, Netherlands

Accepted 14 July 1997

Abstract

Prototypes of Integrated Farming Systems for arable farming are being developed in the Netherlands based on a coherent methodology elaborated in an European Union concerted action. The role of crop protection in Integrated systems is, additional to all other methods, to efficiently control the remaining harmful species, with minimal use of well selected pesticides. The overall aim of more sustainable farming systems is to reduce the exposure of the environment to pesticides in order to prevent short- and long term effects on all species over all the biosphere. An innovative approach to quantify this exposure of the environment to pesticides, based on molecular-chemical properties of the pesticides, is presented. The results of prototyping on an experimental farm in the Netherlands shows that not only drastic reductions in pesticide use are possible but that subsequent careful selection of pesticides can also lead to minimal environmental impact. © 1997 Elsevier Science B.V. Keywords: Arable farming; Environment; Integrated crop protection; Integrated farming; Pesticides; Prevention; Farming systems research; Indicators; Pesticide risk evaluation

I. Introduction

The use of pesticides in current arable farming systems is extremely high due to the almost exclusive choice for pesticides to correct structural problems in farm management such as insufficient crop rotation, susceptible varieties and high nitrogen inputs. The high pesticide use is only one symptom, however a major one, of the shortcomings of current farming in the European Union. Current farming is associated with a complex of environmental, agronomic and ecological problems. In reaction to these problems, Integrated Farming *Tel.: +31 320 29Ill1; fax: +31 320230479.

Systems have been developed as a coherent new vision on agriculture alongside other concepts such as ecological farming. Over the last 15 years these systems that integrate potentially conflicting objectives concerning economy, environment and agronomy are being developed on experimental farms all over Europe (Vereijken and Royle, 1989; Vereijken, 1994); in the last 5 years increasingly also in cooperation with commercial farms (Vereijken 1995, 1997). The methodology of designing, testing, improving and disseminating Integrated and Ecological Farming Systems for arable farming is elaborated in a 4 year European Union Concerted Action involving the leading research teams in Europe. This methodology is

Reprinted from the European Journal of Agronomy 7 (1997) 251-260

320

called prototyping and comprises five steps (Vereijken, 1994, 1995, 1997). After the objectives have been set (1) and transformed into a suitable set of multi-objective parameters (2), appropriate farming methods (comprehensive strategies built on different techniques) that sufficiently integrate the potentially conflicting objectives need to be developed or redesigned (3). Top priority is given to the design of a multifunctional crop rotation. Then nutrient management strategies need to be designed, followed by the design of soil cultivation strategies and the lay-out of an ecological infrastructure on the farm. All these methods are aimed at sustaining quality production with minimum external inputs and environmental hazards. In a theoretical prototype parameters and methods are linked as last check (4) before the testing in practice may start (5). Testing and improving the prototype in general and the method in particular continues until the objectives as quantified in the set of parameters have been achieved. This can either be done on experimental farms or on pilot farms. Dissemination of the results including implementation in practice concludes this approach. This paper is based on prototyping research on the Nagele experimental farm (Wijnands and Vereijken, 1992) in the Netherlands and elaborates the role of the farming method Integrated Crop Protection (ICP) in Integrated systems. This method is complementary to the methods that consider crop rotation, nutrient management, soil cultivation and ecological infrastructure, that were mentioned before. It will be shown how ICP can reduce the input of pesticides drastically. A new concept of quantifying the environmental burden due to pesticide use will be elaborated. This concept is called Environment Exposure to Pesticides (EEP). Minimising the latter is the basic aim for more sustainable farming systems in order to prevent short- and long term adverse effects on all species over all the biosphere. Results of the Nagele farm will demonstrate the perspective of this concept.

designed for three specific regions in the Netherlands and laid out on experimental farms with region-specific crop rotations and cropping systems (Wijnands and Vereijken, 1992). From 1990 to 1993 the tested prototypes were evaluated on commercial farms in a national pilot farm network (Wijnands, 1992; Wijnands et al., 1995). The Integrated prototype for the Central Clay area will serve here as example. The small farm size (2550 ha) in this region encourages farmers to grow cash crops in short rotations needing heavy inputs. Potato is the most profitable crop, followed by sugar beet and vegetables such as onion and cabbage. Cereals are financially less attractive but are needed as break crops. Most rotations are for only 3 or 4 years. Consequently, beet and potato cyst nematodes (Heterodera spp and Globodera spp) cause serious problems, forcing farmers to fumigate soil regularly as a curative or preventive measure. The Integrated prototype for the Central Clay area has been developed since 1979 on the 'Development of Farming Systems' experimental farm at Nagele (central clay region). The farm size is 72 ha and the soil is heavy sandy marine clay (24% clay). Three farming systems were studied until 1991: Integrated, Conventional (rotations see Table 1) and Ecological. In 1991 the experimental layout was drastically Table 1 Crop rotations of the different systems and periods at the experimental farm at Nagele Integrated and Conventional 1986-1990

Integrated and Experimental 1991

Year

Crop

Year

Crop

l

i

One-half ware, one-half seed potato Sugar beet

3

One-halfware, one-half seed potato One-halfdry pea, one-quarter carrot, one-quarter onion Sugar beet

4

Winter wheat

2

2 3

2. Material and methods 4

2.1. Prototypes of integrated systems in the Netherlands Integrated prototypes for arable farming were

Experimental, advanced integrated.

One-half carrot, one-half onion (experimental: one-half carrot, one-half chicory) Winter wheat (Experimental: one-half winter wheat, one-half sugar barley)

321

revised. Because of the promising results of the Integrated prototype (Wijnands and Vereijken, 1992; economically the Integrated system was competitive with the Conventional reference system) and the subsequent progress in policy (Ministry of Agriculture, Nature Management and Fisheries, 1990, 1991), the Conventional reference system was no longer needed. It was therefore replaced by a demonstration-Integrated prototype meeting the policy aims of 2000. Subsequently a new Integrated prototype (Experimental) was designed, aimed at further reductions in inputs of pesticides and nutrients (rotations; see Table 1). Concerning the farming methods that are used in the Integrated system the following specifications can be given. Concerning the Multifunctional Crop Rotation: a potato cropping frequency of 1:4 is considered as an acceptable compromise between a more sound rotation (1:5 or 1:6) and more profitable short rotations (1:3) with more biotic stress and therefore requiring more inputs. The Integrated Nutrient Management strategy applied is based on the environmental safe and agronomic efficient use of manure as a basic source of nutrients and organic matter and is aimed at minimum losses. For more details on Ecological Infrastructure Management and the economic aspects see Wijnands (1994). The ICP strategy that was followed will be elaborated in Section 2.2.

2.2. Integrated crop protection The role of crop protection in an Integrated system is, additional to all the other methods, to efficiently control the residual harmful species, with minimal use of well selected pesticides. ICP focuses on the real problems, namely the residual ones, after all other methods are designed and optimised. Consequently this means that in the design of the multi-objective methods Multifunctional Crop Rotation and Integrated Nutrient Management, all crop protection aspects are taken into consideration. This concerns for instance the choice of the (inter)crops, their frequency and sequence as well as the spatial aspects of the crop rotation (Vereijken, 1994). Moreover a well balanced Ecological Infrastructure Management should enhance the stability of the system. When designing an Integrated Nutrient Management strategy, the interaction between weeds, pests and dis-

eases and soil fertility, and the plant nutritional status are taken into account. The ultimate objective of the Integrated and Ecological systems with respect to pesticides is the same: zero use and zero negative impact on environment and ecology. However whilst an Ecological system radically abandons pesticide use and consequently produces under label for higher prices on special markets, Integrated systems still use pesticides since production for the world market does not allow to radically abandon pesticides. However also for Integrated systems the target can only be zero use of pesticides. This is a major challenge for agronomy and crop protection science. First of all the use of pesticides in an Integrated approach can be minimised by putting maximum emphasis on prevention (resistant varieties, cultural measures such as adapted sowing date and row spacing). Whenever a disease, pest or weed population occurs a correct interpretation of the need for control (guided control systems, thresholds, signalising systems, etc.) can prevent unjustified use of pesticides. Secondly all available non-chemical control measures (mechanical weed control, physical and biological control) should optimally be integrated in effective and manageable control strategies. Pesticides are only necessary in very specific cases. They always have to be integrated in crop- and location specific control strategies. Application methods are preferred that lead to a minimum use, such as seed treatment and row- or spotwise application. The latter techniques require careful integration of chemical and mechanical techniques when applied in weed control. Appropriate dosages and when possible a curative approach (field- and year specific) further reduce the input. The residual required pesticide use then requires a careful selection of pesticides to avoid disturbance of non-target organisms (selectivity) and to minimise the exposure of the environment to pesticides.

3. Pesticides

3.1. Behaviour and impact Current agriculture depends to a great extent on pesticides. It is estimated that world wide some 2.5

322 million t of pesticides are applied annually in agricultural crops. Pesticides can be described as the only group of toxic chemicals which are intentionally dispersed in the environment (The Pesticides Trust UK, information leaflet). Only a fraction of the pesticides gets in contact with its target organisms (directly or indirectly). Pimentel (1995) estimates that in the case of pests, only 0.4% of the pesticide contacts its target pest. Inevitably a large part of the applied pesticides become part of the abiotic environment. Pesticides may volatilise into the air, runoff or leach into surface- and groundwater, be taken up by plants or soil organisms or remain in the soil, depending on pesticide properties, climatic and crop conditions, soil type and 'infrastructure' (slope of fields, nearness of surface water, hydrology, etc.). The environment thus gets exposed to a certain pesticide load. The combination of pesticide properties and 'environmental' conditions determines the 'persistence' of the compounds (adsorption, degradation, photolysis, etc.). Pesticide behaviour in soil (persistence and leaching to groundwater) has been studied extensively and is relatively well known. The total seasonal losses in runoff rarely exceed 5-10% of the total amount applied (Leonard, 1990). The fraction removed by leaching is probably less than 5-10% (Taylor and Spencer, 1990) however both runoff and leaching have a very significant impact on water quality causing world wide serious concern over the past three decades. Volatilisation is the major cause of pesticide loss. Volatilisation losses up to 80-90%, within a few days after application, have been reported (Taylor and Spencer, 1990). A recent study in the Netherlands (in the framework of the evaluation of the crop protection policy) estimates that some 50% of the total pesticide use volatilises (Multi-Year Crop Protection Plan, 1996). The fate of pesticides in the atmosphere is relatively unknown. However by atmospheric transport and deposition (global distillation) many pesticides may be distributed all over the earth (Gregor and Gummer, 1989; Atlas and Schauffler, 1990; Schomburg and Glotfelty, 1991; Simonich and Hites, 1995). Pesticides unavoidably cause ecological effects, since no pesticide is specifically toxic to only one species. Consequently the presence of pesticides in the abiotic environment is potentially a threat for all involved biota (non-target). The magnitude and dif-

ferentiation of this threat is only very partially known and quantified. Pesticide toxicity for humans and some mammals is relatively well known. Much less is known about the effects on other biota, the so-called ecotoxicity. A proper evaluation of the ecotoxicity of a substance is virtually impossible since it involves thousands of different species that react differently when exposed to a certain substance. It not only involves direct toxicity but also mid- and long term effects on, for instance fertility, vitality and population dynamics. This knowledge calls for a radical strategy. A preventive strategy that aims at minimising any potential effect of pesticides on biota. Therefore the exposure of the environment to pesticides should be minimised. This should be reached by minimising the pesticide requirements of farming systems (e.g. by ICP, see Section 2.2) and consequently careful selection of pesticides taking into account the extent to which the environment gets exposed to pesticides. The use of pesticides is currently often quantified as number of treatments, as kg active ingredients or as a relative number expressing the ratio used dose/recommended full field dose. These parameters only quantify use and cropping technique. In Section 3.2 the quantification of pesticide properties in terms of potential presence in the environment will be elaborated. 3.2. Environment exposure to pesticides

EEP is quantified by taking into account the active ingredient properties and the amount used. EEP-air = active ingredient (kg/ha) x vapour pressure (VP at 20-25°C) (Pa). EEP-soil = active ingredient (kg/ha) x50% degradation time (DT50) (days). EEP-groundwater = EEP-soil (kg days/ha) x mobility of the pesticide (-). Mobility = Kom; g o m -" partitioning coefficient of the pesticide over dry matter and water fraction of the soil/organic matter fraction of the soil. The properties of active ingredients of pesticides, i.e. VP, DT50 and Kom, are known under standardised conditions, since this is required for the approval procedures (Linders et al., 1994). For instance the ratio DT50/Kom in the Netherlands is used in model studies to establish the leaching risk as part of the approval procedures.

323

These rather simple calculations do not take into account any division of the compounds over the three compartments of the abiotic environment nor do they relate to the period of the year and the crop conditions (soil cover) during application. EEP quantifies the maximum risk of environment exposure to pesticides and can be used to evaluate pesticide use or to select pesticides. Of course any additional knowledge of ecological effects should be taken into consideration. EEP can be quantified per pesticide, but also be summarised as EEP per crop (sum of EEP per pesticide) or EEP per farm (weighted average of EEP per crop with respect to area). Comparative surveys of available pesticides provide the basis for rational pesticide choice. Evaluation of pesticide use implies quantification of the EEP-water, -air and -soil per pesticide, per crop and per farm. Pesticides then can be ranked by calculating their relative contribution to the EEP per farm (Table 7). This provides a rational basis for targeted improvement in EEP. EEP targets should be achieved by: (1) substitution of the highest ranking compounds by non-chemical measures or lower ranked pesticides, or (2) reducing the used amount by a more appropriate dose or by bandspray or spotwise treatments.

4. Results The results of crop protection in terms of pesticide use and EEP of the Nagele experimental farm over the period 1986-1990 are presented, including an outlook

Table 2 Average marketable crop yields (t/ha) in the Integrated and Conventional farming systemat Nagele in different periods Crop

Ware potato Seed potato Sugar beet Winter wheat Pea Winter carrot Sown onion

1986-1990

1992-1995

Conventional Integrated

Integrated

54.4 33.9 64.3 7.6 5.0 52.2 40.6

53.1 32.0 59.2 9.7 69.3 45.3

54.6 34.3 60.2 6.7 4.7 52.2 31.0

into the period 1992-1996. The crop rotation in both periods is given in Table 1. Yields of the crops in the Conventional and Integrated systems were in general similar, with exception of winter wheat and sown onion (Table 2). For ware potato (cultivar value), seed potato (cultivar value), sown onion (quality) and sugar beet (sugar content and extractability) higher product prices were achieved in the Integrated system. Prices of produce for the other crops were similar in both systems. The costs of pesticides and fertilisers were lower and the costs of seeds and tubers were higher in the Integrated system. The final gross margin per crop was higher in the Integrated system, still with exception of winter wheat and sown onion. At farm level, costs of machinery and labour were slightly higher in the Integrated system, but these did not fully remove the financial advantage of the higher gross margins. As a result the net surplus of the Integrated system was slightly higher than that of the Conventional system (Wijnands and Vereijken, 1992). More details about the physical and financial results of the various systems have been reported by Bos et al. (1992). The yields of the 1992-1995 period show considerable improvement in yields of wheat (cultivar and N-management), onion (N management) and carrot (cultivar and Nmanagement) in the Integrated system. For all crops, the yield levels are now similar to the average 'conventional' yields of the region (Conventional no longer available at the experimental farm). 4.1. Pe s tic id e use

Table 3 specifies the number of ICP interventions. Compared to the Conventional system the annual input of pesticides in kg active ingredients/ha in the Integrated system was reduced by 65%, excluding nematicides and by 90% if nematicides are included (Table 4). The largest reduction in active ingredient use was realised by substituting dichloropropene (DCP), a soil fumigant used to control potato cyst nematodes in the Conventional system, by non-chemical measures such as the use of appropriate cultivars based on detailed monitoring techniques. Herbicide input in the Integrated system was largely replaced by mechanical control and by band spraying or low dose techniques (Tables 3 and 4).

324 Table 3 Number of interventions for crop protection (n/crop) in the Conventional and Integrated farming system (1986-1990) at Nagele Weeds

Conventional Integrated

Pest/diseases

Mechanical

Thermal

Chemical

Total

Chemical

0.9 2.0

0.2

2.6 1.5

3.5 3.7

4.2 1.9

Per herbicide application, the amount of active ingredient used was 2 5 - 5 0 % less. The labour demand increased because the band spraying and mechanical interventions took more time than full field herbicide spraying. In seed and ware potato crops it was not necessary to use herbicides for weed control. Growing winter wheat at wider inter-row spacing (26 cm) enabled herbicides to be replaced by mechanical control. Fungicide input was reduced by using resistant cultivars, moderate nitrogen supply, control thresholds and decision support systems. The largest reduction at farm level was achieved in potato (Table 4 and Table 5). Fungicide input in onion was largely reduced by supervised control based on monitoring initial infestation by Botrytis squamosa and weather conditions (Table 4). Growth regulators were only used in sown onions to inhibit sprouting during storage. Insecticide input was minimal due to low insect pressure and the use of control thresholds, reduced dose techniques and band spraying. Fig. 1 shows the further decrease in pesticide use in

Total

7.7 5.6

the most recent period (1992-1996). The reduction in fungicide use is mainly based on the substitution of the 'old' compounds used to control Phytophthora infestans by a new low active ingredient compound called fluazinam. The reduction in herbicide use is based on a further increased and optimised use of mechanical techniques and appropriate dosage herbicide systems Compared to the Conventional system of 1 9 8 6 1990 the reduction percentage increased. In absolute terms the level of pesticide use is very low (Fig. 2). Does this also mean that the environmental impact of the Integrated system is much lower with respect to pesticides?

4.2. Environment exposure to pesticides Active ingredient input and EEP were quantified for the Conventional and Integrated system in 1988, and to demonstrate the progress that was made also for the Integrated system in 1992 (Table 6). From the Conventional 1988 system (representative for 1986-

Table 4 Annual input of pesticides (kg active ingredients/ha) in the Integrated and Conventional farming system (1986-1990) at Nagele Herbicides

Fungicides

Insecticides

Growth regulator Nematicides

Total

lnte- Conven- I n t e - Conven- I n t e - Conven- I n t e - Conven- l n t e - Conven- I n t e - Convengrated tional grated tional grated tional grated tional grated tional grated tional 0.1 a Ware potato Seed potato 2.0a Sugar beet 1.3 Winter wheat 1.2 Pea 2.1 Winter carrot 1.4 Sown onion 2.7 System average 1.4

2.5 a

4.5a 3.8 3.7 3.3 3.5 9.0 4.0

8.9 4.3 0.0 0.3 0.6 0.0 2.5 2.0

19.6 13.9 0.0 2.3 1.1 0.7 8.6 5.5

0.0 0.3 0.1 0.0 0.2 1.3 0.0 0.2

0.4 0.8 0.3 0.1 0.4 3.7 0.1 0.5

. 0.0 . . 1.8 0.1

.

.

.

. .

. .

0.6 . . 2.3 0.3

-

104.6 135.8

-

-

-

29.7

9.0 6.6 1.4 1.6 2.9 2.7 7.0 3.7

124.1 155.1 4.1 6.7 4.8 7.9 20.0 40.0

325 l 86-90 / 192-96 j X 92-961

Table 5 Number of interventions (n/crop) and fungicide input (kg active ingredients/ha) for Phytophthora infestans control in the Integrated and Conventional farming system (1986-1990) at Nagele Ware potato

Seed potato

Ii

100 90 80

Integrated Conventional Integrated Conventional

70 60

Interventions 6.3 Active 8.9 ingredients

11.4 19.5

2.6 4.2

5.4 12.0

50 40 30

1990) to the Integrated 1988 system the input of pesticides was strongly reduced over all crops (Table 6). In 1992 (representative for 1992-1996) the Integrated system reduced the pesticide input even further, again over all crops. The main cause of the drastic decline in EEP-air, water and-soil going from the Conventional 1988 system to the Integrated 1988 system is that DCP, the soil fumigant used to control potato cyst nematodes, has been replaced by non-chemical measures. DCP is extremely volatile and used in high dosages (80-110 kg active ingredients/ha). The major change going from the Integrated 1988 system to the Integrated 1992 system again occurs in the potato crop. The dithiocarbamates and fentin acetate, fungicides against late potato blight (Phytophthora infestans), were replaced by fluazinam, a new low dosage compound. From the Conventional 1988 system to the Integrated 1992 system the EEP-air, -water and -soil

0 fungicides

==herbicides

C 86-90

1 86-90

1 92-96

X 92-96

Fig. I. Annual input of pesticides (kg active ingredients/ha) in different systems and periods (I, Integrated; C, Conventional; X, Experimental) at Nagele.

20 10 0 ,Q_

t"

~

~

t~

m

¢-

..,

Fig. 2. Reduction (%) in input of pesticides (kg active ingredients/ ha) of the Integrated (I) (1986-1990) and the I and Experimental (X) farming system (1992-1996) in comparison to the Conventional farming system (1986-1990) at Nagele.

were reduced by >>99, 96 and 98%, respectively. The active ingredient use was reduced by 'only' 95% (including nematicides). The basis for the reduction in EEP is ICP. However the beneficial effect of selecting pesticides based on EEP is large as is apparent from the foregoing, especially in EEP-air. Table 7 presents, as an example the pesticides used in the Integrated 1992 system ranked according to their share in the farm average EEP-soil. Fluazinam is present at position 1 and 3 on the list, in ware-and seed potato (32%), respectively. It is used full field, to prevent late blight in potatoes, however in a low dosage system based on the higher resistance of the cultivars cropped in the Integrated system. Then herbicides used in band-spray appropriate dosage systems in sugar beet are present at position 2, 5, 11 and 14 (23%). To reduce the EEP further, harrow treatments might replace the last one or two low dose herbicide applications in sugar beet. Other compounds, that would decrease EEP are not available yet. Then glyphosate is used to control spot-wise perennial weeds in different crops (6, 9, 10; 12%). Pirimicarb and propiconazole are used in winter wheat (10%). However, the crop growth stage during application will largely prevent these compounds to reach the soil. Other compounds on the list contribute only marginally to the EEP-soil per farm.

326 Table 6 Pesticide use (kg/ha, active ingredients) and Environment Exposure to Pesticides (EEP)-air, -water and -soil by crop for the Conventional and Integrated farming system at Nagele in different years Active ingredients

Ware potato Seed potato Winter wheat Sugar beet Sown onion Average c

EEP-aira

1988

1992

1988

Conven- Integtional rated

lntegrated

Conventional

196.8 187.7 5.1 2.1 23.4 60.3

7.2 3.5 1.0 1.8 3.8 2.9

156 152 141 1.1 7.4 86

9.1 6.4 2.6 0.4 9.8 4.5

EEP-waterb

EEP-soilb

1992

1988

1992

1988

Integrated

Integrated

Conven- lntegtional rated

Integrated

Conventional

lntegrated

Integrated

0.9 0.9 1.5 0.1 3.8 1.3

0.5 0.3 0.4 0.4 0.7 0.4

3493 3218 162 103 722 1170

82 66 10 55 47 46

196.6 187.5 5.0 3.0 23.4 60.5

9.7 6.4 2.6 0.4 9.8 4.6

2.0 1.4 0.4 1.6 2.1 1.4

376 185 25 15 401 ! 29

1992

aln log (106 x EEP-air); bsee text; ¢cropping plan.

5. Discussion

The presented approach clearly distinguishes three phases in pesticide use" (1) the use (characterised by number of applications and kg active ingredients/ha);

(2) the exposure of the environment to pesticides (quantified by the EEP) and; (3) the effects on biota. The presented approach focuses on the first and second step. The governmental approval procedures for pesticides guarantee that the 'worst' pesticides

Table 7 Ranking of pesticides based on share in farm level EEP-soil for the Integrated farming system (1992) at Nagele Product

Type a

Active ingredients

Method b

Crop

Share farm level %¢

Cumulative share in farm level

1 2 3 4 5

Shirlan Goltix Shirlan Pirimor Betanal Progress

F H F I H

FF RT FF FF RT

Ware potato Sugar beet Seed potato Seed potato Sugar beet

22 11 10 7 7

22 33 43 50 57

6 7 8 9 10 11

Roundup Royal MH-30 Tilt Roundup Roudnup Betanal Tandem

H GR F H H H

SPOTW FF FF SPOTW SPOTW RT

Sugar beet Sowed onions Spring barley Sowed onions Chicory Sugar beet

7 5 3 3 3 3

64 69 72 75 78 81

12 13 14 15

Chloor-IPC Kerb Pyramin Flow Chloor-IPC

H H H H

Fluazinam Metamitron Fluazinam Pirimicarb Ethofumesate + desmedipham + fenmedipham G lyphosate Maleine-hydrazide Propiconazole Glyphosate Glyphosate Ethofumesate + fenmedipham Chlorpropham Propyzamide Chloridazon Chlorpropham

RT RT FF RT

Sowed onions Chicory Sugar beet Chicory

3 2 2 2

84 86 88 90

Ranking

a GR, growth regulator; F, fungicide; H, herbicide; I, insecticide. bFF, full field; RT, row treatment; SPOTW, spotwise. ¢Share in farm level in % = [(EEP by pesticide x crop share in farm area)/(EEP by farm)] xl00; Crop share in farm area = area of one crop/ total area of farm.

327

from an environmental and ecological point of view are not approved. ICP enables a minimum use of pesticides. EEP is a useful instrument to select then within the range of approved pesticides. EEP in combination with ICP enables a quantitative approach to a stepwise, targeted reduction in pesticide use and environmental impact. Van der Werf (1996) reviewed different approaches to evaluate pesticide impact on environment and biota. The reviewed methods show considerable differences in the parameters that are considered to asses environmental impact. From this review that includes the proposed approach in this article it is clear that EEP is the only approach that takes volatilisation of pesticides into account. It is also the only one that 'on purpose' does not consider effect on biota, since an overall comprehensive assessment is virtually impossible. Overall quantitative scores of 'ecosafety' therefore may easily lead to unjustified classification of a pesticide as being safe. The more radical approach of EEP enables a basic approach towards prevention. The volatilisation losses are obviously the largest ones and thus have to be taken into account when trying to reduce environmental impact. It can be argued, whether the parameter VP is the best to account for volatilisation losses. Henry's law constant (Kh, the ratio of the VP to the water solubility) might also be an appropriate criterion for the volatility of a pesticide. However in a recent study in the Netherlands VP was also chosen in the model calculations as most simple and accurate prediction parameter (Multi-Year Crop Protection Plan, 1996). Both DT50 and Komare in fact exponentially related to the leaching risk. Nevertheless by using the straight forward ratio DT50/Kom the risk of leaching is estimated properly. The presented approach might be extended to consider all involved processes, however the risk of loosing the simple and user-friendly character of EEP should be taken seriously. The Integrated prototype as designed, tested and improved on the Nagele experimental farm for the Central Clay conditions proved to have good perspectives in terms of minimising pesticide input and environment exposure to pesticides. ICP and pesticide selection based on EEP proved to be effective. Since the economic perspectives of the Integrated prototype are equal to those of the current 'conventional' farm-

ing systems, large scale implementation in practice inevitably should be the next step

Acknowledgements The author cordially thanks Dr. Boesten, Ir. Spoorenberg and particularly Dr. Vereijken for the support in developing the concept of Environment Exposure to Pesticides.

References Atlas, E.A. and Schauffler, S., 1990. Concentration and variation of trace organic compounds in the north pacific atmosphere. In: D.A. Kurtz (Editor), Long Range Transports of Pesticides. Lewis, Chelsea, MI, pp. 161 - 183. Bos, A., Janssens, S.R.M. and Krikke, A.T., 1992. Analysis of economic results. In: H.H. Cheng (Editor), More Sustainable Farming Systems for Arable Farming. Themaboekje hr. 14, PAV, Lelystad, pp. 126-181 (in Dutch). Gregor, D.J. and Gummer, W.D., 1989. Evidence of atmospheric transport and deposition of organochiorine pesticides and polychlorinated biphenyl's in Canadian arctic snow. Environ. Sci. Technol., 23: 561-565. Leonard, R.A., 1990. Movement of pesticides into surface waters. In: Pesticides in the Soil Environment. Soil Science Society of America Book Series, No. 2, Madison, WI, pp. 303-349. Linders, J.B.M.J., Jansma, J.W., Mensink, B.J.W.G. and Ottermann, K., 1994. Pesticides: Benefaction or Pandora's Box, A Synopsis of the Environmental Aspects of 243 Pesticides. Report no. 6791014. National Institute of Public Health and Environmental Protection, Bilthoven, 201 pp. Ministry of Agriculture, Nature Management and Fisheries, 1990. Agriculture Structure Memorandum. Government decision (in Dutch). Ministry of Agriculture, Nature Management and Fisheries. DS, The Hague (essentials available in English). Ministry of Agriculture, Nature Management and Fisheries, 1991. Multi-Year Crop Protection Plan. Government decision. (In Dutch). Ministry of Agriculture, Nature Management and Fisheries. SDU, The Hague (essentials available in English). Multi-Year Crop Protection Plan, 1996. Multi-Year Crop Protection Plan. Evaluation emission 1995, background document. IKC-L. Ede, 127 pp. plus annexes (in Dutch). Pimentel, D., 1995. Amounts of pesticides reaching target pests: environmental impacts and ethics. J. Agric. Environ. Ethics, 8: 17-29. Schomburg, C.J. and Glotfelty, D.E., 1991. Pesticide occurrence and distribution in fog collected near Monterey, California. Environ. Sci. Technol., 25: 155-160. Simonich, S.L. and Hites, R.A., 1995. Global distribution of organochlorine compounds. Science, 269:185 l - 1854. Taylor, A.W. and Spencer, W.F., 1990. Volatilisation and vapor transport processes. In: H.H. Cheng (Editor), Pesticides in the

328

Soil Environment. Soil Science Society of America Book Series, No. 2, Madison, WI, pp. 213-269. The Pesticides Trust UK, information leaflet. Eurolink Business Centre, 49 EFFRA Road, London, SW2 IB2, UK. Vereijken, P., 1994. 1. Designing Prototypes. Progress Reports of Research Network on Integrated and Ecological Arable Farming Systems for EU- and Associated Countries (concerted action AIR3-CT927705). AB-DLO, Wageningen, 87 pp. Vereijken, P., 1995.2. Designing and Testing Prototypes. Progress Reports of Research Network on Integrated and Ecological Arable Farming Systems for EU- and Associated Countries (concerted action AIR3-CT927705). AB-DLO, Wageningen, 76 pp. Vereijken, P., 1997. A methodical way of prototyping integrated and ecological arable farming systems (I/EAFS) in interaction with pilot farms. In: M.K. van lttersum and S.C. van de Geijn (Editors), Proceedings of the 4 th ESA Congress, Elsevier, Amsterdam. Vereijken, P. and D.J. Royle, 1989. Current status of Integrated arable farming systems research in Western Europe. IOBC/ WPRS Bull., XII: 5.

Werf van der, H.G.M., 1996. Assessing the impact of pesticides on the environment. Agric. Ecosyst. Environ., 60: 81-96. Wijnands, F.G., 1992. Introduction and evaluation of integrated arable farming in practice. Neth. J. Agric. Sci., 40: 239-250. Wijnands, F.G., 1994. Focus on IAFS prototyping in Nagele experimental farm (NL1). In: P. Vereijken (Editor) 1. Designing Prototypes. Progress Reports of Research Network on Integrated and Ecological Arable Farming Systems for EU- and Associated Countries (concerted action AIR3-CT927705). AB-DLO, Wageningen, pp. 79-84. Wijnands, F.G. and Vereijken, P., 1992. Region-wise development of prototypes of Integrated arable farming and outdoor horticulture. Neth. J. Agric. Sci., 40: 225-238. Wijnands, F.G., van Asperen, P., van Dongen, G.J.M., Janssens, S.R.M., Schrrder, J.J. and van Bon, K.B., 1995. Innovatiebedrijven ge'fntegreerde akkerbouw, beknopt overzicht technische en economische resultaten. PAGV-verslag nr. 196. PAV, Lelystad, 126 pp. (in Dutch with English summary and tables and figures).

1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

329

Use of agro-ecological indicators for the evaluation of farming systems C. Bockstaller a'*, P. Girardin b, H.M.G. van der

Werf b

aAssociation pour la Relance Agronomique en Alsace (ARAA), Laboratoire d'Agronamie, 68021 Colmar Cedex, France blNRA, Laboratoire d'Agronomie, BP 507, 68021 Colmar Cedex, France

Accepted 13 June 1997

Abstract

For the development of integrated arable farming systems (IAFS), tools are needed to evaluate the achievement of agronomic and environmental objectives, in order to optimize the systems. A set of agro-ecological indicators (AEI) is proposed. These indicators estimate the impact of cultivation practices on the agrosystem and its environment. AEI are aimed, first of all, at being used as decision aid tools, to help farmers to adapt their cultivation practices to IAFS requirements, from one cropping year to the next. So far, seven indicators have been elaborated for the evaluation of farming systems: crop diversity, crop succession, pesticide, nitrogen, phosphorus, organic matter and irrigation. The calculation method for the organic matter and pesticide indicators is presented. Possibilities for use of the AEI at the farm and field level, for farmers and decision makers are given with data from a network of 17 commercial arable farms. The elaboration of a single aggregated indicator is discussed. © 1997 Elsevier Science B.V. Keywords: Integrated arable farming; Environmental assessment; Indicator; Organic matter; Pesticide; Agro-ecological indicators; Environmental impact; Decision aid tool; Soil fertility; Hdnin-Dupuis model; Fuzzy logic; Farm network; Multi-criteria method

I. Introduction

Integrated arable farming systems (IAFS), based on the concepts of Integrated Agriculture or Integrated Production (El Titi et al., 1993), are generating an increasing interest as an alternative to conventional intensive farming systems. The results from several pioneer IAFS projects in the 80s and early 90s, concerning profitability as well as environmental and agronomic effects, are promising (Holland et al., 1994). However, for the development of IAFS, tools

* Corresponding author. E-mail: [email protected]

are needed to evaluate the achievement of the objectives (Girardin and Spiertz, 1993), in order to optimize the system (Vereijken, 1992). This kind of evaluation concerning especially the environmental and agronomic objectives needs new methods taking into account specific criteria. Economic criteria used in modern market-oriented agriculture such as the yield or gross margin are no longer sufficient for a global evaluation of agricultural practices, which should include an assessment of their environmental impact. The most obvious approach to environmental impact evaluation is based on direct measurements at the field level. Such a solution is possible on an experimental farm, but its extension to commercial

Reprinted from the European Journal of Agronomy 7 (1997) 261-270

330

farms for routine measurements poses practical problems, because measurements are costly and often time consuming (Sharpley, 1995). Simulation models may be used for impact evaluation, but comprehensive models required for a multi-objective evaluation are not available. Another problem of many models is that they are not adapted for use at farm level, requiring too many input data. In other cases models are not validated for a broad range of conditions (Hansen, 1996). Indicators are an alternative when it is not possible to carry out direct measurements. The term 'indicator' has been defined as a variable which supplies information on other variables which are difficult to access (Gras et al., 1989). Indicators help to understand and to interpret a complex system by: (1) synthesizing data; (2) showing the current state; (3) demonstrating the achievement or not of objectives; (4) communicating the current status to users for management decisions (Mitchell et al., 1995). The parameters proposed by Vereijken et al. (1995) in their methodological framework for prototyping IAFS are either based on direct measurements (e.g. nitrate in ground water) or are calculated from data available on the farm, (e.g. environment exposure to pesticides; EEP). As mentioned before, parameters based on measurements present practical problems for use on a broad scale. The second type of parameter corresponds to what we call indicators. However, the expression of these parameters in physical units as proposed by Vereijken et al. (1995) may not be easily understood by farmers for some parameters (e.g. EEP air expressed in kg/ha per year). As an alternative to indicators expressed in physical units, we propose a set of agro-ecological indicators (AEI) to evaluate the degree of achievement of the IAFS objectives by farming systems. Unlike the parameters proposed by Vereijken et al. (1995), the AEI are calculated with data available on the farm, and thus do not require specific field measurements. Neither are AEI expressed in physical units, but on a scale between 0 and 10, to make them easily understandable by farmers. The IAFS objectives will be restricted to the environmental effects of farming systems (e.g. nitrogen losses) and to some agronomic effects (e.g. on soil fertility). In this way, AEI are estimators of the impact of cultivation practices on the agrosystem and its environment. However, as

mentioned by Mitchell et al. (1995), the assessment of environmental impacts should affect management decisions. We therefore consider AEI to be, first of all, decision aid tools to help farmers adapt their cultivation practices to IAFS requirements, from one cropping year to the next. However, AEI may also be used by decision makers to monitor or to evaluate their agri-environmental policies. The purpose of this paper is to present the principles of elaboration of the AEI, illustrated by the calculation methods of two indicators, and to discuss the possibilities for their use.

2. Elaboration of the AEI

An indicator assesses the impact of a cultural practice on one or several objectives (Table 1). We defined a list of environmental and agronomic objectives in relation to the principles of IAFS (El Titi et al., 1993). Some of them are close to the objectives defined by Vereijken et al. (1994). We did not include economic and social objectives. From the list of objectives a set of indicators was defined and connected to the objectives (Table 1). These indicators are identified as 'key' components for the management of a farming system. Crop rotation (or cropping sequence) and cultivation practices within each crop make up the two parts of an arable farming system. Cultivation practices are assessed by the AEI listed in Table 1. We added the crop sequence indicator (Bockstaller and Girardin, 1996) to that list of AEI. The proposed crop sequence indicator assesses whether the cropping sequence satisfies the requirement of IAFS, for example concerning the use of natural regulation mechanisms, or of nitrogen supply from the previous crop (El Titi et al., 1993). The crop sequence indicator, contrary to the other AEI, is not aimed at being used as an estimator of an environmental impact but as a decision aid tool along with the other AEI, in order to help farmers adapt their farming system to the principles of IAFS. For simplicity of presentation, AEI are constructed in such a way that they take a value between 0 and 10. The value 7 represents the achievement of a minimum level for IAFS requirements (e.g. the organic matter content is maintained at a satisfactory level). A value below 7 indicates that the IAFS requirements are not

331

Table 1 List of agro-ecological indicators in association with their environmental and agronomic objectives Agro-ecological indicators Objectives

Nitrogen Phosphorus Pesticide Irrigation Organic Energy Crop matter diversity

Soil Soil structure cover

Ecological structures

Protection of'.

Ground water X quality Surface water quality Air quality Soil quality Non-renewable resource Biodiversity Landscape quality

X (x)

x x

X

x x (x)

x

x

x

x

(x)

x x

x

x x

Each mark represents an association, (X) means that the objective is not fully assessed by the indicator for the moment. met (e.g. the soil organic matter content decreases) and a value above 7 indicates that the farmer does better than the minimum IAFS requirements (e.g. the soil organic matter content increases). The reference level, corresponding to a value of 7, is based on scientific knowledge or expert judgment. This level may be adapted to local conditions by local experts. For example: maintaining the soil organic matter at a satisfactory level is an IAFS principl e admitted by experts. This level should be quantified according to local conditions (soil type, climate, etc.). The AEI are calculated with data available on the farm (cultivation practices recorded by the farmer, soil analyses, permanent characteristics such as field size, slope, etc.). All indicators, except for the crop diversity indicator, are calculated at the field level and then weighted by the field size to obtain a mean farm value. The time scale of calculation is generally the period between the harvest of the preceding crop and the harvest of the crop in the current year. The calculation algorithm of an AEI is based on available scientific knowledge or on expert judgement. In some cases (e.g. organic matter indicator, see Section 3.1) the indicator is based on a simple model for which the required data are available on the farm. The nitrogen and phosphorus indicators compare the farmer's practices with a sequence of practices corresponding to the minimum requirements for IAFS. A sequence of farmer's practices in accor-

dance with this reference sequence yields an indicator value of 7. Practices deviating from the reference sequence may lead to loss of nitrogen (or phosphorus) to the environment and will yield a sum of plus or minus marks, which is added to the value of 7. These plus or minus marks are based on the equivalence: one point is equal to 30 kg N/ha or 30 kg P2Ofl ha. Thus, in the case of the nitrogen and phosphorus indicators, or when a simulation model is used, the evaluation is quantitative. For other indicators such as crop diversity, crop sequence and pesticide, the evaluation is qualitative because quantitative data and adapted models are not available. So far, seven indicators have been elaborated for the evaluation of farming systems: nitrogen, phosphorus, pesticide, irrigation, organic matter, crop diversity and crop sequence. Indicators for: energy, soil structure, soil cover and ecological structures are planned. Following their elaboration, the indicators should be evaluated in order to improve them, if necessary. Three tests are proposed. The sensitivity test aims to observe the behaviour of the indicator when the value of its input variables is varied. Two examples will be given below. The purpose of the probability test is to study the soundness of the indicator as an estimator of environmental impact. The test consists of the establishment of a relationship between the values taken by the indicator and those taken by an observed or mea-

332

sured criterion which reflects the environmental impact. The usefulness test should be carried out in order to see whether the tools developed are helpful for the target users. These two latter tests will not be elaborated in this paper.

3. Calculation algorithm of two indicators

O t~

.u

3.1. The organic matter indicator The organic matter indicator (lMo) evaluates the effect of farmers' practices on the evolution of soil organic matter (SOM) in order to help farmers adapt their cultural practices to maintain the SOM at a satisfactory level. The calculation of the indicator as given in Eq. (1) is based on the comparison of the organic matter (OM) inputs by manure and crop residues with recommended levels of inputs:

IMo = 7Ax/A a

(1)

where A x is the mean OM input of the 4 preceding cropping years, and AR is the recommended level of OM input. The indicator ranges from 0 to 10. If the indicator is less than 7, the organic matter inputs are not sufficient, if it is equal to or above 7, the recommended levels of input are reached or exceeded. The recommended levels of inputs are expected to maintain a satisfying level of SOM in the long term. They were obtained by running the H6nin-Dupuis model (Mary and Gu6rif, 1994) for several classes of clay and limestone contents in the soil (Table 2). The H6nin-Dupuis model is a monocompartment model of the evolution of SOM. The calculation of OM inputs by crops and manure are based on data of Boiffin et al. (1989). Examples of OM inputs are given in Table 3. The indicator was Table 2 Recommended level of OM inputs (AR) (kg OM/ha) Clay content (%) Limestone content

15-20

20-25

25-30

30-35

0-5% 5-15% > 15%

1085 945 840

945 840 735

910 770 700

840 735 630

6 5 4 i

3 2

1 0 17% 28% 17% 28% 17% Clay Clay Clay Clay Clay 0% 0% 10% 10% 20% CaO CaO CaO CaO CaO

28% Clay 20% CaO

Fig. 1. Sensitivity of the organic matter indicator to soil characteristics (clay and limestone contents), to the crop succession (crop residues incorporated, without manure) and to yield. The indicator was calculated with data from Tables 2 and 3. Crop succession 1: grain maize monoculture (yield, respectively, 11 t/ha and 9 t/ha); crop succession 2: grain maize (11 t/ha)/winter wheat (8 t/ha); crop succession 3: sugar beet (70 t/ha)/winter wheat (8 t/ha)/grain maize (10 t/ha)/winter wheat (8 t/ha). 0, crop suc. 1 (yield 1 ! t.ha-J); O, crop suc. 1 (yield 9 t.ha-~); ~, crop suc. 2; • crop suc. 3; - - - , recommended value.

calculated for a range of clay and limestone contents in soil, for several rotations and yield levels, with data from Tables 2 and 3. Fig. 1 shows the sensitivity of the indicator to these variables.

3.2. The pesticide indicator (Ipest) The environmental impact of a pesticide largely depends on: (a) the amount applied, (b) its rate of degradation, (c) its partitioning to the air, the surface water and the groundwater, (d) its toxicity to the species in those environmental compartments (van der Weft, 1996). Several methods have been proposed to estimate pesticide environmental impact (Levitan et al., 1995; van der Weft, 1996). None of these methods aggregates the four criteria mentioned above into a single output value. The pesticide indicator (Ip~t) we propose is based on an expert system using a collection of fuzzy mem-

333

value depends on a set of 16 decision rules, five of which are given below as an example:

bership functions and decision rules. This technique is robust when uncertain or imprecise data is used. It also allows the aggregation of knowledge which is expressed in every-day language (Bouchon-Meunier, 1993). In a first step, /pest is calculated for each single application of an active ingredient. The value of/pest depends on four modules: P (presence, reflecting amount and persistence), Rgro (risk of groundwater contamination), Rsur (risk of surface water contamination) and Rair (volatilization risk). The value of each of these modules depends on two to four input variables according to fuzzy decision rules which will not be presented here. For all modules the membership to a fuzzy set F (favorable) and a fuzzy set U (unfavorable) has been defined. The value of P depends on the rate applied and its soil degradation half-life. The value of Rgro depends on the leaching potential of the pesticide, the site of application (in the soil, on the soil, or on the crop), the month of application and its toxicity to man. The value of Rsur depends on the run-off risk of the field, the site of application and the toxicity of the pesticide to aquatic organisms. The value of Rair depends on the volatility of the pesticide and its site of application. For the air the toxicity is not taken into account because an appropriate variable is not available. The four modules (P, Rgro, Rsur and Rair) can be either considered individually or aggregated in a single value, for instance by summation, multiplication or a combination of both. We present here a mode of aggregation using decision rules to calculate Ipe~t. For each application of an active ingredient,/pest can take values between 0 (no risk of environmental impact) and 1 (maximum risk of environmental impact). Its

(a) If P is F and Rgro is F and Rsur is F and Rair is F then/pest is 0.0; (b) If P is F and Rgro is F and Rsur is F and Rair is U then/pest is 0.1; (c) If P is F and Rgro is U and Rsur is U and Rair is U then/pest is 0.5; (d) If P is U and Rgro is F and Rsur is F and Rair is F then/pest is 0.5; (e) If P is U and Rgro is U and Rsur is U and Rair is U then/pest is 1.0. Volatilization risk is given less weight than groundwater and surface water contamination risks (rules a, b and c). This weighting is an example of expert judgment. We valued the pollution of air as less 'important' than the pollution of surface water or groundwater. The presence of a pesticide is considered as an environmental risk, even when the risk for each of the environmental compartments is nil (rule d). Examples of calculation for a range of active ingredient applications are given in Table 4. An analysis of the sensitivity of/pest to variation of input variables is presented in Fig. 2. Each input variable was varied from favorable (0%) to unfavorable (100%), while the other input variables are kept at their median value. In a second step the indicator (IpEsr) is calculated for all pesticide applications on a crop within a year. At this level/PEST takes values between 0 (highest risk) and 10 (no risk): •PEST

10 - k ,Y_,/pesti

""

where k is a constant depending on the crop and Ipesti

Table 3 Examples of OM inputs (kg dry OM/ha) obtained with data from Boiffin et al. (1989). Crop yield (t/ha) Crop

6

7

8

9

10

I1

Wheat (straw incorporated) Wheat (straw exported) Grain maize Sugar beet ( l e a v e s incorporated)

470 250 .

550 290 -

620 330 650

700 370 740 .

780 420 820

. . 900

.

.

.

.

.

12 . .

60 . .

980

. . 420

70

80

500

570

. .

OM inputs include straw and root contribution to the steady SOM. They are derived from root, stem and leaves dry matter. These latter values are obtained from the crop yield (expressed at normalized humidity: 16% for wheat and maize, 85% for sugar beet).

334

Table 4 The values of P, Rgro, R.... Rair and/pest for a number of pesticides applied at their recommended rate in a field with medium run-off risk Pesticide (active ingredient) Rimsulfuron Cyfluthrin 2,4-D Parathion EPTC Carbofuran Glyphosate Alachlor Isoproturon Atrazine Lindane

Application

Sitea (%)

Month

Rate

P/S P/S P/S P/S S S P/S P/S P/S P/S S

June July April Aug. April April April April Jan. April April

0.015 0.040 0.600 0.300 3.600 0.600 4.300 2.400 1.800 1.500 1.350

(50) (100) (50) (100)

(100) (0) (10) (0)

P (kg/ha)

Rgro

Rsur

Rair

]pest

0.00 0.07 0.12 0.23 0.50 0.35 0.66 0.52 0.51 0.63 0.88

0.00 0.00 0.20 0.00 0.00 0.65 0.00 0.23 0.31 0.84 0.46

0.18 0.00 0.19 0.00 0.00 0.00 0.00 0.47 0.48 0.34 0.00

0.00 0.98 0.00 0.48 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.02 0.11 0.13 0.15 0.25 0.30 0.37 0.39 0.41 0.55 0.55

ap/s, applied on plant/soil; S, applied in the soil. The percentage (in parentheses) indicates the crop cover of soil at the time of application.

are first of all decision aid tools for farmers. The type of presentation shown in Fig. 3 is inspired by Magnollay (1993). Vereijken (1997) also used this presentation. It gives an overview of the AEI values at the farm level, allows the comparison with r e c o m m e n d e d values and shows the weak and strong points of an arable fanning system. The farmer knows which cultivation practices he should improve according to IAFS requirements. For most of the indicators, results at the field level are available, so they can be used to help the farmers to take into account the differences between the fields

is the value of the indicator (ranging between 0 and 1) for a single application of an active ingredient i. The constant k is chosen such that a value of 7 for/PEST is obtained when a crop protection program satisfies the m i n i m u m r e c o m m e n d a t i o n s for IAFS.

4. Using agro-ecological indicators The AEI were calculated in 1994 and 1995 with data from a network of 17 commercial arable farms of the Rhine plain in France and Germany. The AEI

0.60 [

•

0.55 i

Soil half-life

--÷-- Soil mobility (GUS)

,-, 0.50

:,-,:$-:---.$

--- - Amount applied

.2 0.45 . . . - - Month

.,-~

.~ o.4o - -. - : : t t ' . ' .

" ll"

e

0.35

Runoffrisk

-- o-- Crop cover

0.30 0.25 [ 0.20 /

0

= ,

10

20

~

!

t

,

,

,

~

;

30

40

50

60

70

80

90

100

Transition interval of each input variable (%)

Toxicity-human

-- a - - Volatility =

Toxicity-aquatic

Fig. 2. Analysis of the sensitivity of the pesticide indicator (Ipe~t)to variation of input variables. Each input variable is varied from favorable (0%) to unfavorable (100%), while the other input variables are kept at their median value.

335

Crop diversity IO

I0

IO

Organic matter

Crop sequence . . . . . . Farm vaalue in 1994

:t

---

Farm value in 1995 Recommended value

Phosphorus Io

lo Nitrogen

Fig. 3. Example of use of the agro-ecological indicators at the farm level (90 ha: grain maize, sugar beet, winter rape, winter wheat). as shown by Fig. 4 in case of the organic matter indicator. This type of presentation helps the farmer to adapt his management to the specific conditions of each field (soil type, crops, etc.). Another use for farmers would be the possibility to follow up the evolution of his cultivation practices over several years or to analyze the cropping history. These indicators can also be used by decision makers (politicians, environmentalists, etc.) to follow up the evolution of cultural practices and the influence of an agri-environmental policy on a sample of farms in a given area (e.g. water catchment). Results can be

presented as in Fig. 5, which shows the organic matter indicator. In this example, the differences between the 2 years are minor, because farmers did not change their rotations and organic fertilization management. We used the database Microsoft Access 2.0® to implement the calculation of the AEI. The software is user-friendly, and with the help of a handbook or a short training it can be used by farmers or farmeradvisers. Farmers may need help from their advisers to enter a few parameters (nitrogen soil mineralization, recommended nitrogen and phosphorus fertilizations, etc.).

10 ~. • ""

.~

R e c o m m e n d e d value

9 8 7 6 5 4 3

1 0 2

9

14

14

18

19

25

25

26

26

28

Field number Fig. 4. Example of calculation at the field level for the organic matter indicator.

31

336 Recommended value

10 9 O 8 ~3 7 "k. "

6

5 4 3

1994 m 1995

1 0 1

2 3 4 5 6

7

8 9 1011121314151617 Farm number

Fig. 5. Value of the organic matter indicator in 1994 and 1995 for the 17 farms of the network (1-13: France; 14-17: Germany).

5. Discussion The type of evaluation using AEI yields results which are easy to understand. This is an important feature for decision aid tools. Obviously, when given to farmers or to decision makers, such results should be followed by an interpretation and advice about ways to improve poor indicator values. In this way the AEI will be really used as decision aid tools. The use of a set of indicators may seem an analytical approach to describe a complex system. However, the approach is holistic in the sense that we deal with all cultivation practices within a farming system by means of a set of indicators. Furthermore, one AEI is generally related to several environmental compartments (e.g. in the case of the pesticide indicator) and often does not deal with a single environmental problem. It thus helps farmers to take into account the whole agrosystem and the adjacent ecosystem. On the other hand, because these tools are intended to be used to help farmers, there was a necessity to deal with elements reflecting the farmers' reality. Each indicator is related to a set of practices which are interrelated. The presentation of several AEI, as in Fig. 3, brings

up the question of the importance of each indicator. No comparison is possible between the results of the different indicators because one unit does not have the same meaning in each indicator. This problem is especially acute if there is a need to classify farms or farming systems by means of a set of indicators. A single aggregated indicator resulting from the aggregation of the set of AEI might be used in this case. This aggregation will involve weighting each AEI according to the user and his objectives, which involves a certain degree of subjectivity. Some authors (e.g. Hansen, 1996) do not accept this subjectivity. Another problem of a single aggregated indicator is the compensation which can occur between the values of its components. For instance, low nitrate leaching risk cannot balance a higher risk of pesticide volatilization. Such compensation has no scientific basis and it is not acceptable if global environmental impact is assessed. A single major environmental risk is sufficient to put in question the sustainability of the system. The use of multi-criteria methods (Simos, 1990) can be an alternative approach to aggregate the information supplied by a set of indicators. This approach

337

supplies a solution to the problem of compensation, and to rank or to classify actions (in our case farming systems). In this case, techniques of operational research are used. The researchers in this discipline generally accept the subjectivity inherent to decision making (Roy, 1992). In fact, there are no scientific results or rules to decide which impact is more important (e.g. between risk for soil quality or for water quality). At this stage of our work, we did not introduce any weighting of the indicators for the reasons mentioned above. Neither did Vereijken (1997) in his figure presenting the results of the set of parameters. The main objective of our AEI is to help farmers improve their management. In the perspective of a global approach as in IAFS, we consider that the value of each indicator should be improved and attain at least the reference value of 7.

6. Conclusion The approach presented in this paper is based on a set of indicators which are, first of all, decision aid tools to help farmers adapt their cultivation practices to IAFS requirements. The results of the AEI are expressed in a simple and straightforward fashion, which is an important quality for decision aid tools. Their calculation is based on data available on the farm and has been implemented in an user-friendly software tool, so that farmers or farmer-advisers can calculate the AEI. For other production systems with livestock or perennial crops, our indicators can be adapted and other specific indicators should be developed. The use of a set of indicators raises the question of the relevance of a single aggregated indicator. Several possibilities exist to aggregate the individual indicators. All these methods present a certain degree of subjectivity, which is inherent to human decision making.

Acknowledgements This work is sponsored by the EU (Interreg programme), the Land Baden-Wtirttemberg (Germany) and the Alsace Region (France) as part of the ITADA programme (C. Bockstaller), the 'Agriculture

Demain' programme of the French Research Ministry (C. Bockstaller), and by a EU Research Training Fellowship (H.M.G. van der Werf). The technical assistance of C. Zimmer (programming of the software) is gratefully acknowledged.

References Bockstaller, C. and Girardin, P., 1996. The crop sequence indicator; a tool to evaluate crop rotations in relation to the requirement of Integrated Arable Farming Systems. Aspects of Applied Biology 47, Rotations and Cropping Systems, Association of Applied Biology, Warwick, UK, pp. 405-408. Boiffin, J., K61i Zagbahi, J. and Sebillote, M., 1989. Syst~mes de culture et statut organique des sols dans le Noyonnais: un essai d'application du module de H6nin et Dupuis. In: M. Sebillote (Editor), Fertilit6 et Syst~mes de Production. lnstitut National de la Recherche Agronomique, Paris, France, pp. 234-258. Bouchon-Meunier, B., 1993. La Logique Floue. Presses Universitaires de France, Paris, France, 128 pp. E! Titi, A., Boiler, E.F. and Gendrier, J.P., 1993. Integrated production. Principles and technical guidelines. IOBC/WPRS Bull., 16: 13-38. Girardin, P. and Spiertz, J.H.J., 1993. Integrated agriculture in western Europe: Researchers' experience and limitations. J. Sustainable Agric., 3: 155-170. Gras, R., Benoit, M., Deffontaines, J.P., Duru, M., Lafarge, M., Langlet, A. and Osty, P.L., 1989. Le fait technique en agronomie. Activit~ Agricole, Concepts et M6thodes d'l~tude. Institut National de la Recherche Agronomique, L'Hamarttan, Paris, France, 184 pp. Hansen, J.W., 1996. Is agricultural sustainability a useful concept? Agr. Syst., 50:117-143. Holland, J.M., Frampton, G.K., t~ilgi, T. and Wratten, S.D., 1994. Arable acronyms analysed - a review of integrated arable farming systems research in western Europe. Ann. Appl. Biol., 125: 399-438. Levitan, L., Mervin, I. and Kovach, J., 1995. Assessing the relative environmental impacts of agricultural pesticides: the quest for a holistic method. Agric. Ecosystems Environ. 55: 153-168. Magnollay, F., 1993. R6seau PI: des progr~s mesurables. Revue Suisse Agric., 25: 361-363. Mary, B. and Gu6rif, J., 1994. lnt~r~ts et limites des modules de pr6vision de l'6volution des mati~,res organiques et de i'azote du sol. Cahiers Agric., 4: 247-257. Mitchell, G., May, A. and McDonald, A., 1995. PICABUE: a methodological framework for the development of indicators of sustainable development. Int. J. Sustain. Dev. World Ecol., 2: 104-123. Roy, B., 1992. Science de la d6cision ou science de I'aide ~ la d6cision? Rev. Int. Syst6mique, 6: 980-1472. Sharpley, A.N., 1995. Dependence of runoff phosphorus on extractable soil phosphorus. J. Environ. Qual., 24: 920-926. Simos, J., 1990. Evaluer I'Impact sur l'Environnement. Presses

338

Polytechniques et Universitaires Romandes, Lausanne, Swizerland, 261 pp. Van der Werf, H.M.G., 1996. Assessing the impact of pesticides on the environment. Agric. Ecosystems Environ., 60: 81-96. Vereijken, P., 1992. A methodic way to more sustainable farming systems. Neth. J. Agr. Sci., 3: 209-223. Vereijken, P., 1997. A methodical way of prototyping integrated and ecological arable farming systems (I/EAFS) in interaction with pilot farms. In: M.K. van Ittersum and S.C. van de Geijn (Editors), Proceedings of the 4th ESA Congress, Elsevier, Amsterdam, pp. 56-60.

Vereijken, P., Wijnands, F., Stol, W. and Visser, R., 1994. Progress Report 1. Designing Prototypes. Progress repots of the research network on integrated and ecological arable farming systems for EU and associated countries (Concerted action AIR 3 CT920755) AB-DLO, Wageningen, The Netherlands, 87 pp. Vereijken, P., Wijnands, F. and Stol, W., 1995. Progress Report 2. Designing and Testing Prototypes. Progress repots of the research network on integrated and ecological arable farming systems for EU and associated countries (Concerted action AIR 3 -CT920755) AB-DLO, Wageningen, The Netherlands, 90 pp.

© 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van lttersum and S.C. van de Geijn (Editors)

339

Model-based explorations to support development of sustainable farming systems" case studies from France and the Netherlands W.A.H. Rossing a'*, J.M. Meynard b, M.K.

van Ittersum a

aDepartment of Theoretical Production Ecology, WageningenAgricultural University, P.O. Box 430, 6700 AK Wageningen, The Netherlands bUnit~, d'Agronomie INRA-INA PG, F-78850 Thiverval-Grignon, France Accepted 13 June 1997

Abstract Sustainable land use requires development of agricultural production systems that, in addition to economic objectives, contribute to objectives in areas such as environment, health and well-being, rural scenery and nature. Since these objectives are at least partially conflicting, development of sustainable farming systems is characterized by negotiation about acceptable compromises among objectives. Four phases can be distinguished in the course of farming systems development: diagnosis, design, testing and improvement, and dissemination. During the last decade an approach coined 'prototyping' has emerged as a promising method for empirical farming systems development in Western Europe. Limitations of the approach include: (1) the limited number of systems that can be evaluated, resulting in a lack of perspective on conflicts among objectives, and (2) the expertise-based nature of rules used during systems design which unduly narrows the range of available options and obscures understanding of systems behaviour. In the paper, explorative studies based on transparent models of agronomy and management are put forward to supplement empirical prototyping and to remedy its shortcomings. To illustrate the potential of model-based explorations, two case studies are presented. The first case study deals with diagnosis and design of wheatbased rotations in the Paris Basin of France, aimed at alleviating tactical problems of poor resource-use efficiency within the constraints imposed by existing crop rotations. The second case study addresses design of sustainable bulb-based farming systems in the Netherlands with the purpose of investigating strategic options at crop rotation and farm level to resolve conflicts between economic and environmental objectives. In the discussion, methodological elements of model-based explorations and interaction with stakeholders are addressed, and opportunities for enhanced development of sustainable farming systems are identified. © 1997 Elsevier Science B.V. Keywords: Sustainable agriculture; Farming systems; Cropping systems; Prototyping; Model-based learning; Participatory research

I. Introduction I n m a n y parts of Europe, arable farmers have been very successful in increasing yields per unit area dur*Corresponding author. Tel.: +31 317 484766; fax: +31 317 484892; e-mail: [email protected]

ing the last decades. However, the production techniques that were utilized have resulted in negative side effects" emissions of pesticides and plant nutrients, (in)organic waste, high energy consumption. Public concern is reflected in a suite of national and international policy statements that call for more sustainable agricultural farming systems.

Reprinted from the European Journal of AgronOmy 7 (1997) 271-283

340

For operational purposes, sustainability can be defined as a combination of socio-economic, ecological and agro-technical objectives of agricultural production (WRR, 1995). The weight attributed to these objectives is value-driven and varies among interest groups. Because the objectives are usually at least partially conflicting, development of sustainable farming systems is characterized by negotiation about acceptable compromises among objectives by various stakeholders. Actors include farmers, agricultural industry, consumers and the public sector. Agricultural research contributes to this process by developing methodology to demonstrate consequences of alternative options. During the last decade, a promising empirical methodology for developing sustainable farming systems has been elaborated, coined 'prototyping' (Vereijken, 1994, 1997). Prototyping involves application-oriented design and testing of farming systems in collaboration with commercial farmers or at experimental farms, according to a methodical approach. Four phases can be distinguished: diagnosis, design, testing and improvement and dissemination. In the diagnostic phase the objectives of agricultural production and the value-driven weights attached to them by various interest groups are established, and problems caused by the current system design are identified. The diagnostic phase should result in a strategic alliance of stakeholders with a common motivation to design alternative ways of agricultural production. In the next phase, farmers and researchers set out to design new production systems that better meet the objectives. The result of the design phase is a number of promising theoretical prototypes of sustainable production systems. Following implementation of these theoretical prototypes, monitoring their performance in terms of objectives provides the basis for iterative prototype improvement. This phase of testing and improvement, executed on experimental or commercial farms, results in practical prototypes which have demonstrated acceptable performance in all objectives. Capitalizing upon the insights gained during the first three phases, a larger farmer audience can be addressed during the final phase, which is aimed at dissemination of the prototypes within the farming community. Prototyping suffers from two shortcomings. Only few theoretical prototypes can be tested, resulting in

a lack of information on trade-off among objectives, and systems design is based on expertise summarized in simple rules which unduly narrows the range of available options and obscures understanding of systems behaviour. Model-based explorations can remedy the limitations of empirical prototyping. Models are used because they represent devices for combining detailed information on system components and creating system designs that meet objectives of the various actors involved in farming systems development. The models are used in an explorative, as opposed to predictive, fashion. Rather than aiming at predicting which farming systems are plausible, explorations focus on designs which are possible in relation to the objectives of the actors involved. As a consequence, results of explorative studies are presented as options rather than recipes. While different techniques may be used, including continuous simulation, rule-based simulation or optimization techniques, explorative models are mechanistic and integrate components to create designs at next higher aggregation levels. The mechanistic approach enables elucidation of causes of calculated system behaviour based on insight in component behaviour and provides the opportunity to enhance understanding of systems behaviour. To optimally utilize their potential, model-based explorations should be conducted at different levels of aggregation of agricultural systems. Objectives and constraints at the various levels differ, resulting in conflicting options between levels. For instance, sustainability from a regional perspective may lead to farming systems that are economically not viable. The time horizon adopted in a study determines the type of options available, longer time horizons leading to designs with more futuristic farm strategies, alternative crops or production techniques. This paper focuses on the contribution of modelbased explorations at the level of field, crop rotation and farm to design of sustainable farming systems. Studies aimed at exploring options at regional and global levels are described by de Wit et al. (1988), Veeneklaas (1990), van de Ven (1994), Rabbinge et al. (1994) and Penning de Vries et al. (1995). In the next section, two case studies are described, with emphasis on the role of model-based explorations in different phases of farming systems development. The first case study illustrates the role of explorations dur-

341

ing diagnosis and design of wheat management systems at field and crop rotation levels, with a time horizon of 1-5 years. The second case study exemplifies model-based design at the crop rotation and farm level, with a time horizon of 10-15 years. In the discussion, methodological elements related to model-based explorations and their delivery are discussed and opportunities for enhanced support of development of sustainable fanning systems are identified.

2. Explorations at the level of field, crop rotation and farm: two case studies

2.1. Opportunities for improving wheat-based systems in the Paris Basin of France In the Paris Basin, farmers often have more than half their farm land under cereals (mainly wheat and barley), in rotation with sugar beet, potato, oilseed rape, fodder peas and sometimes vegetables for industry. High input levels of pesticides and nutrients constitute a threat to environmental sustainability of these farming systems. A series of studies were carried out at the aggregation levels of field and crop rotation, aimed at improving wheat management in the short term, i.e. with a time horizon of 1-5 years. A combination of model-based explorations and empirical research was used to diagnose current practices in wheat cultivation and design improved systems. In the explorations, regression models and rule-based simulation models were used, their nature and role differing among phases.

2.1.1. Diagnosis Causes of variation in input efficiency and yield between wheat crops grown on similar soils and as part of similar crop rotations were diagnosed. Data on actual wheat management were obtained through surveys. In each surveyed field, data were collected on crop (total biomass, yield components, crop nutritional status, crop health), soil (available macro-nutrients, structure of the arable layer) and environmental conditions (temperature, radiation and precipitation). Furthermore, information was gathered on wholefarm management, i.e. crop and soil management practices (timing of operations, labour requirement

and equipment used) for all crops in the rotation (Meynard, 1985a; Dor6 et al., 1997). In the Picardie region, the actual yield ranged from 4 to 9 t grain/ha. Variation in yield appeared to be more related to variations in ear and grain density than to variation in weight per grain (Meynard, 1985a). To identify the factors limiting each of these yield components, a yield gap analysis was carried out. Potential size of yield components were explored using a set of linked empirical regression models that related the size of a yield component in the absence of limiting or reducing factors, to critical crop variables. For instance, potential ear density was estimated by the regression model of Masle (1985) in which biomass of the shoot at the start of stem elongation and variety are used as input variables, while potential grain density was described by the regression relation of Boiffin et al. (1981) that uses shoot biomass at the start of flowering as input. Next, the yield gap, defined as the ratio of actual level and potential level of a yield component, was correlated to soil and environmental variables. Correlation and categorical analyses revealed significant association between yield gap and two factors: compacted soil structure and belated applications of nitrogen fertiliser that caused nitrogen deficiency during the first part of stem elongation (Meynard and David, 1992). The same factors decreased nitrogen fertiliser efficiency. For example, uptake of N on compacted soils was reduced by about 30-50 kg/ha compared to soils without compaction, which increased risk of leaching of nitrogen. Belated fertilizer application and soil compaction appeared correlated to aspects of whole-farm work organisation. Sowing of sugar beet was demonstrated to receive priority over simultaneously required fertilizer application in wheat. Ploughing to alleviate soil compaction, generally caused in crops preceding wheat, competed for labour with sugar beet harvesting, and was replaced by less time consuming, but also less effective, shallow tillage. The scale of these problems varied among farms depending on the relative importance of the competing crops and the labour-power and equipment available (Meynard, 1985a; Aubry, 1995). Yield reduction caused by aphids and diseases was found to be positively correlated to early sowing (before mid October), high nitrogen fertilizer input (about 200 kg/ha), high seed rates (about 300 grains/

342 m 2 for early sown crops) and the level of susceptibility of the cultivar, except when fungicides and aphicides had been applied twice (Meynard, 1991). Survey data were summarized in a multiple regression model that predicted yield loss relative to yield of a 'healthy' crop that had received two pesticide applications (Chevallier-G6rard et al., 1994). Input variables included attainable yield, i.e. crop yield in absence of pests and diseases, year, location, sowing date, varietal resistance spectrum and preceding crop. The model was used to explore the economic returns on pesticide application for different combinations of target yield, sowing dates and cultivar resistance, using weather data from 1978 to 1991. The results demonstrated that pesticide application was justified in the vast majority of years for a susceptible cultivar such as Th6s6e that was sown early with a high yield target. In contrast, positive returns on pesticide input were obtained in four out of 14 years only when sowing occurred after mid November and a more resistant variety such as Renan was selected in combination with a lower target yield.

2.1.2. Design The results of the diagnostic phase stimulated the design of low-input wheat management systems that were environmentally friendly and provided economic margins that were at least equal to those of

intensive systems. Three major constraints of prevailing systems should be overcome: soil compaction, belated fertilizer application and disease risk. Working hypothesis during design of these new systems was that by adopting a target yield below the level in conventional systems, it would pay off to sow later, to reduce sowing rates and nitrogen input, and to adopt varieties that, although lower yielding than popular varieties, were more disease resistant. As a consequence, risks of lodging and infestation by diseases and aphids were expected to be lower and costs of growth regulators and pesticides would decrease, while at the same time less labour would be required (Meynard, 1985a). A range of alternative systems was assessed by model-based explorations. In the study, the set of linked regression models that were used during the diagnostic phase was extended to account for the entire growing season. For instance, Shinosaki and Kira's model, modified by Willey and Heath (1969) and Meynard (1985b), was added to calculate aerial biomass at the start of stem elongation using plant density and sowing date as inputs. This model's output was input to calculation of ear density according to the model of Masle (1985). The balance sheet approach (R6my and H6bert, 1977) was used to calculate the effect of nitrogen fertilizer application on yield. The complete set of models is described by Meynard (1985a). Results of the study indicated that

Table 1 Comparison of characteristics of a prototypeintegrated wheat managementsystem and a conventional intensive systemfor the Picardie region in the Paris Basin of France Aspect

Integrated system

Intensive system

Target yield range (t/ha) N requirement (kg/ha) Sowing rate (grain/m2) Sowing before 25 October Sowing after 11 November Nitrogen fertilizer First application: Date Rate (kg/ha) Second application: Date Rate (kg/ha) Growth regulator Fungicides

6.5-7.5 195

8.0-9.0 240

180 250

260 490 15 February + 10 days

40

70

Beginning of stem elongation According to balance sheet method: crop N-requirement minus estimated soil supply No CCC at start of stem elongation According to damage threshold Two fixed applications: at heading and 4 weeks before

Prototype design is based on an explorative model-based study by Meynard (1985a).

343

~~

• Climatic database

• Cropping history Soil type

(20 to 30 years)

• Set of decision rules]

J

i

I

[

DECISIONSIMULATOR

I

i

Crop management (year i, i= I to30) i=l+ 1

Yield, protein content, soil mineral N at harvest, gross margin (year i)

• Frequency distribution of yield, protein content, soil mineral N at harvest, gross margin

• Risks linked to the particular set of decision rules ~l

illl

i

Fig. 1. General Day-out of D6cibi6, a software tool for interactive design of wheat management systems (after Aubry et al., 1992).

the 'integrated system' described in Table 1 would provide the largest returns. Widespread application of these design principles to the diversity of constraints of specific farmers, has been facilitated by the development of the interactive software tools 'D6cibl6' (Aubry et al., 1992; Chevallier-G6rard et al., 1994) and 'Otelo' (Attonaty et al., 1993). D6cibl6 simulates the effects of crop management on wheat yield, gross margin, protein content, and soil mineral nitrogen at harvest for specific fields characterized by cropping history, soil type and weather (Fig. 1). Crop management is described by a set of decision rules, representative for a farmer or proposed by researchers and extensionists. In the decision rules environmental and agronomic conditions are related to actions. For example, a possible rule would be: 'if the wheat crop is in development stage 30 and calculated trafficability of the soil is sufficient, then apply nitrogen dressing calculated according to the balance sheet method'. These generic decision rules are made specific for a particular crop in a particular year by a 'decision simulator'. Simulated crop management is combined with modules of crop growth and development in a 'crop simulator' which contains the agronomic relations described earlier.

These modules can easily be tested and adapted to new varieties and different areas. When run with up to 30 years of historical weather data, D6cibl6 enables exploration of different sets of decision rules, thus providing information to the farmer for selection of the set that is most desirable for his specific objectives regarding grain quality, work organization, or economic and environmental goals. While D6cibl6 is used to explore options at the crop and field level, opportunities for improvement of work organization at the crop rotation and farm level can be explored with the interactive software tool Otelo (Attonaty et al., 1993). Otelo is a rule-based system which simulates consequences of work prioritization on dates of tillage, sowing, harvest, etc. for a given farm. Otelo enables the farmer to explore different ways to reduce competition among activities and assess the possible contribution of changing machinery, manpower, or cropping plan. During exploration, the farmer specifies his decision rules in the same format as described for D6cibl6, for all crops and activities on the different fields of the farm. These decision rules are input to Otelo. The farmer then simulates the dates of the various operations using weather data from the last two or three years. The comparison between simulated and actual dates is a validation of the representation of the farmer's decision rules. Such validation determines the quality of the ensuing explorations, and increases the farmer's confidence in the model. Similar to D6cibl~, risks associated with the farmer's decision rules can be assessed by running Otelo with weather data for a period up to 30 years. Risk may be expressed as probability of exceedence of a threshold value and can be estimated from simulated frequency distributions for sowing date, lateness in fertilization, or soil compaction at sowing or harvest. Various sets of decision rules, cropping patterns or equipment can be compared to those of the farmer. Both in Otelo and in D6cibl6, the agronomist and the farmer iteratively identify the best organization pattern, integrating the characteristics of the farm. Combination of surveys and model-based explorations during diagnosis resulted in a perspective on bottlenecks in existing wheat-based systems at the level of wheat crop and crop rotation. In the design phase, models were used to assess alternative solutions with respect to objectives pertaining to eco-

344

nomic returns, environment and labour availability. Empirical evaluation of the performance of the integrated system that emerged as most promising (Table 1) will be addressed in Section 3.1.

2.2. Opportunities for improving flower bulb based farming systems in the Netherlands Current systems of flower bulb production in the Netherlands use considerable amounts of nutrients and pesticides per unit area. High prices of product and land, relatively low input prices and a defensive attitude among growers towards environmental issues are among the causes for these high input levels. Legislation is aimed at reducing negative environmental side-effects, particularly addressing pesticides and nutrients. To support design of environmentally more acceptable production systems by an association of growers and environmentalists, an explorative study was carried out (Rossing et al., 1997). In the study, fragmented agronomic information was synthesized in a database and a linear programming optimization model was used to explore technical options for flower bulb production with a time horizon of 10-15 years. The choice of time horizon was reflected in the choice of farm sizes (in terms of labour and area, both treated as exogenous variables) and in the choice of production techniques. The study focused on farms located on coarse sandy soils in the west of the Netherlands, allowing rent of land for bulb production on heavier soils further away from the farm. In accordance with the operational definition of sustainability proposed by WRR (1995), a distinction was made between value-driven objectives and factdriven agronomic information. One economic and two environmental objectives were formulated in interaction with the association of stakeholders. The economic objective was represented by maximization of farm gross margin. The environmental objectives were minimization of pesticide input expressed in kg active ingredient (a.i.) averaged over the cropped area and minimization of nitrogen surplus calculated as nitrogen not taken up by the crop and not transferred to a subsequent crop, averaged over the cropped area. Important value-driven constraints that were formulated comprised farm size, the possibility to rent additional land free of soil-borne pests and diseases, and the variety of crops that could be grown.

Agronomic information was synthesized to define management systems for four bulb crops, i.e. tulip, narcissus, hyacinth and lily, and for one break crop, i.e. winter wheat, which has positive effects on soil structure and soil health. Crop management systems were characterized by soil type and soil health, cropping frequency, crop protection regime and nutrient regime. The characteristics were chosen such that a wide array of crop production techniques could be defined that varied distinctively in terms of the objectives of flower bulb production. In addition, inter-crop management systems were defined, such as soil fumigation, inundation, and prevention of wind erosion with straw. Congruent with the time horizon of 1015 years, attention was focused on production techniques still in an experimental stage and techniques derived from other crops, rather than on current practices only. For all specified crop and inter-crop management systems inputs and outputs were formulated using empirical information, expertise and production ecological theory (Rabbinge, 1993; de Koning et al., 1995; van Ittersum and Rabbinge, 1997). Crop and inter-crop management systems were combined to rotations in a multiple goal linear pro-

x t X

60-

X

X

X

X

c.-

m 4010(

io_- 20-

x 60

°T'-"--r-. 80

100

~ 120

b 140

x 160

t

180

Nitrogen surplus (kg N ha-l)

Fig. 2. Calculated maximum farm gross margin (index, see Table 2) associated with combinations of farm-based average pesticide input (kg active ingredient/ha) and farm-based average nitrogen surplus (kg N/ha) for a farm with 15 ha sandy soils, three full time labour equivalent, and optional rent of land. Optional crops on sand: tulip, hyacinth, narcissus, lily and winter wheat; on clay: tulip and narcissus. Points of equal farm gross margin are connected (iso-lines). Each combination of pesticide input and nitrogen surplus for which maximumfarm gross margin is calculated is indicated as x. Arrows indicate development paths (see text and Tables 2 and 3). (Rossinget al., 1997;reprinted with permissionof Kluwer Academic).

345

tulip-lily-wheat to narcissus-wheat and farm gross margin becomes negative. In all steps, the rented land, free of soil-borne pests and diseases remains approximately 11 ha. On this rented area tulip is grown with a relatively moderate pesticide input of 12 kg (a.i.)/ha. In contrast, the results for the development path for nitrogen surplus reduction show that with the defined techniques, N-surplus reduction is only possible at the expense of a considerable reduction in income. A decrease in N-surplus of 30% beyond the levels anticipated for 2000 is associated with a 40% decrease in farm gross margin (Table 3). In the cropping sequence lily is replaced by narcissus, which has a much lower gross margin but higher N-efficiency. Experiences on two experimental farms and current trends in the sector support the conclusion that reducing pesticide use affects farm income less than N-surplus reduction. Remedy may be sought in development of new technologies, aiming at more precise application of nutrients in time and space, or in re-evaluation of strategic choices, such as the current use of alluvial sandy soils for growing the bulk of nutrient-inefficient flower bulb crops. The sensitivity of results to farm size, range of crops, prices and assumptions on input-out-

gramming approach (de Wit et al., 1988; Schans, 1996) to allow evaluation of objectives. By maximising farm gross margin at increasingly tighter constraints on the environmental objectives, the tradeoff between market and environment was explored. The reference situation represents a production system which just meets the (anticipated) governmental targets with respect to pesticide input and nitrogen surplus for the year 2000. Two development paths were assessed, representing gradually reduced pesticide use and N-surplus, respectively (Fig. 2). The development path for pesticide reduction (Table 2) shows that in the first step a substantial reduction in pesticide input may be achieved with relatively little loss of farm gross margin. This is achieved mainly by substituting soil fumigation by inundation and adoption of new low-dosage fungicides in tulip production. No changes in cropping sequence or area rented land occur. Further reduction in pesticide input (step 2) is most economically accomplished by abolishing the use of mineral oil for virus control in lily. The associated yield loss in lily causes a reduction in farm gross margin. Again, no changes in cropping frequency occur. The third step, zero pesticide input, causes major changes: the rotation changes from

Table 2 Exploration of flowerbulb production systems under increasingly tighter constraints on pesticide input (kg active ingredient/ha) for a farm of 15 ha sandy soils, three full time labour equivalent, and possibility to grow tulip, narcissus, lily, hyacinth and winter wheat

Objectives Farm gross margin (indexed) Average pesticide input (kg a.i./ha) Average nitrogen surplus (kg N/ha) Production techniques Fraction area per crop (%) Tulip Lily Narcissus Winter wheat Pesticide input per crop (kg a.i./ha) Tulip Lily Narcissus Winter wheat Area fumigated (ha) Area rented (ha)

Referencea

Step I b

Step 2

Step 3

100 50 140

97 30 140

77 10 140

-4 0 140

33 33

33 33

33 33

-

-

-

33

33

33

18 86

12 78

12 18

-

-

-

-

50 50 0 0

0

0

0

0

1.2 11

0 11

0 10

0 11

aThe reference farming system, equivalent to point A in Fig. 1, just meets the anticipated governmental targets regarding pesticide input and

nitrogen surplus for the year 2000. The associated farm gross margin has index value 100. Zero gross margin has index value 0. bStep 1 results in point E in Fig. 1, step 2 in point F, step 3 in point G.

346 Table 3 Exploration of flowerbulb production systemsunder increasingly tighter constraints on nitrogen surplus for a farm of 15 ha sandy soils, three full time labour equivalent, and possibility to grow tulip, narcissus, lily, hyacinth and winter wheat Referencea

Step Ib

Step 2

Step 3

100 50 140

62 50 100

38 9 90

2 8 55

33 33

33 16 16 33 11

50 50 11

50 50 12

Objectives Farm gross margin (indexed) Average pesticide input (kg a.i./ha) Average nitrogen surplus (kg N/ha)

Production techniques Fraction area per crop (%) Tulip Lily Narcissus Winter wheat Area rented (ha)

33 11

aThe reference farming systemjust meets the anticipated governmentaltargets regarding pesticide input and nitrogen surplus for the year 2000. The associated farm gross margin has index value 100. Zero gross margin has index value 0. bStep 1 results in point B in Fig. 1, step 2 in point C, step 3 in point D.

put relations is reported elsewhere (Rossing et al., 1997). The approach of separating objectives and bio-physical options was much appreciated by the association of growers and environmentalists and resulted in bridging the gap between the two parties involved. The existing polarization appeared to be caused by divergent views on objectives, rather than by disagreement on bio-technical relations. The perspective on the trade-offs among all objectives focused the discussion on preferred development pathways. While the a priori outlook of growers was especially focused on tactical decision making, the study increased awareness of the importance of strategic choices over tactical choices (Rabbinge and Zadoks, 1989). In particular, the importance of introducing a soil health restoring break crop in the rotation, such as winter wheat, and renting healthy land proved to be important strategic options for mitigating the decrease in farm gross margin associated with less pesticide input and lower nitrogen surplus. Based on the results of the study, participating farmers actively promoted research on ecology of soil-borne pests at their experimental station in response to the lack of knowledge that had become apparent during the explorations. Despite uncertainty in a number of the agronomic relations, the results were deemed sufficiently robust for testing and improvement on commercial farms. A major project was formulated and is anticipated to start in 1998. The project envisages continuous train-

ing of selected farmers and extensionists and efforts are currently focused on adapting and extending the exploratory design tools for this educational purpose.

3. Discussion

3.1. Methodological aspects of model-based explorations In the introduction of this paper, model-based explorations were put forward to supplement empirical prototyping and to remedy its shortcomings: the limited number of production systems that can be evaluated and the rules of thumb used during the design process. The case studies demonstrated the capacity of models to explore large numbers of alternative production systems and to enhance understanding of systems behaviour due to the transparency of model components. The case studies differed with respect to model types and aggregation levels. In the Dutch case study, input-output relations stored in databases that were linked to a linear programming model were used to address strategic changes in flower bulb production systems needed to resolve conflicts between economic and environmental objectives. In the French case study, a combination of regression models and rule-based simulation models were used for tactical exploration of wheat management systems aimed at adjustment of bottle-

347

necks within the constraints imposed by existing crop rotations. In the regression models agronomic information was summarized to assess potential production during diagnosis or target production levels during design. The rule-based models were used to mimic farm management decisions, both of a particular farmer and in an explorative sense. In principle, results of these tactical explorations provide inputoutput relations for strategic optimization studies. Establishing such link constitutes an important research area to be developed, as it would improve the coherence and consistency of explorative studies at different time and spatial scales. The case studies indicate that relevant answers require explorative studies at different aggregation levels. Only by combining the opportunities at the crop rotation level with those at the crop level, soil and nutrient management in wheat could be improved without sacrifices in other crops caused by labour constraints. For flower bulbs, a study at the sectoral level would be desirable to explore the implications of various options identified at the farm level, because prices are largely determined by the production volume re alised in the Netherlands. The case study in the Netherlands demonstrated the usefulness of sensitivity analysis to reveal gaps in knowledge relevant to the problem. The consequences of uncertainty in agronomic knowledge were revealed by varying single parameters or parameters in a single relation and assessing the resulting change in model output in the conventional way (cf. Janssen, 1994). However, sensitivity in linear programming models represents a special case, because, typically, uncertainty in agronomic knowledge may have little effect on the realization of objectives, but leads to very different optimal production systems (Scheele, 1992; Hijmans and van Ittersum, 1996). New mathematical techniques are needed to reveal the range of production systems that results in similar levels of satisfaction of objectives. By definition, models are simplified representations of reality, targeted at capturing the essential elements of system dynamics. To be relevant, statements based on model calculations should be accompanied by an indication of their quality. Quality assessment may address model components at the field level, such as a single regression relation or an input-output relation, and compare it to reality. Concepts for model valida-

tion or evaluation at the field level have been described by various authors (e.g. Teng, 1981; Rossing, 1991). A similar approach to quality assessment of models at the level of crop rotations or farms is less useful because the large number of uncontrollable variables impedes classical experimental design. Quality may then be interpreted as the degree to which a model-based systems design that emerged as potentially successful, performs better than an existing system. In such output-oriented assessment of model quality, causality requires attention: the model results must be better for the fight reasons. Approaches to output-oriented assessment of model quality that were developed in the French case study include (1) cropping system experiments based on decision rules, and (2) monitoring of farms that have adopted the prototype systems. In a cropping system experiment alternative wheat management systems are evaluated on-farm for their effects on a set of objectives. Each management systems consists of a specific set of decision rules that emerged as promising from the design phase and was further refined by discussion among farmers, advisers and researchers (Meynard et al., 1996). Variation in production situations among farms is taken into account by executing the experiment in a network of farms. This approach was adopted by Meynard ( 1985a, 1991) for evaluating the simulated design for integrated wheat management (Table 1). During 4 years, 28 on-farm experiments were executed in which this integrated system was compared to the conventional system. Outputs of the integrated system that were compared to the conventional system comprised mean gross margin, yield variability, and risk of nitrate leaching. The integrated system appeared better than the conventional system from both the economic and environmental point of view (Table 4). An important spin-off of farmers being responsible for execution of the experiments in their fields, was the increased credibility of the results to the farming community. In a survey of farms that had adopted integrated wheat production systems, Aubry (1995) showed that farmers modified the underlying decision rules to simplify their decision-making tasks. Farmers classified their large number of different wheat fields in groups which could be treated in similar ways and monitored crop development and diseases in only one field. Further improvement of the relevance of

348

Table 4 Performance of a prototype integrated wheat management system and a conventional intensive system in 28 on-farm experiments during 4 years, on different soil types and with different previous crops in the Picardie region in the Paris Basin of France Aspect Actual yield Mean (t/ha) Highest mean yield (cases out of 28 experiments) Gross margin Highest mean gross margin (cases out of 28 experiments) Standard deviation of yield (t/ha) N-recovery Highest mean N-recovery (cases out of 28 experiments) Fungicide treatment a Average number

Integrated system

7.5 4

Intensive system

8.0 24

21 1.10

7 1.14

22

6

1.4

2.0

Agronomic details of the systems are described in Table !. Soil tillage, sowing date and herbicide application were farmer-specific. In the systems, the same variety was used. aln additional experiments with a more disease resistant variety the average number of fungicide treatments was 0.7.

systems design may be expected when during design the set of fields on the farm sown to a given crop, i.e. a decision-making level intermediate between the crop and the crop rotation, is taken into account. 3.2. Interaction with stakeholders

As was argued in the introduction to this paper, development of sustainable farming systems implies negotiations about change and has an important social dimension. Contributions by model-based explorations to this social aspect can be assessed in terms of 'product', 'process' and 'instrument'. The contribution of explorative studies may be assessed in terms of their envisaged product, i.e. change in perceptions and/or change in actions of actors in the agricultural knowledge chain. Such impact assessment may help to improve design and delivery of explorative studies, but few retrospective studies of this sort have been carried out and reported (Sebillotte, 1996). The funding body informally evaluated the case study on flower bulb systems by asking a journalist to interview the participants. Counterintuitive results of the study were reported with respect to the importance of introducing a soil health restoring break crop in the rotation, and with respect to renting healthy land. In contrast to these strategic options, the a priori attention of the participants had been focused on improving management of pests, diseases and nutrients. The study prompted a range of

activities by the association of farmers and environmentalists. After completion of the study, the association continued discussions, involved other parties from the flower bulb industry, and formulated a proposal for testing and improvement of prototype systems of integrated flower bulb farming that was widely supported. Apart from its outcome, the process, i.e. the execution of an explorative study in itself may contribute to development of sustainable farming systems by stimulating discussions among stakeholders based on scientific information. In both case studies, participants considered communication and reflection on sustainability to be improved as a result of the explorative approach in diagnosis and design. Essential elements in this respect are the clear separation of value-driven objectives and fact-driven options, the quantitative nature of model results that enabled discussion on acceptable trade-offs among objectives, and the transparency of underlying information which improved understanding of system behaviour. Effective contribution to the process of design by stakeholders necessitates research planning in which communication between researchers and stakeholders is explicitly taken into account. For instance, during the flower bulb case study researchers interacted with a delegation of the association once every 6 weeks to discuss general progress, and the association organized two workshops to formulate the value-driven objectives that were used to evaluate options during

349

the study. These frequent social interactions may well have paved the way for the projects 'product' described in the previous paragraph. The instruments, i.e. the models that were used during the explorations served as computer-aided learning tools for stakeholders (Leeuwis, 1993; Cerf et al., 1994; Papy, 1996). Characteristically, models were used to answer 'what-if' questions, with researchers acting as intermediates between model and stakeholders (in the flower bulb case) or process facilitators (in the wheat case). Fast response to 'whatif' questions is necessary to maintain process momentum, requiring further improvement of the linear programming model concerning user-friendliness, flexibility and methodology for sensitivity analysis. While in the flower bulb case the model itself was meant t o b e used by researchers only, the tools Otelo and D6cibl6 were designed for interactive application by farmers and extensionists to a specific farm. Two aspects were found to influence the success of the application (Mousset et al., 1997). Firstly, the users should be able to gain confidence in the tool. Farmers were usually able to recall their agronomic decisions of the preceding three years, which enabled assessment of model quality by comparison of simulation results with actual data. Secondly, the amount of time spent in creating input for the model should be commensurate to the expected value of information of model output. Especially for Otelo, the amount of farm-specific input data is considerable, and the tool is used only for complex situations, requiring innovative solutions.

3.3. Model-based explorations and empirical prototyping The case studies in this paper demonstrated how model-based explorations supplement empirical prototyping during the first two phases of sustainable farming systems development: diagnosis and design. The diagnostic surveys of the prototyping approach were complemented by modeling studies exploring production potential. Their combination enabled identification of constraints in current practices and assessment of opportunities for improvement. Opportunities were elaborated during the model-based design phase to reveal trade-offs among objectives. Although not substantiated by the case studies, sup-

plementation of prototyping by model-based explorations may also be expected during the last two phases of development of sustainable farming systems" testing and improvement, and dissemination. Promising options emerging from the design phase are put to the test empirically. During testing and improvement of these prototypes, explorations can reveal yield gaps or trends in slow processes such as soil organic matter turn-over. At the start of dissemination, empirical prototyping has resulted in prototype systems that have proven their value in practice, while essential elements of production systems are synthesized in models at different levels of aggregation to facilitate extrapolation to new conditions. During all phases model-based explorations may indicate gaps in knowledge of researchers, extensionists and farmers, thus contributing to learning about the system by all actors. The pivotal role of learning appears prominently in farmer-oriented projects that have been successful in stimulating more judicious use of resources (Zadoks, 1989; Kenmore, 1991; Sebillotte, 1996). Two important characteristics of learning are (1) cyclic iteration of experimentation, action, observation and reflection, and (2) repeated switching between aggregation levels, time periods, knowledge domains and farm types. Model-based explorations in interplay with prototyping have the potential of contributing to such non-formal education. To realize this potential requires a research approach in which equal attention is devoted to creating and synthesizing relevant agronomic knowledge and to creating settings in which learning can take place (Chatelin et al., 1994; Okali et al., 1994; R61ing, 1996; Somers, 1997).

References

Attonaty, J.M., Chatelin, M.H.,Poussin, J.C. and Soler, L.G., 1993. Advice and decision support systems in agriculture: new issues. Farm Level Information Systems, Woudschoten, Zeist, The Netherlands. Aubry, C., 1995. Gestion de la solel d'une culture dans l'exploitation agricole. Cas du BW d'Hiver en Grande Culture dans la R6gion Picarde. Doctoral Thesis, Institut National Agronomique Paris-Grignon, 277 pp. Aubry, C., Chatelin, M.H., Poussin, J.C., Attonaty,J.M., Meynard, J.M., G6rard, C. and Robert, D., 1992. D6cibl6: a decision support systemfor wheat management.In: Book of Abstracts,4~.me Congr~s d'lnformatique agricole, Versailles, SAF, Paris.

350

Boiffin, J., Caneill, J., Meynard, J.M. and Sebillotte, M., 1981. Elaboration du rendement et fertilisation azot6e du bl6 d'hiver en Champagne Crayeuse. I. Protocole et mdthode d'6tude d'un probl6me technique r6gional. Agronomie, I: 549-558. Cerf, M., Papy, F., Aubry, C. and Meynard, J.M., 1994. Agronomic theory and decision tools. In: J. Brossier, L. de Bonreval and E. Landais (Editors), Systems Studies in Agriculture and Rural Development. INRA, Paris, pp. 343-356. Chatelin, M.H., Mousset, J. and Papy, F., 1994. In: B.H. Jacobsen, D.E. Pedersen, J. Christensen and S. Rasmunsen (Editors), Farmers' Decision-Making - A Description Approach. Proc. from the 38th EAAE seminar, Institute of Agricultural Economics and the Royal Veterinary and Agricultural University, Copenhagen, pp. 369-381. Chevallier-G6rard, C., Denis J.B. and Meynard J.M., 1994. Perte de rendement due aux maladies cryptogamiques sur b16 tendre d'hiver. Construction et validation d'un mod61e de l'effet du syst6me de culture. Agronomie, 14: 305-318. de Koning, G.H.J., van Keulen, H., Rabbinge, R. and Janssen, H., 1995. Determination of input and output coefficients of cropping systems in the European community. Agric. Syst., 48: 485502. de Wit, C.T., van Keulen, H., Seligman, N.G. and Spharim, I., 1988. Application of interactive multiple goal programming techniques for analysis and planning of regional agricultural development. Agric. Syst., 26:211-230. Dor6, T., Sebillotte, M. and Meynard, J.M., 1997. A diagnostic method for assessing regional variations in crop yield. Agric. Syst., 54: 169-188. Hijmans, R.J. and van lttersum, M.K., 1996. Aggregation of spatial units in linear programming models to explore land use options. Neth. J. Agric. Sci., 44: 145-162. Janssen, P.H.M., 1994. Assessing sensitivities and uncertainties in models: a critical evaluation. In: J. Grasman and G. van Straten (Editors), Predictability and Non-Linear Modelling in Natural Sciences and Economics. Kluwer Academic, Dordrecht, pp. 344-361. Kenmore, P.E., 1991. How rice farmers clean up the environment, conserve biodiversity, raise more food, make higher profits Indonesia's IPM - a model for Asia. FAO Inter-Country Programme for Integrated Pest Control in Rice in South and Southeast Asia, Manila, Philippines, 56 pp. Leeuwis, C., 1993. Of computer, myths and modelling. The social construction of diversity, knowledge, information and communication technologies in Dutch horticulture and agricultural extension. Wageningen Studies in Sociology, Pudoc, Wageningen, Vol. 36, 468 pp. Masle, J., 1985. Competition among tillers in winter wheat: consequences for growth and development of the crop. In: W. Day and R.K. Atkin (Editors), Wheat Growth and Modelling. Plenum Press, New York, pp. 33-54. Meynard, J.M., 1985a. Construction d'itin6raires techniques pour la conduite du b16 d'hiver. Doctoral Thesis, Institut National Agronomique Paris-Grignon. Paris, 258 pp. Meynard, J.M., 1985b. Les besions en azote du b16 d'hiver jusqu'au d6but de la montaison. Agronomie, 5: 579-589. Meynard, J.M., 1991. Pesticides et itin6raires techniques. In: P.

Bye, C. Descoins and A. Deshayes (Editors), Phytosanitaires, Protection des Plantes, Biopesticides. INRA, Paris, pp. 85100. Meynard, J.M. and David, G., 1992. Diagnostic sur l'61aboration du rendement des cultures. Cahiers Agric., 1: 9-19. Meynard, J.M., Reau, R., Robert, D. and Saulas, P., 1996. Evaluation exp6rimentale des itin6raires techniques. In: G. Urbano et al. (Editors), Experimenter sur les Conduites de Culture. DERF! ACTA, Paris, pp. 63-72. Mousset, J., Aslah6, C., Billa, Ph., Boiffin, J., Chatelin, M.H., Chopplet, M., Franqois, M., Gandon, H., Groell, F., His, M., Hopquin, J.P., Klein, D., Papy, F., Qui6vreux, D. and Soler, L.G., 1997. Le conseil Agro-Equipement en Picardie: MECAGRO. Colloque 'Aide h la d6cision et choix de strat6gies dans les entreprises agricoles', Laon (France), 10-11/12/1996, INRA, pp. 195-206. Okali, C., Sumberg, J. and Farrington, J., 1994. Farmer Participatory Research. Rhethoric and Reality. Intermediate Technology Publications, London, 159 pp. Papy, F., 1996. Modelling the farm as a business to help farmers make their decisions: a summary review. In: J.P. Colin, E. Crawford and C. Fillonneau (Editors), Research Methodology for Agricultural Systems Analysis. In press. Penning de Vries, F.W.T., van Keulen, H. and Rabbinge, R., 1995. Natural resources and limits to food production in 2040. In: J. Bouma, A. Kuyvenhoven, B.A.M. Bouman and J.C. Luyten (Editors), Eco-Regional Approaches for Sustainable Land Use and Food Production. Kluwer Academic, Dordrecht, The Netherlands, pp. 65-87. Rabbinge, R., 1993. The ecological background of food production. In: D.J. Chadwick and J. Marsh (Editors), Crop Protection and Sustainable Agriculture. Ciba Foundation Symposium 177, Wiley, Chichester, pp. 2-29. Rabbinge, R., van Diepen, C.A., Dijsselbloem, J., de Koning, G.J.H., van Latesteijn, H.C., Woltjer, E. and van Zijl, J., 1994. 'Ground for choices': A scenario study on perspectives for rural areas in the European community. In: L.O. Fresco, L. Stroosnijder, J. Bouma and H. van Keulen (Editors), The Future of the Land: Mobilising and Integrating Knowledge for Land Use Options. Wiley, New York, pp. 95-121. Rabbinge, R. and Zadoks, J.C., 1989. Disease and pest control. In: J.C. Zadoks (Editor), Development of Farming Systems. Pudoc, Wageningen, pp. 32-39. R6my, J.C. and H~bert, J., 1977. Le devenir des engrais azot~s dans le sol. Comptes Rendus de l'Academie d'Agriculture de France, 63: 700-710. R61ing, N.G., 1996. Towards an interactive agricultural science. Eur. J. Agric. Educ. Ext., 2: 35-48. Rossing W.A.H., 1991. Simulation of damage in winter wheat caused by the grain aphid Sitobion avenae. 3. Calculation of damage at various attainable yield levels. Neth. J. Plant Pathol., 97: 87-103. Rossing, W.A.H., Jansma, J.E., de Ruijter, F.J. and Schans, J., 1997. Operationalizing sustainability: exploring options for environmentally friendly flower bulb production systems. Eur. J. Plant Pathol., 103: 217-234. Schans, J., 1996. MGOPT_CROP, a Multiple Goal Linear Pro-

351 gramming Model for Optimisation of Crop Rotations, version 1.0. AB-DLO, Wageningen, Report 69, 48 pp. Scheele, D., 1992. Formulation and characteristics of GOAL. Technical Working Document (W64), Netherlands Scientific Council for Government Policy, The Hague, The Netherlands, 64 pp. Sebillotte, M., 1996. Les Mondes de l'Agriculture: Une Recherche pour Demain. INRA, Pads, 258 pp. Somers, B.M., 1997. Learning for sustainable agriculture. In: M.K. van Ittersum and S.C. van de Geijn (Editors), Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use. Elsevier Science, Amsterdam, The Netherlands, pp. 353-359. Teng, P.S., 1981. Validation of computer models of plant disease epidemics: a review of philosophy and methodology. Z. Pflanzenkr. Pflanzenschutz, 88: 49-63. van de Ven, G.W.J., 1994. A mathematical approach for comparison of environmental and economic goals in dairy farming at the regional scale. In: L.'t Mannetje and J. Frame (Editors), Grassland and Society. Proceedings of the 15th General Meeting of the European Grassland Federation (EGF), Wageningen, 6-10 Juni 1994, Wageningen Press, Wageningen, pp. 453457. van Ittersum, M.K. and Rabbinge, R., 1997. Concepts in production

ecology for analysis and quantification of agricultural input-output combinations. Field Crops Res., 52: 197-208. Veeneklaas, F.R., 1990. Dovetailing technical and economic analysis. Doctoral Thesis, Erasmus University Rotterdam, The Netherlands, 159 pp. Vereijken, P., 1994. Designing prototypes. Progress report 1. Progress reports of research network on integrated and ecological arable farming systems for EU and associated countries. ABDLO, Wageningen, 87 pp. Vereijken, P., 1997. A methodical way of prototyping integrated and ecological arable farming systems (I/EAFS) in interaction with pilot farms. Eur. J. Agr. 7. 235-250. Willey, R.W. and Heath, S.B., 1969. The quantitative relationships between plant population and crop yield. Adv. Agron., 21: 281331. WRR, 1995. Sustained risks, a lasting phenomenon. Reports to the Government No. 44, Netherlands Scientific Council for Government Policy, The Hague, The Netherlands, 205 pp. Zadoks, J.C., 1989. EPIPRE, a computer-based decision support system for pest and disease control in wheat: Its development and implementation in Europe. In: K.J. Leonard and W.E. Fry (Editors), Plant Disease Epidemiology, Volume II. Macmillan, New York, pp. 3-29.

© 1997 Elsevier Science B. V. All rights reserved Perspectives for Agronomy - Adopting Ecological Principles and Managing Resource Use M.K. van Ittersum and S.C. van de Geijn (Editors)

353

Learning for sustainable agriculture B.M. Somers * Hoofdweg 3, 3233 LH Oostvoorne, The Netherlands Abstract

Policy makers are searching for ways of accelerating the rate of adoption of sustainable agriculture. However, sustainable agriculture is a complex innovation and it is still unclear what is the best way to stimulate farmers to make the change. Compared to other types of innovations, the change to sustainable agriculture imposes various risks on farmers. These risks, not only of an economic and technical nature, also have social and political aspects. In this article it is argued that the characteristics of sustainable farming systems, and the risks involved, imply a gradual learning process instead of a linear adoption process. Through adaptive learning, farmers can lower their levels of perceived risks. The possibility to try out modules of new farming systems and to combine general, scientific knowledge with specific, tacit knowledge form positive incentives for the introduction of sustainable farming systems. There is also a need for a more interactive development of such farming systems. Even more than is usual, researchers should be willing to learn from practical knowledge and should develop a feeling for the policy measures and risks under which farmers are working. Keywords: Adapative learning; Development of sustainable farming systems; Risk perception

1. Introduction

Designing sustainable farming systems is one thing, farmers practising sustainable agriculture on a large scale is another. In a scientific community where researchers have developed specialised fields of knowledge and skills it would be normal to leave the development of new methods to researchers and the introduction of it among farmers to extensionists. This route is a linear process in which sequential stages, from design to diffusion, must be completed. However, Rossing et al. (1995) point to the importance of interaction between several stages in the process of development and introduction of sustainable farming systems. They make a distinction between the following stages: a) designing (prototype); b) testing (including improvements); c) implementation on small scale; and d) implementation on large * Tel: +31 0181 484634; Fax: +31 0181 484071. E-mail: [email protected]

scale. According to Rossing et al. the large-scale implementation is furthered when stages b) and c) are closely intertwined. During these stages a close cooperation between research, extension and a group of well-motivated farmers is important. An efficient introduction of new farming systems on a large scale can take place when region-specific, practically tested knowledge is available in the agricultural community, and when this community is motivated and familiar with (elements of) the new prototype. Therefore, co-operation between different participants is a necessary condition for the diffusion of knowledge about sustainable farming systems. The innovation project "Integrated Arable Farming" (Anonymous, 1994) is a good example of the interaction mentioned above. Also other initiatives in the Netherlands (Somers and R61ing, 1993; Somers, 1995; Van Weperen, 1994) and outside the Netherlands (R61ing and Van de Fliert, 1994; Pretty, 1994) benefit from a close co-operation between farmers and organizations for research and extension. Yet,

354

despite the co-operation between research, extension and a group of well-motivated farmers, the large scale implementation of sustainable arable farming is hampered. When the actual performances are measured in terms of governmental norms about the use of chemicals, it is clear that much still has to be done. In this contribution I will elaborate on some factors that influence farmers' acceptance of sustainable methods and technologies. Usually knowledge, or a lack of knowledge, is mentioned as an important factor in the diffusion of innovations. Knowledge apparently reduces farmers' feelings of risks that are connected with new concepts. Therefore, I will start with recounting some characteristics of innovations that influence their acceptance by farmers. These characteristics refer to farmers' perception of risks. Subsequently I describe the possibilities of adaptive learning and the types of knowledge that are relevant in the case of sustainable farming systems. These specific characteristics of knowledge require changing roles of the actors involved in the development and introduction of alternative farming systems.

2. Types of risks involved with the shift to sustainable farming systems Sustainable farming systems can be conceived of as a set of novel technologies and farming methods that together form a complex innovation. What interests extensionists and policy makers is how to make this innovation acceptable to farmers. In extension science there is a vast amount of literature about the willingness of farmers to adopt new methods and technologies. Rogers (1983) made a compilation of the previous known research into this topic and created a checklist of factors that influence the process of diffusion of innovations. This checklist formed the basis for the studies of Somers and R6ling (1993) and of a workgroup of the Dutch Ministry of Agriculture (Somers, 1996a). In both studies, Rogers' checklist proved to be valuable for understanding the types of risks farmers encounter when they are confronted with complex innovations. Moreover, we became aware of the fact that the shift to sustainable agriculture meant that new requirements are imposed on farmers' pro-

fessional goals and skills. The complexity of the innovation and the diverging, and in short term conflicting goals, are both characteristic features of sustainable agriculture. Farmers are required to find a balance between economic and ecological goals instead of focusing on productivity levels. Besides, the public increasingly appeals to farmers where the care for landscape, tourism and nature is concerned. The implementation of sustainable farming systems is not just "another" innovation to be accepted, but reflects new professional goals in a context of new functions of the agricultural sector. In the studies of Somers and R61ing (1993) and the working group of the Ministry of Agriculture (Somers, 1996a), the following characteristics of innovations were found to influence their acceptance by farmers: The degree of certainty for the farmer that the innovation is advantageous for him: his trust that the innovation has a positive effect on his goals; The technical complexity of the innovation which can have repercussions on the economic performance of the farm; The degree to which the innovation is compatible with existing norms and values: the degree to which the farmer is confronted with social complexity; The possibility to try out the innovation: to try out a module of the total package, to try out the innovation on a part of the farm or to make a choice in the level of input; The effects of the innovation on current farm management: the degree to which the innovation alters the direction of farm management. Most of these factors have to do with the feeling of risk that hampers the acceptance of the innovation by farmers. Very often, risks perceived by farmers are connected to technical risks and negative economic consequences. Yet, we must be aware that the introduction of some innovations is also surrounded by social and political risks. This is well illustrated in the "Innovation Project Integrated Arable Farming" in the Netherlands. Thirty-four of the thirty-eight arable farmers who were involved in the stage of testing and improving the prototypes of sustainable

355

farming systems were questioned in 1994 about their experiences (Van Weperen, 1995). These thirty-four farmers mentioned the following three risk-items that caused most tension during their change to integrated agriculture" economic risks; risks connected to governmental policy; and risks in the social relationship with colleagues. The experience of risks has much to do with the fear of making wrong technical decisions which may adversely affect the financial situation. Governmental pressure of laws and regulations also cause tension, argued several of the interviewed farmers. Too strong pressure or sudden change of governmental regulations could cause a diminishing willingness among farmers to co-operate with the change towards a more sustainable agriculture. There comes a time when farmers think governmental regulations not attainable, particularly when the economic prospects are gloomy because of low prices for arable products. The interviewed farmers also mentioned the fear that the government would use the achievements of the project to oblige all Dutch arable farmers to attain extreme norms for chemical and nutrient input. Other studies also point to this sociopolitical tension that hampers the shift to sustainable agriculture (Van Weperen, 1994; Buurma, 1996). Moreover, the farmers who were farming in an integrated or organic way experienced the distrust of neighbours and colleagues, distrust that was aroused for instance by their higher tolerance for weeds and their diverging strategies for combating pests and diseases. In fact, the integrated and organic farmers were challenging the norms about "good-farming practice". Consequently, as will be shown later in this paper, new norms have to be developed. This process can take place in a group situation, for instance study groups. Thus, there are many different factors that influence the process of diffusion among farmers. Some of these factors point to the economic prospects and technical risks inherent in the innovation. Extensionists and policy makers are searching for ways of accelerating the rate of adoption of sustainable agriculture. The question is how this can be done, since farmers perceive risks in so many areas. There is the economic risk of setbacks in yields and inappropriate marketing strategies. There are unsolved problems in pest and disease management. About many aspects

there is still a lack of knowledge. Moreover, the actual and perceived risks are not confined to economic and technological aspects" there are also social and political risks. In the political arena the seriousness and sources of the environmental problems are constantly contested. The same holds true for political measures, norms and penalties to fight the problems. Forerunners in sustainable agriculture risk the distrust of colleagues who fear that their evidence of newly achieved low input agriculture will be used to create new political norms. The social and political risks that are inherent in a change to sustainable agriculture form strong impediments to its introduction.

3. Gaining knowledge Where innovations are perceived by farmers as risky, gaining knowledge can lower the level of perceived risks. The economist Bayes pointed to the importance of learning in the process of adopting an innovation (Leathers and Smale, 1991; Lindner and Gibbs, 1990). The so-called Bayesian learning model is an adaptive learning model. Crucial in it is the notion that the perceptions of farmers will change when they gain experience with a part of the innovation. This means that farmers must have the possibility to try out one or more modules of the system, to try out the new methods on a part of their farm, or have a choice concerning the level of input. By trying out, the farmer gains extra information with which he can adapt his original perceptions about the risks of applying the innovation. He also discovers whether the information given by researchers or extensionists is relevant for his own farm situation. The notion of reducing perceived risks through adaptive learning coincides with the ideas of Kolb (1984) regarding "learning by doing". His central idea is the interaction between cognitive processes and action. Kolb's theory is incorporated into theories about "learning organizations" that have become very popular in certain circles of management science. According to these management theories, organizations can improve their utilization of the knowledge that is available in the organization at all levels. Scientists in the field of extension can

356

also learn from these theories and support bottomup interactive learning processes as being more effective than top-down linear adoption processes. By shifting the perspective from a (top-down) adoption process to a (bottom-up) learning process, the roles of the participants also change. Applying this notion to the agricultural knowledge network, we will find that roles, attitudes and skills of the people involved change when learning processes and supporting learning processes become the foundation instead of innovations being imposed on farmers "from above".

4. Types of knowledge 4.1. General and tacit knowledge Through gaining knowledge, feelings of risks can be diminished, yet what types of knowledge can we think of? We can make a distinction between two types of knowledge: general knowledge and tacit knowledge. General knowledge is in principle accessible, though not always gratuitous for everybody (Somers, 1996b). General knowledge can be developed in scientific experiments, apart from the specific situations in which the knowledge will be applied. It can be incorporated in standard advice, for instance about the levels of fertilization or time schedules for spraying fungicides. General knowledge consists of facts and figures and it is transferable, orally or through written material. Tacit knowledge, on the contrary, can only be developed in specific enterprises or situations. Tacit knowledge has much to do with "knowing how" and very often is hard to verbalize. It is not readily transferable because it is formed by experience through the years. Tacit knowledge is stored in the heads of people, therefore it is often referred to as "human capital". This type of knowledge is only transferable by learning. Researchers and extensionists can teach the farmer much about indicators to determine critical levels and the nutritional needs of plants. These indicators and critical levels are usually developed on research stations and they are a form of general knowledge that the farmers can rely on. Yet, because of the varying circumstances on farms and, consequently the qualities of the specific knowledge needed, farm-

ers cannot depend solely on knowledge that is developed elsewhere. They not only need knowledge about indicators, but also the experience to observe and interprete these indicators in specific farming circumstances. Many of these interpretative skills are not transferable to farmers: much of it depends on tacit knowledge of the farmer about the precise qualities of his fields and his experiences with the growth of his crops in the past. Much of the knowledge that is required for sustainable agriculture is thus not only very specific, but also internal to the farm. The shift to sustainable agriculture implies that the farmer tries to replace chemical input by knowledge as much as possible. One field of knowledge consists of combating pests, herbs and insects. The farmer must have detailed knowledge and skills in order to observe and judge the threat of pests and diseases. He must know how to take adequate measures in advance in order to prevent the outbreak of pests and diseases. He needs figures in order to balance the pros and contras of different chemicals. In order to diminish his use of pesticides, herbicides and insecticides, the farmer will not follow general spraying schemes, but determine himself the time of spraying, based on the combination of general knowledge about critical levels and his past experiences with combating pests/herbs/insects in this specific crop on his own farm. In short, the farmer combines general and tacit knowledge. Another area of knowledge is the "fine-tuning" of minerals to match the needs of plants. In order to diminish mineral surpluses, the farmer will not follow standard advice about fertilization. Instead, he tries to adapt the administering of nutrients to the plants' needs by combining scientifically developed criteria and his own, farm-internal knowledge. 4.2. Partial and holistic knowledge Applying more sustainable methods has consequences for the whole farm management, both for the short and the long term. Organic farmers are aquainted with the fact that the decisions they take during raising a crop influences the successes and failures of subsequent crops. In an agriculture that depends largely on chemical input this relation is less decisive, because farmers can intervene during the

357

growing season. In integrated arable farming, where the aim is to become less dependent on chemicals, the farmer will prevent or postpone chemical interventions. This requires a special type of knowledge. Kropff (1996) elaborates on this different type of knowledge, for the case of weed control. Instead of short-term, partial knowledge, he focuses on knowledge about the functioning of crop-weed-systems, whereby strategic or long-term aspects must be given more emphasis than usual. In fact, Kropff is discussing holistic knowledge, about "managing coo-systems" instead of short-term combating weeds. However, much of the knowledge that is developed on research stations provides solutions for specific problems. Yet, the farmer who changes towards sustainable agriculture must have the skills to interpret this partial knowledge and estimate their long-term effects. Again, we see that a combination of types of knowledge is needed.

4.3. Farmers' experiments Above we discussed the different types of knowledge that are needed for a succesful turnover to sustainable arable farming. The availability of scientific knowledge is not sufficient in itself to induce a diminishing dependance on chemicals. The feeling of risk is not diminished by scientific knowledge alone. Farmers must develop the skills to connect the scientific knowledge with their own tacit knowledge about the specific characteristics of their land. In several projects in the Netherlands ~ we observe that farmers are experimenting with modules of integrated farming systems, such as trial methods for identifying the nutritional needs of plants and the availability of nutrients in the soil (Somers and R6l The four projects that were evaluated by Somers and R61ing (1993) were: a) "Integrated Arable Farming", a project organized by the Ministry of Agriculture, the Landbouwschap (the apex organization of the three farmers' organizations), formal research and extension institutions; b) "Agriculture and Environment", a project of the Province of Brabant, the Landbouwschap, the local farmers' organization and the local extension institute; c) "Farmers' Bread", a farmers' initiative in the Province of Zeeland, the local farmers' organization, environmental and consumer organizations, millers and bakers; d) "Organic Farming", counting 440 farmers who are organized in an ecological and a biological-dynamic organization and supported by a separate knowledge infrastructure.

ling, 1993). Experimenting with these methods creates confidence in the possibility to farm in a more sustainable way. It moreover creates the willingness to try out other parts of the system. Another observation is that factors other than just knowledge determine the application of sustainable technologies and methods, such as the availability of labour, the desires of the farmer and social and political circumstances (ibid, 1993). It is not only knowledge that counts, but also the situations in which the learning process takes place. This brings us to the changing roles of researchers when they support farmers' learning processes.

5. Towards an interactive development of sustainable farming systems

The introduction of a complex innovation such as sustainable farming systems requires that we have attention for the type of knowledge that is needed, the form in which the knowledge is delivered and the structure in which the knowledge development takes place. The furthering of sustainable agriculture seems to benefit from an intensive interaction between scientific knowledge that has been developed on experimental research stations and experimental knowledge that is developed on farms (Somers and R61ing, 1993). The fact that sustainable farming methods are perceived as risky, requires that standards which are developed elsewhere are tried out locally. On-farm research should get a more important place than it has at the present time. Moreover, the study of Somers and R61ing showed that farmers do not always ask for ready solutions, but also for suggestions and ideas that they can test in their specific farm situation. Groupwork is important, especially where it concerns careful thought about registrations with new methods and the determination of socially acceptable behaviour. New norms about "good"-farming practice can spring from a group process (Van Weperen, 1994). We can speak of "social" learning. Increasingly, social learning takes place in so-called "environmental co-operatives". Environmental co-operatives are local groups of farmers who search for ways to realize environmental goals that are specific for their own locality and for their type of farming. Often, they hold on-farm

358 experiments in order to gain knowledge about environment, nature and landscape. Their aims vary from achieving measurable values of nature to minimizing the input of fungicides. One of these groups is the working group "Soil Based Horticulture Under Glass", on which I will elaborate here. Soil-based horticulture under glass is especially threatened by policy measures such as the requirement to recirculate the drainage water (Somers, 1995). Technical solutions for this problem have not been found until now. Furthermore, over the years the agricultural research for soil-based horticulture has been minimized in favour of horticulture on artificial substratum. The underlying argument for this is the expectation that it is easier to control the input of nutrients and the dose of water in horticulture on artificial substratum. In other words: horticulture on artificial substratum was expected to be less environmentally harmful than soil-based horticulture under glass. Because of this choice, soil-based growers feel that they have lost room for manoeuvring. They think it important to quickly search for solutions and help develop sustainable methods and technologies under "practical circumstances". By means of experiments they started, together with researchers, to search for environmental parameters that are specific for their situation. They think it very urgent to gain knowledge on practical methods and technologies because strict policy measures are already enacted. In some cases they have requested delay of measures until their experiments will yield results. The experience of the working group "Soil Based Horticulture Under Glass" shows that researchers and growers both participate in a learning process. Growers learned to formulate their problems and wishes for research experiments in a very detailed manner. They also established a good relationship with the experimental research station in order to influence the research programmes. For the participating researchers, supporting learning processes implied that they had open minds for forms of knowledge other than just scientific. In general, an important researchers' skill is that he/she can learn systematically from the experiences of farmers and growers. Part of the research actually takes place on the firms of the horticulturists. Researchers also must develop a feeling for the policy restrictions and

risks with which agricultural entrepreneurs must deal. One of the conclusions of the above mentioned working group is that the close co-operation between growers and researchers altered their relationship and roles. The working group "Soil Based Horticulture Under Glass" gives a good example of an interactive development of sustainable farming methods. Yet, in general there are many hindrances for an approach like this. Many impediments to a better support of learning processes can be found on the institutional level (Somers, 1996a). For instance, the way the research institutes are financed cause a lack of flexibility in the research programmes. Also the resistance to cultural changes is found to be an impediment. The ability to support farmers in learning is not only a question of skills and training, but also of self entrepreneurship. An active learning attitude and the willingness to cope with uncertainties form a part of the profile of the modern farmer. Teachers, researchers and extensionists also must change their attitude in order to adequately support farmers' learning processes. However, in general this change meets much resistance (ibid, 1996a).

6. Conclusions

Both the projects that were studied by Somers and Rrling (1993) and the working group "Soil-Based Horticulture Under Glass" show that farmers are willing to contribute to a more sustainable agriculture when the necessary conditions are created that facilitate learning processes. Also other experiments, such as the growing amount of environmental cooperatives are illustrative. Some conclusions we can draw from these experiences are that: a) The introduction of sustainable farming systems is encouraged when the systems bear the possibility of learning by doing. The introduction of a total concept, a system, seems unattainable for a large group of farmers. Following the Bayesian theory about adaptive learning, the introduction of sustainable farming systems will benefit from a modular composition. A modular composition not only lowers the perceived risks, but also is attainable in a context of social and political tensions, b) Due to the social and political impediments, the introduction of sustainable farming

359 systems will benefit from "social" learning: groups of farmers setting their goals and finding ways to realize these together, c) A greater interaction between researchers, extensionists and farmers is needed for taking into account valuable practical experiences of farmers. This requires a more systematic apprehension by researchers of farmers' experiences.

References Anonymous, 1994. Telen met perspectief: Teeltstrategiei~n gericht op een duurzame akkerbouw. IKC/AGV/DLV, Lelystad. Buurma, J.S., 1996. Oorzaken van verschillen in middelenverbruik tussen bedrijven: vuurbestrijding in tulpen. Landbouw-Economisch lnstituut (LEI-DLO), Den Haag. Kolb, D.A., 1984. Experiential Learning. Prentice Hall Inc., Englewood Cliffs, New Jersey. Kropff, M.J., 1996. Strategisch balanceren: onkruidkunde als toegepaste plantenecologie. Inaugurele rede. Landbouwuniversiteit, Wageningen. Leathers, H.D. and Smale, M., 1991. A Baysian Approach to Explaining Sequential Adoption of Components of a Technological Package. Am. J. Agric. Econ., 73(3), 734-742. Lindner, R. and Gibbs, M., 1990. A test of Bayesian Learning from Farmer Trials of New Wheat Varieties. Aust. J. Agric. Econ., 34(1), 21-38.

Pretty, J., 1994. Alternative Systems of Inquiry for Sustainable Agriculture. IDS Bull., Vol.25(2), 37-49. Rogers, E.M., 1983. Diffusion of Innovations (third edition). The Free Press, New York. R61ing, N. and Van de Fliert, E., 1994. Transforming extension for sustainable agriculture: the case of Integrated Pest Management in rice in Indonesia. Agric. and Human Values, Vol 11 (2-3), 96-108. Rossing, W.A.H., Wijnands, F.G. and Krikke, A.T., 1995. Voortgaande vernieuwing in de landbouw: het samenspel van prototypering en toekomstverkenning. In Studiedag KLV, AB-DLO en PELUW, Wageningen 21 november 1995. Eds. A J Haverkort en P A van der Werff. Pp. 115-135. AB-DLO thema's, Wageningen. Somers, B.M. and R61ing, N.G., 1993. Kennisontwikkeling voor duurzame landbouw. NRLO (National Council for Agricultural Research), Den Haag. Somers, B.M. (ed.) 1995. Plan van Aanpak Werkgroep Telen in de grond. NTS (Dutch Federation of Horticulture Study Groups), Honselersdijk. Somers, B.M., 1996a. Zoeken en leren als invalshoek voor kennisprocessen. Spil 137-138/139-140, 34-38. Somers, B.M. (ed.) 1996b. Kennis op Bedrijfsniveau. LandbouwEconomisch Instituut (LEI-DLO), Den Haag. Van Weperen, W., 1994. Balancing the Minerals, Moving Boundaries: Mineral Balance Extension in Dutch Dairy Farming. Agricultural University, Wageningen. Van Weperen, W. (ed.) 1995. Het veranderingsproces: ervaringen van akkerbouwers bij het omschakelen naar een ge'integreerde bedrijfsvoering. Landbouwuniversiteit/IKCdPAGV, Wageningen/Lelystad.

This Page Intentionally Left Blank

361

Author Index Agtiera, F., 145 Ammon, H.U., 227 Arsene, G.G., 237 BaRs, G.R., 67 Bockstaller, C., 329 Bonciarelli, F., 179 Bonet Torrens, M., 123 Bonnet, A.-C., 135 Bosch Serra, A.D., 123 Breman, H., 39 Bryson, R.J., 77 Bullock, P., 29 Cabelguenne, M., 113 Chassin, P., 257 Clark, W.S., 77 Dauzat, J., 87 Debaeke, P., 113, 217 Delprat, L., 257 Dirks, B.O.M., 57 Domingo Oliv6, F., 123 Durand, J.-L., 135 Edmeades, G.O., 155 Elings, A., 155 Ellis, R.H., 67 Eroy, M.N., 87 Etchebest, S., 135

Hadley, P., 67 Hammer, G.L., 99 Hassink, J., 171,245 Haverkort, A.J., 191 Helander, C.A., 309 Hensen, A., 57 Jambert, C., 257 Janssen, B.H., 267 Keating, B.A., 99 Kessler, J.J., 39 Kub~it, J., 245 Langeveld, C.A., 57 Leakey, R.R.B., 19 Leterme, Ph., 237 Lin6res, M., 257 Mary, B., 237 Meinke, H., 99 Melines Pag6s, M.A., 123 Mengel, K., 277 Meynard, J.M., 339 Minguez, M.I., 191 Morison, J.I.L., 67 Morvan, T., 237

Neeteson, J.J., 171 Neumann, R., 49

Feil, B., 227

Orgaz, F., 145

Garibay, S.V., 227 Gastal, F., 135 Gatmt, J.L., 201 Ghesqui6re, M., 135 Girardin, P., 329

Paveley, N.D., 77 Porceddu, E., 3 Rabbinge, R., 3, 99 Rossing, W.A.H., 399

Sanchez, P.A., 19 Scott, R.K., 77 Segers, R., 57 Somers, B.M., 353 Stamp, P., 227 Stockdale, E.A., 201 Stockle, C.O., 113, 217 Struik, P.C., 179 Sylvester-Bradley, R., 77 van den Pol-van Dasselaar, A., 57 van der Weft, H.M.G., 329 van Keulen, H., 99, 191 van Ittersum, M.K., 339 Velthof, G.L., 57 Vereijken, P., 293 Villalobos, F.J., 145 Vos, J., 201 Wheeler, T.R., 67 White, J.W., 155 Whitmore, A.P., 245 Wijnands, F.G., 319 Yang, H.S., 267

This Page Intentionally Left Blank

Subject Index Acacia seyal 41 Africa, sub-Saharan 20, 39 Agricultural Systems 100 Agriculture broadened objectives v, 8, 10 education 4 history 3 infrastructure 26 land use 6, 9 policy, CAP 7 research and development 4, 10, 49, 339, 348 Agro-ecological characterization 29 indicators 329 zones 31 Agroforestry 24, 40, 46 Agroindustry 49 Agronomy broadened objectives v exploration 8, 12 history 3 objectives 8 reorientation vi, 6, 13 synthesis v Allium cepa L. 123 Ammonia volatilization 173, 241 Arable farming 191,268,294, 309, 319 integrated 329 irrigated 194 rainfed 194 AUDPC 82

Breeding 123, 155, 192 Calcium phosphates 281 Carbon dioxide elevated 69, 74 emission 57 Catch crop 179 Cattle 60, 172 Chemical crop protection 49 product innovation 50 synthesis 53 Chlorophyll meter 230 Climate change 33, 57, 67 Coconut 87 Cocos nurcifera 87 Cover crop 184, 229 Crop canopy 78, 81 development 70, 72, 220 disease 82, 182 growth 69, 103, 156, 221 growth models 34, 147, 157, 217 parameters 117 protection 49, 51,297, 312, 320, 345 residues 180, 269 requirements 31, 221 rotation 181,299, 310, 320, 341,344 thermal time 70, 231 yield 51, 71, 73, 116 Cropping systems 45, 99, 184, 227, 277, 339

Biological control 183

Dairy farms 58, 171 Databases, soil, climate 33 Decision aid tool 329, 334, 340 Denitrification 58, 173,258 Diagnostics 206, 341 Dissolved organic carbon (DOC) 257 Drainage 62 Drip irrigation 124 Drought resistance 137 Ear density 341 formation 164 prolificacy 164 Early vigour 146, 153 Ecological infrastructure 299, 312, 315 Eddy correlation 58 Energy efficiency 314 Environment 319 assessment 330 exposure to pesticides 297, 322 impact 331 risk assessment 31, 34 European research network 293 European Union 7, 293 Evapotranspiration 114, 118, 145, 150 Exploration, model based v, 155, 340, 346, 349 Extraction methods, cold water 258 Farm network 302, 330, 334

363

364 Farming system arable 293, 309, 319, 329, 339,353 case study 339 conventional arable 310, 320 ecological arable 294, 306, 311,319 flower bulb based 344 integrated 311,320 integrated arable 293, 306, 310, 319, 329, 353 organic arable 294, 306 sustainable 293, 309, 319, 329, 339, 353 Farms experimental 171, 310, 320 pilot 171,302, 330 Farmyard manure 249, 268, 278, 284, 302, 334 Fertilizer nitrogen 22, 229 phosphorus 23,277 recommendation 206 Festuca arundinacea 136 Festulolium 136 Flower bulbs 344 Fodder crops 195 Food production 10, 30, 35 security v, 19, 29, 32 Forage 136 Fuzzy logic 333 FYM 249, 268, 278,284, 302, 334 Gaseous losses 241 Geographical Information 34 Global warming 33, 57, 67 Glycine max, L. 116 Grain yield 71 Grass canopy structure 139 tiller density 138 Grassland 60, 172, 247 Greenhouse gases carbon dioxide 58, 60, 69, 74 emission estimates 58, 63 flux measurements 59

methane 58, 61 N20 58, 61,207, 257 warming potential 60 Groundwater table 58 Growing season 193 Harvest index 73, 151 Helianthus annuus L. 145 Herbicide 323 History 3 IAFS 329 Ideotyping 123, 135, 145, 161, 193 Immobilization 202, 241 Indicator agro-ecologica1297, 321,329 crop sequence 330 environmental 322, 331 environmental impact pesticides 297, 322, 332 nitrogen status 227 organic matter 332 pesticides 315,322 Integrated arable farming 293,306, 310, 319, 329, 353 crop protection 312, 321 farming systems 311, 319 pest management 51 Intercropping 44, 89 Interspecific hybridization 136 Irrigation 116, 136 Irvingia gobonesis 24 Isotopes ~513C203,259 15N 203, 241 Italian ryegrass 136, 228 LAI 79, 105, 158, 179 Land use diversification 23 intensification 23, 25 sustainable 32 transformation 19 Land Degradation 29 Leaf

area growth 105, 108, 136, 139 area index 79, 105, 158, 197 elongation rate 136, 139 form factor 78 green leaf area index 77 growth, day-night 136 healthy area duration 83, 85 senescence 164 water potential 136 Learning adaptive learning 337 computer-aided learning 348 social learning 357 Leguminous crop 184 Ley crop 184 Light interception 40, 42, 92, 163 sum 70, 84 Linear programming 344 Livestock 173,280 Lolium multiflorum 136, 229 Lolium perenne L. 136, 239 LUE see RUE Maize 258 hybrid 116, 156, 162, 227 open pollinated variety 156 Methane 58, 61 Mineralization 269 Model 100, 217, 339 ACCESS 34 ALMANAC 34 AFRCwheat2 73, 217 biomass 118 CERES 206 comparison 113, 219 complexity 113 CRIES 34 CropSyst 114, 181,206, 217 DAISY 217 D6cibl6 343 DSSAT 34 EUROSEM 34 EPIC 34, 217 exploration v, 155, 340, 346 H6nin-Depuis 332

365 LEACH.M 34 LINTUL 194 maize growth 157 mechanistic 340 N-cycle 204 MIR 93 Musc 93 OILCROP-SUN 147 ORYZA 206 Otelo 343 PAPRAN 196 regression 342 soil organic matter 157, 248, 268 soil-plant cycle 202 soil water 115 SUCROS 87 157 SUNDIAL 34, 205 synthesis v validation 116 winter wheat 73 WOFOST 34 yield 118 yield loss 82 Mulch seeding of mulch 229 living mulch 229 Multi criteria method 330, 336, 341,344 Mycorrhiza 283 Nature and landscape 295 Nematodes 182, 320 Nitrate in maize 229 leaching 173 Nitrogen 99, 101,103 15N 240 accumulation 173 atmospheric deposition 174 availability 40, 223, 314 available reserves 313 budget 171,209 competition for 252 content in plant parts 206, 221,231 diagnostics 206, 233 dynamics 43, 203, 240

fertilisation 58 immobilization 241 in dung and urine 173 in organic matter 173 in soil 229 limitations 99, 110 losses 173, 187, 207 management 172, 196, 316 mineralization 202 models 204, 219 nutrition index 136, 138 plant nitrogen use 105, 110 recovery 42, 241 recycling 43,203, 240 requirements 197, 219 soil nitrogen supply 202, 239 soil- plant transfer 202 status of maize 230 surplus 174, 345 uptake 105, 110, 173, 204, 223,240 use efficiency 196, 205, 341 utilisation 314 Nitrous oxide 58, 61,207, 257 NUE 196, 205, 341 Nutrient balance 21, 43, 185 catch crop 187 loss 41 management 299, 313 residual 184 Objectives 295, 313, 344 Onion 123 Optimalisation 344 Organic farming 294, 306 inputs 249, 296, 344 Organic matter 332 capacity to protect 246 dissolved organic carbon 257 dynamics 253,267 farmyard manure 250, 269 fractionation, density 248 fractionation, size 248 indicator 245, 336 inputs 249, 296, 344 physical protection 246

pools 203 roots and stubble 269 soil type 250 straw 269 Pasture 60 Peat soil drained 58, 62 pasture 58, 60 Perennial ryegrass 136, 140, 239 Pesticide risk evaluation 322 Pesticides 49, 315, 319, 332 pH liming 282 Phosphate adsorption 280 availability 185, 279 fertilizer 277, 284 fixation 280 inorganic 280 mobilization 286 reserves 278 rock phosphate 285 Policy enabling 20, 25 EU-CAP 7 GATT 7 Porous cup vacuum extraction 258 Potato 192, 320 Productivity 20 agricultural 5, 10, 12 labour 5 post-harvest losses 23 soil related constraints 21 Prototyping v, vi, 294, 310, 320, 339, 349 Prunus africana 24 Radiation incident 70, 80, 157, 193 intercepted 78, 84, 93, 161, 193 use efficiency 83, 105, 108, 161,194 Regulations 7, 177 Research and development agriculture 4, 10 chemical crop protection 49

366 participatory 339, 348 Resistance: systemic activated 54 Resource use efficiency vi, 6, 22, 161, 180 Risk and benefit 52, 354 Root competition 43 development 129 distribution 128 length-density 125 mass 148 sampling 125 uptake (N,P,K) 127 Root-shoot relation 127, 148 Root system tree 43 onion 123 RUE 83, 105, 108, 161,181,194 Sahel 20 Scaling, nitrogen budgets 209 Sclerocarya birrea 41 Season length 146, 149 Semi-arid 99 Shorea javanica 24 Simulation model 34, 40, 100, 149, 217,219, 341 Size and density fractionation 248 Slurry 238 Smallholder 22 Soil cultivation 312 depth 149 fertility 20, 46, 181,253,330 moisture 115 organic matter 42, 245 soil-bome fungi 182, 186 test for P 279 type 250 Soil cover index 313

Soil organic matter 44, 46, 203, canopy cover 45 damar 24 245, 332 domestication of indigenous content 269 24 dissolved organic carbon profitable crops 21, 24 258, 261 dynamics 268 Triticum aestivum L. 67, 77, 100, 217, 341 forest soil 257 from maize 265 Volatilization 208, 241,322 molecular weight 261 Solanum tuberosum L. 192 Water 99 Solar radiation 70, 157, 193 availability 40, 116, 156 Sorghum 116 balance 41, 45 Sorghum bicolor, L. 116 deficit 136 Soybean 116 evapotranspiration 145 Straw 269 extraction 43, 46, 106, 137 Sunflower 145 limitation 99, 106, 109 Sustainable loss 41 agriculture v, 179, 309 stress 116, 137 farming systems 186, 293, transport 115 306, 309, 344, 353 use efficiency 145, 150, 194 land use and management 35 Weed(ing) 183,298, 323 production systems v Wheat 99, 341 Sustainability models 100, 217, 341 objectives 340 spring wheat 100 Symbiotic nitrogen fixation 174 winter wheat 67, 77, 217, 341 System analysis 10 Systemic activated resistance 54 Woody plants 40, 45 Temperature 192 gradient 68 high temperature stress 74 sensitivity 73 thermal time 70, 117, 146 Tillage strip tillage of maize 229 minimum tillage 229 Tillering 74 Transpiration efficiency 101, 106, 110, 150 coefficient 116 Tree bush mango 24 crown 41

Yellow rust 77, 82 Yield actual yields 12 attainable yields 12, 191 components 104, 162, 341 gap analysis 341 grain yields 71, 82, 155 loss by disease 51, 82, 182, 341 potential yields 12, 191 seed 149 silage 230 Zea mays L. 155, 227,258

This Page Intentionally Left Blank

Printed in the United Kingdom by Lightning Source UK Ltd. 117018UKS00001B/26

IL!J!L !!11!11IIIJII 2852