![consists of 7 pages, the last of which has the typewritten name
"Evelyn Strickman" and the date 4-30-53, and bears your initials—
does it not?
Mrs. Siegel. Those are the initials of Marion Cohen, who was
casework supervisor at Youth House at that time. That shows she
read it.
Mr. Liebeler. She read it also?
Mrs. Siegel. Yes.
Mr. Liebeler. And we will mark the document Exhibit No. 2.
(Document marked "Exhibit 2.")
Mrs. Siegel. Wait a minute. Let me just correct that. Marion
would have written her own initials. That isn't my handwriting. I
never made an "E" like that. I don't know who did that.
Mr. Liebeler. You have no question, however, that this is the
report prepared by you?
Mrs. Siegel. No; I have absolutely no question. This is my
dictation into the record. I know—that was Sadie Skolnick. That was
the undersupervisor at the time. That is who that S.S. is.
Mr. Liebeler. I have initialed Exhibit 2. So that we are sure we are
talking about the same exhibit, would you initial it also, please?
Mrs. Siegel. Sure. [Witness complies.]
Mr. Liebeler. Exhibit 1 consists of six pages; is that correct?
Mrs. Siegel. Yes.
Mr. Liebeler. After reviewing the report which you prepared in
connection with Lee Oswald back in 1953, is your recollection
refreshed so that you could add anything other than that which is
already set forth in the written report which you prepared at that
time?
Mrs. Siegel. No; I can't add a thing to that.](https://image.slidesharecdn.com/2655895-250528192740-098c98d1/85/Cropsoil-Simulation-Models-Applications-In-Developing-Countries-Matthews-74-320.jpg)
Cropsoil Simulation Models Applications In Developing Countries Matthews
- 1. Cropsoil Simulation Models Applications In Developing Countries Matthews download https://ebookbell.com/product/cropsoil-simulation-models- applications-in-developing-countries-matthews-5311790 Explore and download more ebooks at ebookbell.com
- 2. Here are some recommended products that we believe you will be interested in. You can click the link to download. Modelling Water And Nutrient Dynamics In Soilcrop Systems Proceedings Of The Workshop On Modelling Water And Nutrient Dynamics In Soilcrop Systems Held On 1416 June 2004 In Mncheberg Germany 1st Edition Kurt Christian Kersebaum https://ebookbell.com/product/modelling-water-and-nutrient-dynamics- in-soilcrop-systems-proceedings-of-the-workshop-on-modelling-water- and-nutrient-dynamics-in-soilcrop-systems-held-on-1416-june-2004-in- mncheberg-germany-1st-edition-kurt-christian-kersebaum-4390128 Soil And Crop Sensing For Precision Crop Production Minzan Li https://ebookbell.com/product/soil-and-crop-sensing-for-precision- crop-production-minzan-li-46090696 Microbial Interventions In Agriculture And Environment Volume 3 Soil And Crop Health Management 1st Ed 2019 Dhananjaya Pratap Singh https://ebookbell.com/product/microbial-interventions-in-agriculture- and-environment-volume-3-soil-and-crop-health-management-1st- ed-2019-dhananjaya-pratap-singh-10806270 Push Button Agriculture Robotics Drones Satelliteguided Soil And Crop Management Krishna https://ebookbell.com/product/push-button-agriculture-robotics-drones- satelliteguided-soil-and-crop-management-krishna-5674238
- 3. Crop Rotation And Cover Cropping Soil Resiliency And Health On The Organic Farm Revised Seth Kroeck https://ebookbell.com/product/crop-rotation-and-cover-cropping-soil- resiliency-and-health-on-the-organic-farm-revised-seth-kroeck-48931366 Visual Soil Evaluation Realising Potential Crop Production With Minimum Environmental Impact Ball https://ebookbell.com/product/visual-soil-evaluation-realising- potential-crop-production-with-minimum-environmental-impact- ball-5288570 Soil Microbiology And Sustainable Crop Production 1st Edition Andrew D Noble https://ebookbell.com/product/soil-microbiology-and-sustainable-crop- production-1st-edition-andrew-d-noble-1887584 Soil Salinity And Salt Resistance Of Crop Plants 2024th Edition Sven Qadir Schubert Manzoor https://ebookbell.com/product/soil-salinity-and-salt-resistance-of- crop-plants-2024th-edition-sven-qadir-schubert-manzoor-63693840 Role Of Rhizospheric Microbes In Soil Volume 2 Nutrient Management And Crop Improvement 1st Ed Vijay Singh Meena https://ebookbell.com/product/role-of-rhizospheric-microbes-in-soil- volume-2-nutrient-management-and-crop-improvement-1st-ed-vijay-singh- meena-7156428
- 6. CROP–SOIL SIMULATION MODELS APPLICATIONS IN DEVELOPING COUNTRIES
- 8. Crop–Soil Simulation Models Applications in Developing Countries Edited by Robin Matthews and William Stephens Institute of Water and Environment Cranfield University Silsoe UK CABI Publishing
- 9. CABI Publishing is a division of CAB International CABI Publishing CABI Publishing CAB International 10 E 40th Street Wallingford Suite 3203 Oxon OX10 8DE New York, NY 10016 UK USA Tel: +44 (0)1491 832111 Tel: +1 212 481 7018 Fax: +44 (0)1491 833508 Fax: +1 212 686 7993 Email: cabi@cabi.org Email: cabi-nao@cabi.org Web site: www.cabi-publishing.org ©CAB International 2002. All rights reserved. No part of this publication may be reproduced in any form or by any means, electronically, mechanically, by photocopying, recording or otherwise, without the prior permission of the copyright owners. A catalogue record for this book is available from the British Library, London, UK. Library of Congress Cataloging-in-Publication Data Crop–soil simulation models: applications in developing countries/edited by Robin Matthews and William Stephens. p. cm. Includes bibliographical references. ISBN 0-85199-563-2 (alk. paper) 1. Crops and soils—Computer simulation. 2. Crops— Computer simulation. 3. Soils—Computer simulation. 4. Crops and soils—Mathematical models. 5. Crops— Mathematical models. 6. Soils—Mathematical models. I. Matthews, Robin B. II. Stephens, William, Ph. D. S596.7.C72 2002 630’.1’13––dc21 2001052815 ISBN 0 85199 563 2 Typeset by Wyvern 21 Ltd, Bristol. Printed and bound in the UK by Cromwell Press, Trowbridge.
- 10. Contents Contributors ix Abbreviations x Preface xiii 1 Introduction 1 Robin Matthews Part 1: Models as tools in research 7 2 Models as Research Tools 9 Robin Matthews 3 Crop Genotype Improvement 13 Robin Matthews 3.1 Identification and Evaluation of Desirable Plant Characteristics 14 3.2 Environmental Characterization 19 3.3 G ×E Interactions 22 4 Crop Management 29 Robin Matthews 4.1 Yield Gap Analysis 29 4.2 Soil Surface Management 33 4.3 Planting 33 4.4 Water Management 37 v
- 11. 4.5 Nutrient Management 41 4.6 Pest and Disease Management 47 4.7 Weed Management 49 4.8 Harvesting 53 5 Cropping and Farming Systems 55 Robin Matthews 5.1 New Crops and Cropping Systems 55 5.2 Evaluating Sustainability 58 5.3 Farm Household Models 63 6 Regional and National Planning 69 Robin Matthews 6.1 Linear Programming Approaches 70 6.2 Dynamic Simulation Approaches 74 6.3 Limitations 76 6.4 Impact 77 7 Global Level Processes 79 Robin Matthews and Reiner Wassmann 7.1 Impact of Climate Change on Rice Production 80 7.2 Greenhouse Gas Production 86 7.3 The El Niño-Southern Oscillation 90 Part 2: Models as decision-support tools 93 8 Decision Theory and Decision Support Systems 95 William Stephens 8.1 Decisions, Decisions, Decisions 95 8.2 Characteristics of Decision Making 95 8.3 Definitions of Decision Support Systems 97 8.4 Spatial and Temporal Scale 101 8.5 Application 102 9 Tools to Support Operational Decision Making 105 William Stephens and Tabitha Middleton 9.1 Pest Management 105 9.2 Irrigation Scheduling 109 9.3 Optimizing Fertilizer Application 112 9.4 Multiple Decision Support 113 9.5 Deciding whether to Implement Emergency Relief 114 10 Tools to Support Strategic Decision Making 117 William Stephens and Tabitha Middleton 10.1 Land-use Planning 117 vi Contents
- 12. 10.2 Planning for Climate Change 120 10.3 Crop Forecasting 121 10.4 Irrigation Planning 123 10.5 Assessing the Benefit of Proposed New Technologies 124 10.6 Planning Optimum Farm Management Strategies in Collaboration with Extension Services and Farmers 125 11 Why has the Uptake of Decision Support Systems been so Poor? 129 William Stephens and Tabitha Middleton 11.1 Model Construction Constraints 132 11.2 Marketing and Support Constraints 136 11.3 Technical and Operational Constraints 141 11.4 User Constraints 143 11.5 Other Constraints 144 11.6 Criteria for Success of Decision Support Systems 145 11.7 Risks Associated with using Decision Support Systems 146 Part 3: Models as tools in education and training 149 12 Using Models as Tools in Education and Training 151 Anil Graves, Tim Hess and Robin Matthews 12.1 Introduction 151 12.2 Using Existing Simulation Models 152 12.3 Model Building 163 12.4 Transferring the Systems Approach to Less-developed Countries 165 12.5 Considerations for Education and Training 173 12.6 Limitations and Constraints of Models in the Educational Context 178 Part 4: Have crop models been useful? 183 13 Who are Models Targetted at? 185 Robin Matthews 13.1 Researchers 186 13.2 Consultants 187 13.3 Educators and Trainers 188 13.4 Policy Makers 189 13.5 Extensionists 190 13.6 Farmers 191 14 Impacts of Crop–Soil Models 195 Robin Matthews, William Stephens and Tim Hess 14.1 Limitations to Use 195 Contents vii
- 13. 14.2 Constraints to the Uptake of Crop Models 196 14.3 Characteristics for Impact 198 14.4 Concluding Remarks 205 Part 5: The way forward 207 15 Where to Now with Crop Modelling? 209 Robin Matthews 15.1 Modelling Rural Livelihoods 211 15.2 Contribution to Crop Improvement Programmes 217 15.3 Making Information Available – Decision Support Systems 219 15.4 Integrating Model Use into Research and Extension Projects 221 15.5 Environmental Research 224 15.6 Building Modelling Capacity 226 15.7 Further Crop–Soil Model Development 227 15.8 Summary 229 16 Concluding Remarks 231 Robin Matthews and William Stephens References 235 Appendix: Personal Communications 267 Index 271 viii Contents
- 14. Contributors A. Graves, Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK. T. Hess, Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK. R. Matthews, Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK. T. Middleton, Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK. W. Stephens, Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK. R. Wassmann, Fraunhofer Institute for Atmospheric Environmental Research, Kreuzeckbahnstrasse 19, 82467 Garmisch-Partenkirchen, Germany. ix
- 15. Abbreviations AEGIS Agricultural and Environmental Geographic Information System AMMI Additive Main effects and Multiplicative Interaction ANOVA Analysis of variance APSIM Agricultural Production Systems Simulator AWS Automatic weather station BBF Broadbed-and-furrow CAL Computer-assisted learning CAP Common Agricultural Policy CATIE Centro Agronomic Tropical de Investigacion y Ensenanza CERES Crop Environment Resource Synthesis CIP International Potato Centre CLUES Centre for Land Use Studies CTI Computers in Teaching Initiative DAD (South African) Department of Agricultural Development DFID Department for International Development DNDC Denitrification and decomposition DSS Decision support system DSSAT Decision Support System for Agrotechnology Transfer ENSO The El Niño-Southern Oscillation EPIC Erosion Productivity Impact Calculator EPIPRE Epidemic prevention ES Expert system EU European Union FAO Food and Agriculture Association x
- 16. FARMSCAPE Farmers, Advisers and Researchers Monitoring Simulation, Communication and Performance Evaluation FRS Fertilizer recommendation system FSR Farming systems research GCMs General circulation models GFDL General Fluid Dynamics Laboratory GHGs Greenhouse gases GIS Geographical information system GISS Goddard Institute of Space Studies G ×E Genotype by environment HRI Horticultural Research International IBSNAT International Benchmark Sites Network for Agrotechnology Transfer ICASA International Consortium for Agricultural Systems Applications ICRISAT International Crops Research Institute for the Semi-Arid Tropics IMS Irrigation management services IPM Integrated pest management IRRI International Rice Research Institute IWR Irrigation Water Requirements LDCs Less-developed countries LINTUL Light Interception and Utilization LP Linear programming LUT Land-use type MAFF Ministry of Agriculture Fisheries and Food MARS Monitoring Agriculture with Remote Sensing MDS Minimum data set MERES Methane Emission from Rice Ecosystems METs Multi-environment trials NARCs National Agricultural Research Centres NGO Non-governmental organization NR Natural resources OR Operations research QTL Quantitative trait loci R&D Research and development RLWR Root length/weight ratio REPOSA Research Programme for Sustainability in Agriculture RIL Recombinant inbred line RUE Radiation use efficiency SARP Simulation and Systems Analysis for Rice Production SASEX (South African) Sugar Association Experiment Station SL Sustainable Livelihoods SLA Specific leaf area SOI Southern Oscillation Index Abbreviations xi
- 17. SOLUS Sustainable Options for Land Use SWB Soil water balance UKMO United Kingdom Meteorological Office UNED Universidad Estatal a Distancia USAID United States Agency for International Development WARDA West African Rice Development Association WUE Water use efficiency xii Abbreviations
- 18. Preface From 1990 until 1999, the United Kingdom’s Department for International Development (DFID) funded work on developing a suite of models to address problems relating to crop production in the semiarid tropics, specif- ically the evaluation of rain-water harvesting, and maintenance of soil fer- tility. This work resulted in the PARCH, PARCHED-THIRST, SWEAT and EMERGE models. However, a number of subsequent studies (e.g. Fry, 1996; Stephens and Hess, 1996; Kebreab et al., 1998) showed that uptake and use of these models was limited to non-existent. This has led to questions being asked as to whether crop simulation modelling and systems analysis approaches have any contribution to make in addressing problems in developing countries. A general weakness of all the models was that a clear definition of who potential users were had received scant attention. The models were devel- oped to support the solution of problems in natural resources management, but this was not in response to a known and well-articulated demand from potential users in the natural resources sector of developing countries. The study by Stephens and Hess (1996) identified the first limitation to the uptake of a model as the inability of a potential user to be able to per- ceive a relevance of the model to his/her work, or lack of appreciation of what the model could be used for. All three studies emphasized the need for continued support to end-users of the models if there is to be uptake. Both of these points have been recognized by the Natural Resources Systems Programme (NRSP) of DFID, and in October 1999, a workshop funded by NRSP was held at IACR-Rothamsted to review the current sta- tus of the PARCH suite of models and crop models in general, and to dis- cuss options for taking DFID-funded modelling activities forward, with xiii
- 19. particular emphasis on the application of systems approaches to contribute to the solution of real-world problems in developing countries. This book is a progression from discussion points raised in the wrap-up session of this workshop. The purpose of the work was to make a thorough review of the literature to identify past and current applications of crop–soil sim- ulation models in general, identify the limitations of such models, charac- terize groups of end-users of the models, and to attempt a synthesis of where such models might be useful in the future in contributing to system- based, poverty-oriented research projects in developing countries. Many people have contributed to the ideas in this book in many dis- cussions over several years. Prominent among these are Professor Tony Hunt, at the University of Guelph, Canada; Professor Jim Jones and Professor Gerrit Hoogenboom, at the University of Gainesville, Florida; Dr Walter Bowen, IFDC; Professor Martin Kropff, Wageningen Agricultural University, The Netherlands; Dr P.K. Aggarwal, IARI, India; Dr Ino Lansigan, UPLB, The Philippines; Dr Attachai Jintrawet, Thailand; and Dr Kevin Waldie, University of Reading, UK. We are also grateful to the par- ticipants in the Rothamsted workshop in October 1999 for the useful dis- cussions that set us on the route to writing this book – Dr John Gaunt, IACR-Rothamsted (the workshop organizer); Dr Georg Cadisch, Wye College; Dr Neil Crout, University of Nottingham; Dr John Gowing, University of Newcastle; Mr Gerry Lawson, Institute of Terrestrial Ecology; Mr Frans-Bauke van der Meer, Silsoe Research Institute; Dr Robert Muetzelfeldt, University of Edinburgh; Dr Lester Simmonds, University of Reading; Dr Terry Thomas, University of Wales, Bangor; Dr Geoff Warren, University of Reading; Dr Ermias Kebreab Weldeghiorghis, University of Reading; and Dr Damion Young, University of Newcastle. Substantial use was also made of Internet discussion groups to obtain information and views from the international community involved in modelling agricultural systems. Principally, these were the AGMODELS list- server (AGMODELS-L@UNL.EDU), the DSSAT listserver (DSSAT@LIST- SERV.UGA.EDU), the ESA-AGMODELS listserver (ESA-AGMODELS@ESA. UDL.ES), and the FAO-AGROMET listserver (AGROMET-L@MAILSERV. FAO.ORG). We are grateful to all subscribers of these groups who respond- ed to our questions, and have attempted to give credit to the source where we have included these comments in this book. This publication is in large part an output from a programme develop- ment assignment funded by DFID for the benefit of developing countries. The views expressed are not necessarily those of DFID. A condensed version of Part 1 has been published in Advances in Agronomy (Matthews et al., 2002). xiv Preface
- 20. Introduction Robin Matthews Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK Arable agriculture is a major way in which people interact with the natu- ral resource base in developing countries; this may not always be to the long-term benefit of either, particularly if cropping practices are subopti- mal or inappropriate. Traditional agronomic research has made remarkable advances in recent years in improving some of these practices, but new tools being developed, such as crop and soil simulation models with their ability to integrate the results of research from many different disciplines and locations, offer a way of improving the efficiency and/or reducing the cost of some of this research. Since research organizations cannot afford to generate technologies that are inappropriate, more use is being made of systems methods to ensure that research is relevant (Goldsworthy and Penning de Vries, 1994). Their use in a research programme has the poten- tial to increase efficiency by emphasizing process-based research, rather than the study of site-specific net effects. This is particularly attractive in developing countries, where scarce resources may limit effective agricul- tural research. There is also an increasing need to understand how agricultural systems interact with other segments of society. The population of the world is increasing by over 70 million per year, and it is likely that there will be 1.8 billion more people in the world by 2020 (Pinstrup-Andersen et al., 1999). To meet the demand for food from this increased population, the world’s farmers need to produce 40% more grain by 2020. Moreover, if certain climate change scenarios come to pass, agricultural production in some areas may decrease. There are many cases of land degradation, and a lack of new land that can be brought into agricultural production. How can productivity be increased while ensuring the sustainability of 1 © 2002 CAB International. Crop–Soil Simulation Models (eds R. Matthews and W. Stephens) 1
- 21. agriculture and the environment for future generations? Decision makers need information supplied by research to make informed choices about new agricultural technologies and to devise and implement policies to enhance food production and sustainability. Ultimately, however, it is the farmer who makes the final choices about acceptance of a new technolo- gy or method. Policy makers need to understand the impacts of their deci- sions on the wellbeing of farm households, on the natural resource base, and on the regional or national economy. The users of information gener- ated through research and encapsulated in models are not just farmers but decision makers at all levels in the public and private sectors. There are many types of models that have been published describing various aspects of agricultural production systems, and it is all too easy to be overwhelmed by their sheer numbers. However, as this book evolved from discussions on the crop modelling work already funded by DFID and how this might be taken forward, we have, therefore, restricted our focus to crop simulation models, which may or may not have components describing soil processes and pests and diseases (Penning de Vries, 1990). We have adopted the definition of Sinclair and Seligman (1996) that a crop model is ‘the dynamic simulation of crop growth by numerical integration of constituent processes with the aid of computers’. More specifically, this implies a computer program describing the dynamics of the growth of a crop (e.g. rice, wheat, maize, groundnut, tea, etc.) in relation to the envi- ronment, operating on a time-step an order of magnitude below the length of a growing season, and with the capacity to output variables describing the state of the crop at different points in time (e.g. biomass per unit area, stage of development, yield, canopy N content, etc.). We have not gener- ally included models that only predict some final state such as biomass or yield (Whisler et al., 1986). Nevertheless, we have permitted ourselves to deviate occasionally and consider models and their applications that might be outside this definition, but only if we feel that there are lessons to be learned that are relevant to the way ahead for crop and/or soil simulation modelling in the context of natural resources systems research. In a task of this kind, it becomes necessary to classify model applications in order to provide the basis for some kind of meaningful discussion. There are many different ways that the uses of models can be classified: Passioura (1996), for example, classifies models into two groups – scientific models (i.e. helping with understanding) and engineering (i.e. applying science to solve a problem). Mindful of the broad groups of end-users of crop simu- lation models, we have expanded this classification, and have divided model applications into: (i) those used as tools by researchers, (ii) those used as tools by decision-makers, and (iii) those used as tools by those involved in education, training and technology transfer. We are the first to recognize that this is not a perfect classification and that there is bound to be over- lap between the groups, but have found it to be a useful way of thinking about common characteristics of models from the point of view of the people 2 R. Matthews
- 22. who will be using them. Where a particular application falls into more than one classification, we generally discuss it under both headings, with the focus on the aspects relevant to each classification. We have not attempted to review every single instance of a crop model application, as that would require considerably more time than we had available. Instead, we have attempted to cover all the broad types of uses to which crop models have been put, and have used as many examples of each as possible to illustrate the use of models in that area. We recognize that there will probably be many good examples of model applications that we have not included; we apologize to the people involved and hope that they can appreciate that it is only space and time that prevented us from doing so. We have also focused on applications of crop models in devel- oping countries, although we have drawn substantially on experiences with tropical crops in Australia, as many of the examples there are relevant to possible applications of systems analysis techniques in other tropical coun- tries. We have not generally included examples from temperate agricultur- al systems, except where we feel there were interesting lessons learned that had some relevance to agriculture in developing countries. We also recognize that there is a certain element of unavoidable bias in such a review towards instances where models have been successfully applied. Cases where models have failed or have been unsuccessfully applied are generally not reported in the literature. However, we make no apologies for this bias, in the same way as a plant breeder is not required to apologize for the 99% or more of his material that are ‘failures’. Just as progress in plant breeding is made with the proportion of individuals that are ‘successful’, the aim of this study is to as impartially as possible identify areas where models have been applied successfully, so that future model- ling activities can be focused in those areas. Indeed, we would argue that the notion of a model being a ‘success’ or a ‘failure’ is somewhat mean- ingless, anyway – in research, the most useful model is often the one that fails as it can point the way to new thinking and research (Seligman, 1990). On the other hand, we do recognize that there is a cost to research that has not produced results, and with this in mind we have attempted to make appraisals of the limitations of the models used where possible, and have discussed constraints to their uptake and impact. Before describing the various applications of crop simulation models that have been published, it is perhaps useful to summarize the history of crop modelling to provide some perspective. Sinclair and Seligman (1996) have given an excellent overview of developments, drawing parallels between the growth and development of crop simulation models and human beings. They describe how the infancy stage began after the birth of crop modelling more than 35 years ago with the advent of the mainframe com- puter in the 1960s (Bouman et al., 1996). The first steps for crop models were models designed to estimate light interception and photosynthesis in crop canopies (e.g. Loomis and Williams, 1962; de Wit, 1965). These were Introduction 3
- 23. relatively simple models, but they provided for the first time a way of quan- titatively and mechanistically estimating attainable growth rates of various crops. They showed that the potential yield of a crop could be defined in terms of the amount of solar radiation energy available for the accumula- tion of chemical energy and biomass by plants. The juvenile stage that followed in the 1970s seemed to open up wide areas of research, and led to the development of so-called ‘comprehensive’ models mainly aiming at increasing understanding of the interactions between the crop and the main growth factors. This stage also coincided with rapid advances in equip- ment for field experimentation to provide data describing the various phys- iological processes that were incorporated into the models, which inevitably led to an increase in their complexity. However, this complex- ity meant that the number of parameters required to describe the system in detail increased dramatically. Errors in the values of these parameters obtained from field experimentation often propagated through the model. Other parameters could not be measured directly and had to be estimated. The adolescence stage in the early 1980s saw a re-evaluation of the basic concepts of crop modelling in the light of accumulated evidence. The first of these was the assumption that the reductionist approach of increasing the complexity of a model led to better models. It had become apparent that much of the behaviour of a system could be captured by a few key variables, with the inclusion of further variables only adding marginally to the accuracy of the model, if at all. This led to the emergence of simplified versions of the comprehensive models, or so-called ‘summary models’ (Penning de Vries et al., 1989), and, even more recently, still simpler ‘par- simonious’ models (e.g. ten Berge et al., 1997b; Peiris and Thattil, 1998). In these latter models, a system is modelled with only a few key variables in an attempt to keep a model simple, both so that it is easily understood by potential users, and also that its requirements for input data are marked- ly reduced. Nevertheless, a more detailed crop simulation model may often be used to help develop the simpler model. The second re-evaluation was of the original assumption that a universal model could be developed for each crop, with the realization that the nature of the problem to be solved dictated the nature of the most appropriate model to use. This brought a move towards ‘bespoke’ models built with a specific purpose in mind. The maturity phase in the 1990s brought a growing awareness of the limitations of crop models and a better understanding of the nature of these limitations, some of which we discuss in more detail in this book. Many objections have been raised to the use of deterministic crop growth models, ranging from lack of confidence in the method altogether (e.g. Passioura, 1973; Monteith, 1981), through the problems of their data requirements, the ‘parameter crisis’ (Burrough, 1989b), the stochastic nature of the input data used (Burrough, 1989a), the fact that model results necessarily pertain to single events which causes application problems in spatially and temporally variable environments, to the complaint that the 4 R. Matthews
- 24. models cannot reproduce the actual situation. On the positive side, how- ever, there does seem to be general agreement across the board that the development of such models has brought benefits by providing the oppor- tunity to formulate consistent quantitative statements on the behaviour of the systems under consideration, that the consequences of alternative options can therefore easily be made explicit, and as such, these models form a tangible basis for discussion. In the following chapters, we would like to extend the human develop- ment analogy of Sinclair and Seligman (1996) and describe the first employ- ment these crop models have had, and offer some thoughts about how their job prospects might develop from here, with particular focus on their relevance to agriculture in developing countries. Introduction 5
- 26. Part 1 Models as tools in research
- 28. Models as Research Tools Robin Matthews Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK Crop simulation models were originally developed as research tools, and have probably had their greatest usefulness so far in being part of the research process. The advantages of integrating simulation modelling approaches into a research programme have often been stated – Seligman (1990), for example, lists the following uses of models in research: • identification of gaps in our knowledge; • generation and testing of hypotheses, and an aid to the design of exper- iments; • determination of the most influential parameters of a system (sensitivity analysis); • provision of a medium for better communication between researchers in different disciplines; • bringing of researchers, experimenters and producers together to solve common problems. Boote et al. (1996) see models as providing a structure to a research pro- gramme, and being particularly valuable for synthesizing research under- standing and for integrating up from a reductionist research process, but point out that if the efficiency of research is to be increased, the model- ling process must become a truly integrated part of the research activities. Experimentation and model development need to proceed jointly – new knowledge is used to refine and improve models, and models are used to identify gaps in our knowledge, thereby setting research priorities. Sinclair and Seligman (1996) make a similar point, seeing models as a way of setting our knowledge in an organized, logical dynamic framework, allowing identification of faulty assumptions and providing new insights. 2 © 2002 CAB International. Crop–Soil Simulation Models (eds R. Matthews and W. Stephens) 9
- 29. They propose that models should be seen as aids to reasoning in research and teaching about the performance of a crop or the benefits of alterna- tive management strategies. An interesting example of the use of models to provide new insights into crop processes for future research to focus on is provided by Matthews and Stephens (1998b). During the development of a simulation model for tea (Camellia sinensis), it was found that temperature alone could not be used to simulate the large peak in tea production in September in Tanzania. Various potential mechanisms were evaluated, but the only one that was able to adequately explain this peak was the assumption that the growth of dormant shoots was triggered at the time of the winter solstice, allow- ing a large cohort of shoots to develop simultaneously and reach har- vestable size at the same time (Fig. 2.1). The proposed mechanism, in which shoot dormancy was induced by declining photoperiod and released by increasing photoperiod, was also able to accurately simulate patterns of shoot growth in the northern hemisphere (Panda et al., 2002). An exper- iment was planned to test the hypothesis, but was subsequently cancelled when the company that was to fund the work sold their interests in tea. The matter is not only academic – the large September production peak can often exceed factory processing capacity with a subsequent loss of har- vested material. If the peak could be manipulated with, say, supplemen- tary lighting to offset the photoperiod effect, a more even spread of production over the year might be obtained. Many of the crop model applications discussed in the following pages involve an assessment of risk, so we feel that it is worthwhile to say a little 10 R. Matthews Fig. 2.1. Comparison of observed (——) yields of fully irrigated tea in Tanzania with yields simulated using the CUPPA-TEA model (----) (redrawn from Matthews and Stephens, 1998b).
- 30. at this stage about what risk is, what causes it, and how different people perceive risk in different ways. Risk and uncertainty are inherent in agriculture, particularly in tropical countries. The influence of weather, pests and diseases, and prices and costs, are all unknown in advance to varying degrees. This risk and uncertainty is important as it affects deci- sion making at a number of levels – the same decision made in a low-risk environment may be totally inappropriate in a high-risk environment. Efficient agricultural management has much to do with the management of this risk, both at the household and regional levels. At the household level, a farming family may try to maximize income fluctuations over time; at the national level on the other hand, a government may try to ensure an adequate supply of food to the population in all sectors of society. Risk to crop production may be short term, such as fluctuations in climate or socio-economic conditions, or long term, such as degradation of soil fer- tility. Often production in the short term can be maximized, but sometimes it may be at the expense of an increase in resource degradation in the long term. Also, different people have different perceptions and time scales of risk – an individual farmer may be much more concerned about the risk of crop failure in the next season than the risk of long-term decline in soil fertility, whereas a government may be much more concerned about long- term harm to the environment. Wade (1991) classified farmers into three groups: (i) risk takers – those who aim for high productivity in a good year, and are prepared to accept crop failure or low yields in some years; (ii) risk avoiders – those who are prepared to sacrifice some yield in a good year as long as the risk of crop failure in poor years is minimized; and (iii) those who fall somewhere in between (i) and (ii). Commercial farmers with access to credit and who can afford high levels of inputs would generally fall into category (i), while subsistence farmers with little or no access to means to buffer year-to-year variability would generally fall into category (ii). With access to long sequences of historical weather data, crop models can be excellent tools for assessing the production variability asso- ciated with weather for various strategies (Thornton and Wilkens, 1998). Traditional field experiments to obtain the same assessment of risk associ- ated with a particular strategy would be virtually impossible due to the time and cost involved. In the following chapters describing applications of models as research tools, we have started at the level of the genotype, considering how mod- els may contribute to the process of genotype improvement, then moved to the level of the whole crop, discussing applications aimed at under- standing and improving crop management. We then progress to how individual crops fit into an overall farming system, looking at ways in which these systems can be optimized to meet certain goals and how they con- tribute to the livelihoods of the farmers involved. Finally, we consider how crop models have contributed to the policy-making process at the nation- al and international levels. Models as Research Tools 11
- 32. Crop Genotype Improvement Robin Matthews Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK The goal of any plant breeding programme is the development of new improved cultivars or breeding lines for particular target areas and for specific applications. In general, the time from initial selection of individ- ual plants to the release of a new cultivar can take up to 10–15 years, and in most cases, improvements of only a few per cent are obtained for each new cultivar over current ones. Any new techniques of improving the effi- ciency of the improvement process are, therefore, of considerable interest to plant breeders. Chapman and Bareto (1996) have defined increasing the efficiency of plant breeding as increasing the rate of genetic gain, given particular levels of research resources and genetic variability. The overall process of crop improvement can be subdivided into three phases – a planning and hybridization phase, a segregation and stabiliza- tion phase, and a line evaluation and release phase (Hunt, 1993). The per- sonal time of a breeder can be allocated between these three phases in different ways, but a suggested allocation for a wheat improvement pro- gramme is shown in Table 3.1. Regardless of the type of crop improvement programme, most breeders consider the first phase, the design of a new genotype for a particular environment, the selection of parents with char- acteristics matching this design and the initial hybridization, to be of criti- cal importance, with around 40% of their time being allocated to it (Hunt, 1993). Even with careful matching of parents, the chances of success depend on the numbers of lines evaluated each year. Thus, given the increasingly marginal returns from conventional breeding approaches, it is timely to seek more efficient methods that might help improve the efficiency of this phase. The emergence of simulation models for a large number of crops pro- vides tools that may be useful in helping to improve the efficiency of the 3 © 2002 CAB International. Crop–Soil Simulation Models (eds R. Matthews and W. Stephens) 13
- 33. crop improvement process. Both Shorter et al. (1991) and Lawn (1994) stress the need for an integrated multidisciplinary approach between plant breeders, crop physiologists, and crop modellers. Cooper and Hammer (1996), summarizing the results of a workshop on plant adapta- tion and crop improvement held at the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), highlighted the use of mod- els in crop improvement programmes in three main areas: (i) identification and evaluation of desirable plant characteristics to aid indirect selection methods; (ii) characterization of the target environments for different germplasm; and (iii) partitioning genotype × environment (G ×E) interac- tions to increase the sensitivity of the analysis of variance of trial data. However, for crop simulation models to make significant contributions in these areas, there is a need for much collaborative research to be done between physiologists and plant breeders. Until this research is conduct- ed, and the benefits of shifting resources into the systems approach can be weighed against reducing resources put into the conventional empirical approach, widespread acceptance of the new methods is unlikely (Hammer et al., 1996a). In the following sections in this chapter, work that addresses these three areas is summarized and discussed. 3.1 Identification and Evaluation of Desirable Plant Characteristics Direct selection for crop yield is generally perceived as costly and ineffi- cient because of its low heritability (White, 1998), despite being the main method of selection for superior germplasm up until the present time. Much effort, therefore, has gone into the identification of traits which breeders might select for to increase yield indirectly. Crop models offer a way in which various traits can be evaluated simply and easily. Varying only one plant parameter at a time while keeping the rest of the parameters constant is analogous to the creation of genetic isolines, something that requires a good deal of time and effort in reality. Although a single trait may be of interest, a combination of traits, or a crop ideotype, is more often sought. 14 R. Matthews Table 3.1. Division of time between the different phases of activity in a self- pollinated crop breeding programme (Jensen, 1975). Phase Breeders’ time (%) Technicians’ time (%) 1. Planning and hybridization 40 5 2. Segregation and stabilization 10 10 3. Line evaluation and release 50 85 100 100
- 34. The concept of designing a genotype with optimal characteristics for a particular set of conditions was first used by Donald (1968) who designed a small grain cereal for favourable environments. These ideas were subse- quently expanded to develop a general ideotype applicable to cereals, grain legumes and oil seeds (Donald and Hamblin, 1983). The principal traits of this plant were an annual habit, erect growth, dwarf stature, strong stems, unbranched and non-tillered habit, reduced foliage, erect leaves, determi- nant habit, high harvest index and early flowering. Similarly, Cock et al. (1979) proposed a cassava ideotype with late branching, large leaves and long leaf life, based on a model that used weekly time intervals to simu- late leaf development, crop growth and partitioning between roots and shoots. More recently, simulation studies helped in defining morphologi- cal characteristics of the ‘New Plant Type’ of rice currently being devel- oped at the International Rice Research Institute (IRRI; Dua et al., 1990; Dingkuhn et al., 1991). Desirable traits were identified as: (i) enhanced leaf growth during crop establishment; (ii) reduced tillering; (iii) less foliar growth and enhanced assimilate export to stems during late vegetative and reproductive growth; (iv) sustained high foliar N concentration; (v) a steep- er slope of N concentration from the upper to lower leaf canopy layers; (vi) expanded capacity of stems to store assimilates; and (vii) a prolonged grain filling period. Because of their dynamic nature, crop simulation models also offer the opportunity to explore the effect of changing the rates of various physio- logical processes. Various cotton models, for example, have been used since 1973 to assess the effect on yield of traits including photosynthetic efficiency, leaf abscission rates and unusual bract types (e.g. Landivar et al., 1983a, b; Whisler et al., 1986). Landivar et al. (1983b) concluded that if photosynthetic efficiency was correlated with specific leaf weight, then most of the increased growth would go into the leaf with little overall effect on yield. Hoogenboom et al. (1988) used the BEANGRO model to investi- gate the effects of specific leaf area (SLA), root partitioning, rooting depth and root length/weight ratio (RLWR) on seed yield and water use effi- ciency (WUE) of common bean. Yields increased with an increase in root- ing depth, root partitioning, increased RLWR, and increases in SLA up to 300 cm2 g–1 , beyond which there was no increase. Boote and Jones (1988) performed a similar exercise with PNUTGRO, comparing the effects of 16 parameters on groundnut yield under rain-fed conditions over 21 years. Increasing canopy photosynthesis and the duration of the vegetative and reproductive phases both increased yields over 15%. Jordan et al. (1983) and Jones and Zur (1984) found that for soybean growing in a sandy soil, increased root growth was more advantageous than capacity for osmotic adjustment or increased stomatal resistance. By contrast, the GOSSYM model predicted that doubling stomatal resistance would lead to a 28% increase in yield, a conclusion subsequently supported by improved culti- vars (Whisler et al., 1986). Other examples of the use of crop simulation Crop Genotype Improvement 15
- 35. models to investigate the sensitivity of different genetic traits on yields are in soybean (Wilkerson et al., 1983; Elwell et al., 1987) and groundnut (Duncan et al., 1978). Determining the responses of particular genotypes to environmental characteristics is another important area to which crop simulation models have made a contribution. Such an application was reported by Field and Hunt (1974) to help determine the optimum response of lucerne growth to temperature in eastern Canada. Lower production in the latter part of the season was thought to be due to increased ambient temperatures, although it was difficult to confirm this experimentally due to the con- founding influences of various combinations of day, night and soil temperatures. Using what was known of basic temperature responses from controlled environment experiments, the authors constructed a model to calculate the degree to which seasonal changes in temperature controlled lucerne growth. The results supported the hypothesis, and led to the suggestion that breeding work should be directed at selecting clones with more uniform performance at different temperatures. This was subse- quently explored in actual breeding work by McLaughlin and Christie (1980). A development of this approach has been to use long sequences of his- torical weather data and crop models to test the likely performance of a ‘novel’ genotype in a target environment. Differences in predicted yields from year to year give an estimate of the likely risk faced by a farmer in choosing to grow that genotype. This approach is particularly useful in vari- able environments such as, for example, the semiarid tropics, which are characterized by variability in the amount and temporal distribution of rain- fall. These areas pose special problems for effective selection of improved genotypes, as the relative importance of different growth processes in deter- mining final yield, and consequently the value of different traits, may dif- fer between environments and between years in the same environment. It is expensive, if not impractical, to assess the value of different plant types using conventional multi-site, multi-season trials. This in turn, restricts the amount of information available to evaluate the risks associated with dif- ferent plant traits that farmers are likely to face over longer periods of time. As an example, Bailey and Boisvert (1989) used a crop model coupled to long-term weather data to evaluate the performance of a range of cultivars at several locations in the semiarid areas of India by incorporating eco- nomic concepts of risk efficiency. They found that the ranking of the cul- tivars differed from that obtained with the traditional Finlay and Wilkinson (1963) approach, and depended crucially on the simulation of yields, and therefore on the ability of the model to accurately simulate the crop’s response to water deficits. Using a similar approach, Muchow et al. (1991) explored the conse- quences of maize and sorghum breeders selecting for: (i) greater rate of soil water extraction by the root system and (ii) a higher WUE. Their sim- 16 R. Matthews
- 36. ulations showed that in the first case, the resulting faster exhaustion of soil water supply early in growth led to a 20–25% likelihood of yield loss due to lack of rain later to recharge the profile. In the second case, there was a yield gain in all years if the higher WUE was associated with no change (or increase) in radiation use efficiency (RUE), but a 30% chance of yield loss if the increased WUE was associated with lower RUE. In a subsequent study, Muchow and Carberry (1993) used models for maize, sorghum and kenaf based on the CERES crop models to analyse three crop improvement strategies – modified phenology, improved yield poten- tial and enhanced drought resistance. They found that there was no clear yield advantage of the traits in all years, and that the choice of plant type would depend on the farmer’s attitude to risk. They defined a subset of cultivars as ‘risk-efficient’, characterized by a higher mean yield or lower standard deviation. However, the problem remained of how feasible it is in practice to modify the plant in the way shown by the simulations – a higher transpiration efficiency (g dry matter (DM) (g H2 O) –1 ) was shown to be beneficial, but this is generally a very conservative parameter with little genetic variation. The work highlighted clearly the dangers inherent in using conventional selection techniques alone – traits selected for supe- rior yields in a few years only could be very unrepresentative of their per- formance over a much longer time span – but also underscored the need to temper simulation results with information from field experimentation as to what was realistically achievable. In rice, Aggarwal et al. (1996, 1997) used the ORYZA1 model for inves- tigating effects on grain yield of various traits such as developmental rates during juvenile and grain-filling periods, leaf area growth, leaf N content, shoot/root ratio, leaf/stem ratio, and 1000-grain weight. Because of the lack of feed-backs built into this model, however, changing any one of these parameters generally changed yields in the expected way, with the excep- tion of the phenological parameters, which interacted with year-to-year variability in weather. However, these changes were generally small, and they concluded that all parameters need to be increased simultaneously if there is to be any increase in yields – increasing one parameter alone has little effect. They also made the point that increased N applications might be necessary to express the effects of genotypes with higher yield poten- tial as current N practices may be masking this potential. Yin et al. (1997) also used the ORYZA1 model to investigate the effect of variation in pre- flowering duration on rice yields at IRRI in the Philippines, at Hangzhou in China and at Kyoto in Japan. They concluded that the pre-flowering duration was about right for most modern cultivars – if it was any shorter, yield would be sacrificed, any longer and the number of cropping seasons possible per year would be sacrificed. In another study in West Africa, Dingkuhn et al. (1997) used the same ORYZA1 model to investigate traits that would enhance the competitive ability of rice against weeds (see Section 4.7 for further details). Crop Genotype Improvement 17
- 37. Similar approaches have been used to assess the effects of different phenology in different varieties on grain yield for sorghum (e.g. Jordan et al., 1983; Muchow et al., 1991), rice (O’Toole and Jones, 1987) and wheat (Stapper and Harris, 1989; Aggarwal, 1991). Hammer and Vanderlip (1989) simulated the impact of differences in phenology and radiation use effi- ciency on grain yield of old and new sorghum cultivars. Jagtap et al. (1999) used the CERES-MAIZE model to show that short duration varieties performed better than long-duration varieties at three sites in Nigeria, but that the risk of crop failure was high if N was not applied. Other modellers have used simulation analysis to design improved plant types for specific envi- ronments (Dingkuhn et al., 1991; Hunt, 1993; Muchow and Carberry, 1993). Although much interest was generated in this approach, at the practical level, few plant breeding programmes have adopted it in any significant way. Donald (1968) himself recognized several of the difficulties of the approach, dividing them into conceptual (i.e. whether the approach was valid – is there such a thing as a ‘best’ type?), and practical (i.e. could the approach be implemented – e.g. which selectable characteristics deter- mined the ‘best’ type?). Of the latter, one of the most serious is the fre- quent lack of genetic variability in reality of the characters in question. For example, BEANGRO predicts increases in yields with an increase in days to maturity in the absence of temperature or water deficit, but it has been dif- ficult to breed lines that mature later than existing cultivars (White, 1998). Similarly, most models predict that increasing photosynthesis rates will increase yields, but little success has been achieved in practice so far in selecting for genotypes with increased photosynthetic rates. A second major problem is that characters are often negatively correlated, so that selecting to optimize one results in a suboptimization of another (e.g. Kramer et al., 1982), cancelling out any improvement. Models are not usually able to predict these negative correlations in advance, although Boote and Tollenaar (1994) considered possible compensation between traits such as photosynthesis rate and specific leaf weight, and concluded that there was little potential for selecting for higher photosynthetic rates. Nevertheless, even as recently as 1991, Rasmusson (1991) argued that ideotype design for a particular environment was a useful exercise for a plant breeder, as it helped to focus his/her attention on what was or was not known about the environment, what particular characteristics it was practical to select for, and promoted goal setting for particular traits. However, for many traits, crop models may not yet be sophisticated enough to capture the subtle differences between genotypes. For example, Yin et al. (2000) explored the ability of the SYP-BL crop model to explain yield differences between genotypes in a recombinant inbred line (RIL) pop- ulation of barley, in which a dwarf gene (denso) was segregating. When all input parameters were calibrated using data from one of the seasons, the model could explain only 26–38% of the yield variation between the genotypes in a second season. Apparently, this was partly caused by some 18 R. Matthews
- 38. of the variation being due to plant N status, which the model did not account for. However, when the model was calibrated with values from the first season for only three of the parameters, lodging score, pre- flowering duration, and fraction of biomass partitioned to the spike, and with the other model input parameters held constant at their across- genotype means, the model could explain 65% of the yield variation in the second season. The authors make the point that part of the relatively poor performance of the model may be due to the so-called ‘genotype parameters’ it uses actually varying across environments, and give an ex- ample showing that the post-flowering duration and specific leaf area parameters varied with plant N status. This limitation obviously depends on the model being used. They also observe that in most crop models yield is determined by the availability of assimilate (i.e. source-limited) rather than the availability of sites to receive the assimilate (sink-limited). Again, the relative importance of source and sink approaches will depend on the type of crop – Matthews and Stephens (1998a) use a sink-limited approach to model the yields of tea, as previous work had shown that yields were poorly correlated with the source strength (Squire, 1985). 3.2 Environmental Characterization The aim of any plant breeding programme is to develop improved geno- types for a pre-defined target population of environments. The target pop- ulation could be defined either geographically (e.g. wheat varieties for the Punjab region in India) or in terms of a type of environment (e.g. rain-fed rice cultivation). However, because the definitions of the target environ- ments are generally rather broad, there is usually a range of different indi- vidual environments within any defined population. Traditionally, genotypes are evaluated using multi-environment trials (METs) to evaluate the performance of a genotype in a sample of environments from the tar- get population of environments. However, in many METs, there is no meas- urement of how well the sample environments match the target population of environments. Progress, therefore, is often slow because of the need to sample sufficient environments over sufficient years to be sure that any gain from selection is real. What is needed is a way of characterizing all of the different environ- ments within the target population with an index, so that similar environ- ments with the same index can be grouped together. Representative locations within environments with the same index could then be chosen, with trials established at these locations to evaluate the selected genotypes. Results obtained from trials at these ‘benchmark sites’ can then be extrap- olated with some degree of confidence to other similar environments with the same index. The number of METs that need to be carried out, there- fore, could be greatly reduced. Crop Genotype Improvement 19
- 39. The question then arises as to which indices are the most appropriate to characterize environments. Angus (1991) has reviewed the evolution of approaches to describe climatic variability, ranging from agroclimato- logical indices, simple water balances, through to crop simulation models. These methods vary in their input data requirements and in their complexity. Cooper and Fox (1996) distinguished between direct charac- terization, or characterization based on the measurement of environmen- tal variables such as water availability or the physical or nutrient status of the soil, and indirect characterization, based on measurement or estima- tion of plant responses in a particular environment. As plant breeders are interested in the way a particular genotype performs in different envi- ronments, indirect characterization, taking into account how plants perceive their environment, perhaps gives a more realistic index. One way of doing this that has had some success is to use probe genotypes, where a specific set of genotypes is selected based on their known reaction to an environmental factor encountered in the target population of environments (e.g. Cooper and Fox, 1996). The relative performance of the genotypes which comprise the probe set can then be used to judge the incidence of the environmental factor in METs. A second approach currently being explored is to use crop simulation models to pre- dict how a genotype with a particular set of characteristics will perform under different environments, and to characterize these environments in terms of that genotype’s performance. The model used in this way acts as a means of transforming raw meteorological data into a form that repre- sents the way a plant perceives its environment rather than just a purely physical description. As a comparison of the different approaches to environmental charac- terization, Muchow et al. (1996) calculated three indices of water deficit to characterize target environments at two locations in Australia for grain sorghum. The simplest index was based on rainfall and potential evapo- transpiration only, but poorly characterized the two environments. The sec- ond, a soil water deficit index based on a soil water balance and variable crop factor, and third, a relative transpiration index calculated using a sorghum simulation model, were both successful in identifying groups of seasons having distinct patterns. However, groupings based on the relative transpiration index from the crop model accounted for a higher proportion of the annual yield variation. Chapman et al. (2000b) developed this approach further to use the envi- ronment types (ETs) to replace the location × years (L ×Y) interaction term in the analysis of variance of trial results. Using data from 18 locations over 17 years, they used the sorghum model to generate drought stress patterns that were then grouped using pattern analysis into three environ- ment types: (i) low stress; (ii) severe end-of-season stress; and (iii) medium end-of-season stress. They found that these ETs had more consistent relationships with simulated yields than did categorization of locations and 20 R. Matthews
- 40. years by descriptors such as rainfall and latitude. The implication of these results for plant breeding programmes in the region was that random sam- pling of environments (the current approach) is unlikely to be the most effi- cient way of improving broad adaptation, and that selection of locations representing the three ETs would improve this efficiency. In a companion paper (Chapman et al., 2000a), the same authors argue that weighting geno- type performance by the relative proportions of the three ETs across all sites and all years would improve the precision of the broad adaptation value. Their results indicated that if simple averaging of yields to select genotypes had been employed over the last 80 years, hybrids with adap- tation to a higher frequency of drought environments than the long-term average would have been developed. In another example of this approach, Chapman and Bareto (1996) used a simple model to define the extent of adaptation environments for maize in Central America using phenology and drought tolerance as traits. Monthly minimum and maximum temperature data from 364 base stations in the region were interpolated spatially in a geographical information system (GIS), and then used to develop maps of flowering date and thermal time accumulated up to 70 days after sowing (DAS). A major limitation to the use of crop models to characterize environ- ments in this way, especially in developing countries, is the lack of input data both in spatial and temporal dimensions. This may be because of either poor spatial coverage (i.e. few stations with reliable long-term records) or due to the availability of only monthly mean data rather than the daily data required by the models. Interpolation methods within a GIS such as those used by Chapman and Bareto (1996) go some way towards addressing this problem, although the reliability of data between weather stations is often dependent on the method of interpolation used. However, the availability of agrometeorological data suitable for use with crop mod- els in developing countries is improving gradually all the time, and may not be such a limitation in the future. A second limitation to most of the above approaches for environmental characterization is the failure to take into account socio-economic aspects. Because crop production always takes place in a sociological context, attempts to change cropping practices or recommend certain types of genotypes used by farmers within the target environment may fail if this is ignored (e.g. Fujisaka, 1993). Information on the preference of farmers for such things as plant type, grain/stover ratios and quality of grain for cook- ing and eating would help to ensure that the goals of breeding programmes were consistent with farmer requirements. If possible, information of this nature needs to be represented spatially in a GIS and overlaid onto the biophysical characterization information. Crop simulation models could be used to predict plant type and grain/stover ratios, but most models to date do not incorporate aspects of quality such as taste or cooking characteristics. Crop Genotype Improvement 21
- 41. 3.3 G × × E Interactions As mentioned in the previous section, the traditional way of evaluating genotypes is through large numbers of METs. However, METs are gener- ally conducted for only a few years, and are unlikely to sample the full range of seasonal variability at a specific location, particularly where tem- poral variation is high such as at many locations in the semiarid tropics, for example. Current approaches used by plant breeders involve partition- ing the variation observed in such METs for a desired trait into that due to genotype (G), environment (i.e. location and season, E) and the interaction between genotype and environment (G ×E). The G ×E interaction term is then often treated as a source of error or bias in the analysis of genotypic variation, which has resulted in the theoretical framework on which selec- tion methods have developed being biased towards broad adaptation (Cooper and Hammer, 1996). Where the G ×E term is large, however, the usefulness of using genotype means across the sample environments as an index for selecting superior genotypes is reduced. Recognizing that mean yields across all of the sample environments may hide important differences in response, a number of statistical methods have been developed to analyse G ×E interactions. One approach is to characterize each sample environment by the mean yield of all genotypes grown in the trial at that site, and then use this mean as an index of pro- ductivity of the site (Finlay and Wilkinson, 1963). Yields of individual geno- types across all the environments are then regressed against their corresponding site indices, and the slope of the line is taken as the stabil- ity or responsiveness of the genotype. However, the approach is often crit- icized as the site index violates the assumptions of statistical independence, and the response of genotype performance is assumed to be linearly relat- ed to the site index, which may not always be the case. Moreover, it is difficult to relate the site index to specific environmental factors such as water deficit or temperature stress. New statistical tools to address some of these problems, notably the assumption of linearity between genotype per- formance and site index, are being developed to discriminate between genotypes and to explain G ×E interactions (DeLacy et al., 1996). These include Additive Main effects and Multiplicative Interaction (AMMI) mod- els, and pattern analysis. However, the problem remains of the time and cost of running sufficient METs to generate the data needed for the analysis. Aggarwal et al. (1996) proposed a strategy for increasing the efficiency of this process, by using limited MET data to estimate genotype interaction scores by AMMI analy- sis for all test genotypes on one hand, and to identify groups of genotypes with similar interactions via pattern analysis on the other. Representative genotypes for each group could then be identified and their performance simulated with a crop model over a wider range of target environments. The interaction scores for these new environments are then estimated from 22 R. Matthews
- 42. the simulated responses and combined with the genotype scores from the original MET to extrapolate G ×E interaction effects over the wider range of environments. They used the ORYZA1 rice simulation model to simulate the performance of 26 hypothetical genotypes, which were ‘created’ by random combinations of eight model parameters (leaf N content, fraction of stem reserves, leaf/stem ratio, relative growth rate of leaf area, specific leaf area, spikelet growth factor, and crop development rates before and after anthesis), in ten different environments. These were then grouped into six genotype groups (Fig. 3.1), from each of which one genotype was arbi- trarily selected as the reference genotype representing that group. Data for eight new environments were then generated using the same 26 genotypes. A highly significant positive correlation was obtained between the esti- mated and simulated interaction effects for the new sites, indicating the potential for this type of combination of statistical analysis and crop mod- elling to extend the range of G ×E interaction information. Crop Genotype Improvement 23 Fig. 3.1. Interaction bi-plot for the AMMI2 analysis of simulated G × E interaction data for 26 hypothetical genotypes simulated over ten environments. Lines join the environmental scores to the origin. IPCA scores are the multiplicative inter- action scores for genotypes in the AMMI2 model, boundaries encircle groups of genotypes with similar interaction (after Aggarwal et al., 1996).
- 43. A similar approach for sorghum in Australia was used by Hammer et al. (1996a), who investigated the effects of phenology, stay-green, transpira- tion efficiency, and tillering traits. As in the study by Aggarwal et al. (1996), they found that the partitioning of total variation between genotype, envi- ronment, and G ×E interactions produced by simulation (4, 75 and 15%, respectively) were similar to that observed in the field, and that most of the G ×E interaction variability was due to duration to maturity. In another study, Palanisamy et al. (1993) used a model based on the SUCROS fami- ly of models to predict the ranking of 11 genotypes in variety trials at three locations in India over 4 years. They successfully predicted the rankings of two of the top three genotypes, but concluded that the failure to do so with the other genotypes highlighted the need for further refinement of the methodology. In another study, Acosta-Gallegos and White (1995) used the BEANGRO model to examine the length of the growing season at three sites in the Mexican highlands for 10–18 years. For two sites, long growth seasons and an early onset of the season were associated with greater probability of adequate rainfall. At the other site, total rainfall was lower, and was uncor- related with the onset or length of the season. They proposed two types of cultivars – one with a growth cycle that becomes longer with early plant- ings for the first two sites, and a cultivar with a constant, short cycle for the third site. Similarly, Bidinger et al. (1996) used a simple crop model (Sinclair, 1986) to analyse G ×E interactions for pearl millet in terms of dif- ferences among genotypes in the capture of resources, the efficiency of their use, the pattern of partitioning to economic yield, and their drought resistance. Another way in which crop simulation models may be able to contribute in the area of G ×E interactions is to reduce the amount of unexplained variability in the G ×E term in the analysis of variance (ANOVA) of trial results. It has been recognized for some time that variation due to G ×E interactions is amenable to selection if the environmental basis could be understood (e.g. Comstock and Moll, 1963). This has led to the concept of repeatable and non-repeatable G ×E interactions (Baker, 1988), the repeatable part of which could be used as a basis for selection for specif- ic environments. Crop models offer a way of predicting quantitatively the repeatable portion of this variability. A number of studies have used crop models to understand G ×E interactions for various characteristics, but have not yet, to our knowledge, been used to partition the G ×E interaction vari- ance term to allow greater sensitivity of the analysis of variance. Much of the work to date has focused on G ×E interactions in relation to phenolo- gy, probably because this is one characteristic in which the natural varia- tion is greater than the resolution of most models. For example, Muchow et al. (1991) used a model to show that it was better to use a longer-matur- ing variety of sorghum than the standard cultivar (Dekalb E57+) at one location in Australia, while at another location, there was a 50% chance 24 R. Matthews
- 44. of yield loss from using either a shorter- or longer-maturing variety com- pared to the standard. An important point was that there is no clear advan- tage in all years of selecting a particular cultivar type. They were able to determine the probabilities of particular outcomes by using long-term weather with the crop model, which would have been time consuming and costly by a traditional MET approach. Following on from the studies just described, some interesting work is emerging from the Agricultural Production Systems Research Unit (APSRU) group on linking crop simulation models to models of plant breeding sys- tems as a way of understanding how the efficiency of plant breeding as a search strategy in a particular ‘gene-environment landscape’ could be improved. Chapman et al. (2002) describe an approach demonstrating how the flow of genes through breeding programmes can be investigated by incorporating assumptions about the links between individual alleles of genes and their corresponding phenotypic characteristics into a crop sim- ulation model. In their study, four phenotypic traits were investigated – transpiration efficiency, flowering time, osmotic adjustment and stay-green. Each trait was assumed to be controlled by two genes, each with two alleles, giving five evenly distributed levels of expression depending on the number of alleles with positive effects present in the genotype. The APSIM- SORG model was then used to predict the yield of a crop with a particu- lar genetic complement in a variable set of dry-land environments, grouped into three patterns of drought stress. Yields resulting from every possible combination of alleles were simulated in a large range of different envi- ronments (including different years and locations), and the results used as inputs into the QU-GENE model, which simulates different plant breeding systems (Cooper et al., 1999). They found that different plant breeding strategies resulted in the accumulation of favourable alleles at different rates (Fig. 3.2), and, interestingly, that complex epistasis (gene–gene interactions) and G ×E interactions emerged at the crop level, despite the assumption that the effects of different genes were simply additive. For example, alle- les for the stay-green characteristic were not fixed until those for early matu- rity had first been fixed. The order of selection of different traits is, therefore, important. Yin et al. (1999a) describe initial efforts to use crop models to improve the accuracy of analysis of quantitative trait loci (QTL). To identify QTL for SLA in 94 recombinant lines of barley, measurements of SLA were made at six times during the season (five of these were on the same date for all the lines, while one was at flowering, and hence differed between lines). Based on these measured data, between one and three QTL for SLA were found for all the sampling dates, and a dwarfing gene (denso) was found to strongly affect SLA. However, when a simple model based only on tem- perature was used to rescale the SLA measurements for direct comparison at the same development stage rather than chronological age, fewer QTL for SLA were found, and the presence of the denso gene did not affect SLA. Crop Genotype Improvement 25
- 45. The correlation in the first instance was found to be due to the gene’s effect on the duration of the pre-flowering period, rather than on SLA directly. The authors suggest that the use of crop models in this type of analysis could be useful when investigating traits with a dynamic nature, such as leaf N content, biomass partitioning fractions, WUE and RUE. An interesting suggestion is made by Hammer et al. (1996a) that anoth- er way in which models could help is to explore the interaction between management practice and genotype for different target environments, some- thing that is not often accommodated in crop improvement programmes. They make the point that this interaction can be as important in assessing the value of a genotype as interaction with the physical environment. To our knowledge, there are no examples of studies investigating this aspect. As already mentioned in Section 3.1, a major limitation of using current crop models in accounting for G ×E interactions is the resolution and accu- racy of the model in comparison to the subtle differences between geno- types commonly observed in many well-conducted multi-environment 26 R. Matthews Fig. 3.2. Mean changes in the gene frequency for + alleles associated with four physiological traits: transpiration efficiency (TE, average of five genes), phenology (Ph, three genes), osmotic adjustment (OA, two genes) and stay- green (SG, five genes) given four different selection environments (from Chapman et al., 2002).
- 46. trials. For yield, these differences may be in the order of 500 kg ha–1 or less, which is probably less than, or at least near to, the resolution of most crop models. This level of resolution is due both to uncertainties in the input data used by the model (Aggarwal, 1995), and to inaccuracies intro- duced by the structure of the model itself. It is this last factor that poses a dilemma in the application of crop mod- els to crop improvement programmes – the simulated predictions are an inevitable consequence of the assumptions made in modelling the trait; by their very definition models are simplifications of a complex reality. However, for models to be able to capture the small differences between genotypes, they must be sufficiently detailed to simulate the interactions of growth and development in a particular environment. The dilemma is what constitutes ‘sufficient detail’. One school of thought (e.g. Loomis, 1993) argues that more detailed models are required that are capable of simu- lating processes approaching the gene level. Some attempts have been made in this direction (e.g. Hoogenboom et al., 1997; Yin et al., 1999a, 2000). The criticism already discussed in Section 3.1 that most models do not adequately account for physiological linkages between traits (Lawn, 1994) would also support the argument for greater model sophistication. Often models do not incorporate these linkages because we do not have the knowledge of how they operate, and if they are included, their descrip- tion is usually empirical rather than mechanistic (Mutsaers and Wang, 1999). A contrasting point of view is that simpler crop physiological frameworks that are more readily aligned with plant breeders’ modes of action are required (e.g. Shorter et al., 1991). Hammer and Vanderlip (1989) were able to capture genotypic differences in RUE and phenology with a simple model, but such studies where simulation analyses of variation in a trait have been confirmed in the field are rare. Certainly, it seems logical that if crop models are to be incorporated into a crop improvement programme, it is essential that the parameters are easily and simply obtained, so that breeders can use them and apply them without substantial investment in time and data collection. Cooper and Hammer (1996) argue that crop physiologists have not generally appreciated this constraint faced by breed- ers, and have therefore not been able to adequately extend their often very relevant findings to ‘real life’ breeding programmes. However, it remains to be seen whether it is possible to resolve this dilemma of whether models of sufficient detail to discriminate between genotypes, yet requiring only limited input data, can be developed. The two approaches may not nec- essarily be mutually exclusive – Shorter et al. (1991) have suggested that the best way forward is to take a simple framework as the starting point, and add additional detail as necessary to describe the traits the plant breed- er is interested in. A danger of this approach, which needs to be guarded against, is that the resulting model may reflect the prejudices of the user, and contain only the components that he/she thinks are important. Crop Genotype Improvement 27
- 47. The other major limitation with current models is that not all of the traits that plant breeders are interested in are accounted for by the models (Hunt, 1993). Most crop models are designed to predict crop yield, but few crop improvement programmes focus on this characteristic only. Pest resistance and harvest quality, for example, are often of equal, if not greater, impor- tance, but are not generally included in crop models (White, 1998). Some attempts to take these characteristics into account have been made – Piper et al. (1993), for example, used the SOYGRO model to explore the influ- ence of temperature on oil and protein content in soybean. Similarly, recent advances in coupling pest models to crop models (e.g. Batchelor et al., 1993) should make it easier to assess the effects of pest damage on crops, although further development is obviously needed to take into account complex mechanisms of disease resistance such as increased lignification, or changes in tissue N content. While crop models have the potential to make an important contribu- tion to the crop improvement process, Hammer et al. (1996a) warn that there are many issues faced by plant breeders where modelling may be of limited value. Issues associated with pests and diseases and some soil physical and chemical factors cannot be readily incorporated into existing models owing to lack of knowledge on the complexity of interactions with the crop. They suggest that, in most cases, such issues are best dealt with in other ways. 28 R. Matthews
- 48. Crop Management Robin Matthews Institute of Water and Environment, Cranfield University, Silsoe, Bedfordshire MK45 4DT, UK As crop–soil simulation models are designed to predict crop-level respons- es, it is perhaps not surprising that a large proportion of the work described in the literature in which such models are used is in relation to various management options of a single crop. Much of this modelling work has focused on understanding the interactions between the various factors influencing crop growth and development, such as water and nutrient supply, biotic stresses, and the timing of planting and harvesting of the crop in relation to the prevailing environment. This has led on to using the models to find optimum management practices for these factors in par- ticular environments, generally with the purpose of maximizing yields. In this chapter, we look at examples where models have been used in this way. 4.1 Yield Gap Analysis Before any improvements to crop management practices are made, it is useful to know what the potential yield1 of the crop is in the region of interest, how large the gap is between this potential yield and yields actu- ally being obtained, and what factors are causing any discrepancy between potential and actual yields. Pinnschmidt et al. (1997) define yield gap as the difference between an attainable yield level and the actual yield. It is affected by various constraints and limitations, such as cultivar character- 4 1 Here we define potential yield as that yield determined by solar radiation, temperature, photoperiod, atmospheric CO2 concentration and genotype characteristics only. Water, nutri- ents, and pests and diseases are all assumed to be non-limiting. © 2002 CAB International. Crop–Soil Simulation Models (eds R. Matthews and W. Stephens) 29
- 49. istics, cropping practices, weather and soil conditions, and stresses due to pests, diseases and inadequate water supply. An analysis of the yield gap allows a quantification of the likely benefits to be gained by embarking on a programme to improve crop management, and identification of the factors that it is worthwhile concentrating research resources on. Crop models offer a way of estimating what the potential yield of a crop is, and a step-wise analysis of the various inputs can help identify the limiting factors. An example of such an application is provided by studies conducted at different sites in India, during evaluation of the groundnut model PNUTGRO (Boote et al., 1991; Singh et al., 1994). Using parameters for the standard cultivar Robut 33-1 (=Kadiri 3), the model predicted that poten- tial yield as determined by climatic factors alone was achieved at about one-third of the sites, but at many locations poor growth and low yields could not be attributed to weather conditions. It was concluded that other factors such as soil fertility and pests were causing a yield gap at some sites. Subsequent research focused on these problems. In a similar study in wheat, the WTGROWS model was used to predict potential wheat yields across India (Aggarwal and Kalra, 1994; Aggarwal et al., 1995), which were compared with the economic optimum yield and actual yields across a range of latitudes (Fig. 4.1). Results showed that yields increased with increasing latitude and at more inland sites, primarily because of variation in temperature. Average actual yields were less than 60% of potential yield – although actual wheat yields had increased con- siderably over the preceding 25 years to 3000 kg ha–1 , they concluded that the yield gap was still at least 2000 kg ha–1 . Further analysis suggested that about 35–40% of this gap was due to delayed sowing – most farmers are sowing later than the optimal planting date as rice/wheat systems are becoming more common. Rice matures in October/November, which is the optimal wheat planting date, and as rice is more profitable, farmers try to maximize its yield, which underlines the importance of taking the whole system into account in any analysis. Irrigation inefficiencies and variabili- ty in fertilizer use were other important factors limiting wheat yields. There is no evidence, however, that the findings of this work have been used by planners or to prioritize research, although it should be noted that WTGROWS itself is currently being used for yield forecasting (P.K. Aggarwal, New Delhi, 2000, personal communication). In another example, Pinnschmidt et al. (1997) collected data on crop and pest management practices, soil conditions, weather, crop perform- ance, and biotic and abiotic stresses from 600 plots in farmers’ rice-fields in the Philippines, Thailand and Vietnam. The CERES-RICE model was used to estimate potential and N-limited attainable yields, and a simple empir- ical approach was used to estimate yield trends based on fertilizer N and soil organic matter. The gaps between these predicted attainable yields and actual yields ranged from 35 to 55% in the different countries. In Thailand, it was shown that much of this was due to N limitations, resulting from 30 R. Matthews
- 50. low soil organic matter and low fertilizer inputs. Other factors such as pests and disease damage and water stress were important in the Philippines and Vietnam. This type of information can help in setting priorities in studying and managing yield-limiting factors, although again there is no evidence to date that it has been taken up and used by anyone (H. Pinnschmidt, Denmark, 2000, personal communication). Van Ranst and Vanmechelen (1995) developed three simple crop models to estimate potential yields, water-limited yields and yields limited by soil suitability, for the north-west region of the Cameroon. As a demonstration of the approach, these models were used within a geographical informa- tion system (GIS) framework, and maps were produced of the predicted yields at the three different production levels. However, the authors make the point that the lack of accurate environmental data for operation and validation of the crop models is a serious constraint that must be given urgent priority. Crop Management 31 Fig. 4.1. Potential, economic optimum and actual grain yield of wheat as a function of latitude. Also shown are the simulated yields on 15 December and 1 January sowings to illustrate the contribution of later sowing to the yield gap (from Aggarwal et al., 1995).
- 51. In Mali, public investment in irrigation schemes to try to capitalize on the expected high potential yields of rice due to high solar radiation and adequate water have not been as successful as hoped. In an attempt to identify why the expected yields were not being obtained, Dingkuhn and Sow (1997) used the ORYZA_S rice model to study the spatial, seasonal and year-to-year variability of potential rice yields in the region as a func- tion of planting date. Results indicated that potential yields are primarily driven by temperature, and that the major physiological determinants of yield were: (i) crop duration, which is very variable due to flood-water temperature; (ii) leaf area expansion, which is susceptible to chilling; and (iii) spikelet sterility due to heat or chilling. Yields varied from 4 to 10 t ha–1 . The results were used to propose environment-specific research foci. In a similar example, van Keulen (1975) was able to show, through simulation studies of growth in semiarid conditions in Israel and the Sahel, that in many years production was limited by nutrient deficiency rather than lack of water, as had been commonly thought. This insight led the way for subsequent comprehensive research projects on primary produc- tion in both of these regions. In Zambia, Wolf et al. (1989) used a model to simulate identification of the factors limiting maize yields for the main land-units, and found that rainfall was limiting only if it was less than 800 mm. At higher levels of rainfall, the main constraint to higher yields was nutrients, indicating that there would be a response to fertilizer. It is not known if the results from this work had any impact. An interesting use of a crop model to evaluate possible causes for change in crop yields over time in a given region is provided by Bell and Fischer (1994). Farmers’ yields of wheat in a region of Mexico had increased by nearly 60 kg ha–1 year–1 between 1978 and 1990 due to improved varieties, crop management and weather variation. The CERES-WHEAT model was used to predict potential yields in the region assuming no change in cultivar or management over the time period. The analysis showed that yields would have declined over this period because of increased temperatures, and that the true yield gain, attributed to improvements in genotype and crop man- agement, was in fact 103 kg ha–1 year–1 . However, despite these gains, aver- age farmers’ yields, having risen 50–75% over the period in question, were still considerably lower than the potential yields predicted by the model, indicating that there is still scope for improvement. The main impact of all of these studies has been to focus research activ- ities on the major factors limiting yield, although in some cases there is no evidence that this information has been used. It is difficult to quantify in monetary terms the value of such work, as this depends on the outputs of the downstream research. Nevertheless, it would seem logical that using models to identify limiting factors and prioritize research effort in these areas is a more efficient way forward than carrying out large-scale field experiments and finding out afterwards that the wrong factors were being investigated. Ways of disseminating this information to the relevant 32 R. Matthews
- 52. researchers, however, need to be improved substantially. It is tempting to suggest that if a sound modelling study had been carried out in Mali before rather than after public money had been spent on irrigation schemes, the results might have been more productive than was the case. 4.2 Soil Surface Management The condition the soil is in can have a major influence on the crop that is subsequently grown in it. It is, therefore, of interest to know what effect various soil management practices have on crop growth and yield. Freebairn et al. (1991) used the PERFECT model to simulate the effects of various management practices, such as crop/fallow sequences, tillage and addition of soil ameliorants, to modify different soil physical processes including infiltration, evaporation and erosion. They also used sequences of historical weather data to look at long-term decline (100+ years) in yields associated with soil erosion. Results showed that annual soil loss was much greater when previous crop stubble was removed. In another study, Singh et al. (1999) used a soybean–chickpea sequenc- ing model to extrapolate 2 years of experimental data investigating the effect of two land preparation techniques – broadbed-and-furrow (BBF) and flat – for two depths of soil in India. Using 22 years of historical weather data, the model simulations showed that in most years, BBF decreased runoff from the soil, but had a marginal effect on yields of soybean and chickpea, although these effects tended to be larger in dry years. The decreased runoff was associated with a concomitant increase in deep drainage from the BBF treatments. There is no record for either of these studies of any practical impact they might have had. 4.3 Planting In most environments, the time a crop is sown can have a major influence on its growth during the season, and therefore on its final performance. This is particularly the case in variable environments, or where there is a strong seasonal effect. In many tropical and subtropical regions, for exam- ple, planting decisions await the onset of a rainy season, and the available soil water reservoir is often only partially recharged over the dry season. In such cases, planting too early may result in poor establishment if the soil water status is insufficient, while planting too late may mean that the crop encounters drought stress towards the end of the season, the time in many crops when the economic yield is being determined. For example, Omer et al. (1988) used a crop model and 11 years of climatic data to determine the optimum planting period in the dry- Crop Management 33
- 53. land region of western Sudan, by generating probability distributions of a water-stress index resulting from different planting dates. The analysis showed a distinct optimum planting period of 20 June to 10 July, with planting in early July as the most likely for best production, which agreed well with general experience. In a similar study, Singels (1992, 1993) used the PUTU wheat growth model to determine optimal wheat-sowing strate- gies in South Africa using 50 years of historical weather data. Highest mean production was simulated when the entire available area was planted on the first possible date after 5 May. The starting date of the optimal sowing period identified by the simulations did not differ markedly from those rec- ommended by the South African Department of Agricultural Development, although the last date of the optimal period occurred earlier than those recommended. The analysis indicated that profit-maximizing, risk-averse producers should delay sowing until 9 June and then plant the total available area as soon as favourable sowing conditions occur. A similar conclusion was reached by Williams et al. (1999) for grain sorghum in Kansas – extremely risk-averse managers would generally choose some- what later sowing dates, earlier maturing hybrids and lower sowing rates than less risk-averse or risk-preferring managers. In Australia, Muchow et al. (1991) showed for sorghum that sowing later on a full-soil profile of water was always better than sowing earlier on a half-full profile. Similarly, Singh et al. (1993) and Thornton et al. (1995b) describe work using the CERES-MAIZE model, calibrated for local field conditions in Malawi, to determine the optimum planting window and planting density for a number of varieties currently grown there. In northern India, Aggarwal and Kalra (1994) used the WTGROWS model to show that a delay in plant- ing date decreased wheat yield, in part by subjecting the crop to warmer temperatures during grain filling. These results confirmed experimental data presented by Phadnawis and Saini (1992) for New Delhi. Hundal et al. (1999) used the CERES-RICE model to evaluate the age of seedlings at trans- planting, number of seedlings per hill and plant population for rice grow- ing in the Indian Punjab. Results showed that the optimum date of transplanting for rice was 15 June, but that earlier-transplanted (1 June) rice may perform better if seedling age is reduced from 40 to 30 days. Increasing plant population increased rice yields. Saseendran et al. (1998) also used CERES-RICE to determine the optimum transplanting date for rice in Kerala, southern India. Similarly, Hoogenboom et al. (2001) used CROPGRO-PEANUT to determine optimum planting date for groundnut in Andhra Pradesh, finding that later planting dates had a higher yield poten- tial than earlier planting dates. Farmers, however, prefer to plant early to avoid pest and disease damage prevalent with late planting, which the model does not currently simulate. In Bangladesh, Timsina et al. (2001) used the SWAGMAN Destiny model to evaluate the optimum time of planting of short-duration mung bean dur- ing the period between the main wheat and rice crops. During this time 34 R. Matthews
- 54. (March–June), the rainfall is somewhat erratic, leading to either water deficits or waterlogging, to both of which mung bean is susceptible. They found that where soils were undrained, a March planting was best, where- as for drained soils, planting in April was optimum. In Mozambique, Schouwenaars and Pelgrum (1990) used a crop model to simulate maize production over 28 years for different sowing strategies. They found that the maximum annual production depended almost com- pletely on losses caused by pests and diseases and postharvest losses. However, if the criterion was to minimize periods of food shortage, the preferred sowing strategy depended on water availability. In Australia, Clewett et al. (1991), while designing shallow-dam irriga- tion systems, considered two planting options – the first was to plant as soon as there was sufficient rain to ensure crop establishment, while the second was to delay planting until there was sufficient runoff to provide irrigation so that crop production could be assured. The first option was shown to have the higher long-term mean production, although this was accompanied by a much higher variability of production. Also in Australia, Muchow et al. (1994) assessed climatic risks relative to planting date deci- sions for sorghum growing in a range of soils in a subtropical rain-fed envi- ronment. Yield response was associated closely with differences in leaf area development and degree of depletion of the water resource brought about by differences in sowing date. It was suggested that decision makers could use the information taking into account their risk preferences, but no evi- dence is presented of this having happened. A general approach to generating the information required to assist in making sowing decisions in climatically variable subtropical environments is presented by Hammer and Muchow (1992). The approach involved cou- pling a sorghum growth simulation model to long-term sequences of cli- matic data to provide probabilistic estimates of yield for the range of decision options, such as sowing date and cultivar maturity, for a range of soil conditions. The likely change in the amount of stored soil water with delay in sowing was also simulated to account for the decision option of waiting for a subsequent sowing opportunity. The approach was applied to three locations in subtropical Queensland, Australia. Production risk varied with location, time of sowing, soil water storage and cultivar phenology. The probabilistic estimates presented of yield and change in stored soil water could assist decision makers with risky choices at sowing in subtropical environments. The density of planting is another characteristic that has been investi- gated with crop models. Much early work on determining optimum plant- ing density used static models (e.g. Stickler and Wearden, 1965) which related plant population density to overall yield and to its components, such as yield per plant. More recently, crop models have been used to develop and confirm these relationships for particular environments. Keating et al. (1988), for example, used the CERES-MAIZE model to examine Crop Management 35
- 55. Random documents with unrelated content Scribd suggests to you:
- 56. who had examined Lee Oswald? Dr. Hartogs. No, no. Mr. Liebeler. It was not? Dr. Hartogs. No, no. They didn't know. They called me because they call me very often to give some psychiatric explanations of murderers or something like that. They did not know, and I did not know for sure. Mr. Liebeler. At that time neither one of you were—— Dr. Hartogs. And they selected me. I mean it was a fantastic thing. Mr. Liebeler. It was purely coincidence? Dr. Hartogs. Coincidence that they selected me. Mr. Liebeler. So you made no reference at that time to the examination which you had made of Oswald? Dr. Hartogs. None at all. I didn't know. Mr. Liebeler. Dr. Hartogs, do you have in your possession a copy of the report which you made at the time you examined Oswald? Dr. Hartogs. No. Mr. Liebeler. Have you had any opportunity to examine a copy of that report since the assassination? Dr. Hartogs. No. Mr. Liebeler. So the recollection that you have given us as regards your diagnosis and your recommendations is strictly based on your own independent recollection, plus the reconstruction of your interview with Oswald from the seminar that you recall having given? Dr. Hartogs. Right. Mr. Liebeler. Do you remember anything else that particularly impressed you about Oswald? The FBI report indicates that you
- 57. were greatly impressed by the boy, who was only 13½ years old at the time, because he had extremely cold, steely eyes. Do you remember telling that to the agents? Dr. Hartogs. Yes, yes; that he was not emotional at all; he was in control of his emotions. He showed a cold, detached outer attitude. He talked about his situation, about himself in a, what should I say, nonparticipating fashion. I mean there was nothing emotional, affective about him, and this impressed me. That was the only thing which I remembered; yes. Mr. Liebeler. Now, you recall also that Oswald was a slender and pale-faced boy? Dr. Hartogs. Yes. Mr. Liebeler. Can you remember what particular thing it was about Oswald that made you conclude that he had this severe personality disturbance? What led you to this diagnosis? Dr. Hartogs. It was his suspiciousness against adults, as far as I recall, his exquisite sensitivity in dealing with others, their opinions on his behalf. That is as far as I recall it. Mr. Liebeler. Did you form an opinion as to his intellectual ability, his mental endowment? Dr. Hartogs. Yes; but that I don't recall for sure. It was at least average at that time. Mr. Liebeler. I want to mark "Exhibit 1" on the examination of Dr. Renatus Hartogs, April 16, 1964, in New York, a photostatic copy of a document entitled "Youth House Psychiatrist's Report," indicating a report on case No. 26996; date of admission, April 16, 1953, exactly 11 years ago; date of examination, May 1, 1953, with regard to a boy by the name of Lee Harvey Oswald. I have initialed a copy of this report for identification purposes, Doctor. Would you initial it here next to my initials. (Witness complies.)
- 58. (Photostatic copy of document entitled "Youth House Psychiatrist's Report" marked "Exhibit 1.") Mr. Liebeler. Would you read the report and tell us if that is the report that you prepared at that time? Dr. Hartogs. That is right, that is it. Interesting. Mr. Liebeler. Doctor, is your recollection refreshed after looking at the report that you made at that time? Dr. Hartogs. Yes, yes; that is the diagnosis, "personality pattern disturbance with schizoid features and passive-aggressive tendencies." Yes. Mr. Liebeler. On page 1, at the very beginning of the report, you wrote at that time, did you not, "This 13-year-old, well-built, well- nourished boy was remanded to Youth House for the first time on charge of truancy." Dr. Hartogs. Yes. Mr. Liebeler. On the last page of the report there is a section entitled "Summary for Probation Officer's Report," is there not? Dr. Hartogs. Yes. Mr. Liebeler. And you wrote there, about two or three sentences down, did you not, "We arrive therefore at the recommendation that he should be placed on probation under the condition that he seek help and guidance through contact with a child guidance clinic, where he should be treated preferably by a male psychiatrist who could substitute, to a certain degree at least, for the lack of father figure. At the same time, his mother should be urged to seek psychotherapeutic guidance through contact with a family agency. If this plan does not work out favorably and Lee cannot cooperate in this treatment plan on an outpatient basis, removal from the home and placement could be resorted to at a later date, but it is our definite impression that treatment on probation should be tried out before the stricter and therefore possibly more harmful placement approach is applied to the case of this boy?"
- 59. Dr. Hartogs. Yes. It contradicts my recollection. Mr. Liebeler. Yes. As you now read your report—and it is perfectly understandable that it is something that might not be remembered 11 years after the event; I have no recollection of what I was doing 11 years ago. Dr. Hartogs. I did not know that I made this ambiguous recommendation. Mr. Liebeler. As you read this report and reflect on this report and on the boy, Oswald, as he is revealed through it, do you think that possibly it may have been somebody else that was involved in the seminar or are you convinced that it was Oswald? Dr. Hartogs. No; that was Oswald. Mr. Liebeler. That was Oswald? Dr. Hartogs. Yes. Mr. Liebeler. It would not appear from this report that you found any indication in the character of Lee Oswald at that time that would indicate this possible violent outburst, is there? Dr. Hartogs. I didn't mention it in the report, and I wouldn't recall it now. Mr. Liebeler. If you would have found it, you would have mentioned it in the report? Dr. Hartogs. I would have mentioned it; yes. I just implied it with the diagnosis of passive-aggressive. It means that we are dealing here with a youngster who was hiding behind a seemingly passive, detached facade aggression hostility. I mean this is what I thought was quite clear. I did not say that he had assaultive or homicidal potential. Mr. Liebeler. And in fact, as we read through the report, there is no mention of the words "incipient schizophrenic" or "potentially dangerous" in the report.
- 60. Dr. Hartogs. No; I don't know where she has it from, but these are my words. I use it in other reports, but here it is not. Mr. Liebeler. "Passive-aggressive tendencies" are fairly common in occurrence, are they not amongst people? Dr. Hartogs. No; it is not so common. It is the least common of the three personality traits. It is either a passive-dependent child or an aggressive child, and there is a passive-aggressive child. The passive-aggressive one is the least common. Mr. Liebeler. Would you describe for us briefly what the passive- aggressive tendencies are, how do they manifest themselves, what do they indicate? Dr. Hartogs. They indicate a passive retiring surface facade, under which the child hides considerable hostility of various degrees. Mr. Liebeler. It would indicate to some extent a hiding of hostile tendencies toward others? Dr. Hartogs. Yes. But usually in a passive-aggressive individual the aggressiveness can be triggered off and provoked in stress situations or if he nourishes his hate and his hostility for considerable length of time so that the passive surface facade all of a sudden explodes, this can happen. I said here that his fantasy life turned around the topics of omnipotence and power. He said also that "I dislike everybody," which is quite interesting, I think, also pertinent. Mr. Liebeler. You indicated that his mother was interviewed by the Youth House social worker and is described as such-and-such. That would indicate, would it not, to you that you personally did not see the mother? Dr. Hartogs. That is right. I did not see the mother personally, but the information I have from the Youth House social worker's report. Mr. Liebeler. You indicated in the second sentence of the summary for the probation officer's report, "No finding of
- 61. neurological impairment or psychotic mental changes could be made," did you not? Dr. Hartogs. That is right. Mr. Liebeler. What do you mean when you say that "No finding of psychotic mental changes could be made"? Dr. Hartogs. This child was not suffering from delusions and hallucinations. Mr. Liebeler. Would you couple that with the concept of neurological impairment which indicated no brain damage or anything of that sort which would cause hallucinations or disturbance of the personality? Dr. Hartogs. Yes. Mr. Liebeler. Do you remember the circumstances of Oswald's home environment here in New York at the time he came? Dr. Hartogs. No. Mr. Liebeler. You have no recollection of that. If I were to tell you now that this boy came to New York with his mother, his father having died before he was born, to live with one of his older brothers, and that they lived with the brother here in Manhattan on 92d Street for a short time, after which friction developed, and they then moved to the Bronx, the mother worked all day, to support the child, in a department store here in New York or in Brooklyn, and the boy apparently found difficulty in his relations with others at school because he dressed differently, being from Texas, they lived apparently on the Grand Concourse, which has been described to us at that time as being a generally middle-class Jewish neighborhood, in which the boys did not dress in levis or quite so casually as Oswald did; that he was given some difficulty because of the fact that he did not speak the way the people did in New York, he spoke with a southern Texas accent and did not understand the patois of the city; assuming that those things were true, would that be a
- 62. partial explanation, do you think, of the way that he reacted to you during the interview as reflected in your report? Dr. Hartogs. No; I would not say. This was not the personality disturbance which was the result of the situation of changes or conditioning; this was more deeper going. A personality pattern disturbance is a disturbance which has been existing since early childhood and has continued to exist through the individual's life. It is not the result of recent conditioning. Mr. Liebeler. After reading your report, are you able to form an opinion or did you form an opinion at that time of what might have caused this particular personality pattern disturbance in this boy? Dr. Hartogs. I mentioned it, I think, in the report, the lack of a father figure, the lack of a real family life, neglect by self-involved mother. Yes; I think these are the three factors. Mr. Liebeler. After reviewing the report, do you have any other remarks that you think would be helpful to us in trying to understand what motivated this boy, assuming that he was the assassin of the President? Dr. Hartogs. No. Mr. Liebeler. That you haven't already talked about? Dr. Hartogs. No. Mr. Liebeler. I will ask the reporter to set forth the text of the report at the end of the deposition. I want to thank you very much for giving us the time that you have, and on behalf of the Commission we want to tell you that we appreciate it very much. Thanks very much, Doctor. Dr. Hartogs. Okay. "This 13 year old, well-built, well-nourished boy was remanded to Youth House for the first time on charge of truancy from school and of being beyond the control of his mother as far as school attendance is concerned. This is his first contact with the law.
- 63. "He is—tense, withdrawn and evasive boy who dislikes intensely talking about himself and his feelings. He likes the give the impression that he doesn't care about others and rather likes to keep himself so that he is not bothered and does not have to make the effort of communicating. It was difficult to penetrate the emotional wall behind which this boy hides—and he provided us with sufficient clues, permitting us to see intense anxiety, shyness, feelings of awkwardness and insecurity as the main reasons for his withdrawal tendencies and solitary habits. Lee told us: 'I don't want a friend and I don't like to talk to people.' He describes himself as stubborn and according to his own saying likes to say 'no.' Strongly resistive and negativistic features were thus noticed—but psychotic mental content was denied and no indication of psychotic mental changes was arrived at. "Lee is a youngster with superior mental endowment functioning presently on the bright normal range of mental efficiency. His abstract thinking capacity and his vocabulary are well developed. No retardation in school subjects could be found in spite of his truancy from school. Lee limits his interests to reading magazines and looking at the television all day long. He dislikes to play with others or to face the learning situation in school. On the other hand he claims that he is 'very poor' in all school subjects and would need remedial help. The discrepancy between the claims and his actual attainment level show the low degree of self-evaluation and self- esteem at which this boy has arrived presently, mainly due to feelings of general inadequacy and emotional discouragement. "Lee is the product of a broken home—as his father died before he was born. Two older brothers are presently in the United States Army—while the mother supports herself and Lee as an insurance broker. This occupation makes it impossible for her to provide adequate supervision of Lee and to make him attend school regularly. Lee is intensely dissatisfied with his present way of living, but feels that the only way in which he can avoid feeling too unhappy is to deny to himself competition with other children or expressing his needs and wants. Lee claims that he can get very
- 64. angry at his mother and occasionally has hit her, particularly when she returns home without having bought food for supper. On such occasions she leaves it to Lee to prepare some food with what he can find in the kitchen. He feels that his mother rejects him and really has never cared very much for him. He expressed the similar feeling with regard to his brothers who live pretty much on their own without showing any brotherly interest in him. Lee has vivid fantasy life, turning around the topics of omnipotence and power, through which he tries to compensate for his present shortcomings and frustrations. He did not enjoy being together with other children and when we asked him whether he prefers the company of boys to the one of girls—he answered—'I dislike everybody.' His occupational goal is to join the Army. His mother was interviewed by the Youth House social worker and is described by her as a 'defensive, rigid, self-involved and intellectually alert' woman who finds it exceedingly difficult to understand Lee's personality and his withdrawing behavior. She does not understand that Lee's withdrawal is a form of violent but silent protest against his neglect by her—and represents his reaction to a complete absence of any real family life. She seemed to be interested enough in the welfare of this boy to be willing to seek guidance and help as regards her own difficulties and her management of Lee. "Neurological examination remained essentially negative with the exception of slightly impaired hearing in the left ear, resulting from a mastoidectomy in 1946. History of convulsions and accidental injuries to the skull was denied. Family history is negative for mental disease. "Summary for Probation Officer's Report: "This 13-year-old, well-built boy, has superior mental resources and functions only slightly below his capacity level in spite of chronic truancy from school—which brought him into Youth House. No finding of neurological impairment or psychotic mental changes could be made. Lee has to be diagnosed as 'personality pattern disturbance with schizoid features and passive-aggressive
- 65. tendencies.' Lee has to be seen as an emotionally, quite disturbed youngster who suffers under the impact of really existing emotional isolation and deprivation; lack of affection, absence of family life and rejection by a self-involved and conflicted mother. Although Lee denies that he is in need of any other form of help other than 'remedial' one, we gained the definite impression that Lee can be reached through contact with an understanding and very patient psychotherapist and if he could be drawn at the same time into group psychotherapy. We arrive therefore at the recommendation that he should be placed on probation under the condition that he seek help and guidance through contact with a child guidance clinic, where he should be treated preferably by a male psychiatrist who could substitute, to a certain degree at least, for the lack of father figure. At the same time, his mother should be urged to seek psychotherapeutic guidance through contact with a family agency. If this plan does not work out favorably and Lee cannot cooperate in this treatment plan on an out-patient basis, removal from the home and placement could be resorted to at a later date, but it is our definite impression that treatment on probation should be tried out before the stricter and therefore possibly more harmful placement approach is applied to the case of this boy. The Big Brother movement could be undoubtedly of tremendous value in this case and Lee should be urged to join the organized group activities of his community, such as provided by the PAL or YMCA of his neighborhood."
- 66. TESTIMONY OF EVELYN GRACE STRICKMAN SIEGEL The testimony of Evelyn Grace Strickman Siegel was taken at 2:39 p.m., on April 17, 1964, at the U.S. Courthouse, Foley Square, New York, N.Y., by Mr. Wesley J. Liebeler, assistant counsel of the President's Commission. Evelyn Grace Strickman Siegel, having been first duly sworn, was examined and testified as follows: Mr. Liebeler. Mrs. Siegel, my name is Wesley J. Liebeler. I am a member of the legal staff of the President's Commission investigating the assassination of President Kennedy. Staff members have been authorized to take the testimony of witnesses by the Commission pursuant to authority granted to the Commission by Executive Order No. 11130, dated November 29, 1963, and Joint Resolution of Congress No. 137. Pursuant to the authority so granted to it, the Commission has promulgated certain rules governing the taking of testimony from witnesses, which provide, among other things, that each witness is entitled to 3 days' notice before he or she is required to give testimony. I know you didn't get 3 days' notice of this, but each witness also has the power to waive that notice, and I assume that you will be willing to waive that notice, and go ahead with the testimony since you are here. Is that correct? Mrs. Siegel. Yes. That's correct.
- 67. Mr. Liebeler. We want to advise you also that the rules provide that if you wish to have a copy of your transcript, you may have it at your own expense, at such time as the Commission releases the transcripts, releases the testimony, and that you are entitled to counsel if you wish. You don't have counsel here, and I assume you do not wish it. Mrs. Siegel. No. I do not wish it. Will I be advised when the transcripts are released? Mr. Liebeler. Yes. The Commission understands that you were working as a social worker in 1953 and 1954, at which time Lee Harvey Oswald and his mother lived here in New York City. Before we go into the details of that, I would like to have you state your full name for the record, if you would. Mrs. Siegel. Evelyn Grace Strickman Siegel. Mr. Liebeler. Where do you live? Mrs. Siegel. 1347 River Road, Teaneck. Mr. Liebeler. Where were you born? Mrs. Siegel. New York City. Mr. Liebeler. And am I correct in understanding that you did work in New York as a social worker? Mrs. Siegel. That's correct. Mr. Liebeler. When did you begin working as a social worker? Mrs. Siegel. In March of 1950. Mr. Liebeler. How long did you continue in that work? Mrs. Siegel. I'm still working as a social worker. Mr. Liebeler. In the city? Mrs. Siegel. Yes; on a part-time basis. Mr. Liebeler. Would you outline briefly for us your educational background?
- 68. Mrs. Siegel. A.B., Hunter College; M.S., Columbia University, School of Social Work. Mr. Liebeler. And in 1953, at the time that you did have contact with the Oswalds, you had been doing social work for about 3 years; is that correct? Mrs. Siegel. That's correct. Mr. Liebeler. For whom did you work as a social worker? Mrs. Siegel. Youth House. Mr. Liebeler. Are you still working for Youth House? Mrs. Siegel. No; I'm not. Mr. Liebeler. When did you begin working for Youth House and when did you terminate your employment with Youth House? Mrs. Siegel. I began working for them in January of 1952, and I left in August—well, I left Youth House for Girls, which is part of the same institution setup, in August of 1958. Mr. Liebeler. Would you describe for us briefly the nature of the Youth House as it existed in 1953? Mrs. Siegel. In what aspect? Mr. Liebeler. What kind of institution was it? What kind of people went there? What was done with them there? Will you tell me? Mrs. Siegel. It was a remand center for boys, delinquent boys who had gotten into trouble with the court and were remanded to Youth House for a brief period of diagnostic study. Upon their reappearance in court, so far as I understood it, those children who had been assigned for diagnostic study went back to court accompanied by a report from Youth House, which was given to the judge. Mr. Liebeler. What kind of a report was this? What was in it? What did it say?
- 69. Mrs. Siegel. A full-scale diagnostic study includes a social history taken by the social worker after one or several interviews with the boy and an interview with a parent, as well as an interview with the Youth House psychiatrist; that is, the boy was interviewed by the Youth House psychiatrist. All this material was then typed up and sent to court. Mr. Liebeler. Who was the Youth House psychiatrist? Mrs. Siegel. Dr. Renatus Hartogs. Mr. Liebeler. Did Dr. Hartogs personally interview each boy, or were there other psychiatrists who sometimes interviewed the boys and reported, do you know? Mrs. Siegel. First of all, let me say that not every boy was seen by a psychiatrist or a social worker. Also, the caseload was shared from time to time by other psychiatrists on the staff of Youth House, not by Dr. Hartogs alone. Mr. Liebeler. There was a report of the psychiatrist, then, a report of the social worker, and were there any other reports of any other workers, generally speaking, attached to the court report? Mrs. Siegel. Incorporated into the social worker's report was a report from those workers on the floor where the boy lived, the counselors, so to speak, brief reports as to his behavior and so on. Mr. Liebeler. Those would be given to the social workers; is that correct? Mrs. Siegel. That's right. Mr. Liebeler. And used as a basis for the social worker's report? Mrs. Siegel. Not as a basis for it but incorporated into it. Mr. Liebeler. So as a general proposition, the reports of people from the floor would be before the social worker when she prepared her report and would usually be reflected in the report of the social worker; is that correct?
- 70. Mrs. Siegel. That's correct. Mr. Liebeler. Do you have any recollection of any contact during the course of your work as a social worker for Youth House with Lee Harvey Oswald? Mrs. Siegel. After the President's assassination, the name meant nothing to me. As the biographies in the papers started to appear, and it was said that this boy was in Youth House in 1953, I believe it was, I had a vague stirring of memory, and I then said to my husband that somehow I have a mental picture of this youngster. At the time I attributed him not to me but to another worker. I somehow thought that he was assigned to another worker. But I had a picture of what he looked like, and the only reason that I think I remember him is that he was from Texas, and he was distinctive because he had an accent that was different from most of the children I saw, and he wore blue jeans, which most of our kids didn't wear in those days. And that was all I remembered about it. I remembered absolutely nothing about him at all. Mr. Liebeler. And your recollection of Lee Oswald is still the same as it was at that time? Mrs. Siegel. Sitting in the corner of my office, a slim, skinny little boy. Mr. Liebeler. That is to say, you have not been able to refresh your recollection? Mrs. Siegel. No. Mr. Liebeler. And improve it at all? Mrs. Siegel. No. Mr. Liebeler. Since the—— Mrs. Siegel. No. I must have seen between 400 and 450 boys a year in those days. I don't remember. Mr. Liebeler. Do you remember talking to his mother at all?
- 71. Mrs. Siegel. No; I do not. I don't even know if I saw her. I am terribly curious to see my report again. Mr. Liebeler. How long do you know Dr. Hartogs? Mrs. Siegel. Well, we were associated over a period of from 1952 to 1958—6 years. Mr. Liebeler. Have you seen him since that time? Mrs. Siegel. No; we don't see each other socially at all. Mr. Liebeler. And you haven't spoken to him? Mrs. Siegel. No; I haven't. Mr. Liebeler. About the Oswald case; is that right? Mrs. Siegel. No; I haven't seen him since I left Youth House. Mr. Liebeler. Do you have any recollection that from time to time the psychiatrist, Dr. Hartogs, would give seminars as a technique to instruct or provide examples to the social workers and perhaps the psychologists and other employees of Youth House? Mrs. Siegel. Well, I don't remember that Dr. Hartogs gave the seminars. We all participated in them, social workers and psychiatrists. I remember them vividly. I was a participant, myself. Mr. Liebeler. I didn't mean to characterize Dr. Hartogs' role as being the sole role. Mrs. Siegel. Oh, no. Mr. Liebeler. But there were seminars? Mrs. Siegel. Oh, there were seminars. Certainly. I misunderstood you. Yes; there were seminars which took place weekly. Mr. Liebeler. Do you have any recollection that Lee Oswald was the subject of one of these seminars? Mrs. Siegel. No; I do not.
- 72. Mr. Liebeler. Do you have any recollection of what the reason for Oswald's being remanded to Youth House was? Mrs. Siegel. I only read in the paper that it was truancy. Mr. Liebeler. And you have no independent recollection about it otherwise at all? Mrs. Siegel. No; I do not. Mr. Liebeler. I show you a photostatic copy of a document entitled "Youth House, Social Worker's Report," which is dated Bronx, May 7, 1953, referring to case No. 26996. This report indicates that the social worker involved was Evelyn Strickman, which would at that time have been you; is that correct? Mrs. Siegel. Yes. Mr. Liebeler. And still is? Mrs. Siegel. Yes. Mr. Liebeler. I hand you this document, and tell me if that is the report which you prepared in connection with your work with Lee Harvey Oswald. Are you able to state whether or not that is the report you prepared? Mrs. Siegel. This is indubitably mine. Mr. Liebeler. These reports were prepared shortly after your contact with the boy, with the mother, or prepared from notes that you made of the interview, were they not? Mrs. Siegel. Oh, yes; they were prepared probably during the time he was still at Youth House. Mr. Liebeler. The point being that the report would accurately reflect the interview that you had both with Lee Oswald and with his mother? Mrs. Siegel. As accurately as I could; yes. Mr. Liebeler. And it was prepared on or about the time that you conducted the interview, was it not?
- 73. Mrs. Siegel. Correct, yes; and shortly afterward. (Document marked "Exhibit 1.") Mr. Liebeler. I have marked the photostatic copy of the exhibit as Exhibit 1 to the deposition of Evelyn Strickman Siegel, April 17, 1964, and I have initialed it for purposes of identification. I would ask if you would initial it also so that we can make sure that we are talking about the same thing. (Witness complies.) Mr. Liebeler. I show you another report, which upon examination you will note contains much of the same material as is set forth in the Exhibit No. 1, and ask you if you recognize the sheaf of photostatic copies which I have just shown you and if you can tell me what they are. Mrs. Siegel. This is my report. Just a minute. This is what I dictated into the record before I pulled from it the essential material which should go into the report to the court. Mr. Liebeler. So that the photostatic document that I have just shown you was prepared before Exhibit No. 1, and closer in time to your actual contact with the boy and with the mother? Mrs. Siegel. This is correct. Mr. Liebeler. The one you have in your hand? Mrs. Siegel. Right. Mr. Liebeler. And from the document you hold in your hand you prepared Exhibit No. 1, which is the formal report which was submitted to the court along with the report of Dr. Hartogs and perhaps of other personnel; is that correct? Mrs. Siegel. This is correct. Mr. Liebeler. We will mark the document to which we have just been referring, which is captioned "Oswald, Lee Harvey—Charge: Truancy," and has "Youth House" written at the top of it, and which
- 74. consists of 7 pages, the last of which has the typewritten name "Evelyn Strickman" and the date 4-30-53, and bears your initials— does it not? Mrs. Siegel. Those are the initials of Marion Cohen, who was casework supervisor at Youth House at that time. That shows she read it. Mr. Liebeler. She read it also? Mrs. Siegel. Yes. Mr. Liebeler. And we will mark the document Exhibit No. 2. (Document marked "Exhibit 2.") Mrs. Siegel. Wait a minute. Let me just correct that. Marion would have written her own initials. That isn't my handwriting. I never made an "E" like that. I don't know who did that. Mr. Liebeler. You have no question, however, that this is the report prepared by you? Mrs. Siegel. No; I have absolutely no question. This is my dictation into the record. I know—that was Sadie Skolnick. That was the undersupervisor at the time. That is who that S.S. is. Mr. Liebeler. I have initialed Exhibit 2. So that we are sure we are talking about the same exhibit, would you initial it also, please? Mrs. Siegel. Sure. [Witness complies.] Mr. Liebeler. Exhibit 1 consists of six pages; is that correct? Mrs. Siegel. Yes. Mr. Liebeler. After reviewing the report which you prepared in connection with Lee Oswald back in 1953, is your recollection refreshed so that you could add anything other than that which is already set forth in the written report which you prepared at that time? Mrs. Siegel. No; I can't add a thing to that.
- 75. Mr. Liebeler. Would you say after reviewing the report that you prepared at that time that this boy gave any indication to you back in 1953, that is, as indicated in your report, that he had any violent tendencies or tendencies in this direction, in the direction of violence? Mrs. Siegel. Well, I can only say from what I wrote in that report that apparently this was a youngster who was teetering on the edge of serious emotional illness. Now, whether that included violence I am not prepared to say. Mr. Liebeler. You couldn't say that one way or the other from the material set forth in your report; is that correct? Mrs. Siegel. Yes; I would say that is correct. Mr. Liebeler. Can you think of anything else that you would like to add to the record after reviewing these reports that you think might be helpful to the Commission in its work? Mrs. Siegel. I am sorry, there is nothing I can add. Mr. Liebeler. I have no more questions. I want to thank you very much on behalf of the Commission. Mrs. Siegel. Not at all. It is a real tragedy. Mr. Liebeler. Thank you very much, Mrs. Siegel. Mrs. Siegel. Yes; not at all. Thank you. Goodbye.
- 76. TESTIMONY OF NELSON DELGADO The testimony of Nelson Delgado was taken on April 16, 1964, at the U.S. Courthouse, Foley Square, New York, N.Y., by Mr. Wesley J. Liebeler, assistant counsel of the President's Commission. Nelson Delgado, having been first duly sworn, was examined and testified as follows: Mr. Liebeler. My name is Wesley J. Liebeler. I am a member of the legal staff of the President's Commission investigating the assassination of President Kennedy. Staff members have been authorized to take the testimony of witnesses by the Commission pursuant to authority granted to the Commission by Executive Order No. 11130, dated November 29, 1963, and Joint Resolution of Congress No. 137. Under the Commission's rules for the taking of testimony, each witness is to be provided with a copy of the Executive order and of the joint resolution, and a copy of the rules that the Commission has adopted governing the taking of testimony from witnesses. The Commission will provide you copies of those documents. I cannot do it at this point because I do not have them with me, but we will provide you with copies of the documents to which I have referred. Under the Commission's rules for the taking of testimony, each witness is entitled to 3 days' notice before he is required to come in and give testimony. I don't think you had 3 days' notice. Mr. Delgado. No.
- 77. Mr. Liebeler. But each witness can waive that notice requirement if he wishes, and I assume that you would be willing to waive that notice requirement since you are here; is that correct? Mr. Delgado. Yes. Mr. Liebeler. We want to inquire of you this morning concerning the association that the Commission understands you had with Lee Harvey Oswald during the time that he was a member of the United States Marine Corps. The Commission has been advised that you also were a member of the United States Marine Corps and were stationed with Oswald in Santa Ana, Calif., for a period of time. Mr. Delgado. Yes. Mr. Liebeler. Before we get into the details of that, would you state your full name for the record, please? Mr. Delgado. Nelson Delgado. Mr. Liebeler. You are now in the United States Army; is that correct? Mr. Delgado. That is correct. Mr. Liebeler. What is your rank? Mr. Delgado. Specialist 4. Mr. Liebeler. What is your serial number? Mr. Delgado. RA282 53 799. Mr. Liebeler. Where are you stationed? Mr. Delgado. I am stationed at Delta Battery, 4th Missile Battalion, 71st Artillery, in Hazlet, N.J. Mr. Liebeler. How long have you been in the Army? Mr. Delgado. I joined the Army on November 1, 1960. Mr. Liebeler. What kind of work do you do in the Army?
- 78. Welcome to our website – the perfect destination for book lovers and knowledge seekers. We believe that every book holds a new world, offering opportunities for learning, discovery, and personal growth. That’s why we are dedicated to bringing you a diverse collection of books, ranging from classic literature and specialized publications to self-development guides and children's books. More than just a book-buying platform, we strive to be a bridge connecting you with timeless cultural and intellectual values. With an elegant, user-friendly interface and a smart search system, you can quickly find the books that best suit your interests. Additionally, our special promotions and home delivery services help you save time and fully enjoy the joy of reading. Join us on a journey of knowledge exploration, passion nurturing, and personal growth every day! ebookbell.com