Sustainability of rice in the global food system 9712201074, 9789712201073

Citation preview

SUSTAINABILITY OF RICE IN THE GLOBAL FOOD SYSTEM EDITED

BY

N. G. DOWLING S. M. GREENFIELD K. S. FISCHER

1998

Pacific Basin Study Center*

INTERNATIONAL RICE RESEARCH INSTITUTE *Now part of the East Asia Center on Population, Resources, and Welfare—EACOPRAW—at the University of California, Davis

The International Rice Research Institute (IRRI) was established in 1960 by the Ford and Rockefeller Foundations with the help and approval of the Government of the Philippines. Today IRRI is one of 16 nonprofit international research centers supported by the Consultative Group on International Agricultural Research (CGIAR). The CGIAR is cosponsored by the Food and Agriculture Organization of the United Nations (FAO), the International Bank for Reconstruction and Development (World Bank), the United Nations Development Programme (UNDP), and the United Nations Environment Programme (UNEP). Its membership comprises donor countries, international and regional organizations, and private foundations. As listed in its most recent Corporate Report, IRRI receives support, through the CGIAR, from a number of donors including UNDP, World Bank, European Union, Asian Development Bank, and Rockefeller Foundation, and the international aid agencies of the following governments: Australia, Belgium, Canada, People's Republic of China, Denmark, France, Germany, India, Indonesia, Islamic Republic of Iran, Japan, Republic of Korea, The Netherlands, Norway, Philippines, Spain, Sweden, Switzerland, United Kingdom, and United States. The responsibility for this publication rests with the International Rice Research Institute. The designations employed in the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of IRRI concerning the legal status of any country, territory, city, or area, or of its authorities. or the delimitation of its frontiers or boundaries. Copyright International Rice Research Institute 1998 Los Baños, Philippines Mailing address: P.O. Box 933, Manila 1099, Philippines Phone: (63-2) 845-0563, 844-3351 to 53 Fax: (63-2) 891-1292, 845-0606 Email:[email protected] Telex: (IT) 40890 Rice PM; (CWI) 14519 IRILB PS Cable: RICEFOUND MANILA Home page: http://www.cgiar.org/irri Riceweb: http://www.riceweb.org Riceworld: http://www.riceworld.org Courier address: Suite 1009, Pacific Bank Building 6776 Ayala Avenue, Makati Metro Manila, Philippines Tel. (63-2) 891-1236, 891-1174, 891-1258, 891-1303

Suggested citation: Dowling, NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute. 404 p.

EDITORS: Bill Hardy, Domenic Fuccillo DESIGNED& PRODUCEDBYTHE CPSCREATIVE & PRODUCTIONTEAM COVER DESIGN Juan Lazaro IV COVER PHOTOGRAPHY: Lingkod Sayo COVER CONCEPT AND ART DIRECTION: Albert Borrero DIGITAL IMAGING: Raul Ramiro Jr PAGE MAKEUP, ARTWORK, AND COMPOSITION: Erlie Putungan PRINT PRODUCTION MANAGEMENT Millet Magsino

ISBN 971-22-0107-4

Contents PREFACE V.W. Ruttan CHAPTER1 Introduction and overview S.M. Greenfield and N.G. Dowling

V

1

Part I: Food Security

CHAPTER2 Global food needs and resource limits J.G. Speth

11

Part II: Food Systems CHAPTER 3

Sustaining food security in Asia: economic, social, and political aspects M. Hossain

19

CHAPTER 4

A stable landscape? Social and cultural sustainability in Asian rice systems F. Bray CHAPTER5 Sustainability, food systems, and rice: exploring the interactions K.A. Dahlberg

45

67

Part III: Rice Production Systems: Challenges for Rice Research in Asia CHAPTER 6

Challenges for rice research in Asia K.S. Fischer

95

CHAPTER 7

Genetic enhancement of rice yields

99

S. Peng and D. Senadhira CHAPTER 8

Intensification of rice production systems: opportunities and limits W. Reichardt, A. Dobermann, and T. George

127

CHAPTER 9

Importance of rice pests and challenges to their management M.B. Cohen, S. Savary, N. Huang, O. Azzam, and S.K. Datta

145

CHAPTER 10

Weeds: a looming problem in modern rice production M. Olofsdotter, A. Watson, and C. Piggin

165

CHAPTER 11

Management of water as a scarce resource: issues and options in rice culture S.I. Bhuiyan, T.P. Tuong, and L.J. Wade

175

iii

CHAPTER 12 Securing the future of intensive rice systems: a knowledge-intensive resource management and technology approach L. M. L. Price and V. Balasubramanian

193

CHAPTER 13 Rice and the global environment R. Wassmann, T.B. Moya, and R.S. Lantin

205

CHAPTER 14 New Frontier Projects: beyond the pipeline J. Bennett, J.K. Ladha, V. Schmit, and J. Sheehy

225

Part IV: Rice Production Systems: Biodiversity CHAPTER 15

Protecting the diversity of tropical rice ecosystems K.S. Fischer

247

CHAPTER 16

Rice genetic resources M.R. Bellon, D.S. Brar, B.R. Lu, and J.L. Pham

251

CHAPTER 17

Biological diversity of rice landscapes K. Schoenly, T. W Mew, and W. Reichardt

285

Part V: Economic Considerations CHAPTER 18

The economic value of genetic improvement in rice R. E. Evenson

303

CHAPTER 19

Food, energy, and the environment: implications for Asia's rice agriculture V. Smil

321

Part VI: Case Studies CHAPTER 20

Rice production constraints in China Justin Yifu Lin

335

CHAPTER 21

Priorities and opportunities of rice production and consumption in India for self-sufficiency R.S. Paroda

357

CHAPTER 22

Conclusions: a potential research agenda S. M. Greenfield

391

APPENDIX 1. LIST OF PARTICIPANTS

396

ABOUT THE AUTHORS

399

iv

Preface

Rice is the primary food grain consumed by almost half of the world's population. During the past half century, rice has become available to consumers on increasingly favorable terms. Rice yields have risen more rapidly than demand arising out of population and income growth. These gains have resulted from the development of new and more productive rice varieties, increased intensity of fertilizer use, expanded irrigated area, improved crop protection, and the development and use of better management practices by agronomists and farmers. The success in generating rapid growth in rice yields, often referred to as the Green Revolution, has given rise to excessive complacency on the part of national governments and international aid agencies. While yields have continued to rise at the farmer level, maximum yield in trials at the International Rice Research Institute (IRRI) and at other leading rice research centers has remained static for almost two decades. Does this imply a new biological ceiling on rice yields that will limit them to the 8-10 t ha-1 now being achieved by the best farmers in the most favored ricegrowing areas? This concern has led to a new and broader rice research agenda, focusing on the new possibilities being opened up by advances in molecular biology and genetic engineering for plant breeding and crop protection. Researchers are working to develop knowledge-intensive farming systems, and attempting to ensure the conservation of rice germplasm diversity and to expand the use of underexploited relatives of cultivated species. This book represents the best single source of knowledge available on the state of efforts to develop the scientific and technical basis for a second Green Revolution-forthe advances necessary to sustain the increases in yield that have been achieved in the past and that will be needed to meet the demands that consumers will place on the world's rice farmers in the first half of the 21st century. V.W. RUTTAN

V

CHAPTER 1

Introduction and overview S.M. Greenfield and N.G. Dowling

Over the past decade, increasing concern has been expressed about “global sustainability” and “sustainable development.” Although sounding somewhat similar, these terms do not, at least initially, cover the same areas. Global sustainability involves the sustainability of the world’s institutions that are dedicated to meeting the needs of rising demands for goods and services in the face of changes in the natural environment. Sustainable development involves the ability to sustain the course of global development vis-à-vis the need to protect the environment, ecosystems, and the world’s population. Obviously, these concerns merge when the drive to sustainably develop any one sector reaches a point where an unacceptable restraint jeopardizes its continued viability. The importance of understanding and ultimately ensuring the sustainability of the world’s societal institutions in the face of increasing developmental pressure and natural changes is well recognized. Also recognized is the fact that information deficiencies and the complex nature of the problem severely limit our ability to intelligently address alternative strategies for avoiding a potentially deleterious future. This subject has been under discussion at the Pacific Basin Study Center for a number of years as a search was made for an approach that might help solve the problem. We realized that although the ultimate objective would be to address the question of global sustainability, reality—in the form of the recognized complexity of the subject, little understood interrelationships and feedback mechanisms, and inadequate databases—dictated the need to first scale the problem appropriately. Stated another way, sustainability involves the continued ability of our societal institutions to meet the current and future needs of their client populations. Achieving sustainability also requires that we meet these needs without compromising the ability of future generations to meet their own socioeconomic needs. When we attempt to address the question of global sustainability, we immediately realize the complexity of the problem, particularly if we are determined to deal with the entire mix of institutions that define human interactions with the planet. To avoid many of these complications, we chose to first focus on the single issue of agriculture and, in particular, the problem of sustainable rice production and distribution. This decision was based on the fact that most of the problems involved in striving to understand the global

1

sustainability macrocosm, particularly when confronting the uncertainty of potential global change, are present when we study the agricultural sector and food security. Further, the quasi-global characteristics of rice production and distribution provide a microcosm of the macro issues that we must address. Because of the relatively rich research base on rice, an examination of the crop, from production through consumption, should shed light on the important interacting roles played by technical, climatic, cultural, ecological, economic, social, religious. political, and geographical factors and their temporal and spatial variations. In addition. such an investigation, if structured correctly, should provide some insight into potential policy options for use globally and locally. This potential to provide insight into impact is clearly illustrated when we consider the current and projected situation for rice production and consumption worldwide. Rice provides about 40–45% of the calories consumed in the Pacific Basin, and as much as 70% in Vietnam, Cambodia, and Bangladesh. Rice is one of the world’s primary food crops; 90% of the world’s rice is grown in Asia, and almost all of it is consumed there—a third in China and a fifth in India. Without considering substitute foods, rice production must grow 60% by the year 2020 to keep pace with Asia’s increasing population (IRRI Toward 2000 and Beyond, International Rice Research Institute, 1989). The amount of land available for cultivated rice production is not increasing, partly because of the urbanization of the world’s population. It has been estimated that up to 85% of all arable land in Southeast Asia is currently under cultivation. Therefore, to meet the projected demand, a 3% increase in yield per hectare per year on the remaining arable land is needed. But recent studies indicate that such an increase is unsustainable: in fact, yields in many areas are declining, despite the use of chemical fertilizers and pesticides. New rice varieties, through genetic engineering, mechanization, and a better understanding and use of soil chemistry, may be useful for increasing yield. Other important considerations involve ideal farm size, economies of scale in differing regions, and the social dislocation that would result from any large change in existing systems. In addition, because of rapidly escalating populations in Asia and the region’s apparently limited ability to increase rice production to meet the growing demand, sustainability of this crop must clearly be considered, almost from the beginning, on a more global basis. In particular, the rice-producing capacity of the United States and its ability to adjust to an expanding demand can become a key factor in determining the security of this important food product. The complex nature of the problem described is clear. Any attempt to seek understanding and, ultimately, strategies to deal with such a problem must first address the need to integrate the input of the many disciplines involved in considering both technical and policy issues. This integration provides a way to achieve a common language and effective method of communication among participants, and promotes a strong inter- and intradisciplinary interaction. To date, the problem of sustainability at any institutional level has not been adequately structured and defined to the point where we might expect the required inte-

2

Greenfield and Dowling

gration and interaction to occur. But we are not implying that the problem is not recognized. The United Nations Food and Agriculture Organization (FAO) and the United Nations Development Programme (UNDP), for example, have shown a growing interest in global food security and have held numerous conferences on the subject. They organized an international food summit, held in Rome in November 1996. Although such activities are crucial to alerting and sensitizing decision makers to the magnitude of the problem, they must be supported by in-depth studies designed to provide the understanding and analytical tools needed to explore alternative strategies that could help mitigate the problem. The first steps must be ones that help to: • define the problem (in this case, sustainability of rice as a viable part of the food chain) and its complex aspects (i.e., technical and policy), • develop a set of realistic scenarios, • explore the ability to simulate these scenarios, and • determine the limits of our knowledge and information and hence the next steps that must be taken to remove these limitations. If we design an effective integrated approach to address the problem, then we will be well on our way to taking these four steps and, more importantly, the desired interaction and integration will have started and effective communication will be established. This international network of people can begin the process of sharing information and understanding that can ultimately lead to the development of effective strategies and policies to allow the world to cope with the problem of sustainability. What was obviously required to carry this out was a cost-effective process whereby interactions • would occur over a vastly extended time period, • would not require participants to participate continuously, but would allow them to leave and reenter the “discussion” and still have a sense of what occurred during their absence, • would ultimately provide easy and convenient access and involvement for policyand decision makers, and thus provide an ongoing forum that would serve as a resource and “sounding board” as ideas surfaced, conclusions were reached, global and local policies were developed and decisions formulated, and • would provide a semipermanent communication network of the disciplines involved that would allow the rapid exchange of information and ideas, and encourage collaborative efforts to address the issues involved. The approach adopted drew upon rapidly developing Internet/World Wide Web capabilities to establish a relatively permanent international electronic network that could grow and promote, among a broad set of participants, the desired interaction and exchange of ideas and information. Via this approach, papers were prepared, distributed, and commented on, and ideas were exchanged and discussed. In addition, participants believed that information could be exchanged; electronic conferences held; questions raised, discussed, and answered; simulations tried; and scenarios, strategies, and policies developed, explored, and placed before decision makers.

Introduction and overview

3

All the participants believed strongly that the use of such a network could promote an international common purpose and could allow all participants to express their concerns and have them considered within the context of the whole. This could help solve some policy problems before they became insurmountable. In essence, we can view this approach as a phased effort imbedded in an ongoing, interactive Web forum. The purpose of what was essentially the first phase of a long-term effort was to set up and begin the crucial dialogue among participants that would ultimately result in a working forum capable of addressing complex technical and policy issues. The objectives of this phase were as follows: 1. Demonstrate how rice sustainability represents a microcosm of global food security. 2. Place rice production within the context of the general food system (define). 3. Begin developing the understanding that underlies an ability to provide the advice required for decisions in response to perceived problems under a broad range of potential scenarios (optional choices, “societal costs,” problem avoidance vs. the search for “permanent” solutions, economic viability, etc.). 4. Explore where and how these elements come together in the pursuit of research and informational needs and policy options. 5. Begin the effort to determine what we must know to permit a timely analysis of sustainability in contrast to our current knowledge. The process through which these objectives were addressed is represented by a group of commissioned papers designed to stimulate discussion. These papers, carefully chosen to provide a spectrum of current thinking, were prepared under an NSF/ EPA (National Science Foundation/Environmental Protection Agency) grant. Abstracts were made available on a Web page specifically designed for this conference and made part of the Conference on the Web (COW) procedures with software developed and implemented by San Francisco State University. With the COW software, access to this Web page was limited to those invited to participate in the conference (see the list of participants in Appendix 1). All participants had the ability to download any of the full papers they desired to read and critique. Over a specified time period, the invited participants had the opportunity to submit comments or supplementary materials on any of the papers (or subjects represented by the papers) available through the conference Web page. Comments submitted through COW were available to all conference participants, who could thus make their own papers or data available to each other. This book contains all of the papers prepared for this first Web conference on the sustainability of rice production. As such, it represents not just a collection of the originally commissioned papers; it also contains the final version of these papers as modified by the comments and dialogue of the participants. This book could be viewed as the first volume of a series that begins by exploring the technical and socioeconomic aspects of a global problem, including a sense of what is known now and a

4

Greenfield and Dowling

research agenda designed to close the existing information gap. Whether subsequent volumes are prepared will depend on how seriously the problem is viewed by global decision makers. Whether or not these subsequent volumes are ever prepared, these same decision makers and people around the world will ultimately “write” the final volume. The overall objective of this book is to examine rice production from many aspects of the rich academic and policy base, not only from the side of production and economics but also from the involvement of generations of people of many cultures. Professor Vernon Ruttan, well known for his involvement in international agricultural policy, provides a preface and evaluation of the volume. James Gustave Speth, director general of the UNDP, which together with the FAO sponsored the United Nations Conference on Food Security in November 1996, explains the international concern about the growing world population and declining food supplies, especially in countries with the lowest average annual incomes. These include some countries in Asia and Africa that need attention to policies and research to provide a sustainable food supply. In the part of the book dealing with food systems, Hossain considers the important aspects of the food supply: (1) water, land, and labor scarcities, (2) the importance of rice in the diet, (3) the cost of growing rice and the impact of income production, and (4) the greater dependence of poor countries on rice. This chapter points out the need for improved farm management, mechanization, and help in developing nonfarm employment. The General Agreement on Tariffs and Trade (GATT) may pressure high-income countries out of rice production and favor poor countries such as Vietnam. Demand for rice in high-income countries will decrease, whereas in poor countries, where population will increase, demand for rice will increase (such as Vietnam and Pakistan, among others). Bray presents three cases—from late imperial China, contemporary Vietnam, and contemporary Japan—to illustrate the potential of small-scale wet-rice farming as a sustainable basis for a diversified rural economy. She concludes that planners must find ways to strengthen rural economies and increase both food output and the numbers of people to whom the local economy can provide a livelihood. Dahlberg examines the elements for maintaining a sustainable rice system and developing a larger framework and structures of regenerative food systems. The nature and structure of regenerative food and fiber systems are evaluated based on the health and regenerative capacities of biological and social systems. This framework is then applied to historical and current rice cultures to understand how rice fits into efforts to create more sustainable food systems. Some future research questions are raised. Parts III and IV, which deal with rice production systems, are a compilation of papers prepared by specialists in rice at the International Rice Research Institute. These papers were specially prepared for this book and they cover diverse scientific subjects. These papers present current and pending research on the sustainability of rice in the food supply over the next 30 years. Each paper suggests the most important questions that must be addressed in that field.

Introduction and overview

5

Part III contains nine chapters: Fischer explains the overall picture of rice in the Asian region. Peng and Senadhira review the genetic development of rice since the Green Revolution and analyze aspects of plant development that are being studied to improve the nutritional value of the plant. • Reichardt et al present soil nutrient conditions in both wet-rice and dry-rice regimens. Rice grown in flooded fields produces higher yields, but most of the world’s rice is rainfed. Because water is a scarce resource, some areas have potential for intensified crop production. • Cohen et al examine work in progress to quantify risk probability and risk magnitude of damage to crops from insects, plant diseases, and weeds. • Olofsdotter et al show the importance of developing integrated weed management systems in which several control measures are combined and herbicide use is minimized. • Bhuiyan et al analyze how water management and rice production systems can be improved to obtain more rice per unit of water supplied. • Price and Balasubramanian address the need for knowledge-intensive resource management encompassing “smart” equipment and increased information to farmers to improve production and environmental quality. • Wassmann et al examine the key effects of agricultural production on the environment, including the benefit of increased CO2 for rice yield and the effect of methane emission from rice. • Bennett et al discuss research strategies to enhance rice plants through N2 fixation, apomixis, and perenniality. They conclude with an overview of the challenges to achieving higher rice yield from the perspectives of systems analysis and mathematical modeling. Part IV addresses the use and conservation of biological diversity for agroecosystems, especially rice. It also describes the use and management of genetic diversity of the rice gene pool and the indigenous biota of rice landscapes. • Fischer analyzes the importance of biodiversity to the sustainability of rice. • Bellon et al investigate the threats and challenges that the conservation of rice genetic diversity faces from changing socioeconomic and cultural conditions, as well as from the development and widespread adoption of modern varieties. • Schoenly et al explain the rich biodiversity of microbial, floral, invertebrate, and vertebrate populations found in tropical rice fields. The challenge is to find the best ways to inventory, characterize, and assess such diversity and interconnected communities. Part V addresses economic considerations. • Evenson looks at the major features of the genetic improvement of rice and review studies that attempt to value rice genetic resources. He assesses the comparative role of genetic improvement and its prospects in the Second Green Revolution.

• •

6

Greenfield and Dowling

•

Smil analyzes the contradictions in rice agriculture, from the declining use of rice in richer Asian diets to the growing demand for rice from the increasing population. The environmental benefit of rice production for paddy biodiversity may outweigh the negative impacts of nitrogen and methane buildup. Part VI features two case studies. • Lin presents a country case study of rice yield under field conditions in China. He compares actual yield with experimental plot yields that indicate a higher yield potential, based on genetic differences as well as natural conditions (weather, soil, etc.). • Paroda examines the complex efforts in India to support the infrastructure for increasing rice yield and improving socioeconomics and policies for food security. India needs to increase rice productivity by 3% annually, by using technology, increasing genetic yield, and exploiting abundant untapped opportunities in the rice environment. The final chapter summarizes the research questions raised by the authors. Many of these questions are already being examined. A case is made for an integrated approach and understanding that could ultimately permit us to rationally address potential solutions to the problem of sustainability. Because of the ongoing, dynamic nature of this project, the editors would appreciate any constructive suggestions or comments from interested readers of this initial volume.

Notes Authors’addresses: San Francisco State University, San Francisco, California. [email protected] and [email protected] Acknowledgments: The framework of this effort was based on the work of a steering committee chaired by William Rains. Members were Richard Howitt, Peter Lindert, and Shu Geng of the University of California, Davis; Kenneth Fischer of the International Rice Research Institute (IRRI); Kenneth Dahlberg, Western Michigan University; and Stanley Greenfield and Noreen Dowling of the Pacific Basin Study Center. IRRI gave staff members time to prepare several papers presented here and its Communication and Publications Services helped edit and print this book. Ronald Miller persevered in elaborating the Web site and Donya Khalife helped prepare the papers for this publication. Funding for this effort was provided by the National Science Foundation in conjunction with the United States Environmental Protection Agency under NSF/EPA grant number 9602476. We also wish to thank the following reviewers for their contributions: Stephan Brush, Colin Carter, Shu Geng, Daniel Sumner, and Edward Taylor, University of California, Davis; J. Berkowitz, University of Connecticut; and Kenneth Cassman, University of Nebraska. Citation: Dowling NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute.

Introduction and overview

7

Part I: Food Security

CHAPTER 2

Global food needs and resource limits J.G. Speth

Rice, the staple food for the largest number of people today, is important for development. We have learned much about the potential of agricultural technology from work in rice, and it is in the area of rice that the most striking growth has taken place.

Food supply: growth and crises Breakthroughs in cereal technology have helped to increase grain production. The production of 10 major food crops in developing countries increased by 74% in the past two decades, with yield advances from technology contributing to 70% of this growth in output. Investments in agricultural research at the Consultative Group on International Agricultural Research (CGIAR) centers and in national research and extension systems have been instrumental in this increased productivity. Developing countries use high-yielding varieties on 74% of the area producing rice, 70% of wheat area, and 57% of maize area. Food grain availability has been increasing and, according to the International Food Policy Research Institute (IFPRI), 150 million fewer people go to bed hungry than 25 years ago, and an additional 1.5 billion people in developing countries are being fed with the incremental production. The availability of staple food has grown in all regions of the world except subSaharan Africa, where this remains a critical concern. There, population growth has outstripped growth in agricultural production. Food imports rose by 185% between 1974 and 1990, and food aid by 295%.

Food for sustainable human development Too often, agriculture and its development have been perceived in narrow food-supply terms alone. Technical specialists in agriculture can contribute to increasing production potential, output, and efficiency. But the social, institutional, and policy dimensions are critical in translating scientific knowledge into the reality of fuller lives. These dimensions have been much harder to address as effectively as have scientific developments. The focus of attention should shift from the world’s food needs to people’s food needs. The world’s food needs are primarily driven by market prices, and an abstract

11

unfulfillable goal, whereas people’s food needs are a more tangible fulfillment of needs for a healthy life. Sustainable food security is a fundamental aspect of sustainable human development. It fuses the goals of household food security and sustainable agriculture. A commitment to sustainable food security requires that we address not only increasing agricultural production but also income and land distribution, dietary needs, women’s status and opportunities, and the protection and regeneration of the resource base for food production. The recent World Food Summit held at the Food and Agriculture Organization of the United Nations (FAO) has clearly emphasized the more people- and environment-oriented approach to guide future investments.

Food strategy for sustainable human development: three policy areas We can identify three components of the sustainable human development approach, as it relates to food and agriculture: food and participatory development, food and environmental sustainability, and food and sustainable livelihoods.

Food and participatory development Broad-based economic growth that is equitable and anticipatory is central to eradicating poverty and meeting food needs. Agricultural development is an important instrument for this growth, but only insofar as it is accompanied by dynamic nonfarm economic growth. The United Nations Development Programme’s (UNDP) Poverty Strategy Initiative, through participatory methods, helps smallholders and communities identify and implement a range of actions to improve their livelihoods. Though the rural poor are immensely skilled in generating livelihoods under adverse conditions, they mostly operate with no improved inputs or information, low prices and distant markets for their produce, and virtually no institutional support for harnessing and managing natural resources. Migration to cities only serves to bloat cities and offers partial and temporary solutions to problems of rural poverty. Achieving sustainable livelihoods for the rural poor is thus crucial for balanced and sustainable economic growth. Partnership between local communities and development planning and programs can be enhanced by building and supporting local institutions that enable broad-based participation. For instance, the conventional top-down process of agricultural technology development and transfer has had some success. But when farmers are actively involved, and technology development and transfer take into account local needs and conditions, the results are far greater and the benefits more broadly shared. Institutional development is critical. Decentralization and democratic governance facilitate local participation. For small farmers to increase their productivity, they need access to information, services, improved technologies, and markets. Most existing institutions in developing countries fail to meet small-farmer needs. To meet these crucial needs, UNDP supports capacity building in a variety of ways, via inno-

12

Speth

vative approaches and pilot programs such as the Sustainable Agriculture Network and Extension Program, to build technical capacity and networks for field-based nongovernmental organizations (NGOs) and farmers’ organizations to link with national research and extension systems.

Food and environmental sustainability Broad-based participatory agricultural development that enables environmentally sound growth and innovative approaches to resource management are prerequisites to meeting the overlapping goals of the poverty and environmental agendas. We are now at a juncture where we may have to reinvent our approach to agricultural research and development. Supply-side policies alone may, in fact, be detrimental to both the environment and food security when resource-poor areas and people are marginalized. An appropriate response will start with a recognition and better documentation of the immense potential for resource cycling and conservation in agriculture for promoting food security and environmental benefits. Can ecosystem-wide impacts of agriculture on biological diversity be mitigated by changes in agricultural practices, technologies, and land use patterns? Will environmentally sound agricultural production practices necessarily mean a sacrifice in economic efficiency, yields, and output? The answers will require work in both the scientific and policy arenas. Biodiversity is an important aspect of environmental sustainability. By endangering biodiversity, our present habitation and agricultural practices threaten future productivity in some of the most fertile areas of this planet. One solution would be to promote the development of more diverse sets of improved varieties, coupled with major reforms in seed production policies to support localized seed farms and seed marketing systems. These are huge challenges to both crop scientists and policymakers, but not insurmountable ones. They have risen to the challenge before when global food supplies were threatened in the 1970s. Rice breeders and scientists have led the way in the past and can do so again. Although small farmers are well positioned to adopt labor-intensive agroecological production methods and to produce the high-value produce for which demand is increasing most rapidly, policy frameworks and rural services in many countries can often be biased against the poor and fail to address their real needs. This is particularly true for women farmers, who constitute a disproportionate share of the rural poor and whose incomes contribute most to family well-being, and yet who are denied equal access to development opportunities. The bulk of UNDP’s resources devoted to environmental activities goes to help countries protect and manage the natural resources that are essential to the basic needs of poor people in low-income countries. Four focus areas are sustainable agriculture and food security, water resources and the aquatic environment, renewable energy and energy conservation, and forest management. Other important areas for UNDP include its work on combating desertification and drought in all affected regions of the world.

Global food needs and resource limits

13

Building national capacity for an integrated approach to environmental and development objectives is a key goal for UNDP. In a recent program conducted jointly by the public sector and civil society organizations in Zambia, we supported training of about 18,000 women farmers to improve food legume production and storage for environmental sustainability. Legumes are an important source of both food value and income in Zambia. By reaching women with improved technological solutions, a breakthrough in sustainable food security could be achieved. UNDP also works closely with the Global Environment Facility to incorporate development and long-term food security goals in all environmental programming. In Ethiopia, we demonstrated the potential of farmer-based conservation of their rich plant genetic resources. Genetic diversity is critical for agriculture in marginal lands to survive, improve its productivity, and be environmentally friendly for long-term food security. A similar integration of environment and development has been demonstrated in an integrated coastal zone management program in Belize.

Food and sustainable livelihoods The emphasis on environmental sustainability cannot be separated from an emphasis on the livelihoods of people, especially for those who live in ecologically fragile areas. Environmental and human health concerns often intersect. One example is the development of agricultural systems that promote biological diversity and human nutritional needs. As soon as minimal calorie needs are met, the natural human response is to diversify the diet. Though this can be obvious in regions where demand for grains for human consumption has reached a plateau, a strong case can also be made for agricultural diversification in the newly emerging agricultural growth areas such as sub-Saharan Africa. Raising agricultural productivity is clearly necessary, as are improvements in agricultural technology on marginal lands, where some 500 million poor people live today. In other regions of the world, where agricultural production and the economy have been on a steady upward trend, the structure of demand is changing. Food needs continue to grow, but food demand will become more complex as incomes rise. Population and income growth will continue to increase demand for food, as will the success of poverty eradication efforts. In much of Asia, and particularly in most of the rice-producing countries of Southeast Asia, incremental demand will occur primarily in noncereal food groups. Cereals for human consumption will remain important, but diminishing at the margin. Other uses of cereals—such as for livestock feed and industrial and energy uses—are growing rapidly. There is a need for such uses, but agricultural policies need to distinguish where incremental investments would have the greatest benefit to the majority of the people, and particularly for the poor. The changing structure of demand offers a unique opportunity for investing in areas where land-intensive agriculture is not a feasible option, such as hillsides. Highvalue, labor-intensive crops, agroforestry, and other biomass-enriching options are

14

Speth

well suited to smallholder agriculture characterized by intensive agroecological care. This will require the careful identification of ecological and economic options, market and local institutional support for inputs, outputs, and technical information, and investment for rural income diversification. UNDP is currently developing a new initiative for dryland regions through the enhancement of natural resources and diversified livelihoods as a means of eradicating poverty and reversing the vicious circle of poverty and environmental degradation.

Conclusions Despite recent progress in the production of cereals and in the ability of the world to produce enough food to meet everyone’s food needs, the world food situation is beset by crises and by a considerable backlog of hunger and malnutrition. Today, some 840 million people are hungry or face food insecurity. Poverty-related hunger and malnutrition account for 1,700 deaths every hour, mainly children. Thirty percent of the global population lives in households too poor to obtain food for basic needs, and one child in three is underweight by the age of five. At the World Food Summit, UNDP emphasized that although food production needs to grow, the world already produces enough food to nourish everyone. UNDP also argued for a strong link between food security and poverty reduction. In partnership with FAO and the CGIAR, we aim to ensure that agriculture addresses not only better efficiency in raising output but also efficiency in eliminating poverty and in protecting the environment and natural resource base. This is the challenge of sustainable food security, which we, at UNDP, recognize as one of the greatest challenges facing us today.

Notes Author’s address: United Nations Development Programme. [email protected] Citation: Dowling NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute.

Global food needs and resource limits

15

Part II: Food Systems

CHAPTER 3

Sustaining food security in Asia: economic, social, and political aspects M. Hossain Will Asia be able to sustain favorable food balances and further improve food security for low-income households? This paper addresses these issues by assessing the impact of recent socioeconomic developments on the organization of the production of rice, the dominant staple food in Asia. It also analyzes the forces that influence the trends in demand and supply of rice and examines the political factors that could affect the trade-off between pursuing self-sufficiency in domestic production and achieving self-reliance through trade to sustain food security. The paper argues that the rapid increase in rural wages associated with growing economic prosperity, and changes in the tenancy market from sharecropping to fixed-rent tenancy and ownership cultivation, will put an upward pressure on the cost of rice cultivation in middle- and highincome countries and regions that have achieved a high level of productivity. The comparative advantage in rice cultivation will shift from irrigated to rainfed environments. The uncertainty in achieving food security through international trade because of the thin and volatile rice market would encourage middle- and high-income Asian countries to maintain a safe capacity of producing this staple grain through market interventions, although such action is not economically efficient.

Asia has an impressive record of feeding its ever-growing population despite limited land resources. The Green Revolution contributed to a growth in the production of staple grains at nearly 3% per year over the past three decades, keeping pace with population growth and the increase in per capita food consumption brought about by rising incomes and urbanization. A per capita income growth of 2-6% gave many food-deficit countries adequate purchasing power to meet shortages through commercial imports. Yet, despite improvements in food availability, many low-income countries still face food insecurity. Recent World Bank estimates indicate that about 1.1 billion people still live in poverty, and 840 million suffer from hunger, 70% of them in Asia (World Bank 1992, Bender and Smith 1997). Dramatic changes in Asia’s economic situation may affect demand-supply balances for staple grains. Middle- and high-income countries will experience a decline in per capita consumption of rice, the dominant food staple, because of food habit changes associated with rising income and urbanization. Population growth will remain a major force behind the substantial increase in total demand for staple grains

19

for the next 30-50 yr. Also, the demand for maize (corn) and other grains will increase substantially as the consumption of livestock products expands with further income growth. On the supply side, prosperous Asian countries will find it increasingly difficult to sustain producers’ interest in rice farming. The move toward free trade in agricultural production, initiated by the Uruguay Round of the General Agreement on Trade and Tariffs (GATT), will further dampen incentives for rice farming in these countries. The potential for increased productivity created by the dramatic technological breakthrough in the late 1960s has been almost fully exploited, particularly for the irrigated ecosystem. Without further technological advances, it will be difficult to maintain growth in rice production at historical rates. As rice production loses the race against population growth, sustaining food security becomes the major challenge for land-scarce, low-income countries. Affluent Asians could buy rice on the world market by offering higher prices, but the prospect of generating exportable surpluses outside Asia is limited. If the rice supply fails to increase with demand, the price will increase and the market will reallocate scarce supplies from low-income to high-income consumers, a shift that could aggravate poverty in low-income countries. Because poverty alleviation is a major political objective, governments in countries with food surpluses may raise trade barriers to protect their domestic consumers, a reaction that may induce high-income countries to continue their inefficient domestic production of rice. The question is, Will Asia be able to sustain favorable food balances and further improve food security for low-income households? This paper addresses these issues by assessing the impact of socioeconomic developments on the organization of rice production, analyzing the forces governing demand-supply balances, and examining political factors that could affect the trade-off between pursuing self-sufficiency in domestic production and achieving self-reliance through trade to sustain food security.

Rice: the dominant food staple and way of life in Asia Importance of rice in the economy and culture In Asia, rice is the principal staple food and the most important source of employment and income for rural people. Asia’s hot and humid climate during the long and heavy monsoon season, and the fertile land along the river basins of the major deltas that are regularly flooded, provide the most favorable agroecological environment for rice cultivation. The production of the other staple grain, wheat, raised in a rice-wheat sequence, is limited to the subhumid subtropics, in central China and in the foothills of the Himalayas in South Asia (Huke and Huke 1992). Most Asian nations depend on imports from outside the region for the supply of wheat, whose consumption is low but growing with rapid urbanization and changes in food habits. Maize is produced in sizable amounts in the sloping uplands of Indonesian outer islands, the Philippines, Thailand, and Vietnam; it rarely competes with rice for land resources. With the fast increase in the demand for livestock products following rapid

20

Hossain

Table 1. Level of food consumption (kg capita -1 yr 1 ) in selected Asian countries, 1992.

Country Philippines Bangladesh India Myanmar China Vietnam Indonesia Thailand Malaysia South Korea Japan

Population (million) 64 113 884 44 1,184 70 189 57 19 45 124

Rice (unhusked) 133 220 108 301 141 226 209 200 143 157 93

Wheat 28 20 56 3 83 4 14 9 36 45 41

Maize 20

–

10 2 27 6 25 1 2 18 21

Total cereals

Meat

Fish

181 241 201 309 256 236 248 210 182 225 157

19 3 4 7 30 15 8 21 50 30 39

32 8 4 15 10 14 14 25 24 58 75

Source: FAO Agrostat database, 1994.

economic growth, the importance of maize as a source of human nutrition dwindles, as it is being increasingly used as livestock feed. Among the cereals, the demand for maize has been increasing at the fastest rate. Table 1 illustrates the overwhelming importance of rice in the Asian diet. More than 250 million farm households in Asia depend on rice for their livelihood. A typical farm household grows rice along with many other subsistence crops in rice-based farming systems. Farms specializing in the production of a single crop are rarely found, except for plantations where a few perennial crops are grown and in regions where land distribution is highly unequal, such as in the Philippines. More than half of the rice produced is consumed by members of farm households. The marketable surplus varies depending on farm size and rice-growing environment (ecosystem). The surplus for the urban population and the rural landless occurs mostly on irrigated land (nearly 70% of total production) on farms with holdings of more than 2 ha. Rice farms in the upland and rainfed lowland ecosystems are mostly subsistenceoriented. A number of in-depth village studies conducted by IRRI, in collaboration with policy research institutions in national systems (David and Otsuka 1994, David et al 1994, Sudaryanto and Kasryno 1994, Isvilanonda and Wattanutchariya 1994, Hossain et al 1994, Upadhyaya and Thapa 1994, Ramasamy et al 1994, Yifu Lin 1994), estimated the average farm household income at US$1,000 per year, of which 36-57% came from rice cultivation (Table 2). A large portion of the off-farm and nonfarm income came from providing wage labor in rice farming, processing, trade, and transport of agricultural products and inputs. Because rice plays such an important role in the lives of its producers and consumers, it is little wonder that it occupies such an important position in Asian culture (Huggan 1995). Rice is mentioned in all the scriptures of the ancient civilizations of Asia. Its cultivation was considered as the basis of the social order and occupied a major place in Asia’s religions and customs. The Emperor of Japan is the living embodiment of the God of the Ripened Rice Plant. In Balinese (Indonesia) myth, Lord

Sustaining food security in Asia: economic, social, and political aspects

21

Table 2. Average farm household income (US$ yr -1 ) by source in selected Asian countries, 1985-88.

Country

Bangladesh China India (T. Nadu) Indonesia (Lampung) Nepal Philippines Thailand

Total household income 977 871 1,010 721 1,105 1,072 1,763

Sources of income (%) Rice

Nonrice

Nonfarm

38 43 52 36 43 57 49

30 30 36 44 46 18 20

32 27 12 20 11 25 31

Source: Compiled from unpublished data collected from household surveys under the collaborative IRRI-NARS project on the Differential Impact of Modern Rice Technology in Favorable and Unfavorable Production Environments. For country case studies on the production, organization, and impact of modern rice technology, see David and Otsuka 1994.

Vishnu created rice and God Indra taught mankind how to raise it. China has a saying that “the most precious things are not jade and pearls, but the five grains,” of which rice is the first. Death is symbolized in Taiwan by chopsticks stuck into a mound of rice. Japanese did not use the terms breakfast, lunch, and dinner; the three meals were asa gohan (morning rice), hiru gohan (afternoon rice), and ban gohan (evening rice). In China and Bangladesh, a polite way to greet a visitor is to ask, “Have you eaten your rice today?” Even the names of automobile giants Toyota and Honda have their roots in the rice paddies. The characters for Toyota (originally Toyoda) mean “bountiful rice field” and Honda “main rice field.” Debts, taxes, rent, and wage payments to agricultural laborers and rural artisans are still sometimes paid in rice.

Social organization of production Rice is cultivated on a small scale in fragmented landholdings. The average size of a farm ranges from 0.43 ha in China to 0.8 to 1.5 ha in most other countries (Table 3). A1-ha farm is often divided into a large number of parcels. Only in Thailand, Myanmar, and northwestern and southern India are farm holdings larger, around 3.5 ha. Farm size varies with population density and land productivity. Regions with fertile land and a developed irrigation infrastructure generally have small farms. A high incidence of rural-rural migration redistributes people from low to high productive areas. The adjustment of population pressure on land across regions within a country is limited only in countries where land reform laws prohibit the transfer of cultivation and ownership rights, such as in the Philippines and India (Otsuka 1991). Rice cultivation is highly labor-intensive. In low-income countries with a labor surplus, all farm operations are done manually and use more than 150 d of labor for each ha during a crop season. Transplanting seedlings and controlling weeds alone require 80 d of labor ha -1 (Sidhu and Baanante 1984). This work is done mostly by women (Paris 1996). A high degree of seasonality in farm operations, which depend on the rainfall pattern, requires the use of hired labor even on very small farms. Tra-

22

Hossain

Table 3. Size and structure of operational holdings and tenancy in selected Asian countries, 1985-88.

Country

Bangladesh China India (T. Nadu) Indonesia (Lampung) Nepal Philippines Thailand

Year of survey

1987 1988 1987 1987 1987 1985 1987

Average farm size (ha) 0.87 0.43 3.54 1.60 1.95 1.58 3.52

Distribution of operational holdings (households, %) 1 m) P deficiency Zn deficiency Fe deficiency Salinity Fe toxicity AI toxicity Organic acids and H,S toxicity

2

3

4

+ +

+ + +

+ +

5

6

7

8

9

+ +

+

+

+ +

+ + +

+

+

+

+ +

+ +

+ + +

+ + +

10

11

12

+

+ +

+

+

+ +

+

+ +

+ +

+ + +

a 1 = Terai region, Nepal; 2 = Cuttack, India; 3 = Khon Kaen, Thailand: 4 = Lopez, Philippines: 5 = Gampaha, Sri

Lanka; 6 = Claveria, Philippines; 7 = Sitiung, Indonesia, 8 = Pusa. India; 9 = Bangsang, Thailand: 10 = San Jose, Philippines; 11 = Castuli, Philippines; 12 = Unit Tatas, Indonesla.

Role of biotechnology in rice yield improvement Genetic engineering techniques offer new opportunities for accelerating breeding progress, increasing selection efficiency, and transferring genes across species and genetic barriers. The challenge for plant breeders is to capitalize on these novel techniques. Tissue culture. Tissue culture allows somaclonal variation and in vitro selection, thereby shortening the breeding cycle. Somaclonal variation was used to overcome some difficult problems of breeding for salinity tolerance in rice (Senadhira et al 1994). Hagonoy, the salt-tolerant improved rice cultivar released in the Philippines, was produced by F1 anther culture, a technique used extensively at IRRI in its NPT breeding program. Most rainfed rice grown in Asia is photoperiod-sensitive. Because it is cultivated only once a year, developing a new variety takes about 10 yr. This period could be reduced to 3–4 yr with anther culture. DNA probes. With conventional techniques, breeders have to rely on the phenotypic expression of genes. Selection efficiency is substantially reduced by this expression when interactions occur. Furthermore, approaches such as gene pyramiding for enhanced adaptability are impossible to undertake with conventional methods. With molecular biology techniques, breeders can detect alleles of interest in their materials by using DNA probes and by chemical or immunological assays. These techniques have numerous advantages. The tests have unlimited capacity and are nondestructive, rapid, and reliable (as high as 100%). The biggest advantage is their ability to detect in one screening the presence or absence of any number of alleles of interest. Most of the problems described in earlier sections could be overcome by

118

Peng and Senadhira

these marker-aided selection (MAS) techniques. A prerequisite to developing MAS techniques for a trait is precise phenotyping, which, in turn, demands an understanding of the physiological mechanism of the trait and its inheritance pattern. For drought and most soil-related stresses, these prerequisites are still lacking. Development of MAS techniques for tolerance of flooding and salinity is in progress at IRRI. Gene transformation. Transformation will open new opportunities to solve old problems. One good example is the control of stem borer and sheath blight. There are no known sources of resistance to this insect and this disease and chemical control is costly and unacceptable. Transformation with genes producing insecticidal proteins such as endotoxins of Bacillus thuringiensis (Bt ) and tripsin inhibitors should reduce stem borer damage. Similarly, the chitinase-producing gene can suppress the sheath blight pathogen. Floating-rice cultivars, when transformed with the Bt gene, could substantially increase the yields of very deeply flooded rice lands. Apomixis, if transferred to rice from other species, will revolutionize hybrid rice cultivation. Biotechnology tools, especially MAS techniques, will certainly provide solutions to most problems associated with breeding improved rice with tolerance for abiotic stresses. We urgently need to intensify research on the genetics and physiological mechanisms of tolerance traits. Priority should be given to drought, P and Zn efficiency, and iron toxicity tolerance. Cooperation among breeders, geneticists, stress physiologists, plant nutritionists, and biotechnologists is vital to produce the rainfed rice cultivars that we need for the future.

References Ahn JK. 1986. Physiological factors affecting grain filling in rice. PhD dissertation. University of the Philippines Los Baños, Laguna, Philippines. Akbar M, Khush GS, HilleRisLambers D. 1985. Genetics of salt tolerance in rice. In: Rice genetics. Manila (Philippines): International Rice Research Institute. p 399-409. Akita S. 1988. Physiological bases of heterosis in rice. In: Hybrid rice. Manila (Philippines): International Rice Research Institute. p 67-77. Akita S. 1989. Improving yield potential in tropical rice. In: Progress in irrigated rice research. Manila (Philippines): International Rice Research Institute. p 41-73. Ashraf M, Akbar M, Salim M. 1994. Genetic improvement in physiological traits of rice yield. In: Slafer GA, editor. Genetic improvement of field crops. New York (USA): Marcel Dekker, Inc. p 413-455. Austin RB. 1993. Crop photosynthesis: can we improve on nature? In: International crop science I. Madison, Wis. (USA): Crop Science Society of America. p 697-701. Austin RB, Bingham J, Blackwell RD, Evans LT, Ford MA, Morgan CL, Taylor M. 1980. Genetic improvements in winter wheat yields since 1900 and associated physiological changes. J. Agric. Sci. (Cambridge) 94:675-689. Boje-Klein G. 1986. Problem soils as potential areas for adverse soil-tolerant rice varieties in South and Southeast Asia. IRRI Research Paper Series 119. Chmg TT, Armenta-Soto JL, Mao CX, Peins R, Loresto GC. 1986. Genetic studies on the components of drought resistance in rice ( Oryza sativa L.). In: Rice genetics. Manila (Philippines): International Rice Research Institute. p 387-398.

Genetic enhancement of rice yields

119

Chang TT. 1964. Varietal differences in lodging resistance. Int. Rice Comm. Newsl. 13(4): 1-11. Chang TT, Loresto GC. 1985. Lodging in rice. Rice Production and Extension, Management Specialist Training Course, September 1985. Manila (Philippines): International Rice Research Institute. Chang TT, Vergara BS. 1972. Ecological and genetic information on adaptability and yielding ability in tropical varieties. In: Rice breeding. Manila (Philippines): International Rice Research Institute. p 431-453. Chaubey CN, Senadhira D, Gregorio GB. 1994. Genetic analysis of tolerance for phosphorus deficiency in rice (Oryza sativa L.). Theor. Appl. Genet. 89:313-317. Cho DY, Ponnamperuma FN. 1971. Influence of soil temperature on the chemical kinetics of flooded soils and the growth of rice. Soil Sci. 112(3):184-194. Coronel P, Akita S, Yoshida S. 1990. Aluminum toxicity tolerance in rice (Oryza sativa L.) seedlings. In: van Bensichem, editor. Plant nutrition: physiology and applications. Kluwer Academic Publishers. Dai Q, Peng S, Chavez AQ, Vergara BS. 1995. Effects of UVB radiation on stomatal density and opening in rice (Oryza sativa L.). Ann. Bot. 76:65-70. Dingkuhn M, Penning de Vries FWT, De Datta SK, van Laar HH. 1991. Concepts for a new plant type for direct seeded flooded tropical rice. In: Direct seeded flooded rice in the tropics. Manila (Philippines): International Rice Research Institute. p 17-38. Donald CM. 1968. The breeding of crop ideotypes. Euphytica 17:385-403. Ekanayake IJ, Garrity DP, Masajo TM, O’Toole JC. 1985. Root pulling resistance in rice: inheritance and association with drought tolerance. Euphytica 34:905-913. Elmore CD. 1980. The paradox of no correlation between leaf photosynthetic rates and crop yields. In: Hesketh JD, Jones JW, editors. Predicting photosynthesis for ecosystem models. Vol. II. Boca Raton (Fla., USA): CRC Press. p 155-167. Emes MJ, Wilkin CP, Smith P, Kupkanchanakul K, Hawker K, Charlton WA, Cutter EG. 1988. Starch utilization by deepwater rice during submergence. In: The 1987 international deepwater rice workshop. Manila (Philippines): International Rice Research Institute. Evans LT. 1990. Raising the ceiling to yield: the key role of synergisms between agronomy and plant breeding. In: Muralidharan K, Siddiq EA, editors. New frontiers in rice research. Hyderabad (India): Directorate of Rice Research. p 103-107. Evenson RE. 1994. Varietal yield and trait value estimates: rice in India. In: Proceedings of the international workshop on rice research prioritization. Manila (Philippines) and New York (USA): International Rice Research Institute and Rockefeller Foundation. Fischer RA. 1985. Number of kernels in wheat crops and the influence of solar radiation and temperature. J. Agric. Sci. 100:447-461. Fischer RA. 1994. Key issues in wheat yield potential. In: Cassman KG, editor. Breaking the yield barrier. Manila (Philippines): International Rice Research Institute. Fukai S, Cooper M. 1994. Stress physiology in relation to breeding for drought tolerance: a case study of rice. In: Rainfed lowland rice: physiology of stress tolerance. Manila (Philippines): International Rice Research Institute. Gent MPN. 1995. Canopy light interception, gas exchange, and biomass in reduced height isolines of winter wheat. Crop Sci. 35:1636-1642. Gifford RM, Thorne JH, Hitz WD, Gianquinta RT. 1984. Crop productivity and photoassimilate partitioning. Science 225:801-808.

120

Peng and Senadhira

Gregorio GB, Senadhira D. 1993. Genetic analysis of salinity tolerance in rice ( Oryza sativa L.). Theor. Appl. Genet. 86:333-338. Hamblin J. 1993. The ideotype concept: useful or outdated? In: International crop science I. Madison, Wis. (USA): Crop Science Society of America. Hitaka H. 1969. Studies on the lodging of rice plants. Jpn. Agric. Res. Q. 4(3):1-6. Horton P, Ruban AV. 1992. Regulation of photosystem II. Photosynth. Res. 34:375-385. Howeler RH, Cadavid F. 1976. Screening of rice cultivars for tolerance to Al toxicity in nutrient solution as compared with field screening method. Agron. J. 68:551-555. Huang H. 1988. Japonica and indica differences in large vascular bundles in culm. Int. Rice Res. Newsl. l3(1):7. Hunt LA, van der Poorten G, Pararajasingham S. 1991. Post-anthesis temperature effects on duration and rate of grain filling in some winter and spring wheats. Can. J. Plant Sci. 71 :609-617. IRRI (International Rice Research Institute). 1977. Annual report for 1976. Manila (Philippines): IRRI. 418 p. IRRI (International Rice Research Institute). 1979. Annual report for 1978. Manila (Philippines): IRRI. 478 p. IRRI (International Rice Research Institute). 1984. Terminology for rice-growing environments. Manila (Philippines): IRRI. 35 p. IRRI (International Rice Research Institute). 1989a. IRRI toward 2000 and beyond. Manila (Philippines): IRRI. IRRI (International Rice Research Institute). 1989b. Implementing the strategy: work plan for 1990-1994. Manila (Philippines): IRRI. p 15-17. IRRI (International Rice Research Institute). 1993a. Rice research in a time of change: IRRI’s medium-term plan for 1994-1998. Manila (Philippines): IRRI. p 35-36. IRRI (International Rice Research Institute). 1993b. Program report for 1992. Manila (Philippines): IRRI. 316 p. IRRI (International Rice Research Institute). 1995. Program report for 1994. Manila (Philippines): IRRI. 311 p. Iwasaki Y, Mae T, Makino A, Ohira K, Ojima K. 1992. Nitrogen accumulation in the inferior spikelet of rice ear during ripening. Soil Sci. Plant Nutr. 38(3):517-525. Janoria MP. 1989. A basic plant ideotype for rice. Int. Rice Res. Newsl. 14(3): 12-13. Johnson GN, Young AJ, Scholes JD, Horton P. 1993. The dissipation of excess excitation energy in British plant species. Plant Cell Environ. 16:673-679. Khatiwada SP. 1995. Variability and genetics of tolerance for aluminum toxicity in rice ( Oryza sativa L.). MS thesis. University of the Philippines Los Baños, Laguna, Philippines. 58 p. Khush GS. 1995. Breaking the yield frontier of rice. GeoJournal 35(3):329-332. Khush GS, Aquino R. 1994. Breeding tropical japonicas for hybrid rice production. In: Virmani SS, editor. Hybrid rice technology: new developments and future prospects. Manila (Philippines): International Rice Research Institute. Khush GS, Peng S. 1996. Breaking the yield frontier of rice. In: Reynolds MP, Rajaram S, McNab A, editors. Increasing yield potential in wheat: breaking the barriers. Mexico: International Maize and Wheat Improvement Center. p 36-51. Kirk GJD, Bajita JB. 1995. Root-induced iron oxidation, pH changes and zinc solubilization in the rhizosphere of lowland rice. New Phytol. 131:129-137.

Genetic enhancement of rice yields

121

Kishore GM. 1994. Starch biosynthesis in plants: identification of ADP glucose pyrophosphorylase as a rate-limiting step. In: Cassman KG, editor. Breaking the yield barrier. Manila (Philippines): International Rice Research Institute. p 117-119. Kropff MJ, Cassman KG, Peng S, Matthews RB, Setter TL. 1994. Quantitative understanding of yield potential. In: Cassman KG, editor. Breaking the yield barrier. Manila (Philippines): International Rice Research Institute. p 21-38. Kuroda E, Ookawa T, Ishihara K. 1989. Analysis on difference of dry matter production between rice cultivars with different plant height in relation to gas diffusion inside stands. Jpn. J. Crop Sci. 58(3):374-382. Ladha JK, Kirk GJD, Peng S, Reddy K, Reddy PM, Singh U, Bennett J. 1995. Opportunities for increased nitrogen use efficiency from improved rice germplasm in lowland rice ecosystems. Paper presented at the Workshop on Nutrient Use Efficiency of the Cropping Systems, 13-15 Dec 1995, International Rice Research Institute, Los Baños, Philippines. Loss SP, Siddique KHM. 1994. Morphological and physiological traits associated with wheat yield increases in Mediterranean environments. Adv. Agron. 52:229-276. Mackill DJ, Amante MM, Vergara BS, Sarkarung S. 1993. Improved semidwarf rice lines with tolerance to submergence of seedlings. Crop Sci. 33:749-753. Majumder ND, Borthakur DN, Rakshit SC. 1989. Heterosis in rice under phosphorous stress. Indian J. Genet. 49:231-235. Makino A, Mae T, Ohira K. 1983. Photosynthesis and ribulose-1,5-bisphosphate carboxylase in rice leaves. Plant Physiol. 73:1002-1007. Makino A, Mae T, Ohira K. 1984. Relation between nitrogen and ribulose-1,5-bisphosphate carboxylase in rice leaves from emergence through senescence. Plant Cell Physiol. 25(3):429-437. Mazaredo AM, Laureles EV, Setter TL. 1996. Growth and yield of modern deepwater rice: comparisons with modern irrigated rice. Field Crops Res. 46:105-116. McDonald DJ, Stansel JW, Gilmore EC. 1974. Breeding for high photosynthetic rate in rice. Indian J. Genet. 34A:1067-1073. Mishra SB, Senadhira D, Manigbas NL. 1996. Genetics of submergence tolerance in rice (Oryza sativa L.). Field Crops Res. 46:177-181. Moeljopawiro S, Ikehashi H. 1981. Inheritance of salt tolerance in rice. Euphytica 30:291-300. Moormann FR, van Breemen N. 1978. Rice: soil, water, land. Manila (Philippines): International Rice Research Institute. 185 p. Murata Y. 1961. Studies on the photosynthesis of rice plants and its cultural significance. Bull. Natl. Inst. Agric. Sci. Jpn. Ser. D, 9:1-169. Narayan KK, Krishnaraj S, Sree Rangaswamy SR. 1990. Genetic analysis for salt tolerance in rice. In: Rice genetics II. Manila (Philippines): International Rice Research Institute. p 167-173. Neales TF, Incoll LD. 1968. The control of leaf photosynthesis by the level of assimilate concentration in the leaf: a review of the hypothesis. Bot. Rev. 34:107-125. Nelson CJ. 1988. Genetic associations between photosynthetic characteristics and yield: review of the evidence. Plant Physiol. Biochem. 26:543-554. Neue HU. 1991. Adverse soil tolerance of rice: mechanisms and screening techniques. In: Deturk P, Ponnamperuma FN, editors. Rice production on acid soils in the tropics. Kandy (Sri Lanka): Institute of Fundamental Studies. p 243-250. Neue HU, Lantin RS, Cayton MTC, Autor NU. 1990. Screening of rices for adverse soils tolerance. In: El Bassam N, Dambroth M, Loughman BC, editors. Genetic aspects of plant nutrition. Kluwer Academic Publishers.

122

Peng and Senadhira

O’Toole JC. 1982. Adaptation of rice to drought-prone environments. In: Drought resistance in crops with emphasis on rice. Manila (Philippines): International Rice Research Institute. p 195-213. Ookawa T, Todokoro Y, Ishihara K. 1993. Changes in physical and chemical characteristics of culm associated with lodging resistance in paddy rice under different growth conditions and varietal difference of their changes. Jpn. J. Crop Sci. 62(4):525-533. Ookawa T, Ishihara K. 1992. Varietal difference of physical characteristics of the culm related to lodging resistance in paddy rice. Jpn. J. Crop Sci. 61(3):419-425. Padmaja Rao S. 1987. Panicle type: few structural considerations for higher yield potential in rice. Indian J. Plant Physiol. 30(1):87-90. Peng S, Cassman KG. 1998. Upper thresholds of nitrogen uptake rates and associated nitrogen fertilizer efficiencies in irrigated rice. Agron. J. (accepted). Peng S, Cassman KG, Kropff MJ. 1995. Relationship between leaf photosynthesis and nitrogen content of field-grown rice in the tropics. Crop Sci. 35:1627-1630. Peng S, Khush GS, Cassman KG. 1994. Evolution of the new plant ideotype for increased yield potential. In: Cassman KG, editor. Breaking the yield barrier. Manila (Philippines): International Rice Research Institute. p 5-20. Penning de Vries FWT. 1991. Improving yields: designing and testing VHYVs. In: Systems simulation at IRRI. IRRI Research Paper Series 5. Manila (Philippines): International Rice Research Institute. p 13-19. Ponnuthurai S, Virmani SS, Vergara BS. 1984. Comparative studies on the growth and grain yield of some F 1 rice ( Oryza sativa L.) hybrids. Philipp. J. Crop Sci. 9(3):183-193. Quatrano RS. 1987. The role of hormones during seed development. In: Davies PJ, editor. Plant hormones and their role in plant growth and development. Kluwer Academic Publishers. p 494-514. Richards RA. 1996. Increasing yield potential: source and sink strength. In: Reynolds MP, Rajaram S, McNab A, editors. Increasing yield potential in wheat: breaking the barriers. Mexico: International Maize and Wheat Improvement Center. p 134-149. Ruban AV, Young AJ, Horton P. 1993. Induction of nonphotochemical energy dissipation and absorbance changes in leaves. Plant Physiol. 102:741-750. Sasahara T, Takahashi T, Kayaba T, Tsunoda S. 1992. A new strategy for increasing plant productivity and yield in rice. Int. Rice Comm. Newsl. 41:l-4. Scobie BM, McDonald D, Willey R. 1993. Investment in rice research in the CGIAR: a global perspective. Rome (Italy): TAC Secretariat, FAO. Senadhira D, Li GF. 1989. Variability in rice grain-filling duration. Int. Rice Res. Newsl. 14(1): 8-9. Senadhira D, Neue HU, Akbar M. 1994. Development of improved donors for salinity tolerance in rice through somaclonal variation. SABRAO J. 26(1-2):19-25. Setter TL, Conocono EA, Egdane JA, Kropff MJ. 1995. Possibility of increasing yield potential of rice by reducing panicle height in the canopy. I. Effects of panicle on light interception and canopy photosynthesis. Austr. J. Plant Physiol. 22:441-451. Setter TL, Peng S, Kirk GJD, Virmani SS, Kropff MJ, Cassman KG. 1994. Physiological considerations and hybrid rice. In: Cassman KG, editor. Breaking the yield barrier. Manila (Philippines): International Rice Research Institute. p 39-62. Shorter R, Lawn RJ, Hammer GL. 1991. Improving genotypic adaptation in crops: a role for breeders, physiologists and modellers. Exp. Agric. 27:155-175. Sinclair TR, Horie T. 1989. Leaf nitrogen, photosynthesis, and crop radiation use efficiency: a review. Crop Sci. 29:90-98.

Genetic enhancement of rice yields

123

Singh G, Singh S, Gurung SB. 1984. Effect of growth regulators on rice productivity. Trop. Agric. 61:106-108. Slafer GA, Calderini DF, Miralles DJ. 1996. Generation of yield components and compensation in wheat: opportunities for further increasing yield potential. In: Reynolds MP, Rajaram S, McNab A, editors. Increasing yield potential in wheat: breaking the barriers. Mexico: International Maize and Wheat Improvement Center. p 101-133. Slafer GA, Rawson HM. 1994. Sensitivity of wheat phasic development to major environmental factors: a re-examination of some assumptions made by physiologists and modellers. Austr. J. Plant Physiol. 21:393-426. Stark DM, Timmerman KP, Barry GF, Preiss J, Kishore GM. 1992. Regulation of the amount of starch in plant tissues by ADP glucose pyrophosphorylase. Science 258:287-292. Steponkus PL, Shahan KW, Cutler JM. 1986. Osmotic adjustment in rice. In: Drought resistance in crops with emphasis on rice. Manila (Philippines): International Rice Research Institute. p 181-194. Suge H. 1988. Physiological genetics of internode elongation in submerged deepwater rice. In: Proceedings of the 1987 International Deepwater Rice Workshop. Manila (Philippines): International Rice Research Institute. p 275-286. Tanaka A, Los R, Navasero SA. 1966. Some mechanisms involved in the development of iron toxicity symptoms in the rice plant. Soil Sci. Plant Nutr. 12:32-38. Teare ID, Peterson CJ, Law AG. 1971. Size and frequency of leaf stomata in cultivars of Triticum aestivum and other Triticum species. Crop Sci. 11:496-498. Terashima K, Akita S, Sakai N. 1995. Physiological characteristics related with lodging tolerance of rice in direct sowing cultivation. III. Relationship between the characteristics of root distribution in the soil and lodging tolerance. Jpn. J. Crop Sci. 64:243-250. Thach TD. 1994. The genetic association between elongation ability and submergence tolerance in rice. MS thesis. Central Luzon State University, Muñoz, Nueva Ecija, Philippines. 65 p. Tsunoda S. 1962. A developmental analysis of yielding ability in varieties of field crops. IV. Quantitative and spatial development of the stem-system. Jpn. J. Breed. 12:49-56. Tu ZP, Lin XZ, Cai WJ, Yu ZY. 1995. Reprobing into rice breeding for high photosynthetic efficiency. Acta Bot. Sin. 37(8):641-651. Turner NC, O’Toole JC, Cruz RT, Yambao EB, Ahmed S, Namuco OS, Dingkhun M. 1986. Response of seven diverse rice cultivars to water deficits. 2. Osmotic adjustment, leaf elasticity, leaf extension, leaf death, stomatal conductance and photosynthesis. Field Crops Res. 13:273-286. Venkateswarlu B, Vergara BS, Parao FT, Visperas RM. 1986. Enhanced grain yield potentials in rice by increasing the number of high density grains. Philipp. J. Crop Sci. 11:145-152. Vergara BS. 1988. Raising the yield potential of rice. Philipp. Technol. J. 13:3-9. Virmanj SS. 1994. Prospects of hybrid rice in the tropics and subtropics. In: Virmani SS, editor. Hybrid rice technology: new developments and future prospects. Manila (Philippines): International Rice Research Institute. p 7-19. Virmani SS, Aquino RC, Khush GS. 1982. Heterosis breeding in rice, Oryza sativa L. Theor. Appl. Genet. 63:373-380. Wada G, Matsushima S. 1962. Analysis of yield determining processes and its application to yield prediction and culture improvement of lowland rice. Proc. Crop Sci. Soc. Jpn. 31:1518.

124

Peng and Senadhira

Woperies MCS, Kropff MJ, Hunt ED, Sanidad W, Bouman J. 1993. Case study on regional application of crop growth simulation models to predict rainfed rice yields, Tarlac Province, Philippines. In: Bouman BAM, van Laar HH, Zhaoqian W, editors. Proceedings of an international workshop on agroecological zonation of rice. Wageningen Agricultural University. Yeo AR, Yeo ME, Flowers SA, Flowers TH. 1990. Screening of rice (Oryza sativa L.) genotypes for physiological characters contributing to salinity resistance and their relationship to overall performance. Theor. Appl. Genet. 79:377-384. Yoshida S. 1973. Effects of temperature on growth of the rice plant (Oryza sativa L.) in a controlled environment. Soil Sci. Plant Nutr. 19:299-310. Yoshida S. 1981. Fundamentals of rice crop sciences. Manila (Philippines): International Rice Research Institute. Yoshida S, Coronel V. 1976. Nitrogen nutrition, leaf resistance, and leaf photosynthetic rate of the rice plant. Soil Sci. Plant Nutr. 22(2):207-211. Yoshida S, Parao FT. 1976. Climatic influence on yield and yield components of lowland rice in the tropics. In: Climate and rice. Manila (Philippines): International Rice Research Institute. p 471-494. Yuan LP, Virmani SS, Mao CX. 1989. Hybrid rice: achievements and future outlook. In: Progress in irrigated rice research. Manila (Philippines): International Rice Research Institute. p 219-223. Yuan LP. 1994. Increasing yield potential in rice by exploitation of heterosis. In: Virmani SS, editor. Hybrid rice technology: new developments and future prospects. Manila (Philippines): International Rice Research Institute. p 1-6. Zeigler RS, Puckridge DW. 1995. Improving sustainable productivity in rice-based rainfed lowland systems of South and Southeast Asia. GeoJournal 35(3):307-324. Zelitch I. 1982. The close relationship between net photosynthesis and crop yield. Bioscience 32:796-802.

Notes Authors’ address; S. Peng, International Rice Research Institute (IRRI), P.O. Box 933, Manila, Philippines. [email protected]; D. Senadhira is deceased. Citation: Dowling NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute.

Genetic enhancement of rice yields

125

CHAPTER 8

Intensification of rice production systems: opportunities and limits W. Reichardt, A. Dobermann, and T. George

Intensification of rice systems implies the disturbance of existing equilibria in soil by extensive submergence and elevated levels of agrochemicals in nutrient and pest management. In keeping pace with the deployment of ever higher yielding rice varieties, nutrient management risks adversely affecting the agronomic and environmental sustainability of rice lands. The first signs of declining productivity reported from onstation field experiments have been linked to reduced soil N-supplying capacity. Furthermore, neglect of non-N mineral fertilizers has frequently led to depletion in K, P, S, and Zn. In regions with rapidly progressing intensification, inputs of organic carbon as residue or as manure have been discontinued. On the other hand, the organic matter pool of rice-cropping systems can be seen as a mechanistic key to nutrient supply. With microbial biomass as its most rapidly recycled segment, the organic phase serves as a source of biocatalysts governing nutrient supply and as a nutrient pool by itself. The dynamics of the organic matter phase in flooded soils are fundamentally different from those in aerated soils. Green manure derived from N-fixing organisms has its merits in less intensified systems where it can provide sufficient N at N rates below 100 kg ha-1. Options for sustaining the most intensified resource bases would have to include a demand-driven integrated inorganic/organic nutrient management and rotation cropping, the latter mainly in response to periodic annual shortages of irrigation water. As a prerequisite for the rotation of rice with upland crops, however, an efficient, fine-tuned nutrient and pest management would have to be established. In tropical wetlands, intensive rice cropping is dealing with a greater diversity of habitats and biological and biogeochemical functions over space and time than other agroecosystems. In accordance with ecological theory, this is likely to confer maximum stability and sustainability on agricultural wetlands.

Intensification and its impact on the soil resource base Sustainability About 30 years after the Green Revolution in Asia, the sustainability of intensified rice production systems can be viewed from different perspectives that reflect seemingly conflicting interests. The application of more recent concepts in agroeconomics, however, could bridge the gap between economic and ecological goals. Replacing the gross domestic product with the net domestic product has become a conceptual ad-

127

vance with far-reaching consequences. It requires agronomists to include impacts on environmental capital in agronomic yield equations (Swaminathan 1991). It was remarkable that agronomists and not ecologists triggered the resurging interest in sustainable rice production. Following dramatic yield increases since the 1960s, productivity in farmers’ fields in Southeast Asia became stagnant or even declined from the mid–1980s (Flinn and De Datta 1984, Cassman and Pingali 1995a, Cassman et al 1996). Given the expected increase in future demand for rice and plant breeders’ capacity to develop varieties for higher and more stable yields, the limiting capacity of the soil resource base has become a crucial issue. Opportunities to enhance yield through improved nutrient, water, and pest management will have to be balanced against the hidden risk of degrading the agronomic and environmental quality of rice-growing areas and beyond.

Rice cropping before the Green Revolution Wetland rice is the only major crop that was grown for many centuries and possibly millennia in monoculture without major soil degradation (Bray 1986, Uexkuell and Beaton 1992). Soil flooding and puddling maintained favorable soil properties for rice growth (Ponnamperuma 1972), and traditional rice-growing patterns were geared for stability instead of high yields. Traditional long-duration varieties (130–210 d) with low harvest index and yield were grown and, in many areas, much of the straw remained in the field (Uexkuell and Beaton 1992). Bunds protected rice fields from soil erosion. Floodwater buffered the soil temperature and allowed ample growth of N2-fixing microorganisms (Roger 1996). Suspended particles and soluble nutrients from rainfall and irrigation water contributed to an indigenous nutrient supply covering the demand of extensively grown crops. Current rainfall contributions to annual nutrient inputs to irrigated rice fields of Asia are estimated to be in the range of 1–10 kg N ha-1, 0.2–2 kg P ha-1, 3–10 kg K ha-1, and 5–20 kg S ha-1. Low net total nutrient inputs may have supported yields of 1–2 t ha-1. In traditional irrigated rice systems where net total nutrient removal as well as daily nutrient uptake rates were low, nutrient additions from natural sources were an important component of the overall nutrient balance, and even poor soils had the capacity to supply enough nutrients to sustain yields of 1–2 t ha-1. Such systems originated in river valleys and deltas of Asia and they remained unchanged for hundreds of years.

Intensification effects on physicochemical properties of flooded rice soils The invention and widespread adoption of high-yielding, early maturing semidwarf indica varieties in the 1960s led to a rapid intensification in the tropical lowlands of Asia. New varieties such as IR8 had a short growth period and greater yield potential because of more efficient biomass partitioning, were short-statured and lodging-resistant, and responded well to fertilizer N additions. The use of external inputs such as fertilizers, water, energy, and pesticides increased and the diversity of rice varieties used in irrigated systems decreased. The higher yield potential of modern varieties

128

Reichardt et al

promoted private and public investments in irrigation infrastructure. Tillage and management intensity improved through extension programs and soils remained submerged for longer periods. Two to three rice crops per year became a reality. Average grain yield reached 4.9 t ha -1 in 1991 (Cassman and Pingali 1995b) and harvesting techniques changed. To facilitate land preparation for the next crop, farmers started to cut the entire crop and remove or burn the straw (Uexkuell and Beaton 1992). Since the mid-1980s, trends of declining factor productivity have been noted in long-term rice monoculture and, later, rice-wheat experiments (Flinn and De Datta 1984, Cassman et al 1995, Nambiar 1995). There is evidence that the declining productivity trends come from a gradual degradation of soil quality caused by intensive cropping. Reduced soil N-supplying capacity was identified as a driving force, despite conservation or even an increase in total soil organic matter content (Cassman et al 1995, Cassman and Pingali 1995a). Depletion of soil nutrient reserves, buildup of soil pests, physicochemical changes in the soil caused by increased submergence, and changes in soil microflora were also listed as possible causes of the productivity decline, but universal mechanisms have not yet been identified. There are numerous examples of soil nutrient depletion other than soil N in intensive rice systems. In productive soils of the alluvial floodplains of South and Southeast Asia, P and K rarely limited rice productivity before these systems were intensified (Kawaguchi and Kyuma 1977, De Datta and Mikkelsen 1985, Bajwa 1994). In most early fertilizer trials with modern varieties, no significant responses to P or K additions were observed, whereas tremendous yield gains could be achieved by applying N fertilizer. Depletion of extractable soil P to a level that significantly reduced N use efficiency and grain yield was first shown in long-term experiments in the Philippines (De Datta et al 1988). Similar effects were noted in long-term experiments in China. Across 11 sites in five countries, the negative P balance averaged -7 to -8 kg ha -1 per crop in zero-P treatments, whereas fertilizer P rates of 17-25kg ha -1 were required to maintain the P balance or to increase total soil P (Fig. 1; Dobermann et al 1996b). Potassium deficiency has become a constraint in soils that were previously not considered as K-limited (Chen et al 1992, Mohanty and Mandal 1989, De Datta and Mikkelsen 1985, Uexkuell 1985, Dobermann et al 1996c, Oberthuer et al 1996). Modern rice varieties require similar amounts of K and N (20 kg of each per ton of grain yield). Most rice farmers in Asia do not apply much fertilizer K, and, as a result of intensification, straw was increasingly removed from the field. In long-term experiments at 11 sites, the K balance was highly negative in all NPK combinations tested (-34 to -63 kg ha -1 per crop cycle, Fig. 1) and even fertilizer K application at an average rate of 40 kg ha -1 in the +NK and +NPK treatments was not enough to match the K removal at most sites (Dobermann et al 1996c). Examples of K depletion observed in farmers’ fields include alluvial, illitic soils in India (Tiwari 1985), lowland rice soils of Java, Indonesia (Sri Adiningsih et al 1991), and vermiculitic clay soils of Central Luzon, Philippines (Oberthuer et al 1996). Although researchers started to raise concern about the danger of negative K balances and soil K depletion many

Intensification of rice production systems: opportunities and limits

129

Fertilizer input

Plant uptake

Recycled with stubble

Net balance

Fig. 1. Partial net K and P balance for one rice crop in five different fertilizer treatments. Values shown are averages and standard deviations (error bars) of long-term experiments at 11 sites in five countries sampled in 1993. Stubble was recycled at five sites and all straw was removed at six sites, reflecting standard farmer practices for each location.

130

Reichardt et al

years ago (Uexkuell 1985, De Datta and Mikkelsen 1985, Kemmler 1980), this has not yet led to a significant improvement of K management. Intensification also contributed to more widespread occurrence of S and Zn deficiencies in marginally productive lowland rice soils (Uexkuell and Beaton 1992, Blair et al 1978). The removal or burning of straw or the replacement of sulfur-containing fertilizers with non-S fertilizer (Yoshida 1981) contributed to S depletion in several rice areas. More recently, however, increased air pollution and S deposition associated with rapid industrial development seem to counteract this trend in some parts of South and Southeast Asia. Little is known about Zn balances in traditional and intensive irrigated rice culture. Zinc deficiency is usually associated with leached ultisols and oxisols with high pH or high amounts of organic matter, but Zn depletion may also occur in nonalkaline soils with ZnS formation (Oberthuer et al 1996). On marginally productive and highly weathered soils, the increased supply of N intensified deficiencies in K, P, and Zn, resulting in the spread of a nutritional disorder known as iron toxicity (Ottow et al 1981). There is limited quantitative information on the effect of prolonged submergence on the soil’s physicochemical properties and its effects on nutrient supply. Though most irrigated rice lands are probably not prone to salinization, the long-term use of poor-quality irrigation water may cause undesirable changes in soil chemistry. Because of the precipitation of carbonates, soil pH may increase (Marx et al 1988). In some areas where groundwater is the irrigation source, high net additions of Ca and Mg may result in reduced K availability because of a wide (Ca + Mg)/K ratio (Dobermann et al 1995). We do not have enough quantitative information about the importance of such processes for sustaining soil quality.

Effect of intensification on the organic phase Crop intensification increases the total pool of organic matter in the soil because of intensified root formation and root exudation, and decreased mineralization processes under anoxic conditions (Olk and Cassmann 1995). Photosynthetic primary production in the floodwater provides another soil organic matter source. An average fraction of 1-5%of soil organic matter accounts for living biomass (Anderson and Domsch 1980, Inubushi and Watanabe 1986). This consists mainly of heterotrophic microorganisms and represents an easily available pool of nutrients with a rapid turnover rate (Lee 1994). Traditional rice cultivation owed most of its sustainability to the continuous replenishment of the organic matter pool (Bray 1986). As a result of the faster turnaround time between intensified crops, farmers eliminate the entire crop from the field and often burn the straw (Uexkuell and Beaton 1992). The intensified cultivation of higher yielding, less photoperiod-sensitive varieties with shorter growth periods required roughly a doubling of the soil nutrient supply. Mineral fertilizers can rapidly and efficiently satisfy this growing demand when bypassing removal in nutrient cycling or the retarding sequences of sequestration and remobilization (Broadbent 1984).

intensification of rice production systems: opportunities and limits

131

Tropical wetland soils are known for their rapid decomposition of organic debris (Kimura et al 1990). Nevertheless, total soil organic carbon content seems to be conserved or even enhanced in long-term trials under intensive double or triple cropping for decades, though it is partially attenuated by decreasing bulk density (Cassman and Pingali 1995a,b). This may be an indirect evidence of fertilizer-induced, largely microbial, soil organic matter production (Broadbent 1984). Although biomass pools in the photosynthesis-dominated floodwater subsystem are small, its autotrophic productivity can reach 600 kg ha-1 over a cropping period (Saito and Watanabe 1978). Intensification of lowland rice crops implies extended periods of submergence. Thus, anoxic conditions prevailing in the bulk soil can both slow down the primary attack of extracellular enzymes on particulate organic matter (Reichardt 1986) and modify the metabolic pathways of microbial mineralization (Schink 1988). Finally, humification processes involving the buildup of phenolic compounds depend strongly on the chemical composition of the organic input in the submerged system. In contrast to green manure, rice straw may release high amounts of phenolic compounds (Tsutsuki and Ponnamperuma 1987). Polymerization and mineralization of phenols are delayed under anaerobic conditions, which favor the accumulation of young, lowhumified soil organic matter that is rich in phenols (Ye and Wen 1991, Palm and Sanchez 1991, Becker et al 1994a, Olk and Cassmann 1995). N-containing aromatic compounds could give a mechanistic explanation for the declining endogenous N supply in continuously flooded anaerobic fields (Cassmann et al 1995).

Have rice cropping systems become less sustainable since the advent of the Green Revolution? Irrigated rice systems in tropical Asia will remain the major source of food production in the region, but their sustainable management represents an enormous challenge. Because of increased cropping intensity and yields, the pressure on the soil resource base has increased tremendously. Yield decline, changes in organic matter quality, and nutrient depletion seem to indicate that modern intensive rice systems are less sustainable than the traditional rice culture practiced for thousands of years. Both increased nutrient demand and prolonged submergence seem to cause gradual changes in soil quality that need to be managed. Some sustainability issues in irrigated rice, such as negative nutrient balances, clearly result from inadequate soil and crop management. Loss of indigenous nutrient supply and negative nutrient balances are the key factors that may reduce the ability of the soil resource base to sustain high rice yields. The seed and fertilizer package approach used during the Green Revolution in Asia did not address such problems adequately. Nutrient management practices of most rice farmers in Asia focus on optimizing short-term gains rather than sustaining soil quality over the long run (Uexkuell and Beaton 1992). Exploiting native soil fertility prevails over maintaining or enhancing soil fertility. The diverse nature of the soil resource base, particularly the large variation in indigenous nutrient supply, has not been taken into account adequately. The importance of returning at least part of the rice straw for soil organic

132

Reichardt et al

matter conservation and the nutrient balance is well known, but, for various reasons, is widely neglected in current field management. Therefore, some negative trends in productivity are probably reversable through site-specific nutrient management approaches that focus on optimizing nutrient use efficiency in combination with longterm soil fertility management (Dobermann et al 1996a). Such a fine-tuning of system performance will probably significantly improve the productivity and sustainability of intensive rice systems. The preservation of natural resources depends on a system’s environmental sustainability. The latter is often viewed in terms of biodiversity (Schoenly et al, this volume, Chapter 17). There are numerous examples of a reduction in genetic richness and organismic diversity caused by agronomic land use, both among flora and fauna (Schoenly et al 1996a,b, this volume, Chapter 17) and among microorganisms (Torsvik et al 1990). Microorganisms serve key functions that are also crucial for agronomic sustainability (Chapin et al 1997). This refers to nutrient cycling as well as to the incidence of pathogens and their antagonists. A preliminary comparison of a wetland soil left fallow with a continuously cropped soil of the same texture indicates that intensive irrigated rice cropping can substantially reduce microbial functional diversity (Fig. 2; Reichardt et al 1996). Problems with agronomic sustainability such as declining yields have been observed in a few long-term experiments with good nutrient management in both predominantly anaerobic (rice-rice) and anaerobic-aerobic (rice-wheat) cropping systems. Identifying the causes for this remains a challenge. At this stage, at what productivity level intensive rice systems can become environmentally and economically sustainable is an open question.

Options for sustaining the soil resource and functions Monocropping versus diversification Cassman and Pingali (1995b) discussed a reduction in the intensity of flooded rice monocropping by diversifying into higher-value nonrice crops in rotation with rice. In flooded rice systems with two to three crops per year, diversification means providing an aerated upland crop phase. Provision of an aerated phase between two flooded rice crops would reverse the buildup of phenol-rich humic compounds (Olk et al 1996) that may cause a reduction in N availability (Cassman et al 1995). An aerated phase long enough to grow an upland crop is justified, if the total productivity is maintained or even enhanced, provided the quality of the resource base is not adversely affected. Recent research in favorable rainfed lowland rice systems (George et al 1992, 1993, 1994, 1995) indicates that total productivity can indeed be increased by proper management of dry-season and dry-to-wet-season transition vegetation including grain and green manure legume crops. Without an enhancement in total productivity, short periods of aerated phase between flooded rice could likely provide the same benefits as a whole aerated crop in terms of soil aeration, organic matter decomposition and

Intensification of rice production systems: opportunities and limits

133

Fig. 2. Distinct similarity profiles based on 58 phospholipid fatty acid biomarkers, with positive Biolog® test scores of microbial functions in submerged fallow and triple-cropped rice soils at the IRRI farm, Laguna, Philippines.

formation, and microbial activity. On the other hand, an aerated soil phase may not always be possible because of the heavy clay texture of rice lowlands in the humid tropics. In regions with coarse-textured soils where rice-vegetable rotation cropping is practiced, the excessive use of agrochemicals for the dry-season crops already threatens to lower the quality of the entire resource base. Rice-wheat systems are facing soil quality problems, too (Hobbs et al 1996, Nambiar and Abrol 1989). Thus, diversification alone does not necessarily solve the problems associated with intensification in the irrigated rice lowlands, although it can be part of an overall solution.

134

Reichardt et al

It may be that crop diversification would become mandatory in the current intensively cultivated irrigated rice systems with the prospect of water becoming a premium commodity. Developing high-value rice-nonrice crop systems such as ricesoybean or rice-vegetables is a real possibility with the ever-increasing urban demand for water.

Balancing nutrient inputs and outputs Substantial quantities of N, P, K, and S are removed from the soil with each crop. Maintaining nutrient balances is therefore a prerequisite for sustaining the resource base. Although our understanding of nutrient cycling in the intensive irrigated rice system has improved, this has not led to measurable improvements in nutrient management practices in farmers’ fields. Farmers' decisions about fertilizer application are often more affected by socioeconomic factors (market availability, prices, availability of money) than by biophysical needs. Farmer adoption of existing technology and recommended practices is confounded by considerable field-to-field variability in the indigenous soil nutrient supply. To achieve and sustain average yields greater than 7-8 t ha-1 , the nutrient use efficiency from both indigenous and external sources will have to be increased. This implies that nutrient management recommendation domains would have to shift from large regions to farms, single fields, or even single parcels within a larger field. Knowledge-based objectives and tactics for management differ for each essential nutrient (Dobermann et al 1996a). Adjusting the quantity of applied N to variations in the indigenous N supply is as important as timing, placement, and source of applied N (Peng et al 1996, Cassman et al 1996). Because nutrients such as P and K are not easily lost or added to the root zone by the biological and chemical processes affecting N, their management requires a long-term strategy that emphasizes maintenance of soil nutrient supply to ensure that crop growth and N use efficiency are not limited. Diagnosis of potential deficiencies is the key management tool for nutrients such as Mg, Zn, and S. Once identified as a problem, deficiencies can be alleviated by regular or irregular (single) measures as part of a general fertilizer/soil use recommendation (Dobermann et al 1996a). Straw management is a key leverage point for maintaining a positive balance of most nutrients, particularly for N and K (Becker et al 1994b, Dobermann et al 1996c, 1998). Increasing combine or stripper harvesting may provide new opportunities for better crop residue recycling. Implementing site-specific management will only be successful if the additional labor required is restricted to a minimum, if the economic gain is sufficient, and if suitable easy-to-use decision aid tools become available. In many Asian countries, facilities for more sophisticated farmer support need to be built up. Included among these are soil-testing laboratories and a soil-testing program (perhaps with the involvement of the private sector), fertilizer recommendation services, objective information about new fertilizer products, and the use of mass media (radio, TV, newspapers) for extension of new technologies. Because the transition to farm- or field-

Intensification of rice production systems: opportunities and limits

135

specific management will take time, other cost-effective ways to increase nutrient use efficiencies must be further explored over the shorter term (Pingali et al 1998).

Managing the organic phase in rice soil/floodwater systems More than 100 million tons of rice straw are estimated to be produced annually in Southeast Asia, but only a small fraction is presently reincorporated into the soil (Blair et al 1995). Also, enrichment of the organic phase with N2-fixing green manure has ceased in many areas with progressing intensification, as its main purpose of providing sufficient N was no longer served (Becker et al 1994a,b, George et al 1998). With a change in economic conditions, however, the use of green manure could resume, possibly supplementing the use of inorganic N fertilizer as part of integrated nutrient management strategies. Fallow periods of only 40-60 d in intensive systems would limit the use of leguminous green manure to the fastest growing short-duration legumes such as the stem-nodulating Sesbania rostrata (Singh et al 1991, Ventura and Watanabe 1993, George et al 1993, 1998). Because green manure is chosen for its capacity to accumulate N from N2 fixation, its performance is judged in terms of the agronomic efficiency of its N component (Morris et al 1986, Becker et al 1994b). This efficiency is comparable to that of inorganic fertilizer only at N levels below 100 kg ha-1 (Singh et al 1991). Nitrogen input, however, would not be the only criterion to justify the use of green manure. In China, a number of K-rich green manure plants have proved successful as potash fertilizers with the beneficial side effect of enhanced protein content in the grain (Peng and Yi 1992). Grown in situ, however, such green manure plants do not contribute to a net addition of K to the soil. The complex effects of organic matter inputs on soil quality improvement are also reflected in soil reclamation practices. Green manuring has been shown to be effective in accelerating the reclamation of saline and sodic soils (Singh et al 1991). Management of organic matter has not kept pace with the recent intensification of rice systems. Success hinges on a clearer understanding of how the network of biogeochemical pathways is regulated. Part of the mineral fertilizers is also assimilated by a dynamic, metabolizing matrix of active biomass and organic matter that forms the system’s food web (see Nannipieri et al 1994, Clarholm 1994; Fig. 3). Improved management of the organic matter in the soil means that nutrient release is keeping pace with crop demand. This is achieved by making use of the dynamics of the biota in the rice soil/floodwater systems as the most labile fraction of organic matter (Nannipieri et al 1994). Current knowledge gaps concerning nutrient supply to lowland rice are further illustrated by the fact that soil nutrient analyses refer to the submerged, anoxic bulk soil, whereas an envelope of oxygen surrounding the roots creates a completely different microenvironment for nutrient uptake (Armstrong 1967, Kirk et al 1993, Revsbech and Reichardt, unpublished).

136

Reichardt et al

Fig. 3. Central role of microbial biomass in nutrient immobilization and release. Pom = particulate organic matter, Dom = dissolved organic matter, Som = soil organic matter.

Is organic farming a viable option in highly intensified systems? Organic farming, a practice that uses only organic inputs for production, is sometimes claimed to be the way to attain sustainable crops. Yet there are considerable doubts whether organic inputs alone can sustain high levels of production in intensified systems without polluting the resource base, including its drainage area. For example, the excessive use of green manure will cause the release of nitrate and ammonium from the unused fraction of the organic input. There is growing evidence that this can lower the quality of the resource base, the same as the excessive use of inorganic N fertilizers (George et al 1993, 1994, 1998). A serious practical problem for intensive organic rice farming on a large scale would be generating the quantities of organic nutrients (not to mention their transportation costs) that are required to compensate for nutrient depletion after each harvest. So far, we have not seen convincing evidence that the supply of nutrients from organic sources to highly intensive cropping systems can be managed on a large scale. There seems to be much more potential for improving the integrated use of nutrients from inorganic and organic sources as appropriate to sustain productivity at high yield levels. The bottom line is that rice production must keep pace with the demand for rice by the ever-increasing rice-eating population.

Intensification of rice production systems: opportunities and limits

137

Caveat: moving intensified rice cultivation to the uplands? Wetland rice production systems apparently owe much of their sustainability to flooding. Processes in the floodwater component and submergence of the soil create the chemical and biological basis for a continuous renewal of the system’s soil fertility (Ponnamperuma 1984, Roger 1996). Hence, it seems inevitable that most rice is produced in irrigated systems. Yet the majority of the world’s rice area is rainfed. Further, these less productive rice areas, in particular the uplands, are inhabited by the poorest farmers. Productivity gains should obviously be achieved in these presently less productive areas. But how sustainable will these rainfed systems be if we intensify rice production? Indications are that intensification is possible, but only in the limited, more favorable rainfed areas, including the uplands. If total rice production on all rice land were maintained at the same level as in irrigated systems (4.9 t ha-1), global production would reach 727 million t. But because of the marked differences in production systems, real production falls 207 million t short of that figure (Prasad et al 1995). Because of the increasing water shortage, the potential and sustainability of less water-intensive alternatives to the present irrigated rice systems will have to be explored. Limited areas in the uplands, where the rainy season is relatively free of drought spells and the land is flat to moderately sloping, have the potential for intensified crop production that includes rice. Though severely deficient in nutrients and usually highly acidic (Sanchez 1976), the highly weathered tropical upland soils possess the best physical properties for supporting crop production (Sanchez and Logan 1992). Phosphorus limitation is an example of a serious constraint even in traditional upland rice production (Fig. 4). Yet substantial productivity gains are possible once nutrient deficiencies and problems associated with soil acidity have been overcome (Sanchez and Logan 1992, Cassman et al 1993, Uexkuell and Mutert 1995). An example of a favorable rainfed upland is the cerrado ecosystem in Brazil. Its approximately 100 million ha of highly acidic upland soil could be reclaimed. In addition, large-scale mechanization in rice production is possible, as the first attempts in a few areas have shown. For the less favorable, fragile agroecosystems, the obstacles to sustaining the resource base outweigh the gains in most instances and regions.

Opportunities for short-term measures of environmental sustainability Agronomic sustainability, which implies stable productivity, is reflected in measures such as annual yield records, partial factor productivity, or nutrient balances that can be monitored with each crop. Useful as they are, such tong-term records can only give an incomplete account of the total factor capacity of the resource base to sustain high yields. Being production-targeted, they do not include aspects of the environmental quality of the flooded resource base. Ultimately, productivity and environmental quality of a rice field are both linked to processes in the organic phase. Here, microorganisms are the main carriers of biocatalytic functions (Chapin et al 1997). They affect nutrient supply to the crop as well as the cycling of bioelements, which is a crucial function in any ecosystem. This

138

Reichardt et al

Figure 4. Phosphorus uptake by traditional upland rice in response to near-non-limiting applications of P with and without nitrogen and potassium in Southeast Asian uplands, 0 P = no P, P = 50 kg P ha -1 , and P + NK = 50 kg P ha -1 plus 100 kg N ha -1 and 50 kg K ha -1. Columns under each country are not significantly different by LSD (0.05) if indicated by the same lowercase letter. (George, unpublished.)

linkage allows us to look for promising combined measures of yield- and ecosystemrelated sustainability (Matson et al 1997). Rapid progress in microbial ecology provides a number of options for short-term assays to quantify the sustainability of biocatalytic functions in the soil environment. One such category of assays targets certain enzymatic processes as potential indicators of functional imbalances in an ecosystem (Reichardt et al 1993, Reichardt 1996). The relatively new discipline of ecotoxicology, which focuses on man-made damage to ecosystem functions and environmental health, has adopted biochemical techniques that were designed for holistic analyses of ecosystem functions. Another category on which potential measures of environmental sustainability are based involves the concept of functional microbial diversity and richness (Atlas 1984, Coleman et al 1994). It partly requires advanced techniques such as biomarker analysis (Tunlid and White 1992, Reichardt et al 1997). An alternative methodology is based on substrate mineralization patterns (Zak et al 1994, Reichardt et al 1996, 1997). The principle on which commercially available test kits such as Biolog (Zak et al 1994, Schoenly et al, this volume, Chapter 17) are already based might eventually allow rapid tests of functional sustainability to be carried out in farmers’ fields. Notwithstanding the potential role of biodiversity as an indicator of an agroecosystem’s sustainability (Chapin et al 1997, Matson et al 1997), intensified lowland rice production systems are composed of an extremely large number of very diverse microbial subhabitats in space and time (Schoenly et al, this volume, Chapter 17). The floodwater compartment with its primary production in particular is viewed as a major supporter of the system’s sustainability (Roger 1996). Moreover, under conditions that are conducive to aquaculture at the same time, a potent “natural” toxicity testing system could become available to farmers (Dela Cruz et al 1992).

Intensification of rice production systems: opportunities and limits

139

References Anderson JPE, Domsch KH. 1980. Quantities of plant nutrients in the microbial biomass of selected soils. Soil Sci. 130:211-216. Atlas RM. 1984. Use of microbial diversity measurements to assess environmental stress. In: Klug JM, Reddy CA, editors. Perspectives in microbial ecology. Proceedings of an International Symposium on Microbial Ecology, East Lansing, Mich. p 540-545. Armstrong W. 1967. The oxidizing activity of roots in waterlogged soils. Physiol. Plant. 20:920926. Bajwa MI. 1994. Soil potassium status, potash fertilizer usage and recommendations in Pakistan. Potash Review No. 3/1994. Subject 1, 20th suite. Base1 (Switzerland): International Potash Institute. Becker M, Ladha JK, Simpson IC, Ottow JCG. 1994a. Parameters affecting residue nitrogen mineralization in flooded soils. Soil Sci. Soc. Am. J. 58:1666-1671. Becker M, Ladha JK, Ottow JCG. 1994b. Nitrogen losses and lowland rice yield as affected by residue nitrogen release. Soil Sci. SOC. Am. J. 58: 1660-1665. Blair GJ, Mamaril CP, Momuat E. 1978. Sulfur nutrition of wetland rice. IRRI Research Paper Series 21. Manila (Philippines): International Rice Research Institute. Blair G, Lefroy R, Konboon Y, Wonprasaid S, Naklang K. 1995. Carbon and nutrient pools in rice cropping systems. In: Fragile lives in fragile ecosystems. Manila (Philippines): International Rice Research Institute. p 161-172. Bray F. 1986. The rice economies: technology and development in Asian societies. London (UK): B. Blackwell. Broadbent FE. 1984. Plant use of soil nitrogen. In: Nitrogen in crop production. Madison, Wis. (USA): ASA-SSSA. p 171-182. Cassman KG, Singleton PW, Linquist BA. 1993. Input/output analysis of the cumulative soybean response to phosphorus on an Ultisol. Field Crops Res. 34:23-36. Cassman KG, De Datta SK, Olk DC, Alcantara J, Samson M, Descalsota JP, Dizon M. 1995. Yield decline and the nitrogen economy of long-term experiments on continuous, irrigated rice systems in the tropics. In: Lal R, Stewart BA, editors. Sustainable management of soils. Michigan (USA): Lewiston Publishers, CRC Press Inc. p 181-222. Cassman KG, Gines GC, Dizon M, Samson MI, Alcantara JM. 1996. Nitrogen-use efficiency in tropical lowland rice systems: contributions from indigenous and applied nitrogen. Field Crops Res. 47:l-12. Cassman KG, Pingali PL. 1995a. Extrapolating trends from long-term experiments to farmers’ fields: the case of irrigated rice systems in Asia. In: Barnett V, Payne R, Steiner R, editors. Agricultural sustainability in economic, environmental, and statistical terms. London: J. Wiley & Sons. p 64-84. Cassman KG, Pingali PL. 1995b. Intensification of irrigated rice systems: learning from the past to meet future challenges. GeoJournal 35:299-305. Chapin FS, Walker, BH, Hobbs RJ, Hooper DU, Lawton JH, Sala OE, Tilman D. 1997. Biotic control over the functioning of ecosystems. Science 277:500-504. Chen FX, Chen YA, Zou CM. 1992. Potassium balance in red earth paddy soil of Southern Hunan and its application. Proceedings of the International Symposium on Paddy Soils. Nanjing (China): Chinese Academy of Sciences. p 193-194. Clarholm M. 1994. The microbial loop in soil. In: Ritz K, Dighton J, Giller KE, editors. Beyond the biomass. Chichester: J. Wiley. p 221-230.

140

Reichardt et al

Coleman DC, Dighton J, Ritz K, Giller KE. 1994. Perspectives on the compositional and functional analysis of soil communities. In: Ritz K, Dighton J, Giller KE, editors. Beyond the biomass. Chichester: J. Wiley. p 261-270. De Datta SK, Gomez KA, Descalsota JP. 1988. Changes in yield response to major nutrients and in soil fertility under intensive rice cropping. Soil Sci. 146:350-358. De Datta SK, Mikkelsen DS. 1985. Potassium nutrition of rice. In: Munson RD, Sumner ME, Bishop WD, editors. Potassium in agriculture. Madison, Wis. (USA): ASA, CSSA, SSSA. p 665-699. Dela Cruz CR, Lightfoot C, Costa-Pierce BA, Carangal VC, Bimbao MP. 1992. Rice-fish research and development in Asia. ICLARM Conf. Proc. No. 24. Manila (Philippines): International Center for Living Aquatic Resources Management. Dobermann A, Cassman KG, Peng S, Pham Sy Tan, Cao Van Phung, Sta. Cruz PC, Bajita JB, Adviento MAA, Olk DC. 1996a. Precision nutrient management in intensive irrigated rice systems. In: Attanandana T, Kheoruenromne I, Pongsakul P, Vearasilp T, editors. Proceedings of the International Symposium on Maximizing Sustainable Rice Yields Through Improved Soil and Environmental Management. Khon Kaen, Thailand. p 133-154. Dobermann A, Cassman KG, Mamaril CP, Sheehy JE. 1998. Management of phosphorus, potassium, and sulfur in intensive, irrigated lowland rice. Field Crops Res. 56:113-138. Dobermann A, Cassman KG, Sta. Cruz PC, Adviento MAA, Pampolino ME 1996b. Fertilizer inputs, nutrient balance, and soil nutrient-supplying power in intensive, irrigated rice systems. III. Phosphorus. Nutr. Cycl. Agroecosys. 46:111-125. Dobermann A, Sta. Cruz PC, Cassman KG. 1995. Potassium balance and soil potassium supplying power in intensive, irrigated rice ecosystems. In: Potassium in Asia. Proc. of the 24th International Colloquium of the International Potash Institute, 21-24 Feb. 1995, Chiang Mai, Thailand. Basel (Switzerland): IPI. p 199-234. Dobermann A, Sta. Cruz PC, Cassman KG. 1996c. Fertilizer inputs, nutrient balance, and soil nutrient-supplying power in intensive, irrigated rice systems. I. Potassium uptake and K balance. Nutr. Cycl. Agroecosys. 46:l-10. Flinn JC, De Datta SK. 1984. Trends in irrigated rice yields under intensive cropping at Philippine research stations. Field Crops Res. 9:l-15. George T, Ladha JK, Buresh RJ, Garrity DP. 1992. Managing native and legume-fixed nitrogen in lowland rice-based cropping systems. Plant Soil 141:69-91. George T, Ladha JK, Buresh RJ, Garrity DP. 1993. Nitrate dynamics during the aerobic soil phase in lowland rice-based cropping systems. Soil Sci. Soc. Am. J. 57:1526-1532. George T, Ladha JK, Garrity DP, Buresh RJ. 1994. Legumes as “nitrate catch” crops during the dry-to-wet transition in lowland rice cropping systems. Agron. J. 86:267-273. George T, Ladha JK, Ganity DP, Torres RO. 1995. Nitrogen dynamics of grain legume-weedy fallow-flooded rice sequences in the tropics. Agron. J. 87:l-6. George T, Buresh RJ, Ladha JK, Punzalan G. 1998. Recycling in situ of legume-fixed and soil nitrogen in tropical lowland rice. Agron. J. (In press.) Hobbs PR, Harrington LW, Adhikary C, Gin GS, Upadhyay SR, Adhikary B. 1996. Wheat and rice in the Nepal Tarai: farm resources and production practices in Rupandehi district. NARC, International Maize and Wheat Improvement Center (CIMMYT), Mexico, D.F. Inubushi K, Watanabe I. 1986. Dynamics of available nitrogen in paddy soils. II. Mineralized N of chloroform-fumigated soil as a nutrient source for rice. Soil Sci. Plant Nutr. 32:561577.

Intensification of rice production systems: opportunities and limits

141

Kawaguchi K, Kyuma K. 1977. Paddy soils in tropical Asia: their material nature and fertility. Honolulu, Hawaii (USA): The University Press of Hawaii. Kemmler G. 1980. Potassium deficiency in soils of the tropics as a constraint to food production. In: Priorities for alleviating soil-related constraints to food production in the tropics. Manila (Philippines): International Rice Research Institute. p 253-275. Kimura M, Watanabe I, Patcharapreetcha P, Panichsapatana S, Wada H, Takai Y. 1990. Quantitative estimation of decomposition process of plant debris in paddy field. Soil Sci. Plant Nutr. 36:33-42. Kirk GJD, Begg CBM, Solivas JL. 1993. The chemistry of the lowland rice rhizosphere. Plant Soil. 155/156:83-86. Lee KE. 1994. The diversity of soil organisms. In: Hawksworth DL, editor. The biodiversity of microorganisms and invertebrates: its role in sustainable agriculture. Wallingford (UK): CAB International. p 73-87. Mam DB, Gilmour JT, Scott HD, Ferguson JA. 1988. Effects of long-term water management in a humid region on spatial variability of soil chemical status. Soil Sci. 145:188-193. Matson PA, Parton WJ, Power AG, Swift MJ. 1997. Agricultural intensification and ecosystem properties. Science 277:504-509. Mohanty SK, Mandal LN. 1989. Transformation and budgeting of N, P and K in soils for rice cultivation. Oryza 26:213-231. Moms RA, Furoc RE, Dizon MA. 1986. Rice responses to a short-duration green manure. I. Grain yield. Agron. J. 78:409-412. Nambiar KKM, Abrol IP. 1989. Long-term fertilizer experiments in India: an overview. Fert. News. 34(4):11-20. Nambiar KKM. 1995. Long-term experiments on major cropping systems in India. In: Barnett V, Payne R, Steiner R, editors. Agricultural sustainability in economics, environmental, and statistical terms. London: John Wiley & Sons. Nannipieri P, Badalucco L, Landi L. 1994. Holistic approaches to the study of populations, nutrient inputs and fluxes: limitations and future research needs. In: Ritz K, Dighton J, Giller JSE, editors. Beyond the biomass. Chichester: John Wiley. p 231-246. Oberthuer T, Dobermann A, Neue HU. 1996. How good is a reconnaissance soil map for agronomic purposes? Soil Use Manage. 12:33-43. Olk DC, Cassman KG. 1995. Characterisation of two chemically extracted humic acid fractions in relation to nutrient availability. In: Lefroy RDB, Blair GJ, Craswell ET, editors. Soil organic matter management for sustainable agriculture. ACIAR Proceedings No. 56. Ubon, Thailand, 24-26 Aug 1994. ACIAR, Canberra, Australia. Olk DC, Cassman KG, Randall EW, Kinchesh P, Sangers LJ, Anderson JM. 1996. Changes in chemical properties of organic matter with intensified rice cropping in tropical lowland soil. Eur. J. Soil Sci. 47:293-303. Ottow JCG, Benckiser G, Watanabe I. 1981. Iron toxicity as a multiple nutritional stress of rice. Giessener Beitrage zur Entwicklungsforschung, I. 7:203-215. Palm CA, Sanchez PA. 1991. Nitrogen release from the leaves of some tropical legumes as affected by their lignin and polyphenolic contents. Soil Biol. Biochem. 23:83-88. Peng S, Garcia FV, Laza RC, Sanico AL, Visperas RM, Cassman KG. 1996. Increased N-use efficiency using a chlorophyll meter on high-yielding irrigated rice. Field Crops Res. 47:243-252. Peng KL, Yi FX. 1992. The selection and application of potassium-rich green manure. In: Chen L, editor. Proceedings of studies on green manure in China. Shenyang (China): Liaoning University Press. p 66-69.

142

Reichardt et al

Pingali PL, Hossain M, Pandey S, Price L. 1998. Economics of nutrient management in Asian rice systems: towards increasing knowledge intensity. Field Crops Res. 56:157-176. Ponnamperuma FN. 1972. The chemistry of submerged soils. Adv. Agron. 24:29-96. Ponnamperuma FN. 1984. Effects of flooding on soils. In: Kozlowski TT, editor. Flooding and plant growth. Orlando, Fla. (USA): Academic Press. p 9-45. Prasad ASR, Prasad GSV, Muralidharan K. 1995. Fragile ecosystems for rice cultivation? In: Fragile lives in fragile ecosystems. Manila (Philippines): International Rice Research Institute. p 369-393. Reichardt W. 1986. Enzymatic potential for decomposition of detrital biopolymers in sediments from Kiel Bay. Ophelia 28:369-384. Reichardt W. 1996. Ecotoxicity of certain heavy metals affecting bacteria-mediated biogeochemical pathways in sediments. In: Calmano W, Foerstner U, editors. Sediments and toxic substances: environmental effects and ecotoxicity. Berlin: Springer Verlag. p 159-178. Reichardt W, Heise S, Piker L. 1993. Ecotoxicity testing of heavy metals using methods of sediment microbiology. Environ. Toxicol. Water Qual. 8:299-311. Reichardt W, Briones A, Padre B, de Jesus R, Mascariña G. 1996b. Dynamics of soil microbial communities and sustainable nutrient supply in highly intensified rice cultivation. In: Attanandana T, Kheoruenromne I, Pongsakul P, Vearasilp T, editors. Proceedings of the International Symposium on Maximizing Sustainable Rice Yields Through Improved Soil and Environmental Management. Khon Kaen, Thailand. p 887-896. Reichardt W, Mascariña G, Padre B, Doll J. 1997. Microbial communities of continuously cropped irrigated rice fields. Appl. Environ. Microbiol. 63:233-238. Roger PA. 1996. Biology and management of the floodwater ecosystem in ricefields. Manila (Philippines): International Rice Research Institute, and France: ORSTOM. Saito M, Watanabe I. 1978. Organic matter production in rice field floodwater. Soil Sci. Plant Nutr. 28:427-440. Sanchez PA. 1976. Properties and management of soils in the tropics. New York (USA): John Wiley. Sanchez PA, Logan TJ. 1992. Myths and science about the chemistry and fertility of soils in the tropics. In: Lal R, Sanchez PA, editors. Myths and science of soils of the tropics. SSSA Spec. Publ. 29. Madison, Wis. (USA): ASA, SSSA. p 35-46. Schink B. 1988. Principles and limits of anaerobic degradation: environmental and technological aspects. In: Zehnder AJB, editor. Biology of anaerobic microorganisms. New York (USA): John Wiley. Schoenly KG, Cohen JE, Heong KL, Arida AS, Bamon AT, Litsinger JA. 1996a. Quantifying the impact of insecticides on food web structure of rice-arthropod populations in a Philippine farmer’s irrigated field: a case study. In: Polis G, Winemiller K, editors. Food webs: integration of patterns and dynamics. New York (USA): Chapman and Hall. p 343-351. Schoenly KG, Cohen JE, Heong KL, Litsinger JA, Aquino GB, Barrion AT, Arida G. 1996b. Food web dynamics of irrigated rice fields at five elevations in Luzon, Philippines. Bull. Entomol. Res. 86:451-466. Singh Y, Khind CS, Singh B. 1991. Efficient management of leguminous green manures in wetland rice. Adv. Agron. 45:135-189. Sri Adiningsih J, Santoso D, Sudjadi M.1991. The status of N, P, K and S of lowland rice soils in Java. In: Blair G, editor. Sulfur fertilizer policy for lowland and upland rice cropping systems in Indonesia. Melbourne (Australia): Australian Centre for International Agricultural Research. p 68-76.

Intensification of rice production systems: opportunities and limits

143

Swaminathan MS. 1991. Sustainable agriculture: need for conceptual clarity. In: Hawksworth DL, editor. The biodiversity of microorganisms and invertebrates: its role in sustainable agriculture. Wallingford (UK): CAB International. p 273-287. Tiwari KN. 1985. Changes in potassium status of alluvial soils under intensive cropping. Fert. News 30:17-24. Torsvik V, Sake K, Sorheim R, Goksoyr J. 1990b. Comparison of phenotypic diversity and DNA heterogeneity in a population of soil bacteria. Appl. Environ. Microbiol. 56:776781. Tsutsuki K, Ponnamperuma FN. 1987. Behaviour of anaerobic decomposition products in submerged soils. Soil Sci. Plant Nutr. 33:13-33. Tunlid A, White DC. 1992. Biochemical analysis of biomass, community structure, nutritional status, and metabolic activity of microbial communities in soil. In: Stotzky G, Bollag JM, editors. Soil biochemistry. Vol. 7. New York (USA): Marcel Dekker. p 229-262. Uexkuell HR von. 1985. Availability and management of potassium in wetland rice soils. In: Wetland soils: characterization, classification and utilization. Manila (Philippines): International Rice Research Institute. p 293-305. Uexkuell HR von, Beaton JD. 1992. A review of fertility management of rice soils. In: Kimble JM, editor. Characterization, classification, and utilization of wet soils. Proceedings of the 8th International Soil Correlation Meeting (VIII. ISCOM). Soil Conservation Service. Lincoln, Neb. (USA): United States Department of Agriculture. p 288-300. Uexkuell HR von, Mutert E. 1995. Global extent, development and economic impact of acid soils. Plant Soil 171:l-15. Ventura W, Watanabe I. 1993. Green manure production of Azolla microphylla and Sesbania rostrata and their long-term effects on rice yields and soil fertility. Biol. Fert. Soils 15:241248. Ye W, Wen QX. 1991. Characteristics of humic substances in paddy fields. Pedosphere 1:229239. Yoshida S. 1981. Fundamentals of rice crop science. Manila (Philippines): International Rice Research Institute. Zak JC, Willig MR, Moorhead DL, Wildman HG. 1994. Functional diversity of microbial communities: a quantitative approach. Soil Biol. Biochem. 26: 1101-1108. Notes Authors’ address: International Rice Research Institute (IRRI), P.O. Box 933, 1099 Manila, Philippines. [email protected] [email protected] [email protected] Acknowledgments: We would like to thank J.C.G. Ottow, Giessen, Germany, and W.H. Patrick Jr., Baton Rouge, Louisiana, USA, for reviewing an earlier draft. Citation: Dowling NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute.

144

Reichardt et al

CHAPTER 9

Importance of rice pests and challenges to their management M.B. Cohen, S. Savary, N. Huang, O. Azzam, and S.K. Datta

New technologies and a greater understanding of the rice ecosystem are contributing to more effective and sustainable pest management in farmers’ fields. Prioritizing research on rice pests (insects, plant diseases, and weeds) has been made difficult by a lack of systematic survey data on pest losses in different ecosystems and under different production conditions. To bridge this knowledge gap, work in progress is quantifying risk probability and risk magnitude of pest injuries. Surveys in farmers’ fields have been conducted at hundreds of sites in several countries to quantify the risk probability for various pests. Experiments at IRRI have manipulated pest levels under varying production situations to quantify the magnitude of yield loss across these conditions. Researchers are applying biotechnology to produce rice varieties with improved resistance to insects and diseases. Marker-aided selection can improve the efficiency of rice breeding, and be used to “pyramid” multiple genes for resistance to a given pest. Plant transformation enables us to introduce novel resistance genes from any organism into rice. Varietal resistance to pests has many desirable features, such as environmental safety and convenience for farmers, but has suffered from a lack of durability as pest populations adapt to new resistant varieties. Researchers are using DNA fingerprinting to enhance understanding of pest population genetics and behavioral studies of insects to develop resistance management strategies for the sustainable use of resistant cultivars in farmers’ fields.

In agriculture, pests (or biotic constraints) can be defined as organisms that cause economic loss. Among the pests that attack rice are insects, microorganisms (viruses, bacteria, and fungi) that cause plant disease, weeds, and even vertebrates such as rats and birds. Pest management has been a dynamic area of research at IRRI since its establishment, driven by the advent of new technologies and improvements in understanding of the rice ecosystem. The roles of two new technologies in pest management, marker-aided selection and genetic engineering, are covered later in this chapter. We also discuss two examples of improved ecosystem understanding: quantification of pest-associated yield losses under different crop production conditions and new approaches to the sustainable use of pest-resistant rice varieties. An ecosystem component now recognized to be of tremendous importance, the beneficial arthropods and microorganisms that feed upon or compete with pest organisms, is reviewed by

145

Schoenly et al (this volume, Chapter 17) in their paper on arthropod biodiversity and rice landscapes. The goal of pest research at IRRI is to provide technologies and knowledge that contribute to integrated pest management (IPM) in farmers’ fields (IRRI 1994). IPM has become a term with diverse meanings (Waage 1996). As used by IRRI, it refers to achieving low and stable pest populations and reducing chemical pesticide use by improving farmer understanding of the crop ecosystem and combining biological, cultural, and chemical tactics. Pesticides can be major expenses for resource-poor farmers, are often hazardous to the environment and human health (Pingali and Roger 1995), and can exacerbate pest problems by disrupting naturally occurring biological controls (Way and Heong 1994).

Importance of rice biotic constraints and prioritizing research for their management This section addresses a number of questions. How important are rice biotic constraints under current agricultural scenarios? How reliable is our assessment of these constraints? What are the implications of foreseeable agricultural changes for the importance of rice biotic constraints? What are the avenues to both predict and manage these constraints in yet-to-come production situations in a sustainable way? All these questions cannot, of course, be answered in detail here. Rather, this section tries to bring into perspective the close association between changes in production situations and damage caused by rice pests. Such a link implies that the introduction of new agricultural technologies — changes in production situations (De Wit 1982) — will have an effect, positive or negative, on damage from pests. One avenue that we offer for addressing this issue is risk analysis, similar in principle to the approach used in industry (Rowe 1980). Pest populations building up in crops may have economic, social, and political consequences (Zadoks and Schein 1979). These consequences stem from the diversity of effects or injuries caused by pests (Zadoks 1967): direct losses (in yield, in quality, or costs of replanting) or indirect losses (at the farm, community, or consumer level). Measurement of yield losses therefore only provides a limited view of the impact of pests on crops and societies. Yield loss, however, is associated with a comparatively precise and simple operational definition. Quantitative information on yield losses from pests is necessary to develop policies, to set research priorities, to assess the progress made in protecting crops, and to develop efficient IPM schemes (Zadoks and Schein 1979, Teng 1983). Such information represents level 1 of a process leading to the implementation of a systems approach in pest management (Teng and Savary 1992). Yield loss data attributable to pests are thus all the more necessary when agricultural systems are undergoing rapid and important transformations, such as the rice-based cropping systems of tropical Asia (Hossain, this volume, Chapter 3), so that the risk associated with such changes can be assessed from a plant protection viewpoint (Savary et al 1997).

146

Cohen et al

A review of the literature on rice diseases and insect pests was done over the period 1960-93 using five criteria: (1) reports address rice production in tropical Asia, (2) their main objective is to measure yield loss, (3) they provide descriptions of experimental and sampling designs, (4) they describe the techniques used for both manipulating disease levels (if applicable) and measuring yield variation, and (5) they provide quantitative information on yield losses. Reports were sorted according to the rice ecosystem involved (irrigated, rainfed lowland, flood-prone, and upland; Khush 1984), and ranked by their representativeness with respect to space, time, scale, and injury. Assessments of representativeness of yield loss data (James 1974, Madden 1983) attributable to rice pests over time, space, and scale were based on the proportion of studies conducted over more than 1 year, on the proportion of studies conducted at more than one location, and on the proportion of studies conducted at the plot (>1 m2) or field level, respectively. Representativeness of injury was judged much more difficult to assess. The standard deviation of the proportion of studies using inoculations, spontaneous infection, or chemical control was used as an index. A low standard deviation in one group of studies (e.g., yield losses caused by bacterial diseases) would indicate flexibility in addressing a particular issue and a balance among approaches. The main result of this review is the surprisingly limited number of published reports that we can rely on. Considerable discrepancies have also been found among rice ecosystems in a number of studies, as most of them concentrated on the irrigated ecosystem. Most studies conducted in this ecosystem, however, were conducted at one location and in one season, whereas many studies in the other three ecosystems pertain to several locations in two or more seasons. We need to better document yield losses in ecosystems other than those in the irrigated one. The potential for extrapolation of results in the irrigated ecosystem deserves consideration, and the representativeness of studies conducted in other ecosystems cannot compensate for their small numbers. Perhaps, more importantly, improving the representativeness of yield loss data in all four ecosystems is necessary to better define the needs of rice production systems.

A risk-analytical approach for setting priorities The dynamics of harmful agents may lead to injury—visible signs of their biological activity on the standing crop. Injury may lead to damage and yield loss. Damage may or may not, in turn, lead to yield loss and a reduction of crop value in economic terms (Zadoks 1985). Our focus is on damage, which closely depends on injury via a damage function, which in turn may affect losses via a loss function. Changes in patterns of cropping practices (e.g., inputs) may dramatically alter the physiological reaction of a crop to injury, and therefore the shape of the damage function. Similarly, the occurrence of two different injuries, simultaneously or in sequence, may also modify the damage function. As a result, the damage function, which is the basis of the threshold theory (Zadoks 1985) in plant protection, is very complex, being a product of numerous processes. Changes in patterns of cropping practices are also known to strongly

Importance of rice pests and challenges to their management

147

influence population dynamics, and therefore injuries (Teng and Savary 1992, Savary et al 1994). The variation of damage with changing production situations and injury profiles was recently addressed using a risk-analytical approach (Savary et al 1997), which involves two steps: assessing the risk probability P (i.e., the probability of a given injury occurring in a given production situation) and determining the risk magnitude (i.e., the damage associated with that injury). Risk probability can be assessed from surveys in farmers’ fields, whereas risk magnitude can be measured in field experiments where both injuries and cropping practices are varied. The risk associated with a particular pest is then determined (Rowe 1980): R = P × M.

Risk probability: surveys of injuries in farmers' fields Survey procedures (Elazegui et al 1990, Pinnschmidt et al 1995) have been used in farmers’ fields of different countries—the Philippines, India (eastern Uttar Pradesh), Thailand, and Vietnam. The procedures entail quantification of injuries caused by insects, pathogens, and weeds, as well as a description of cropping practices. Figure 1 shows the strong variation in risk probabilities for nine injuries in a few selected rice production situations, especially for two diseases: sheath blight (ShB) or brown spot (BS). Rice tungro disease (RTD) is detected in one production situation at a low risk probability. Insect injuries (deadhearts, DH, and whiteheads, WH, caused by stemboring caterpillars; and whorl maggot, WM) are omnipresent, often with high risk probabilities. Weed infestation (WA, weed above the rice crop canopy, and WB, weed below the rice crop canopy) appears to be the most common constraint in all production situations, with mostly a high risk probability.

Risk magnitude: measuring yield losses in controlled experiments Over the past several years, IRRI has been conducting a series of experiments in which four input factors have been varied (potential yield of the rice cultivar, crop establishment method, water management, and fertilizer supply) and nine injuries have been manipulated (bacterial blight, RTD, BS, ShB, WM, DH, WH, WA, and WB). Each experiment includes noninjured controls for each pattern of cropping practice it addresses, which provide estimates of attainable yields (Ya), and therefore a means to empirically measure yield losses: YL = Ya - Y. The resulting experimental yield loss database covers a range of attainable yields from 1 to 11 t ha-1, reflecting the variation in production situations. These data were analyzed with multivariate techniques. One of the resulting empirical models shows significant interactions of Ya and injuries with yield loss variation. In other words, the same injury will have different consequences (Fig. 2), depending on the production situation considered (represented by Ya).

Risk estimates across production situations The empirical model was also used to estimate the risk magnitude (% yield loss) in several production situations. Two arbitrary injury levels (high and low) were also considered, based on the range of observed injuries. The resulting estimates for risk

148

Cohen et al

Fig. 1. Risk probability: probability of occurrence of nine injuries in selected production situations in four countries. Only four examples are shown. Each histogram represents a particular production situation at a site or in a country (e.g., CL3, Central Luzon). The horizontal axis indicates injuries: ShB = sheath blight, BLB = bacterial leaf blight, RTD = rice tungro disease, BS = brown spot, WM = whorl maggot, DH = deadhearts, WH = whiteheads, WA = weed above the rice crop canopy, WB = weed below the rice crop canopy. The vertical axis indicates the proportion (0 to 1) of affected fields.

Importance of rice pests and challeneges to their managament

149

Fig. 2. Damage functions for three injuries: weeds above the rice crop canopy, sheath blight injury, and rice tungro disease. (A) Effects of weeds above the crop canopy (area under the disease progress curve in % days). (B) Effects of sheath blight (maximum severity in %). (C) Effects of RTD injury (% area affected × symptom score). Increasing injury levels are indicated on the horizontal axes. Increasing damage is indicated on the vertical axes in relative terms (%). Damage functions are shown for a range of attainable yields, from 2 t ha -1 (Ya2) to 11 t ha-1 (Ya11).

magnitude were then multiplied by the estimates for risk probabilities in the corresponding production situations (Fig. 2) to produce estimates for risk. Figure 3 shows the considerable variation in risks depending on injuries and production situations.

Revisiting the concepts of threshold and crop loss profile The concepts of injury profile (Pinstrup-Andersen et al 1976) and thresholds (Stem 1973) are of central importance to plant protection. In the context of changes in production situations, these concepts need to be adapted to account for variations in attainable yield and allow examination of injury combinations. One useful approach is to consider the yield of a crop as a response surface (Teng and Gaunt 1980). The risk-analytical approach illustrates well, in an empirical way, the fact that changes in production situations must be factored in when setting priorities for pest management. This approach implies the same principles as a systems-analytical one, in which, according to Rabbinge (1993): • potential yield is defined by factors such as crop genotype, radiation, or temperature; • potential yield is then limited to an attainable yield by factors such as water and nutrient availability; and • attainable yield is reduced in turn by factors such as injuries. Both the systems-analytical (Elings and Rubia 1994) and the risk-analytical approaches are used at IRRI. Their combination might provide a solid empirical basis to model extrapolations.

150

Cohen et al

Code for production situation Site(s), country Mean yield and (minimum-maximum values) Cropping season Crop establishment method Fertilizer supply Weed management level Water management level Other distinctive attributes of the production situation

• • • • • •

CL3 Central Luzon, Philippines Mean yield = 5.3 t ha -1 (3.0–7.3) Dry season Direct seeded High fertilizer Good weed management (herbicides) Good water management

• • • • •

IN6 Uttar Pradesh, India Mean yield = 4.6 t ha -1 (2.5–6.2) Rainy season Transplanted No/low fertilizer Good weed management Poor water management Previous crop: fallow

• • • • • •

VT1 Thailand and Vietnam Mean yield = 3.5 t ha -1 (2.1–5.8) Rainy season Direct seeded (predominant) Medium/low fertilizer Good weed management Moderate water management

VT2 Eastern Thailand Mean yield = 3.1 t ha -1 (1.9–4.4) Rainy season Transplanted or direct seeded Medium to low fertilizer Poor weed management Poor water management

Fig. 3. Estimated risk magnitudes in selected production situations in four countries. The horizontal axis indicates injuries caused by ShB (sheath blight), BLB (bacterial leaf blight), RTD (rice tungro disease), BS (brown spot), WM (whorl maggot), DH (deadhearts), WH (whiteheads), WA (weed above the rice crop canopy), and WB (weed below the rice crop canopy). For each injury, two levels, high (H) or low (L), are shown. (Adapted from Savary et al 1997.)

Importance of rice pests and challenges to their management

151

New directions for host-plant resistance to diseases and insects in rice For thousands of years, farmers have recognized that some crop varieties produce a larger yield of good quality than other varieties, under conditions of similar insect and pathogen populations and other environmental factors. In the modern discipline of host-plant resistance (HPR), plant breeders, entomologists, and plant pathologists identify genes that confer pest resistance and introduce them into suitable agronomic backgrounds. HPR is a key component of IPM systems in rice and has been a focus of research at IRRI almost since the founding of the institute. Multiple pest resistance has been a feature of IRRI varieties released since the 1960s, and this has made immense contributions to increased and stabilized yields (Khush 1995). The use of rice germplasm as a source of genes for pest resistance is reviewed by Bellon et al elsewhere in this book (Chapter 16). Here we review new approaches to increase the efficiency of breeding for pest resistance, introduce resistance genes from outside the rice gene pool, and enhance the durability of resistant varieties in farmers’ fields.

DNA marker-assisted selection Marker-assisted selection (MAS) enables plant breeders to improve the efficiency of breeding when an important trait, which is difficult to assess, is tightly linked to a trait that is easily measured. Although the development of molecular biology and DNAbased markers has vastly expanded the potential of MAS in plant breeding, breeders have for many years also used morphological markers. For example, a gene for resistance to brown planthopper (BPH) is closely linked to a gene specifying purple coleoptile color in some traditional rice varieties grown in northeast India. When a resistant plant with a purple coleoptile is crossed with a susceptible plant with a green coleoptile, more than 95% of the F2 plants showing purple coleoptile are also resistant to BPH. In this case, coleoptile color is a morphological marker that is used to help select for BPH resistance. Unfortunately, few morphological markers are known. They tend to be particular to certain rice varieties, and most morphological markers are mutations that are deleterious to rice plants. A homozygous locus is indistinguishable from a heterozygous one when a dominant allele is involved. Moreover, the usefulness of the approach is limited to traits controlled by single major genes; it does not apply to many agronomically important traits that are governed by many unlinked genes. The advent of molecular markers has enormously increased the power of MAS. The most commonly used DNA markers are restriction fragment length polymorphism (RFLP) markers. Other kinds of DNA markers have been developed recently. In MAS, target genes are detected based on the genotype as determined by the DNA markers and not on the phenotypic expression of the genes. Figure 4 illustrates the principle and genetic basis of marker-assisted identification of a target gene. We assume that a locus on a rice chromosome is responsible for a character such as resistance to blast. The donor parent carries a resistance allele (R) linked to a DNA marker allele (m) and the recipient parent carries a susceptible allele

152

Cohen et al

Fig. 4. Schematic diagram of genetic basis of marker-assisted selection with codominant DNA markers.

Importance of rice pests and challenges to their management

153

(S) linked to a DNA marker allele (M). The genetic distance between the DNA marker locus and the gene locus is the recombinant frequency (r in Fig. 4, r = 0.05) as determined when the gene was first mapped. After crossing the donor parent with the recipient, we obtain an F1 hybrid that is heterozygous in both the target resistance gene and the DNA marker locus as indicated by the DNA banding pattern in gel analysis. The selfing of the F1 hybrid produces the segregating F2 population. Based on the banding pattern of DNA markers, individuals of the segregating population can be classified into three groups (MM, Mm, and mm). Within the mm group, the majority of the plants carry the resistance allele (R) because of linkage. The smaller the r value (the distance between the marker and target gene), the higher the proportion of the plants in the mm group carrying the R allele. Therefore, selection based on the DNA marker permits selection of the target gene unless the selected individual carries a recombinant chromosome. If the recombination frequency between the DNA marker and the target gene does not change from gene mapping population to breeding population, a relation discussed later, we would be able to select the target gene based on the DNA marker with a predictable rate of accuracy. The MAS technique has many advantages in rice breeding in that it can be used at any time and at any growth stage of rice. This advantage is obvious when there are two or three breeding seasons in a year, but the pest can only be collected and analyzed once a year. Selection of minor genes (quantitative trait loci) is difficult with the conventional approach because of epistasis of gene actions and environmental effect. Identification of target genes by markers can avoid these problems. IRRI’s success in using MAS to pyramid genes for bacterial blight (BB) resistance demonstrates the power of MAS in improving breeding efficiency. Bacterial blight caused by Xanthomonas oryzae pv. oryzae is one of the most destructive diseases of rice throughout the world, but has been successfully controlled in many areas through the deployment of resistant varieties (Khush et al 1989). Nineteen rice genes conferring resistance to BB have been identified (Kinoshita 1995), several of which have been incorporated into modern rice varieties. The Xa-4 gene has been of particular importance, but large-scale and long-term cultivation of varieties carrying Xa-4 in Indonesia, India, China, and the Philippines has led to a significant shift of the dominant BB host races in these countries (Mew et al 1992). In many areas, rice varieties with Xa-4 have become susceptible to BB. One way to delay such a breakdown in BB resistance is to pyramid multiple resistance genes into rice varieties. This method, however, can be difficult or impossible with a conventional approach because of epistasis of gene actions, particularly when a breeding line already carries a gene such as Xa-21, which shows resistance to all BB races when the varieties are being developed. With a conventional approach, a breeding line with Xa-21 only cannot be distinguished from a breeding line with Xa-21 plus some other genes. DNA markers were used to assist in the pyramiding of four BB resistance genes (Huang et al 1997). All possible combinations of the four resistance genes were obtained (Table 1). The pyramided lines show a wider spectrum or higher level of resis-

154

Cohen et al

Table 1. Plants containing two or more bacterial blight resistance genes were selected based on DNA marker analysis (modified from Huang et al 1997). Race Lines

IRBB4 IRBB5 lR66999-1-1-5-2 lR66700-3-3-3-4-2 IR24 IRBB50 IRBB51-1 IRBB51-2 IRBB51-3 IRBB51-4 IRBB52 IRBB53-1 IRBB53-2 IRBB53-3 IRBB53-4 IRBB54-1 lRBB54-2 IRBB54-3 IRBB55-1 IRBB55-2 IRBB55-3 IRBB55-4 IRBB56-1 IRBB56-2 IRBB57-1 IRBB57-2 lRBB57-3 IRBB58-1 lRBB58-2 IRBB58-3 IRBB59-1 IRBB59-2 IRBB59-3 IRBB60-1 IRBB60-2

Gene combinations

Xa-4 xa-5 xa-13 Xa-21 – Xa-4/xa-5 Xa-4/xa-13 Xa-4/xa-13 Xa-4/xa-13 Xa-4/xa-13 Xa-4/Xa-21 xa-5/xa-13 xa-5/xa-13 xa-5/xa-13 xa-5/xa-13 xa-5/Xa-21 xa-5/Xa-21 xa-5/Xa-21 xa-13/Xa-21 xa-13/Xa-21 Xa-21/xa-13 xa-13/Xa-21 Xa-4/xa-S/xa-13 Xa-4/xa-5/xa-13 Xa-4/xa-5/Xa-21 Xa-4/xa-5/Xa-21 Xa-4/xa-5/Xa-21 Xa-4/xa-13/Xa-21 Xa-4/xa-13/Xa-21 Xa-4/xa-13/xa-21 xa-5/xa-13/Xa-21 xa-5/xa-13/Xa-21 xa-5/xa-13/Xa-21 Xa-4/xa-5/xa-13/Xa-21 Xa-4/xa-5/xa-13/Xa-21

1

2

3

4

5

6

Ra R S R S R+ R R R R R+ R R R R R+ R+ R+ R R R R R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+

S R S R S R S S S S R R R R R R+ R+ R+ R R R R R R R+ R+ R+ R R R R+ R+ R+ R+ R+

S R S R S R S S S S R R R R R R+ R+ R+ R R R R R R R+ R+ R+ R R R R+ R+ R+ R+ R+

S MS S R S R R R R R R R R R R R R R R R R R R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+

R R S R S R+ R R R R R+ R R R R R+ R+ R+ R R R R R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+ R+

S S R R S S R R R R R R R R R R R R R R R R R R R R R R R R R R R R R

a R = resistant, S = susceptible, MS = moderately susceptible, R+ = highly resistant.

tance to the bacterial pathogen. This effect can be seen from the pyramided lines carrying xa-4/xa-13. The variety IRBB4, carrying Xa-4, was resistant to races 1 and 5 but susceptible to other races. On the other hand, IR66699, carrying xa-13, was resistant to race 6 only. The lines with both Xa-4 and xa-13 showed resistance to races 1, 5, and 6 as did their parents. Furthermore, these pyramided lines showed resistance to race 4, which can infect both parents (IRBB4 and IR66699).

Importance of rice pests and challenges to their management

155

Rice transformation Although techniques such as embryo rescue have widened the pool of germplasm available for improvement of cultivated rice, several important insects and diseases remain for which sources of resistance have not been found in the genus Oryza. Among these pests are ShB and the complex of caterpillar pests known as stem borers. Even for genes that occur in the rice gene pool, plant transformation may be preferable to conventional backcrossing of resistance genes into elite cultivars, for example, by allowing the genes to be introduced without disrupting complex genetic traits such as grain quality, or by increasing the level of expression of existing resistance genes. These applications of genetic engineering can be illustrated by recent achievements in resistance to ShB, BB, and stem borers. Lin et al (1995) transformed the indica variety Chinsurah Boro II with a rice chitinase gene. The transgenic plants express the chitinase gene constitutively, rather than only after fungal infection of the plant has taken place, as do normal rice plants. In greenhouse tests, the transgenic plants show enhanced resistance to the ShB pathogen, Rhizoctonia solani. Song et al (1995) cloned a gene conferring resistance to another rice disease, BB, from an African species of wild rice, Oryza longistaminata. This gene, Xa-21, has been shown to confer resistance to BB when genetically engineered into IR64 (Song et al 1995) and IR72 (Tu et al 1998). Several rice varieties have now been transformed with toxin genes from Bacillus thuringiensis ( Bt ), and been shown to have enhanced resistance to stem borers (e.g., Fujimoto et al 1993, Wunn et al 1996, Ghareyazie et al 1997, Wu et al 1997, Datta et al 1998). The production of these initial transgenic lines has been important in demonstrating that the foreign gene “constructs” used in their transformation can function well in rice. Larger numbers of transgenic lines are now being produced at IRRI and numerous other institutions, and are being screened in containment greenhouses to identify those that perform best. Some of the best lines will eventually be evaluated under field conditions. Field tests of transgenic rice have already begun in China, and by the year 2000 field tests will likely be under way in several other Asian countries. Three methods have been used successfully for rice transformation (Fig. 5): protoplast transformation, particle bombardment, and Agrobacterium-mediated transformation. Protoplasts are plant cells freed of their cell wall by enzymatic digestion. Protoplasts can uptake foreign DNA after treatment with polyethylene glycol, a neutral polymer, or application of an electric current in a process known as electroporation. In the biolistic method, also known as particle bombardment, DNA associated with tiny gold particles is shot into cells with a burst of high pressure. Agrobacteriummediated transformation makes use of a species of plant parasitic bacterium, Agrobacterium tumefaciens, that harbors a virus capable of inserting its DNA into plant chromosomes. Progress in the efficiency of transformation and tissue culture regeneration is still needed, particularly for indica varieties.

156

Cohen et al

Fig. 5. Protocol for production of fertile transgenic rice plants using biolistic and protoplast systems. (Modified after Datta 1995.)

Sustainable deployment of pest-resistant rice cultivars Plant pathogens and insect pests have demonstrated an impressive capacity to adapt to resistant cultivars. In addition to the inherent genetic potential of pests to respond to selection imposed by resistance genes, the “breakdown” of resistance in many rice varieties has been accelerated by the release of cultivars containing simple genetic resistance backgrounds (often a single major gene for each target pest) and by the

Importance of rice pests and challenges to their management

157

deployment of some of these cultivars as monocultures over vast areas. Resistance breakdown has led to episodes of yield instability and to a continuing need to identify new resistance genes and incorporate them into new varieties. Novel approaches to improve the durability of resistance to insects and diseases are being pursued at IRRI, and are as relevant to genetically engineered cultivars as they are to conventional ones. These approaches include increasing the genetic complexity of resistance to particular pests and the strategic deployment of cultivars in farmers’ fields. Diseases. With advances in molecular genetics, new tools have become available to better understand and deploy resistance genes for rice diseases (McCouch et al 1988, Hamer 1991, Leach et al 1992). One new approach relies on analyzing the genetics of resistance in traditional cultivars that have demonstrated durable resistance in farmers’ fields and identifying “gene tags” that can be used to incorporate resistance from such cultivars into modern varieties. This information can lead to the strategic deployment of a diversity of resistance genes either within fields or among fields, as an alternative to the large-scale cultivation of varieties with single resistance genes. Tagging resistance genes permits us to develop sets of near-isogenic lines (NILs) in which different resistance genes are introduced into a common genetic background. This allows us to characterize individual genes and to pyramid them (based on the spectrum of their resistance to various pathogen populations) in a marker-assisted breeding program (Table 1) or release the NILs as multilines (mixtures of cultivars that are genetically similar except for their disease resistance genes). Varietal mixtures have been deployed successfully in various crop systems (Wolfe 1985, Schaffner et al 1992). Analysis of the genetic basis of durable resistance to the blast fungus (Pyricularia grisea) in Moroberekan, a traditional West African upland rice cultivar, showed that the resistance consists of multiple major and minor genes (Mackill and Bonman 1992). Characterization of these genes from recombinant inbred populations led to the production of NILs that carry them separately (Wang et al 1994). These NILs are now being used to dissect the effects of major and minor genes for disease resistance and to evaluate the intrafield diversification deployment strategy to control rice blast (Chen, Zeigler, and Nelson, unpublished data). In the upland rice ecosystem, planting mixtures of rice cultivars in a field has been a traditional practice (Bonman et al 1986). Durable resistance to BB in China, Indonesia, and the Philippines has also been attributed to a complex (quantitative) resistance in traditional cultivars (Lee et al 1989, Mew et al 1992). The availability of NILS for BB (Ogawa 1993, Ikeda et al 1990) has allowed scientists to further characterize these genes and to identify gene tags that are useful for selection in a breeding program (Yoshimura et al 1995, McCouch et al 1991, Ronald et al 1992). Further testing of the spectrum of resistance for each of the identified resistance genes has helped researchers design a more targeted combination of genes in a pyramid breeding line. Experiments were conducted in farmers’ fields in the Philippines to evaluate the deployment of nine varietal combinations that included various resistance backgrounds to BB (pure stands, two-component mixtures, and two-gene pyramids). Initial results

158

Cohen et al

showed that both pyramids and partial resistance from “broken down” major genes reduced BB severity (Ahmed et al 1997). Cultivars with durable sources of resistance to blast and BB have been identified and the genetic basis of their durability is being studied. But no durable resistance has been reported for RTD. Most of the deployed cultivars carry a major gene for resistance to the vector, green leafhoppers. Breeding for resistance to RTD is further complicated by the fact that the disease is caused by two viruses—rice tungro spherical virus (RTSV) and rice tungro bacilliform virus (RTBV)—and one of them (RTBV) depends on the other for its transmission. Several NILS, carrying genes for resistance to the spherical virus, have been developed and are being characterized (Sebastian et al 1996, Ikeda and Imbe, unpublished data). Because RTBV transmission depends on RTSV, it has been difficult to screen for resistance to RTBV. Dasgupta et al (1991), however, showed that the cloned RTVB-G, strain can be singly inoculated into rice using agroinfection. This technique was used to confirm the previously identified tolerance (Ikeda and Imbe, unpublished data) in Utri Merah and Balimau Putih to RTBV (Dahal et al 1992, Sta Cruz and Assam, unpublished). Durable resistance depends not only on the inherent properties of the resistance in a cultivar but also on how and where the cultivar is grown. In South Sulawesi, Indonesia, genetic resistance to the vector of rice tungro viruses works in concert with appropriate planting time and other cultural practices for RTD management (Sama et al 1991). Insects. In 1996, farmers in the United States became the first to begin commercial production of crops genetically engineered with insecticidal toxins from the bacterium Bacillus thuringiensis (Bt ). Large numbers of “Bt rice” lines are under evaluation in containment greenhouse facilities at IRRI and other institutions, and smallscale field tests of some lines began in China in 1997 (Ye Gongyin, Zhejiang Agricultural University, personal communication). Once lines are identified that perform well in small-scale field tests, more multisite testing will probably be required by national seed boards, as is the case for all new varieties. Thus, it will be several years before Bt rice becomes available to farmers. The important potential benefits of Bt rice include reduced yield losses to stem-boring caterpillars and a reduction in chemical insecticide applications against these pests. But Bt toxins are insecticides and, like conventional chemical insecticides, insects may quickly adapt to them unless Bt plants are carefully designed and deployed. With the recent development and release of Bt crops, “resistance management” for transgenic insect-resistant crops has become a very active area of research (Gould 1996, 1998). One strategy that has been much discussed is combining two or more genes for insect resistance within a single cultivar. With almost 100 kinds of Bt toxins having been identified (Schnepf 1995), this at first seemed a promising approach for Bt crops. Because insects that carry mutations conferring resistance to a novel toxin are relatively rare when the toxins are first deployed, it seems reasonable to assume that insects resistant to two novel toxins will be much rarer. But toxin combinations will enhance durability only if mutations conferring resistance to one toxin do not confer resistance to the second, and it is now known that some insect mutations can confer

Importance of rice pests and challenges to their management

159

resistance to even highly divergent kinds of Bt toxins (Gould et al 1992). A greater assurance of durable resistance can be achieved if a Bt toxin is combined with a second, unrelated type of toxin. Consequently, there has been a great deal of privateand public-sector research on identifying toxins with the desirable characteristics of Bt toxins, such as high effectiveness against insect pests at low doses and an absence of mammalian toxicity. Some promising new toxins have been identified (Carozzi and Koziel 1997). Whether transgenic insect-resistant plants contain one toxin or multiple toxins, it is known that insect resistance to the toxins can be slowed by the use of refuges. Refuges are periods of time or areas of space in which a toxin is not used and they serve to maintain toxin-susceptible insects in local populations. Because alleles that confer insecticide resistance are generally recessive, mating between resistant and susceptible insects usually produces susceptible progeny. Temporal refuges can be established by rotating varieties or using gene “promoters” that drive the expression of toxin genes only at certain stages of plant growth, for example, at the reproductive stage but not at the vegetative stage. Spatial refuges can be established within fields by sowing mixtures containing seeds of toxic and nontoxic plants, or among fields by planting some fields to nontoxic plants. Which spatial scale is most effective is highly dependent on the biology of the target pest species. Studies of important aspects of rice stem borer biology, such as larval movement among plants and dispersal of adult insects among fields, are under way at IRRI (Cohen et al 1996). In the United States, Monsanto requires all farmers growing their “Bollgard” Bt cotton to plant a proportion of their land to non-Bt cotton, to serve as a refuge. It will not be possible to maintain refuges in this way for Bt rice in Asia, where there are hundreds of millions of small farmers. It remains to be seen whether a sufficient level of “unstructured refuges,” arising as a result of some farmers growing non-Bt rice varieties by preference or by chance, will be maintained in rice-growing areas. The amount of refuge area required will be lower if two toxins are used, because insects having resistance to both toxins will be rare. Thus, the use of multiple toxins may be particularly important for sustainable use of transgenic insect-resistant rice varieties.

References Ahmed HU, Finckh MR, Alfonso RF, Mundt CC. 1997. Epidemiological effect of gene deployment strategies on bacterial blight of rice. Phytopathology 87:66-70. Bonman JM, Estrada BA, Denton RI. 1986. Blast management with upland rice cultivar mixtures. In: Proceedings of the Symposium on Progress in Upland Rice Research, Los Baños, Laguna. Manila (Philippines): International Rice Research Institute. p 375-382. Carozzi N, Koziel M, editors. 1997. Advances in insect control: the role of transgenic plants. London (UK): Taylor & Francis. Cohen MB, Aguda RM, Romena AM, Roderick GK, Gould FL. 1996. Resistance management strategies for Bt rice: what have we learned so far? In: Rice genetics III. Proceedings of the Third International Rice Genetics Symposium; 16-20 Oct 1995. Manila (Philippines): International Rice Research Institute. p 749-752.

160

Cohen et al

Dahal G, Dasgupta I, Lee G, Hull R. 1992. Comparative transmission of three tungro isolates by green leafhopper (GLH). Int. Rice Res. Notes 17:3. Dasgupta I, Hull R, Eastop S, Pogi-Pollini C, Blakebrough M, Boulton MI, Davies JW. 1991. Rice tungro bacilliform virus DNA independently infects rice after Agrobacterium - mediated transfer. J. Gen. Virol. 72:1215-1221. Datta SK. 1995. Polyethylene glycol mediated direct gene transfer to indica rice protoplast and regeneration of transgenic plants. In: Potrykus I, Spangenberg G, editors. Gene transfer in plants. Berlin (Germany): Springer Pub. p 66-74. Datta K, Vasquez A, Tu J, Torrizo L, Alam MF, Oliva N, Abrigo E, Khush GS, Datta SK. 1998. Constitutive and tissue-specific differential expression of cryIA(b) gene in transgenic rice plants conferring resistance to rice insect pests. Theor. Appl. Genet. (In press.) De Wit CT. 1982. La productivité des pâturages sahéliens. In: Penning de Vries FWT, Djiteye MA, editors. La Productivité des Pâturages Sahéliens - Une Etude des Sols, des Végétations et de l’Exploitation de cette Ressource Naturelle. Pudoc, Wageningen, Netherlands. p 2235. Elazegui FA, Soriano J, Bandong J, Estorninos L, Jonson I, Teng PS, Shepard BM, Litsinger JA, Moody K, Hibino H. 1990. Methodology used in the IRRI integrated pest survey. In: Crop loss assessment in rice. Manila (Philippines): International Rice Research Institute. p 243-271. Elings A, Rubia EG, editors. 1994. Analysis of damage mechanisms by pests and diseases and their effects on rice yield. SARP Research Proceedings. DLO (Wageningen): Wageningen Agricultural University and Manila (Philippines): International Rice Research Institute. 279 p. Fujimoto H, Itoh K, Yamamoto M, Kyozuka J, Shimamoto K. 1993. Insect resistant rice generated by introduction of a modified -endotoxin gene of Bacillus thuringiensis. Biotechnology 11:1151-1155. Ghareyazie B, Alinia A, Menguito CA, Rubia L, de Palma JM, Liwanag EA, Cohen MB, Khush GS, Bennett J. 1997. Enhanced resistance to two stem borers in aromatic rice containing synthetic cryIA(b) gene. Mol. Breed. 3:401-414. Gould F. 1996. Deploying pesticidal engineered crops in developing countries. In: Persley GJ, editor. Biotechnology and integrated pest management. Wallingford (UK): CAB International. p 264-293. Gould F. 1998. Sustainability of transgenic insecticidal cultivars: integrating pest genetics and ecology. Annu. Rev. Entomol. 43:701-726. Gould F, Martinez-Ramirez A, Anderson A, Ferre J, Silva FJ, Moar WJ. 1992. Broad spectrum resistance to Bacillus thuringiensis toxins in Heliothis virescens. Proc. Nat’l. Acad. Sci. USA 89:7986-7990. Hamer JE. 1991. Molecular probes for rice blast disease. Science 252:632-633. Huang N, Angeles ER, Domingo J, Magpantagy G, Singh S, Zhang G, Kumaravadivel N, Bennett J, Khush GS. 1997. Pyramiding of bacterial blight resistance genes in rice: markerassisted selection using RFLP and PCR. Theor. Appl. Genet. 95:313-320. Ikeda R, Khush GS, Tabien RE. 1990. A new resistance gene to bacterial blight derived from O. longistaminata. Jpn. J. Breed. 40:280-281. IRRI (International Rice Research Institute). 1994. Integrated pest management: the IRRI perspective. IRRI Information Series No. 3. Manila (Philippines): IRRI. 20 p. James CW. 1974. Assessment of plant diseases and losses. Annu. Rev. Phytopathol. 12:27-48.

Importance of rice pests and challenges to their management

161

Khush GS. 1984. Terminology for rice growing environments. Manila (Philippines): International Rice Research Institute. Khush GS. 1995. Moderm varieties: their real contribution to food supply and equity. GeoJournal 35:275-284. Khush GS, Mackill DJ, Sidhu GS. 1989. Breeding rice for resistance to bacterial blight. In: Bacterial blight of rice. Manila (Philippines): International Rice Research Institute. p 207217. Kinoshita T. 1995. Report of Committee on Gene Symbolization, Nomenclature, and Linkage Groups. Rice Genet. Newsl. 12:9-153. Leach JE, Rhodes ML, Vera Cruz CM, White FF, Mew TW, Leung H. 1992. Assessment of genetic diversity and population structure of Xanthomonas oryzae pv. oryzae with a repetitive DNA element. Appl. Environ. Microbiol. 58:2188-2195. Lee EJ, Zhang Q, Mew TW. 1989. Progress in irrigated rice research. Manila (Philippines): International Rice Research Institute. Lin W, Anuratha CS, Datta K, Potrykus I, Muthukrishnan S, Datta S. 1995. Genetic engineering of rice for resistance to sheath blight. Bio/Technology 13:686-691. Mackill DJ, Bonman JM. 1992. Inheritance of blast resistance in near-isogenic lines of rice. Phytopathology 82:746-749. Madden LV. 1983. Measuring and modeling crop losses at the field level. Phytopathology 73: 1591-1596. McCouch SR, Abenes ML, Angeles R, Khush GS, Tanksley SD. 1991. Molecular tagging of a recessive gene, xa-5, for resistance to bacterial blight of rice. Rice Genet. Newsl. 8:143145. McCouch SR, Kochert G, Yu ZH, Wang ZY, Khush GS, Coffman WR, Tanksley SD. 1988. Molecular mapping of rice chromosomes. Theor. Appl. Genet. 76:815-829. Mew TW, Vera Cmz CM, Medalla ES. 1992. Changes in race frequency of Xanthomonas oryzae pv. oryzae in response to rice cultivars planted in the Philippines. Plant Dis. 76:10291032. Ogawa T. 1993. Methods and strategy of monitoring race distribution and identification of resistance genes to bacterial blight ( Xanthomonas oryzae pv. oryzae ) in rice. JARQ 27:7180. Pingali PL, Roger PA, editors. 1995. Impact of pesticides on farmer health and the rice environment. Manila (Philippines): International Rice Research Institute. Pinnschmidt HO, Nguyen Dan Long, Tran Tan Viet, Le Dinh Don, Teng PS. 1995. Characterization of pests, pest losses, and production patterns in rainfed lowland rice of the Mekong River Delta. In: Denning GL, Vo-Tong Xuan, editors. Vietnam and IRRI: a partnership in rice research. (Manila) Philippines: International Rice Research Institute. p 223-242. Pinstrup-Andersen P, de Londono N, Infante M. 1976. A suggested procedure for estimating yield and production losses in crops. PANS 22:359-365. Rabbinge R. 1993. The ecological background of food production. In: Crop protection and sustainable agriculture. Ciba Foundation Symposium 177. Chichester: John Wiley and Sons. p 2-29. Ronald PC, Albano B, Tabien R, Abenes L, Wu K, McCouch S, Tanksley S. 1992. Genetic and physical mapping of the rice bacterial blight resistance locus Xa-21. Mol. Gen. Genet. 236:113-120. Rowe WD. 1980. Risk assessment approaches and methods. In: Conran G, editor. Society, technology, and risk assessment. New York: Academic Press. p 3-29.

162

Cohen et al

Sama S, Hasanuddin A, Manwan I, Cabunagan R, Hibino H. 1991. Integrated management of rice tungro disease in South Sulawesi (Indonesia). Crop Prot. 10:34-40. Savary S, Elazegui FA, Moody K, Teng PS. 1994. Characterization of rice cropping practices and multiple pest systems in the Philippines. Agric. Syst. 46:385-408. Savary S, Elazegui FA, Pinnschmidt, Teng PS. 1997. Characterization of rice pest constraints in Asia: an empirical approach. In: Teng PS, Kropff M, ten Berge HFM, Dent JB, Lansigan FP, van Laar HH, editors. Systems approaches for agricultural development. Vol. I. Dordecht: Kluwer. p 83-98. Schaffner D, Koller B, Mueller K, Wolfe MS. 1992. Response of populations of Erysiphe graminis sp. hordei to large-scale use of variety mixtures. Vortraege fuer Pflanzenzuechtung 24:317-319. Schnepf HE. 1995. Bacillus thuringiensis toxins: regulation, activities and structural diversity. Current Opinion Biotechnol. 6:305-312. Sebastian LS, Ikeda R, Huang N, Imbe T, Coffman WR, McCouch SR. 1996. Molecular mapping of resistance to rice tungro spherical virus and green leafhopper. Phytopathology 86:25-30. Song W-Y, Wang G-L, Chen L-L, Kim H-S, Pi L-Y, Holsten T, Gardner J. Wang B, Zhai W-X, Zhu L-H, Fauquet C, Ronald P. 1995. A receptor kinase-like protein encoded by the rice disease gene, Xa-21. Science 270:1804-1806. Stem VM. 1973. Economic thresholds. Annu. Rev. Entomol. 18:259-280. Teng PS. 1983. Estimating and interpreting disease intensity and loss in commercial fields. Phytopathology 73:1587-1590. Teng PS, Gaunt RE. 1980. Modeling systems of disease and yield loss in cereals. Agric. Syst. 6:131-154. Teng PS, Savary S. 1992. Implementing the systems approach in pest management. Agric. Syst. 40:237-264. Tu J, Isabelida O, Zhang Q, Mew TW, Khush GS, Datta SK. 1998. Transgenic rice variety IR72 with Xa-21 is resistant to bacterial blight. Theor. Appl. Genet. (In press.) Waage J. 1996. Integrated pest management and biotechnology: an analysis of their potential for integration. In: Persley GJ, editor. Biotechnology and integrated pest management. Wallingford (UK): CAB International. p 37-60. Wang G, Mackill DJ, Bonman JM, McCouch SR, Nelson RJ. 1994. RFLP mapping of genes conferring complete and partial resistance to blast resistance in a durably resistant rice cultivar. Genetics 136:1421-1434. Way MJ, Heong KL. 1994. The role of biodiversity in the dynamics and management of insect review. Bull. Entomol. Res. 84:567-587. pests of tropical rice-a Wolfe MS. 1985. The current status and prospects of multiline cultivars and variety mixtures for disease resistance. Annu. Rev. Phytopathol. 23:251-273. Wu C, Fan Y, Zhang C. Oliva N, Datta SK. 1997. Transgenic fertile japonica rice plants expressing a modified cryIA(b) gene resistant to yellow stem borer. Plant Cell Rep. 17:129132. Wunn JA, Kloti A, Burkhardt PK, Ghosh Biswas GC, Launis K, Inglesias VA, Potrykus I. 1996. Transgenic indica rice breeding line IR58 expressing a synthetic cryIA(b) gene from Bacillus thuringiensis provides effective insect pest control. Bio/Technology 14:171-176. Yoshimura S, Yoshimura A, McCouch SR, Baraoidan MR, Mew TW, Iwata N, Nelson RJ. 1995. Tagging and combining bacterial blight resistance genes in rice using RAPD and RFLP markers. Mol. Breed. 1:375-387.

Importance of rice pests and challenges to their management

163

Zadoks JC. 1967. Types of losses caused by plant diseases. In: FAO papers presented at the symposium on crop losses. Rome: Food and Agriculture Organization. p 149-158. Zadoks JC. 1985. On the conceptual basis of crop loss assessment: the threshold theory. Annu. Rev. Phytopathol. 23:455-473. Zadoks JC, Schein RD. 1979. Epidemiology and plant disease management. New York: Oxford University Press. 427 p.

Notes Authors’ addresses: M. Cohen, S. Savary, O. Azzam, and S.K. Datta, International Rice Research Institute, P.O. Box 933, 1099 Manila, Philippines. N. Huang, Applied Phytologics, Inc., 4110 N. Freeway Blvd., Sacramento, CA 95834. USA. [email protected] [email protected] [email protected] [email protected] [email protected] Citation: Dowling NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute.

164

Cohen et al

CHAPTER 10

Weeds: a looming problem in modern rice production M. Olofsdotter, A. Watson, and C. Piggin

Weeds are a constant problem in all rice-growing areas. The reduced availability of water and labor is the driving force that changes cultural practices in rice production. The shift from transplanted to direct seeding of rice aggravates the problem because weeds and the crop emerge together and it is more difficult to use early flooding for weed control. Herbicide use is increasing in Asia because herbicides are cheaper than hand labor and easy to apply. Herbicides have negative effects, however, such as changes in weed flora that result in the increase of hard-to-control weed species, environmental contamination, and selection for herbicide-tolerant weed biotypes. Thus, it is becoming increasingly important to develop integrated weed management systems in which several control measures are combined and herbicide use is minimized. More tools are therefore required to complement good agronomic practices. IRRI research has shown that allelopathy and biological control have the potential to increase weed control. Some rice cultivars can suppress weed growth by more than 50% under field conditions. Research on biological control shows promising results for controlling some of the major weeds in rice.

Rice production must increase from 500 to 800 million t in the next 25 yr to meet projected world rice demand. In addition, this increase must be sought through sustainable agricultural practices to ensure a long-term food supply. Rice production systems are changing rapidly in response to the declining availability of labor and water in rural areas. Direct seeding is being used increasingly to reduce dependence on labor for transplanting. Agrochemicals are also used more frequently to reduce losses from weeds that have traditionally been controlled by flooding. These changes in cultural practices are bringing new selection pressures for weeds and a need to develop better systems of integrated and sustainable weed management.

Weed management Weeds are the major biological constraint in most rice-growing areas of the world. Unlike the periodic outbreaks of insect pests and plant diseases, weeds are ever-present and threatening. Problems associated with weeds in rice are mounting dramatically in South and Southeast Asia because of the reduced availability of affordable labor,

165

decreased availability of adequate irrigation water, and the shift in crop establishment from transplanting to direct seeding. The lack of suitable weed control alternatives has led to a reliance on herbicides in many rice-producing areas, and their use is increasing. Herbicides are generally less expensive than manual labor, very effective, and easy to use. These desirable features, however, are major disincentives to the development of alternative control strategies. The shift to direct seeding has been accompanied by the widespread use of herbicides and has led to a shift from relatively easyto-control broadleaf weeds to more difficult-to-control grass weeds, especially weedy rice. The continuous use of herbicides naturally selects for tolerant species, leading to the development of herbicide-resistant weeds. Increased herbicide use also poses a threat to human health and the environment. Our challenge is to create an environment favorable to the rice crop and unfavorable to weeds, where minimal labor, water, and chemical herbicide inputs are required for weed control.

Integrated weed management No single weed-management strategy will solve all weed problems in rice (Hill et al 1994). Preventive, physical, managerial, biological, and chemical control methods need to be combined to attain acceptable weed management with minimal use of herbicides (Watson 1992). Integrated weed management emphasizes managing the weed population rather than eradicating weeds (Altieri 1987, Kropff and Moody 1992). Table 1 illustrates this shift from weed control to weed management, which develops

Table 1. Differences between weed control and weed management. Structure

Weed control

Weed management

Goal

Maximize crop yield and profits.

Optimize long-term farm productivity.

Objectives

Eradicate weeds from the crop.

Maintain weeds below level of significant competition with the crop.

Approach

Use one or two of the easiest, most effective methods suited to the crop.

Balance the best available methods suited to the farming system.

Action

Employ full-tillage technology, apply full rates of herbicides.

Employ minimum tillage, minimum effective rates of herbicide, and integrated agronomic practices to increase competitive ability of the crop.

Outputs

Near-perfect weed elimination, high crop yield.

Substantial reduction of weed pressure, optimum farm profit.

Application

Wide geographical regions.

Adapted to specific locations/areas.

Source: Kon (1993).

166

Olofsdotter et al

long-term strategies to minimize problems caused by weeds in the farming system. Long-term decision making in much of the developing world, however, is constrained because day-to-day issues determine the survival of most subsistence farmers and, understandably, short-term strategies with immediate benefits often prevail (Kon 1993).

Changes in weed flora Changes in cultural practices associated with rice production contribute to changing the weed flora. Increased herbicide use, mechanized tillage, variable water availability, crop establishment by direct seeding rather than transplanting, fertilizer use, mechanized harvesting and seed cleaning, and consolidation of farm units into larger holdings are some factors that cause shifts in weed flora (Haas and Streibig 1982, Liebman and Janke 1990). As these practices become widespread and extensively used, weed flora selected over time are tolerant of the weed control practices employed. Selected weed species are often difficult to control. Increased herbicide use is the most important factor responsible for shifts in weed flora (Haas and Streibig 1982). For example, phenoxy-acid herbicides such as 2,4-dichlorophenoxy acetic acid (2,4-D) have been used extensively in rice and cereals to control broadleaf weeds, resulting in the increase of hard-to-control grass weeds tolerant of 2,4-D.

Weed control with less labor and less water Labor for transplanting and hand weeding is becoming more expensive and difficult to find. As a result, farmers have been forced to switch to direct seeding, thus losing the early season advantage of flooding with transplanting to suppress early weed growth, especially of hard-to-control grasses such as barnyardgrass (Echinochloa crusgalli ). Rapid industrialization and urbanization compete with agriculture for limited water resources. In addition, much irrigation infrastructure is also poorly maintained, causing water shortages in rice production. Water conservation measures in rice production, such as intermittent flooding and shallow water depths, generally make weed control more difficult. With labor and water shortages and the shift to direct seeding, farmers have few weed control alternatives other than to increase herbicide use. Reliance on herbicides. With less labor and water for weed control, herbicide use in South and Southeast Asia is increasing exponentially. Even with increased herbicide use, crop losses caused by weeds have not declined and may have increased (Heong et al 1995). This effect may be attributable to the continuous use of the same selective herbicides that select for herbicide-tolerant weeds. Few, if any, economical alternatives to herbicides are now available, and this factor exacerbates herbicide dependency. Increases in herbicide use and dependency will have environmental and sociological costs, such as contamination of surface water and groundwater, adverse effects on nontarget organisms, and risks to human health. In the United States, rice herbicides have been detected in well water. They have polluted agricultural drains and rivers and have caused off-tastes in potable water supplies in California (Cornacchia

Weeds: a looming problem in modem rice production

167

et al 1984). Similar adverse environmental effects are occurring in South and Southeast Asia as herbicide use increases. We expect that these adverse effects will be even greater as rice production systems become more intensive, given the close proximity of rice fields to the water supply within the entire community. Herbicide use can be reduced when combined with good husbandry. Recommended herbicide rates are set to ensure that the product will perform over a wide range of environmental conditions, and control the more difficult species. Many farmers in South and Southeast Asia are already applying herbicides at less than recommended rates and achieving satisfactory control. Water, land preparation, seeding, and weed control are closely interrelated in rice production. Good land preparation reduces weed infestations and permits more efficient water use (Heong et al 1995). During 1990-93, in the Muda area in Malaysia, herbicide use declined as a result of an extension campaign on integrated weed management (Ho 1994). Farmers who applied proper land leveling and water management needed to apply herbicide only once, whereas farmers using poor cultural practices had to apply herbicides three or four times to control weeds. Resistance to herbicides. Reliance on herbicides for weed control brings biological repercussions, such as selection and enrichment of genes that confer herbicide resistance in weed populations. Resistant biotypes are common, normally have vigor, and are difficult to control with other herbicides. Some weed populations are accumulating resistance mechanisms and have resistance to many herbicide groups. Resistance to at least 15 classes of herbicides by more than 100 weed species worldwide has been reported, and the area infested with herbicide-resistant weeds is increasing (Jasieniuk et al 1996). Propanil has been used in the United States since 1960 for grass control in rice. The continuous use of propanil on 70% of the rice area in Arkansas has led to the development of resistant populations of E. crus-galli (Carey et al 1992). Of equal or greater concern is the rapid appearance of resistance to newer generation herbicides, including the sulfonylureas. Resistant populations of four weeds—Sagittaria montevidensis, Cyperus difformis, Scirpus mucronatus, and Ammannia auriculata—have developed in California after only 5 yr of bensulfuron field use and resistant populations have been found at 72 sites throughout rice-growing areas in California (Pappas-Fader et al 1994). Transgenic herbicide-resistant rice. Major research efforts are being directed toward developing herbicide-resistant field crops (Dekker and Duke 1995). The primary focus of this research is to incorporate genes conferring resistance to broadspectrum herbicides such as glyphosate and glufosinate. Both herbicides are environmentally relatively benign. Transgenic herbicide-resistant rice cultivars, including glufosinate-resistant (Datta et al 1992) and sulfonylurea-resistant (Li et al 1992) ones, have already been developed. Transgenic glufosinate-resistant rice has been fieldtested in Louisiana with no substantial negative agronomic or quality differences between the transformed and original parent material (Braverman and Linscombe 1994). The cited reason for the keen interest in developing glufosinate-resistant rice is to control “red rice.”

168

Olofsdotter et al

Weedy rice and red rice are undesirable early shattering off-types that are morphologically very similar to, and naturally cross with, cultivated rice. Their ability to hybridize with cultivated rice is a major concern if herbicide-resistant rice cultivars are to be released widely. Kerlan et al (1992) reported that the risk of gene dispersal from outcrossing of transgenic glufosinate-resistant rapeseed with weedy Brassica spp. was limited. Mikkelsen et al (1996), however, recently demonstrated that transgenic rapeseed could cross with weedy relatives, producing transgenic weedlike plants after only two generations of hybridization and backcrossing. These findings with rapeseed suggest that herbicide-resistant rice and weedy rice have the potential to hybridize and confer herbicide resistance to “weedy” rice. This would have major adverse effects if herbicide-resistant “weedy” rice flourished. The potential for and consequences of such a transfer of herbicide resistance need to be thoroughly considered before transgenic herbicide-resistant rice is released.

Reduction in herbicide dependency and alternatives Herbicides alone cannot be relied upon to solve all weed problems in rice. More tools are needed besides good husbandry and a judicious use of minimal amounts of herbicides. Improvement of rice germplasm to enhance weed-suppressing capacity can help in minimizing herbicide use. IRRI has several novel research programs under way to develop allelopathy in rice and to achieve biological control of weeds using indigenous fungi. Competitive cultivars suppress weeds through the efficient capture of available nutrients, light, and water, whereas allelopathic cultivars have been identified to suppress weeds through the release of chemicals into the environment. Traditionally, competition has been thought of as the most important factor in plant interference. But recent research has shown that there is a good potential to use allelopathy in rice to reduce weed growth significantly in the field (Olofsdotter and Navarez 1996). Allelopathy is the release by a plant of chemical compounds that affect the growth and development of other living plants. Dilday et al (1991), in observing 10,000 rice accessions in nonreplicated seed increase plots, reported that 3.5% showed allelopathic potential against ducksalad ( Heteranthera limosa). One allelopathic accession was also able to control 72-95%of a mixed population of Ammannia coccinea and Bacopa rotundifolia (Lin et al 1992). Because it is difficult to separate the effects of competition and allelopathy in the field, laboratory experiments have been used to eliminate competition as a cause of observed crop-weed interference (Olofsdotter and Navarez 1995). At IRRI, laboratory screening and field experiments showed that 19 of 111 rice cultivars tested suppressed the growth (dry matter) of E. crus-galli by >40% in the dry and wet seasons of 1995. Eight of these cultivars reduced weeds by >50% in both growing seasons. Suppression in the field was comparable to root reduction observed in laboratory screening, suggesting that allelopathy was the major part of the interference found (Navarez and Olofsdotter 1996, Olofsdotter and Navarez 1996).

Weeds: a looming problem in modern rice production

169

The rice accessions that showed allelopathic activity have different origins and are in different stages of improvement. Characterization of the chemical(s) involved, physiological cost of the mechanism, ecotoxicology, and incorporation of allelopathic potential into a breeding program are some of the challenges to be met. Although many questions remain unanswered, new knowledge on allelopathic and competitive abilities is likely to result in a wider use of weed-suppressing rice cultivars. Pathogenic fungi that occur naturally on weeds also offer an environmentally sound aid to control weeds in rice (Watson 1994). Biological weed control research began in 1991 at IRRI and focuses on the following major weeds of rice: Echinochloa crus-galli and E, colona, Eleusine indica, Fimbristylis miliacea, Cyperus rotundus, C. iria, and C. difformis, Mimosa invisa, Monochoria vaginalis, and Sphenoclea zeylanica. This research focuses on the discovery and propagation of indigenous fungal pathogens from weed hosts for weed control. Disease and mortality of target weeds are promoted by augmenting natural pathogen populations through the inundatory application of high levels of spore suspensions. Virulent indigenous pathogens with biocontrol potential have been isolated from all of the targeted species except M. vaginalis. Optimum conditions for disease expression and damage to the weed have been determined under regulated environmental conditions for most of the weedpathogen systems under study. A spore suspension of an Alternaria species was used to effectively control Sphenoclea zeylanica in a farmer’s field in Central Luzon (Mabbayad and Watson 1995). The control provided by a standard herbicide treatment of 2,4-D was inferior. This trial was repeated thrice in Laguna and once in Leyte with the collaboration of staff from the Visayas State College of Agriculture. Similar results were found. S. zeylanica is the only plant species susceptible to the Alternaria isolate. Six pathogenic fungi have been isolated from Echinochloa species. Of these, two were virulent on three Echinochloa species (nonpathogenic to rice) and needed a relatively low dew period duration compared with other fungi tested (Zhang et al 1996). Three of the Echinochloa pathogens produce chemicals that are active on the three Echinochloa species tested. Three of the chemicals are known phytotoxins, whereas the fourth appears to be a novel compound. Initial field trials with the fungus Exserohilum monoceras provided 50-80% control of Echinochloa species. Two additional Echinochloa pathogens have provided similar levels of control in the field. Leaf wetness duration is a critical factor and can limit the performance of these fungi for weed control in tropical areas where evaporation is high. An oil emulsion has overcome this limitation, but it was slightly phytotoxic to rice. In pot experiments, a dry-powder formulation that floated on the water surface effectively delivered the inoculum to the target weeds with no damage to rice. In studies on other weeds, a virulent pathogen from Eleusine indica has been evaluated for biocontrol potential. Three Curvularia spp. have been isolated from Cyperus spp. and Fimbristylis miliacea that demonstrate different degrees of virulence on these sedges. Mimosa invisa was controlled in the field by applying a spore

170

Olofsdotter et al

suspension of a fungal pathogen without any damage to mungbean, but this isolate damaged some upland rice cultivars. Challenges to be met in the biocontrol program are optimization of propagule production and infection in the field, technology transfer, and integration within weed management systems at the farm level. As these challenges are met, fungal pathogens will become more attractive and more widely used in integrated weed management.

References Altieri MA. 1987. Agroecology: the scientific basis of alternative agriculture. Boulder, Colo. (USA): Westview Press. p 173-185. Braverman MP, Linscombe SD. 1994. Field evaluation of genetically engineered glufosinateresistant rice lines. Baton Rouge, La. (USA): Agricultural Experiment Station, Louisiana State University. 37(3):29. Carey VF III, Talbert RE, Baltazar AM, Smith RJ Jr. 1992. Evaluation of propanil-resistant barnyardgrass in Arkansas. Proceedings of the Rice Technical Working Group, 1992. Little Rock, Ark. (USA): Rice Technical Working Group. 24:120. Cornacchia JW, Cohen DB, Bowes GW, Schnage RJ, Montoya BL. 1984. Rice herbicides: molinate and thiobencarb. CSWRCB special project report. Sacramento, Calif. (USA): State Water Resources Control Board. 84-4sp. Datta SK, Datta DK, Soltanifar N, Donn G, Potrykus I. 1992. Herbicide resistant Indica rice plants from IRRI breeding line IR72 after PEG-mediated transformation of protoplasts. Plant Mol. Biol. 20:619-629. Dekker J, Duke SO. 1995. Herbicide-resistant field crops. Adv. Agron. 54:69-116. Dilday RH, Nastasi P, Lin J, Smith RJ Jr. 1991. Allelopathic activity in rice (Oryza sativa L.) against ducksalad (Heteranthera limosa (Sw.) (Willd.). In: Proceedings of a Symposium on Sustainable Agriculture for the Great Plains. Washington, D.C. (USA): United States Department of Agriculture. p 193-201. Haas H, Streibig JC. 1982. Changing patterns of weed distribution as a result of herbicide use and other agronomic factors. In: LeBaron HM, Gressel J, editors. Herbicide resistance in plants. New York John Wiley & Sons. p 57-79. Heong KL, Teng PS, Moody K. 1995. Managing rice pests with less chemicals. GeoJournal 35(3):337-349. Hill JE, Smith RJ Jr, Bayer DE. 1994. Rice weed control: current technology and emerging issues in temperate rice. Austr. J. Exp. Agric. 34:1021-1029. Ho NK. 1994. Integrated weed management of rice in Malaysia: some aspects of the Muda irrigation scheme’s approach and experience. In: Sastroutomo SS, Auld BA, editors. Appropriate weed control in Southeast Asia. Proceedings of an FAO-CAB International workshop, Kuala Lumpur, Malaysia, 17-18 May 1994. Wallingford (UK): CAB International.

p 83-97. Jasieniuk M, Brûlé-Babel AL, Morrison IN. 1996. The evolution and genetics of herbicide resistance in weeds. Weed Sci. 44:176-193. Kerlan MC, Chevre AM, Eber F, Baranger A, Renard M. 1992. Risk of assessment of outcrossing of transgenic crop seed to related species. I. Interspecific hybrid production under optimal conditions with emphasis on pollination and fertilization. Euphytica 62:145-153.

Weeds: a looming problem in modern rice production

171

Kon KF. 1993. Weed management: towards tomorrow (with emphasis on the Asia-Pacific Region). Proceedings of the 14th Asia-Pacific Weed Science Society Conference. Brisbane: Asia Pacific Weed Society. p 1-9. Kropff MJ, Moody KM. 1992. Weed impact on rice and other tropical crops. In: Combellack JH, Levick KJ, Parsons J, Richardson RB, editors. Proceedings of the First International Weed Control Congress. Melbourne: Weed Science Society of Victoria Inc. p 123-126. Li Z, Hayashimoto A, Murai N. 1992. A sulfonylurea herbicide resistance gene from Arabidopsis thaliana as a new selectable marker for production of fertile transgenic rice plants. Plant Physiol. 100:662-668. Liebman M, Janke RR. 1990. Sustainable weed management practices. In: Francies CA, Flora CB, King LD, editors. Sustainable agriculture in temperate zones. New York: John Wiley & Sons. p 111-143. Lin J, Smith RJ Jr, Dilday RJ. 1992. Allelopathic activity of rice germplasm on weeds. Proceedings of the Southern Weed Science Society. 45:99. Mabbayad MO, Watson AK. 1995. Biological control of gooseweed ( Sphenoclea zeylanica Gaertn.) with an Alternaria sp. Crop Prot. 14:429-433. Mikkelsen TR, Andersen B, Jørgensen RB. 1996. The risk of crop transgene spread. Nature 380:31. Navarez D, Olofsdotter M. 1996. Relay seeding technique for screening allelopathic rice ( Oryza sativa). In: Brown H, Cussans GW, Devine, MD, Duke, SO, Fernandez-Quintanilla C, Helweg A, Labrada RE, Landes M, Kudsk P, Streibig JC, editors. Proceedings of the Second International Weed Control Congress. Slagelse (Denmark): Department of Weed Control and Pesticide Ecology. Vol. 4. p 1285-1290. Olofsdotter M, Navarez D. 1996. Allelopathic rice for Echinochloa crus-galli control. In: Brown H, Cussans GW, Devine MD, Duke SO, Fernandez-Quintanilla C, Helweg A, Labrada RE, Landes M, Kudsk P, Streibig JC, editors. Proceedings of the Second International Weed Control Congress. Slagelse (Denmark): Department of Weed Control and Pesticide Ecology. Vol. 4. p 1175-1181. Olofsdotter M, Navarez D. 1995. Approaches in rice allelopathy research. In: Proceedings of the 15th Asian-Pacific Weed Science Society Conference. Vol 1A. Tsukuba (Japan): Weed Science Society of Japan. p 315-321. Pappas-Fader T, Turner RG, Cook JF, Butler TD, Lana PJ, Caniere M. 1994. Resistance monitoring programs for aquatic weeds to sulfonylurea herbicides in California rice fields. In: Proceedings of the Rice Technical Working Group, 1994. New Orleans, La. (USA): Rice Technical Working Group. 25:165. Watson AK. 1992. Biological and other alternative control measures. In: Combellack JH, Levick KJ, Parsons J, Richardson RB, editors. Proceedings of the 1st International Weed Control Congress. Vol I. Melbourne: Weed Science Society of Victoria Inc. p 64-73. Watson AK. 1994. Biological weed control and prospects for bioherbicides development in the Philippines. Phillip. J. Weed Sci. Special Issue 64-81. Zhang WM, Moody K, Watson AK. 1996. Responses of Echinochloa species and rice ( Oryza sativa) to indigenous pathogenic fungi. Plant Dis. 80:1053-1058.

172

Olofsdotter et al

Notes Authors’ addresses: M. Olofsdotter, International Rice Research Institute (IRRI), P.O. Box 933, 1099 Manila, Philippines, and the Royal Veterinary and Agricultural University, Frederiksberg, Denmark; A. Watson, IRRI and McGill University, Quebec, Canada; C. Piggin, IRRI. [email protected] and [email protected] [email protected] and [email protected] [email protected] Citation: Dowling NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute.

Weeds: a looming problem in modem rice production

173

CHAPTER 11

Management of water as a scarce resource: issues and options in rice culture S.I. Bhuiyan, T.P. Tuong, and L.J. Wade Rice culture is known for its crucial dependence on an adequate supply of water. But how much water is really needed for producing the crop? How can water management and rice production systems be improved to obtain more rice per unit of water supplied? How can the agrochemicals associated with rice production be managed for minimal impact on the quality of water resources that are vital for sustainable agriculture? These and other related questions must be adequately addressed to achieve the needed rice production growth in Asia, where water for agriculture is becoming increasingly scarce. This paper addresses these issues in a holistic perspective that elucidates options from the farm to the irrigation system and basin level. Prospective technological innovations in areas such as crop management and varietal development, which should improve water use efficiency in rice culture, are also discussed.

Importance of water in rice culture: present and future As an aquatic plant, rice grows better and produces higher grain yields when grown in a flooded soil than when grown in dry soil. Besides supplying water to meet the plant’s evapotranspiration demand, the ponded water layer also helps suppress weed growth and increase the availability of many nutrients (De Datta 1981). Unlike other food crops, rice suffers from water stress even at soil water contents that exceed field capacity. Thus, a reliable and adequate water supply is crucial for high yield performance of rice. Because rice culture evolved in response to the amount and reliability of water supply, distinct ecosystems for rice have evolved. They have been characterized as upland, rainfed lowland, irrigated, and flood-prone, defined by their agrohydrology (IRRI 1989). In modern rice culture, the degree of control over water determines the level of production technologies employed by rice farmers. In rainfed ecosystems, variability in the amount and distribution of rainfall is the most important factor affecting crop growth and yield. The planting season begins with the onset of the monsoon rain. Inseason drought is common, however, and limits the yield potential of rice. In addition, alternate wet-and-dry field conditions cause nitrogen (N) loss and high weed infestation. Farmers in most rainfed ecosystems therefore use the less risky traditional varieties and small amounts of fertilizers. Except in favorable rainfed areas

175

with shallow water depths, modern rice technologies have contributed little to improved rainfed rice yields, mainly because of a lack of water control. Irrigation has contributed significantly to the success of the Green Revolution in Asia. In the past three decades, the growth in rice yield in the irrigated ecosystem has been 2.6% per year (Hossain 1995). Only about 55% of rice land is irrigated, but it produces 76% of worldwide production (IRRI 1993). By 2025, the Asian population is expected to increase by 53% and the demand for rice by 69% (Hossain 1995). Although more recent estimates expected lower future demand for rice (about a 40% increase, M. Agcaoili-Sombilla, personal communication, 1997), the increase is still substantial. Irrigated rice lands will have to satisfy a large proportion of this additional demand and at the same time allow the development of other crops. We will also need appropriate technologies for rainfed systems to meet the growing demand for rice. This paper examines the current supply and quality of water for rice culture, and opportunities for increasing water use efficiency while improving or sustaining water quality over the coming decades.

Water—a scarce and declining resource Present scarcity and future scenarios Fresh water is a finite resource. Only 38 million km3 of water, or 2.7% of all the water on Earth, is fresh or nonsaline and suitable for consumption by terrestrial plant and animal life (Sarma 1986). About 76% of this amount is held in permanent ice caps and glaciers, and 11% is held in formations at depths greater than 1 km. Only about 4.5 million km3 of fresh water is available for consumption, of which 97% is present as underground water within 1 km depth. Only 0.14 million km3 of water is present in lakes, rivers, and the atmosphere. Large-scale irrigation development has slowed considerably since the early 1980s, because engineering and environmental costs of exploiting new but feasible sources of water are increasing. Although the total water use in Asia, about 85% of which is used in agriculture, has increased by nearly 3% annually from 1950 to 1990, per capita water availability has declined by 40-60%over the same period (Gleick 1993). India, Pakistan, the Philippines, and Vietnam are expected to suffer sharp declines in per capita water availability over the next two decades (Fig. 1; IRRI 1998). The urban population in Asia is expected to increase from about 35% of the total population in 1990 to more than 50% in 2025 (IRRI 1998). If demand outpaces supply of water resources, intersectoral competition will intensify, with adverse effects on agricultural water availability and food production, as well as on environmental quality. In the Angat multipurpose project in Luzon, Philippines, for example, the amount of water diverted for Metropolitan Manila increased consistently at about 10% per annum during 1980-95, with a corresponding decrease in the supply for irrigation to its 28,000-ha rice fields. A similar diversion of irrigation water to the urban sector is occurring in the Jatiluhur irrigation project of West Java, Indonesia, and in the Guangxi Autonomy Region of China. Because urban and industrial demands are likely to receive priority over irrigation, agricultural productivity would

176

Bhuiyan et al

Fig. 1. Per capita available water resources in selected Asian countries (after Gleick 1993).

be reduced in such irrigation systems, especially in years with low water supply at their sources.

Gap between water “need” and “use” in rice culture The use of water in traditional rice culture is highly inefficient. For each kg of irrigated rice, about 5,000 L of water are diverted at the source of the canal system (IRRI 1995). The actual field-level need is only about 25-30% of that amount. The gap between the “need” and the “use” of water in rice culture can be understood clearly by looking into the two major water-consuming components in transplanted rice culture: land preparation and crop irrigation. Land preparation. Preparing land for crop establishment normally involves supplying enough water to saturate the soil (land soaking) and maintain a water layer for plowing, harrowing, puddling, and leveling before rice seedlings are transplanted. The amount of water required for land preparation is about 150–250 mm, depending on the initial soil water condition and soil type. But the actual amount used for this purpose may be as high as 1,500 mm (Ghani et al 1989). Rice is grown mostly in clayey soils and land soaking for the wet-season rice crop generally starts following the long dry period when the soil is cracked. In fields with permeable subsoil, up to 60% of the water applied for land soaking may move down the cracks, bypassing the topsoil matrix (Tuong et al 1994, Tuong and Cabangon 1996). Most of this water is lost from the field through lateral drainage. Cracks in clayey soils may not close even

Management of water as a scarce resource: issues and options in rice culture

177

after prolonged wetting; therefore, bypass flow through cracks may continue until the field is puddled. Another reason for excessive water use during land preparation is the long period over which farmers continue land soaking and tillage activities. In the transplanted rice system, farmers keep the main field flooded during the 1-month period when seedlings are grown in small seedbeds until ready for transplanting. If the canal that serves a block of farms has slow-flowing water, 2 months or more are taken before all farmers in the canal service area can complete land preparation (Valera 1977). Most of the water applied to the field during this period is lost by runoff, seepage, percolation, and evaporation. Crop irrigation. Irrigation water supplied to the cropped field is used by evapotranspiration from the rice field, deep percolation, seepage, and overland runoff (Fig. 2). Excessive percolation loss can occur even in puddled fields through some nonpuddled spots that are unintentionally omitted during land preparation and through the under-bund areas that remain porous. Tuong et al (1994) found that a 1% nonpuddled area can increase the percolation water loss by a factor of 5. Under-bund percolation caused a further 2- to 5-fold increase in water loss by percolation, depending on the size of the field. Farmers prefer to maintain a relatively high water depth to control weeds and reduce the frequency of irrigation (and hence labor cost), and to store water as insurance against possible shortage, but percolation loss increases as the depth of water standing in the field increases (Tabbal et al 1992, Tuong et al 1994). Water loss from target areas by seepage and surface overflow also increases with greater water depths.

Fig. 2. Components of water balance in a rice field.

178

Bhuiyan et al

Water quality deterioration—causes and consequences Soil erosion. Soil erosion in upper watershed areas and consequent sedimentation downstream are constantly undermining the quality of irrigation water in reservoirbacked, canal-supplied rice production systems. Inappropriate land use and deforestation in the upper watershed areas aggravate the erosion-sedimentation problem. At the reservoir level, excessive sedimentation will fill up the reservoir’s active storage space, resulting in a reduction of the project’s service capacity and useful life. A survey of eight reservoirs in India established that the sedimentation rates in seven of them were 2.9 to 16.5 times higher than expected (Dogra 1986). Below the reservoir, canals silt up quickly when the sediment load is high. The cost of desilting canals to maintain their flow capacity is high. Irrigation-induced waterlogging. Waterlogging and salinization of the soil are the most pervasive damages caused by badly designed or poorly managed irrigation systems. Irrigation-induced waterlogging is a major problem in tropical Asia, but its actual extent is not well established. Estimates suggest that about 6 million ha of irrigated land are waterlogged in India. About 22% of the irrigation systems surveyed in the Philippines have 5% or more of their areas waterlogged. A high rate of water loss by seepage and percolation from rice fields and unlined canals may raise the underlying water table, thus affecting low-lying areas first. Heavy rainfall in the wet season may cause large-scale inundation, with severely affected areas becoming unproductive. Salinization. Irrigated areas are basically large evaporation pans where distilled water is returned to the atmosphere in the vapor phase, and salts remain behind in the soil. Therefore, salinity buildup is often associated with irrigated agriculture. In the humid tropics, salinity normally does not build up because of strong leaching of the soil by high rainfall in the monsoon season. In the semiarid tropics, however, rainfall is low, and salinization is aggravated when salty groundwater rises from continued percolation and seepage from irrigated fields and leaky irrigation canals. When the water table rises to within about 2 m from the soil surface, salts are brought up to the crop root zone by soil capillarity (Khosla et al 1980). Salinization reduces crop yields and may eventually cause complete crop failure, forcing farmers to abandon the land. Postel (1989) estimated that 36% of irrigated land in India, 15% in China, and 20% in Pakistan has been damaged by irrigation-induced salinization. Contamination by leachate from reclaimed acid sulfate soils. Reclamation of acid sulfate soils, which cover significant areas in Thailand, Vietnam, and Indonesia, involves leaching of acidic toxicity from the crop root zone. The process contaminates surface water, which may affect crops and soils in surrounding areas (Dent 1992, Minh et al 1997a). Acidic toxicities, especially aluminum, are particularly hazardous to fish and aquatic organisms whose threshold concentrations are far less than those for plant roots (van Breemen 1993). In Indonesia, Klepper et al (1990) reported a 10-fold reduction of fish yield in areas reclaimed from acid sulfate soils.

Management of water as a scarce resource: issues and options in rice culture

179

Nitrate-nitrogen contamination. Increased use of N fertilizer may result in a higher NO3 content in groundwater commonly used for domestic water consumption. High concentrations of nitrate-nitrogen (NO3-N) in drinking water, usually those in excess of 10 ppm, are considered unsafe for human consumption (Viets and Hagemen 1976), as they may cause methemoglobinemia (blue-baby syndrome) and carcinogenic effects (Follet and Walker 1989). In a study of two large irrigation systems in Luzon, Philippines, where farmers have been growing two rice crops per year using moderate levels of N fertilizers for several decades, NO3-N concentrations found in groundwater were very low (Castañeda and Bhuiyan 1991). In contrast, a study conducted at Batac, Ilocos Norte, Philippines, has shown high NO3-N concentrations in groundwater. At the Batac site, the combination of very intensive land use (two or three crops per year, with one rice crop in the wet season), relatively light-textured soil (loam), application of heavy doses of N in the dry season to upland crops grown after wet-season rice (average rate applied was 348 kg N ha-1), frequent irrigation, and the rise of groundwater to shallow depths in the wet season has resulted in a dry-season average groundwater NO 3-N concentration of 9.7 ppm (Gumtang et al 1998). Some groundwater samples exceeded the average by 3–4 times, making the water extremely hazardous for consumption. More research is needed on the process of nitrate pollution of groundwater from rice fields. Pesticide contamination. Pesticides in fresh surface water may enter the food chain. Although acute toxicities have a lethal effect on the fish population (Lim and Ong 1977), sublethal and chronic exposures to pesticides are more insidious and difficult to identify. Sublethal exposure to pesticides may suppress reproduction and result in pesticide-resistant strains of fish (Cheng 1990). In a recent case study in two irrigated rice areas in Luzon, Philippines, where nearly all farmers used pesticides, many of the pesticides reached the shallow groundwater aquifers beneath irrigated rice fields. Endosulfan was found in about 80% of the samples, monocrotophos in 54%, butachlor in 24%, methyl parathion in 24%, chloropyrifos in 7%, and carbofuran in 6% of the samples (Castañeda and Bhuiyan 1996). Endosulfan and butachlor are considered moderately hazardous and the rest extremely hazardous to human health. Their concentrations are still far below the daily acceptable intakes based on toxicological standards (FAO/WHO 1977). With the increasing cost of labor and consequent shift to direct seeding of rice, however, herbicide use has increased, especially in the Philippines, Thailand, and Malaysia. The levels of many groundwater pesticides, other than the six cited above, are not clear. Few data are available on herbicide concentration in groundwater. Nor is their persistence behavior in groundwater understood. Therefore, we need to study the contamination potential of new and untested insecticides and herbicides, and to develop appropriate policies to safeguard water quality.

180

Bhuiyan et al

Opportunities for increasing irrigation efficiency in rice Farm-level opportunities Land preparation phase. Water loss in land preparation through bypass flow can be curtailed by eliminating or reducing soil crack formation and water flow into cracks. Shallow dry-tillage of the soil soon after harvesting the previous crop allows the topsoil to act as mulch, thus reducing soil dehydration and its consequent cracking. Also, small soil aggregates formed by the tillage block the cracks and reduce water flow. Tuong and Cabangon (1996) found that in the clay soil of the IRRI experimental farm, shallow dry-tillage could save about 200 mm of water in land preparation. Because of increased access to high-powered tractors in rural areas, dry-tillage will become more feasible for farmers. This technique is practiced extensively in the Muda irrigation project of Malaysia and is credited with water savings and timely rice crop establishment benefits in the project area (Ho et al 1993). Shortening land preparation time reduces water loss in the irrigation system (Wickham and Sen 1978). Most rice irrigation systems allow water to be available to farmers for much longer periods than necessary to complete all irrigation activities. Consequently, farmers’ land use schedules and water use remain inefficient. Tailoring water delivery periods to the near optimal duration for land preparation and crop growth will reduce water wastage. To be successful with such actions, users must have confidence in the reliability of water delivery and the benefits of strict scheduling. Crop growth phase. When a rice field of medium soil type was maintained at a nearly saturated soil condition and weeds were controlled by herbicides, about 45% less water was consumed, without any yield loss, than when the standard continuous shallow (5-2 cm deep) submergence was maintained (Tabbal et al 1992). The difference was attributable mostly to reduced percolation loss because of the absence of standing water in the field. If the weed pressure is high, however, shallow flooding can be maintained from the beginning up to the panicle initiation stage when the field is fully shaded by the crop, and then the continuous saturated soil regime can be established for the remaining period. This practice will save significant amounts of water without reducing yield, but it requires additional labor and supervision. For farmers using a canal irrigation system to adopt such measures, reliable water deliveries must be maintained, an uncommon feature in Asian rice-producing countries. Plastic sheet lining of the bund faces, or sealing the faces with sticky mud at the beginning of the season, can reduce lateral movement of water into the bund, and hence reduce under-bund percolation loss. Development of unpuddled spots can be eliminated by good land leveling and by careful puddling activity. Recent breeding work has continuously reduced the duration of the rice growth period. Such a reduction results in less water demand and has contributed greatly to increasing water use efficiency, especially with yield improvement of new varieties.

Management of water as a scarce resource: issues and options in rice culture

181

Opportunities with wet-seeded rice culture. Availability of early maturing rice varieties and effective herbicides, increasing cost of labor, and declining profitability of rice production have encouraged rice farmers in some countries to switch from transplanted to direct-seeded rice systems. Direct seeding comes in two forms: wet seeding and dry seeding. In wet-seeded rice (WSR), pregerminated seeds are broadcast onto the puddled soil. After the crop is established, WSR is maintained in much the same manner as transplanted rice (TPR). Changing from a transplanted to a wet-seeded rice system automatically advances the farmers’ crop establishment schedule, with a shorter period taken to prepare the land, as seeds require only 24-36 h of soaking and incubation before they are ready for wet seeding. In contrast, in the transplanting period, seedlings are nurtured in the seedbed for 1 month, and farmers have no reason to complete land preparation before the seedlings are ready for transplanting (Bhuiyan et al 1995). The WSR system required 27% less water to complete land preparation than the TPR (Table 1). Because farms were better leveled for facilitating germination, WSR farmers were able to maintain less water depth during crop growth, and the crop had better lodging resistance. Furthermore, WSR gives higher yield than TPR when water stress occurs (Table 2) (Bhuiyan et al 1995). Where WSR is properly introduced, farmers on their own are adopting land and water management practices that lead to better water use efficiency (Bhuiyan et al 1995). Promoting this development involves little cost or risk of failure. In certain areas where the WSR system has been introduced, its popularity has spread very quickly. More studies are needed to determine why WSR adoption is still limited, and how it can be spread more widely. Further opportunities with dry-seeded rice culture. Dry-seeded rice (DSR) technology offers further opportunity for significant water savings in irrigation systems by making more efficient use of rainfall for land preparation and crop establishment. In DSR, nonpregerminated seeds are sown onto dry-plowed soil that is dry or moist, Table 1. Water use, water use period for land soaking and land preparation, and water depth maintained in the field in wet-seeded rice (WSR) and transplanted rice (TPR) in Maligaya, Philippines, 1990-91 dry season.

Water use (mm) Land preparation Crop irrigation Total Time taken to complete land preparation (d) Water depth (cm) at Crop establishment Crop growth Yield (t ha -1 )

WSRa

TPRb

740 1,010 1,750

890 1,300 2,190

6

24

1 6 7

a Turnout service area = 68 ha. b Turnout service area = 35 ha. Source: Bhuiyan et al 1995.

182

Bhuiyan et al

3 7 6.5

Table 2. Yields of wet-seeded rice (WSR) and transplanted rice (TPR) under different water regimes, Maligaya, Philippines, 1990-91 dry season. Yield (t ha -1) Water regime a

W1 W2 W3 W4 W5 W6

WSR

TPR

7.6 7.3 7.0 6.1 6.4 5.2

7.4 6.7 6.3 5.3 6.0 4.2

Difference b 0.2ns 0.6** 0.7** 0.8** 0.4* 1.0**

a W1 = fully irrigated, 5-7 cm depth (no stress), W2 = saturated soil throughout growing season, W3 = mild vegetative stress (10 d without watering starting at 30 DAS or 9 DAT), W4 = severe vegetative stress (same as W3, but stress continued for 20 d), W5 = mild reproductive stress (10 d without watering starting 50 DAS or 29 DAT), W6 = severe reproductive stress (same as W5 but stress continued for 20 d). bns = not significant, ** = significant at 1%, * = significant at 5%. Source: Bhuiyan et al 1995.

but unpuddled. In contrast to TPR, which consumes a large amount of irrigation water in preparing the land for crop establishment, DSR is first established and nurtured by (premonsoon) rainwater as a nonirrigated crop. Later in the season, when the canal water supply has been started, rice may be fully irrigated. Data from Malaysia’s Muda irrigation scheme indicate that this practice could save up to 500 mm of irrigation water (Ho Nai Kin et al 1993). In 1991, when irrigation water could not be released because of very low reservoir storage, farmers grew DSR that yielded an average of 3.9 t ha-1 . In a similar situation in 1978, these farmers could not grow any rice because TPR was their only choice and it could not be established for lack of water. Adoption of DSR allows the irrigation system to achieve better rainfall use and reservoir water conservation.

Irrigation system-level opportunities The irrigation system is more than the sum total of the farms and canals when it comes to the issue of efficiency of water use. Critical determinants of system efficiency are the capacity to control and deliver water in a timely manner, the use of delivered water, communication with water users, quality of feedback, and commitment to cooperation. A recent study of 15 irrigation systems in South and Southeast Asia indicated little systematic measurement of performance by system managers. Wide gaps existed between operational targets and actual achievements, little feedback came from the field, and farmers could not respond to information if it was available. Governments were spending less and less money on system operation and maintenance. Only a few cases showed evidence of concern for maintaining the physical resource base necessary for productive agriculture. The study concluded that improving managerial capacity should be the first step toward performance improvement of these systems

Management of water as a scarce resource: issues and options in rice culture

183

(Murray-Rust and Snellen 1993). Few systematic efforts have been made to remedy defects at the systems level. Recent years have seen a global recognition of the value of consulting and involving water users in water management plans. As the value of water is better internalized by users and is priced realistically, a system of joint responsibility should result in better water use efficiency. A recent review of 208 World Bank-funded irrigation projects around the world showed that Asian rice irrigation systems were unique in that their water efficiency problems stem from incompatibility between design concept and operational objectives. Their design is aimed at slow, continuous water delivery, whereas they are expected to be operated as reticulated systems with capacity to deliver water on demand (World Bank 1994). The problem is exacerbated by the opposing climatological conditions of excess water during the monsoon months—when rice is essentially the only crop grown—and water scarcity during the dry season—when many different crops, including rice, are grown. Most irrigation systems have problems in handling the wet-dry-wet transitions, leading to major sacrifices in water efficiency. Large rice irrigation systems in the humid tropics are mostly designed and operated for continuous flow of canal water regardless of the amount of rainfall occurring in their command area. Nonuse of rainfall and complete dependence on canal supply in the early part of the wet season lead not only to wasted water but also to delayed planting. Adoption of DSR or WSR systems should allow a more efficient use of rainfall and facilitate more intensive cropping. Better use of rainfall in the field enables conservation of water in the reservoir to increase the service area of dry-season irrigation, when water scarcity is acute. Improved use of rainfall will also reduce waterlogging problems in lower areas of the system. In large pump-supported irrigation systems, better use of rainfall can be translated directly into economic benefits derived from reduced pumping cost.

Basin- or watershed-level opportunities The watershed or water basin is the third and final geographic focus in this analysis of water efficiency issues and options in rice culture (the other two being the farm and the irrigation system). Because basin water has multiple uses, off-site effects of increasing water efficiency at the farm or irrigation system level must be carefully assessed. Because water bodies in a basin or watershed are interconnected through the hydrological cycle, water quality must be maintained in lakes, rivers, reservoirs, irrigation and drainage canals, and groundwater. When downstream flow from an irrigation system is the source of water, for example, increased water efficiency upstream may adversely affect the downstream enterprise. Another example is the possible effect of lowering the water table in the groundwater aquifer that supplies water for domestic use, which must depend on pumping from shallow water tables in the same aquifer. Similar issues of water quality should also be considered and properly addressed.

184

Bhuiyan et al

Opportunities for increasing crop water use efficiency in rainfed rice culture Opportunities with crop management technologies Timeliness of establishment and crop intensification. Crop water use efficiency may be increased by improving the timeliness of rice culture relative to the prevailing seasonal conditions. Because up to 700 mm of cumulative rainfall may be needed to complete land preparation for transplanting (Saleh and Bhuiyan 1995, My et al 1995), a significant part of the growing season may be lost by waiting for adequate rains for soil puddling. Late transplanting may reduce productivity because of the consequent reduction in crop duration and yield potential, especially in the traditional, strongly photoperiod-sensitive varieties. Late transplanting may also expose the crop to greater risk from late-season drought. Direct dry seeding allows earlier establishment of the crop than transplanting or wet seeding, because less water is required for land preparation and crop establishment. Earlier seeding offers the prospect of earlier harvest, especially if shorter-duration, less photoperiod-sensitive varieties are used. Such an earlier harvest may reduce exposure of rice to late-season drought, which is often responsible for the greatest yield loss in rainfed lowlands (Fukai et al 1995). Reduced crop duration, with improved synchronization of sensitive stages with periods of the growing season expected to be more favorable on average, should further improve mean yield and its reliability. Earlier harvest may then permit farmers to grow a short-duration postrice crop on residual moisture (Saleh and Bhuiyan 1995, Pascua et al 1998). Thus, the dry-seeded rice system should make the best use of rainwater and offer the prospect of increased cropping intensity in rainfed conditions. Establishment, seedling vigor, and weed control. The traditional system of transplanting rice on puddled soils offers major benefits for weed control. Dry seeding of rice may expose the crop to a number of risks during crop establishment. A rain break after sowing could result in seed loss. The stand may be thinned or lost to seedling drought, or seedling vigor may be impaired. Weeds may emerge before or with the rice seedlings. Seedlings of dry-seeded rice lack the early size advantage of transplants. Rice’s adaptation to anaerobic soil conditions is not helpful until water is ponded and weeds are submerged in the bunded fields as rainfall intensifies later in the season. Transplanting is therefore a compromise for yield stability—less yield may be lost from weed competition but at the cost of a lower yield potential from delayed sowing, reduced system intensification from any foregone second crop, and lower crop water use efficiency. In contrast, dry seeding may offer the prospect of a higher yield potential with the opportunity for a second crop, as long as a suitable plant stand relatively free from weeds can be established. To fully capitalize on the potential advantages of direct dry seeding for both yield and yield stability, research is needed to develop integrated strategies for reliable establishment of direct dry-seeded crops with adequate management or suppression of weeds. The potential contribution of short-residual, postemergence herbicides should be fully explored.

Management of water as a scarce resource: issues and options in rice culture

185

Nutrient balance and sustainability. The traditional system of transplanting rice on puddled soils also offers advantages for nutrient availability, because nutrients are less available in drying soils. A change to dry seeding, with the consequent advance in sowing time, would help capture nitrate formed during the dry-wet transition at the beginning of the growing season, thus reducing N loss to seepage and groundwater (George et al 1994). If the system is intensified with a short-duration legume, some additional benefits may accrue to the N balance, if N loss during the aerobic to anaerobic transition can be minimized (Ladha et al 1996). Given the greater threat from early weed competition in dry seeding, however, proximity of nutrient supply to the roots of the emerging seedling may be important for early vigor, especially for less mobile elements such as phosphorus. Manipulation of controlled-release fertilizer and root system development may be the key to optimizing nutrient release and capture in fluctuating water environments of the rainfed lowlands (Wade et al 1997a). Longterm changes in soil organic matter content and soil nutrient-supplying capacity require further clarification in these contrasting soil conditions (Wade and Ladha 1995).

Opportunities with crop improvement technologies Inherent in the performance of dry-seeded rice in water-limited conditions is the need to establish a uniform, vigorous stand of rice capable of competing with weeds. Although crop management is the basis of any effective strategy for establishing a good stand and competing with weeds, selection of improved cultivars may also be helpful. Most cultivars, whether traditional or improved, have been selected for performance in transplanted conditions. Breeding lines are now evaluated under dry-seeding conditions, and selections are made for quality of plant stand and for seedling vigor (Sarkarung et al 1995). A rapid increase in plant height and rapid expansion of leaf area are usually considered advantageous for weed competitiveness. But yield tradeoffs by incorporating canopy traits for greater competitiveness are not likely to impede crop performance under water-limited conditions (Bastiaans et al 1997). In addition to plant characteristics for more effective integrated weed management, rice varieties can also be selected for reduced exposure to drought. Drought resistance can be achieved by three strategies: escape, avoidance, and tolerance (Ludlow and Muchow 1990). Breeders have been most successful in manipulating drought escape, where exposure to drought is minimized by reducing crop duration or minimizing coincidence of sensitive stages with periods of the growing season in which water deficit is likely. The change to direct seeding, together with selection of short-duration, photoperiod-insensitive cultivars, is a drought escape strategy. This strategy also provides some opportunity to partition water use more efficiently by devoting a larger proportion to grain production. Further gains in water use efficiency should be possible by exploiting drought avoidance and drought tolerance. With the former, the plant avoids drought by extracting additional reserves of soil water, such as by having a superior root system. With the latter, the plant tolerates some desiccation by physicochemical changes, such as osmotic adjustment.

186

Bhuiyan et al

Much effort is currently being directed to developing molecular markers for a greater maximum rooting depth (Champoux et al 1995), a capacity to penetrate hardpans (Ray et al 1996), and a capacity to osmotically adjust to declining water availability (Lilley and Ludlow 1996). Related efforts in physiology are examining whether incorporation of those traits would result in greater extraction of water from the soil in all conditions, or whether other factors could also be involved. Oxygen supply, chemical and physical barriers, rate of stress onset, and root signals could impede water extraction under some conditions, especially in rainfed lowlands (Wade et al 1997b). Further research is required to understand root growth control and water extraction in various rice environments as well as opportunities for their genetic enhancement. Improved water extraction should also be associated with improved nutrient uptake, reduced percolation loss, and reduced accession of nitrates to groundwater. Work to use marker-aided selection for improved drought tolerance is now commencing for maximum rooting depth, hardpan penetration capacity, and osmotic adjustment.

Opportunities for sustaining water quality Nutrient management Nitrogen losses when the wet season begins will be reduced by a change to direct dry seeding because the rice should capture available nitrate before it is lost to denitrification and leaching and water is lost to evaporation and percolation (Wade et al 1998a). Further benefits to nutrient balance and levels of soil organic matter may accrue from incorporation of a legume into the dry-seeded rice system, with the effect dependent upon the duration of the dry period (Ladha et al 1996).

Salinity control Salinization hazard can be reduced by decreasing percolation loss and providing effective drainage facilities for leaching salt from the crop root zone. At the farm level, reduced land preparation period, shallow water depths during crop growth, and proper bund repair will reduce the risk of waterlogging and salinization. The extreme situation, found when the amount of percolation water is reduced, can be avoided by allowing farmers to grow rice only in less permeable soil (Millington 1996). A concomitant use of surface water and groundwater to control water table depth and to maintain a favorable balance of water quality from the two sources offers an opportunity for controlling irrigation-induced soil salinity development (Abrol 1987). In acid sulfate soil areas, it is judicious to limit leaching to periods with high surface water runoff, so that acid and toxic products of the leaching process are easily transported and diluted as much as possible (van Breemen 1993). At the beginning of the rainy season, when the river discharge is low, leaching can reduce the environmental hazard to surface runoff water. Leaching acid sulfate soils with floodwater at the end of the rainy season can improve rice yields (Minh et al 1997b).

Management of water as a scarce resource: issues and options in rice culture

187

Pest management Research has shown unnecessary pesticide use in Asian rice culture (Heong et al 1994). In the Philippines, for example, 80% of the insecticide sprays that farmers applied were found to be unnecessary (Heong et al 1995). In Indonesia, after the introduction of integrated pest management programs and the withdrawal of pesticide subsidies, insecticide use was reduced substantially, with no decline in rice productivity (Ruchijat and Sukmaraganda 1992). Rice breeders, by applying modern biotechnology, may also succeed in developing varieties with insect and disease resistance and thus lessen dependence on the heavy use of pesticides. We may also substitute the more hazardous category chemicals with less hazardous ones without affecting crop productivity (Pingali and Rola 1995). Via judicious water management, we can reduce herbicide use without sacrificing rice yield. Weed control in WSR is adequate with half the recommended dose of herbicide, especially when land is prepared 7-10 d between primary and secondary tillage to allow seeds to germinate after the primary tillage (Bhagat et al 1996). Shallow flooding during the first 45 d after transplanting, followed by maintenance of a saturated-soil regime for the rest of the season, achieved the same yield as conventional water management, but with more than a 30% savings in water (Table 2; Bhuiyan et al 1995).

Conclusions A water crisis for rice is fast approaching. We therefore need to analyze future scenarios and options to guide research directions and national water policies toward more rice production with less water. It is tempting to assume that water efficiency in rice production will improve as water becomes a scarcer resource for rice farmers. But will things really develop that way? A recent analysis of a large number of irrigation systems in both arid and humid areas did not find any significant correlation between water scarcity and irrigation system performance (World Bank 1994). The study indicated that groundwater projects in wet areas did better than those in dry areas. As water scarcity increases, the politically and socially powerful members of the rural community may find ways to secure the limited amounts for themselves first, ignoring the needs of others in the system. In short, we do not have a rational way to prepare for the impending water shortage, to overcome it, and to minimize its effect on food production. Demand for water delivery for the rice crop is often too high and not sustainable. Practical means of addressing the issue have not been available for public-sector irrigation systems because operating agencies did not have full control over water. Sustainable means of improving control over the resource must be found to increase water efficiency in rice irrigation systems. Although some improvements in our capacity to handle the decreasing availability of water for rice culture seem feasible, no clear picture has emerged on how severe the present level of deterioration of water quality is and what can be expected in the future. Consciousness of water-quality problems as affected by agricultural practices,

188

Bhuiyan et al

such as nitrogen and pesticide use, has just begun to grow in most Asian rice-producing countries. It will be unwise to permit water-quality-degradation processes to continue until the problem has magnified to dangerous proportions. Studies of the degree of water-quality degradation and correction options are essential. The basic ingredients of developing and implementing water-efficient rice production systems seem to be in place now. The immediate challenge lies in tailoring these systems to suit local conditions and farming communities and, at the same time, in changing long-standing practices and operational procedures in the farm and irrigation system. A special challenge in addressing these issues is that new ideas and initiatives have to be tested without adversely affecting farmers’ production or income.

References Abrol IP. 1987. Salinity and food production in the Indian Subcontinent. In: Jordan WR, editor. Water and water policy in world food supplies. Proceedings of the Conference, 26-30 May 1985. Texas A&M University, College Station, Texas, USA. p 109-113. Bastiaans L, Kropff MJ, Kempuchetty N, Rajan A, Migo TR. 1997. Can simulation models help design rice cultivars that are more competitive against weeds? Field Crops Res. 51(1/2):101-111. Bhagat RM, Bhuiyan SI, Moody K. 1996. Water, tillage and weed interactions in lowland tropical rice: a review. Agric. Water Manage. 31(1996):165-184. Bhuiyan SI, Sattar MA, Khan MAK. 1995. Improving water use efficiency in rice irrigation through wet-seeding. Irrig. Sci. 16(1):1-8. Castañeda AR, Bhuiyan SI. 1991. Nitrate-nitrogen concentrations in shallow groundwater underneath ricefields. Phil. J. Crop Sci. 16(2):57-62. Castañeda AR, Bhuiyan SI. 1996. Groundwater contamination by ricefield pesticides and some influencing factors. J. Environ. Sci. Health 31:83-90. Champoux MC, Wang G, Sarkarung S, Mackill DJ, O’Toole JC, Huang N, McCouch SR. 1995. Locating genes associated with root morphology and drought avoidance in rice via linkage to molecular markers. Theor. Appl. Genet. 90:969-981. Cheng HH. 1990. Pesticides in soil environment: processes, impacts and modeling. Madison, Wis. (USA): Soil Science Society of America. Dent D. 1992. Reclamation of acid sulphate soils. In: Lal R, Stewart BA, editors. Advances in soil science. New York: Springer-Verlag. 17:79-122. De Datta SK. 1981. Principles and practices of rice production. New York: John Wiley and Sons. 618 p. Dogra B. 1986. The Indian experience with large dams. In: Goldsmith E, Hildyard N, editors. United Kingdom: Wadbridge Ecological Center. p 201-208. FAO/WHO (Food and Agriculture Organization/World Health Organization). 1977. Pesticide residues in food. Report of the 1976 Joint Meeting of the FAO Panel of Experts on Pesticide Residues and the Environment and the WHO Experts Group on Pesticide Residues, Rome, Italy. Follet RF, Walker DJ. 1989. Groundwater quality concerns about nitrogen. In: Nitrogen management and groundwater protection. Amsterdam (Netherlands): Elsevier Science Publishers B.V.

Management of water as a scarce resource: issues and options in rice culture

189

Fukai S, Rajatsasereekul S, Boonjung H, Skulkhu E. 1995. Simulation modeling to quantify the effect of drought for rainfed lowland rice in northeast Thailand. In: Fragile lives in fragile ecosystems. Proceedings of the International Rice Research Conference, 13-17 Feb 1995. Manila (Philippines): International Rice Research Institute. p 657-674. George T, Ladha JK, Garrity DP, Buresh RJ. 1994. Legumes as nitrate catch crops during the dry-to-wet season transition in lowland rice-based cropping systems. Agron. J. 86:267273. Ghani MA, Bhuiyan SI, Hill RW. 1989. A model to evaluate intensive vs extensive irrigation practices for rice production in Bangladesh. Agric. Water Manage. 20:233-244. Gleick PH, editor. 1993. Water in crisis: a guide to the world's fresh water resources. New York (USA): Oxford University Press. Gumtang RJ, Pampolino MF, Tuong TP. 1998. Groundwater dynamics and quality under intensive cropping systems. J. Exp. Agric. (In press.) Heong KL, Escalada MM, Vo Mai. 1994. An analysis of insecticide use in rice: case studies in the Philippines and Vietnam. Int. J. Pest Manage. 40: 173- 178. Heong KL, Escalada MM, Lazaro AA. 1995. Misuse of pesticides among rice farmers: Leyte, Philippines. In: Pingali PL, Roger PA, editors. Impact of pesticides on farmer health and the rice environment. Norwell, Mass. (USA): Kluwer Academic Publishers, and Manila (Philippines): International Rice Research Institute. p 97- 108. Ho Nai Kin, Cheng CM, Murat M, Ismail MZ. 1993. MADA's experiences in direct seeding. Paper presented at the workshop on Water and Direct Seeding for Rice, 14-16 June 1993, Muda Agricultural Development Authority, Ampang Jajar, Alor Setar, Malaysia. Hossain M. 1995. Sustaining food security for fragile environments in Asia: achievements, challenges, and implications for rice research. In: Fragile lives in fragile ecosystems. Proceedings of the International Rice Research Conference, 13-17 Feb 1995. Manila (Philippines): International Rice Research Institute. p 3-23. IRRI (International Rice Research Institute). 1989. IRRI toward 2000 and beyond. Manila (Philippines): IRRI. IRRI (International Rice Research Institute). 1993. IRRI rice almanac, 1993-1995. Manila (Philippines): IRRI. IRRI (International Rice Research Institute). 1995. Water-a looming crisis. Manila (Philippines): IRRI. IRRI (International Rice Research Institute). 1998. Medium-term plan. Manila (Philippines): IRRI. Khosla BK, Gupta RK, Chawla KL. 1980. Evaluating the field hydraulic conductivity and soil salinization under conditions of high water table. In: Proceedings of the international symposium on salt affected soils. Karnal (India): Central Soil Salinity Research Institute. p 256-264. Klepper O, Hatta GM, Chairuddin G. 1990. Environmental impacts of the reclamation of acid sulphate soils in Indonesia. IARD J. 12(2):29-34. Ladha JK, Kundu DK, Van Coppenolle G, Peoples MB, Carangal VR, Dent PJ. 1996. Legume productivity and soil nitrogen dynamics in lowland rice-based cropping systems. Soil Sci. Soc. Am. J. 60:183-192. Lilley JM, Ludlow MM. 1996. Expression of osmotic adjustment and dehydration tolerance in diverse rice lines. Field Crops Res. 48:185-197. Lim GS, Ong SH. 1977. Pesticide benefits and side effects. Southeast Asian workshop on pesticide management, Bangkok, Thailand.

190

Bhuiyan et al

Ludlow MM, Muchow RC. 1990. A critical evaluation of the traits for improving crop yields in water limited environments. Adv. Agron. 43:107-153. Millington P. 1996. The basin management context. Keynote presentation at the Regional Consultation Workshop Towards a Policy for Water Resource Development and Management in the Asian and Pacific Region, The Asian Development Bank, 10-14 May 1996, Manila, Philippines. Minh LQ, Tuong TP, van Mensvoort MEF, Bouma J. 1997a. Contamination of surface water as affected by land use in acid sulphate soils in the Mekong Delta, Vietnam. Agric. Ecosys. Environ. 61 (1): 19-27. Minh LQ, Tuong TP, van Mensvoort MEF, Bouma J. 1997b. Tillage and water management for increasing riceland productivity in flood prone acid sulphate soil areas of the Mekong Delta, Vietnam. J. Soil Tillage 42:l-14. Murray-Rust DH, Snellen WB. 1993. Irrigation system performance assessment and diagnosis. IIMI, ILRI, and IIHEE. 148 p. My TV, Tuong TP, Xuan VT, Nghiep NT. 1995. Dry seeding rice for increased cropping intensity in Long An Province, Vietnam. In: Denning GL, Xuan VT, editors. Vietnam and IRRI: a partnership in rice research. Manila (Philippines): International Rice Research Institute, and Hanoi (Vietnam): Ministry of Agriculture and Food Industry, p 111-122. Pascua SR, Ladha JK, Ventura W, Agustin EO, Valencia DV, Marcos TF, Sta. Cruz PC, Obien SR. 1997. Yield trends and apparent nutrient balances in intensified and diversified cropping systems of Ilocos Norte, Philippines. Exp. Agric. (In press.) Pingali PL, Rola AC. 1995. Public regulatory roles in developing markets: the case of pesticides. In: Pingali PL, Roger PA, editors. Impact of pesticides on farmer health and the rice environment. Manila (Philippines): International Rice Research Institute. p 391-412. Postel S. 1989. Water for agriculture: facing the limits. World Watch Paper 93. Washington, D.C. (USA): World Watch Institute. Ray JD, Yu L, McCouch SR, Champoux M, Wang G, Nguyen HT. 1996. Mapping quantitative trait loci associated with root penetration ability in rice ( Oryza sativa L.). Theor. Appl. Genet. 92:627-636. Ruchijat E, Sukmaraganda. 1992. Material integrated pest management in Indonesia: its successes and challenges. In: Ooi P, Lim G, Ho T, Manulo P, Waage J, editors. Integrated pest management: the Asia-Pacific region. CABC and ADB. p 329-347. Saleh AFM, Bhuiyan SI. 1995. Crop and rainwater management strategies for increasing productivity of rainfed lowland rice systems. Agric. Sys. 49:259-276. Sarkarung S, Singh ON, Roy JK, Vanavichit A, Bhekasut P. 1995. Breeding strategies for rainfed lowland ecosystem. In: Fragile lives in fragile ecosystems. Proceedings of the International Rice Research Conference, 13-17 Feb 1995. Manila (Philippines): International Rice Research Institute. p 709-720. Sarma PBS. 1986. Water resources and their role in food production. In: Swaminathan MS, Sinha SK, editors. Global aspects of food production. Oxford, Riverton, N.J. (USA): Tycooly International. p 171-198. Tabbal DF, Lampayan RM, Bhuiyan SI. 1992. Water-efficient irrigation technique for rice. In: Murty VVN, Koga K, editors. Soil and water engineering for paddy field management. Proceedings of the International Workshop on Soil and Water Engineering for Paddy Field Management, Bangkok, Thailand, 28-30 Jan 1992. Bangkok: Asian Institute of Technology. p 146-159.

Management of water as a scarce resource: issues and options in rice culture

191

Tuong TP, Cabangon RJ. 1996. Reducing bypass flow during land soaking of cracked rice soils. In: Kirichoff G, So B, editors. Management of clay soils in rainfed rice-based cropping systems. Proceedings of a workshop held at the Bureau of Soil and Water Management, Quezon City, Philippines, 20-24 Nov 1995. Proceedings No. 70. Canberra (Australia): Australian Centre for International Agricultural Research. p 237-242. Tuong TP, Wopereis MCS, Marquez J, Kropff MJ. 1994. Mechanisms and control of percolation losses in irrigated puddled rice fields. Soil Sci. Soc. Am. J. 58:1794-1803. Valera A. 1977. Field studies on water use and duration for land preparation for lowland rice. Ph.D. dissertation. University of the Philippines at Los Baños, Philippines. van Breemen N. 1993. Environmental aspects of acid sulphate soils. In: Dent DL, van Mensvoort MEF, editors. Selected papers on the Ho Chi Minh City symposium on acid sulphate soils. International Institute for Land Reclamation and Improvement Publication No 53. Wageningen (Netherlands). p 39 1-402. Viets FG Jr, Hageman RH. 1976. Factors affecting the accumulation of nitrate in soil, water, and plants. Agriculture Handbook No. 413. Washington, D.C. (USA): Agricultural Research Service-United States Department of Agriculture. Wade LJ, Ladha JK. 1995. The fate of organic matter and nutrients in lowland rice systems. In: Lefroy RDB, Blair CJ, Craswell ET, editors. Soil organic matter management for sustainable agriculture. ACIAR Proceedings Number 56. Canberra (Australia): ACIAR. p 115119. Wade LJ, George T, Ladha JK, Singh U, Bhuiyan SI, Pandey S. 1998a. Opportunities to manipulate nutrient-by-water interactions in rainfed lowland rice systems. Field Crops Res. 56:93-112. Wade LJ, Moya TB, Pantuwan G, Regmi KR, Samson BK. 1998b. Research at IRRI on rice root systems for drought resistance. In: Morita S, Abe J, editors. Perspective on ideotype of rice root system. Singapore: World Scientific Publishing. (In press.) Wickham TH, Sen LN. 1978. Water management for lowland rice: water requirements and yield response. In: Soils and rice. Manila (Philippines): International Rice Research Institute. p 649-669. World Bank. 1994. A review of World Bank experience in irrigation. Report No. 13676. Washington, D.C. (USA): World Bank. 126 p.

Notes Authors’ address: International Rice Research Institute, P.O. Box 933, Manila, Philippines. [email protected] [email protected] [email protected] Citation: Dowling NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute.

192

Bhuiyan et al

CHAPTER 12

Securing the future of intensive rice systems: a knowledgeintensive resource management and technology approach L.M.L. Price and V. Balasubramanian Scientific achievements in increasing yields have been fast and profound in Asia's intensive rice systems, but farmers' knowledge and corresponding practices have not kept pace, particularly in disease, pest, nutrient, and water management. Knowledgeintensive resource management and technology can be used to fine-tune farmer management to enhance profitability and environmental protection in high-productivity systems. This paper identifies, defines, and discusses two strategies: (1) KIT-P, knowledge physically embedded in machines and instruments that provide field-level information to farmers, and (2) KIT-H, knowledge embedded in the farmers themselves and composed of information directly linked to cognition and acquired through a process of learning and experimentation. Making more knowledge and information available to farmers is one way of addressing the problems of resource depletion in both quantity and quality, of degradation of the environment, and of increased health risks caused by lack of appropriate knowledge in managing changes in cropping systems. Knowledge-intensive approaches are expected to serve farmers in decision making and in enhancing precision as they come to terms with the Green Revolution of the past and face future challenges.

International agricultural research has generated technologies that have changed the face of rice production in Asia. No longer do we hear the predictions of massive starvation in Asia that were given prior to the Green Revolution. The annual productivity growth of rice, the staple food of Asia, kept pace with population growth from the 1960s to the mid-1980s (Herdt and Capule 1983, Dalrymple 1986, Hossain and Fischer 1995). The Green Revolution strategy, started in the early 1960s, was based on the use of modern rice varieties, assured irrigation, and subsidies for fertilizers, pesticides, and farm equipment. It provided food security to people for more than three decades and minimized the extension of food crop cultivation to ecologically fragile, marginal lands. But we are only beginning to understand that intensive ricecropping systems and crop and resource management technologies used for the Green Revolution may have adversely affected the environment.

193

Although IRRI continues to work to increase productivity to meet predicted demands for rice through plant breeding and biotechnology, its rice scientists also realize that they have reached a new frontier that requires more expertise on the part of the farmer. Environmentally sound management practices can greatly reduce stress on the resource base, raise farm profits, and improve farmer health and productivity. New knowledge-intensive management practices and technologies are most urgently needed where traditional knowledge has not been applicable to methods of intensive rice production from the Green Revolution. Farmers are unable to manage nutrient inputs to meet crop and soil needs, are erroneously applying insecticides, and are illequipped to understand and control diseases such as blast and tungro. In addition, farmers rapidly change their rice cultivation systems in response to higher costs, particularly labor. The labor savings of direct-seeded rice bring issues of weed management and herbicide use to the foreground. The future of intensive rice systems that looms before us is one of ever-decreasing natural and human resources. Scientific achievements in increasing yields have been fast and profound in Asia’s intensive rice systems, but farmers’ knowledge and corresponding practices have not kept pace, specifically in disease, pest, nutrient, and water management. This paper calls for increased attention to knowledge-based management and technology application at the farm level to tackle the problems of highly productive agriculture. We identify key productivity and environmental considerations, explore knowledge-intensive approaches for crop management, and suggest directions that will help to develop strategies to generate and diffuse knowledge-intensive resource management approaches and technologies.

Environmental degradation and dwindling resources Intensive cropping methods pursued in the past three decades have led to a significant depletion of resources both in quantity and quality, degradation of the environment, and increased health risks to producers and consumers (Cassman and Pingali 1995, Gardner 1996). The long-term implications of using chemical inputs and regarding them as routine, prophylactic practices were not known at that time. The Green Revolution and government and development agencies focused mainly on the goal of increasing yields of target crops with improved varieties and routine application of inputs at recommended rates.

Nutrients Soil degradation is caused by nutrient imbalance (deficiency or toxicity), salinity/ alkalinity, waterlogging, subsoil compaction, and declining organic matter quality and soil N supply (Cassman et al 1993, Kundu et al 1995). Because nutrient management needs are predominantly farm- and field-specific, knowledgeable decisions on how much and when to apply nutrients are required. Decision-making success is typically expressed in crop yields.

194

Price and Balasubramanian

Rice yield response follows a diminishing return function with increasing N application. Yields could decrease further because of lodging and increased incidence of pests and diseases at high N levels. Good crop and water management on research stations has resulted in higher fertilizer N (50-60%efficiency) (Cassman et al 1994), but such precision management practices must be economically viable when all costs are considered. To obtain the projected grain yield of 8 t ha-1 in irrigated rice by the year 2025, it is necessary to apply 280 kg N ha-1 at 33% fertilizer N recovery efficiency (Cassman and Pingali 1995). This means that urea fertilizer applied to irrigated rice in Asia would increase from 15.5 to 43.6 million t—nearly a 300% increase in N for a 63% increase in yield. By increasing fertilizer N recovery efficiency to 50%, it is possible to reduce the N application rate from 280 to 187 kg ha-1 and the urea fertilizer need from 43.6 to 29.1 million t—still a 200% increase in fertilizer N application for a 63% increase in yield (Cassman and Pingali 1995). Governments in Asia are moving away from fertilizer subsidies—a practice begun in the 1970s. Although the subsidy structure may have accounted for excessive growth in fertilizer inputs to intensive rice systems, farmers’ current decisions on what can profitably be used in crop production may be guided more by fertilizer market prices. The situation is essentially the same because decision making on inputs is driven by price and external recommendations. Efficient fertilizer use will produce healthy plants that are less vulnerable to pests and diseases and to lodging. Optimal crop management requires farmer knowledge of matching inputs to crop production needs (Pingali et al 1995). The overuse or improper use of nutrient inputs is highly damaging to crops and the environment (FFTC 1994). Excess nitrates pollute not only the soil and groundwater but also the produce itself. Nitrous oxide released from denitrification of nitrates pollutes the air. Recent studies indicate increased nitrate levels above the permissible limit of 10 ppm NO 3-N in well water because of excess fertilizer application to the pepper crop after rice in Batac, northern Philippines (J.K. Ladha, IRRI, Philippines, 1996, personal communication). Nitrate in food or drinking water is a hazard to human health. Therefore, to minimize health risks, new methods or products must be developed to achieve a more efficient use of nutrient sources. Efficient fertilizer use will minimize water pollution by nitrates and phosphates, and will reduce the accumulation of free nitrates in food.

Pesticides Likewise, pesticide inputs are historically not matched to crop needs in intensive systems. The recommendation for crop protection set in the 1970s, consisting of prophylactic calendar-based spraying, was not based on actual pest infestations and crop loss calculations. This recommendation has been adopted by farmers as a standard procedure; in many places, calendar-based chemical pest control, particularly for insects, is still being recommended. Such an approach to crop protection undermines environmental integrity, host-plant resistance, human health, and, ultimately, the prof-

Securing the future of intensive rice systems

195

itability of the farming enterprise (Heong et al 1995, Widawsky 1996, Price 1995, Rola and Pingali 1993). Optimal use of inputs is a growing concern. Factors other than meeting crop needs affect optimum productivity and farmer welfare. Health effects from pesticide exposure include a whole range of medical problems from acute pesticide poisoning to symptoms of ill health from long-term exposure (Pingali et al 1992, Rola and Pingali 1993). Rola and Pingali (1993) conclude that “prolonged and frequent exposure to pesticides impairs farmers’ health and hence their productivity. The more frequent the insecticide applications, the higher are the health costs, treatment costs, and opportunity cost of time lost. Explicit accounting for health costs substantially raises the cost of using pesticides. The value of the crop lost to pests is invariably lower than the cost of treating pesticide-caused [human] disease. When health costs are factored in, the natural control (‘do nothing’) option is the most profitable and useful pest control strategy.”

Conserving resources Increasing input efficiency not only minimizes environmental pollution and health risks but also conserves the nonrenewable sources of fertilizers, such as fossil fuels and minerals. At the projected rate of consumption, known oil reserves will limit food production in 50 yr, phosphorus deposits will be depleted in 90 yr, and other minerals (K, Mg, trace elements) will become increasingly scarce. Technologies that recycle nutrients efficiently and maximize biological N fixation have to be used increasingly to prolong the availability of these nonrenewable resources.

A knowledge-intensive approach defined Two primary knowledge-intensive crop and resource management approaches exist: (1) knowledge imbedded in machines/instruments, and (2) knowledge imbedded within farmers themselves. The first encompasses physical technology—in which the knowledge or expertise to enhance decision making is imbedded in the physical technology itself, termed here “knowledge-intensive technology-physical” (KIT-P). The second is knowledge as it is held by people—which is directly linked to cognition and is acquired and retained through a process of learning and experimentation (KIT-H). Information is narrower in scope than knowledge and implies a random collection of material rather than an orderly synthesis. Knowledge is thus a system of cognition and interpretation; it is dynamic in that it builds upon itself empirically through trial and error. While knowledge acts as a foundation for building upon, it is also a foundation for interpretation; it acts as a filter through which we interpret our new observations. KIT-P provides farmers with information to enhance precision, but the data generated must still be interpreted so that appropriate action can be taken. KIT-H provides knowledge, but an orderly cognitive synthesis must take place in the minds of farmers if KIT-H approaches are to be valid.

196

Price and Balasubramanian

Traditional research produced the Green Revolution seed technology, but seed technology is becoming increasingly complex. Host-plant resistance (HPR), for example, requires a different management method. Studies have shown that the message of resistance has not filtered down to farmer behavior and farmers continue to spray resistant varieties. Although HPR is an effective substitute (economically) for pesticides, there is increasing evidence that HPR must be accompanied by knowledge as a substitute for pesticides (Widawsky 1996).

Knowledge imbedded in physical technology (KIT-P) KIT-P is distinct from other modem technologies in that it enhances decision making through information, whereas other modern technologies, such as seed technology, although the product of much scientific knowledge, do not. An excellent example of KIT-P is the Ag Leader Yield Monitor 2000 developed by Ag Leader Technology in Ames, Iowa (USA). This KIT harvester combines a digital device, global positioning system, and transducer behind a plate. The amount harvested is continuously measured by the force striking the plate and values are fed into the processor. Values are corrected for factors that include the speed of the combine and moisture content of the crop. One acre of land might be broken up into 500 measurement units. Farmers can then identify sections of their fields that have production shortfalls. With this information, they can call in outside assistance for soil testing, make investment decisions on upgrading selected field areas, and use their data to validate claims made by commercial enterprises such as projected yields of seed by private-sector enterprises or government agricultural extension agents (Hapgood 1995).

Knowledge imbedded in human beings (KIT-H) KIT-H approaches to resource management are concerned with imparting learning that fits the structured cognition of farmers in an orderly and synthetic fashion. In building knowledge-intensive approaches to resource management, scientists are operating on principles in a scientific tradition, one with explicit notions and methods of verification of cause-and-effect relationships. Farmers—within their various cultural perspectives and traditions—also have understandings of cause-and-effect relationships that must be addressed in any knowledge-intensive approach. The science of building knowledge involves theory, fact, observation, and probability. Western science has provided us with the documented history of the importance of a framework within which to interpret facts. The theoretical context within which the facts are interpreted will ultimately influence interpretations and conclusions drawn from experiments. Observations and facts are not sufficient— contextualizing theory must be included. Before the introduction and acceptance of statistical probability, facts and observations were at the mercy of equally plausible interpretations. Science no longer seeks the certainty of predicting all instances but uses statistics as a tool to gauge the degree to which a theory will give us predictive power.

Securing the future of intensive rice systems

197

When we develop a knowledge-intensive approach, we are in essence drawing on a great tradition and the lessons learned along the way. Transferring only one component of the scientific process—for example, only facts or observations—may be at the expense of full absorption and sustainability of knowledge. Integrated pest management with the farmer field school approach (IPM FFS) is an example of KIT-H. The pest management package, however, includes both information and knowledge. Although information may be easily incorporated, knowledge of crop management must compete with knowledge systems already in place. The IPM FFS approach involves a system in which observations are made, facts are highlighted, and observations and facts are placed in a framework of ecological theory. Farmers also learn about probability of infestations (through monitoring techniques) and yield loss (economic thresholds). Social reinforcement of the learning process occurs through the learning interactions of FFS students.

Combining physical technology and human learning: the chlorophyll meter method for better timing of N application Significant spatial and temporal crop yield variations are common in any field because of variations in microclimate, soil type, soil flora and fauna, organic matter content and quality, nutrient status, drainage, pest and disease incidence, and weed infestation (Hapgood 1995). Hapgood maintains that the same input will not produce the same output from one year to the next, nor do any two fields on one farm produce the same yield in the same year. IRRI researchers are developing improved techniques to predict soil N supply and in-season plant N status. Peng et al (1995) adapted the chlorophyll meter method to measure the leaf N status of rice and to synchronize N application with crop demand. The meter readings (also called SPAD values) are calibrated with rice leaf N concentrations and critical meter values are established to determine the need for N application. For example, 35 is the critical SPAD value for transplanted semidwarf indica varieties in irrigated systems during the dry season; whenever the meter reading falls below 35, a topdressing of 30-40 kg N ha-1 is recommended. The chlorophyll meter can be used to handle soil variability by adjusting N application to crops, based on variable soil N supply and crop demand in different fields or different parts of a large farm. It can also be used to diagnose soil or crop problems that affect plant N uptake and yield. Several factors—such as cultivar, plant population, stage of growth, and biotic and abiotic stresses that cause leaf chlorosis—affect chlorophyll meter readings (Peterson et al 1993, Turner and Jund 1994). Therefore, the SPAD meter should be calibrated on the basis of cultivar group, system of crop establishment, plant density, and environmental conditions prevalent in each location. If meter readings are properly calibrated according to cultivar group under local conditions, the chlorophyll meter can be a good tool for fine-tuning the N fertilization of a crop and for correcting N deficiency within the same season. For proper calibration, an education component to accompany the physical technology must be developed.

198

Price and Balasubramanian

National agricultural research systems are evaluating the chlorophyll meter method in farmers’ fields. Early results indicate that the method works well on transplanted rice; the critical SPAD value may require some adjustment for different seasons and systems of crop establishment (e.g., direct-seeded rice) (Turner and Jund 1994, Balasubramanian et al 1998).

Measuring KIT-H in farmers The popularity of IPM FFS as a KIT-H approach continues to spread across the globe. The challenge lies in conducting empirical investigations and developing measures of how knowledge is absorbed, acted upon, and transferred on a farmer-to-farmer basis. Knowledge as it is expressed in farmer cognition and decision making can be measured with a combination of ethnosemantic elicitation, analysis of cognitive domains (factual knowledge), and expert systems/knowledge-based systems to model decision making. (Knowledge-based systems computer software allows for both absolute and probabilistic statements, reasoning from the rule antecedent or the consequence [“if ” and “then” in sequence]. Ethnosemantic elicitation is a well-tested tool used in anthropology by ethnobotanists and ethnozoologists to capture the structure of cognitive domains.) Procedural knowledge is contained in the rules, whereas the values of the variables included in the rules represent the factual knowledge (Guillet 1989). Together, procedural and factual knowledge systems represent one possible strategy for revealing the relationship between knowledge and action on the part of farmers. Locating values representing cognitive absorption, decision making, and transfer is a high priority and is needed to develop accurate measures of returns to investment and techniques for analyzing the economic impact of knowledge and knowledge transfer.

The process of knowledge incorporation The transfer of knowledge based on scientific principles aimed at altering farming practices requires a good fit between the knowledge system of the farmers and that of the scientists. If new components were added to the existing knowledge system and if these were couched in familiar terms, there would be latitude for experimentation on the local level that could eventually develop into a functional fit. A scientific (versus local or indigenous) interpretation may not be feasible because of the high cost of education and uncertain desirability of replacing the foundation of indigenous practices, many of which may be environmentally sound. Much of what we currently see as mismanagement in intensive rice systems is the farmer response to a lack of appropriate knowledge in managing Green Revolution changes in cropping systems. The “blanket recommendation approach” gave farmers information without understanding—it provided information but did not expand knowledge. This led to the continuation of practices deemed scientifically unsound on the basis of contemporary research. Farmers continue to engage in what we now know as dangerous behavior (in terms of health, productivity, and environmental protection). But farmers’ behavior is consistent with their assessments of the probability

Securing the future of intensive rice systems

199

of success, given their interpretation of the options available. These judgments are embedded in and filtered through their knowledge base coupled with information that remains outside that base. From their perspective, farmers act to enhance their probability of obtaining success. Nazarea (1996) provides evidence that farmers slowly lost confidence in indigenous knowledge (the ethnoscientific knowledge base) during the Green Revolution accompanied by a desire to manage their crops on scientific principles. It is therefore probable that (1) these farmers have become more reliant on external recommendations and (2) appropriate indigenous/local knowledge is absent or farmers are not willing to use the indigenous knowledge that is appropriate to managing intensive systems. Documenting actual decision making is necessary to bring to the foreground constraints to the implementation of knowledge if absorption is present. Economic constraints may force farmers to act in ways inconsistent with their environmental/agronomic knowledge base. For example, labor demand for monitoring environmental phenomena and calculating thresholds is high. Labor is a common production constraint in intensive rice systems and it may constrain the implementation of knowledge-intensive crop management practices. It is therefore important to evaluate impact with instruments that uncover these distinctions in decision making.

Support systems Systematic attempts to develop, test, transfer, and track knowledge and knowledgeintensive physical technologies need multidisciplinary planning and strategic research, government support, and farmer participation. Diffusion of knowledge-intensive technologies to farmers is more difficult than distribution of improved seeds of new varieties. Several institutions and organizations are involved in this process—education and training groups, extension services, banks for credit, input suppliers, machinery companies and contractors, traders, market outlets, rural infrastructure, and policymaking bodies. All have to perform effectively in a coordinated manner to maximize adoption of knowledge-intensive management technologies.

Training and education Wherever feasible, farmers should be involved as opinion givers or as active partners in the generation, adaptation, and diffusion of new knowledge and technologies. Farmer participation and contributions are considerable in evaluating a new knowledge or technology. For a completely new KIT-P, they play a consultative role, providing a valuable input. We can reinforce the individual farmer’s opinions and assessments in group discussions. Wherever possible, working with farmer groups or associations is better. KIT-H can also be best served with a model that incorporates active learning among farmers and a social group to reinforce learning. Farmers will need support for various combinations of KIT-P and KIT-H. For example, pheromone traps are being tested to control yellow stem borers in rice (K. Krishnaiah, Directorate of Rice Research, Hyderabad, India, 1996, personal commu-

200

Price and Balasubramanian

nication). Farmers should therefore know about the life cycle of the yellow stem borer, the threshold levels of damage, or how borers are attracted by pheromones before they can correctly apply the pheromone technology in their fields. Once the principles are well understood, farmers themselves can make minor changes in the application of a technology to increase its effectiveness. Similarly, farmers must be educated on the proper use of pest- and disease-resistant rice varieties before such varieties are deployed.

Institutional support Availability of agricultural credit, timely supply of inputs, availability and quality of contract services and machinery for different farm operations, and repair and maintenance services in rural areas will influence the rate of adoption of new KIT-P. In promoting new machines to Asian farmers, it is important to implement certain steps to increase and sustain adoption—standardize the new machines and spare parts to assure quality, provide a warranty for specified periods, train farmers in the correct use and maintenance of the machines, assure after-sales service and follow-up inspections and advice, and provide repair services in rural areas within easy reach of farmers. Similar iterative steps have to be developed and tested for knowledge-intensive resource education (KIT-H).

Policy support The lack of a mechanism to take promising technologies to the field for farmer evaluation and the absence of government action plans to mobilize necessary institutional and policy support often hinder farmer adoption (Tandon 1989). The national bureaucracy and government policies must be favorable to the process of technology assessment, adaptation, and promotion. Some countries have realized the importance of farmer assessment and use of new technologies in achieving impact on food production. They are therefore developing technology assessment units for on-farm evaluation and adaptation of new technologies in target areas. A good example is Indonesia, where 17 assessment institutes for agricultural technologies are being developed to undertake location-specific adaptive research and technology evaluation.

Conclusions Intensive cropping methods pursued in the past three decades have led to a significant depletion of resources in both quantity and quality, degradation of the environment, and increased health risks to producers and consumers. Much of what we currently recognize as mismanagement in intensive rice systems is a farmer response to the lack of appropriate knowledge on managing Green Revolution changes in cropping systems. The “blanket recommendation approach” gave farmers information without understanding—it provided information but did not expand knowledge. This paper has examined the potential of knowledge-intensive technologies for enhancing appropriate precision management of intensive rice production systems. It stresses the importance of distinguishing between physical technologies in which

Securing the future of intensive rice systems

201

knowledge is embedded in machines and instruments (KIT-P) and knowledge that is embedded within farmers as human beings (KIT-H). Both approaches require systematic multidisciplinary planning and strategic research, government support, and farmer participation in attempts to develop, test, transfer, and track knowledge and knowledge-intensive physical technologies. The potential for knowledge-intensive technologies is tremendous—not only for the protection of crops, natural resources, and human health of present and future generations of farmers but also for empowering farmers to validate claims of commercial agricultural enterprises and extension alike.

References Balasubramanian V, Morales AC, Cruz RT, Abdul Rachman S. 1998. On-farm adaptation of knowledge-intensive nitrogen management technologies for rice systems. Nut. Cyc. Agroecosys. (In press.) Cassman KG, Kropff MJ, Gaunt J, Peng S. 1993. Nitrogen use efficiency of rice reconsidered: what are the key constraints? Plant Soil 155/156:359-362. Cassman KG, Kropff MJ, Yan ZD. 1994. A conceptual framework for nitrogen management of irrigated rice in high-yield environments. In: Virmani SS, editor. Hybrid rice technology: new developments and future prospects. Manila (Philippines): International Rice Research Institute. p 81-96. Cassman KG, Pingali PL. 1995. Intensification of irrigated rice systems: learning from the past to meet future challenges. GeoJoumal 35(3):299-305. Dalrymple DG. 1986. Development and spread of high-yielding rice varieties in developing countries. Washington, D.C. (USA): Metrotec, lnc. FFTC (Food and Fertilizer Technology Center). 1994. Fertilizer use and sustainable food production. FFTC Newsletter 104(June 1994):4-5. Gardner G. 1996. Preserving agricultural resources. In: Brown LR et al, editors. State of the world 1996. New York (USA): World Watch Institute and W.W. Norton and Company. p 78-94, 208-213. Guillet D. 1989. Expert-systems applications in anthropology. Anthropol. Quart. 64(2):59-67. Hapgood F. 1995. High-tech harvest. Inc. Technol. 3:52-55. Heong KL, Escalada MM, Lazaro AA. 1995. Misuse of pesticides among rice farmers in Leyte, Philippines. In: Pingali PL, Roger PA, editors. Impact of pesticides on farmer health and the rice environment. Massachusetss (USA): Kluwer Academic Publishers. Herdt RW, Capule C. 1983. Adoption, spread, and production impact of modern rice varieties in Asia. Manila (Philippines): International Rice Research Institute. Hossain M, Fischer KS. 1995. Rice research for food security and sustainable agricultural development in Asia. GeoJoumal 35(3):286-298. Kundu DK, Ladha JK. 1995. Enhancing soil nitrogen use and biological nitrogen fixation in wetland rice. Exp. Agric. 31:261-277. Nazarea VD. 1996. Ethnoecology of the Manupali watershed. Unpublished Phase I Report of the Ethnoecology of the Manupali Watershed Project. SANREM CRSP, University of Georgia, Georgia Station, Griffin, Georgia, USA.

202

Price and Balasubramanian

Peterson TA, Blackmer TM, Francis DD, Schepers JS. 1993. Using a chlorophyll meter to improve N management. A Nebguide in Soil Resource Management: D-13, Fertility. Lincoln, Neb. (USA): Cooperative Extension, Institute of Agriculture and Natural Resources, University of Nebraska-Lincoln. Pingali PL, Hossain M, Pandey S, Price L. 1995. Economics of nutrient management in Asian rice systems: toward increasing knowledge intensity. Manila (Philippines): International Rice Research Institute. Pingali PL, Marques CB, Palis FG. 1992. Farmer health impact of long-term pesticide exposure: a medical and economic analysis for the Philippines. Paper presented at a workshop on Measuring the Health and Environmental Effects of Pesticides, 30 March-3 April 1992, Bellagio, Italy. Price LML. 1995. IPM “no early spray” farmer practice: farmer attitudes toward natural resource exploitation of rice field flora and fauna in the Philippines. Manila (Philippines): International Rice Research Institute. Rola A, Pingali PL. 1993. Pesticides, rice productivity, and farmers’ health. Manila (Philippines): International Rice Research Institute. Tandon HLS. 1989. Urea supergranules for increasing nitrogen efficiency in rice: an overview. In: Kumar V, Shrotriya GC, Kaore SV, editors. Soil fertility and fertilizer use. Vol. III. Urea supergranules for increasing nitrogen use efficiency. New Delhi (India): Indian Farmers Fertilizer Cooperative. p 10-22. Turner FT, Jund MF. 1994. Assessing the nitrogen requirements of rice crops with a chlorophyll meter method. Austr. J. Exp. Agric. 34:100l-1005. Widawsky D. 1996. Rice yields, production variability, and the war against pests: an empirical investigation of pesticides, host-plant resistance and varietal diversity in China. Ph.D. dissertation. Stanford University, Stanford, California, USA.

Notes Authors’ addresses: L.M.L. Price, Wageningen Agricultural University, Wageningen, The Netherlands; V. Balasubramanian, International Rice Research Institute, P.O. Box 933, 1099 Manila, Philippines. [email protected] [email protected] Citation: Dowling NG, Greenfield SM, Fischer KS, editors. 1998. Sustainability of rice in the global food system. Davis, Calif. (USA): Pacific Basin Study Center, and Manila (Philippines): International Rice Research Institute.

Securing the future of intensive rice systems

203

CHAPTER 13

Rice and the global environment R. Wassmann, T.B. Moya, and R.S. Lantin

The productivity and sustainability of natural resources ultimately depend on favorable climatic conditions that are currently being altered by human activities. The key process for changing the atmospheric environment is the combustion of fossil fuels, but agricultural activities are also associated with the release of trace gases that affect the radiation balance of the Earth. The ambivalent role of agriculture, one of the most important sectors affected by global change as well as one of the contributors to a changing environment, has prompted IRRI to investigate the interaction of rice cultivation and changing climate. Irrigated rice production at ambient growth temperature (25 °C) will benefit from increased atmospheric CO2 . Increased rates of CO2 assimilation and decreased rates of maintenance (dark) respiration at elevated CO 2 result in increased plant biomass accumulation. Grain yield also increases with rising atmospheric CO2 concentration. Concomitant temperature increases, however, could entail substantial losses in future yield because rice yields are extremely sensitive to temperature increases during the grain-filling stage, which can lead to abundant spikelet sterility. The coupling of crop models to future climate scenarios for the main riceg-rowing areas has given diverging results, from an 11% increase to a 12% decrease, depending on the model and scenario. The most significant contribution by rice fields to global change stems from the emission of the greenhouse gas methane. Methane formation in wetland rice fields is an important component of carbon cycling in the predominantly anaerobic soils. The quantity of methane emitted to the atmosphere is regulated by inherent soil and climate properties as well as agricultural practices. The shift from organic manure to mineral fertilizers substantially reduces methane emission. Likewise, the flux is reduced by intermittent drying of soils. New, high-yielding cultivars also reduce methane emission compared with traditional varieties. These findings help identify promising strategies to mitigate methane emission without yield losses, but they still have to be corroborated and improved by field experiments.

In recent years, public discussion on environmental issues has largely focused on the effects caused by enhanced concentrations of trace gases in the atmosphere. The recognition of a fundamental anthropogenic effect on the atmospheric composition and radiation balance of the Earth is gaining more and more acceptance in the scientific community. The detection of mechanisms leading to ozone destruction by chlorofluorocarbons (CFCs) in the stratosphere was recently acknowledged by the Nobel

205

Prize committee; the effect of greenhouse gases on the global climate was thoroughly reassessed and confirmed by an independent group of scientists, the Intergovernmental Panel on Climate Change (IPCC 1990, 1992). But climatic parameters such as temperature are characterized by pronounced spatial variability and genuine dynamics in different time scales. The relatively short time span of available observations impedes an ultimate proof of ongoing global warming, but indications to corroborate an anthropogenic impact on the global environment are compelling enough to urge against complacency. The carbon dioxide (CO2) level in the atmosphere has increased by approximately 32% from the preindustrial concentration of 270 ppm to a current concentration of 335-360 ppm (IPCC 1990). As the world population increases and the demand for energy rises, increased burning of fossil fuels will continue to drive levels of atmospheric CO2 upward. The IPCC “business-as-usual” scenario predicts that atmospheric CO2 concentrations will rise to 530 ppm by the year 2050 and could exceed 700 pprn by 2100 (IPCC 1990). This increase will significantly affect the physiological basis of plant production. The increasing ultraviolet-B (UV-B) exposure caused by ozone depletion in the stratosphere poses a further threat of unknown dimension to the productivity of agricultural systems. Since 1991, IRRI has been examining the impact of climate change on rice cultivation as well as the specific contribution of rice fields to the global budget of greenhouse gases (Neue et al 1995). These studies include different approaches at various levels including the physiological base of rice plants and the microbial community, element cycling in rice ecosystems, and regional and global trends in rice yields under a changing climate. Such interdisciplinary efforts are indispensable for sound decisions and technology development to cope with the food demand within the coming decades and beyond.

Agriculture in a changing global environment The increase in CO2 concentration is the key part of the greenhouse effect, accounting for approximately 50% of the projected increase in mean surface temperature (IPCC 1990). The imbalance of global sources and sinks in the atmospheric CO2 budget is caused primarily by combustion of fossil fuels. Net releases of CO2 by the agricultural sector are mainly related to land use changes, such as deforestation. Continuous cropping systems such as rice cultivation encompass high fluxes of CO2 , but input and output are balanced in sustainable production (Bronson et al 1998). Changes in soil organic carbon (C) (caused by intensified use of fertilizers), however, have a large potential to sequester C from the atmosphere (Cassmann et al 1995), although the significance of this CO2 sink is still unknown. The major contribution of rice fields to the greenhouse effect derives from the emission of methane (CH4), which is ultimately linked to the submergence of soils. Nitrous oxide (N 2O), another greenhouse gas, is emitted from virtually all cropping systems with high nitrogen (N) inputs, including intensive rice cultivation (Rennenberg et al 1992). In spite of an increasing number of emission records, estimates of global

206

Wassmann et al

emissions of greenhouse gases are still tentative. The broad range of these estimates prevents a definite assessment of rice cultivation’s part, and thus the amount that modified rice cultivation might curtail greenhouse gas emissions on a global scale. The need to increase rice production in the near future is imperative. Strategies to reduce greenhouse gas emissions improve the nutrient and C balance and therefore represent one component of advanced resource management in rice fields. In countries where rice cultivation predominates, rice research could play a crucial role in developing feasible mitigation strategies on a national level, a goal stipulated in the United Nations Framework Convention on Climate Change. But the largest share of historical and current greenhouse gas emissions has come from developed countries and from the energy sector. Concerted efforts for the widest possible cooperation are therefore essential to forestall changes in the global environment and possible effects on agricultural production.

Growth and yield response of rice to enhanced CO2 concentration The projected increase in CO2 concentration will significantly affect the physiological basis of plant production. Most plants grow under suboptimal levels of CO2 to achieve maximum photosynthetic capacity. But the beneficial effect of higher CO2 levels on plant growth may be outweighed by concomitant changes in other environmental factors (Rosenzweig and Parry 1994). Global increases in CO2, along with other trace gases such as CH4 and N2O, will trap outgoing thermal radiation and lead to higher temperatures at the Earth’s surface. Therefore, emphasis has to be given to the synergistic effects of CO2 and temperature on crop growth, weed competition, and water demand.

Photosynthesis and respiration Figure 1 shows the leaf photosynthetic rate of IR72, which was grown in flooded fields from germination to maturity under different temperature regimes and CO2 concentrations. The photosynthetic rates of plants exposed to elevated CO2 levels (ambient +200 and ambient +300 ppm of CO2) exceeded the rates of plants grown in ambient air by 35-60%,whereas the CO2 increases of 200 and 300 ppm did not show significant differences. Higher temperature resulted in higher photosynthetic activity until flowering. Plants use photoassimilates to build up structural biomass, but a portion of the assimilates is allocated to respiration. Respiration supplies the energy to maintain biochemical and physiological processes of growth and development. In respiration models, these functions are divided into two components—some respiration is associated with the maintenance of existing biomass and some with the synthesis of new tissue (Baker et al 1992, Kropff et al 1995). Ziska and Bunce (1993) found that maintenance respiration decreased at higher CO2 levels, but the reasons for this phenomenon are not clear. Plants growing at high CO2 may be constructing and maintaining

Rice and the global environment

207

Fig. 1. Leaf photosynthetic rates of IR72 grown in open top chambers at the IRRI farm in the 1995 dry season. T (amb) = ambient temperature, T +4 °C = ambient temperature + 4 °C, C (amb) = ambient CO 2 concentration, C (+200) = ambient CO2 concentration +200 ppm, C (+300) = ambient CO 2 concentration +300 ppm.

less energetically expensive biomass and thus use less CO2 (Bunce 1994). As long as suppression of respiration does not reduce the supply of energy for vital plant functions, it may increase net photosynthesis (Imai 1995). But a CO2-induced modification in respiration rates may result in a lack of energy to repair strained tissues. Also, stomatal aperture may decrease partially because of an inability to maintain the ionic gradients responsible for the opening mechanism of cells (Bunce 1994).

Biomass accumulation and yield Aboveground biomass showed distinct patterns for ambient and increased temperatures (Fig. 2). Under the ambient temperature regime at the IRRI farm, the elevated

208

Wassmann et al

Fig. 2. Increment in aboveground biomass (g m-2 d -1) of IR72 grown in open top chambers at the IRRI farm in the 1995 dry season. T (amb) = ambient temperature, T+4 ºC = ambient temperature +4 ºC, C (amb) = ambient CO 2 concentration, C (+200) = ambient CO 2 concentration +200 ppm, C (+300) = ambient CO 2 concentration +300 ppm.

CO2 levels resulted in a distinct boost during the grain-filling stage. This boost was not observed at higher temperatures when the increments remained in a relatively stable range throughout the reproductive and ripening stages. Table 1 summarizes the agronomic characteristics of the mature plants. Harvested biomass (aboveground plus roots) increased with higher CO2 concentrations, an effect observed for both tempera-

Rice and the global environment

209

Table 1. Plant growth properties of IR72 grown in open top chambers at the IRRI farm in the 1995 dry season.

T (amb)a Variable

Green leaf area Leaf weight Stem weight Root weight Root-shoot ratio Panicle weight 1,000-grain weight Filled spikelets Harvest index

T (+4 °C)

Units

(cm2 hill -1 ) (g m -2 ) (g m -2 ) (g m -2 ) (g m -2 ) (g) (%)

C (amb)

C (+200)

C (+300) C (amb)

C (+200)

C (+300)

1,185 b 285.5 b 471.0 b 390.0 b 0.19 b 726 c 24.8 a 84.7 a 0.47 a

1,046 b 287.3 b 589.2 ab 501.5 b 0.23 a 1,076 a 24.6 a 84.8 a 0.46 a

1,103 b 289.3 b 635.3 a 516.5 a 0.23 a 1,099 a 24.9 a 85.1 a 0.47 a

1,561 a 289.3 b 643.6 a 423.0 b 0.19 b 854 b 23.4 ab 80.5 ab 0.39 b

1,394 ab 371.2 a 683.6 a 489.1 a 0.20 b 930 ab 23.4 ab 77.2 b 0.38 b

1,349 ab 301.4 b 469.4 b 270.0 c 0.16 c 671 c 22.3 b 82.4 a 0.39 b

a T (amb) = ambient temperature, T (+4 °C) = ambient temperature plus 4 °C. C (amb) = ambient CO concentra2 tion, C (+200) = ambient CO 2 concentration +200 ppm, C (+300) = ambient CO 2 concentration +300 ppm. In a row, means followed by the same letter are not significantly different at the 5% level by Duncan's multiple range test.

ture regimes. The impact of the rise in temperature depended on the CO 2 level— lower biomass under ambient temperatures and higher biomass under a higher CO 2 concentration. Rice yield increased with increasing atmospheric CO2 at a given temperature regime. The average rice yields at the intermediate and high CO2 were 1 t ha-1 higher than for rice grown at ambient CO2 (Fig. 3). The yield increment accrued from increased weight per panicle at an increasing CO2 concentration, whereas the weight of individual grains was fairly stable over the CO 2 treatments (Table 1). In sum, irrigated rice production at ambient growth temperature (25 °C) will benefit from increased atmospheric CO2 as such. The increased rates of CO2 assimilation and decreased rates of maintenance (dark) respiration at elevated CO2 resulted in increased plant biomass accumulation. Grain yield also increased with increasing atmospheric CO2 concentration.

Environmental limitation and management options for exploiting CO2 effects The actual impact of enhanced CO2 levels on agriculture will depend on the availability of water and nutrients as well as climatic factors. The ultimate linkage between atmospheric CO2 and temperature results in a number of uncertainties about the overall benefit from CO2 “fertilization.” Several synergistic pathways of CO2 and temperature were shown above, but temperature also affects plant development independently, for example, through shorter vegetation periods and spikelet sterility. A temperature increase of 4 °C accelerated plant development until maturity by 4 d in our experiment. Shorter cropping periods may allow a shift in planting dates and the introduction of long-maturing varieties at some locations. The potential benefit of

210

Wassmann et al

Fig. 3. Grain yield at maturity (t ha -1) of IR72 grown in open top chambers at the IRRI farm in the 1995 dry season. T (amb) = ambient temperature, T +4 °C = ambient temperature +4 °C, C (amb) = ambient CO 2 concentration, C (+200) = ambient CO 2 concentration +200 ppm, C (+300) = ambient CO 2 concentration +300 ppm.

such modifications, however, may be limited by the sensitivity of rice plants to the actual temperature regime at specific plant stages. High temperatures during flowering may result in abundant spikelet sterility, which was shown to be a decisive mechanism in determining rice yields in a future climate (Matthews et al 1995). A small change in the mean temperature or even an altered temperature pattern could cause pronounced effects on production because of the extreme sensitivity of spikelet fertility to temperature at a very distinct and short period of time. Crop models coupled to global climate-change scenarios yield different results, depending on the model and the scenario used. Simple crop models usually indicate higher rice production. As an example, Leemans and Solomon (1993) predicted an 11% increase. The more sophisticated IBSNAT model showed a 2–4% reduction in global rice production (Rosenzweig and Parry 1994). These losses are mainly attributed to low latitudes; crop yields in mid- and high latitudes are predicted to increase. Matthews et al (1995) coupled the ORYZA and SIMRIW models to different climate-

Rice and the global environment

211

change scenarios and obtained an overall impact on rice production in Asia ranging from +6.5% to -12.6%. The average of these computations suggests that rice production in Asia may decline by 3.8% (Matthews et al 1995). But the detrimental effects of an increase in temperature may be ameliorated by varietal adaptation. The level of adaptation required (e.g., for spikelet fertility) is within the genotypic variation currently present in environments with hot climates (Matthews et al 1995). Rice cultivars exhibit a range of adaptation to changing global CO2 and temperature. Although some varieties may be unable to cope with changing CO2 and temperature, others may be able to optimize them for increased vegetative and reproductive growth. Of the 22 species (other than sativa) of the genus Oryza, commonly called the wild relatives of rice, several possess photosynthetic characteristics equal to and, in some respects, superior to modern cultivars. At increased growth temperature, some varieties may be insensitive to high CO2 levels; others may experience reduced growth and yield. The mechanisms of varieties that exploited high CO2 and temperature for vegetative and reproductive growth must be further investigated. Enhanced levels of atmospheric CO2 can influence the competitive ability of rice against weeds—namely, those with a C4 metabolism. Plant species that follow the C3 photosynthetic pathway produce a primary compound consisting of 3 carbon atoms, whereas others produce mainly a compound consisting of 4 carbon atoms (C 4 pathway). C4 plant species have CO2 -concentrating mechanisms that enhance photosynthetic potential at ambient CO 2 concentrations. This supplementary mechanism results in a lower response to increasing CO2 levels in the atmosphere compared with C3 plants. Rice, a C3 plant, may sharpen its competitive edge against a C4 weed in the future when atmospheric CO2 rises. IR72 produced more biomass than Echinochloa crus-galli— a C4 weed species—at increased CO2 levels in the glasshouse (data not shown). Photosynthetic rates of IR72 grown at two CO2 levels and three N levels decreased at high CO2 when N was not applied (data not shown). In the future, rice growth and yield responses to increasing CO2 levels may depend on available N. Aboveground biomass and yield increased with increasing CO2 levels even when phosphorus (P) was not applied, but growth and yield benefits will increase further with increasing CO2 when P is applied (Seneweera et al 1994). Overall, rice ecosystems will absorb more CO2 for the production of biomass, which will also involve an increased turnover of soil organic C. The bulk of the biomass as well as the easily degradable soil organic matter will be released in the form of CO2 after harvest, but a long-term sequestration of C may occur in the enhanced formation of relatively inert organic compounds in the soil. Intensified rice production increased the amount of inert organic material in the soil (Cassmann et al 1995); this process could become more significant in the future with a further increase in productivity by “CO2 fertilization.” The quantification of this C sink, which could act as a negative feedback mechanism to an increase in CO2 , is a crucial question in the overall assessment of rice production and global change.

212

Wassmann et al

Possible effects of increased UV-B radiation One class of atmospheric trace gases, the CFCs, has a twofold effect on the environment. This group of gases contributes—with minor importance—to global warming, but the real threat derives from the catalytic degradation of ozone in the stratosphere (Cicerone 1987). The stratospheric ozone layer filters out much of the short-wave component of the solar spectrum before it penetrates the Earth’s surface. Increasing radiation will have severe effects on human health and on terrestrial and marine ecosystems. But the ban implemented on CFCs in industrialized countries is expected to reverse the declining trend of stratospheric ozone concentrations within the coming decades. Ultraviolet radiation with wavelengths from 280 to 320 nm (termed UV-B) is readily absorbed by biochemical molecules, such as proteins and nucleic acids, resulting in a destruction of chemical bonds (Tevini and Teramura 1989). The natural UV-B radiation in the tropics is considerably stronger than that in the higher latitudes because the ozone layer is thinner in the tropics and the solar angles are higher. The depletion in the ozone layer in these regions with significant rice cultivation corresponds to 1.6-3.1% (NASA 1988). On the other hand, a large portion of the UV-B radiation in the tropics and subtropics is captured by clouds, especially during the monsoon season. Rice plants are relatively resistant to enhanced UV-B radiation. In field studies, the modern varieties disseminated by IRRI did not show a significant change in growth and yield as a result of enhanced UV-B radiation (Dai et al 1995). A screening of 188 cultivars originating from various locations confirmed that rice plants cope with relatively high UV-B radiation (Dai et al 1994). This resilience appears to be related to the effective mechanisms of DNA repair that are stimulated by other components of the solar spectrum, such as UV-A radiation. These findings led to the conclusion that increased UV-B exposure will not cause significant yield losses in global rice production. The impact of enhanced UV-B in some areas (such as outside the humid tropics) should be considered separately. Furthermore, increased UV-B radiation could have indirect effects on plant competitive interactions, biodiversity of rice cultivation, and pest-pathogen relationships in rice systems.

Greenhouse gas emissions from rice fields Processes involved in methane emissions Methane is generated in the anaerobic layers of rice soils (Fig. 4). The organic material converted to CH4 is derived mainly from soil organic matter, plant-borne material, and—if applied—organic manure (Neue 1993). Methane is produced in the last step of different biochemical pathways. The decomposition rate of the organic material determines the availability of immediate precursors of methane and, thus, the in situ rates of CH4 production. Methane production requires a redox potential of less than -200 mV, which is commonly found in rice soils 2 wk after flooding (NeUE 1993). But the upper micro layer ( < - - - - data- - - - > L1D L1Q TIL L2C L2SU L2SS crop soil weather T12 3 4 2 7 8 9 5 10 11 App B

x x x

x

x

x

x

x

x

x

x x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

suggest absolute accuracy. Do not restrict yourself to obtaining the same answers, but apply the models for your own crop, soil and weather data. Adapt, shorten, or extend programs when desired. Readers are encouraged to examine the varying aspects of model behaviour to explain the results. Simulation exercises are particularly effective when undertaken by small groups of students or during courses. Running models will help greatly in obtaining a better grasp of simulation and of the entire model. It is only by running, adapting and improving crop models for stated objectives that the knowledge and insights gained can become effective in agricultural research, extension or planning. Glossary CSMP The simulation language Continuous System Modelling Program (IBM, 1975). Function A user defined mathematical, physical or biological relation between one or more inputs and one output (FORTRAN). Listing A printed version of a module or a program. MACROS Modules of an Annual CROp Simulator. Mimick To reproduce the behavior of a small system or subsystem with xiv

equations that are not based on the processes involved. Module A set of statements in a computer language that together describe a system or a large part of a system; a set of data characterizing a crop, soil or weather; a set of functions and subroutines. Model A simplified representation of a system. A submodel is a model of a subsystem. Simulation model A module that represents the relevant processes of a system, usually in the form of a computer program. Program A complete set of modules and individual statements, and data sets of a particular module and of its driving variables (also called: simulation program). SAHEL A model for simulating the water balance of free-draining soils with a deep water table. SAWAH A model for simulating the water balance of soils with impeded drainage, often with a high water table and partially saturated. Simulate To create a model to study a system and to use a program of the model to reproduce the behavior of the system. Subroutine A user defined mathematical, physical or biological relation between one or more inputs and one output (FORTRAN). System A part of the real world consisting of parts that interact and change. The environment of the system exerts influence on it, but the system does not affect its environment. When a system is part of a larger system, it is referred to as a subsystem.

xv

1

Introduction to crop growth modelling

Chapter 1 attempts to answer the question ‘Why create a crop growth model?’ and to identify methods of determining what factors should be included. Section 1.1 introduces systems analysis and the dynamic simulation of living systems in their physical environment. Section 1.2 narrows the focus to systems analysis and simulation of annual field crops. Section 1.3 deals with the possibilities for using crop growth models in research and education, and their practical application. Some technical aspects of simulation techniques are discussed in Section 1.4. 1.1

Modelling crop growth

Growing a crop is complex. Some activities, such as planting or seeding, are always needed; others, such as irrigation, fertilization and spraying fungicides are optional. A farmer combines activities effectively because he has a concept or model of how the crop will react to its environment and to husbandry practices. In this sense farmers use multidisciplinary models. However, these mental models are somewhat crude and are difficult to improve or to explain to others.

1.1.1

Descriptive and explanatory models

A crop model is a simple representation of a crop. It is used to study crop growth and to compute growth responses to the environment. Crop models in common use can be distinguished as descriptive and explanatory models.

Descriptive models A descriptive model defines the behaviour of a system in a simple manner. The model reflects little or none of the mechanisms that are the cause of the behaviour. Creating and using this type of model is relatively straightforward (Figure 1). Descriptive models often consist of one or more mathematical

Figure 1. A scheme to indicate how real world observations are brought into a descriptive model.

1

equations. An example of such an equation is derived from successively measured weights of a crop (Figure 2). This equation is helpful to determine quickly the weight of the crop when no observation was made. However, the growth rate of the crop will not be the same when soil, crop husbandry practices or weather are different. Large deviations can result from differences in weather patterns between years (Figure 3). Adapting the starting point of the regression equation and of the maximum value are possible in hindsight, but predicting these parameters for other fields and in other years is usually too inexact for specific production studies. In theory, it is possible to derive the required constants and equations from many experiments with acceptable accuracy. In practice, however, many variables influence growth patterns. Some, such as soil texture, are constant; others, such as the properties of new cultivars and crop husbandry practices, constantly evolve. Thus, it is impossible to quantify adequately all variables through extensive field experiments. Descriptive models are therefore of value only for situations where interpolation between observations is sought and there is no attempt to quantify the background of the shape of the biomass curve.

Figure 2. The course of the dry weight of a maize crop in the Netherlands in 1972. Crosses represent observations, the line the regression equation BM = 12.0 / (1.0 + 23.0 · e -0.08 · T ), where BM is the biomass in t ha-1 , 12.0 is the maximum value of BM, T is the time in days since emergence and 1.0, 23.0 and 0.08 are constants.

2

Figure 3. The dry weight of maize crops under optimal conditions in different years in the Netherlands. (Source: Sibma, 1987).

Explanatory models An explanatory model consists of a quantitative description of the mechanisms and processes that cause the behaviour of a system. These descriptions are explicit statements of the scientific theory and hypotheses. To create an explanatory model, the system is analyzed and its processes and mechanisms are quantified separately. The model is built by integrating these descriptions for the entire system. An explanatory crop growth model contains descriptions of distinct processes such as photosynthesis, leaf area expansion and tiller induction. Crop growth is a consequence of these underlying processes (Figure 4). Each process must be quantified in relation to environmental factors, such as radiation and temperature; and in relation to the crop status, including leaf area, development stage and nitrogen content. Growth rates can then be computed for any stage of the growing season, depending on the actual crop status, the soil and current weather. All important factors can be accounted for in this way, provided there is sufficient theory and data to quantify them. Models considered in this book are mostly of the explanatory type. The ‘behaviour’ of the model, i.e., the growth rate for any stage, can be explained by the basic physiological, physical and chemical processes and by the effects of environmental factors on them. De Wit (1970) noted that the hierarchical levels of explanatory processes and of explained behaviour are characterized by time coefficients of different orders of magnitude, and that they are usually the subject of study in different scientific disciplines. The ex3

Figure 4. A scheme to indicate how real world observations are analyzed and integrated into an explanatory model to simulate behaviour of the system.

planatory approach to modelling goes deeper by at least one hierarchical level, and sometimes two or more, than the descriptive approach.

1.1.2

Simulation with explanatory models

Simulation models are relatively simple representations of systems in the world around us. A system is defined here as any well delineated part of the real world. The user identifies a system on the basis of objectives and on the intrinsic structure of the world as measured and observed. For an agronomist, a system may be a rice crop; its elements, plant organs (such as leaf, stem and root) and processes (such as growth and transpiration) interact strongly. Weather is a driving variable because it exerts an important driving, or regulating effect on the crop. The crop, on the other hand, has virtually no impact on the weather. In general, driving variables influence the system and its behaviour, but the reverse is not true. Behaviour is the sum of all processes in a system, for example, the growth of a rice crop during a season. A system is ‘dynamic’ when its states change over time. This may be ‘continuous’, as when its behaviour and states change relatively slowly, or ‘discrete’ when changes occur fast or are large (e.g. a tractor changing from the state used to not-used). A model is dynamic when it simulates the behaviour of a system. State variables in models represent quantities, which may be tangible (such as weight), or abstract (such as development stage). Rate variables represent rates of change of state variables. The photosynthesis rate is an example of a rate variable. In this context, ‘simulation’ is the study of a system and the computation 4

of its behaviour using a dynamic model. In explanatory simulation models of dynamic systems, such as those of crops, it is assumed that the rate of change can be closely approximated by considering the rates of processes to be constant during short time periods. This is the state variable approach. In crop simulation time periods must be short compared to the duration of the growing season; often one-day periods are chosen. The biomass formed in such a short time period equals the multiple of growth rate and time period. This is added to the quantity of biomass already present. The growth rate is then recalculated. The new rate is slightly different because environmental conditions or the internal status of the plant will have changed a little. Calculations of rate variables and updating state variables are repeated in sequence until the entire growing season has been covered. Van Keulen & Wolf (1986) give a clear example of this procedure for a growing crop. Fortunately the number of processes of prime importance for simulating crop growth is limited and detailed calculations to quantify these processes are unnecessary. For example, calculating the efficiency of synthesis of each biochemical compound in a biomass is usually unnecessary; averages for classes of compounds are sufficient. It is also unnecessary, and even counter-productive, to include dynamic aspects of cell physiology in crop growth models. Explanatory models can be of practical use even though knowledge of the processes often does not reach the cellular level. Indeed, it is of limited importance whether processes at the explanatory level (the general level of physiology and soil physics in this book) are descriptive or quantified in more detail (e.g. at the biochemical level), as long as their quantification is valid within the range of conditions for which the model is used. The more detail desired in the results of a model, the more detail the model itself must contain, and hence the more explanatory processes should be included.

1.1.3

Development of explanatory models

Large and complex explanatory models have been developed over the last decade in many places, including Wageningen in the Netherlands. Development has been relatively slow, because among other things, some essential topics were insufficiently understood. These large comprehensive models contain a wealth of information, but are unwieldy. Their use is limited to reassessing hypotheses, for sensitivity analyses and for reference and comparison with other models. These models are seldom used except by the scientists who created them. This not only limits their usefulness, but also undermines the credibility of models and model builders. In recent years, summary models have been derived from several comprehensive models. These models retain much of the scientific basis and quality of the comprehensive models, but are simpler and much easier to use. Explanatory models are of three types: preliminary, comprehensive and 5

summary. Explanatory models develop gradually from one type, or phase, to the next. Preliminary models have a simple structure because insights at the explanatory level are still vague. Comprehensive models represent a system in which essential elements are thoroughly understood and incorporate much of this knowledge. Summary models are abstracts of comprehensive models. Summary and comprehensive models are currently used where weather or soil water limit crop production. Models vary considerably in their value for scientific research, education and applications (Table 2). Examples are given in Chapter 1.3. The content of models considered in Chapters 2-5 is mostly at the summary phase, though parts are still scarcely beyond the preliminary phase. For more extensive introductory reading see Brockington (1979), Dent & Blackie (1979), Loomis et al. (1979), Penning de Vries & van Laar (1982), Penning de Vries (1983), van Keulen & Wolf (1986), or Rabbinge et al. (1989). 1.2

1.2.1

Crop Production Levels and processes

Levels of crop production

De Wit proposed a classification of systems of crop production based on growth-limiting factors (de Wit & Penning de Vries, 1982; Penning de Vries & van Laar, 1982) and distinguishes four levels of plant production. The crop production systems at any of these levels can be considered as members of a broad class of systems. In order of decreasing yield, these levels are:

Production Level 1 The crop has ample water and nutrients and produces a higher yield than at any other Production Level. Its growth rate depends only on the current state of the crop and on current weather, particularly radiation and temperature.

Table 2. The relative values of important aspects of models in different phases of development.

Preliminary model Comprehensive model Summary model

6

Scientific value

Instructive value

Applicability value

+++ +++ +

++ + +++

+ ++ +++

With a full canopy, the growth rate of field crops is typically between 150 and 350 kg ha-1 d -l of dry matter. This is the ‘potential growth rate’ and the crop yield ‘the potential yield’. These growth conditions are realized on very intensive arable and grassland farms in Western Europe and often in glasshouses.

Production Level 2 The growth rate is limited only by the availability of water for at least part of the growing season. This situation seldom occurs spontaneously, but in semiarid regions applying fertilizers can result in crop growth at this Production Level. This may also occur in other climates under intensive cropping on light soils. Production Level 3 The growth rate of the crop is restricted by nitrogen shortage for at least part of the growing season and by water shortage or poor weather for the remainder. This situation occurs frequently in agricultural systems all over the world. Nitrogen shortage occurs particularly in crops when fertilizer is not intensively applied. In the natural environment, even nitrogen-efficient plants cannot always absorb sufficient nitrogen. Production Level 4 Crop growth is restricted by low phosphorus and other mineral nutrients in the soil for at least part of the growing season. The growth rates are 10-50 kg ha -1 d-1 and the growing season often lasts less than 100 days. This situation usually occurs in heavily exploited areas where no fertilizers are used. Rarely do cases fit exactly into one of these Production Levels, but it is practical to reduce specific cases to one of these four categories. This focuses attention on the dynamics of the main environmental factor and on the response of the crop to it. Environmental factors that have no regulatory effect can then be disregarded, because they do not determine the growth rate. The growth rate then sets the absorption rate or efficiency of use of non-limiting factors. If, for example, plant growth is limited by nitrogen, there is little use in studying CO2 assimilation or transpiration to understand the current growth rate. All emphasis should be placed on nitrogen availability, the nitrogen balance and the plants response to nitrogen. This analysis of plant production systems allows for considerable narrowing of the subject of study and permits more rapid research progress. Growthreducing factors, such as diseases, insect pests and weeds, can occur at each of these Production Levels and give them, in a sense, an extra dimension. The fact that actual situations are often more complex does not contradict the general usefulness of this scheme of Production Levels as a basis for distinction between causes and consequences of plant growth. Note that this use of Production Levels has a crop physiological basis and is 7

not related to descriptions of production systems based on agronomic practice or crop ecology, such as the irrigated, rainfed lowland, deep water and upland production systems in rice growing (IRRI, 1984).

1.2.2

Principal processes of the Production Levels

This systematic analysis of crop production can be taken a step further to formulate simple systems and models at the four levels of plant production. At Production Level 1 the intensity of radiation, the interception of light and the efficiency of energy use in the plant are key factors for understanding the growth rate. Figure 5, a relational diagram, indicates the essence of models at Production Level 1. Light is a driving variable. Assimilated carbohydrates are stored, usually briefly, in an easily accessible form, such as starch (‘reserves’), and are later used for maintenance or growth. Temperature is an external variable that can modify growth rates and photosynthesis. In growth processes, reserves are converted into ‘structural biomass’ with a specific efficiency. Structural biomass consists of those components that are not mobilized again for maintenance or growth processes elsewhere in the plant. The partitioning of biomass between roots, leaves, stems and storage organs is strongly related to the physiological age of the crop, which itself is a function of temperature.

Figure 5. A relational diagram of a system at Production Level 1. Light and temperature are driving variables; the photosynthetic efficiency is a constant. Rectangles represent quantities (state variables); valve symbols, flows (rate variables); circles, auxiliary variables; underlined variables, driving and other external variables; full lines, flows of material; dashed lines, information flow (symbols according to Forrester, 1961).

8

At Production Level 2, key factors are the degree of exploitation of soil water and the efficiency of its use by the crop (Figure 6). Water shortage leads to stomatal closure and to a simultaneous reduction of CO2 assimilation and transpiration. Water use efficiency is the ratio of photosynthesis and transpiration rates. The ratio of the actual transpiration rate and the potential rate provides the link between the carbon and water balance. The extent to which the potential transpiration, and consequently the potential photosynthesis rate, is realized, depends on the availability of water. The amount of water stored in the soil is a buffer between rainfall and capillary rise and the processes by which water is lost. This buffering capacity and the simultaneous water loss through transpiration and non-productive processes cause the growth rate to depend only indirectly on rainfall. The relation of plant growth to the principal driving variable of this system is indirect (rather than direct, as at Production Level 1). At Production Level 3, nitrogen in plant tissues is distinguished by two fractions: mobilizable and immobilizable nitrogen (Figure 7). The amount of nitrogen that can be mobilized for growth of new organs is often considerable.

Figure 6. A relational diagram of a system at Production Level 2. Water shortage is the main limiting factor. Rectangles represent quantities (state variables); valve symbols, flows (rate variables); circles, auxiliary variables; underlined variables, driving and other external variables; full lines, flows of material; dashed lines, information flow (symbols according to Forrester, 1961).

9

Figure 7. A relational diagram of a system at Production Level 3. Nitrogen shortage is the main limiting factor. Rectangles represent quantities (state variables); valve symbols, flows (rate variables); underlined variables, driving and other external variables; full lines, flows of material; dashed lines, information flow (symbols according to Forrester, 1961).

The concentration of nitrogen in mature tissue may reduce to half or a quarter of its maximum value before the tissue stops functioning. Growth is directly related to the rate of nitrogen absorption only after the internal nitrogen reserve is used. This internal reserve of nitrogen makes the increase in plant dry matter at any moment largely independent of the current absorption of nitrogen. The relation of nitrogen uptake and growth is, therefore, quite different from that of water uptake and growth. The mobilizable fraction consists of enzymes and membrane proteins that are broken down and exported as amino acids; not all can be considered reserves, because cells cannot function without them. The immobilizable fraction of nitrogen in the tissues is tied up in stable proteins. The growth rate at this Production Level is primarily determined by the availability of nitrogen from the soil and the internal reserve. Hence the rate of CO2 assimilation is a consequence of the growth rate. The availability of nitrogen from the soil resembles that of water; a variable amount of inorganic nitrogen is present in the soil and most of it is readily available to roots that are sufficiently close. Soil microflora may compete with plants for this nitrogen and other processes may also interfere. Nitrogen in organic matter in the soil is not available to crops. But mineralization, ie., breakdown of organic matter by microbes, releases nitrogen to the inorganic pool. Crucial processes of crop growth at Production Level 4 are similar to those at 10

Production Level 3 (Figure 8). The concentration of phosphorus in ageing tissue decreases in the same way as nitrogen; and, as with nitrogen, plants also have an internal reserve of phosphorus. But the processes that make phosphorus available to roots differ considerably from those for nitrogen. Plants require a much higher root density for adequately exploring the soil for phosphorus; and the quantity of dissolved phosphorus in the soil is so small that the rate of its replenishment determines the phosphorus supply to roots. Mycorrhiza may enhance phosphorus uptake by increasing the explored volume of soil. Both organic and inorganic compounds in the soil may provide and capture dissolved phosphorus. Chapters 2,3,4 and 5 consider models for situations with ample nutrients for crop growth. For models of crops in situations with severe shortages of nitrogen and phosphorus, see van Keulen (1982), Hansen & Aslyng (1984) and van Keulen & Wolf (1986). 1.3

1.3.1

Uses of crop growth models

Determining when to use a simulation model

De Wit et al. (1978) stated: ‘In our opinion, simulation models, if they are useful at all, should form a bridge between reductionists, who analyze proc-

Figure 8. A relational diagram of a system at Production Level 4. Phosphorus shortage is the main limiting factor. Rectangles represent quantities (state variables); valve symbols, flows (rate variables); underlined variables, driving and other external variables; full lines, flows of material; dashed lines, information flow (symbols according to Forrester, 1961).

11

esses separated from their physical, chemical or biological background, and generalists who are interested in the performance of whole systems in which the individual processes operate in their natural context. Both the reductionist and the generalist should recognize their work in the simulation program. By comparing detailed output, the generalist can independently evaluate how the model operates with field data, and the reductionist can determine whether the treatment of processes that form the basis of the simulation model correspond with his ideas. To the reductionist, simulation can be a guide to areas where research is most promising for further understanding of the system studied. To the generalist, simulation extends his capability to envisage how a whole system functions.’ Scientists who are generalists follow a bottom-up approach to modelling. By using a mathematical model, they can describe their subject more clearly and study the implications more easily. As the generalist’s model is improved and elaborated, it gives a broader view and covers more topics. As yet, understanding the real world of agriculture and biology is still far from complete. Further improvement of crop growth models is necessary. Research administrators, policy makers and industrial leaders may use a top-down approach when seeking the solution to a problem or surveying possibilities. They first identify the problem and then determine specifications to which the answer must comply. One or more techniques from mathematics or information sciences, such as statistics, linear programming, simulation, data base management and expert systems, can be used to provide these answers. Statistics can create clarity about the relations between variables that fluctuate. Linear programming can help determine the optimal combination of many inputs and factors to achieve a certain goal, such as optimal land use (see Kingwell & Pannell, 1987). Simulation techniques can cope with complex relations between state and rate variables in a system and enable computation of a systems behaviour in specific circumstances and new environments. Data base management techniques allow storage of masses of data plus the relations between them; each item remains traceable individually or in groups, making information quickly, efficiently and completely available to many people. Expert systems, an emerging tool, can help select the best choice from many possibilities in a question-and-answer dialogue, the computer playing the role of the expert. All techniques require the availability of sufficient basic data and knowledge. For simulation, the relations between all principal variables of the system and the values of key constants must be known. This should not be overlooked. If these relations and values are not known, it may be better not to choose simulation for the practical problem; it may be better to devote a major research effort to accumulate the necessary knowledge. Finding basic data to use in this book was not easy. The amount in open literature is limited and an inventory for a range of crops was difficult to make. This was true even for parameters that are quite commonly used in science and 12

for which good examples are documented, such as the maximum rate of leaf photosynthesis. This is symptomatic of the fact that we are still in the early stages of applying crop growth models. Potential users of crop growth models frequently ask scientists to produce or adapt a model for a specific crop and a specific problem. Model developers look more towards potential users, not least, because they may sponsor their work. A brief discussion of where modelling stands in terms of usefulness to non-modellers is presented in Subsection 1.3.4. Models can also be appreciated for their value for research and for instruction, see Subsections 1.3.2 and 1.3.3. It is worthwhile to recall that the usefulness of a model changes considerably as it evolves (see Table 2 Subsection 1.1.3). Any tool can be adequate or inappropriate, depending on the goal for which it is applied. This holds for models just as much as for mechanical tools. A particular crop growth model can be very suitable for achieving a specific goal, but totally inappropriate for another. The user and the developer must carefully define objectives before using, adaptating, or developing a model. Though this may appear obvious, many modelling efforts have suffered from underspecified objectives (IIASA, 1980). Defining objectives can also help prevent excessive optimism about results and unconsciously pushing objectives higher and higher. It also helps resist the temptation to let the model derail and become an encyclopaedia of science. Models are not yet at a stage (and may never reach the stage) when they can be used or applied without understanding how they work. As yet few persons are trained to work with crop growth simulation models. 1.3.2

Using models to guide research

For the past 15 years, simulation models have been used, among other things, to determine how far crop growth in different situations can be explained from documented theory and data. Objectives for this modelling have been broadly, but not explicitly defined (Ng & Loomis, 1984). In the process of developing models, topics were identified where crucial insights were missing, and hence research was needed. McKinion (1980) documented this research progress of a group in southeastern USA. An example from Wageningen describes briefly how modelling helped guide research. In the sixties, attempts to compute the photosynthesis rate of crop canopies yielded several explanatory models, e.g. the model by de Wit (1965). These were static models; time was not included as a variable. Their results were used, among others, to estimate potential food production for certain areas of the world and hence to provide perspectives and goals for crop husbandry and breeding (de Wit, 1967; Linneman et al., 1979). These efforts stimulated quantitative research on leaf photosynthesis (e.g., Louwerse & van Oorschot, 1969). Next, a preliminary dynamic simulation model was constructed. This in13

cluded an abstract of the static photosynthesis model. Respiration was first taken as a fixed fraction of photosynthesis, and later as a fixed fraction per day of the biomass, plus an amount proportional to the growth rate. Adding a functional equilibrium between root and shoot growth (by which crops attempt to maintain an optimal water content) made this model an ELementary CROp growth Simulator (ELCROS) (de Wit et al., 1970). However, respiration was quantified unsatisfactorily and in subsequent research it was determined that respiration due to growth is directly related to the chemical composition of the new biomass (Penning de Vries et al., 1974). Respiration caused by maintenance could only be partially quantified and is a weak part of crop growth models even today. Micrometeorology was introduced in the models (Goudriaan, 1977) to simulate, among others, the effect of canopy resistance on heat and gas exchange. This refined and improved the simulation of transpiration. Next, the BAsic CROp growth Simulator (BACROS) was evolved (de Wit et al., 1978). BACROS has been evaluated with field crop experiments with two-week periodic harvests. But as this comprehensive model simulates growth processes with time intervals of hours, it also needed evaluation on a shorter time scale. A mobile laboratory was created (Alberda et al., 1977) to measure assimilation and transpiration continuously in the field. This led to another round of checking, correcting and improving BACROS. The major conclusions from research about the discrepancies between observed and measured gas exchange were that stomatal resistance is often controlled by the photosynthesis rate (Louwerse, 1980), and that the contribution of diffuse light to canopy photosynthesis had been underestimated (Lantinga, 1985). Wageningen scientists now use BACROS as a reference model and yardstick for developing other models and as a basis for developing summary models, such as SUCROS (a Simple and Universal CROp growth Simulator) (van Keulen et al., 1982). Current research is aimed at further improving BACROS and SUCROS, among others, by research on plant morphology and maintenance respiration and by combining them with models from other scientific fields.

1.3.3

Using models in education

Simulation models can be used for hands-on learning about the behaviour of a system in different situations when the real system is too large, too slow, or too expensive for teaching purposes. Examples are weather models, models of the annual cycle of farm activities and airplane flight simulators. There is no difference to the trainee between descriptive and explanatory models as long as they reproduce reality sufficiently well. Crop growth simulation models have, as yet, seldom been used in this way. Huke’s interactive model (1985) of a rice farm in Bangladesh demonstrates several options and hazards involved in growing crops, but is not meant to be a realistic simulation of the real world. 14

Studying quantitative crop physiology with a model is also not yet common. Educative models should be thoroughly evaluated, and must be lucid, well documented and physically available to students; few models score high on all criteria. Summary type models are best for this purpose and some have recently been published. Learning with crop models has been part of the International Post Graduate Courses in Wageningen for some years (Penning de Vries et al., 1988) and it has been incorporated in the training program of the Wageningen Agricultural University. Some books in the series of Simulation Monographs teach crop physiology in a whole crop context. To allow more widespread use of models by non-specialists, simplified access is generally needed. This may be achieved either by simplifying the model itself to only a few lines, or by adding a program before the model. The latter program, called an ‘interface’, can ask a novice-user for the most crucial inputs and suggest defaults for others. Most model simplifications, so far, are of the first group. Technical limitations in simulation techniques have made this unavoidable. However, with the advent of powerful but cheap computers, this is no longer a strong argument. Providing interfaces between the end-user and well-evaluated models may be a better development. 1.3.4

Application of crop growth models

Application refers here to using a model outside the scientific discipline in which it was constructed. In the near future, opportunities for applying simulation models for crop production lie particularly in the domains of potential crop production and crop production with temporary water shortage but ample nutrients (i.e., Production Levels 1 and 2). These domains include predicting short term yield, extrapolating and interpolating crop performance over large regions and simplifying and combining with other models to create links with other sciences. Applying models can lead to more effective use of existing knowledge for extension, agronomic and cropping systems research and breeding, to more efficient experimentation and for further integrating the scientific disciplines involved in crop production. Some examples are given in this book; others, concerning rice, are given elsewhere (Penning de Vries, 1987). A broad overview of crop simulation models and their applications is presented by Whisler et al. (1986). A survey of the consequences of different crop husbandry measures, such as different planting densities, can be pursued with a model and alternatives can easily be compared (Ng & Loomis, 1984). Results of different timings and dosages of fertilizer applications can be simulated. This opens the door for improving the efficiency of fertilizers and biocides for specific cases, and for reducing the loss of excess inputs to the environment. Guided management, such as this, is already in use on a large scale for optimizing fungicide application in wheat in the Netherlands (Zadoks et al., 1984) and is under development for nitrogen fertilization in arable crops. Crop modelling is also used in 15

irrigation scheduling in the USA. Crop yields can be predicted some time before harvest by using expected weather data. This is important for crops where trade or planning post-harvest operations starts early. Because long-range weather forecasts are not yet reliable, predicting weather is not yet very exact. However, using a range of reasonable weather patterns, a ‘fork’ of yield expectations can be determined. Figure 9 shows this for a tulip bulb crop in the Netherlands. The further the season progresses, the shorter and narrower the fork becomes and the better commercial options can be considered. Models are not yet being used for this type of prediction. Crop performance can be predicted for climates where the crop has not been grown before, or not grown under optimal conditions. This has been used for wheat in Zambia (van Keulen & de Milliano, 1984) and Southeast Asia (Agganval & Penning de Vries, 1988). Using simulation models in this manner has been successful in several cases. Even though it may not be accurate, it is probably as good as an experts opinion, and is easier to get! However, it is not

Figure 9. The course of biomass of a tulip bulb crop in the Netherlands, planted on November 1. Each pair of lines begins from observed values. The lower branch of each pair represents expected growth in a very poor season, the upper branch in a very good season. Repeating the simulation after a month narrows the range of expected values (Source: Benschop, 1986).

16

used very often, possibly because of the lack of essential data (Versteeg, 1985) and possibly because few agronomists realize how modelling can help to establish what yield is possible. Van Keulen & Wolf (1986) have used a crop growth model to estimate yield levels of various food crops on a regional scale where there were no or insufficient external inputs to ensure potential growth. It is expected that this type of simulation can help strengthen regional development and agricultural planning in developing countries (de Wit et al., 1988). A particular form of extrapolation is that in which the physiological or morphological characteristics used in the model are modified. Some crop characteristics are known or expected to vary between cultivars. With simulation techniques, a breeder can survey the impact that breeding for specific characteristics may have (Landivar, 1979, de Wit et al. 1979, Ng & Loomis, 1984). Few documented examples of this type of simulation exist. It is also possible to explore the effects of the increasing ambient CO2 concentration on crop yield, harvest index and water use, and can help breeders to anticipate future requirements (Goudriaan et al., 1984). An almost infinite number of combinations of soil type, weather and agricultural practices exist. Experimenting in all desirable situations is impossible, but using models increases the human capacity. With sequential years of weather data, estimates can be provided for weather-related variability in yield and water use. A helpful technique is generating a long series of daily values for precipitation from historic records of only a few years (Subsection 6.2.5). Variability in yields of sensitive crops or cropping sequences due to variations in weather can be tested with large sets of weather data, which speeds crop assessment by many years. This has been undertaken for Faba bean crops in Western Europe (Grashoff et al., 1987) and for rice in the Philippines (Morris, 1987). Combining a crop growth model with a model for a related biological or physical system, but with a similar time coefficient, can yield an extremely powerful means to investigate interactions between both. Good examples are combinations with pest, disease and weed models (Rabbinge et al., 1989). Strong interactions between crop growth and disease or pest development make this combination potentially interesting for interactive crop management, e.g., a combination model could be used to determine the timing for spraying fungicides, to avoid unnecessary sprayings. Simulation models can probably also be used to derive simple decision rules for farmers and extension services. For example, a wheat crop in a certain area of India is known to become water-stressed when rains fail for five consecutive days (ICAR, 1977) and irrigation is needed. This knowledge has been acquired from long experience and many field trials. Simulation models, supplied with the appropriate crop, soil and weather data can help to derive quickly such rules of thumb for new situations. Moreover, the decision rules can be made more specific and for smaller areas than is possible by relying only on field trials. 17

1.3.5 Basic data The practical value of simulation results depends on the quality of the data characterizing the crop, the weather and the soil. Collecting crop, soil and weather data suitable for simulation is not easy. The amount of data needed is always large and too much to collect first-hand. Hence, existing data sets and literature must be consulted. Much has been published in national and international journals about crops and soils; weather data are reported in bulletins. However, it is often disappointing to discover that published data are difficult to use. For example, experimental conditions are usually insufficiently described; measurements may be have been taken at the wrong time for the anticipated simulation; physical units may be difficult to convert to standard ones; and scientific terms may be used in different ways. When using published data, one must judge the measuring techniques used and on the relevance of the environmental conditions in which the observations were made. Thorough comparative analyses of species or varieties in field conditions are rare (e.g., Cook & Evans, 1983) and full sets of key characteristics for a single species obtained in one trial have not yet been published. Sample data for several crops, soils and climates are given in the next chapters to allow the reader to get an impression of these values and to be able to use them in exercises with the models. This may be helpful to readers without access to good libraries. The crop data presented are from papers identified from an extensive literature search for well-documented experiments and measurements in realistic conditions. The physiological and morphological data per crop were all tested in the three climates presented in Section 6.1 and the resulting simulations yielded acceptable growth and yield curves. This implies that the crop data form consistent sets, not that they are good enough to give a firm prediction of crop performance under other conditions. The data presented may provide a fair starting point for research about the relative effect of changes in crop or soil characteristics (e.g., percentage of yield increase). However, care should be taken to verify and improve on the crop and soil data presented, particularly if simulation of crop performance in absolute terms is attempted (e.g., yield in kg ha -1, or water use efficiency in mm water kg-1 dry matter). A certain amount of common sense remains indispensable for judging simulation results. 1.4 Modelling techniques

1.4.1

Simulation techniques

The simulation techniques used and the biological, soil physical, soil chemical and microclimatological systems considered in this book are simple and straightforward. This is possible because:

18

— the models used here dynamically simulate crop behaviour using only a general explanatory level of physiological and soil physical knowledge, with far less detail than in comprehensive models. The time period for integration is usually one day, sometimes less. — several complex biological and physical processes are programmed as subroutines (see Appendix B). These are well defined and quantified and can be used without going into their scientific details. — a simulation language is used which permits the modeller to focus on scientific problems, rather than on programming. The real world as observed and quantified is simulated as closely as possible. The book contains all details needed to simulate the basic growth and water balance processes for a variety of crops and soils at Production Levels 1 and 2, as a function of weather, cultivation practices, cultivar characteristics and soil types. For those persons interested in applying the models in different situations and for different goals, alternative and more extensive formulations are given where appropriate. The book explains little about programming techniques. However, a few considerations are given in this section and some details about programs are provided in Sections 3.4 and 5.4. All models are written in CSMP, with FORTRAN used in subroutines. A basic knowledge in simulation techniques and CSMP can be obtained from Basstanie & van Laar (1982), Goudriaan (1982a) and Leffelaar & Ferrari (1989). Extensive textbooks on dynamic simulation in CSMP are: de Wit & Goudriaan (1978), Brockington (1979), Penning de Vries & van Laar (1982) and Rabbinge et al. (1989). The manual for CSMPIII (IBM, 1975) contains full technical background and specifications. CSMP was developed for mainframe computers, but is nowadays also used on minicomputers and personal computers. Exercises and answers are found at the end of Chapters 2-5 to allow readers hands-on experience. Emphasis in these exercises is on three crops: rice (because of its importance as a cereal crop); potato (as a common and productive tuber crop in cool regions); and soya bean (an important leguminous crop in tropical and subtropical areas).

1.4.2

Duration of integration periods

Continuous simulation implies simulation according to the state variable approach using relatively short time periods, so that the value of state variables changes only a little in each period (Subsection 1.1.2). How short is relatively short? As a rule of thumb, the time period of integration is about 0.1 times the time coefficient (Leffelaar & Ferrari, 1989). The time coefficient of a system with an exponentially changing state variable (i.e., the rate of change is proportional to the value of the state variable) is equal to the time in which that state variable increases or decreases e-fold (e, the base of the natural logarithm, equals 2.73). A young crop growing exponentially doubles its weight 19

about once per week, so that the corresponding time coefficient is 10 days. Time periods of one day are thus sufficient for accurate simulation. The time period can be larger in later growth phases when changes in the crop are relatively slow. Simulation with time periods as long as 10 days has been performed successfully with small (van Keulen, 1976) and large models (Jackson et al., 1983; van Keulen & Wolf, 1986). In such models it is implicitly assumed, either that the environment does not change considerably within the time period of 2-10 days, or that the effect of the average condition equals the averaged effect of changing conditions. Often neither assumption is correct for field crops. The soil water content, an important variable, can change greatly within a few days (Figure 10) and diseases can develop very quickly. Moreover, several processes respond to environmental conditions in a non-linear manner. For example, one or two days in a decade may be so dry and hot that the crop is waterstressed and photosynthesis is reduced. In such cases averaging temperature over a longer period gives a higher total photosynthesis than with a day by day calculation.

Figure 10. Simulated changes of the soil water content under a millet crop in a semi-arid zone in Mali during the rainy season (line) and observations (points) (Source: Jansen & Gosseye, 1986).

20

A one-day time period is, therefore, a logical choice for many crop growth models. Moreover, weather data are usually supplied on a daily basis. There is usually little interest in changes of state variables for crops over periods shorter than one day. It can be argued that simulation with one-hour time periods would provide even more accurate results. This may be correct once comprehensive models are developed much further and computer capacity is virtually unlimited. As yet, using one-day time periods appears to be adequate. Temperature fluctuations during the growing season are usually well within the extremes of response curves for agricultural crops. The response is almost linear in this range and averaging over 24 hours is quite acceptable. But a shorter time period should be used if the temperature often exceeds the threshold beyond which important processes stop or respond differently (Figure 11). Simulation using shorter time periods is especially necessary for crops which accumulate starch during the day, for this may lead to declining photosynthesis. For example, the canopy photosynthesis light response curve of the potato crop measured in the morning and afternoon can show hysteresis for this reason (Subsection 2.1.4). An objective of the model presented here is to be able to simulate crop growth in moderately unfavourable environmental conditions. Therefore, a version of the crop growth model was developed to handle quarter-day time periods at Production Level 1. The rate of crop transpiration is in the order of 5 mm d-1. A crop of 5000 kg dry matter ha-1 contains water roughly equivalent to a water layer of 5 mm. A

Figure 11. The response curve of maximum leaf photosynthesis to temperature. If daily temperature fluctuates within ranges A or B, one-day time periods are appropriate; if temperature often fluctuates over ranges C or D, shorter time periods should be taken.

21

20% decrease in available water is generally sufficient to cause severe drought stress. In other words, the daily flux of water through a crop is many times larger than the amount a crop can lose in a day without suffering. To simulate the dynamics of a crop water balance, the model should proceed with time periods of a few hours or less. Several scientists have used this method (e.g. de Wit et al., 1978). It is not necessary to follow this procedure here, because the ratio of water use to photosynthesis does not depend on the rate of water use (Section 4.1). Therefore, fluctuations in the transpiration rate during the day (due to radiation, air humidity, windspeed and precipitation) are unimportant and the total or average transpiration rate is sufficient. The simulation of the water balance for partially saturated soils requires time periods in the order of hours or less, because water fluxes between soil layers can be relatively fast. A model with short time periods is attached to the crop model with one-day time periods (Section 5.3).

1.4.3

Continuous and discrete simulation

Using a time period of one day (1 d) or a fraction of a day (e.g., 0.25 d), provides a continuous simulation of growth, despite the fact that inputs change in a discontinuous way (i.e., they have differing values from one day to the next). Such a change is a first order discontinuity (Goudriaan, 1982a). The CSMP integration method RECTangular or the Euler integration, is completely adequate in these cases. Other methods should not be used with the models described in this book. Zero order discontinuities are sudden changes of quantities in the model. Simulation of discrete processes requires techniques other than continuous simulation, hence, CSMP is not suitable (the language SIMULA is appropriate for discrete models: van Elderen, 1987). However, elements of discrete simulation, enter into crop growth simulation models. Harvesting is a clear example for all biomass is removed in a very short time. In water balance simulation, withdrawal of water from thick soil layers is relatively slow and its simulation is continuous, but the fast withdrawal of water from top soil layers is simulated more or less discretely. Water infiltrates into soil layers quickly and this can be simulated in a discrete fashion by adding a certain amount of water to soil layers (Section 5.1). These cases are made explicit by dividing by the integration period, DELT. (This is done even when DELT equals 1.000 to maintain the correct use of dimensions in the program.) If properly carried out, there is no problem in mixing discrete and continuous simulation such as this. However, if modellers prefer to avoid mixing continuous and discrete simulation, much shorter integration periods can be used. De Wit et al. (1978) used this method with the model PHOTON, derived from BACROS. However, the same problem recurs at a more detailed physiological level, such as for stomatal opening and closing. Further elaboration of the model may overshoot the original problem for which it was built. Another method which avoids 22

mixing discrete and continuous simulation is to construct model sections that run with different time periods (bypass method, Goudriaan, 1977, Section 5.3). The running average concept is used a number of times in the programs presented in this book. A running average is an average of a variable in which recent values count more than older values. A running average ( AV ) of the variable (V ) can be obtained in CSMP by the statement: AV = INTGRL (0., (V – AV)/TC) TC is the time coefficient of the adjustment of AV to new values of V. Because of this constant adjustment, previous values of V become less and less important. A running average can be calculated for state, intermediate, external and even for rate variables. 1.4.4

Model and data management

The models discussed in Sections 3.4 and 5.4 are presented as complete modules. Depending on the system simulated, one or more modules are to be used. Modules need to be combined with crop, soil and weather data, and with a specific set of subroutines and functions. A module with all subroutines and functions used in this book is given in Appendix B. The user can select the functions and subroutines needed, and add these between the CSMP statements STOP and ENDJOB. Size related problems may occur on personal computer with only 512 K RAM with programs that exceed the size of the largest programs presented in this book (Section 5.4). Putting all functions and subroutines in a subroutine library reduces the program size sufficiently to avoid these problems and speeds compilation significantly. This procedure is described in the CSMP (IBM, 1975) and FORTRAN (e.g., IBM, 1984) manuals. Because of these features, using a subroutine library is recommended. When working frequently with different sets of crop, soil and weather data, it is useful to keep the program modules, and the modules with crop, soil, and weather data in different files on the computer, and to only join sets together when a specific combination must be run. Figure 12 provides a diagram of possible choices for crop growth simulation at Production Level 1. For Production Level 2, extra crop and soil sections should be inserted before the END statement (Section 5.4). This procedure is not part of CSMP, but can be written in the operating system language of most computers. The procedure saves space and, more importantly, makes any correction or addition to a program or a data set effective immediately and wherever used. Easy changing of structure or data in models is both an asset and a liability. It is a liability because it is tempting to make changes without fully investigating whether the change is an improvement. To ensure that making a change in the model is an asset, always document all changes in a notebook; execute dimen23

Figure 12. Alternative modules (L1D and L1Q) and alternative crop and weather data sets can be combined with a terminal section (T12) to complete a program for crop growth at Production Level 1.

sion analysis of new equations rigidly; make checks of balance calculations by comparing the accumulated net flux with the amount retained; and compare new data with values already known and evaluated. 1.4.5

Evaluation

Often, the first thorough test of a model is the comparison of its behaviour with that observed of the real world in a similar situation. This behaviour includes, for example, the general shape of the time course of variables, the presence of discontinuities and the qualitative sensitivity of output to parameter values. However, be aware that aspects of model behaviour that seem counterintuitive at first, sometimes turn out to be realistic. If the behaviour of the model qualitatively matches that of a system in the observed world, a quantitative comparison and evaluation of the predictive success of the model 24

should be made. At this stage, statistical tools can be useful. But even when sufficient and accurate data are available, a model cannot be proven to be correct. Model behaviour can sometimes be falsified and one or more model components must then be in error. Calibrating a model is the adjustment of some parameters so that the model matches one set of observed or measured data. Simulation of this data set with the calibrated model is a very restricted form of evaluation. Calibrating several parameters simultaneously degrades simulation into curve fitting. Sensitivity analysis is a procedure in which the value of a parameter is increased or decreased by a certain percentage and the effect on the behaviour of the model recorded. Sensitivity analysis can be part of the evaluation of a model, but it is particularly useful when determining the accuracy with which parameters have to be established experimentally. Behavioural analysis is a useful form of sensitivity analysis, particularly if it is possible to critically discuss results with crop specialists. Further information about sensitivity analysis, evaluation, validation and verification can be found in Baker & Curry (1976), Penning de Vries (1977) and Steinhorst et al. (1978).

25

2

Assimilation and dissimilation of carbon

This chapter discusses and quantifies the assimilatory and dissimilatory processes of crop growth. Figure 13 provides an example of the relative importance of these processes for two crops. It indicates the partitioning of the total carbon captured by harvest-ripe crops during assimilation and dissimilation processes and the corresponding amount of biomass produced. The relative importance of these processes can vary a lot between crops and growth conditions. Photosynthesis and remobilization are carbohydrate sources; their simulation is discussed in Sections 2.1 and 2.2. Simulation of respiration processes related to maintenance and growth are discussed in Sections 2.3 and 2.4. Section 2.5 contains exercises.

Figure 13. An example of the relative importance of assimilatory and dissimilatory processes for two crops, showing the partitioning of carbon (C) captured over these processes and the corresponding amount of biomass formed. (1) the amount of C in biomass formed directly from assimilates; (2) the amount of C in biomass formed from remobilized carbohydrates; (3) and (4) the amount of C lost in growth and maintenance processes, respectively; (5) the excess energy involved in maintenance and NO 3 - reduction and (6) the energy involved in photorespiration are both expressed as C-equivalent. Gross photosynthesis is 1 + 2 + 3 + 4, net photosynthesis is 1 + 2, respiration is 3 + 4.

27

Morphological development and biomass partitioning is discussed in Chapter 3, and two complete modules for simulating crop growth at Production Level 1 can be constructed from Chapters 2 and 3. They are presented in Listing 3, a basic crop growth module with one-day time periods (L1D) and Listing 4, a crop growth module with quarter-day time periods (L1Q). 2.1

2.1.1

Photosynthesis

Introduction

Photosynthesis is the driving force behind growth at Production Level 1. In the programs presented here, daily photosynthesis is computed for whole canopies by using a summary model. Daily photosynthesis is the basis for computing the rate of crop growth. The three most important factors in photosynthesis are: the photosynthesis light response curve of leaves, the radiation intercepted by leaf canopies and the distribution of light within canopies. These are presented below, together with demonstrations of how they can be integrated and used to yield totals for each day of canopy photosynthesis under various conditions. Environmental variables considered are radiation, temperature and the ambient CO2 concentration. Photosynthesis was discovered in the later part of the eighteenth century by Priestley and Ingen Housz and since then has been extensively studied. The new Encyclopedia of Plant Physiology devoted two volumes (5 and 6, 1979) to this process alone. Photosynthesis comprises very complex processes by which plants reduce CO2 and form organic molecules using absorbed radiation energy. These molecules are conveniently represented by the molecule glucose (C6H12 O6). Glucose molecules either serve as the building blocks for virtually all organic constituents in plants, or are respired to provide energy for metabolic processes. Three groups of plants are often distinguished on the basis of their biochemical mechanism of photosynthesis: C3, C4 and CAM plants (Bidwell, 1983). Among the major agricultural crops, pineapple is the only CAM plant, so this group is ignored here. The nature of the difference between C3 and C4 crops is of little significance for the purpose of this book, but it is important to recognize that C4 crops generally perform much better than C3 crops in warm climates, but less so in temperate regions. Morphological and biochemical details can be found in the Encyclopedia of Plant Physiology, Volume 6 (1979). ‘Photosynthesis’ is used here for gross photosynthesis and includes photorespiration, an intensive process in C3 plants that is intimately coupled with photosynthesis itself. Photorespiration does not lead to any product and is almost completely suppressed in C4 plants. Suppression of dark respiration processes during photosynthesis is discussed in Subsection 2.3.3. The rate of leaf photosynthesis is conveniently expressed per unit of leaf area (counting only upper sides). Canopy photosynthesis is the sum of the 28

contributions of all leaves, stems and sometimes reproductive organs. Only photosynthesis of leaves, by far the most important, is considered extensively here.

2.1.2

Leaf photosynthesis

The photosynthesis light response curve The response of leaf photosynthesis to absorbed light can be described as a curve that relates the rate of gross photosynthesis (PL, kg CO2 ha-1 h-1 ) to the intensity of absorbed radiation (PAR, J m -2 s-1 ) exponentially (Figure 14). The exponential form corresponds best with most observations (Goudriaan, 1982b):

This type of curve is characterized by two parameters: the slope at the origin (PLEA, kg CO2 ha -1 h -l (J m -2 s -1) -1) and the rate at saturated light intensity (PLMX, CO2 kg ha-1 h -l). The initial efficiency of the use of absorbed light characterizes, in particular, the biophysical processes and has a fairly constant value. The maximum rate depends strongly on plant properties and environmental conditions and particularly reflects biochemical processes and physiological conditions.

Figure 14. The response curve of gross photosynthesis of a single leaf versus the intensity of absorbed radiation (PAR). PLEA, the tangent of angle alpha, is the initial efficiency of light use, PLMX the maximum level to which the exponential curve rises.

29

Light absorption by leaves Only radiation from a part of the light spectrum (400-700 nm) is effective for photosynthesis. This photosynthetically active radiation (PAR) is about 50% of solar radiation (Subsection 6.1.2). A fraction of the radiation reaching leaves is reflected or transmitted. The values of reflectivity and transmissivity, complements of absorptivity for PAR, are remarkably similar among leaves of healthy crops (Sinclair et al., 1971). Moreover, reflectivity and transmissivity are numerically often about equal and have a value of 0.1 each (Goudriaan, 1977). When leaves are obviously yellow or are extremely thin, there is not enough chlorophyll (less than 30 microgram cm-2) to absorb light and transmission doubles or quadruples (A1berda, 1969; Lin & Ehleringer, 1982). Only when thick leaves carry reflecting hairs (Ehleringer, 1976) can reflection double or quadruple. The fraction of PAR absorbed by leaves is assumed here to always be 0.80 (the complement of the constant SCV in the function FUPHOT and in the subroutine SUPHOL, Appendix B). Initial light use efficiency The theoretical minimum energy requirement for reduction of a CO2 molecule is about 9.5 quanta (PAR) for light of wavelengths 540-670 nm, but the lowest values observed in C3 plants in sunlight in the absence of photorespiration are about 13 quanta per molecule (Farquhar & von Caemmerer, 1982). The difference is caused by light absorption by non-photosynthetic pigments and by the lower (on average 10%) light use efficiency at other wavelengths (McCree, 1982). Photorespiration, induced in C3 plants by O 2, increases the energy requirement to at least 15 quanta per CO2 molecule. Light use efficiency (PLEI) such as this can be converted into the more practical measure of a rate of 0.48 kg CO2 per hectare of leaf surface and per hour per J m-2 s-1 of PAR (de Wit et al., 1978). This is typical for all C3 species at relatively low temperatures (around 10 °C). For C4 plants the initial light use efficiency is about 0.40. It is lower than that of C3 plants at low temperatures because suppression of photorespiration is a costly process. The relative importance of photorespiration increases as temperature increases and the initial efficiency (PLEA) goes down as a result. The value for a C4 crop is reduced to 0.3 at relatively high temperatures (around 30 °C) and to 0.0 at even higher temperatures. Its value remains constant at temperatures up to 45 °C in C4 crops and at higher temperatures drops quickly to 0.0 (Ehleringer & Bjorkmann, 1976; Berry & Downton, 1982). These data, summarized in Table 3, can be of help when values for a specific case are unavailable. Some data of the initial light use efficiency for different species are included in Table 4. Environmental factors other than temperature have little effect on how efficiently plants use radiation. For simulation of the effect of photorespiration on light use efficiency, consult Goudriaan (1982b). 30

Table 3. Typical values of the initial efficiency of use of absorbed light for photosynthesis by individual leaves at different temperatures for the main crop types (kg CO2 ha-1 h -l per J m-2 s-1 of absorbed PAR).

Crop type

Temperature (°C)

C3 C4

0 0.5 0.4

30 40 50 10 20 0.5 0.45 0.3 0.1 0.01 0.4 0.4 0.4 0.4 0.01

Maximum rate of leaf photosynthesis The value of the maximum rate of leaf photosynthesis (PLMX) at high light intensities and normal CO2 concentration is of major importance for crop growth simulation. Its value is usually 25-80 kg CO2 ha-1 h-1, but can exceed this range. An example is given in Subsection 4.1.5 for a very high maximum photosynthesis rate of a maize crop. To quantify PLMX, it can be considered that during photosynthesis a physical and a biochemical process run parallel. CO2 diffuses from the ambient air to the carboxylation sites in the cells, and the total diffusion resistance and the concentration gradient set an upper limit to PLMX. Simultaneously CO2 is converted into glucose in a biochemical chain reaction. RuBPCase is the most prominent and presumably the most rate-limiting enzyme (it may form 50% or more of leaf protein in C3 plants). Its concentration and maximum activity also determine the maximum value of PLMX. The capacities of the physical and biochemical processes tend to adjust to each other so that, in theory, either could be used to quantify PLMX. Yet, predicting the value of PLMX for a particular case from physiological or physical parameters is still inaccurate. Therefore, accurate measurements of PLMX are indispensable. Table 4 provides some data on PLMX determined under standard conditions. The maximum rate of leaf photosynthesis per unit of leaf area is strongly related to leaf thickness and temperature. Quantifying these relations experimentally requires sophisticated equipment and much work. If PLMX measurements are lacking, the influence of leaf thickness and leaf temperature may be quantified by considering PLMX at standard temperature and thickness to be a value characteristic to a species or cultivar. Leaf thickness (more properly, specific leaf weight in kg dry matter per ha of leaf surface) plays a role because thicker leaves usually have more RuBPCase per unit surface than thin leaves. Differences in leaf thickness are the major cause of differences between the maximum rates of photosynthesis of plant cultivars (Charles-Ed31

Table 4. The initial efficiency of use of absorbed light (PLEI, measured at a low temperature) and the maximum rate of leaf photosynthesis (PLMX) of individual leaves of different species at an optimum temperature at 340 vppm CO2 , and for a characteristic specific leaf weight (given in Table 19). Values refer to field grown crops.

C3 /C4

PLEI PLMX Temp kgha-1h-1 ºC kgha-1 h-1 *(Jm-2s-1)-1

Reference

Barley Cassava Cotton Cowpea Faba bean Groundnut

C3 C3 C3 C3 C3 C3

0.40 0.50 0.40 0.50 0.48 0.50

35 35 45 64 35 50

25 25 35 25 25 30

Maize Millet Potato Rice Sorghum Soya bean

C4 C4 C3 C3 C4 C3

0.40 0.40 0.50 0.40 0.45 0.48

60 70 30 47 70-10 40

25 25 20 25 30-35 30

Sugar-beet Sugar-cane Sunflower Sweet potato Tulip Wheat

C3 C4 C3 C3 C3 C3

0.56 0.40 0.45 0.50 0.50 0.50

38 70 60 30-35 40 40

20 25 28 25 15 10-25

Dantuma, 1973 estimate Muramoto et al., 1965 estimate van Laar CABO pers.com. Bhagsari & Brown, 1976; Bhagsari et al., 1976; Pallas & Samish, 1984 Sibma CABO pers.com. Jansen & Gosseye, 1986 Teubner, 1985 Yoshida, 1981 Eastin, 1983 Beuerlein & Pendleton, 1971; Dornhoff & Shibles, 1970 Sibma CABO pers.com. estimate Rawson & Constable, 1980 Hahn & Hozyo, 1983 Benschop, 1986 van Keulen & Seligman, 1987; Joliffe & Tregunna, 1968

Species

wards, 1981) and between plants grown in the field and in phytotrons. The simplest form of the PLMX — thickness relation is adopted here by using a strict proportionality for a normal range of leaf thicknesses of 200-600 kg ha-1 (Listing 3 Line 53; this listing contains a complete crop module (L1D, see Chapter 7)). This implies that leaf thickness is a crucial variable in the simulation programs in this book (Section 3.3). Evidence to support this view for many crop species with an adequate nutrient supply is provided by Khan & Tsunoda 32

(1970), Figure 15, Gulmon & Chu (1981), Bjorkman (1981) and Gifford & Evans (1981). Examples to the contrary also exist, for example in sugar-beet PLMX is negatively correlated with leaf thickness, unless the latter is corrected for the increase in organic acid content associated with age (Sibma, CABO, personal communication). The proportionality factor for C3 crops at an optimal temperature for photosynthesis is about 0.1 kg CO2 ha-1 h-1 per kg leaf ha-1, while that of C4 crops is 0.15-0.2. The thickness – photosynthesis ratio depends strongly on the nitrogen content of the leaf, but this effect is not relevant here because this ratio is applied at Production Levels 1 and 2 where nutrients are in ample supply and because the rate is measured at a fixed stage of crop development. Attention should always be given to the influence of temperature on photosynthesis. The curve relating PLMX to temperature shows an optimum (Figure 16). C3 plants generally perform better than C4 plants at low temperatures (less than 15 °C), and vice versa at high temperatures (more than 25 °C). Table 5 presents data on this relationship for several crops. However, these relationships should not be copied rigidly as many exceptions have been recorded (e.g., Berry & Downton, 1982). Breeding may have modified the PLMX temperature response curve. Miedema (1982) indicates that photosynthesis at 10 °C and at 20 °C in cold-tolerant maize is similar to that in subtropical maize

Figure 15. The maximum rate of leaf photosynthesis as a function of specific leaf weight in wheat (Source: Khan & Tsunoda, 1970).

33

Figure 16. The maximum rate of leaf photosynthesis as a function of leaf temperature for a potato (C 3 ) and a sorghum (C4 ) crop. Data from Tables 4 and 5.

at temperatures 5 °C higher (Table 5). A shift of this magnitude has contributed greatly to the large increase in maize production in the Netherlands since 1970. A shift of 3-4 ºC in the temperature — PLMX response curve was found by Kwon (1984) between an Indica-Japonica and a more cold-tolerant Japonica rice variety. Natural selection has led some C3 species to have a response curve similar to that of typical C4 species (Werk et al., 1983). Leaf temperature can be several degrees above or below air temperature, depending on environmental conditions and the moisture status of the soil (Subsection 4.3.2). However, for simplicity it is assumed here that average leaf temperature is equal to average air temperature. Adaptation of individual plants to other temperature regimes can lead to modification of the actual temperature response curve (Berry & Downton, 1982). To allow for adaptation such as this, the response introduced in a model may have a temperature optimum that is 5 °C broader than that determined in a short-term experiment.

2.1.3

Canopy photosynthesis

The principle of canopy photosynthesis If the photosynthesis rate was proportional to the light intensity and if all leaves had identical properties, then canopy photosynthesis would simply be equal to the multiple of the quantity of light absorbed and the light use efficiency. However, leaves become saturated at high light intensities and they are all exposed differently to radiation. Hence, the relation of canopy photosyn34

Table 5. The effect of leaf temperature on the maximum rate of leaf photosynthesis of several crop species. (See original publication for the cultivar used.)

Barley (Joliffe & Tregunna, 1968) FUNCTION PLMTT = -20.,0.001, 0.,0.01, 5.,0.4, 10.,0.7, 15.,0.9,... 20.,1., 25.,1., 30.,0.9, 35.,0.8, 40.,0.5 Cotton (El-Sharkawy & Hesketh, 1964a; Ludwig et al., 1965) FUNCTION PLMTT = 0.,0.001, 10.,0.3, 20.,0.6, 25.,1., 30 ., l.,... 35.,0.8, 40.,0.5, 50.,0.001 Phaseolusvulgaris(Jones, 1971) FUNCTION PLMTT = –10.,0.01, 0.,0.01, 10.,0.59, 15.,0.76 ,... 20.,0.93, 25.,1., 30.,0.92, 35.,0.84, 40.,0.75 Groundnut (Bhagsari & Brown, 1976; Pallas & Samish, 1974) FUNCTION PLMTT = 0.,0.001, 10.,0.3, 20.,0.6, 25.,1.,... 30.,1., 35.,0.8, 40.,0.5, 50.,0.001 Maize (Hofstra & Hesketh, 1969) FUNCTION PLMTT = 0.,0.01, 5.,0.01, 10.,0.1, 15.,0.5, 20.,0.8,... 25.,1., 35.,1., 40.,0.9, 45.,0.75, 50.,0.07 Millet (Jansen & Gosseye, 1986) FUNCTION PLMTT = 0.,0., 10.,0., 20.,1., 40.,1., 50.,0. Potato (van Heemst, CABO, personal communication) FUNCTION PLMTT = –20.,0., –5.,0.01, 5.,0.02, 15.,0.8, 20.,1.,... 25.,1., 30.,0.8, 37.,0.0 Rice (van Keulen, 1976) FUNCTION PLMTT = 0.,0.01, 10.,0.01, 20.,1., 35.,1., 42.,0.01 Sorghum (El-Sharkawy & Hesketh, 1964a) FUNCTION PLMTT = 0.,0.001, 10.,0.01, 15.,0.3, 20.,0.6, 25.,0.9,... 30.,1., 40.,1., 45.,0.9, 50.,0.8, 55.,0.4, 60.,0.001 Soya bean (Hofstra & Hesketh, 1969) FUNCTION PLMTT = 0.,0.001, 10.,0.3, 20.,0.6, 25.,0.8, 30.,l.,... 35.,1., 40.,0.8, 45.,0.4, 50.,0.001 Sugar-beet (Hofstra & Hesketh, 1969; Hall & Loomis, 1972) FUNCTION PLMTT = 0.,0.01, 5.,0.01, 10.,0.75, 20.,1., 35.,1.,... 40.,0.9,45.,0.01 Sunflower (El-Sharkawy & Hesketh, 1964a) FUNCTION PLMTT = 5.,0.1, 10.,0.5, 15.,0.7, 20.,0.9, 25.,0.95,... 30.,1., 35.,1., 40.,0.7, 45.,0.3, 50.,0.01 Sweet potato (van Heemst, CABO, personal communication.) FUNCTION PLMTT = –5.,0.01, 5.,0.02, 15.,0.8, 20.,0.9 ,... 25.,1., 30.,1., 35.,0.9, 40.,0.5, 45.,0.001 Wheat (van Keulen & Seligman, 1987) FUNCTION PLMTT = 0.,0.0001, 10.,1., 25.,1., 35.,0.01, 50.,0.01

35

thesis to light intensity is curved and varies greatly for different situations (Figure 17). Canopy photosynthesis is the sum of the rates of photosynthesis of all leaves. It is expressed in kg CO2 per hectare of ground surface and per day. To compute its value, canopies are thought to be divided into relatively thin ‘layers’ of leaves containing 0.1-1 m2 of leaf surface per square meter of ground surface. Light intensity in the top layers is highest and decreases towards the base of the canopy. Light must be distinguished as a fraction of direct light coming from a point source (the sun) and a fraction of diffuse light coming from all directions (Subsection 6.1.2). Extinction of light (PAR) occurs with different coefficients: 0.50 for direct and 0.72 for diffuse radiation when the leaf angle distribution is spherical (i.e., leaf surfaces distributed like the surface on a globe) (Figure 18). Its average extinction coefficient is about 0.6 for a canopy with erect leaves and 0.8 for one with horizontal leaves (Goudriaan, 1977). Leaves grow at varying angles from the horizontal. This can be represented by a cumulative leaf angle distribution curve (Figure 19). The intensity of radi-

Figure 17. Photosynthesis light response curve of canopies with different characteristics and at different latitudes at June 15: LAT ALV PLMX PLEA 40 5 50 0.5 Curve 1 2 20 50 0.3 Curve 2 0 5 100 0.4 Curve 3 0 5 0.4 70 Curve 4 0 2 35 0.4 Curve 5

36

Figure 18. The approximate reduction of the intensity of the total global radiation from the top of the canopy downwards (full line), and the diffuse radiation fraction of the total global radiation (dashed lined) at each level of the canopy. The direct and diffuse component are equal above the canopy. The curves are computed with an extinction coefficient of 0.5 for total radiation and 0.7155 for diffuse radiation.

Figure 19. Cumulative leaf angle distribution of a maize crop (Source: de Wit, 1965).

37

ation on individual leaves is anywhere between the intensity of diffuse light only (shaded leaves) and that of diffuse plus direct light (sunlit areas perpendicular to solar rays). The actual leaf angle distribution within a normal range has little effect on canopy photosynthesis (de Wit, 1965). Clustering of leaves, relatively important for young plants and row crops, reduces canopy photosynthesis less than may be expected. The contribution of photosynthetic area other than that of leaves to canopy photosynthesis is discussed in Section 3.3.

Modelling canopy photosynthesis Many simulation models have been developed for canopy photosynthesis (Hesketh & Jones, 1980). A pioneer model by de Wit (1965) has been improved and expanded (Goudriaan, 1977, 1986; de Wit et al., 1978; Spitters, 1986; Spitters et al., 1986) and because of its versatility and documentation its approach is followed in this book. An excellent way to calculate the daily gross photosynthesis of a canopy with a summary model was presented recently by Goudriaan (1986, 1988). It uses a specific way of integrating the instantaneous rate of leaf photosynthesis in time (three points between noon and sunset, times two, assuming the morning to be equal to the afternoon) and in space (three depths in the canopy). The rate of leaf photosynthesis is estimated from the actual leaf photosynthesis light response curve and the amount of direct and diffuse light at that time of day and depth in the canopy. The path of radiation intensity during the day is assumed to be sinusoidal. Results obtained using this summary model were extensively compared with those of the comprehensive models and found to agree very closely. The FUPHOT function calculates canopy photosynthesis in this way (Listing 3 Line 52, Listing 4 Line 72). Both crop characteristics PLMX and PLEA are adjusted for temperature and other conditions. (A decrease of PLMX and PLEI with depth in the canopy is discussed in Subsection 2.1.4). The extinction of PAR in the canopy is characterized by the extinction coefficient; its value is a constant in the standard photosynthesis function (KDIF in FUPHOT). Canopy photosynthesis calculations have been computed year-round for different geographical latitudes and for leaf areas up to 10 m 2 m -2 (Figures 20 and 21). They show how much canopy photosynthesis varies as a result of variations in light and that the variations are not proportional with leaf area. The relation between canopy photosynthesis and the maximum rate of leaf photosynthesis (PLMX) also shows a less than proportional relationship (Figure 22). Canopy photosynthesis does not increase significantly above a leaf area index of 4 or 5, with a spherical leaf distribution at any value of PLMX. A three – to – four-fold increase in PLMX, from 20 kg CO2 ha -1 d -1 , would only double the rate of canopy photosynthesis on a clear day. The effect on a cloudy day is even smaller. On days that are not fully overcast or clear, radiation can be distributed 38

Figure 20. The simulated course of daily canopy photosynthesis throughout the year at 0°, 20°, 40° and 60° northern latitude for a C3 crop (PLEA = 0.5, PLMX = 40., dashed lines), and at 0° and 60° for a C 4 crop (PLEI = 0.4, PLMX = 90., full lines), all at a leaf area of 5 m2 m -2 .

Figure 21. The relation of canopy photosynthesis with leaf area at two dates and for two values of PLEA (0.3 and 0.5), and both with LAT = 50., PLMX = 40.

39

Figure 22. The simulated rate of canopy photosynthesis on a fully clear 21st June in Wageningen for different values of the maximum rate of leaf photosynthesis and at several values of the area of the leaves (in m2 m-2). Dashed lines are the response curves under overcast conditions.

quite unevenly over the day. Using a comprehensive model it was established that daily canopy photosynthesis is some 10% lower when all radiation is concentrated at noon on a partly cloudy day, as compared to a constantly cloudy day (Figure 23). This has only a moderate effect on canopy photosynthesis and is disregarded here. Certain parameters are almost invariable in the comprehensive model for canopy photosynthesis. In other parameters the variability normally encountered has little effect, and such features are either not mentioned or kept constant. Table 6 provides a list of the major assumptions in the summary model. The radiation intercepted on any day gives rise to a certain amount of photosynthesis (PCGW). This photosynthesis rate is used as is to compute the rate of crop growth (Listing 3 Line 39), or, in the module with quarter-day time periods (Listing 4 Line 71) it is divided by daylength to express it as a rate per 24 hours of constant radiation.

2.1.4

Special cases for photosynthesis

Carbohydrate accumulation Starch and glucose sometimes accumulate in leaves. Net photosynthesis is 40

Figure 23. The simulated rate of canopy photosynthesis as a function of the fraction of clear sky radiation when the level of cloudiness is constant all day (full line) or when the same amount of radiation is concentrated around noon (dashed line).

Table 6. Assumptions concerning computation of daily canopy photosynthesis in a summary model.

– Leaf photosynthesis responds instantaneously and fully to changes in light intensity, i.e., leaf movements in the wind or clouds are without after-effects. – PLMX and PLEI have a constant value from top to bottom of the canopy (FUPHOT) unless specified per layer (SUPHOL). – The leaf angle distribution is spherical (FUPHOT) unless specified differently (SUPHOL). – Leaf thickness matters a great deal; leaf form and size are unimportant. – Leaves do not cluster much. Crops are not grown in rows. – The CO2 concentration inside the canopy equals that above it. – Leaf temperature is on average equal to air temperature.

41

impaired when the glucose level is too high; starch probably has less effect. This reduced photosynthesis is partially caused by the glucose concentration stimulating respiration (Azcon-Bieto et al., 1983) and partially by a lowering of PLMX. To simplify simulation of this phenomenon, only the effect on PLMX is quantified (an effect of sugar concentration on maintenance respiration is discussed in Subsection 2.3.2). The carbohydrate level fluctuates during the day and PLMX may fluctuate consequently. Van Keulen & Seligman (1987) use 30% of leaf weight as the carbohydrate level above which PLMX is reduced to almost zero and this approach is followed in module L1Q (Listing 4, Line 75). The cut off level (30%) and the extent to which photosynthesis is reduced (30%) are estimates only and are used for all species for lack of information. Moreover, no distinction is made between starch and glucose which is probably necessary. If carbohydrate accumulation is important, use the L1Q quarter-day time period module (Section 3.4). Carbohydrate accumulation in leaves can occur: – in crops where the export of carbohydrates can be slower than its production (as in potato crops where leaf photosynthesis, even at constant external conditions, decreases during the day and the canopy photosynthesis light response curve shows hysteresis, Bodlaender (1986)); – in situations where daytime photosynthesis is less restricted than growth during any 24-hour period (as in spring in temperate climates when night temperatures are still low, but leaf temperatures during the day permit high photosynthesis); – as a result of water or nutrient stress; – as a result of small sink size of storage organs, such as immediately after heir initiation (elaborated in Subsection 3.2.5); and – as a result of phloem transport being blocked by insect pests or diseases. High CO2 concentrations At the site of carboxylation CO2 is bound to an enzyme in competition with O2 . Carboxylation leads to photosynthesis, oxidation leads to photorespiration. The higher the CO2/O2 ratio, the lower the photorespiration and the higher the maximum rate of photosynthesis and the initial light use efficiency. The maximum rate of leaf photosynthesis is about proportional to the CO2 concentration below the normal level of 340 cm3 CO2 m-3 air (340 vppm). The proportionality holds up to CO2 levels of about 700 vppm in many C3 species. On the other hand, in C4 plants concentrations beyond 340 vppm increase photosynthesis little or not at all (Figure 24). These effects have been well investigated and reviewed (Encyclopedia of Plant Physiology, Vol 6, 1979). The CO2 concentration in greenhouses can be much higher or lower than 340 vppm. The ambient CO2 concentration rises annually by 1-2 vppm (Goudriaan, 1987), which, in many cases, causes a slow increase in photosynthesis. This topic has been discussed extensively in the literature (e.g., Berry & Downton, 1982; Gates et al., 1983; Subsection 6.1.7). 42

Figure 24. The photosynthesis light response curve of sunflower leaves at three ambient CO2 concentrations (Source: Goudriaan & van Laar, 1978b).

The consequences of high or low CO2 concentrations on crops can be simulated by adjusting PLMX in relation to the ratio of the new CO2 concentration and 340 vppm. In this way Goudriaan et al. (1984) established that the overall effect of an increased CO2 concentration (C) on net assimilation (A ) of a canopy can be described by Equation where o refers to 340 vppm and x to the new situation; crops and 0.8 for C3 crops.

1

is about 0.4 for C4

A layered canopy Maximum leaf photosynthesis in a senescing crop declines in time. The oldest leaves in the base of the canopy are affected first. Many diseases also affect the crop in this way. In some cereal varieties the top leaves are clearly more erect than leaves in the lower layers. To deal with layers with different characteristics, Goudriaan (1986, 1988) extended the FUPHOT function to the SUPHOL subroutine. Computations are per leaf layer. The radiation at the bottom of one layer is the input to the next lower layer. Up to five layers are distinguished, each with its own values for maximum leaf photosynthesis and initial light use efficiency. The same variable names as for FUPHOT are retained, but these are now names of arrays with as many elements as there are layers. The areas of active and dead leaves are also specified per layer. Dead 43

leaves in any layer are supposed to provide shade to the leaves in that layer. The CSMP modules Listing 3 (LlD) and Listing 4 (LlQ), presented in Chapter 7, can be expanded to simulate a crop with two different leaf layers as follows: STORAGE PCGCL(5), PLMX(5), PLEA(5), ALVL(5) ,ALVDL(5) PCGC, PCGCL = SUPHOL (2, PLMX, PLEA, ALVL, ALVDL, Fl, F2,... RDTM, DATE, LAT) PROCEDURE ALVL, ALVDL = ALVPRO (ALV) ALVL(1) = AMIN1 (2.5, ALV) ALVL(2) = ALV — ALVL(1) ALVDL(l) = 0.0 ALVDL(2) = 0.2 * ALVL(2) ENDPROCEDURE The number of layers considered (two in this example) is the first number of the SUPHOL inputs. The STORAGE declaration must precede the new arrays. The statements for leaf area are in a ‘procedure’, meaning that they are sorted as a group with the variable inputs and outputs defined in the first line. CSMP cannot sort indexed variables and using a procedure is a correct alternative method. Leaf photosynthesis characteristics can also be computed per layer: PROCEDURE PLMX,PLEA = PLMPRO (PLMXP,PLEI,SLA,TPAD) PLMX(1) = PLMXP * (SLA / SLC) * AFGEN(PLMTT,TPAD + 1.0) PLMX(2) = PLMXP * (SLA / SLC) * AFGEN(PLMTT,TPAD - 1.0) PLEA(1) = PLEI * AFGEN(PLETT,TPAD + 1.0) PLEA(2) = PLEI * AFGEN(PLETT,TPAD - 1.0) ENDPROCEDURE A difference between the layers is created here by calculating different temperatures. The original statements to calculate PLMX and PLEA are eliminated at this point. The distribution of leaf angles per layer must be specified. The common situation (and implicit in FUPHOT) is a spherical distribution, that is 13.4% of leaf area with angles from the horizontal to 30 degrees (first angle class), 36.6% with angles between 30 and 60 degrees (second class) and 50% with almost erect leaves (third class). A crop with erect leaves, such as the rice variety IR8, has an angle distribution of 8%, 17% and 75% respectively, averaged over all layers. A crop with an erect leaf angle distribution in the top layer and a spherical distribution in the bottom layer can be specified as: STORAGE F1(5), F2(5) TABLE Fl(1-2) = 0.08,0.134, F2(1-2) = 0.17,0.366 This defines the fraction of first angle class of the upper and of the lower layers (Fl) and that of the second angle class (F2) in the upper and the lower 44

layer. The value for the third angle class is, by definition, the complement of the first plus the second, and is not specified. The SUPHOL subroutine is included in Appendix B and is an alternative to FUPHOT. The rate of photosynthesis per leaf layer (PCGCL) is also an output of the subroutine (because there is more than one output this calculation is in the form of a subroutine rather than a function). The CSMP output statements can handle indexed variables, so that the photosynthesis per layer (PCGCL(1) and PCGCL(2)) can be printed.

Altitude Reduced leaf photosynthesis with increasing altitude is rarely considered. Yet, at higher elevations the absolute CO2 concentration is lowered (Subsection 6.1.7). The effect on photosynthesis is not identical to correspondingly lowering the CO2 concentration at sea level, because the ratio of the partial pressures of CO2 and O 2 remains the same and photorespiration is not stimulated. As a result, photosynthesis in C3 crops decreases with increasing elevation only at about half the rate as when CO2 is lowered at sea level (Figure 25). Air humidity and pollution A direct effect of low air humidity on stomata, and hence on photosynthesis, has been reported (e.g., El Sharkaway et al., 1984). It may occur in sensitive

Figure 25. The simulated rate of canopy photosynthesis under clear and overcast conditions for a C4 and a C3 crop as a function of elevation.

45

crops in very dry weather. It is further discussed and modelled in Subsection 4.1.7. Air pollution, in particular SO2, has been reported to lower leaf photosynthesis in several species (Berry & Downton, 1982; Kropff, 1987). However, it is still difficult to quantify this effect for crop models. 2.2

2.2.1

Remobilization

Introduction

The first source of carbohydrates for maintenance and growth is photosynthesis (Section 2.1); the second source is internal mobilization or redistribution. The relative importance of remobilization, synonymous with redistribution was shown in Figure 13. Redistribution permits glucose formed before flowering and stored as polysaccharides, such as starch, to enter storage organs (seeds, fruits, tubers) and can allow a high growth rate to continue for a period, in spite of low radiation levels. In simulation, this glucose can be treated in the same way as glucose from photosynthesis. Towards the end of the growing season and in conditions of acute energy shortage, remobilization may also involve protein breakdown and degradation. Protein breakdown is an active and complex process that yields amino acids as well as glucose. In calculating energy balance, however, the error made by assuming that all remobilization is starch hydrolysis is small. Efficiency and the rate of remobilization in established plants and during germination or sprouting are discussed here briefly; pre-flowering remobilization of starch is not considered. Postflowering redistribution of protein (nitrogen) can be an important determinant of the duration of the reproductive period (see Subsection 3.2.6). Accumulation of carbohydrate reserves is discussed in Subsection 3.2.4.

2.2.2

Temporary storage

A common feature among crops is that some glucose is deposited in stems or roots as starch and some of this is mobilized weeks later. At flowering, 20% or more of the weight of vegetative organs may consist of mobilizable starch, particularly in cereals. There is little data published on stem reserve contents around flowering, but Table 7 shows some indicative values. The magnitude of the fraction in any particular case is probably also dependent on weather and crop husbandry. To be mobilized starch must be hydrolyzed into glucose. This is a ‘passive’ process (i.e., it does not require additional energy). Mobilization of glucose requires only a small amount of energy (Subsection 2.4.3). The amount of glucose produced in remobilization is included in Listing 3, Line 39 and Listing 4, Line 30. 46

Table 7. The fraction of stem weight at flowering consisting of remobilizable carbohydrates (starch, sucrose plus glucose). Data are unpublished results provided by scientists at the Centre for Agrobiological Research, (CABO), Wageningen, unless indicated otherwise.

Species

Fraction

Barley Cotton Faba bean Maize Millet Potato Rice Sorghum Soya bean Sugar-cane Sunflower Sweet potato Tulip Wheat

0.3 0.1 0.45 0.35 0.1 0.2-0.4 0.25 0.2 0.18 0.5 0.1 0.35 0.1 0.4

Source estimate

estimate

Hodges et al., 1979 Hanway & Weber, 1971

Hahn & Hozyo, 1983 Benschop, 1986

Most stored carbohydrates are redistributed to the storage organs, so that regulating this process is not crucial to simulating yield. But regulation does affect growth dynamics and is therefore briefly considered. Redistribution is probably induced when, on consecutive days, the total demand for sugars exceeds the supply. A simple view is that redistribution starts once stems stop growing, and then continues at a rate of 0.1 d-1 of the redistributable starch (Listing 3 Line 35). An alternative hypothesis is that starch is remobilized when the growth rate of the developing storage organ drops below a certain level (in kg ha-1d -1 , not a relative rate). This level can be set to the highest two-day running average (Subsection 1.4.3) of the growth rate that the storage organs previously attained. After induction, remobilization proceeds at a rate of 0.2 d-1-1 , or at 0.1 d-1 in crops with a relatively long reproductive period. This level and rate are chosen without an experimental basis, but in many cases yield a reasonable pattern of stem weight loss. This hypothesis is programmed in Listing 4 Lines 42 and 45-47 of LlQ, the quarter-day time period module. (The provision that the average storage organ growth rate must be at least 10 kg ha-1 d -1 more than 47

the maximum that it previously attained, avoids triggering remobilization too early). The level of triggering and the remobilization rate could be made dependent on crop type or external conditions (such as temperature), however, this is not attempted because of the lack of basic data. Leaves lose weight during senescence. A sizeable fraction of the biomass is broken down and used for respiration or remobilization (functionally of similar value) before individual leaves die. This fraction is estimated at 0.5 (Listing 4 Lines 24, 30). It can be assumed that the same regulation holds for remobilization from leaves during senescence, as for that of starch from stems, but it probably becomes effective at a later stage. This is achieved by triggering the process when the storage organ reaches 80% of the maximum average growth level previously attained. The effective remobilization rate from leaves is estimated at 0.15 d-1 (Listing 4 Lines 40, 43). The amount of glucose that results from the breakdown of structural leaf material is affected by its carbon content (Listing 4 Line 30). It is assumed that dying roots do not contribute carbohydrate to growing points. Remobilization from leaves during senescence is ignored in the simpler program of Listing 3 (module LlD).

2.2.3

Germination

Reserves in seeded or planted material are reconverted and mobilized as glucose and amino acids in the very early stages of plant growth. It is also a key process in regrowth after cutting or ratooning. The rate of germination and early development in field conditions is difficult to simulate adequately. This is partly because environmental conditions for very young plants are difficult to assess. Crop growth modellers generally avoid this problem by initializing the simulation run at the time that 10-100 kg of dry matter has already been formed (see also Subsection 3.1.3). The thermodynamic efficiency of germination is high (Penning de Vries & van Laar, 1976), but the efficiency attained in the field is lower because the fraction remobilized from the seed is often incomplete and because organic components leak into the soil. The germination process of highly viable (more than 80%) seed is estimated to yield about 0.25 g dry weight of seedling per g dry seed in cereals, 0.35 g g-1 in seeds of leguminous species and 0.45 g g-1 in seeds rich in lipids. In this approximation, it is assumed that seedlings do not fully exhaust their seeds, because photosynthesis takes over the carbohydrate supply before that stage. It is expected that the efficiency of the sprouting process of tuber and root crops corresponds with that of cereals. Ng & Loomis (1984) and Ingram & McCloud (1984) simulated the sprouting process of potato. Bulb crops provide an extreme example. Most of the vegetative biomass is formed from carbohydrates in the motherbulb, and even in darkness a beautiful plant can grow from it. After flowering, photosynthesis provides the carbo48

hydrates to fill the new bulbs. Benschop (1986) provides an example of simulating the growth of a tulip bulb crop. 2.3

2.3.1

Maintenance

Introduction

Respiration, like photosynthesis, has been studied for almost 200 years (Steward & Bidwell, 1983). However, the regulatory mechanisms of the processes at the whole plant level are still fairly new territory. Traditionally, but inappropriately, respiration has been regarded as complex, but basically a single process. However, maintenance respiration and growth respiration are processes that occur at different rates and with their own regulation, but they have CO2 production in common (Figure 26). Quantifying the intensity of maintenance processes suffers considerably from a lack of understanding its basis, yet this process alone consumes 15-30% of the assimilates of a whole

Figure 26. A relational diagram of respiration processes in crop growth. Valve symbols indicate fluxes, the circle an intermediate variable, the rectangle a state variable and the underlined variables are input variables or constants.

49

growing season (Figure 13). Therefore, the processes and assumptions underlying simulation of maintenance respiration are discussed here more extensively than those of other carbon balance processes. Living organisms continuously use energy to maintain their current biochemical and physiological states. Respiration provides this energy. Though CO2 is only a byproduct, measuring the CO2 production rate is, nevertheless the best way to quantify maintenance respiration. Maintenance can be considered at different levels of biological organization. Maintenance is only considered at the cellular level in crop growth modelling. Maintaining the biomass of leaves and root system is regarded as a balance of separate growth and loss processes.

2.3.2

Biochemistry and regulation of maintenance

Three components of maintenance at the cellular level are distinguished: maintenance of concentration differences across membranes, maintenance of proteins and a component related to the intensity of metabolism (Penning de Vries , 1975).

Concentration differences This maintenance process is made up of activities which maintain the concentration differences of organic and inorganic ions and of neutral molecules across cell membranes. These processes keep up electrical and pH gradients and counteract spontaneous leakage. The membranes involved are those of cell organelles (mitochondria, chloroplasts) and of tonoplast and plasmalemma (enveloping the vacuole and cytoplasma, respectively). This process requires considerable energy, because though membranes only measure 8-25 nm across, the difference between both sides of the membranes can be appreciable (90 mV, 1 pH unit) and because the total membrane surface is large (typically 2-20 m2 per g dry matter). Active transport of one molecule through one membrane probably requires, on average, the energy of one ATP (irrespective of the gradient), while two molecules follow passively. Protein turnover Decomposition of some proteins and synthesis of others is continuous. Respiratory processes provide energy for this turnover. The rate of enzyme turnover varies: from almost none for the bulk of the proteins, to several times per day for some enzymes in key metabolic positions. Their overall average turnover rate has been estimated to be about 0.10 d-1 in active leaf, stem and root tissue, but is probably much smaller (deducted from data by Huffaker & Miller, 1978; Wittenbach et al., 1982; Bidwell, personal communication). Protein degradation uses an insignificant amount of energy; the cost of protein resynthesis resembles that of synthesis. 50

Metabolic activity The biochemical basis of metabolic activity, the third component of maintenance respiration, is poorly understood. The component is thought to be related directly to the overall metabolic activity of the crop (Penning de Vries, 1975; McCree & Kresovich, 1978; Amthor, 1984). The few data available indicate that its value is low in phytotron plants grown at low light. However, in field crops with high growth rates it is roughly equal to the sum of the other components and is therefore significant. Gross photosynthesis may be used as a measure of the overall metabolic activity in a crop. The third maintenance respiration component is then equal to 10-20% of daily photosynthesis (an estimate, based, as yet, on few observations). The course of CO2 production in a crop suggests that this maintenance component follows gross photosynthesis with a time coefficient of 1-2 days. The process takes place mainly in the leaves. An alternative explanation of the maintenance respiration component is that a high photosynthetic activity leads to a high glucose level in leaves, which in turn can partially uncouple ATP production from glucose oxidation, i.e., lower the efficiency of energy production (Azcon-Bieto et al., 1983). Up to three times as much glucose can then be combusted to provide the same amount of energy. CO2 production increases correspondingly. This promising hypothesis awaits direct experimental support. Rates of maintenance respiration The unit energy costs of the first and second maintenance components (concentration differences and protein turnover) are reasonably well quantified. Thermodynamic efficiency of the process is high (about 40%) if uncoupling does not occur. But the rates and regulation of those processes are insufficiently known to calculate the intensity of maintenance processes. Direct measurements of the energy required for maintenance are also unavailable. Even measuring the CO2 production rate is difficult, because this rate is low and often confounded with CO2 evolution from other processes. The best available measurements are those in which other sources of respiration are avoided as much as possible (cf. Forward, 1983; Amthor, 1984). Values of the rate of maintenance respiration of field grown leaves measured in this way are 0.03-0.08 g CO2 per g dry matter per day at 20°C (for temperate crops) or at 30 °C (for tropical crops) and less for plants grown in phytotrons. In general, the more active the tissue and the higher the nitrogen concentration, the higher the rate of maintenance respiration. Table 8 provides respiration rates for several field crops. These rates are the result of maintaining concentration differences and of protein turnover in leaves. The metabolic component of maintenance respiration is assumed to be proportional to the photosynthesis rate. The values of Table 8 may not be applicable to any specific simulation and observed rates should be obtained if possible. 51

Table 8. Rates of maintenance respiration of leaves of field crops at a reference temperature.

Species

Rate of maintenance respiration (g CO2g-1d -1)

Temperature

Barley Cotton Faba bean

0.03 0.038 0.017

23/18 30 25

Field bean

0.018 0.027

25 25

Maize (temperate)

0.032 0.026

24/18 25

Millet Rice cv IR58

0.03 0.02

25 25

Sorghum

0.01

30

Sunflower

0.029, 0.073 0.025 0.060

30 20 25

Wheat

0.016

20

Reference

(°C) Ryle et al., 1973 Amthor, 1984 Penning de Vries, 1975 (phytotron plants) Amthor, 1984 Penning de Vries, 1975 (phytotron plants) Amthor, 1984 Penning de Vries, 1975 (phytotron plants) Jansen & Gosseye, 1986 estimate based on Yoshida, 1981 McCree, 1974 (phytotron plants) Amthor, 1984 Amthor, 1984(shoot) Penning de Vries, 1975 (phytotron plants) van Keulen & Seligman, 1987

In spite of such uncertainties there are indications of differences in the rates of maintenance respiration between lines and cultivars of several crop species (e.g., Wilson, 1982; Gifford & Jenkins, 1982; Spitters, Stichting voor Plantenveredeling (SVP) personal communication). In some cases the differences are as large as 30-50% and this is of interest to plant breeders.

Non-leaf tissues The literature provides insufficient observations on the rate of maintenance respiration of non-leaf tissue to provide a list of species-specific data; values range from 0.005-0.09 g CO2 g -1 d-1 . For the roots of annual plants use 0.015 g CO2g -1d -1 (at 20 or 30 °C for temperate and tropical crops, respectively) and 52

0.010 or less for stem tissue. (The rate in young stem tissue is 1.5-2 times higher; this is ignored here for the consequences are small.) The composition of storage organs up to 1000 kg ha-1 is similar to that of young stems and has the same relative maintenance cost. All biomass above this threshold is assumed to be biochemically stable and maintenance free. Amthor (1984) lists data for maintenance respiration of roots that are generally higher than those of leaves. However, these rates include the cost of ion uptake by roots, which are included here in the cost of growth (Subsection 2.4.3).

Temperature Temperature has a direct effect on the rate of maintenance respiration. It corresponds to a doubling of the rate for each 10 °C rise in temperature (McCree, 1974) up to temperatures that will kill plants (45-60 °C). This dependence between temperature and rate of maintenance respiration corresponds with the biological concept of a Q 10 with the value of 2.0 (Listing 3 Line 67, Listing 4 Line 92). Lower and higher Q 10 values are sometimes reported (Amthor, 1984), but the value of 2.0 appears to be a reasonable average. It is assumed that the third component of maintenance respiration (metabolic activity) is not directly dependent on temperature. Air pollution might increase maintenance respiration (Berry & Downton, 1982), but the extent is not yet quantified. 2.3.3

Reduction of maintenance respiration during photosynthesis

At high light intensities the photosynthesis light response curve deviates considerably from the straight line set by the initial efficiency. Leaves then absorb more energy than can be channeled into CO2 reduction. The excess energy equals the difference between normal canopy photosynthesis and that of a canopy with a very high maximum photosynthesis rate (Figure 27). Part of the excess is used for maintenance (cf. Graham & Chapman, 1979, vol 6 p. 154; Bidwell, 1983; Amthor, 1984). However, this side benefit of photosynthesis is limited to upper leaves during the bright hours of the day, for this energy cannot be stored or exchanged between cells or organs. It is estimated that this excess energy covers half the daytime cost of the basic maintenance processes in leaves and half the cost of metabolic activity. For a whole season this side benefit of photosynthesis amounts to as much as one third of the total cost of maintaining the crop, or a yield gain of 1000-3000 kg ha-1. (It is expected that energy for nitrate reduction also comes from this source (Subsection 2.4.3). Photosynthesis appears to be more than only CO2 reduction! Figure 13 sector 5 reflects this extra benefit.) This view of a direct interaction between photosynthesis and respiration has no implications for interpreting photosynthesis light response curves or for measuring the maximum value of the curves.

53

Figure 27. Photosynthesis computed with a normal maximum leaf photosynthesis rate (35 and 70 kg CO2 ha-1 h -1 respectively) and with a very high value (1000), for a leaf area of 2 and 5 m 2 m -2, respectively. The vertical distance between corresponding lines represents the energy available for other energy-consuming processes in green cells.

2.3.4

Programming maintenance processes

Maintenance has absolute priority over growth and related energy-demanding processes. All carbohydrates required for crop maintenance are subtracted from daily photosynthesis and the remainder are left for growth (Listing 3 Line 39). In the module L1Q with a daily cycle in stored sugars, priority for maintenance processes is achieved by stopping growth processes when carbohydrate reserves drop below a certain threshold (5% of leaf dry weight), while maintenance continues (Listing 4 Lines 29, 58). The basic processes of maintenance respiration are simulated in a straightforward manner (Listing 3 Lines 61-66, Listing 4 Lines 84-91). Table 8 presents species-specific data for leaves. The metabolic component is assumed to be equal to 20% of the daily gross photosynthesis (Listing 3 Line 69) or to 20% of the value of the running average of gross photosynthesis with a time coefficient 54

of one day (Listing 4 Lines 94-95). To simulate the course of individual processes more realistically, this fraction of maintenance respiration is not subtracted directly from photosynthesis, though numerically it would be the same. Daytime maintenance respiration of leaves is estimated to be reduced by 50% due to excess energy. This is achieved in the model by multiplying with 0.75 when computing on a 24-hour basis (Listing 3 Line 63) or with 0.5 for daytime respiration only (Listing 4 Line 88). In both cases, the cost of the metabolic component is reduced by 50%. The effect of air temperature on maintenance is expressed as an exponential function without limits. The reference temperature at which the relative effect of temperature is unity is chosen to be a fixed value (see Table 8, Subsection 2.3.2) and not subjected to adaptation processes by the plant. If insufficient photosynthetic products are available to meet maintenance respiration demands, which may occur in heavy crops on very cloudy days, structural material is sacrificed to provide energy to maintain some processes at a lowered rate. Whole chloroplasts can be consumed in the process (Wittenbach et al., 1982). After a few days of energy shortage, irreparable damage occurs and many cells die. A full dynamic simulation of these processes is not yet possible. Therefore simulation is stopped by a FINISH condition when leaf maintenance respiration requires more energy than photosynthesis supplies for three consecutive days (Listing 3 Lines 46,47,114; Listing 4 Lines 65-67, 145). Negative net photosynthesis causes some structural material, particularly proteins, to breakdown. To allow this situation to occur for three consecutive days reflects that limited damage is recoverable; the length of this period is chosen arbitrarily. 2.3.5

Special cases

Wasteful respiration has been demonstrated to occur in several species. This involves a normal mitochondrial respiration, but electron pairs are led along a pathway that yields only one, instead of three, ATP. This is called uncoupled respiration. It has been suggested that this mechanism allows elimination of excess carbohydrates, such as those reaching roots, as a result of inaccurate translocation regulation (Lambers, 1979). Wasteful respiration is expected to be relatively unimportant in field crops at Production Level 1 when sink size limitation is not common (Chapter 3.1). This is because, as a rule, simulated potential yields do not exceed significantly experimental potential yields. Wasteful respiration is therefore not explicitly included in the models described here. Its eventual effect can be simulated by increasing the rate of maintenance respiration up to threefold and by slightly decreasing the conversion efficiency (Penning de Vries et al., 1974). When carbohydrates accumulate to high levels in leaves, wasteful maintenance respiration can also be expected to occur. In these cases, the models here reduce the photosynthesis rate (Subsection 2.1.4). However, stimulating maintenance respiration would 55

have been an alternative solution. There is no data on the effect on maintenance respiration of high carbohydrate levels in the storage cells of sugar-beet or sugar-cane, but it is assumed not to induce wasteful respiration.

2.4 Crop growth 2.4.1

Introduction

‘Growth’ is defined here as the biochemical conversion of reserve substances into ‘structural dry matter’. Structural dry matter consists of the organic components that remain at the end of the plant’s life, that is, are not normally broken down. In contrast, ‘reserves’ are components that only exist temporarily and are used for maintenance or growth within hours or days (available reserves) or after a few weeks (shielded reserves). Two aspects of crop growth need particular consideration here: the rate of growth of the entire crop and the efficiency of the growth process. The rate of growth respiration is related to both (Figure 26). (See Subsection 3.2.2 for calculation of the growth rate of separate organs). The crop growth rate is the multiple of the assimilates used for growth and the efficiency of the process. Efficiency is quantified in Subsection 2.4.3, and can be characterized by separate parameters for leaves, stems, roots and storage organs.

2.4.2

Rate of crop growth

Simulation models using a one-day time period of integration, usually assume that any glucose produced during a day and remaining after the day’s maintenance processes, will be used for growth. This is adequate for practical purposes. The carbohydrates available for growth processes are then related directly to daily photosynthesis (Listing 3 Line 39). Temperature does not affect this. Glucose production and consumption do not occur at the same rate over quarter-day time periods. Simulating this process requires introducing a buffering pool of reserves. All products from photosynthesis (and from remobilization, if any) are treated as if they are assembled in this pool from which all consuming processes draw. This pool of reserves is physically dispersed in the crop. Most are found in the active leaves, where reserves may account for up to 30% of the dry weight. (At these values net photosynthesis is reduced to avoid a further build up, see Subsection 2.1.4). In many cases, a starch or glucose pool is also formed in stems and roots. This pool is not available short term, but over many days or weeks. These are shielded reserves (see also Subsection 2.2.2 Temporary storage and Subsection 3.2.4 Assimilate partitioning). With time periods in the order of hours, the growth rate can be directly related to the level of available reserves (e.g. de Wit et al., 1978; Penning de 56

Vries et al., 1979; Ng & Loomis, 1984). Neither of these approaches to growth rate control is quite appropriate for the module with a quarter-day time periods. However, as simulation of the dynamics of the reserve pool is not the objective, its programming can be a compromise. Hence, the rate of use of available reserves in the pool is set at 0.85 d-1 above the lower threshold of 5% (Listing 4 Line 58; 0.85 = 1.0 – (1.0 – 1.5 · DELT)4). Both numbers are somewhat arbitrarily chosen, but yield acceptable patterns of daily fluctuations in the leaf glucose level (Figure 28), and crop yields are insensitive to these numbers. (By changing the factor 1.5 d-1 to 0.5, a crop with slow carbohydrate export is simulated; and changing it to 4.0 (= 1. / DELT) avoids any accumulation. The effect of temperature on the magnitude of this fraction may be considered). The reserve level fluctuations are somewhat exaggerated because there is no feedback between growth rate and level during the quarterday time period while, in reality, there is. The difference in structure of a module with 24-hour and quarter-day time periods is schematized in Figure 29. Implicit in this formulation is that the crop’s growth rate at Production Level 1 is almost fully source-dependent, that is, a higher rate of photosynthesis results in a higher growth rate. However, there are two important exceptions: low temperature may reduce the growth rate more than photosynthesis, and

Figure 28. The simulated course of the amount of reserve carbohydrates relative to leaf weight in wheat, in Wageningen, using quarter-day time periods.

57

Figure 29. A relational diagram of major growth processes when simulating with oneday time periods (a) and with quarter-day time periods (b). The empty circle represents the daily and complete partitioning of gross photosynthesis (PCG) over maintenance respiration and growth processes. For full names of variables see Listing 12.

there may not be enough growing points to accept the carbohydrates. In both cases, the real growth rate is below that which carbon balance permits. A pool of shielded reserves provides a buffer for carbohydrates not immediately used for growth. The quarter-day time period module distinguishes these as potential rates of carbohydrate use for growth (Listing 4 Lines 58-63) and realized growth rates (Lines 51-56). Where the growth rate limits photosynthesis rather than vice versa, care must be taken to properly quantify this constraint. A dynamic approach to simulating the sink size for carbohydrate in cereals is given in Subsection 3.2.5. Though, in general, sink size has no effect on growth, there is insufficient knowledge of non-cereal crops to simulate sink size properly. In these cases, for example, when organs are too small or too few in number to accept all carbohydrates available to them, a rough approximation must be made. This type of situation may be handled by calculating a maximum growth rate for the storage organ (Listing 4 Lines 35-36). Though 58

individual seeds of cereals often grow at a constant rate, the number of growing seeds changes. Hence, the simplest approximation is similar for all species: the maximum relative growth rate, times the actual weight of the storage organ, plus a small basic amount (to permit very small storage organs to grow relatively faster). The value of 0.35 g g-1 d -1 may be used for this maximum relative rate for temperate crops and 0.50 g g -1 d-1 for tropical crops unless better data are available. A similar reasoning and programming of sink-size limitation could be applied to other organs after pruning young leaves and growing points damaged by insects. Ng & Loomis (1984) provide an example for potato. Temperature usually has no direct effect on daily growth. However, low night temperatures may provide an exception, for reserves can then accumulate to levels high enough to reduce photosynthesis or stimulate wasteful respiration (Subsections 2.1.4, 2.3.5). This feature is an important addition for modelling spring and winter crops (cf., van Keulen & Seligman, 1987).

2.4.3

Growth efficiency and growth respiration

Biochemical research has led to a relatively accurate procedure to derive the growth efficiency and the concomitant CO2 evolution (Penning de Vries et al., 1974, 1983; McDermitt & Loomis, 1981; Forward, 1983). The approach, applicable to all species of higher plants, is summarized in Listing 1. ‘Growth’ consists of biosynthetic processes per se, that is, conversion of glucose into other organic components, plus translocation of the glucose from the source to the growth site, plus (in the case of legumes) the cost of nitrogen reduction. The weight ratio of product to substrate ranges from 0.35 to 1.0 g g -1; the lowest values apply to compounds with a high combustion heat. The efficiency of carbon use during the growth of these crops is 68-86%, while 78-86% of the combustion heat of the substrate is retained; leguminous crops score about 10% lower in both respects (these numbers are produced with Listing 1). Growth respiration is defined as the CO2 evolution resulting from growth processes. Growth efficiency and growth respiration are two aspects of the same process (see Figure 26, Subsection 2.3.1).

Biosynthesis Innumerable different organic components of structural dry matter exist. For considerations of biosynthesis efficiency, however, only five relatively uniform groups need to be distinguished: nitrogenous compounds (particularly proteins, but also nucleic acids, nucleotides, free amino acids and peptides), carbohydrates (cellulose, hemicellulose, starch), lipids (fats, fatty acids, oils), lignin and organic acids. Table 9 indicates some characteristics of these five groups. The fact that the molecules of the five groups are assembled in superstructures, such as organelles, has no implications for the cost of synthesis. 59

Table 9. Some characteristics of the five major groups of plant components and minerals.

Carbohydrates Proteins Fats Lignins Organic acids Minerals(K,Ca,P,S)

Heat of combustion (kJ g-1)

Nitrogen content

Carbon content

(g g -1 )

(g g -1)

17.3 22.7 37.7 29.9 13.9 0.0

0.0 0.151 0.0 0.0 0.0 0.0

0.451 0.532 0.774 0.690 0.375 0.0

Found particularly in cellwal1, vacuole enzymes,membranes membranes cell wall vacuole vacuole,cytoplasma

Pathways over which plants synthesize the most common compounds have been unravelled and quantified, and balance equations have been made for biosynthesis of simple and more complex organic molecules using stoicheiometry (e.g., Dagley & Nicholson, 1970). By weighing these equations according to the relative occurrence of specific molecules in its group, equations were obtained that characterize synthesis of such a group as a whole. For instance: 3.030 g glucose + 0.352 g 02 + 1.000 g fat + 1.606 g CO 2 + 0.776 g H2O This balance equation provides the amount of carbohydrates required for synthesis of fat (3.030 g g-l ) and the CO2 production factor for this process (1.606 g g-1). Although intricate for large molecules, such derivations are basically straightforward computations. Only protein synthesis is calculated in two steps: proteins are formed from amides and glucose at the growth site, and amides and glucose are formed in the photosynthesizing cells. (Amides are not explicitly considered here to avoid confusion, for the implications for growth simulation are negligible). Table 10 provides the data that characterize biosynthesis of the five groups of organic constituents (for their derivation see Penning de Vries et al. (1983), but with two corrections: ion uptake uses more energy (see below) and protein synthesis from amides and glucose was not explicit). Sensitivity analysis has shown that the variability in the amino acid constitution of different proteins results in substrate requirements that usually differ less than 5% from the average. This is probably also correct for other organic components. For consistency within the programs used here and for checking the carbon balance (Subsection 3.4.4), very precise numbers are given in Table 10. They reflect an average situation and rarely need to be adapted. The num60

Table 10. Glucose required and CO2 produced during formation of organic components (1); plus related transport (2), expressed in g per g of the product formed (1+2). Values for N reduction (3) applying to leguminous crops includes the cost of transport.

Carbohy- Proteins hydrates **

Fats

Lignins

Organic acids

Minerals

Biosynthesis(l)* glucose 1.211 CO2 0.123

1.793 0.679

3.030 1.606

2.119 0.576

0.906 – 0.045

0.000 0.000

Transport (2) glucose CO2

0.064 0.093

0.094 0.138

0.159 0.234

0.112 0.164

0.048 0.070

0.120 0.176

Growth (1+2) glucose CO2

1.275 0.216

1.887 0.817

3.189 1.840

2.231 0.740

0.954 0.025

0.120 0.176

Reduction (3) glucose CO2

0.897 1.316

* from Penning de Vries et al. (1983) ** synthesis via amides, no N reduction

bers in Table 10 are thought to be correct for all species of higher plants. Moreover, they are insensitive to the rate of growth per se and to temperature, and probably also to water stress and other environmental factors. Transport and uptake Transporting carbohydrates from sources to sinks requires at least three active steps: loading the phloem, transport within the phloem and uptake by the sink cell. The first and last processes consist of traversing cell membranes, but exactly how many is unknown. Here the lowest reasonable estimate is used. Unloading the cells is passive and loading is active across one membrane, which requires 5.3% of the energy content of transported glucose (i.e., one ATP per glucose molecule per passage). This estimate was used to compute the values of Table 10. (If the cost of transport is higher in particular cases, these numbers can be increased and new growth requirements computed). In C4 species, some transport may be passive and the cost of loading correspondingly lower (Moorby, 1981). Transporting sucrose in the phloem is an active 61

process, but the amount of energy involved is probably insignificant. The uptake of inorganic constituents from the soil solution is an active process for some ions and passive for others. It has been estimated that, on average, three monovalent ions are brought across one membrane per ATP, but it is difficult to estimate the number of membranes to be passed. Assuming the minimum of three crossings (into and out of the endodermis, and into the sink cell) and an average molecular (equivalent) weight of 40, 0.12 g glucose per g minerals is needed (Table 10). In high salt plants, this figure could be twice as high (Veen, CABO, personal communication). In a few crops (e.g., rice) silica (SiO2 ) makes up to 20% of the vegetative dry weight. It is assumed that silica is absorbed as silicate at the same cost as other minerals. It is assumed that there is no metabolic energy involved in water uptake.

Nitrogen reduction Nitrogen (N) enters plants in the form of nitrate (NO3 - ), ammonium (NH4 + ), or bi-nitrogen (N2). The first and last form need reduction before N can be assimilated. The direct cost of reduction of nitrate-N to amino-N equals 1.27 gram glucose per gram nitrate, or 0.852 gram glucose per gram protein (Table 1 in Penning de Vries et al., 1974). This glucose is actively transported. N reduction could be an important addition to the cost of protein synthesis (Table 10). However, in crops amply supplied with nutrients, the bulk of nitrate reduction occurs in the leaves during photosynthesis. Excess NADPH2 and ATP generated by chloroplasts (see also Subsection 2.3.3 and Figure 13 in the introduction to Chapter 2), is probably used for this process and N reduction comes essentially free to the crop. This assumption is implicit in the modules for simulating at Production Levels 1 and 2. Rhizobia-bacteria in nodules on roots reduce N in leguminous crops. These bacteria live symbiotically with the host plant, reducing N2 and providing amino-N to the host while receiving carbohydrates in exchange. The rate of N2 reduction by the rhizobia in good conditions can be as high as the rate of absorption of NO3 - from the soil (i.e., in the order of 10 kg ha-1d-1). The costs of reducing N2 per gram N are about the same as those for reducing NO3 - in an effective combination of rhizobium strain and host cultivar on good soils and with an appropriate water supply. The cost can be many times higher in other conditions (Minchin et al., 1981). For simulations of legume growth, the minimum reduction cost is incorporated in the computation of the carbohydrate requirement for growth. The direct cost of N2 reduction to a leguminous crop is 1000-2000 kg glucose per hectare per season. In the presence of NO3- in the soil, many legumes obtain most of their N in that form and reduce it in the leaves at no energy cost. The mechanism may not be quite clear, but from an energy balance point of view NO3- reduction in leaves is advantageous (de Visser , 1984).

62

Simpler biochemical analyses Biochemical analysis of the bulk of major organs appears to be sufficient to derive growth efficiency and growth respiration. However, performing such biochemical analyses routinely and accurately is laborious. An alternative method was developed to characterize the biochemical composition of biomass (Vertregt & Penning de Vries, 1987). This method only requires measurement of the carbon content of the dry matter (C, in g kg-1) and its ash content (A, oxidation at 550 °C, g kg-1 ); the nitrogen content (N, g kg-1 ) is determined only if the crop is expected to reduce all its N in the roots. These measurements can be taken relatively simply and quickly. The C, A and N contents, the glucose required (CRG, g g-1 dry matter) and the CO2 produced during growth (CPG, g g-1 dry matter) can be calculated directly and with ample accuracy by: CRG = ((5.39 * C + 0.80 * A + 5.64 * N - 1191) * 1.053) / 1000 CPG = (4.24 * C + 1.17 * A + 8.28 * N - 1744 + CRG * 77.7) / l000 N is to be set to 0.0 if reduction occurs in the leaves at no energy cost. (See also consistency check in Listing 5 Line 14)

2.4.4 Simulating biosynthetic processes The relative costs of forming plant components are the sum of those for synthesis, transport and N reduction (see Table 10 Subsection 2.4.3). These characteristic values can then be used in models where the five groups of organic components are explicitly distinguished (e.g., de Wit et al., 1978). From the growth efficiency viewpoint, the composition of organs often does not change significantly under optimal nutritional conditions. Typical biochemical compositions can be given for the bulk of the leaves, stems and roots of all crops, while storage organs for each crop can also be typified. (Leaf and stem tissue of rice can contain 15-20% of the dry weight in SiO2; this is an exception large enough to be accounted for.) Thus, the entire growth process can be characterized by two parameters: the carbohydrate requirement and the CO2 production factor. These parameters are assumed to be similar for the leaves of all non-legume crops, the roots of all non-legume crops, and the stems of all non-legume crops. Storage organs of crops are quite different in this respect. Table 11 gives the values for these parameters for many crops. They are produced with the program Listing 1, with the biochemical composition as input plus the parameter LEG = 1. for crops paying their own N reduction costs and LEG = 0. for all other cases. Only rarely may there be a need for the detailed approach. If there are reasons to assume that all N2 reduction occurs in roots rather than in leaves, then either conversion constants must be adapted (recalculate them with Listing 1 and set parameter LEG = l.), or another energy-consuming process must be added to the roots. The amount of glucose required for N reduction (see Table 10 Sub63

Table 11. Carbohydrate requirement (1) (g g-1 dry matter) and CO2 production (2) (g g-1 dry matter) for growth of important tissues. The carbon fraction in the dry matter is given in column (3). The numbers in column (4) are percentages of the dry weight in carbohydrates, proteins, fats, lignins, organic acids and minerals respectively.

(1) Glucose

(2) CO2

Vegetative organs non-leguminous and non-rice crops: Leaves 1.463 0.461 Stems 1.513 0.408 Roots 1.444 0.406 Vegetative organs rice crops: Leaves 1.326 0.408 1.326 Stems and roots 0.365 Vegetative organs leguminous crops: Leaves 1.687 0.790 Stems 1.603 0.540 Roots 1.534 0.537 Storage organs: Cassava, tuber 1.297 0.259 Cotton, boll 1.861 0.748 Cowpea, pod+seed 1.653 0.698 Faba bean, pod+seed 1.740 0.816 Field bean, pod+seed 1.668 0.717 Groundnut, pod+seed 2.518 1.433 Maize, cob+grain 1.491 0.384 Millet, ear+grain 1.477 0.391 0.675 Pigeon pea, pod+seed 1.652 Potato, tuber 1.285 0.274 Rice, inflor.+grain 1.462 0.357 0.377 Sorghum, ear+grain 1.473 1.238 Soya bean, pod+seed 2.161 0.263 Sugar-beet, beet 1.294 Sugar-cane, whole tops 1.478 0.392 0.719 Sunflower, inflor.+seed 1.862 0.287 Sweet potato, tuber 1.328 0.412 Tomato, fruit 1.424 0.347 Wheat, ear+grain 1.415 0.272 Yam, tuber 1.286 Source: Penning de Vries et al. (1983)

64

(3) Carbon fraction

(4) Biochemical composition

0.459 0.494 0.467

52;25;5;5;5;8 62;10;2;20;2;4 56;10;2;20;2;10

0.419 0.431

53;20;4;4;4;15 58;8;2;15;2;15

0.459 0.494 0.467

52;25;5;5;5;8 62;10;2;20;2;4 56;10;2;20;2;10

0.448 0.540 0.471 0.473 0.472 0.616 0.491 0.484 0.476 0.439 0.487 0.486 0.527 0.446 0.484 0.549 0.453 0.457 0.471 0.440

87;3;1;3;3;3 40;21;23;8;4;4 61;22;2;7;4;4 55;29;1;7;4;4 60;23;2;7;4;4 14;27;39;14;3;3 75;8;4;11;1;1 69;9;4;12;3;3 60;20;2;10;4;4 78;9;0;3;5;5 76;8;2;12;1;1 72;9;3;12;2;2 29;37;18;6;5;5 82;5;0;5;4;4 57;7;2;22;6;6 45;14;22;13;3;3 84;5;2;3;3;3 54;17;4;9;8;8 76;12;2;6;2;2 80;6;1;3;5;5

section 2.4.3) is enhanced when respiration is partially uncoupled (de Visser, 1984). The parameters for carbohydrate requirements relate carbohydrate used to biomass produced (Listing 3 Lines 28-31, Listing 4 Lines 31-35). Growth respiration is calculated in Listing 3 Lines 71-75 and Listing 4 Lines 97-101. Note that the transport-related part of growth respiration evolves in the photosynthesizing leaves, while the respiration related to biosynthesis evolves in the growing cells. Respiration due to transport of remobilized carbohydrates (Listing 3 Line 76, Listing 4 Lines 102-104) is added to growth respiration. 2.5

Exercises

Use the following data for the exercises in all chapters: — Crops: for rice (cv IR36) those of Listing 3; for soya bean (cv Hawkeye) or for potato (cv Favorita) as indicated in various Tables. Initial leaf and stem weight, development stage and rooting depth are respectively: 35., 15., 0.0 and 0.15 for soya bean and 170., 90., 0.2 and 0.15 for potato. Suppose that RMCLV is 0.03 at 20 °C for potato. Start wet season crops at DATEB = 197. and dry season crops at 349. (except rice: DATEB = 52.). — Weather: use radiation data of Los Baños, 1984 (Listing 11) plus the approximations: TABLE RAINT(1-365) = 40 * 0., 50., 159 * 0. , . . . 300., 29 * 4., 200., 29 * 0., 15 * 20., 75 * 0., 15 * 10. TABLE TPMT(1-365) = 365 * 32. TABLE TPLT(1-365) = 365 * 24. TABLE WDST( 1-365) = 365 * 1. TABLE HUAAT(1-365) = 365 * 2. Use the data for loamy soil (Listing 10) for exercises of Chapters 4 and 5. All layers have the same characteristics. For a fine sandy soil or a light clay soil, replace the water contents at saturation, field capacity, wilting point and air dry with those from Table 27. For SAWAH, use 5 layers of 0.2 m each. For sand, use soil type type 4, set EES to 30. , CSA to 0.05 and CSB to 15.. For clay, the numbers are 17, 10., 0.2 and 5., respectively.

2.5.1

Photosynthesis

T1. Make a relational diagram of the processes and variables which determine canopy photosynthesis. T2. Use the leaf photosynthesis equation of Subsection 2.1.2 for PLMX 30. and 60., and for PLEA 0.5 and 0.3. Draw the curves. As an alternative to the exponential relation, a Blackman curve and a hyperbola are sometimes used. Their equations are, respectively:

65

PL = AMINl(AL * PLEA, PLMX), and PL = PLMX * AL/(AL + ALH) ALH, the radiation intensity where PL = 0.5 * PLMX, equals 100 J m-2 s-1. Draw these curves for the same values of PLMX and PLEA. Compare the shapes of the curves. T3. Determine the impact of an increase in specific leaf weight of rice from 370 kg ha -1 by 10% on canopy photosynthesis on a clear day in Figure 22 for leaf areas of 1, 3 and 10 ha ha-1. T4. What is the maximum value that PLMX can attain in a C3 and a C4 crop, assuming a minimum diffusion resistance of 100 s m-l in both, an ambient CO2 concentration of 340 vppm and a CO2 concentration in the leaf of 136 in a C4 crop and 238 vppm in a C3 crop? What does this imply for the capacity of the diffusion process? T5. Why does the maximum rate of canopy photosynthesis of temperate regions exceed that in the tropics? Why are there two dates with maximum photosynthesis at the equator and only one in temperate zones? T6. Why is the degree of the photosynthesis reduction as a result of unequal distribution of light during a day with clouds, compared to a day with equal cloud distribution (Figure 23), not unexpected? S1. Determine the impact of an increase in specific leaf weight from 370 kg ha-1 by 10% on canopy photosynthesis on day 244 (Sept 1) for leaf areas of 1, 3 and 10 ha ha -1. (Suggestion: replace the definition of ALV and SLA by parameters with these values.) Compare these results with Figure 22 and exercise T2. S2. Determine the gain in yield of a potato crop in Los Baños when breeding shifts the photosynthesis-temperature response curve higher by 5°C. Assume planting on day 349 (Dec 15). Which process benefits most from the shift? Is the yield realistic? S3. Compare the potential rice yield of a standard IR36 crop with a spherical leaf distribution in the dry season (DATEB = 52.) with a crop in which all leaves are vertical (leaf angles 60-90°) and one in which all are horizontal (030°). Use the SUPHOL subroutine as in Subsection 2.1.4 without the temperature difference between layers and without dead leaves. What percentage of photosynthesis is provided by the second layer? S4. Suppose a disease causes rice grains to increase in weight not more than 10% d-1 (once the 10 kg ha-1 threshold is exceeded). This leads to a carbohydrate buildup in the leaves and to a large reduction in photosynthesis. How much photosynthesis would be lost during a standard season? S5. Construct a figure similar to Figure 13 for a dry season rice crop. Use the modules of Listings 3 and 5 and Appendix B. Add integrals to accumulate the specific fluxes of carbon during the season.

66

2.5.2

Remobilization

T7. Make a relational diagram of the processes and variables which determine remobilization. T8. Estimate the amount of reserves in rice, potato and soya bean crops at flowering. With how many days of photosynthesis to the amounts correspond? T9. Estimate the leaf weight of a rice, a potato and a soya bean crop after emergence and before photosynthesis plays a role, assuming that the seed rate (dry matter) is 200, 330 and 400 kg ha-1 respectively. S6. Determine the sensitivity of rice grain yield for the fraction stem reserves (0.0 to 0.5) in the L1D and L1Q modules. What is the reason for the difference in result between both programs? (Suggestion: print rates at intervals of 2.5 d for some daytime output.) What do you notice about the final stem weight? 2.5.3

Maintenance respiration

T10. Make a relational diagram of the processes and variables which determine maintenance respiration. T11. Why is maintenance respiration in leaves more intensive than in other tissues? Is the response to temperature similar in all organs? T12. Why is wasteful respiration too small to explain the large reduction in leaf photosynthesis at high sugar levels? S7. Breeding might yield soya bean varieties with a basic maintenance respiration rate in leaves which is only half of that in Table 8. How much more leaves, stems and beans (pod plus beans) would this produce, supposing all other factors remain equal (wet season crop)? And how much will be produced if the metabolic component of maintenance respiration and the rates in stems, roots and storage organs are also reduced by 50%? (Suggestion: replace constants for stem, etc., by a parameter). S8. Determine the rate of maintenance respiration of a rice crop and of its organs during the four time periods of day 95 with the module L1Q. Why are the fluctuations of rates of leaves and stems not parallel? 2.5.4

Growth rate and growth respiration

T13. Make a relational diagram of the processes and variables which determine the growth rate and growth respiration of the crop (not for organs). What other information flows could be indicated in Figure 26? T14. Some of the glucose produced on a bright day is carried over to the next day. The L1D module ignores this. Does this cause errors in simulation? T15. How should growth respiration be measured in crops and whole plants? T16. The chemical composition of rubber may be approximated by (C5H8)n. Estimate the amount of glucose required for growing one ton of rubber and the 67

concurrent CO2 production. How does rubber synthesis compare with other organic compounds? S9. What is the effect of doubling or halving the rate coefficient of reserve use on soya bean yield in Los Baños in the wet season? What do you notice about the results? S10. Suppose there is a soya bean cultivar identical to Hawkeye except that its storage organs have a biochemical composition the same as Faba beans (Table 11). How much more would it yield in Los Baños? Explain the difference. S11. Suppose another soya bean cultivar is identical to Hawkeye except that its N-fixation is inefficient and requires 10 times more energy than normal. How much would it yield in Los Baños? 2.6

2.6.1

Answers to exercises

Photosynthesis

T1. Carefully distinguish the types of variables and the relations involved. Use the symbols of Figure 5 to draw the diagram. T2. The photosynthesis light response curve calculated with the exponential function and PLMX = 30., PLEA = 0.5 is found in Figure 14. The response curve for the Blackman equation is the broken line. The hyperbolic response approaches PLMX more closely at higher light intensities than the exponential curve. T3. At a specific leaf weight of 370 kg ha-1 , PLMX equals 47.0 (Listing 5). Canopy photosynthesis on a clear day is 350, 690 and 850 kg CO2 ha-1 for leaf areas of 1, 3 and 10, respectively. With a 10% increase in leaf thickness, PLMX becomes 51.7, and photosynthesis is 380, 720 and 900 kg ha-1 respectively. T4. For a C3 crop: (340-238) = 112 vppm = 112 cm3 CO2 m-3 = 203 mg m-3. This concentration difference across a resistance of 100 s m-1 yields a leaf photosynthesis rate of 73 kg CO2 ha-1 h-1. For the C4 crop, the result is 133 kg CO2 ha-1 h-1. The diffusion process can easily keep up with the biochemical processes. T5. Maximum daily total global radiation at 50° latitude exceeds the maximum at the equator by 11 %. Moreover, this amount is received during a 34% longer daytime period. Together, this leads to a higher maximum daily total canopy photosynthesis. The radiation at the equator peaks twice a year (March 21, September 21) so that maximum canopy photosynthesis reaches a maximum also twice a year. See Figure 20. T6. The increase in canopy photosynthesis per unit increase of radiation decreases continuously (Figure 17). Splitting an amount of radiation in unequal portions over a day leads to a lower daily total. S1. Canopy photosynthesis is 275.58, 534.60 and 651.05 kg CO2 ha-1 on day 244 for an ALV of 1, 3 and 10, respectively when SLA = SLC. When SLA 68

increases 10%, photosynthesis increases to 285.70, 550.04 and 666.74 respectively. The relative increase is smaller than that in Figure 22 and exercise T2 because day 244 is not fully clear. S2. The breeding effect can be evaluated by substituting TPAD with TPAD-5 in the equations for PLMX and PLEA. The normal crop yields 15136 kg ha-1 and the adapted crop yields 22532 kg ha -1, both at Julian day 8, after more than a year of growth. The effect on PLEA is larger than that on PLMX. The example underlines the fact that the potato data set is of limited value at these high temperatures, because, among other factors, crop development is faster and because leaf and stem ageing play a more prominent role. S3. Potential rice yield for a spherical leaf distribution is 7851.2 kg ha-1 , for erect leaves 8186.3 and for horizontal leaves 6404.6. The second leaf layer contributed 23.6%, 27.2% and 10.6% to the total photosynthesis, respectively. S4. Run L1Q with PARAM GSORM = (0.5,0.1). Restricted growth reduces photosynthesis by 5044 kg CO2 ha-1 . Note the very high level of reserves in the crop and the poor yield. S5. The relative amount of carbon involved in sectors 1-6 can be obtained by: FCREM = LSTR * 1.111 * 0.947 / (PCGW * 0.682) GAC = GLV * FCLV + GST * FCST + GSO * FCSO + GRT * FCRT *fraction C in remobilized carbohydrates and *growth of actual C in the crop, respectively CS1 = INTGRL(0.,GAC * (1. - FCREM)) CS2 = INTGRL(0. ,GAC * FCREM) CS3 = INTGRL(0.,12./44. * RGCR) CS4 = INTGRL(0.,12. / 44. * RMCR) CS5 = INTGRL(0., GCS5) GCS5 = (RMMA + RMLV * 0.33) * 12. / 44. +12. / 30. * . . . (GLV * 0.15 + GST * 0.1 + GRT * 0.1 + ... GSO * 0.12) * 0.852 CS6 = INTGRL(0.,12./44. * (PCGC4 - (PCGW + GCS5)) PCGC4 = FUPHOT(70. ,PLEI,ALV, RDTM,DAT,LAT) CSUM = CSl + CS2 + CS3 + CS4 + CS5 + CS6 + 1.E-10 RS1 = CS1 / CSUM RS2 = CS2 / CSUM RS3 = CS3 / CSUM RS4 = CS4 / CSUM RS5 = CS5 / CSUM RS6 = CS6 / CSUM CS1-6 are cumulative values per sector. The pie chart of Figure 13 is constructed with RS-values at maturity. For rice, the 6 RS-values are 0.396, 0.027, 0.095, 0.160, 0.112, 0.210, respectively. 69

2.6.2

Remobilization

T7. Distinguish carefully the types of variables and relations involved. Use the symbols of Figure 5 to draw the diagram. T8. Assuming a stem biomass of 4000 kg ha-1 at flowering, the amount of starch is 1000, 720 and 800 kg ha-1 for rice, potato and soya bean, respectively. Photosynthesis is then about 600 kg CO2 ha-1d-1 , so that the reserves correspond with 2-3 days of gross photosynthesis. T9. With efficiencies as given in the text, crop growth starts with 100, 165 and 300 kg ha-1 in rice, potato and soya bean respectively. About 50% of this biomass is in leaves, 50% in roots. S6. The grain yield simulated with the L1D module is considerably enhanced by increasing the fraction of stem reserves to 0.50 (from 7691.4 to 8898.6 kg ha-1), but it decreases to 6429.1 when the stem contributes none. In the L1Q module, these yields are 7159.4, 7500.4 and 6521.3 kg ha-1, respectively. The response is smaller for the high reserve fraction because high glucose levels during remobilization reduce photosynthesis. The final weight of structural biomass of the stem changes in accordance to the stem reserves.

2.6.3 Maintenance respiration T10. Distinguish carefully the type of variables and relations involved. Use the symbols of Figure 5 to draw the diagram. T11. Leaf maintenance respiration is more intensive than stem, root and storage organ maintenance processes because the fraction of non-storage protein is largest and because the large metabolic component of maintenance respiration is located in the leaves. The response to temperature of the basic processes is characterized by a Q10 of 2.0 The overall relative response to temperature in leaves is smaller than in other organs, because the metabolic component is not directly affected by temperature. T12. Reduction of the normal ratio of 3 ATP to 1 oxygen (O) to only 1 ATP O-1 stimulates maintenance only three-fold, which is insufficient to lower leaf photosynthesis as much as is observed. S7. The potential wet season yield of soya bean is 3692.4 kg ha-1 (cf., Figure 63). With leaf maintenance respiration reduced, the yield rises to 3952.8 kg ha-1. All maintenance respiration rates reduced by 50% would increase yield to 4179.2 kg ha-1. S8. At day 95, the rates of leaf maintenance respiration are 19.0, 11.3, 14.6 and 25.2 kg CO2 ha-1d-1 at 0, 6, 12 and 18 hours respectively. Stem maintenance is then 8.3, 10.0, 12.9 and 11.3. The values for the roots are 2.6, 3.1, 3.9 and 3.4. The fluctuations of the rates are not parallel because maintenance respiration in stems and roots is only affected by temperature, while the large metabolic component in leaves has its own dynamics. (Hint: use the END

70

CONTINUE statement at day 94 and reset the PRDEL to 0.25 to suppress excessive output).

2.6.4

Growth rate and growth respiration

T13. Distinguish carefully the type of variables and relations involved. Use the symbols of Figure 5 to draw the diagram. In Figure 26, the photosynthesis rate affects the rate of maintenance respiration (in L1D directly, in L1Q via PCGDV); temperature affects maintenance; photosynthesis in L1D and the reserve level in L1Q affect the growth rate and growth respiration rate. T14. Too much biomass is formed on the bright day simulated with L1D. If this occurs in the vegetative period, then too much leaf area is simulated to develop and the second day starts with too much leaf area. However, the error is small unless in reality crop growth was severely limited. In these cases, use the L1Q module. The error is always negligible after the period when leaves are formed. Growth rates simulated with the L1D module fluctuate more than is expected to occur in the real crop. T15. Most growth respiration evolves in the growing points. These are usually difficult to access or include properly in equipment. The rate also varies during the day, making a rate for any particular moment difficult to measure and to interpret. It is advisable to estimate the respiration rate of the entire plant or crop over 24 hours, and to subtract measured or estimated maintenance respiration (cf., McCree, 1974). Respiration of a growing fruit or tuber is easier to measure (Schapendonk & Brouwer, 1984). T16. Applying the equations in Subsection 2.4.3 shows that 3.754 ton glucose is required and 2.289 ton of CO2 is released. Rubber is even more expensive to produce than fat, assuming that the biochemical pathways involved are not exceptional from the point of view of energy efficiency. S9. The Soya bean yield goes up from 3676.6 kg ha-1 to 3708.0 when the rate coefficient is 0.75 and up to 3775.8 when it is 3.0. The effect on yield is small, but in the biomass at flowering is it large. A reduced rate in the carbohydrate supply leads to a lower growth rate in the vegetative mass. Photosynthesis after flowering is not greatly affected, but a larger mass requires more carbohydrates for maintenance. S10. The Soya-Faba-bean would yield 4274.1 kg ha-1, 828.2 kg ha-1 more than the soya bean. The lower fat and protein content of Faba bean mean that more weight can be produced from the same amount of carbohydrates. S11. The yield would be only 987.6 kg ha-1 in Los Baños due to poor growth. Most of the difference in biomass with respect to the good crop develops during the pod filling period.

71

3

Morphological development and assimilate partitioning

This chapter discusses simulation of the morphological development of crops. Complete models for crop growth at Production Level 1 can be constructed by combining these processes with the assimilation and dissimilation processes of Chapter 2. Listing 3 (basic module L1D for crop growth) and Listing 4 (crop growth module L1Q with quarter-day time periods) provide two complete modules for these processes. There is less known about the mechanisms of morphological processes and their regulation in crops, in comparison with assimilation and dissimilation. There are also more differences between species and cultivars from a morphological, than from a physiological point of view. 3.1

3.1.1

Crop development

Introduction

Many changes occur as a crop grows. Some changes, such as those of weight and leaf area, are easy to quantify, while others, such as plant age and phenological development, are more difficult. Nevertheless, it is essential to quantify crop phenology because the important process of partitioning of new biomass depends directly on this expression of age. The ‘development stage’ of a crop quantifies its physiological age and is related to its morphological appearance. Development stage is a state variable in crop growth models. The development stage cannot be expressed simply as chronological age, because several environmental factors, such as temperature and waterstress, can speed up or reduce the rate of phenological development. Daylength is crucial in some crops to induce flowering. Contrary to what is suggested by intuition, the rate of crop growth per se has no effect on the rate of phenological development, as long as the growth rate is not very low. The concept of development stage is used to characterize the whole crop; it is not appropriate for individual organs. The development stage has the value of 0.0 at emergence, 1.0 at anthesis and 2.0 at maturation. It is dimensionless and its value increases gradually. The development rate has the dimension d-l. The multiple of rate and time period yields an increment in stage. Figure 30 shows development stages of a rice crop. The rate of phenological development can be affected by temperature differently in the vegetative stage than in the reproductive stage. Daylength has an effect only in the vegetative stage. These differences indicate that the physiological process of development is not the same before and after anthesis. 73

Figure 30. Development stages of a rice crop (Source: Stansel, 1975).

Physiological or biochemical methods of characterizing and measuring the development stage of the crop are yet unknown. Phenological development is still not understood enough to provide an explanatory model of this process, hence, descriptive modelling is used here. Plants in a field do not flower simultaneously. The date of anthesis refers to the first date that 50% of the fertile tillers carry or have carried open flowers. 74

3.1.2

Vegetative phase

A first approximation of the development rate in the vegetative stage is the inverse value of the duration of the period between emergence and anthesis. This value is equal to the development rate constant, when temperature has been constant and daylength has had no effect. Table 12 gives typical values of this constant for different crops and for some important cultivars. The effects of temperature and daylength on the development rate of several species are summarized from the literature and presented in Tables 13 and 14. The extensive Handbook of Flowering (Halevy, 1985) provides a review of relevant literature and contains many data. Roberts & Summerfield (1987) analyzed the effects of temperature and daylength on the duration of flowering of several crops. Temperature is often the dominant factor influencing plant development in temperate climates. The development stage can then be expressed as a temperature sum (degree.days, sum of average temperatures above a lower threshold, e.g. Vanderlip & Arkin (1977), Warrington & Kanemasu (1983), van Heemst (1986)). This assumes aproportionality between temperature and development rate, which is easy to grasp, but has limited validity. Here, the more flexible description of a non-linear relation of development rate with temperature is preferred (Roberts & Summerfield, 1987). Figure 31 provides an example. The relative effect of temperature can be expressed as a multiplication factor (Listing 3 Lines 97, 99, Listing 4 Lines 125, 127). It is difficult to extract data on the temperature-development rate relationship for different crops from the literature and only a few are presented in Table 13. Moreover, responses among cultivars can differ considerably. Modellers must therefore often make their own approximations. As long as daylength does not influence development, a reasonable approximation of the development rate constant and the effect of temperature on it can be obtained from field data plus the intuitive knowledge of crop experts. The temperature that affects the phenological development process can be taken as equal to the daily average air temperature at the height of the shoot’s growing point. Only when day or night temperatures regularly reach values where the response is non-linear, is another procedure of weighing temperatures required or should shorter time periods be taken. The temperature of the growing point can be higher or lower than the air temperature at two metres due to insolation, transpiration from the growing point and heat transfer from the soil. This topic is relatively unimportant and solutions may be species specific. Daylength is important for photoperiod-sensitive crops that are common in the tropics. Subsection 6.2.2 gives a calculation of the photoperiodic daylength. Long days speed up the development rate in long-day plants, but reduce it in short-day plants. Some plants require a certain minimum or maximum night length before flowering is triggered, but in most cultivars phenolog75

Table 12. The rate of crop development in the vegetative stage at a reference temperature and reference daylength.

Species

Barley Cassava Cotton cv BarLXl Cowpea cv TVu1188 Faba bean cv Minica Groundnut cv Robut 33-1 Maize, grain cv XL45 cv Pioneer Maize, silage,cv LG11 Millet, early late Potato cv Mara (late) cv Favorita (early) Rice cv Nipponbare Sorghum CSH6 Soya bean cv Hawkeye cv Jupiter Sugar-beet Sugar-cane Sunflower cv Sobrid cv Relax Sweet potato Tulip cv Apeldoorn Wheat, winter cv Arminda cv UQ189 Wheat,spring cv Miriam

Rate (d-1)

Tempe- Daylength rature (°C) (h)

0.032 0.017 0.016 0.022 0.033 0.030

19 23 30 23 20 25

0.033

25

0.025 0.0265 0.016 0.015 0.010 0.035 0.0555 0.014 0.020 0.020 0.025 0.024 0.0069 0.025 0.013 0.0195 0.006 0.010

28 25 25 30 30 18 18 25 28 30 23 27 20 25 22 18 27 15

12 12 12 10 10 13 ? 12.6 16 12.5 13 16 13 -

Warrington & Kanemasu, 1983 Sibma,CABO, pers.comm. Sibma,CABO, pers.comm. Jansen & Gosseye, 1986

0.015 0.0195

20 20

13 14

de Vos,CABO, pers.comm. Angus et al., 1981

0.020

25

14

de Vos,CABO pers.comm.

16 12.5 12 12 12 -

Source: Halevy (1985), unless otherwise specified - daylength not relevant ? unknown

76

Reference

Grashoff,CABO, pers.comm. Saxena et al., 1983

van Heemst, 1986 van Heemst, 1986 Horie, 1987 Huda et al., 1984 Patron, pers.comm (vernalization important)

unpublished results Hahn & Hozyo, 1983 Benschop, 1986

Table 13. The ratio of the development rate in the vegetative phase at a certain temperature to that rate at reference temperature (Table 12) for different crops. Data in CSMP FUNCTION style: temperature first and the multiplication factor next.

Species

Temperature response

Barley Cassava Cotton Cowpea cv TVu1188

4.,0.01, 19.,1.0, 24.,1.0, 30.,0.01 15.,1., 30.,1. 12.,0.01, 30.,1.0, 40.,1.0 19.,0.77, 23.,1., 27.,1.32 (use smallest values from Tables 13, 14 at any date). –10.,0.01, 5.,0.01, 12.,0.5, 20.,1.0, 30.,1.2 –10.,0.01, 0.,0.01, 20.,0.9, 25.,1.0, 35.,1.2 6.,0.01, 20.,1., 32.,1.2, 40.,1.2 8.,0.01, 14.,0.25, 19.,0.65, 28.,1.0, 35.,0.9 0.,0.3, 10.,0.3, 15.,0.75, 25.,1., 35.,1.2 10.,0.5, 20.,1.0, 30.,1.2, 40.,1.2 (Jansen & Gosseye, 1986) 7.,0.01, 18.,1.0, 29.,0.01 (van Heemst, 1986) 10.,0.1, 19.,0.8, 25.,1., 27.,1.1, 32.,1.2, 40.,1.0 16.,0.25, 20.,0.6, 24.,0.85, 28.,1., 32.,1. (Horie, 1987) 7.,0.01, 30.,1.0 (Huda et al., 1984) 0.,0.01, 10.,0.5, 20.,0.9, 27.,1.0, 30.,1.1, 40.,1.2 0.,0.01, 5.,0.5, 20.,1., 25.,0.5 10.,0.01, 19.,0.01, 23.,1.0, 27.,1., 32.,0.01, 50.,0.01

Faba bean cv Minica Groundnut Maize cv XL45 cv Pioneer,LG11 Millet Potato Rice cv Nipponbare Sorghum CSH6 Soya bean Sugar-beet Sugar-cane Sunflower cv Sobrid cv Relax Sweet potato Tulip cv Apeldoorn Wheat, winter cv’s UQ189, Arminda Wheat, spring cv Miriam

7.,0.01, 12.5,0.62, 22.,1.0, 35.,1.3 10.,0.3, 15.,0.75, 18.,1.0, 25.,1.1, 35.,1.2 10.,0.1, 16.,0.5, 27.,1.0, 36.,0.8 –10.,0.01, 0.,0.2, 5.,0.4, 10.,0.7, 15.,1.0, 20.,1.4, 25.,2.0 (Benschop, 1986) 2.,0.0, 10.,0.68, 15.,0.89, 20.,1.0, 25.,1.03, 30.,1.05, 35.,1.06, (Angus et al, 1981) –10.,0.01, 0.,0.01, 20.,0.9, 25.,1.0, 35.,1.2

Source: based on Halevy (1985), unless otherwise specified

77

Table 14. Effect of daylength on the development rate of sensitive cultivars, expressed by the acceleration relative to the reference daylength (Table 12). Data in CSMP FUNCTION style: daylength (h) first and the relative multiplication factor second. The type of sensitivity indicated in the second column: LDP = long day plant, SDP = short day plant, IDP = intermediate daylength plant, DNP = daylength neutral plant.

Species

Type

Sensitivity

0.,0.1, 11.,0.1, 16.,1., 20.,1.2, 24.,1.2 strong effect but highly variable among cv’s 0.1., 12.,1., 12.5,0.5, 14.,0.1, 24.,0.1 0.,1.4, 10.7,1.4, 11.7,1.2, 13.3,0.9,... 15.,0.6, 24.,0.6 (use smallest values from Tables 13, 14 at any date) cv TVu1188 DNP,SDP 0.1., 12.,1., 13.,0.75, 17.,0.5, 24.,0.1 Faba bean almost none Groundnut SDP,DNP 12.,1., 14.,0.95, 16.,0.9 Maize cv XL45 (Warrington & Kanemasu, 1983) SDP,DNP 0.,1.0, 12.,1.0, 12.5,0.9, 13.,0.6, 24.,0.6 Millet almost none Potato DNP,SDP 8.,1., 12.,1., 13.,0.5, 14.,0.33, 24.,0.33 Rice (Vergara & Chang, 1985; sensitive only in cv BPI76 period DS 0.2-0.7, otherwise always 1.0) 0.,1.0, 10.,1.0, 12.,0.9, 14.,0.7, 16 ., 0.01, ... cv Nipponbare 24.,0.01 (Horie, 1987) SDP,DNP 0.,1.0, 13.6,1.0, 14.3,0.65,24.,0.1 Sorghum cv CSL (Huda et al., 1984) 12.6,1.0, 13.,0.85, 13.5,0.75, 13.6,0.5 Soya bean cv Jupiter SDP (Patron, personal communication) LDP 12.,1.0, 16.,1.0, 24.,10. Sugar-beet (large differences among cv’s) IDP 0.,0.1, 11.5,0.1, 12.,0.5, 12.5,1., 13.5,0.5 ,... Sugar-cane 13.5,0.5, 14. ,0.1, 24. ,0.1 (when days shorten and plenty water; otherwise 0.1) SDP,DNP 0.,1.0, 13.,1.0, 17.,0.9, 24.,0.9 Sunflower cv Sobrid SDP 10.,1.25, 13.,1., 17.,0.1, 24.,0.1 Sweet potato almost none Tulip SDP,DNP 10.,0.29, 11.,0.55, 12.,0.75, 13.,0.89,... Wheat cv UQ189 14.,1.0, 15.,1.08, 16.,1.14, 17.,1.18 (Angus et al., 1981)

Barley Cassava Cotton Cowpea

LDP LDP SDP,DNP SDP,DNP

Source: based on Halevy (1985), unless otherwise indicated

78

Figure 31. The response curve of the development rate to temperature for the wheat cultivar QU189. The full line represents the relationship before flowering, the dashed line the relationship after flowering. Values are relative to that at 25 °C. (derived from Angus et al., 1981).

ical development is a continuous process slowed by unfavourable daylengths. An example for a moderately sensitive wheat cultivar is shown in Figure 32. Daylength effect can also be translated into a multiplication factor of the development rate (Robertson, 1968). Ng & Loomis (1984) use this approach to simulate the photoperiodic induction of potato tubers. Roberts & Summerfield (1987) consider the effect of daylength on the development rate should be added to, rather than multiplied by, that of temperature. Given the inaccuracy of the data, it is difficult to judge which assumption is most appropriate. Sensitivity to daylength can change during the vegetative phase. This is well documented for rice (Table 14; Vergara & Chang, 1985). When plants are only sensitive between certain development stages (such as 0.2 and 0.7 for rice) DRED in Listing 3 Line 97 or in Listing 4 Line 125 can be replaced by: INSW((DS – 0.2) * (0.7 – DS), l., DRED) (note that the daylength effect on rice is exerted over a fraction of the vegetative period, so that its values should be derived correspondingly). A list of daylength sensitivity in crop species is given in Table 14. In spite of much research, neither the mechanism of photoperiodicity, nor the reasonable data base needed for modelling the effects of daylength exists. The degree of sensitivity is characteristic for a variety, and can be modified or suppressed by breeding. Many modern crop varieties are bred to have little, or no sensitivity to daylength to allow easier manipulation in cropping systems and over larger 79

Figure 32. The response curve of crop development rate to daylength for wheat cultivar QU189. Values are relative to that at a daylength of 14 hours. (derived from Angus et al., 1981).

areas. However, sensitivity may be a desirable trait for crops in areas where the planting date can vary a lot, but the harvest date must be constant. Modellers must be aware of the different effects of daylength among individual crop varieties, so that new observations are required for new varieties. Sugar-beet and sugar-cane are harvested before they reach the flowering stage. In these crops photoperiodicity can be strong. Although the development stage can still be expressed as a fraction of the time from emergence to anthesis, quantifying the development rate is more difficult. It may then be estimated that the development stage at harvest is close to 1.0. An alternative is to relate the development stage to the number of leaves formed (as effected by Loomis et al., (1979) for sugar-beet), or simply to the temperature sum.

3.1.3

Initialization

The development stage begins at 0.0 when simulation starts at seedling emergence. Often, however, simulation starts when young plants are already well established (Subsections 2.2.3 and 3.3.3), using observed or assumed quantities of leaves and roots as initial values. Values of 0.1-0.25 for the initial development stage of field crops are common, and as high as 0.5 for transplanted rice. The effect of transplanting in rice can be approximated by reducing its development stage by 0.2. For winter wheat in a temperate climate, simulation may start in spring at development stage 0.33. No generally valid initial values can be given, because they depend strongly on the experimental 80

situation and on local management practices. In tuber and bulb crops, seeding material has already undergone a certain physiological ageing before planting. A normal development stage for a potato crop at emergence is about 0.2. An estimate of the initial development stage in a specific situation can be made with the development rate constant (Table 12), the response to temperature (Table 13) and the actual temperatures. It can also be obtained as a ‘temperature sum’ in cool climates (e.g., Gupta et al., 1984). Its calculation is basically straightforward, but requires good temperature data of the soil at the appropriate depth. Vernalization is not considered here. It is implicitly assumed that crops requiring low temperatures before flowering is induced, such as winter cereals, have been sufficiently exposed to cold.

3.1.4

Reproductive phase

The reproductive period is defined here as the period after flowering until maturity. Simulating this development process proceeds in the same manner as that of the vegetative period. The development rate constant for the reproductive period and for the vegetative period are numerically different, as is the effect of temperature. Daylength has no effect. Some data are presented in Tables 15 and 16; note that such data are difficult to obtain and not very precise; they are also specific to cultivars within species, although less so than in the vegetative phase. If specific data are lacking, it can be assumed that there is no effect from temperature (i.e., duration of the reproductive phase is fixed) or that the effect from temperature is the same as in the vegetative phase. This simulation of the development process contains no explanatory mechanism for the crop ripening. Simulation is halted by imposing an end when the development stage reaches the value of 2.0, by including the statement FINISH DS = 2.0, or FINISH DS = 0.95 for sugar crops and for sweet potato that are harvested before flowering. In the module TIL (Listing 2) for tillering and grain formation, simulation is ended when grains reach their maximum weight (cf., Subsection 3.2.5). Crops that continue to produce branches plus leaves and flowers, while fruits or seeds are being filled are called indeterminate crops (Subsection 3.2.2). Many cultivars are fully or largely indeterminate particularly among leguminous crops and cotton. Reproductive and vegetative growth are parallel as long as conditions remain favourable. This situation can be described as a very slow progress of the development stage after flowering.

81

Table 15. The development of crops after flowering at a reference temperature for each crops.

Species

Rate (d-1)

Barley cv Grit 0.021 Cotton cv BarLXl 0.01 Cowpea 0.03 Faba bean cv Minica 0.0185 Groundnut cv Robut 33-1 0.012 Maize cv Pioneer 0.017 cv LG11 0.021 Millet 0.024 Potato cv Mara (late) 0.015 cv Favorita (early) 0.0225 Rice 0.038-0.046 Sorghum cv CSH6 0.050 Soya bean cv Hawkeye 0.014 cv Jupiter 0.027 Sunflower cv Relax 0.028 Tulip cv Apeldoorn 0.010 Wheat, winter cv UQl89 0.0275 cv Arminda 0.0255 Spring cv Miriam 0.020

Temperature (°C) 16 ? ? 18

Reference

Grashoff, CABO, pers. comm. Saxena et al., 1983

25 25 25

Sibma, CABO, pers.comm. Sibma, CABO, pers.comm. Jansen & Gosseye, 1986

18 18 28 27

van Heemst, 1986 van Heemst, 1986

27 28 18 15 20 20 25

Huda et al., 1984

Patron, pers.comm. Benschop, 1986 Angus et al., 1981 de Vos, CABO, pers.comm. van Keulen, CABO, pers.comm

– = not relevant ? = unknown

3.2

3.2.1

Assimilate partitioning

Introduction

Assimilate partitioning is the process by which assimilates available for growth are allocated to leaves, stems, roots and storage organs. Though only part of the total biomass is harvested, all components are important for allocation of new dry matter even before the economic products are formed. New biomass invested in leaves gives a quick high return from photosynthesized 82

Table 16. The ratio of the development rate after flowering at a certain temperature to that rate at reference temperature (Table 15) for different crops.

Species

Temperature response

Barley Chickpea Faba bean Groundnut Maize Millet Potato Rice

16.,1.0 21.,1.0, 37.,2.0 0.,0.0, 18.,1.0, 25.,1.23, 35.,1.5 0.,0.0, 10.,0.5, 21.,1.0, 30.,1.2, 40.,1.2 as in Table 13 for Pioneer & LG11 25.,1., 35.,1. 7.,0.0, 18.,1. 29.,0.0 10.,0.45, 19.,0.75, 25.,0.9, 28.,1.0,... 30.,1.1, 40.,1.1 7.,0.0, 27.,1.0, 47.,0.0 (Huda et al., 1984) 27.,1.0, 35.,1.0 0.,0.3, 10.,0.3, 15.,0.75, 25.,1., 35.,1.2 O.,O., 20.,0.9, 23.,1., 32.,1.2 as in Table 13 cv’s UQl89 (Angus et al., 198l), Arminda 10.,0.14, 15.,0.66, 20.,1.0, 25.,1.23,... 30.,1.4, 35.,1.5 cv Miriam as in Table 13

Sorghum Soya bean Sunflower Sweet potato Tulip Wheat, winter,

Wheat, spring,

products, whereas biomass invested in roots and stems gives a slower, more indirect return (when there is no water or nutrient shortage). Crop growth and development, discussed in Sections 2.4 and 3.1, should not be confused with distribution of new biomass over plant organs for these are different processes. The distribution pattern is a function of physiological age. The form and number of leaves, stems, roots and storage organs are not generally considered at this level of crop growth modelling. The ‘storage organ’ includes the economically valuable product and its hull or supporting tissue; the relative weights of these parts differ greatly between crops (Table 17). There is little quantitative information on the internal control of carbohydrate distribution in crops, but there are interesting explanatory approaches to this problem (see e.g., Horie et al., 1979; Cock et al., 1979; Dayan et al., 1981; Vos et al., 1982; Ng & Loomis, 1984; van Keulen & Seligman, 1987). There are now some submodels but these are still too specific to implement them into the summary models here. The models here, therefore, rely on quantifications that are, at best, halfway between descriptive and explanatory. A starting point here is the assumption that the growth rate is basically 83

Table 17. Major economic components of storage organs of different crops. Farm yields refer (in most cases) to products with a non-zero moisture content.

Crop

Storage organ

Major component of dry weight

Cassava Cotton Cowpea Faba bean Field bean Groundnut Maize Millet Pigeon pea Potato Rice

tuber bolls pods pods pods pods cobs ears pods tuber panicle

Sorghum Soya bean Sugar-beet Sugar-cane Sunflower Sweet potato Wheat

panicle pods beet millable cane heads tuber ears

tuber 100% lint 35%, seed 65% seed 75-85% seed 75-85% seed 75-85% seed 60-75% grain 70% grain 60% seed 75-85% tuber 100% rough rice l00%, paddy 80%, polished rice 72% seed 70-75% seed 60-80% beet 100% sugar 9-13% seed 44% tuber 100% grain 85%

Source: Penning de Vries et al. (1983)

source-dependent: the more carbohydrates supplied, the faster is growth. Temporary storage of starch around flowering occurs in many crops and is considered in Subsection 3.2.4. Two ways of simulating assimilate partitioning are presented. The first method (Subsection 3.2.2) is appropriate for determinate crops. With modification, this method of biomass distribution also approximates partitioning in indeterminate crops. The second method, specifically referring to cassava, uses a principle probably applicable to most crops, though to a smaller extent (Subsections 3.2.3 and 3.2.4). Simulating the formation of sink size in rice is discussed in Subsection 3.2.5. Loss of biomass during senescence and by root exudation are briefly considered in Subsections 3.2.6 and 3.2.7.

84

3.2.2

Allocation of new biomass in source-limited crops.

When simulating the allocation of new biomass in determinate crops a distribution key is used. Carbohydrates available for growth in any time period are partitioned over organs according to this key, independent of the available amount of carbohydrate. The key changes slowly with the development stage of the crop. Generally, the pattern is such that the largest share is initially attributed to leaves and roots, then to stems, and ultimately to the storage organ. Figure 33 shows a typical distribution pattern for rice. Because the sum of the fractions of the shoot is 1.0 by definition, a cumulative presentation is easier to read (Figure 34). Assimilate distribution is shown for soya bean as an example of a leguminous crop (Figure 35) and potato as an example of a tuber crop (Figure 36). The potato example shows why late-planted seed potatoes, in which the phenological development is already well advanced, immediately produce tubers and have small leaves and stems. The idea of a fixed dry matter distribution pattern has been used by different authors (cf., Vanderlip & Arkin, 1977; van Keulen et al., 1982; van Heemst, 1986). In this model, glucose rather than biomass is partitioned to organs, as this is closer to reality. Note that a certain pattern of carbohydrate allocation does not translate into a predetermined distribution of dry matter at harvest time and into a fixed harvest index. Data on this distribution pattern are crucial

Figure 33. The pattern of carbohydrate partitioning to the organs in rice as a function of development stage. The fraction root is the fraction from the total crop growth; the sum of the fractions for growth of leaf, stem and panicle plus grain always equals 1.00.

85

Figure 34. The pattern of carbohydrate partitioning to the organs in rice as a function of development stage. Data from Figure 33; shoot fractions are presented cumulatively.

Figure 35. The cumulative pattern of carbohydrate partitioning in soya bean as a function of development stage.

86

Figure 36. The cumulative pattern of carbohydrate partitioning in potato as a function of development stage.

inputs to simulation models and deserve proper attention. It is practical to first distinguish the distribution between shoot and root, and then to distinguish leaves, stems and storage organs as shoot subfractions. (The advantage of this two-step procedure becomes more evident when water stress is simulated: Subsection 4.3.3). ‘Stems’ are defined in a functional rather than a morphological manner and include stem proper, leaf sheaths and stem-like petioles. Below-ground storage organs, such as tubers and beets, are treated as part of the shoot. Programming of this method of carbohydrate partitioning is straightforward (Listing 3 Lines 40-44, Listing 4 Lines 59-63). Simulating the allocation of assimilates applies equally well to crops with a fixed development pattern, such as cereal crops, many legume crops and also to bulb crops, such as tulips. A growth phase of only leaves, stems and roots is followed by a period in which these organs and the reproductive or storage organ grow together. In the final phase only the storage or reproductive organ increases in weight. The middle phase is short in cereals and can be long in legumes. The time at which the reproductive or storage organs begin to grow often coincides with flowering, but is directly related only in cereals. In other crops the relation is indirect; potato tubers are induced before flowering, legume pods start to fill weeks after the first flowers appear, and in sugar-beet the flowering stage should not be reached before harvest. Data on carbohydrate allocation patterns, such as those of Figures 34-36, are obtained from field crops with a series of harvests and crop growth analy87

ses. Data on dead and removed material should be added to the biomass data of the standing crop. Table 18 presents a number of such patterns, derived from the patterns of increase in biomass (i.e., the derivative to time of the weight of live and dead biomass), accounting for the efficiency with which glucose is converted into structural dry matter (Table 11). Note that this analysis applies to the growth process of organs; death or biomass removal are governed by other factors and must be treated separately (Subsection 3.2.6). Significant differences in allocation patterns can exist between cultivars as Table 18 shows for maize, potato, rice and wheat and it is important to establish these patterns for the simulated cultivar whenever possible. Van Heemst (1986) found that the pattern of the fraction of assimilates going to leaves occurred about 0.1 unit of development stage later in a late potato cultivar, than in an early one and that the fraction for stems was 0.2 later. The overall pattern remained the same (but there is a large difference in their development rates, Subsection 3.1.2, Table 12). Van Heemst also showed that the lines can shift a little in response to management practices, but that chemical growth retardants had little or no effect on the allocation pattern. The direct effects of environmental conditions on the allocation pattern of new biomass are probably small (van Heemst, 1986) and are not considered here. The effect of water stress is discussed in Subsection 4.3.3. A lowered shoot-root ratio is often reported for growth at unfavourably low temperatures, and the indirect effect (through the development rate) of actions or circumstances that increase or decrease soil and canopy temperatures on partitioning can be considerable. Direct effects of moderate nutrient shortage on carbohydrate allocation patterns are probably small. The descriptive basis of the carbohydrate allocation patterns implies that this approach may be inappropriate in conditions in which a deviating morphological development of the crop can be expected, for example, after severe pruning. In indeterminate crops, growth continues as long as environmental conditions remain favourable. Physiological aging slows down and vegetative and reproductive organs grow simultaneously. Existing parts age, but new branches or tillers are formed and rejuvenate the crop. The pattern of assimilate allocation then remains stable. This may be approximated by reducing the rate of phenological development to a low value (Subsection 3.1.4). 3.2.3

Allocation of new biomass in cassava

Another method of regulating dry matter allocation has been described and modelled for cassava by Cock et al. (1979). Cassava is a tropical crop that grows in areas with periodic droughts and takes 8-20 months to mature. Stems and leaves are produced all the time, and storage roots, the economic product, are formed once the plants are a few months old. Starch can make up 80% or more of the storage roots. The protein-rich leaves are also sometimes eaten. 88

Table 18. Distribution keys for glucose allocation to leaves (CALVT), to stems (CASTT) and to the entire shoot, including storage organs (CASST), as a function of the development stage in healthy crops of different species. Data are in CSMP style: in each pair, development stage is the first number and the fraction of assimilates the second.

Barley FUNCTION CALVT = 0.,0.82, 0.25,0.70, 0.51,0.55, 0.6,0.50,... 0.72,0.23, 0.83,0.01, 0.95,0.0, 2.1,0.0 FUNCTION CASTT = 0.,0.18, 0.25,0.30, 0.51,0.45, 0.6,0.50,... 0.72,0.77, 0.83,0.99, 0.95,1.0, 1.21,0.0, 2.1,0.0 FUNCTION CASST = 0.,0.35, 0.51,0.45, 0.72,0.85, 0.95,l.0,... 2.1, 1.0 Cotton FUNCTION CALVT = 0.,0.59, 0.32,0.63, 0.55,0.56, 0.77,0.45,... 1., 0.37, 1.12, 0.09, 1.24, 0.28, 1.47, 0.02, 1.6, 0. 0, 2.1, 0.0 FUNCTION CASTT = 0.,0.41, 0.32,0.37, 0.55,0.44, 0.77,0.55, 1.,0.61,... 1.12, 0.48, 1.24, 0.58, 1.47, 0.31, 1.6, 0.43, 2.1, 0.36 FUNCTION CASST = 0.,0.5, 0.77,0.75, l.,l.0, 2.1,l.0 Cowpea FUNCTION CALVT = 0.,0.81, 0.16,0.80, 0.39,0.85, 0.61,0.83,... 0.86,0.64, 1.06,0.50, 1.45,0.30, 1.73,0.26, 2.0,0.07, 2.1,0. FUNCTION CASTT = 0.,0.19, 0.16,0.20, 0.39,0.15, 0.61,0.17,... 0.86,0.36, 1.06,0.49, 1.45,0.57, 1.73,0.48, 2.0,0.22, 2.1,0. FUNCTION CASST = 0.,0.50, 0.16,0.63, 0.39,0.76, 0.61,0.80,... 1.06,0.85, 2.1,0.85 Faba bean FUNCTION CALVT = 0.,0.5, 0.54,0.60, 1.,0.25, 1.2,0.01, 2.1,0. FUNCTION CASTT = 0.,0.5, 0.54,0.40, 1.,0.50, 1.2,0.25,... 1.42,0.14, 1.51,0.01, 1.71,0., 2.1,0. FUNCTION CASST = 0.,0.5, 0.54,0.7,1.,0.8,1.2,1.,2.1,l. Groundnut FUNCTION CALVT = 0.,0.63, 0.47,0.56, 1.02,0.58, 1.25,0.52,... 1.48,0.30, 1.7,0.07, 2.1,0.0 FUNCTION CASTT = 0.,0.37, 0.47,0.44, 1.02,0.42, 1.25,0.4,... 1.48,0.22, 1.7,0.07, 2.1,0.07 FUNCTION CASST = 0.,0.4, 0.47,0.70, 1.25,0.95, 1.5,0.99, 2.1,l.0 Maize (silage maize, cv LG11) FUNCTION CALVT = 0.,0.66, 0.47,0.69, 0.56,0.66, 0.65,0.58,... 0.82,0.29, 1.,0.0, 2.1,0.0 FUNCTION CASTT = 0.,0,34, 0.47,0.31, 0.56,0.34, 0.65,0.42,... 0.82,0.71, 1.,0.70, 1.16,0.0, 2.1,0.0

89

FUNCTION CASST = 0.,0.5, 0.51,0.75, 1.24,1.0, 2.1,l.0 Maize (grain maize, cv Pioneer) FUNCTION CALVT = 0.,0.49, 0.35,0.59, 0.67,0.20, 1.0,0.12,... 1.18,0.09, 1.37,0.0, 2.1,0.0 FUNCTION CASTT = 0.,0.51, 0.35,0.41, 0.67,0.66, 1.0,0.64 ,... 1.18,0.31, 1.37,0.0, 2.1,0.0 FUNCTION CASST = 0.,0.5, 0.67,0.75, 1.37,1.0, 2.1,1.0 Millet FUNCTION CALVT = 0.,0.7, 0.26,0.7, 0.40,0.67, 0.58,0.64,... 0.7,0.60, 0.84,0.45, 1.,0.28, 1.24,0.05, 1.5,0.0, 2.1,0.0 FUNCTION CASTT = 0.,0.3, 0.26,0.3, 0.40,0.33, 0.58,0.36,... 0.7,0.40, 0.84,0.50, 1.,0.61, 1.24,0.50, 1.5,0.0, 2.1,0.0 FUNCTION CASST = 0.,0.5, 0.25,0.5, 0.50,0.75, l.,l.0, 2.1,l.0 Potato, cv Favorita (early) FUNCTIONCALVT = 0.,0.77, 0.37,0.67, 0.83,0.39, 1.14,0.l0,... 1.35,0.0, 2.1,0.0 FUNCTION CASTT = 0.,0.16, 0.37.0.15, 0.83,0.16, 1.14,0.15,... 1.35,0.02, 1.57,0.0, 2.1,0.0 FUNCTION CASST = 0. ,0.5, l.,l.0, 2.1,l.0 Potato, cv Mara (late) FUNCTION CALVT = 0.,0.80, 0.15,0.80, 0.46,0.65, 0.76,0.38,... 1.03,0.05, 1.17,0.0, 2.1,0. FUNCTION CASTT = 0.,0.20, 0.15,0.20, 0.46,0.23, 0.76,0.22,... 1.03,0.22, 1.17,0.0, 2.1,0. FUNCTION CASST = 0.,0.5, 1.,1.0, 2.1,l.0 Rice, IR36. See Listing 5, Chapter 7. Rice, IR64 FUNCTION CALVT = 0.,0.577, 0.515,0.577, 0.625,0.520,... 0.71,0.409, 0.82,0.278, 0.995,0.0, 2.5,0.0 FUNCTION CASTT = 0.,0.423, 0.515,0.423, 0.625,0.480,... 0.71,0.591, 0.82,0.722, 0.995,1.0, 1.0,0.39, 1.25,0.09,... 1.4,0.0, 2.5,0.0 FUNCTION CASST = 0.,0.862, 0.515,0.862, 0.625,0.844,... 0.71,0.844, 0.82,0.940, 1.25,1.0, 2.5,l.0 Sorghum FUNCTION CALVT = 0.,0.55, 0.4,0.55, 0.6,0.56, 0.74,0.56,... 0.86,0.06, 1.,0.07, 1.2,0.05, 1.4,0.01, 1.6,0., 2.1,0. FUNCTION CASTT = 0.,0.45, 0.4,0.45, 0.6,0.44, 0.74,0.42,... 0.86,0.78, 1.,0.56, 1.2,0.24, 1.4,0.04, 1.8,0.0, 2.1,0.0 FUNCTION CASST = 0.,0.5, 0.4,0.7, 0.6,0.8, 1.3,1.0, 2.1,l.0

90

Soya bean FUNCTION CALVT =

0.,0.71, 0.25,0.61, 0.5,0.65, 0.75,0.85,... 1.,0.70, 1.2,0.54, 1.4,0.32, 1.6,0.18, 1.8,0., 2.1,0. FUNCTION CASTT = 0.,0.29, 0.25,0.39, 0.5,0.35, 0.75,0.15,... 1.,0.3, 1.2,0.33, 1.4,0.18, 1.6,0., 2.1,0. FUNCTION CASST = 0.,0.5, 0.5,0.7, 1.,0.8, 1.2,1., 2.1,l. Sugar-beet FUNCTION CALVT = 0.0,0.7, 0.35,0.36, 0.52,0.08,0.7,0.07,... 0.86,0.05, l.l,0.0 FUNCTION CASTTT = 0.0,0.3,0.35,0.32, 0.52,0.23, 0.7,0.20,... 0.86,0.05, l.l,0.0 FUNCTION CASST = 0.,0.6, 0.35,0.9, 0.52,1.0, 1.1,l.0 Sunflower cv Relax FUNCTION CALVT = 0.,0.74, 0.42,0.68, 0.51,0.50, 0.57,0.40,... 0.66,0.35, 0.73,0.30, 0.82,0.23, 0.90,0.20, 1.,0.14,... 1.28,0.05, 1.54,0.0, 2.1,0.0 FUNCTION CASTT = 0.,0.26, 0.42,0.32, 0.51,0.50, 0.57,0.60,... 0.66,0.65, 0.73,0.67, 0.82,0.69, 0.90,0.66, 1.,0.48,... 1.28,0.10, 1.54,0.10, 2.1,0.0 FUNCTION CASST = 0.,0.5, 0.73,0.75, 1.,1.0, 2.1,1.0 Sweet potato FUNCTION CALVT = 0.,0.66, 0.22,0.63, 0.49,0.59, 0.69,0.31,... 1.,0.07, 1.22,0.03, 1.35,0.01, 2.1,0.0 FUNCTION CASTT = 0.,0.34, 0.22,0.37, 0.49,0.40, 0.69,0.38,... 1.,0.24, 1.22,0.42, 1.35,0.11, 2.1,0.0 FUNCTION CASST = 0.,0.5, 1.,1.0, 2.1,l.0 Wheat, winter FUNCTION CALVT = 0.,0.90, 0.33,0.85, 0.43,0.83, 0.53,0.75,... 0.62,0.56, 0.77,0.20, 0.95,0.09, 1.14,0.05, 1.38,0., 2.1,0. FUNCTION CASTT = 0.,0.10, 0.33,0.15, 0.43,0.17, 0.53,0.25,... 0.62,0.44, 0.77,0.80, 0.95,0.64, 1.14,0.62, 1.38,0., 2.1,0. FUNCTION CASST = 0.,0.5, 0.33,0.5, 0.53,0.75, l.,l., 2.1,l. Wheat, spring FUNCTION CALVT = 0.,0.90, 0.19,0.83, 0.26,0.85, 0.45,0.82,... 0.6,0.32, 0.86,0.15, 1.,0.26, 1.26,0.0, 2.1,0.0 FUNCTION CASTT = 0.,0.10, 0.19,0.17, 0.26,0.15, 0.45,0.18,... 0.6,0.68, 0.86,0.85, 1.,0.34, 1.26,0.27, 1.5,0.0, 2.1,0. FUNCTION CASST = 0.,0.4, 0.32,0.5, 0.6.0.75, l.,l.0, 2.1,l.0

91

Cock et al. suggest that cassava leaves and stems usually grow at a rate below that permitted by the carbohydrate supply. The rate of leaf and stem growth is not source-limited, but sink-limited. The growth rate of leaves plus branches is proportional to leaf area (Figure 37). ‘Excess’ carbohydrates are stored as starch in storage roots as a reserve for times of energy shortage (‘shielded reserves’). The accessibility of these reserves is much lower than that of glucose in the leaves which are called ‘available reserves’ (Subsection 2.4.2). (The relations of the processes in which reserves are involved are illustrated in Figure 29 part b Subsection 2.4.2). Cock’s formulation implies that the growth rates of leaves and stems are constant during much of the growing season, while that of shielded reserves fluctuates considerably with weather and crop conditions. These shielded reserves provide a buffer which the crop uses to regrow leaves after severe insect attacks. To simulate this type of assimilate distribution, the rate of leaf and stem growth must be computed independently of the carbohydrate supply. Cock’s model computes the rate of leaf production and growth per leaf, relates stem growth to this, and derives the amount of carbohydrate involved. The excess

Figure 37. The growth rate of shoots and storage roots in cassava as a function of leaf area (derived from Cock et al., 1979).

92

photosynthate goes to the storage roots. The model realistically simulates dry matter distribution. 3.2.4

Formation of shielded reserves

The storage behaviour of cassava is not unique, for storing starch or sucrose occurs to a certain extent in many species. However, it is usually noticed less because it is a temporary phenomenon that is over at harvest time (except for sugar-cane and sugar-beet). Carbohydrates are commonly stored in stems or roots for some weeks, particularly when young storage organs are insufficiently developed to handle the total flow of assimilates. A sizeable amount of shielded and available reserves can then be built up in vegetative tissues. Reserve buildup may be to such an extent that it may temporarily or even permanently hamper canopy photosynthesis (Subsection 2.1.4). Van Heemst (1986) reported the growth rate of a late potato cultivar which appeared to be somewhat lower than that of an early cultivar, possibly due to the lack of a sufficiently large sink size when tubers were initiated. A permanent reduction of photosynthesis of this nature is not included in the model; it could be approximated by making PLMX a state variable, the value of PLMX decreasing when the reserve level is too high. A simple way to deal with the formation of shielded reserves is to assume that a certain fraction of the increase in stem weight will be available for redistribution after flowering (Listing 3 Lines 17, 35). This fraction is assumed to consist only of starch. Some data on the magnitude of the remobilizable fraction are given Table 7 (see Subsection 2.2.2). Formation of shielded reserves can be simply simulated by adding those carbohydrates that growing organs cannot absorb to the fraction of the stem weight that consists of starch. This may continue until the maximum level (estimate: Table 17 + 10%) is reached. This assumption is programmed in Listing 4 Lines 32-33, 37-38; where the complex calculation of the growth of stem weight accounts for the different amounts of glucose required to produce one kilogram of starch or stem. The sink size of leaves, stems and roots in a vegetative crop is usually large enough to accept all carbohydrates provided. But young storage organs may not have the capacity to grow; though they may be large in number, they have too small a sink size. 3.2.5

Modelling morphological development in rice

Carbohydrate production in cereals can be limited by the capacity of the grains to use them (Cock & Yoshida, 1973; Evans & Wardlaw, 1976). In the case of rice this is easy to understand, as the size of grains is physically restricted by the size of the hull. The maximum size of the grain is a variety-specific characteristic. TO simulate this effect, the ‘sink size’ of the grains must be quantified (i.e., their capacity to absorb available carbohydrates). It is then 93

essential to keep track of the number of grains. Grain setting is the end result of a series of events, so that the processes of tillering and floret formation must be considered. Tillers dying is only approximated as a dynamic simulation of age groups with different light interceptions is not justified. A module for morphological development in rice is presented to simulate the phenomenon that is often referred to as the sink-source relationship (Listing 2, module TIL to simulate development of tillers, florets and grains in rice). This approach, developed by van Keulen & Seligman (1987) for wheat also facilitates modelling damage by pests and diseases. In the module TIL, the formation rate of plant parts, such as tillers, florets and grains, is assumed to depend on the net carbohydrate supply to the crop. The larger this supply, the higher the organ formation rates, so long as their numbers do not exceed certain limits. Formation rates of organs, in numbers per hectare per day, are equal to the difference between potential number and current number, divided by an appropriate time constant (Lines 7, 16, 23). The time coefficient is 15 days for tiller formation, 7 days for floret formation, and 3 days for grain initiation. The time coefficient for tillers dying is set at 14 days (Line 8), assuming that tillers only die slowly and when carbohydrates are lacking. Each type of organ forms during a restricted developmental period. The potential numbers of these plant parts at any moment, equals the carbohydrate supply of that day, divided by the daily requirement of carbohydrates for forming and maintaining one tiller, one floret and one grain. The carbohydrate required for florets and grains is a constant. The older the plant and the larger most of the early tillers, the more carbohydrates are required to initiate new tillers. This effect is mimicked by making the carbohydrate required for initiating new tillers a function of the development stage of the crop (Line 12). The number of tillers that will be formed in a rice crop in a specific simulation depends on environmental conditions. The maximum number of grains equals the number of florets (Line 25). The sink size of the storage organ equals the number of grains multiplied by the maximum growth rate of one grain (Line 32). The maximum growth rate per individual grain is estimated to be the weight of mature kernels of that variety, divided by half the grain filling period. Hence, if sufficient grains have been formed sink size will rarely limit growth. The module itself is straightforward. Acronyms are explained in Listing 12, Chapter 7. The data for this module are derived from field experiments with the rice varieties IR36 and IR64; data may be different in varieties with other tillering characteristics or panicle structures. Simulation stops when the grain weight reaches its maximum value (Line 42). The TIL module (Listing 2) interacts with the main module L1Q (Listing 4) by providing the maximum growth rate for storage organs, while using as inputs the carbohydrate supply for growth and the morphological development rate. The initial weight of leaves and planting density are related. When adding the module TIL to the crop growth model L1Q, Line 36 (Listing 4) has to be deleted. It is not appropriate 94

to combine this module with the simpler model of Listing 3 (see Table 1 in the Reader's guide).

3.2.6 Senescence and death Senescence refers to the loss of capacity to carry out essential physiological processes and to the loss of biomass. It is particularly important in the case of leaves, for even at Production Level 1 senescence occurs towards maturation. The fundamental processes involve physiological ageing and protein (enzyme) breakdown. These processes are difficult to quantify. Hormones are important as messengers, but it is not known how (de Wit & Penning de Vries, 1983). In addition, nutrient remobilization, in particular nitrogen, often plays a crucial role. If senescence is not very important for the crop or for the research objectives, the simple descriptive approach in which the relative rate of loss is a function of the development stage can be used (Listing 3 Lines 33-34). Listing 5 contains a numerical example that corresponds with observations on a rice crop in the Philippines (functions LLVT and LRTT). Description such as this usually results in a loss of 40-60% of leaf area at harvest time. Loss of absorbing roots is handled in the same descriptive manner. These numbers can be used as default values, but should be calibrated to mimick specific situations. High temperatures accelerate senescence. This environmental effect can be easily added to the program provided that data are available to quantify it. The contribution of senescing leaves and roots to the pool of reserves is disregarded here. A more mechanistic approach to senescence is by setting the death rate of leaves and roots to a certain fraction per day once the conditions for growth deteriorate; except for their reserves, stems do not lose weight. Deteriorating conditions are defined as a drop in the growth rate of the storage organ below a previously attained level (Listing 4 Line 43) (see also Subsection 2.4.2). Other definitions of deteriorating conditions, or of values for relative death rates are also possible. Leaves that drop from the plant have generally lost some weight, even without diseases. This indicates that some biomass was used before the leaves died either for respiration or remobilization (proteins). Both processes increase the amount of carbohydrates available for growth. It is assumed that the leaves that dropped contributed half their original weight to the carbohydrates pool (Listing 4 Lines 24, 30). When the biomass from live leaves becomes smaller, the specific leaf weight, the maximum leaf photosynthesis and canopy photosynthesis consequently decrease. This favours remobilization, so that senescence is characterized by positive feedback. The crop dies when photosynthesis becomes lower than leaf maintenance on several consecutive days (Subsection 2.3.4). Sinclair & de Wit (1976) argued that the process of nitrogen and carbohydrate redistribution 95

from leaves is the major cause of senescence in leguminous crops. Cassava leaves have a fixed life span of about 80 days. The so-called boxcar train method (Rabbinge et al., 1989) can be used to mimick this type of leaf ageing. 3.2.7

Absorbing roots and excretion

The absorbing roots of annual crops grow almost exclusively in the vegetative stage and stop when the storage organ starts to gain weight. (Storage roots are treated as part of the shoot, Subsection 3.2.2.). At initialization, root weight is often taken to be equal to shoot weight. The root mass has a dry weight of 500-2000 kg ha-1 around flowering. For lack of good field data, partitioning of assimilates between root and shoot is carried out so that the above mentioned pattern and values are obtained. Though this may not be quite accurate, root mass is usually only one tenth or less of the final above-ground biomass at harvest, so errors do not cause many problems. Proper data to simulate the growth of root systems are difficult to obtain. The effect of water stress on root growth is discussed in Subsection 4.3.3. Excretion of organic substances into the soil by the roots occurs through exudating and sloughing off of root tips. Estimates for excretion run from 0-20% of the assimilated carbon (Woldendorp, 1978). This process is not included in the programs here, but could be incorporated as a fixed fraction of root growth or of gross photosynthesis (just as for maintenance related to metabolic activity, Subsection 2.3.2.). The crop carbon balance check should then be adjusted accordingly (Subsection 3.4.4).

3.3 Leaf area

3.3.1

Introduction

The leaf area of a crop is usually expressed as the total surface of live leaves, one-sided, per unit of soil surface. It is given in ha ha -1 or m2 m-2 and often called leaf area index. The leaf area of an established crop has a value of 3-6 ha ha -1, or even up to 10 ha ha-l in very dense canopies. More than 80% of light is intercepted when the leaf area reaches 3 ha ha-1, and 5 ha ha-1 in a canopy with erect leaves, which is called a closed canopy situation (see Figure 21 Subsection 2.1.3). The amount of leaves and the rate at which leaves are formed at the start of the growing season are of considerable importance for the final yield. This makes partitioning of new biomass between leaves and other organs important. Growth of leaf weight is computed in earlier Subsections (2.4.2, 3.2.2). It must now be determined with how much leaf area it corresponds. ‘Specific leaf weight’ is used for this purpose. This is defined as the dry weight of leaves (no reserves, only structural dry matter) with a total one-sided leaf area of one 96

hectare. Petioles and leaf sheaths around the stems are not included. The specific leaf weight for individual leaves ranges from 200 to 800 kg ha-1, though the average for entire canopies is rarely more than 600 kg ha-1 . The growth of leaf area is related here to growth in leaf weight. The specific leaf weight of new leaves may change with crop age. Growth of leaf area can also be simulated independently of leaf weight (e.g. Johnson & Thornley, 1983). Both approaches yield a pattern of leaf area development that is approximately realistic provided that parameter values are properly chosen. A general and explanatory simulation of the development of the leaf area of different crops does not yet exist; crop-specific models have been constructed by Cock et al. (1979), Horie et al., (1980), and Ng & Loomis (1984). A descriptive rather than an explanatory approach cannot be avoided. All leaf area is usually treated as being equally effective in photosynthesis and transpiration if exposed to the same conditions. When using the SUPHOL subroutine, however, leaf layers can have different characteristics (Subsection 2.1.4).

3.3.2

Thin and thick leaves

When leaves are formed in a young crop, a certain area of relatively thin leaves seem to be more effective for quick growth than only half the area with leaves twice as thick. However, if the maximum rate of leaf photosynthesis increases proportionally with leaf thickness (as discussed in Subsection 2.1.2), this compensates for the lack of a large leaf area. Consider a leaf biomass of 2000 kg ha-1 with a leaf surface of 3 ha of thick leaves, or 5 ha of normal leaves, or 8 ha of thin leaves. Each thickness corresponds with a maximum rate of leaf photosynthesis. The canopy photosynthesis rates that correspond with these combinations lie on a hyperbola when plotted as a function of leaf photosynthesis and leaf area. Computation of canopy photosynthesis on fully clear days for other amounts of biomass in which leaf thickness is varied, shows similar curves. Compensation of area by thickness appears to be almost complete, because the canopy photosynthesis isolines follow almost rectangular hyperbolas (Figure 38). In full sunlight there is no difference in whether the crop invests biomass in thin or thick leaves; only very thick leaves are disadvantageous. However, on overcast days greater leaf area is always more effective than thicker leaves. Factors other than radiation level also affect the importance of thin versus thick leaves, such as shading competitors (making thin leaves more advantageous) and water use efficiency (making a minimal area advantageous). This calls for a general pattern of thin leaves early in the life of the crop and thick leaves later. Crops in humid climates often show this phenomenon distinctly, but crops in semi-arid regions less so, or not at all. Observations indicate that there is a tendency for leaves formed early to be thinner than later ones, but the extent varies a great deal between crops. For example, Sibma (1987) re97

Figure 38. Combinations of leaf area and maximum rate of leaf photosynthesis that lead to a certain fraction of the maximum daily total of canopy photosynthesis (indicated next to the lines). Isolines are shown for fully clear skies (full lines) and overcast skies (dashed lines). The amount of photosynthesis is expressed relative to the maximum of 1244 kg CO2 ha-1 d-1 in clear, and to 349 in overcast situations. The values are computed for a canopy with ALV = 10., PLMX = 100., PLEA = 0.4 and at LAT = 50. and DATE = 166.

ported that the specific leaf weight of the first leaves of maize was 200 kg ha-1 or less and of the last leaves almost four times as much (Figure 39). On the other hand, cassava in a dry climate produces leaves with a fairly constant specific leaf weight. Leaves of plants grown indoors usually have thinner leaves than field plants, because they are grown at a much lower light intensity.

3.3.3

Modelling growth of leaf area

Several approximations can be used to simulate growth of leaf area. The simplest is to assume that the specific leaf weight is a crop characteristic and that it is constant in time and throughout the canopy. This value is called the specific leaf weight constant. Leaf area is determined by dividing the weight of live leaves by the specific leaf weight (e.g., van Keulen et al., 1982). In many 98

Figure 39. The specific leaf weight at different levels in a maize canopy during a growing season in Wageningen (Source: Sibma, 1987).

cases, this is a fair approximation provided that its value is measured at a proper time, such as at the end of the phase when most assimilates go to the leaves (e.g., development stage 0.5 for rice, Figure 34). Table 19 presents values of specific leaf weight constants for different crops at about this stage. A more realistic approximation of development of leaf area takes into account that new leaves formed early in the life of the plant are thinner than leaves formed later. The specific leaf weight of new leaves is then found by multiplying the specific leaf weight constant with a factor that depends on the development stage of the crop (Listing 3 Line 91, Listing 4 Line 119). Figure 40 gives an example of the relation between specific leaf weight and development stage. Insufficient data were found in the literature to derive more than a few of these crop specific relations (Table 20). Descriptive functions such as these should be used carefully and checked whenever possible. A low specific leaf weight at the beginning of leaf growth speeds up growth of leaf area. Hence this method of simulating leaf area development enhances the simulated date of canopy closure by several days (or even as much as two weeks), in comparison to using a constant value of the specific leaf weight (Figure 41). However, the effect on leaf biomass and on final yield tends to be small. The average specific leaf area of the crop is used to derive the standard 99

Table 19. The specific leaf weight constants for different crops are average values for the entire canopy. Leaf refers to leaf blades (one surface only) excluding petioles and leaf sheaths. The specific stem weight is an average for the growing season and includes stem proper, petioles, branches and leaf sheaths.

Species

Specific leaf weight (kg ha-1)

Specific stem weight (kg ha-1)

Barley Cassava Cotton Cowpea Faba bean Groundnut Maize Millet Potato Rice (IR36) Sorghum

325 450 490 450 315 600 450 440 300 440 400

625 2000 2000 2114 2000 2775 1200 2000 1000 2500

Soya bean

400

2100

Sugar-beet Sugar-cane Sunflower Tulip Wheat, winter spring

500 700 540 710 425 500

1900 7825 2900 1080 1080

Reference

Hearn, 1969 Grashoff et al., 1987 Bhagsari & Brown, 1976 Sibma, 1987 Jansen & Gosseye, 1986

Sivakumar et al., 1979; McCree, 1983 Dornhoff & Shibbles, 1970 Hanway & Weber, 1971

Rawson & Constable, 1980 Benschop, 1986

Values without reference were obtained from colleagues at CABO, Wageningen. maximum rate of leaf photosynthesis that serves as an input in computing canopy photosynthesis. The fact that thicker leaves are usually at the top of the canopy and thinner leaves at the bottom has only small implications for canopy photosynthesis and is disregarded here. Different growing conditions, such as those caused by different plant densities or fertilization level in maize (Sibma, 1987), and by irrigation in potato (Ng & Loomis, 1984), have little effect on specific leaf weight. The influence of these environmental factors may also be small in other crops; they are disregarded here. The change in the specific weight of new leaves with development stage may be a reflection of the change in carbohydrate supply to 100

Figure 40. The specific weight of new leaf area as a function of crop development stage in maize (Source: Sibma, 1987).

growing tissue (cf de Wit et al., 1970). The rate of leaf area loss is computed in direct relation to the rate of leaf weight loss, assuming that the average value of the specific leaf weight applies (Listing 3 Line 89, Listing 4 Line 117). Van Keulen & Seligman (1987) calculated the rate of leaf area loss in wheat independently of leaf weight loss. They put it at 5% d-1 once the leaf area exceeds the value of 6 m2 m-2 to account for mutual shading. The constant life span of cassava leaves (Subsection 3.2.6) is also reported to be due to young leaves that are produced at a constant rate shading old leaves (Cock et al., 1979). Simulation of leaf area for a specific case sometimes appears to be unrealistic and disturbs progress in a study. It is then advisable to cut the positive feed back loop (leaf weight – leaf area – photosynthesis – growth – leaf weight) and introduce the observed (or a chosen) leaf area development as a forcing function.

101

Table 20. The specific leaf weight relative to the values of Table 19, as a function of development stage. Of each pair of values, the first is the development stage, the second the fraction.

Barley FUNCTION SLT = 0.,l., 0.51,1.05, 0.60,0.71, 0.72,1.41, 2.1,l. Cotton FUNCTION SLT = 0.,0.85, 0.32,0.85, 0.55,1.04, 0.77,0.95,... 1.,1.27, 1.12.1.25, 1.24,1.22, 2.1,1.22 Faba bean FUNCTION SLT = 0.,l.l, 0.54,1., 2.1,1. Groundnut FUNCTION SLT = 0.,l., 0.47,1.05, 1.02,0.85, 1.25,1.29,... 1.48,0.95, 1.70,0.93, 2.1,0.93 Maize (cv.LG11) FUNCTION SLT = 0.,0.6, 0.5,1., 2.1,1.2 Maize (cv.Pioneer) FUNCTIONSLT = 0.,0.5, 0.5,0.9, l.,l., 1.5,1.1, 2.1,1.2 Millet FUNCTION SLT = 0.,0.64, 1.,1., 2.1,l. Potato FUNCTION SLT = 0.,1.2, 0.07,1.2, 0.9,0.83, l.0,l.0, 2.1,l.0 Rice FUNCTION SLT = 0.,0.5, 0.5,1., l.71.2, 2.1,1.2 Sorghum FUNCTION SLT = 0.,0.6, 0.4,0.8, 0.6,1.3, 1.,1.1, 2.1,1. Soya bean FUNCTIONSLT = 0.,0.8, 1.,0.8, 1.2,1., 1.5,1.1, 1.8,1.25,... 2.1,1.25 Sugar-beet FUNCTION SLT = 0.,1.0, 0.35,1.1, 0.52,1.2, 0.7,1.2,... 0.86,1.3, 1.1,1.3 Sunflower FUNCTION SLT = 0.,l., 0.42,1., 0.51,1.48, 0.57,0.79, 0.66,0.73,... 0.73,0.9, 0.82,1.39, 0.9,1.32, 1.,1.53, 1.16,1., 2.1,l. Wheat, winter FUNCTION SLT = 0.,1., 0.33,1.1, 0.36,1.06, 0.43,1.5,... 0.53,1.05, 0.62,1., 0.77,0.85, 0.95,1.07, 1.14,1., 2.1,l. Wheat, spring FUNCTION SLT = 0.,0.67, 0.55,0.67, 0.6,1., 2.1,1. Source: see Table 19

102

Figure 41. Development of leaf area (dashed lines) and grain weight (full lines) of a maize crop simulated with a specific leaf weight increasing as in Figure 40 (o) and simulated with a fixed value of 450 kg ha-1 (x).

3.3.4

Green surfaces other than leaves

The overall contribution of organs other than leaves to gross photosynthesis of the canopy can be positive or negative. It is positive when additional photosynthesis outweighs additional shading, as is the case in very open crops and when most leaf area has already died. The net contribution of organs other than leaves is negative when the light absorbed by the non-leaf surface is used less efficiently than it would have been by leaves. Non-leaf photosynthesis is often only a minor addition to gross photosynthesis and may then be disregarded. However, the contribution of stems to the total green area can be significant for small grains and is therefore included in both crop modules (Listing 3 Lines 87, 90, Listing 4 Lines 115, 118). The effective surface area of stems is determined by dividing their weight by a 'specific 103

stem weight’. This constant has a value varying from 625 to 8000 kg ha-1 depending on species (see Table 19 Subsection 3.3.3). The factor 0.5 is added to the area calculation because, unlike for leaves, only the upper surface is active. The specific stem weight is not given as a function of crop development, but this relation may be introduced if the need arises. It is assumed that stem area has the same photosynthesis properties as leaf area. Fruits usually intercept less than a few percent of radiation. However, sunflower provides an interesting exception. Its large heads all point in the same direction and intercept 10-20% of radiation, but barely photosynthesize. It can be simulated that a modified sunflower crop with flowers low in the canopy would yield about 10% more than the existing crop. The contribution of green non-leaf area to photosynthesis can be evaluated with the programs presented. If stems are in the same positions with respect to radiation as leaves, then their contribution is proportional to their share in the green area. If the distributions of the angles of leaf and non-leaf areas are quite different, and, if these areas are at different positions in the canopy, the contributions of non-leaf area to photosynthesis can be evaluated with the SUPHOL subroutine (Subsection 2.1.4). There is no reason to treat products resulting from photosynthesis by stems or fruits any differently from those of leaves, for the assimilates are transported in the same basic forms and are not formed in their destined cells. Hence all growth costs are the same. 3.4

Two modules to simulate potential crop growth

3.4.1 Introduction The background of different processes of crop growth and suggestions for their programming have been given in Chapters 2 and 3. The major components are combined in two modules (L1D, Listing 3 and L1Q, Listing 4) which can be used to simulate the growth of annual crops. The first module is basically the sum of the simplest approaches. The time period for this module is one day (24 hours) and since it simulates at Production Level 1, it is called L1D. The second module (L1Q) includes some of the more detailed approaches and its time period is a quarter of a day. L1D and L1Q must be supplemented with data sets for a crop and weather. Data to typify crops are given in many of the preceding Tables. Listing 5 uses rice as an example of how crop data can be added; Listing 11 (Chapter 7) shows this for weather data. Not all crop characteristics in this data set are required for all modules, but including more data than is required causes no problems. The simulation program is to be completed with module T12 (Appendix B), which contains several special functions. Figure 12 in Subsection 1.4.4 illustrates how sections and data of Chapters 2 and 3 may be combined into a full program. Table 1 in the Reader’s guide shows how sections discussed through104

out the book can be combined. The simulation language CSMP is used throughout and references are provided in Subsection 1.4.1. FORTRAN is used in the functions and subroutines. Though, in the main, the programs presented are straightforward or explained in the text, a few peculiarities are discussed in this Section. Line numbers in listings are used only for identification and are not part of the CSMP program. Lines starting with an asterisk (*) contain comment. All names of variables are explained in Listing 12, Chapter 7). Biological and physical constants and precisely known biological data are given in the listings with three or more significant digits. Approximate values are given with one or two digits. The modules contain a number of constants. They are not combined as each represents a single process: 1.467 stands for g CO2 produced per g of glucose, 0.682 is its inverse; 0.053 refers to the fraction of glucose sacrificed during intercellular transport to provide energy for this process, 0.947 is its complement; 1.111 represents the yield in glucose from starch hydrolysis, 0.900 is its inverse; 0.2727, 0.400, and 0.444 are the carbon fractions in CO2, glucose and starch respectively. The small value 1.E-10 is added in some cases to avoid division by 0.0, which would halt the simulation. The function AINT is used to truncate values; the output of the function AMOD is equal to its first input except when it exceeds the second input; the second input is then subtracted a number of times until a value between zero and the second input remains. Data in CSMP AFGEN functions should cover a range of values that is wider than the range in which inputs are expected. This ensures that extrapolation outside the data, causing unexpected results or irrelevant warnings, does not occur. This is important when using data from Tables 13 and 14 (Subsection 3.1.2), Table 16 (Subsection 3.1.4) and others. 3.4.2

Basic crop growth module with one-day time periods (L1D)

Module L1D (Listing 3) can be used when growth limited by sink size does not need to be considered and when environmental conditions for crop growth are favourable. Hence, it will often be appropriate. L1D should not be used when environmental conditions are unfavourable (e.g., when day or night temperatures considerably exceed the range where the temperature response curves for photosynthesis, respiration and phenological development are more or less linear). It is comparable with the SUCROS models by van Keulen et al. (1982) and by Spitters et al. (1989). Careful reading and practice are required to become familar with this module. Explanations and background were given in Chapters 2 and 3. The module contains an initial and a dynamic part. Before the initial section starts memory is reserved (Lines 3, 4) for the weather data. Line 2 specifies that the value of IDATE is an integer number. Two additional variables are included on Lines 2 and 4, again, to facilitate combining this module with those 105

described in Chapters 4 and 5 (Listings 7-11). IDATE, an integer, is the truncated value of DATE, and ranges from 1 to 365. DATE is the sum of TIME elapsed since simulation started and DATEB, a parameter representing the Julian date at which simulation starts. The initial section (Lines 5-11) is followed by the dynamic section starting in Line 12. Three dummy variables are introduced (Lines 8-10) and used (Lines 40, 51, 97) to ease combining this with other modules (Section 4.4). Actual weather data are read from tables using IDATE as input (Lines 103, 105, 106) and standard weather data are derived (Line 104). See Chapter 6 for further details. Ensure that manipulating DATE does not lead to values lower than 1 or higher than 365: CSMP does not reject an instruction to select data outside the TABLES, but results will be nonsense. The initial value of TIME (Line 113) should be 0.0. The FINish TIMe of 1000. is never reached: simulation always stops when either the FINISH conditions of maturity (DS = 2.0), or that of severe carbohydrate shortage (CELVN = 3.0) is reached. For meaning and implications of the TITLE, PRINT, PRTPLOT and PAGE statements, for the TIMER variables DELT, PRDEL, OUTDEL and FINTIM, and for run control statements, such as FINISH, refer to the CSMP manual (IBM, 1975), or to Basstanie & van Laar (1982). The last variables (Lines 120-125) are solely for convenient presentation of output.

3.4.3

Crop growth module with quarter-day time periods (L1Q)

Module L1Q (Listing 4) can be used for simulating crops in situations where temperature fluctuates a great deal, when the dynamics of plant reserves are important and when tillering and grain formation in cereals is under study. This module is organized in the same way as L1D. Module L1Q should not be combined with water balance modules L2SU and L2SS. To incorporate the module TIL (which simulates tillering and grain formation) into L1Q, substitute TIL for Line 36. Formation and use of available carbohydrates (WAR) is programmed in Listing 4 Lines 14, 29, 30, 51. The rate of use is derived from the growth rates of the organs (Lines 52-56), these being equal to potential growth rates based on carbohydrate availability unless a reduction occurs, such as that resulting from sink size limitation (Line 35). In each 24-hour cycle (i.e., sunrise to sunrise) the time period for integration (DELT) is equal twice to half the daytime (0.5 · daylength) and twice to half the nighttime (0.5 · (24h – daylength)). Though DELT itself remains equal to 0.25, all rates in integrals are multiplied with a correction factor for daylength, FADL (Line 144). This factor is larger than one, if the day part is longer than six hours, and vice versa. Simulation starts at 0.00 h, and output is printed at midnight (when PRDEL is a whole number). Printed rates shows their mid106

night values (so that photosynthesis is always 0.0). (In LlD, printing time corresponds with sunrise.) The variable DTIME indicates the starting time of the fraction of the day that the simulation has reached (first quarter: DTIME = 0., second quarter: DTIME = 0.25, etc.). The variable NIGHT signals whether it is day or night for calculating radiation intensity (Line 131). Temperatures at different times during the day are reconstructed from the minimum and maximum temperatures by the function FUTP (Subsection 6.1.3). The INTeGRaL function is used for various state variables to compute the running average of a variable (Line 47; Subsection 1.4.4) and to retain the maximum value that a variable reached (Line 46). The daily total of a rate can also be calculated with an INTeGRaL function; the content of such integrals is reset to 0.0 each new day (Lines 65, 66). 3.4.4

Functions and subroutines (module T12)

Module T12 (Appendix B) with special functions and subroutines completes the CSMP program. This terminal section can be used with the crop and soil modules at Production Levels 1 and 2. Several of its functions and subroutines are used in only one main program section and can be deleted if not used. Special remarks that are required to understand or use T12 are given here. Readers are advised not to change the contents of functions and subroutines as this can be intricate. Most variables inside these functions and subroutines are not defined in Listing 12. The functions (one output) and subroutines (several outputs) are carefully checked and can be used within the limits and conditions previously described. END closes the DYNAMIC section with its model structure definitions; data for reruns can be entered after END (Basstanie & van Laar, 1982). END may be replaced by END CONTINUE to allow output specifications (such as PRDEL) to change during a simulation run. This can be useful for checking fluctuation of rates during a few interesting days without the burden of a huge output for the full growing season. STOP terminates the section where reruns can be specified. It is followed by FUNCTIONS and SUBROUTINES, placed alphabetically. These program sections, written in FORTRAN, are placed here to maintain maximum lucidity in the main programs (CSMP permits their specification in the main program in several other ways). It is recommended that functions and subroutines be included in a subroutine-library (Subsection 1.4.4). Calculations involving photosynthesis, growth and respiration are complex. Even experienced programmers easily make errors when rewriting a program or adapting functions and parameters. To avoid some of the most obvious errors a check on the C balance (FUCCHK function) is included. This consists of a comparison of the total net amount of C that entered and that is retained in the crop. The totals must be identical, but relative differences up to 1% are 107

allowed for rounding off errors. In the models described here, the relative difference is usually less than 0.1% . If it does exceed 1%, check the consistency of the crop input data and of all statements that were changed, removed, or added. WIR represents the total reserves (starch) formed since simulation started and must be expressed as an integral function. The FUPHOT function is generally used to compute the rate of gross photosynthesis of the canopy. The SUPHOL subroutine is an alternative to FUPHOT which permits computing photosynthesis of a canopy (PCGC) consisting of layers with different characteristics (Subsection 2.1.4). The FUTP function approximates the fluctuation of air temperature during the day for the L1Q module. The FUVP function calculates the saturation vapour pressure (in kPa) that corresponds with a given temperature. Total daily radiation on fully clear days and daylength, both astronomical and photoperiodical, result from the SUASTR subroutine. Both FUPHOT and SUASTR call the subroutine for astronomical computations SUASTC. The ENDJOB statement terminates the CSMP program.

3.4.5

Two examples of application of the modules

The first example addresses the potential growth of cowpea in Mali. Cowpea is an important crop in Sahelian countries, but hardly ever reaches its potential production level because of low soil phosphorus, water stress and insect problems. As a result, the potential yield of this crop is hardly known. The L1D + T12 modules can help provide an estimate of this crop’s potential. The simulated growth curve (Figure 42) was obtained without calibrating the basic crop data. Simulation began at the date that the real crop started to grow. Data collected by P. Gosseye in an experiment in similar conditions, given in the same figure, show a similar pattern. This similarity provides some credibility for when simulating the potential yield of cowpea for other West African countries. The second example simulated a field experiment with a silage maize crop (cv LG11) in Wageningen in 1985 using the module with quarter-day time periods (L1Q + T12). This was a cloudy year with a low final yield. Nutrients and water were always in ample supply. Accumulation of starch may have occurred on the few very bright days which could have resulted in reduced photosynthesis. Measuring with extensive equipment (Louwerse & Eikhoudt, 1975) of some crop plants within an enclosure of approximately 2x1x1 m provided a continuous record of CO2 exchange. The measured CO2 exchange rates were summarized over the same periods as simulation occurred (i.e., midnight till sun-up, sun-up till noon, etc.; adjusted for the position of Wageningen with respect to GMT and summertime). Corrections were also made for a 15% reduction of radiation by the enclosure. Figure 43 shows the results of the simulation and measurement for net CO2 exchange of the above-ground dry matter for both day and night and the daily totals. Simulated nighttime 108

Figure 42. Simulated (left) and measured weight of a cowpea crop in Niono, Mali (Source: crop data from Gosseye, in Haverman (1986)).

respiration is higher than measured, but it is not certain whether the measured data are better than those simulated. Daily totals of net CO2 exchange are fairly equal. High radiation caused the real crop plants to absorb a little more CO 2 than is simulated, possibly due to canopy disturbance by placing the enclosure. There is more variation in the real assimilation rates over the half-day periods, which may be due to an uneven distribution of radiation over the day not accounted for in the model. Reduced canopy photosynthesis due to carbohydrate buildup did not occur in the simulation. Photosynthesis of the field crop on bright days was not smaller than the simulated values and it is concluded that CO2 assimilation was not limited by sink size. Crop growth was established by periodic harvests (Sibma and Louwerse, CABO, personal communication). The final biomass was about 16.000 kg dry matter ha-1 , which was simulated closely (15.500 kg ha-1) with a realistic distribution over the organs. This also supports the conclusion that photosynthesis was not restricted by the capacity to absorb its products.

109

Figure 43. Simulated (PCNSHQ) and observed (NFOT) net photosynthesis of a maize crop in Wageningen on a few selected days of the 1985 growing season (crop data from Louwerse and Sibma, CABO personal communication).

3.5

Exercises

See Section 2.5 for an introduction to the exercises.

3.5.1

Morphological development

T1. Make a relational diagram of the processes and variables which determine morphological development. T2. Estimate how much longer the growing season of IR36 lasts at 400 m elevation, as compared to sea level. T3. How much earlier will potato cv Favorita be ready for harvest in comparison to cv Mara if planted at DS = 0.2 at 18 °C? How much at 26 °C? T4. Are the responses in Table 13 of the average rice variety and Nipponbare to temperature really different? S1. Plot daylength for photoperiodicity as a function of date for latitudes 110

–30°, 0°, 30° and 60°. What is the impact of daylength on the development rate of soya bean (cv Hawkeye) at day 80 at these latitudes? (Suggestion: make a MERGED PRTPLOT of DLP with L1D; eliminate the call for FUPHOT and the superfluous parts in the terminal section). S2. Program a strong daylength sensitivity in rice during the development stage period 0.2–0.7 (multiplication factor 0.5 at 13 h and 0.33 at 14 h, such as cv BPI-76 in Vergara & Chang, 1985). By how much is the growing season shortened due to the effect of daylength by transplanting at dates 250 and 350 as compared to day 150? And by how much in the northern Philippines (18°N) and in the southern Philippines (6°N), supposing all other things remain the same?

3.5.2

Assimilate partitioning

T5. Explain the difference between biomass and assimilate partitioning. T6. Estimate how much senescing leaves contribute to crop yield in potato, rice and soya bean. T7. In what environmental conditions do rice grains fail to reach their normal maximum weight? T8. Make a relational diagram of the processes and variables which determine development of tillers, florets and grains (see Listing 2). T9. Explain the meaning and formulation of CELVN and CELV in Listings 3 and 4. T10. Assume the leaf area of a rice crop to be 7.5 m2 m-2 at flowering, 6.0 at DS = 1.5 and 1.5 at maturity. Propose a function for L1D to reproduce this reduction in leaf area. S3. Determine the effect on grain weight and leaf area of a breeding program that leads to IR36 with the assimilate partitioning pattern of IR64 (Table 18). Did you expect a large effect? Propose a more effective assimilate partitioning for IR36 in Los Baños. S4. Determine the effect on grain yield and its components of doubling and halving the time coefficient for tiller formation in rice (do not change other formation rates) in a sowing density experiment with 0.068, 0.2, 0.68, 2.0 and 6.8 million plants ha-1 . (Use modules TIL and LlQ). What is the optimum density? What do you notice about the number of grains per panicle? Is the pattern of tiller development and death realistic? S5. Compare the effects of a heavily overcast period of three weeks before or after flowering on rice grain yield and yield components. Set radiation in overcast conditions to 17% of clear sky radiation (0.15 · RDTC, Section 6.1).

111

3.5.3

Leaf area

T11. How much is the rate of canopy photosynthesis with a leaf area of 3.0 m2 m-2 and PLMX = 40.0 at a day that is 50% fully clear and 50% fully overcast, according to Figure 38? Does it correspond with Figure 22? Why is the fraction of the reference canopy photosynthesis in overcast conditions much higher than the fraction for clear sky conditions? S6. Is the contribution of photosynthesis by soya bean pods significant? Estimate. Then evaluate the estimate with the model assuming that 2500 kg pod weight has a surface area of 1 ha (one side only). 3.5.4

Simulation models

T12. For what non-physiological reason is the simulation result of potato growth nonsense, if the response of development rate to temperature is specified as in Table 13 and the average temperature is between 30 and 35 °C? Is this a very special situation? T13. What kind of C-balance errors are detected and which are not detected by FUCCHK? 3.6 3.6.1

Answers to exercises Morphological development

T1. Carefully distinguish the types of variables and the relations involved. Use the symbols of Figure 5 to draw the diagram. T2. The growing season lasts 1. / 0.013 + 1. / 0.028 = 113 d at sea level (Table 12). At 400 m elevation, the temperature is, on average, 2 °C lower. The growing season will then last about 7% = 8 days longer (Table 13) T3. From DS = 0.2 to 1.0 in Mara at 18 °C takes 22.9 days and in Favorita 6.4 days less. From DS = 1.0 to 2.0 takes 66.7 days and 44.4 days respectively, so that Favorita is 28.7 days earlier. At 26 °C, interpolation in the functions of Tables 13 and 16 is necessary. Favorita can be harvested after 188 days and Mara only after 269 days. T4. If relative values are calculated for the same temperatures the responses are almost the same in the range 22-32 °C; Nipponbare develops more slowly at 16-22 °C. The responses below 16 and above 32 °C cannot be compared. The assumption that the responses above 22 °C are identical will be difficult to prove wrong with an experiment. S1. The result is a plot with three sinusoidal curves with different phases or amplitudes. The photoperiodic daylength at the equator is almost constant at 12.55 hour. The effects on the development rate at day 80 are 0.97978, 1.00, 1.00 and 0.86687, respectively. S2. FUNCTION DRDT = 8.,l. , 12.,1., 13.,0.5, 14.,0.33,... 112

24.,0.33 and DRV = ... as in Subsection 3.1.2. In Los Baños, the growing season decreases from 135 to 102 and 93 days; in the north from 142 to 102 and 93 days, and in the south from 121 to 104 and 98. This is all due to daylength effects because the temperature for these exercises is kept constant. In reality the effects of daylength and temperature are intertwined.

3.6.2

Assimilate partitioning

T5. Assimilate partitioning implies that glucose distribution to organs is regulated and occurs according to a certain key. Biomass results from glucose after growth took place. Biomass partitioning implies that biomass allocation is regulated; and it suggests feedback from the organs to the distribution mechanism on the amount of glucose required. The first hypothesis seems more appropriate. T6. Assuming 2500 kg ha-1 of leaves at flowering and only 50% of that still attached to the stems at maturity, the dropped leaves provided an equivalent of about 600 kg ha-1 of glucose. This corresponds with 460 kg ha-1 of potato tubers, 410 of rice grain and 280 of soya beans. T7. When photosynthesis plus remobilization in the grain-filling period is much less than is anticipated during flowering, then there are too many grains to be completely filled. This can occur when the light level is very low (see exercise S4), during water stress, when diseases are present and at low temperatures. T8. Distinguish carefully the types of variables and the relations involved in tiller, floret and grain development. Use the symbols of Figure 5 to draw the diagram. T9. CELVN is the number of consecutive days without carbohydrate export from leaves and stems. The crop is assumed to die when the value reaches 3. Stems are included because they contribute to photosynthesis. CELVN increases by 1. each day that CELV is negative; it is reset to 0.0 if CELV becomes positive again (no after-effect). CELV in L1D is the gross photosynthesis minus maintenance of leaves and stems; the fraction of leaf maintenance that is contributed by excess energy from photosynthesis is subtracted. In LlQ, CELV is in principle the same, but the gross photosynthesis is to be accumulated during 24 hours and reset each day. T10. The reproductive period lasts about 30 days. In the first 15 days, loss is 20%, in the second 15 days 75% of what remained. This is reproduced by: FUNCTION LLVT = 0.,0., 1.,0., 1.5,0.03, 2.,0.15 S3. Grain yield (rough rice) of IR36 is 7691.4 kg ha -1 and 6358.2 of R36 x 64; leaf area at flowering is 9.9 m2 m-2 and 11.4, respectively. The new cross is a crop with too much vegetative biomass. Lowering the share that goes to leaves by 0.1 and reducing the share to the stem to 0.1 and 0.0 at DS 1.0 and 1.25, respectively, decreases leaf area to 8.8 and boosts yield to 8200.1 kg ha -1. 113

S4. The planting densities are obtained with WLVI = (0.68,2., 6.8,20., 68.) in kg ha-1. WSO is in kg ha-1, NTI in 1.E6 ha-1 , NGR in 1.E8 ha-1 . The results at maturity are:

The standard density (68 plants m-1) appears to be slightly below optimal with the standard TCFT. The grain number is much more constant than the tiller number, so that the number of grains per tiller decreases substantially when the tiller number increases. The pattern of tiller development in a field experiment was roughly similar to this simulation for the standard case. S5. Run the program first to find the date of flowering (DATEF, equals 56.). Then replace ‘RDTMT(IDATE)’ in Line 131 by LIGHT, and add:

The grain yield drops from 7574.1 kg ha-1 to 3251.3 with early clouds, and to 3928.1 with late clouds. Late clouds do not reduce the number of tillers per plant 4.571, but early clouds reduce it considerably 2.828. The adjustment to low light levels due to late clouds causes tillers to carry much fewer grains, but the grains formed fill completely. Early clouds result in a low number of tillers, the few grains present are filled before maturity and the crop stops growing too early.

3.6.3

Leaf area

T11. Approximately 0.54· 1144.· 0.5 + 0.77· 311.· 0.5 = 429 kg CO2 ha-1 d-1 . This is similar to the rate in Figure 22. Canopy photosynthesis increases little at low radiation with PLMX beyond 25., but increases a lot in full 114

light. S6. The contribution to canopy photosynthesis by soya bean pods is probably small since the additional green area is small. To simulate this, add the growth of pod area (= GSO / 2500 0.5) to ALV and adjust SLA as for stem material. In the standard conditions for soya bean, the yield increases by only 65.3 kg ha-1 with the extra green area.

3.6.4

Simulation models

T12. Because the temperature exceeds the specified range and CSMP extrapolates the relation from the last two data points, the development rate becomes negative. The problem is cured by adding 40.,0.01 to the functions DRVTT and DRRTT. Always take care that inputs do not exceed the specified range. T13. Assimilated carbon remains in the plants or is lost by respiration. FUCCHK continuously checks whether all carbon is accounted for. Omitting one of the respiration rates causes an error. Another occurs when FCLV is unequal to CRGLV * 12. / 30. – CPGLV * 12. / 44. FUCCHK does not detect incorrectly calculated rates of photosynthesis, remobilization or respiration, nor incorrect carbohydrate partitioning (except when the sum of the fractions is unequal to 1.00).

115

4

Transpiration and water uptake

This chapter describes how to simulate transpiration and water uptake, the principal processes of crop water balance. Potential transpiration is discussed in Section 4.1, and water uptake, the limiting factor for transpiration during water shortage, in Section 4.2. Section 4.3 discusses the effects of water shortage on other physiological processes, such as phenological development. 4.1

4.1.1

Transpiration without water stress

Introduction

When there is sufficient soil water (as at Production Level 1) the photosynthesis rate largely determines the transpiration rate. When water is in short supply, the inverse is true, for the rate of water uptake from the soil is then of crucial importance. Both situations can be seen in Figure 6, Subsection 1.2.2, where with ample soil water, transpiration equals potential transpiration and photosynthesis equals potential photosynthesis; when there is water shortage, uptake is less than potential transpiration, and due to this stress, photosynthesis is below potential photosynthesis. This section is summarized in module L2C for crop processes at Production Level 2 (Listing 7). In many cases it is important to consider crop water use. First, because it can be an important topic in itself in field or regional water balance studies, whether water is plentiful or not. Second, because water shortage is a very common phenomenon and water stress, brief or long, reduces the rate of crop growth and can affect dry matter distribution and hence the economic yield. Water loss and uptake and the efficiency of water use by crops have been the topics of many studies, reports and reviews. The entire Volume 12B of the Encyclopedia of Plant Physiology (1982) is devoted to them. Other reports on transpiration and crop production modelling are: Slatyer (1967), Feddes et al. (1978), Doorenbos & Kassam (1979), Tanner & Sinclair (1983), and van Keulen & Seligman (1987). The time period for simulating transpiration is 24 hours. The instantaneous effects of the driving forces of transpiration, irradiance and air-drying, are almost proportional to transpiration without water stress. Therefore, it is assumed that the daily total radiation and the average daytime drying power are proportional to daily transpiration. Amounts of transpiration per day are quite large in comparison to crop water content. Time periods even smaller than those for simulating transient carbohydrate levels (i.e., 0.1 cm cm-3) for different crop species.

Species

Rate of increase in rooted depth (m d-1)

Maximum Reference rooted depth (m)

Barley Cotton

0.03 0.025 0.03 0.028 0.014 0.06 0.06 0.04 0.02 0.014 0.02 0.01 0.05 0.035 0.02* 0.02* 0.018 0.012

1.5 1.8 1.8 1.0 0.7 1.0 2.0 1.0 1.0 0.8–1.0 0.4–>0.8 0.3 1.4 1.7 1.2 0.4 1.3 1.8

Cowpea Faba bean Maize Millet Potato Rice(up1and) Rice(low1and) Sorghum Soya bean Sugar-beet Tulip Wheat winter spring

Day et al., 1978 Bassett et al., 1970 Taylor & Klepper, 1974 Haverman, 1986 Grashoff et al., 1987 Sibma, 1987 Taylor & Klepper, 1973 Jansen & Gosseye, 1986 Gregory & Reddy, 1982 Vos & Groenwold, 1986 Yoshida & Hasegawa, 1982 Sharma et al., 1987 Kaigama et al., 1977 Stone et al., 1976 Brown & Biscoe, 1985 Benschop, 1986 Gregory et al., 1978 van Keulen&Seligman,l987

* estimate

lack of specific data, the effect of temperature on root extension is supposed to equal that of photosynthesis. The effect of water stress on the rate of increase in rooted depth is supposed to equal that of water uptake in the layer where the root tips are found (Listing 8 Line 27, Listing 9 Line 74). The effect of anaerobic conditions on root extension downwards is handled by setting the rooted depth increase to zero at depths below 0.2 m when there is less than 5% air in the soil (Listing 9 Line 73). Roots grow down to a certain maximum depth if they are not restricted by soil conditions. The maximum depth depends on the plant species and ranges from 0.5-1.5 m or more. Table 25 gives some approximate values for the maximum rooted depth. Significant differences between cultivars for this characteristic are reported for upland rice (Gupta & O’Toole, 1986) and are also expected to exist within other species. Sensitivity analysis has established this as an 137

important characteristic, though little is known about it in field crops. Maximum rooted depth should be determined around flowering in soil profile pits, either by using root observation tubes (Vos & Groenwold, 1983), or indirectly by monitoring (with neutron probes) the depths from which water is drawn when drainage is insignificant. A very dense soil offers mechanical resistance which hampers the extension of roots downwards and reduces the maximum attainable depth. An obvious case is where shallow soil lies on bedrock. High soil densities can also be found at depths of 0.30-0.80 m in deep soils, particularly just below the ploughed layer. Its presence may be intentional, such as during soil preparation in irrigated rice where a hard pan is needed to reduce percolation of irrigation water. A compact layer can also develop unintentionally, such as when harvesting crops with heavy machinery. A physical limitation to rooted depth is approximated by specification of a maximum depth as a soil characteristic; the shallowest of the rooted depths set by the soil and by the crop is used (Listing 7 Line 33). Note that cracks, tunnels from animals or decayed roots, and other irregularities can make dense layers more penetrable for roots than the soil density measurement of a uniform piece of soil may indicate (Subsection 5.1.4). Loss of rooted depth in a senescing root system may be added to a simulation model, but field data are needed to calibrate this effect.

4.2.4

Anaerobic conditions

Plants with roots in fully saturated soils generally suffer from stress. For an extensive review of physiological effects of excess water see Jackson & Drew (1984). Root systems of agricultural crops that are developed in aerobic soils do not have aerenchym and degenerate within several days when anaerobic conditions are imposed. Root permeability first decreases and uptake slows down. Root cells disintegrate and die when their metabolism no longer provides sufficient energy (i.e. , O2) for maintenance. Hence, wilting is sometimes, though not always, observed after flooding. Flooding quickly depletes the O2 in the soil and the supply is then almost nil. Anaerobic conditions occur on heavy soils following intensive rainfall and when the groundwater table is very high. Those conditions can be simulated with the module L2SS (described in Section 5.3). Rice in irrigated or rainfed lowland soils has an effective root system in anaerobic conditions, because its roots develop aerenchym tissue that provides air channels (Yoshida, 1981). The rate of diffusion through the narrow channels provides sufficient O 2 to permit roots to extend to about 0.2 m. Several other crops develop roots with aerenchym, but not as extensively as rice. The effect of flooding on water uptake is approximated with the FUWS function, in a similar manner to the effect of water shortage (Figure 55, right hand part of the graph). The effect is assumed to be proportional to the soil 138

water content between field capacity and saturation, and independent of the transpiration rate. To mimick the non-water stress effect of flooding which occurs after a root system is established, a FINISH condition similar to CELVN (Subsection 2.3.4) can be added to the program, stating that crop death occurs if flooding lasts a certain number of days. This maximum flooding period is dependent on the species and its development stage. Some species grow a new root system with aerenchym in anaerobic conditions. Rice does so extensively, but other crops, including soybean, wheat and sunflower, also have this capacity. The regrowth rate of an effective root system after flooding also effects the degree of crop survival. This is not considered here. Diseases are common after soils are flooded and cause much damage. This is due to the crop’s physiological condition and the high humidity accompanying flooding. This is not considered here. 4.3

4.3.1

Non-stomatal effects of water stress

lntroduction

This section discusses simulation of the effects of water stress on the physiological processes of plants which are not mediated by stomata. The degree of

Figure 55. The relation between the soil water content and the stress multiplication factor on the rate of water uptake. WCWP, WCFC and WCST represent the soil water content at wilting, field capacity and saturation, respectively. The dashed line represents either a more drought resistant species under the same field conditions, or the same species under a lower evaporative demand.

139

stress can not be simulated explicitly because of the relatively long integration period. The ratio between actual transpiration (TRW) over potential tranpiration (TRC) is used to represent the degree of stress. When this ratio is above 0.5, the effects on physiological processes are usually small. The most significant influence of water stress is the indirect effect through reduced photosynthesis, particularly if the stress occurs during a sensitive period, such as during grain initiation in cereals (Subsection 3.2.5). This definition of water stress implies that the effects of stress are over as soon as the soil is moist again. However, after prolonged stress this is not correct. To account for aftereffects, van Keulen (1982) computed a particular running average of the relative transpiration deficit (RTDA and RTD, respectively) to characterize water stress. This running average increases when the relative deficit exceeds 0.4 and decreases when it is less. Van Keulen chose a 10-day time coefficient for buildup and breakdown of the average value. This resembles the simulation of buildup and breakdown of hormone levels. In van Keulen’s model, the running average, representing water stress, affects biomass partitioning, leaf photosynthesis characteristics and leaf senescence. This description of the effect of water stress can be achieved by replacing TRW / (TRC + 1.E-10) in Listing 7 Lines 10-11 by (1.0 – RTDA), where:

4.3.2

Crop development during water stress

From the limited amount of data available, it appears that a moderate level of water stress often has no direct effect on the rate of crop development (Halevy, 1985; Section 3.1). However, there are exceptions, for the physiological development of some crops slows under stress and this lengthens the vegetative period. In other crops water stress stimulates physiological development, for example, the development rate of Faba beans increases once the actual/ potential transpiration ratio is below 0.7 and this increases up to twice the normal value to a ratio of 0.0 (Grashoff, CABO, personal communication). The development rate for rice in rainfed lowland and upland situations decreases under moderate stress so that flowering and maturity are postponed by 7-10 days (Buresh, IRRI, personal communication). Morphological development of sugarcane stops, or even reverses, under stress. Acceleration or deacceleration of development can be represented in the model as a relation between the stress level and a multiplication factor for the development rate (Listing 3 Line 97; Listing 7 Line 10). Very little data on this relation is reported in the literature. Without explicit data, the effect is assumed to be negligible. A freely transpiring crop cools itself so much that the leaf temperature can 140

be several degrees below the air temperature, even at high radiation intensities (Idso et al., 1981). The temperatures of water-stressed leaves can be several degrees above those from non-stressed leaves (Gupta & O’Toole, 1986). Reduced transpiration under nutrient stress can have the same result. Growing points do not have the same temperature as leaves, but have a temperature between that of leaves and the air. Increased temperature can accelerate flowering and shorten the crop’s life cycle by several days (Seligman, ARO, personal communication). This temperature increase could be approximated from the reduced transpiration rate, but it is difficult to provide a fair level of accuracy. No general solution is proposed here. Extreme levels of water stress may kill part of the plant. If new sprouts or tillers emerge, the crop rejuvenates and returns to an earlier development stage. No general solution is proposed here for simulating the effects of extreme stress levels.

4.3.3

Carbohydrate partitioning

Carbohydrate partitioning between shoot and root under water stress is altered in favour of the root biomass. Brouwer (in de Wit et al., 1978) described the biological principle of the mechanism; roots are formed in proportion to the demand from shoots for water. Yet it is difficult to quantify the growth stimulation of root biomass in response to water stress. It is assumed that up to a moderate stress level (actual/potential transpiration rate is 0.5) there is no significant effect on partitioning. At higher stress levels during the vegetative phase, the share that goes to the roots increases by up to 50% of the amount that otherwise would go to the shoot (Listing 7 Line 9, and Listing 3 Line 40). The flow of carbohydrates to storage organs in vegetative crops, such as sugarbeet, increases under stress at the expense of the flow to leaves. This is not included in the program. Other differences between species in this respect are not yet known. Although water stress stimulates root growth relative to shoot growth, water stress in the layer with root tips reduces the root extension rate to greater depth. Drought in upper layers does not stimulate Faba bean roots to grow deeper (Grashoff et al., 1987). It is assumed that the relative partitioning of carbohydrates between leaves, stems and storage organs is not affected by water stress. The impact on formation of reserves is an indirect effect. Clearly, these are only approximations, though experience with field crops indicates that they are often acceptable (Section 3.2). 4.3.4

Leaf area

The effect of water stress on the growth of leaf area occurs through the effect of stress on root-shoot partitioning, and hence on an increase in leaf weight. Water stress can lead to higher values of the specific leaf weight, but this is not 141

simulated here. An explicit way to deal with the effect of stress on the growth of leaf area is described by van Keulen & Seligman (1987) and is specific for the development of leaf weight in wheat. Severe water stress can lead to progressive death and removal of leaf area. It is impossible to simulate this process dynamically, because the extreme values reached during the day are of critical importance. These are not obtained with the modules described here. Heterogeneity of soil environment is also important. Modellers should use experimental data to mimick leaf and stem death under severe stress for their crop and their situation. Van Keulen (1982) reduces biomass of wheat and grasslands by 0.1-0.2 d-l when water stress exceeds a certain level and reduces leaf area correspondingly to mimick gradual crop death. Some crops use leaf rolling as a means of reducing leaf area when waterstressed, such as rice (Gupta & O’Toole, 1986). Rolling increases diffusion resistances and reduces the area of exposed leaves. Other crops move or fold leaves when under stress to intercept less radiation. In these ways, they avoid drought, save some water and have a better chance of survival. But they are also less productive during the stressed period. These phenomena can be simulated by replacing the leaf area (ALV) by the effective leaf area (ALVE) in all appropriate rate equations and functions and by reducing the effective leaf area in relation to the stress level. For example: ALVE = ALV * AFGEN(ALVRT, TRW / (TRC + 1.E-10)) FUNCTION ALVRT = 0.0,0.4, 0.4,0.4, 0.5,0.5, 0.6,0.8, 0.8,1., 1.0,... 1.0 This numeric example for leaf rolling is derived from O’Toole & Cruz (1980) for lowland and traditional upland rice cultivars. Reduced leaf area due to wilting can be similarly mimicked. (In this approximation, an ‘implicit loop’ creates a pitfall, for it is constructed by making the effective leaf area a function of the actual/potential ratio, because the actual transpiration itself is a function of the effective leaf area. A loop, such as this, can be solved in CSMP by putting in the DYNAMIC section ‘FALVE = ...’ as the last line instead of ‘ALVE =....’ (as above), and ‘ALVE = IMPL(ALV, 0.05, FALVE)’ as the first line. CSMP will try each time period again with different values for ALVE until all rate equations are balanced, and only then proceeds with integration. For further information on the IMPL function, see IBM, 1975).

4.3.5

Leaf photosynthesis characteristics

Severe water stress can reduce the capacity of leaves to photosynthesize. This may cause damage to chloroplast structure and adaptation to lower actual photosynthesis rates. Van Keulen (1982) suggested considering the maximum leaf photosynthesis rate (read: amount of intact RudPCase) and the initial efficiency (read: amount of chlorophyl) as state variables. Both state variables 142

decrease under severe stress by 0.0-0.05 d-1 and have only a limited recovery capacity. It is a good approach, but the parameters are not general and need to be defined for each case. Readers may want to explore this further. If temperatures are so high that the photosynthesis apparatus might be damaged, attention should be directed to the most vulnerable leaf layers when calculating leaf temperature. The simulation of transpiration given here is not sufficiently detailed for this purpose.

4.3.6 Maintenance respiration Maintenance respiration under water stress may intensify due to ion gradients increasing when the osmotic value of cytoplasma increases. On the other hand, the flux of excess energy increases when photosynthesis is reduced under stress (Subsection 2.3.3). Therefore, it is assumed that the carbohydrate requirement for maintenance respiration is unaffected by water stress. The energy for maintenance in leaf cells in daytime is still provided even when CO2 assimilation has almost stopped. 4.4

Exercises

See Section 2.5 for an introduction to the exercises.

4.4.1

Transpiration without water stress

T1. Make a relational diagram of the processes and variables which determine canopy transpiration without water stress. Where is the water use coefficient? How is stomatal regulation indicated? T2. Is it important to establish whether the maximum resistance of the leaf exceeds 2000 s m-1? If so, suggest ways how to establish it. T3. How is the transpiration coefficient (H2O/CO2) converted into the water use coefficient (water/dry matter)? S1. Evaluate the importance on yield of leaf rolling in rice during the wet season on loamy soil, as discussed in Subsection 4.3.4. What happens if the precipitation in the beginning was slightly more (29 • 6.), but the downpour of 300 mm occurred only at day 260 and no rain fell at the end of the growing season (15 • 0.)? Explain the result. S2. Suppose that soya bean stomata can be made to regulate at a CO2-internal/CO2-external fraction of 0.4. How is yield affected in the wet and in the dry season on a loamy soil, and how is the average water use efficiency and average stress level affected? What is the consequence of a disease that induces loss of stomatal control, leaving them fully open all the time? Why is the water use coefficient in both seasons similar? Do you expect a similar result on sandy soil?

143

4.4.2

Water uptake

T4. Make a relational diagram of the processes and variables which determine water uptake under water stress. S3. Determine the numerical value of the effect of stress on water uptake for all soil water contents between air dry and saturation for values of TRC from 0.5 to 10 mm d-1, for ALV from 0.5 to 10 m2 m-2 , and for the water stress and flooding sensitivity coefficients between 0.0 and 1.0. Is a value of – 1.0 for WSSC or WFSC biologically meaningful?

4.4.3 Non-stomatal effects of water stress T5. Explain the operation of the statements to compute a running average for the relative transpiration deficit (Section 4.3.1). What are the maximum and minimum value of RDTA? T6. The effect of high or low values of the water stress sensitivity coefficient on crop production in monoculture is often not as large as one might expect. Why not? Why can this be different if weeds are present? T7. The actual rooted depth attained in simulations is usually 0-3 cm more than the maximum specified. Explain. What are the consequences? S4. Add to the rice model (Listings 3, 5, 7, Appendix B) that the development rate is unaffected down to a water stress level of 0.9, but reduced by 30% when the stress level reaches 0.5. Run it with the SAHEL water balance. Reduce the precipitation by replacing 29· 4., 200. with 30· 0.0. By how much is the crop delayed and is yield decreased? Explain. S5. What would be the difference in rice yield if a breeding program could make rice absorb water more effectively compared to ineffectively (WSSC equals 1. and 0., respectively)? Assume that the development response to stress and precipitation are those from the previous exercise. How does it compare to increasing the maximum rooted depth by 0.1 m? Explain the results. S6. Add the equations about the running average of transpiration deficit (Section 4.3.1) to the dry season rice program on loamy soil. What is the effect on yield and water use efficiency? 4.5

Answers to exercises

4.5.1 Transpiration without water stress T1. Carefully analyse the processes and variables. Draw the diagram using the symbols presented in Figure 5. Compare your result with Figure 6. The water use coefficient is the ratio of actual transpiration and actual photosynthesis. The stomatal resistance is mainly determined by photosynthesis and the CO2-internal/external fraction. T2. It is unimportant because the absolute value of the transpiration rate is 144

already very low when this maximum becomes effective. T3. The conversion of carbohydrates into dry matter often yields about 0.5 g g -1. CO2 weighs 44/30 times as much as the carbohydrates. The water use coefficient is therefore about 3x as high, the actual ratio depending on conditions and the species. S1. Replace ALV by ALVE in all rate equations and in Lines to calculate WSEl-3. The rainy season yield is 6146.9 kg ha-1 at maturity with leaf rolling; without response to water stress it is 6186.0 on the same date. The stress on this soil was too mild to give the responsive crop an advantage in survival and it lost a little in productivity. The late rain in the ‘dry’ wet season is just in time for the leaf-rolling crop to survive the serious drought, so that it yields 4295.0 kg ha-1 at day 290. The non-adaptive crop had less water left in the soil when the late rain arrived. Net photosynthesis was negative from day 258 onwards and the crop died before the rain arrived at day 261. (Note: use FUPHOT without the restriction on ALV imposed by SUERRM to allow the IMPL function to perform preliminary computations over a wide range.) S2. Use LlD, L2C, SAHEL and T12 with soyabean data for the wet and dry season, the standard sandy soil and short weather data set. FIEC = (0.64,0.4, 0.95). The average stress level may be computed as the accumulated ratio of TRW and TRC divided by TIME. The yields are 3452.0, 3446.4 and 2124.9 kg ha -1, respectively, in the wet season, with average water use coefficients (kg H20 kg -1 CO2) of 101.3, 72.2 and 212.9, and average stress levels of 0.994, 1.000 and 0.822. In the dry season, the numbers are: for yield 2418.4, 4620.8, 406.0 (early death); for water use coefficient 111.1, 78.2, 202.5, and for average stress levels 0.870, 0.990, 0.699, respectively. The water use efficiencies in both seasons are the same because there are almost no differences in seasons in the short data set. The effect of changing FIEC from 1.0 to 0.4 is much more pronounced when less water is available, such as on a sandy soil, or with less rain.

4.5.2

Water uptake

T4. Carefully analyse the processes and variables involved. Draw a diagram using the symbols of Figure 5. S3. Use only Line 17 from Listing 7 and the FUWS function from Appendix B. Make reruns for combinations with multiple value parameters and with TRC = TIME. The output statements PRTPLOT WSEl and PAGE MERGE make it easy to check results. WSEl plotted versus TRC is a graph with curved lines at different positions. Negative values of WSSC or WFSC imply that uptake is restricted by roots or soil even at field capacity. This could occur if the soil moisture contains salt or when roots are damaged.

145

4.5.3

Non-stomatal effects of water stress

T5. This running average is more complex than that in Subsection 1.4.3, because the rate of change of RDTA depends on another variable RTD. RDTA decreases when RTD is low, and increases when RTD increases. Multiplication with ‘1. - RDTA’ prevents RDTA from exceeding 1.0. Without stress, RTD and RDTA are equal to 0.0. T6. The WSSC and WFSC coefficients do not make more water available to the crop, but increase the rate at which it can be absorbed. The difference is significant only if a crop grows under moderate water stress, but not if it is in soil that dries and then recieves water again. When a crop is competing with weeds, crop water absorption may slow down while weeds continue to grow apace (cf. , Lof, 1976). The crop's share of water is then smaller and crop yield is reduced. T7. The final rooted depth exceeds the maximum specified because the daily increment is several centimetres and growth stops only when the maximum is exceeded. The time period for integration is relatively long for simulation of root growth. If the specified maximum rooted depth is equal to the profile depth, roots ‘stick’ out of the simulated system and these root tips cannot absorb water. As a result, the total water uptake is a few percent too low. S4. The yield is 5706.5 ykg ha-1, harvested at day 292. Without any development rate response to stress, the yield is 236.6 kg ha-1 less and maturity is reached 2 days earlier. The longer growth duration permits higher production because stress was not severe. S5. The wet season differences are small. The difference between the WSSC extremes is 152.8 kg ha-1 in grain yield and 2 days in maturity date. This soil appears to provide a large buffer. Increasing the rooted depth by 0.1 m raises the yield by 219.2 kg ha-1 and advances maturity by 2 days (reference: values of the previous exercise, delayed development). The advantage is small because the soil provided just enough water at a rooted depth of 0.7 m and stress develops only in the last days. Rooted depth of 0.8 m still has water left. The difference would be larger in a drier year. S6. The formulation with RTDA leads to a reduced water use efficiency (WUDM = 76.0 versus 95.9 kg dry matter kg-1 water at TIME = 70.; WUPC remains almost the same). There is now a little more water stress and this leads to premature death (at DS = 1.598). Values of RTD below 0.4 have less effect on growth than the original equation, while those that exceed 0.4 have progressively more effect.

146

5

Soil water balance

Simulation of the soil water balance is discussed in this chapter. Though rooted in soil physics, the main simulation modules are relatively simple because the water balance is only considered in relation to crop growth and many details are hidden. Section 5.2 discusses a module (L2SU, Listing 8) to simulate soils that drain excess water freely and quickly and uses a one-day time period for integration. Section 5.3 discusses an alternative module (L2SS, Listing 9) simulating soils in which drainage is impeded, and in which saturated layers may develop. This second module is more complex and simulates fluxes with time periods much smaller than 24 hours, while crop simulation continues with one-day time periods. Section 5.4 presents details of the two modules and provides examples of their use as part of a full simulation program. 5.1

5.1.1

Soil physics for simulating crop growth

Introduction

Water in the rooted part of the soil is quite mobile. Its distribution over time and depth is important because it determines the amount of water available to the crop. Water enters the soil as rain or irrigation water and often by capillary rise from the groundwater table. It leaves the soil by evaporation, drainage, or is taken up by roots. Gravity and gradients in moisture suction cause water movement within the soil. Soil physical characteristics (e.g., hydraulic conductivity) codetermine the rate of water flow. Horizontal (lateral) inflow or outflow can be significant in hilly regions and in plots next to ditches and waterways. A soil profile generally consists of layers of different soil types with distinct physical characteristics. To represent the vertical heterogeneity in water content, physical characteristics and root activity, the soil is divided into horizontal compartments or layers. The thickness and physical characteristics of each layer must be specified. The mathematical equations describing the soil processes are the same for all layers, but because the value of the variables and constants in the equations vary from layer to layer, the outcome is specific to each. Spatial variability of the soil water content at given depth in the field is caused by spatial heterogeneity of physical characteristics of soil layers, irregularities at the surface, artificial drainage structures and by heterogeneous root distribution. This variability is not included in these models. Two soil water balance modules are discussed: a simple module (L2SU, 147

Listing 8) for situations where the groundwater table is very deep (so that no water flows upwards to the rooted profile) and where the hydraulic conductivity of the soil does not limit downward water flow; and a more complex module (L2SS, Listing 9) for situations where hydraulic conductivity may limit water transport in the soil and where water can flow from the groundwater to the rooted profile. Neither module is suited to handle water infiltration into soils that crack when drying, as in vertisols (Eswaran, 1985); water uptake patterns in these soils are different, for roots prefer to grow along ped faces that offer little mechanical resistance. See also Table 1. 5.1.2

Soil texture

Soils differ in chemical and physical properties and in morphological characteristics and can be classified according to several systems. A widely used classification system is that described by the United States Department of Agriculture (1975). Another system uses hydrological conditions as the classification key (e.g, Kanno, 1956, 1962; cited by Moorman & van Breemen, 1978, p. 51). A classification based on physical characteristics is used here, since these are important for the soil water balance. The number of pores and the distribution of pore sizes determine the hydraulic properties of a soil (e.g., Schuh & Bauder, 1986). To a large extent, the distribution of particle sizes affects the distribution of pore sizes. After removing soil particles larger than 2.0 mm in diameter from a soil sample, three particle classes are distinguished by size: sand (0.05-2.0 mm in diameter), silt (0.002-0.05 mm) and clay (2 m below the root zone), but the dynamics of water infiltration into the soil, redistribution between layers, and evaporation are then simulated more crudely. The simulation module for this situation is L2SU (i.e., Production Level 2, soil, unsaturated) which is presented in Listing 8. L2SU is based on models by van Keulen (1975), Stroosnijder (1982) and Jansen & Gosseye (1986). Its concept is described by the acronym SAHEL, for Soils in semi-Arid Habitats that Easily Leach. A brief description is given here. Figure 59 illustrates some of its basic features. The inflow and outflow of water in separate layers is simulated on a daily basis. Inflow into the first layer is from rainfall. Field capacity is the highest water content that the soil can attain. The amount of water that cannot be stored in one layer, drains into the next layer or out of the profile. Water is extracted from layers by evaporation and transpiration. The soil profile is divided into three layers and each is considered to be homogeneous. Thickness and physical characteristics of each layer are inputs. The upper layer should be 0.10-0.20 m thick, the second 0.2-0.4 m, and the third 0.4-1.0 m. Their sum should slightly exceed the maximum rooted depth. The model can be extended to account for more heterogeneous situations by adding more layers. The soil characteristics needed for L2SU are the volumetric water contents of the soil layers at field capacity, wilting point and when air dry (Subsection 5.1.3). Some values for common soil types are given in Table 27. The size of clods on the surface and the surface albedo must be included. Listing 10 gives an example of how soil data can be presented.

5.2.2 Infiltration The infiltration rate (mm d-1) is equal to rain minus interception and runoff. Irrigation can be treated in the same way as rain. Not all water that reaches the surface infiltrates the soil, especially during heavy rain. Runoff from a field can be 0-20% of precipitation, and even more 155

Figure 59. A graphical representation of changes in the soil water content of three layers due to infiltration (top) transpiration and evaporation (bottom). (Source: Stroosnijder, 1982).

156

Table 27. Typical soil water contents for different soil types (calculated from Table 26 with the equation of Subsection 5.1.3).

Water content (cm 3 cm -3 ) at

Course sand Fine sand Loam Light clay Heavy clay

Air dry

Wilting point

Field capacity

Saturation

0.005 0.005 0.01 0.05 0.18

0.01 0.03 0.11 0.24 0.36

0.06 0.21 0.36 0.38 0.49

0.40 0.36 0.50 0.45 0.54

on unfavourable surfaces (Stroosnijder, 1982). It is negligible under proper irrigation and soil management conditions. Runoff occurs when the rate of water supply at the soil surface exceeds the maximum infiltration rate and the accumulated excess exceeds the surface storage capacity. In reality, the maximum infiltration rate is influenced by the water content at the surface. The simplest solution for calculating runoff is used in L2SU: it is assumed to be a constant fraction of precipitation (Listing 8 Line 40). This is not always satisfactory. The runoff fraction can be calculated as a function of daily precipitation (e.g., Jansen & Gosseye, 1986), and of the soil water content. However, the necessary data for calibrating will not often be available. Several other attempts have been made to describe the variation in runoff due to surface conditions (e.g., Davidoff & Selim, 1986; see also Subsection 5.3.5). A detailed consideration of runoff is warranted when its quantification is crucially important, such as for establishing a crop in semi-arid zones. Run-on of water from adjacent fields may also occur and can be added to precipitation. However, it is difficult to quantify or measure this term in the field.

5.2.3 Soil water movement When a soil layer is filled beyond field capacity, water percolates into the next lower layer. Most drainage occurs within 24 hours (except in heavy soils), and as one-day time periods are used in L2SU, it is only a small over-simplification to assume that all drainage occurs within one day. Simulation is therefore straightforward: if on any day more water infiltrates a soil layer than can be held by that layer, then the excess water drains into the next layer (Lines 41157

43). If more water enters the deepest layer than can be retained, the excess is lost as deep percolation. Some upward flow is implicitly simulated as the contribution of each layer to soil evaporation. Soil layers can become no wetter than field capacity. Limiting the infiltration rate into the second layer is a simple method of mimicking a water-logged top layer. Runoff should then be increased by the water that is in excess of the saturated top layer, plus maximum surface storage. Choose the value of maximum infiltration rate such that realistic periods of waterlogging result. Dynamic simulation of waterlogging can be undertaken with the module L2SS (Listing 9) when the saturated conductivity of the impermeable layer is known.

5.2.4

Evaporation

Surface evaporation is important for bare soils, but it is much less than transpiration under a well-developed crop canopy. Water can evaporate until the soil is air dry. The water content is then only about one third, or less of the permanent wilting point (see Table 27, Subsection 5.2.1). The amount of water held between wilting point and air dry can be lost by evaporation, but is inaccessible to crops (see Figure 59, Subsection 5.2.1). Similarly, rain or irrigation must first wet the soil till the wilting point is reached and then provide the water required by the crop. The potential soil evaporation rate is determined by the same energy balance processes as that of leaf transpiration. Therefore, the potential evaporation rate is also computed using the Penman approach (Section 4.1) in the SUEVTR subroutine. The first step of this calculation does not consider canopy shading, but includes the effect of the crop on windspeed near the surface (Listing 7 Lines 41-47). Shading is accounted for in the next step; the extinction coefficient for shortwave radiation together with near infrared radiation is about 0.5 (Listing 7 Line 40). This potential rate applies for the day at which water infiltrates into the soil; a maximum is the extent to which the upper layer can be depleted (Listing 8 Lines 48, 49). Surface roughness, characterized by clod height, affects the resistance of the boundary layer. It is used to calculate soil evaporation in the same way as leaf width (Listing 7 Line 45). Windspeed near the soil surface is less than at canopy or screen height; atmospheric resistance is related to crop height and density. Sensitivity of crop growth for these variables is low. Effects of soil tillage and formation of mulches or ridges on evaporation are not considered. Reflectivity of the soil surface for solar radiation affects its energy balance. Its value depends on the surface colour and moisture content of the upper layer (Menenti, 1984). The values for the albedo or whiteness of a dry soil surface runs from 0.15-0.4 (Table 28). The dependence on soil humidity is represented simply as a negative proportionality (Listing 7 Line 44). Although not entirely correct, the average water content of the upper soil layer is used to calculate reflectivity. The surface emissivity for long wave radiation is between 0.9 and 158

Table 28. Albedo values for wet and dry soils.

Surface type

Wet

Dry

Dune sand Sandy loam Clay loam Clay

0.24 0.10-0.19 0.10-0.14 0.08

0.37 0.17-0.33 0.20-0.23 0.14

Source: ten Berge (1986)

1.0 for all moisture contents (ten Berge, 1986). In SUEVTR its value is set equal to 1.00. The sensitivity of crop growth to both surface characteristics is fairly low. The evaporation rate diminishes as soon as the topsoil starts drying. In L2SU, this is assumed to happen the day after the last rainy day, the latter being defined as a day with at least 0.5 mm of precipitation (Listing 7 Lines 57-58). The reduction of the evaporation rate over time is mimicked by using the observation that cumulative evaporation is proportional to the square root of time (Stroosnijder, 1982; see also Subsection 5.3.6; Listing 8 Lines 50-51). The evaporation proportionality factor (the rate on the first day of this sequence) is assumed to be equal to 60% of the potential soil evaporation. Rains too small to trigger resetting of days since the last rain are added to the evaporation, since they are assumed to be lost the same day. All soil layers contribute to evaporation, but the top layer considerably more than the bottom layer (see Figure 59, Subsection 5.2.1). The actual partitioning at any moment depends on the depth and thickness of layers, their water content, and a soil specific extinction coefficient that is used to mimick the upward flow. Partitioning is calculated in Listing 8 Lines 52-61. The extinction coefficient is approximately 10 m-1 for heavy and 30 m-1 for light soils. The calculation of the potential rate of soil evaporation from incident energy is not correct if the soil is an important net source or sink of energy during a one-day period. This is rarely the case in the tropics, but in temperate climates the soil is a net sink of heat during spring and releases heat in the autumn or fall. The soil may also act as a source or sink if abrupt changes occur in air temperature or radiation intensity, such as those associated with the passing of large scale meteorological systems. This is significant to soil surface temperature and to the interpretation of thermal remote sensing imagery, but it is generally insignificant for crop growth modelling.

159

5.2.5

Water uptake and transpiration

Water uptake by the root system and partitioning over the rooted layers is described in Section 4.2 and depicted in Figure 59 (Subsection 5.2.1). The actual water uptake from soil layers is calculated in Listing 8 Lines 11-15, the effect of water stress on uptake in Lines 17-19, and the root depth in each layer in Lines 21-23. Maximum available water (i.e., all water held between field capacity and wilting point) is from 0.5-2.5 mm water per cm of rooted depth. If evaporation could be avoided, a C3 crop could produce 170-800 kg ha-1 total dry matter on the water stored in each 10 cm of rooted depth and a C4 crop about twice as much (Subsection 4.1.6). Obviously, water stored in the soil provides an important buffer in periods of little rain. Dry season cropping is, in fact, possible in many climates, provided that at the start there is a saturated soil profile and at least 0.5-0.7 m of rootable profile (see Subsection 5.4.6 for an example). A crop dies from water stress even before the lower soil layer reaches wilting point. The rate at which water is extracted near wilting point is so low that photosynthesis provides insufficient energy for maintenance respiration, and the crop dies. In simulation, water is always withdrawn from the entire soil layer as soon as roots enter it. The withdrawal rate is proportional to the root length in the layer. When the maximum rooted depth is reached, but roots do not extend to the bottom of the soil profile, water is slowly extracted from below the rooting zone.

5.2.6

Soil temperature

Soil temperature can affect the downward extension rate of roots. It is approximated by a simple running average of air temperature (Listing 7 Line 60). A more sophisticated treatment requires consideration of thermal soil properties, which are usually unknown. An example of a deterministic analysis is given by de Wit & Goudriaan (1978). A detailed simulation study is presented by ten Berge (1986). 5.3

5.3.1

A water balance module for soils with impeded drainage

Introduction

There are many poorly drained soils and many situations where the groundwater table is between 0.0-1.5 m below the surface during part of, or the entire growing season (e.g., Marletto & van Keulen, 1984). In soils where lowland rice is grown, downward water flow is usually impeded by a hardpan formed by puddling. A perched water table may then develop near the root zone. Capillary rise provides a significant water supply during dry spells. 160

When downward water flow is limited or groundwater is perched near the root zone, the L2SU module for water balance simulations should not be used. In these cases, the Simulation Algorithm for Water flow in Aquic Habitats (SAWAH) is more appropriate. The SAWAH subroutines are included in Listing B. The main subroutine, SUSAWA, makes use of several other subroutines to compute the rates of change of water contents in soil layers. These rates are integrated and their interaction with the crop is simulated in the soil water balance module L2SS (Listing 9). This module simulates the water balance for crop growth at Production Level 2 with partially saturated soils. All subroutines are presented in Appendix B. Technically, L2SS and L2SU differ: in the number of layers simulated, in the complexity of the flow computations, in the time period used and, consequently, in running time on the computer. L2SS plus SAWAH is more deterministic, but also requires more input data. L2SS is to be combined with L2C and L1D (Table 1). Simulation of the soil water balance processes becomes more accurate when the simulated soil layers are made thinner, and more layers are considered. The shape of the soil moisture profile is then approximated more precisely and soil and root properties can be defined in more detail. A disadvantage of thin layers is the associated short integration period in explicit numerical solution schemes (Subsection 1.4.2). This is due to the low capacity of thin layers (i.e., the amount of water that can be absorbed or released to reach a given change in potential). The deterministic simulation of soil water flow requires integration periods very much smaller than those for crop processes or parameteric simulation of waterflow as in the L2SU module. Redistribution of water in the soil during any one day resulting from sudden changes is simulated with short time periods. The L2SS module may therefore be used for a crop-soil system where such sudden changes occur once every 24 hours due to extraction of water from the soil by roots, due to precipitation or evaporation, and to a possible change in groundwater depth, all of which are simulated as instantaneous events.

5.3.2

Integration periods

L2SS with the SAWAH subroutines simulates water redistribution induced by root uptake, rainfall, evaporation and capillary rise. These processes can occur relatively fast. Simulation of water flow requires time periods of 0.0010.1 day, depending on the thickness and hydraulic properties of the soil layers. Thus, SAWAH covers every one-day period of the crop in many small steps. The resulting changes are the rates computed in L2SS. Interaction between the crop and the soil occurs every 24 hours at sunrise when the simulation day begins. The integration period in SAWAH can be explicitly chosen (parameter DTFX), but it is recommended that it is established by SAWAH. Minimum 161

and maximum time periods should be specified (the minimum DTMIN and maximum DTMX, Listing 10 Line 15). The actual integration period is the smallest of those specified and computed. When a fixed time period is chosen, SAWAH does not check whether the time period is larger than the allowed time coefficient. However, it does sometimes reduces the prescibed time period because it cannot exceed the time required to saturate a layer or to remove all its available water. A maximum time period is specified to ensure that at least a minimum number of repetitions occur each day. The maximum time period overrides the fixed time period if the latter is too long. The last integration period of each day completes the 24 hour cycle.

5.3.3

Soil water movement: unsaturated flow

The daily rate of change of water content of each soil compartment is an output of the SAWAH submodel. Water exchange between soil layers is governed by the hydraulic conductivity of the soil and the local gradient of the hydraulic head. Water flows towards the location with the lowest hydraulic head. To keep the calculation time within acceptable limits, the number of layers for a soil profile of 0.7-1.5 m is limited to 10. The upper layer should be at least 0.05-0.10 m thick (Subsection 5.3.6); the thickness of other layers may vary. All layers have their own pFcurve and hydraulic conductivity function. The use of only 10 soil layers implies that these layers are relatively thick and that the simulation of the soil water profile is somewhat crude. This can be viewed as a price for faster computation, but in practice the soil data needed to characterize many soil layers individually are often not available. The inaccuracy associated with the use of thick layers is reduced by applying the concept of ‘matric flux potential’. Consider the flux density or Darcy equation, which expresses the flux q (cm d-l ) as a function of hydraulic conductivity k (cm d-1) and the gradient of the hydraulic head H (cm): Equation 4 where z (cm) is the space coordinate, positive downwards, –h (cm) is the matric suction, -k (dh / dz) is the matric component of the flow, and k is the gravity component. In numerical simulation of flow processes, differentials such as dh, are replaced by differences, such as h(i) – h(i-1), the index referring to a layer number. A difference such as this is called a ‘finite difference’ to stress the contrast with ‘differential’, an infinitely small increment. A finite difference is represented here by the symbol . This distinction is the basis for several ‘tricks’ applied in numerical simulation. One of these is the use of the matric flux potential instead of hydraulic conductivity. The present context does not allow 162

elaboration on the theory and only the most relevant equations are presented (for theory and application of matric flux potential, see Klute, 19.52; Gardner, 19.58; Raats, 1970; Shaykewich & Stroosnijder, 1977; for measurements see ten Berge et al., 1987). The matric flux potential is defined as: Equation 5 It can be substituted into the flux density equation: Equation 6 Because it is assumed that the flux between the centers of two adjacent compartments is constant with depth, the integrated form of Equation 6 is: Equation 7 The second term on the right-hand side is the gravity term. Its value can be approximated by assuming a linear course of the matric suction between the centres of adjacent compartments i and i-1: Equation 8

Since this only applies to the gravity term it is not a severe simplification of the overall flow process. The flux density equation is then rewritten as: Equation 9

The integral on the right-hand side is identical to FD . The term 1 / D h represents the gravity component and 1 / D z the matric component of the flow. The matric flux potential ‘weighing’ therefore also applies to the gravity term. A complication arises when intrinsic soil properties, such as hydraulic conductivity and pF-curve, change with depth, as is often the case. The gradient of the matric flux potential has a straightforward meaning (i.e., it is a flux density) only if the soil material is homogeneous. An averaging procedure with values of two compartments is used for layered soil. The integral term of Equation 9 is then replaced by: Equation 10 163

k1 and k2 are the conductivity functions of the two adjacent layers. This is analogous to the procedure recommended by Vauclin et al. (1979), with an additional weighted averaging over the matric suction. After each time period the new water content in each layer is obtained by subtracting the outflow from the inflow during that period and adding the resulting change to the previous value of the water content. The matric suction in each compartment is assessed again for the new moisture content; the whole procedure is repeated in the next time period. The small simulation program SWD (short for: soil water dynamics, Listing 6) combines the basic processes of water flow in unsaturated soils, infiltration, capillary rise and redistribution. As a deterministic model, the program is based on the flux density equation and the mass conservation equation. The equations are solved for each set of two adjacent compartments and for each short time period. Soil water movement is the result of a gradient in total water potential. When flow across the upper and lower boundaries of the profile is permitted, as in Listing 6, water continues to flow as long as the difference between suction and gravitational head is not constant throughout the profile. With fixed suctions at both boundaries, a steady state profile finally develops, characterized by the same flux across all compartments. SWD can be modified to block drainage or evaporation (set fluxes at bottom or top of the profile to zero), or to extract water from layers (e.g., to mimick transpiration). SWD cannot handle non-homogeneous profiles (and this cannot be overcome by defining the parameter values KST, WCST, KMSA and MSWCA for each layer separately). Since it simulates only unsaturated conditions, saturation of one or more soil compartments cannot be handled. However, the program can be used for evaluating complex models when these are applied for simplified sets of inputs. The program is the basis for exercises at the end of this chapter. To prevent problems with rounding off when using Personal Computers, Line 33 resets K(I) to zero when it approaches that value.

5.3.4

Soil water movement: saturated flow

Saturated soil sections (a section consists of one or more layers) have different solutions for the flow equation from those with unsaturated flow. In a way, this situation is less complex, because the dependence of the transport coefficient on moisture content has vanished and at saturation only hydraulic conductivity is relevant. On the other hand, the pressure gradient at layer interfaces can no longer be calculated from the moisture contents of neighbouring layers. The entire set of saturated layers is now involved, because the weight of the overlaying water contributes to the local water potential in every layer. Moreover, different soil layers may have different values for saturated conductivity, so that the conductivity of an individual layer no longer determines the saturated flow through that layer. The product of local pressure gradient and local conductivity must now be equal at every point in the sat164

urated section, since no changes in moisture content can occur. Only at the edges of the saturated section may moisture content change, related to expansion or contraction of the particular saturated section of the soil profile. To simulate saturated flow, first the saturated sections of the profile are identified (Figure 60). At the saturated – non-saturated transitions, the matric suction is supposed to be zero. This usually closely approximates the real situation. Suction is not always zero when the boundary of a saturated layer coincides with the top or bottom of the soil profile. The pressure may be positive at the top of the profile, for water may be ponded on the surface. At the bottom of the profile the pressure depends on the depth of the groundwater table. If groundwater is present within the modelled part of the soil profile, the water pressure at the lower boundary is positive. For deeper groundwater tables, the pressure at the base of the profile is negative. The total change in hydraulic head over a saturated section is calculated by adding the drop in gravitational head to the difference in pressure head between two edges. For n saturated layers within a section, a total of n + 1 unknowns must be solved: n times the difference in hydraulic head over a layer, plus the flux through the saturated ‘package’. There are n + 1 independent linear equations; n times the flux density equation over a layer and the

Figure 60. An example of water redistribution in a layered profile as simulated with the L2SS + SAWAH modules. Hatched zones indicate saturation. Low conductivity in Layers 5 and 9 cause a double perched water table and ponded water on the surface. The pressure head HP at the top of the profile is then positive (HPTP), at the bottom it is negative (HPBP) and it is zero at the other edges of saturated sections. Arrows indicate the flow direction (single-headed arrows) or possible flow directions (doubleheaded arrows).

165

summation of changes in hydraulic head over each layer, which yield the total difference in hydraulic head over the entire saturated section. This set of n + 1 linear equations is written in matrix form and solved for each saturated section of the profile. In most cases there will be only one saturated section, though separate saturated sections can occur after intermittent heavy showers or when the soil has several slightly permeable layers. Groundwater depth is a boundary condition and an essential input to the model. Its value should be obtained by direct observations in the field for which simulations are made (see also Subsection 5.3.8).

5.3.5 Infiltration and runoff Infiltration depends on rainfall intensity, soil properties and surface storage capacity. The amount of rain (or irrigation water) that cannot infiltrate or be stored on the surface is lost as runoff. Usually, only daily precipitation total, not the intensity, is known. Rainfall can be constant for 24 hours, or may come in one short but heavy rainstorm. Where water is normally received over brief periods, e.g., when short showers predominate and on irrigated fields, the extreme of instant supply comes closer to reality than the other extreme of a constant supply. In SAWAH it is assumed that all daily precipitation or irrigation water is usually received instantaneously at the start of the day. A water layer corresponding to the daily total of rainfall is then ponded on the surface. The rate with which the thickness of this ponded water layer changes is an output of SAWAH; the thickness of the layer is a state variable in L2SS (WL0QT, Listing 9 Line 50). Infiltration is simulated with the equations for unsaturated waterflow (Subsection 5.3.3) including both matric and gravity forces, as long as the surface compartment is not saturated. The effect of positive water pressure is taken into account. This hydrostatic pressure head is due to the weight of the free water layer and is equal to the depth of that layer. When the surface layer becomes saturated, the model switches to the equations for saturated flow, again taking into account the positive surface pressure. If, after a day of infiltration, not all rain has entered the soil, the remainder is left ponded on the surface. When the amount of ponded water exceeds the surface storage capacity, the excess is considered as runoff. The maximum amount of water that can be stored on the surface (WL0MX, m) can be computed from surface characteristics (Driessen, 1986a):

Equation 11

166

where d is surface roughness (cm), s is the clod/furrow angle (degree), j is the average slope angle of the land (degree). These characteristics are illustrated in Figure 61. The clod/furrow angle is normally between 30 and 45 degrees, the slope angle of the field in general does not exceed 17 degrees ( = 30%), while the surface roughness or furrow depth depends on the way the land is cultivated: about 0.2 m for contour plowing, 0.06-0.08 m for tillage with light equipment, and 0.01-0.02 m for untilled soil (Driessen, 1986a). Equation 11 is not relevant for bunded fields. For rice, the effective bund height is numerically equal to the surface storage capacity. The surface storage capacity is not calculated in the L2SS module, but is given as a situation specific parameter. Infiltration is computed with rain as input; runoff is an output (Listing 10 Lines 54-58).

5.3.6

Evaporation

Computing the actual soil evaporation is complicated in SAWAH in comparison with the ‘square-root-of-time relation’ used in L2SU (Subsection 5.2.4). The simple equations cannot cope with significant capillary supply from groundwater. Moreover, they do not deal with root water uptake. The formulation below implicitly incorporates both processes.

Figure 61. A schematic representation of the surface storage capacity of a field, WL0MX. WL0MX is equal to the shaded area divided by the length X. (Source: Driessen, 1986a).

167

The water loss rate is equal to the potential evaporation rate in saturated soil (Subsection 5.2.4). The evaporation rate does not depend on soil properties or soil condition. Topsoil drying is associated with a decreased evaporation rate, for water then evaporates below the surface and water vapour moves upward by diffusion. Formation of a dry surface layer gradually impedes further water loss. The rate of topsoil drying, and hence the rate at which evaporation decreases, depends on soil characteristics, water content and on the potential evaporation rate. Calculating the development of a dry surface layer is crucial for an accurate determination of soil evaporation. There are two solutions for calculating soil evaporation in a deterministic way. The first is a numerical solution by simulating the flow processes involved; the second is an analytical solution. A purely numerical solution requires simulation of very thin surface layers to account for the steep gradient in the soil moisture content; this is impractical. Analytical solutions, however, are only available for homogeneous soils with uniform initial conditions. Therefore, a combined approach is necessary. A basic assumption is that an ‘evaporation front’ exists at the surface or at some depth zE in the soil. This is a sharp transition where liquid water is transformed into vapour. All transport is in the liquid phase below the front, while only vapour transport occurs above the front. The evaporation rate at any moment is a function of the depth of the evaporation front and of the rate of vapour diffusion through the dry layer. The balance of liquid supply to the front, and vapour diffusion away from it, determines the movement of the front itself. At a front depth zE (cm), the rate of water vapour loss from the soil (e, cm d-1 ) can be expressed as: Equation 12 c2 is taken to be a soil-specific constant, though its value depends somewhat on air humidity, wind speed and soil temperature. To compute the supply of liquid water to the evaporation front, some assumptions made by Parlange in describing infiltration can be applied to the description of evaporation (Giraldez, personal communication; for details, see Parlange (1971) and for applications, Smith & Parlange (1978); Giraldez & Sposito (1985)). The amount of water that has evaporated after a given period of time (E, cm) is given by:

Equation 13

168

where is the water evaporated from above the front, c1 /e is the amount of water extracted from below the front, c1 is a function of the initial moisture content the final moisture content (air dry) of the top layer and the (liquid) soil water diffusivity D (which is a function of moisture content q itself) : Equation 14 (‘Initial’ in this paragraph refers to the water content at the start of the day, and not to the value at the start of the simulation). Combining Equations 12-14 and further developing the terms leads to an expression of the rate at which the evaporation front sinks into the soil: Equation 15 The front depth is regarded as a state variable (Listing 10, Line 52); its change over a day is calculated in SAWAH. c3 is defined as: Equation 16 zE increases rapidly if c1 is small compared to c2, i.e. when vapour diffusion is rapid compared to the supply of liquid water from the subsoil. zE also increases rapidly when the difference is small. A high initial moisture content also increases the value of c1, as it increases the upper boundary of the integral in Equation 14. A high initial moisture content, therefore, results in low c3 values and in the slow progress of z E . The value of z E is reset to zero when the topsoil is saturated by rain or rising groundwater. The integral in Equation 14 is calculated as an explicit exponential function of ,q : Equation 17 The parameters A and B are constants for the soil, c4 is about 0.5 (dimensionless) for all soils, e is the base of natural logarithm, and q s is the water content at saturation. The above concept was developed for an ‘ideal’ soil, i.e., deep, with uniform initial moisture content. It can also be used as an approximation for a layered soil provided that the top layer is at least 0.05-0.10 m thick. c1 and c3 are kept constant during each day, with q en q i referring to the toplayer. The Equations 12, 15-17 are incorporated in the module T12 (Appendix B, Subroutine SUZECA), and the parameters used are (Listing 10 Line 19): c2 (about 0.1 cm2 d-1 ), A (0.005-0.5 cm2 d-1 ), B (5-15, dimensionless). Realistic combinations of A and B should give c1 values ranging from 5.0 to 50.0 cm2 d-1 with exceptions up to 400 cm2 d-1 . Coarse soils tend to have high B values combined with low A values. Low B and intermediate A values are recom169

mended for silty soils, and low A and low B values for heavy soils. (Note that the units of z E and e in Equations 12-17 are not equal to the units of the corresponding variables ZEQT and EVSW in Listing 12.)

5.3.7

Transpiration

Actual water uptake from the rooted soil is almost the same as described in Sections 4.2 and 5.2 (Listing 9 Lines 64-72) though its appearance is different (see Subsection 5.4.4). The same relation of water stress and water uptake as in the module L2SU applies here (Line 68), but there is a difference in the possibility of anaerobic conditions developing in layers. The extent of anaerobic conditions is assumed to be proportional to the water content between field capacity and saturation. Anaerobic conditions reduce the capacity of the root system to extract water from the soil in most species (Subsection 4.2.4).

5.3.8

Flow to and from the groundwater

The pressure head at the bottom of the profile is equal to the distance between the bottom of the profile and the water table. The pressure head is negative when the groundwater table is below the profile. The soil below the profile is then assumed to be in equilibrium with the water table. The corresponding matric suction at the lower boundary is then used to calculate the matric flux potential at this depth; the flux (upwards or downwards) is subsequently calculated from the difference in matric flux potential between the lower boundary and the centre of the lower layer. Using pressure head at the bottom of the profile as a boundary condition has an advantage over introducing the ground water level directly. This is because the depth of the water table is often measured in piezometer tubes where water transport is faster than in the soil, particularly in soils with low conductivity. Thus, a soil section below the ground water table may be non-saturated for a while after the level in the piezometer has moved upwards to ‘pass’ that section. The reverse may also occur, leaving the soil saturated, although the piezometer readings show that the ground water level dropped. However, such data are rarely available for crop growth studies. The profile simulated need not be deeper than the bottom of the tube in which the groundwater level was measured. A concious choice of soil profile depth saves much computer time and the results will be almost unaffected.

5.3.9

Lateral flow

Lateral inflow or outflow of water can be significant for the water balance of small fields. The extent of lateral flow depends on the position of the field in the landscape and its distance from ditches and canals. Lateral flow occurs mainly in saturated soil layers. In contrast to rainfall, 170

lateral flow is very difficult to measure in experiments. Angus & Zandstra (1980) demonstrated the principle of simulating lateral flow between a cascade of rice fields, but soil data are lacking to apply it in actual situations (Whisler, 1983). Net lateral flow leads to a change in the groundwater table. Since the water table is an input into the water balance module, the net contribution of lateral flow to the soil water balance is implicitly accounted for and needs no further consideration. 5.4 Simulation modules for water balance and crop growth

5.4.1

Simulating the effects of water stress (module L2C)

The effects of water stress on crop growth and development can be simulated with the module L2C (Listing 7) for crop-related calculations at Production Level 2. The scientific basis for the module was presented in Chapter 4. Abbreviations used are explained in Listing 12. L2C is to be combined with the basic module (LlD), a soil module (L2SU or L2SS), and crop, soil, and weather data (such as that in Listings 5, 10, and 11 respectively), and with the terminal module T12 (Appendix B), as indicated in Table 1 in the Reader’s guide. The L2C module provides the soil module with data on the potential transpiration rate, rooted depth, potential transpiration per cm rooted depth, leaf area and crop height. The L2C module requires from the soil module the rate of water uptake (actual transpiration) and the effect of soil temperature and water stress on root elongation. Lines 2-4 of Listing 7 are to be placed in the INITIAL of the full simulation program. The water stress level is quantified by the ratio of actual transpiration and potential transpiration. Water stress always affects photosynthesis (Line 11), carbohydrate partitioning once the ratio is less than 0.5 (Line 9), and crop development depending on its sensitivity (Line 10). Air humidity may reduce photosynthesis independently of water stress (Line 12). Weather data (discussed in Chapter 6) are incorporated in Lines 51-59. Soil temperature is approximated by a running average of air temperature (Line 60). The DSLR variable (Line 57, 58) keeps track of the number of days since the last day with at least 0.5 mm of rain (or irrigation, if applicable). It is used as an indicator of potential or less than potential soil evaporation in the L2SU module. If no rain was received DSLR is also used to calculate the actual evaporation rate for free-draining soil (Listing 8 Lines 47-51). Statements to compute potential soil evaporation (Lines 40-47) are added to this crop module because they apply to both soil modules. To compute potential soil evaporation, the SUEVTR subroutine is called (Line 41, 42). (The ‘leaf resistance’ in this statement has no equivalent for soil and is set to 0.0. Soil evaporation continues at night, so that the fraction of the day is set to 1.00 (in 171

contrast to canopy transpiration) and the average air temperature is used. The SUEVTR subroutine and FURSC function are explained in Subsection 5.4.5.) Two auxillary variables are created to ease checking results, or to produce characteristic values for water use efficiency in Lines 62-63. The following crop data are required (see Listing 5): – for water uptake: the sensitivity of the species for water stress and flooding (Line 42); – for transpiration: leaf width, crop height (as a function of development stage) and the internal/external CO2 fraction (Lines 37, 39, 42); – for crop development: the relation between stress and development rate (Line 36); – for root growth: the maximum depth and growth rate (Line 43). The effect of temperature on root growth is assumed to equal that on photosynthesis.

5.4.2

Two soil water balance modules

The L2SU module to simulate the water balance of free-draining soils for crop growth at Production Level 2 is an alternative to the L2SS module for soils with impeded drainage. The choice of module depends on the objective of the study, on environmental conditions and on the availability of soil data. Sections 5.1, 5.2 and 5.3 provide the background of the processes. Abbreviations used in the water balance modules are defined in Appendix A. Both soil modules use the potential rate of transpiration, rooting depth (and the ratio of both), leaf area and crop height, calculated in the crop module L2C as inputs. As with the check on the carbon balance (Subsection 3.4.4), there is a check on the water balance at each time period. Both soil water balance modules contain a number of statements that must be placed before or in the INITIAL section of the total CSMP program before running it. They are presented here, together with the DYNAMIC parts for reasons of clarity.

5.4.3

Module for free-draining soil (L2SU, SAHEL concept)

The water balance module of a soil with three layers, freely draining at the bottom is contained in L2SU, Listing 8. The numbers of layers can be extended by repeating the separate statements for new layers (the calculations of CKWFL and CKWIN must then be adapted). The upper soil layer should be between 0.05 and 0.3 m thick and the other layers between 0.2 and 0.5 m. The simulated soil layers may correspond with actual layers in thickness and characteristics. For accurate calculations it is a disadvantage to include soil layers below the rooted depth. Three soil characteristics per layer are specified: water content at field capacity, wilting point and air dry (Subsection 5.1.3). An example is given in Listing 10. The degree of water stress is calculated per layer with the FUWS function (see Figure 55, Subsection 4.2.2). Stress in any layer affects water uptake 172

(Lines 12-14); but only stress in the deepest layer where root tips are located affects the rate of root growth (Line 27). Rooted depth is calculated per layer by repeated subtractions (Lines 21-23). A limit condition ensures that roots do not grow beyond their maximum depth (Listing 8 Line 32). The actual daily evaporation, obtained with Lines 47-51, is equal to the potential rate on days when there is at least 0.5 mm of rain. After rain, evaporation diminishes proportionally with the square root of time (Lines 50-51). The partitioning of actual evaporation over layers is according to an exponential extinction with depth, assuming the centre of the layer to be halfway between its middle and upper boundaries (Lines 55-61). Water percolating from the deepest layer (Line 43) is only used to check the consistency of the water balance (Line 66). Factors 0.1, 10. and 1000. in the module are used to maintain proper units. Dividing by DELT converts amounts into rates (Subsection 1.4.3). The total amount of water in the soil profile is an auxillary variable (Line 39); not all the water is available to the crop. 5.4.4

Module for soil with impeded drainage (L2SS, SAWAH concept)

The water balance simulation module for soil with impeded drainage (L2SS) is presented in Listing 9. Rates of change of the water content of layers are computed by the subroutines that together represent SAWAH. In L2SS, these subroutines are all invoked through the subroutine SUSAWA (Lines 37-39, 54-56). The SAWAH subroutines are part of Appendix B. The L2SS module looks different from the L2SU module in three respects: some variables are indexed (they refer to arrays of similar variables for individual layers); some are within DO-loop (the computations are performed for I, the DO-loop runner, with values increasing from 1 to NL, the number of layers); and, the sequence of the statements is not affected by CSMP, but determined by the modeller. FORTRAN is used because repeating lines for 10 layers (as when using CSMP) is clumsy and makes the program too large to conveniently read or modify. The L2SS module also looks different from the L2SU module because most computations are performed by a single large subroutine (SUSAWA) which calls many other subroutines. Indexed variables must be placed within PROCEDURES. CSMP sorts the contents of the entire PROCEDURE as a single statement, assuming that the arguments to the left of the equal sign are outputs and to the right of the equal sign are inputs. For further implications, check the CSMP manual (IBM, 1975). Place statements related to organization of the module (STORAGE, FIXED) in the INITIAL of the crop part before running the program; those beginning with ‘/’ (in column 1) link subroutines and the main program with each other and must be placed immediately below the STORAGE statements. The number of layers (NL, often 10) is explicitly given in the calls for the 173

subroutine SUSAWA. The module can be used with a smaller number of compartments by lowering the value of NL. No changes are needed in the STORAGE statement (Line 4, 5) in the call for the COMMON blocks (Lines 6-9), or in the dimensioning of variables in the subroutines. NL should be two or more. Increasing the number of layers above 10 will rarely be necessary and cannot be done without adapting subroutines and dimensions of variables. The soil water contents of layers are initialized in SUSAWA in the INITIAL. This is identified by the first input being 1 (Line 38). Many arguments in the subroutine are not now used, so that the list of input arguments for sorting the PROCEDURE that carries SUSAWA has been shortened. When a switch parameter (WCLIS) has a value –1.0, initial water contents are set at values in equilibrium with the soil water table (Line 17). (If user specified water contents are preferred, set the switch to 1.0 and the initial soil moisture contents from TABLE WCLMQI(1-10) are then effective). The DYNAMIC section seems short. The rate of water uptake is defined here as in the L2SU module (Subsection 5.2.5), but in FORTRAN style (Lines 64-72). No more water than is available is extracted for transpiration in any one time period (Lines 70, 71). The effect of water stress on root growth is calculated in Lines 73 and 74. The majority of the soil water balance calculations are performed in the SAWAH subroutines, i.e., the unsaturated and saturated flow in all layers. Changes in water contents of the layers, in the level of ponded water, and in the evaporation front depth, are all rate variables that are integrated in L2SS to yield the new values of state variables after every day. Data characterizing the soil are indicated by specifying the soil type. Indicative values for common types are given in tables KMSAlT, KMSA2T, KMSMXT, KSTT, MSWCAT and WCSTT of Listing 11. The numbers in table TYL (type of layer) specify the soil type for subsequent layers. Other soil types can also be used by replacing some of the present values. The module requires the observed (or estimated) depth of the groundwater table as an input and forcing function (Lines 22, 57). The water table depth has a negative value when above the surface, and positive below it. The groundwater table is interpreted in SAWAH as a piezometer pressure at the lower profile boundary.

5.4.5

Functions and subroutines

A number of functions and subroutines for handling the soil water balance are given in Appendix B. They are described here briefly, in alphabetical order. Some are relatively complex, but crop growth modelling does not require understanding of the exact procedure of computations. All subroutines beginning with SUST are only used for saturated soil compartments. Most names in the first lines of the subroutines and functions are defined in Appendix A. Note that the L2SU module uses fewer subroutines than L2SS. 174

User defined functions look like other CSMP statements in the main program. In defining them, however, the output name does not appear, only the name of the function itself. Calls for subroutines in the main program are statements with several inputs and outputs (left of the equal sign). In defining the subroutines, all outputs follow the inputs on the right-hand side. The order of arguments should not be changed. The FUCCHK function is explained in Subsection 3.4.4. The FUPHOT function is explained in Subsection 3.4.4. The FURSC function calculates canopy resistance. It is simple and will be adequate in many situations. Inputs are windspeed, leaf area, crop height and the height of windspeed measurement (Subsection 6.1.6). When more emphasis is put on micrometeorology, another function may be needed (cf., Goudriaan, 1977). The FUVP function is explained in Subsection 3.4.4. The FUWCHK function checks whether the integral of all fluxes into and out of the soil, corresponds to the change in water content of the entire profile since the beginning. When the relative difference between both exceeds 1%, a warning is printed. The program should then be checked for errors made while modifying the program. The FUWCMS function calculates the volumetric water content at a specified value of the moisture suction. The FUWRED function calculates windspeed near the soil surface under a leaf canopy, using equations based on Goudriaan (1977). Inputs are leaf width, total leaf area, crop height and windspeed at reference height. The FUWS function quantifies the degree of reduced water uptake in a layer due to stress. It requires as inputs (in the following order), the potential transpiration rate of the canopy, leaf area, water content of the soil layer, sensitivity coefficients for water stress and excess, and the soil water content at wilting point, field capacity and saturation. FUWS can be used for all soil water contents; its value runs from 1.0 (no stress) till 0.0 (see Figure 55, Subsection 4.2.2). Stress is generally due to water shortage, but can also result from excess water. The SUASTC and SUASTR subroutines are explained in Subsection 3.4.4. The SUCONV subroutine converts units of variables between the main program and subroutines. The direction of the conversion is governed by the first argument, all others are variables to be converted. The SUERRM subroutine is used in many subroutines to check whether variable X has a value between a reasonable minimum and maximum value. If not, an error message is written to a file with unit number NUNIT (6 for FOR06.DAT in CSMP) and the program is stopped. If XMIN or XMAX equals -99.0, no minimum or maximum is effective. Appendix A provides a list of the error messages. The SUEVTR subroutine computes two evaporation rates of a canopy or soil surface from a few surface characteristics and weather data. The first rate is 175

related to the daily total radiation, the second to the drying power of the air. Both refer to a single surface. The subroutine is used to compute potential canopy transpiration and potential soil evaporation. Inputs (in this order) are: radiation on a fully clear day, the measured radiation for that day, the reflection coefficient of the surface for solar radiation, the fraction of the day that evaporation is significant, the average temperature of the evaporating surface, the air humidity, and the leaf, boundary and canopy resistance (see also Subsection 6.2.4). The incoming long wave radiation is calculated from the average daytime temperature. Crops only transpire during the day, so only the daytime fraction of thermal radiation need consideration. The SUGRHD subroutine determines the gravitational head at the top and bottom of layers. Only thickness and the number of layers are required as inputs. The SUINTG subroutine performs rectangular integrations to update the water contents for small increments of time during the one-day time period. It also determines the size of these increments (DT). Inputs are the rates of change of water content in each layer, soil moisture contents and parameters for the minimum and maximum size of the time period. Outputs to SUSAWA are updated state variables (WCL, WL0); SUSAWA then returns the total change of each state variable over that day to the L2SS module for real integration. The SUMFLP subroutine calculates the matric flux potential for a layer with a given water content and soil properties. SWICH3 = 1 computes the matric flux potential using analytical integration. The equation used must be adjusted if the hydraulic k(h) relation in SUMSKM is modified. A numerical Gaussian integration can be chosen (SWICH3 = 2) as an alternative if no analytical solution exists or is feasible. See ten Berge & Jansen (1989) for further details. The SUMSKM subroutine calculates the hydraulic conductivity for a layer from its suction. SWICH3 has the same meaning as in SUMFLP. The k(h) relations in SUMSKM and SUMFLP must always correspond. The SUPHOL subroutine is explained in Subsection 3.4.4. The SUSAWA subroutine is the key subroutine for soil water balance. It is used in the INITIAL (first input argument is 1) to determine initial water contents, and in the DYNAMIC (first argument is 2) to compute rates of water redistribution for saturated and unsaturated layers, for change in the thickness of the surface water layer and for change in the evaporation front depth. Many inputs are not used in the first call; these are therefore not included in the PROCEDURE statement (Listing 9 Line 14). In the second call, all variable inputs are used (Lines 54-56). SUSAWA causes rates to be integrated with short time periods (in SUINTG); it maintains its own time scale, TIMTOT, which goes from 0.0 till DELT for each time period of the simulation program. Rain is ponded on the surface immediately after the subroutine is called, and root water uptake and evaporation are then extracted. Water is subsequently redistributed between layers for the remainder of the 24 hours. SUSAWA uses 176

SUZECA to compute the depth of the evaporation front and derives from it the maximum allowable rate of soil evaporation. Inputs to SUSAWA are: a switch argument; water contents of the layers and ponded water; number of layers; variables: crop water uptake, potential evaporation, rain and the soil water table depth; and constants: layer thicknesses and soil types, integration period of crop simulation (DELT, here always one day), the minimum, maximum, and fixed time period for SUSAWA, the maximum level of ponded water and three evaporation constants. Outputs are: the change in water content of the soil layers, of the ponded water level, the actual soil evaporation rate, runoff, daily drainage or capillary supply to the bottom of the profile, the rate of change of all water in the profile and of evaporation front depth, and the total profile depth. The SUSEFL subroutine selects between fluxes calculated at the boundaries of saturated sets of layers in the saturated and non-saturated soil. Inputs are indexes of saturated layers and potential fluxes. Output is the array of net flows into layers (FLX). The flux through a saturated set (from SUSTFL) is usually not equal to the flux at the boundaries of the set (from SUUNST). It is assumed that the fluxes across the outflow end of the saturated set and across the internal interfaces are those computed by SUSTFL. The SUSLIN subroutine is called in SUSAWA and calculates certain soil characteristics from the basic data. Inputs to the subroutine are soil types and layer thickness, number of layers and groundwater table depth. Outputs are the soil water contents of layers in equilibrium with the water table and the depth of the upper boundary of each layer. The subroutine checks whether the soil characteristics are between reasonable extremes, and writes them into the CSMP output for inspection by the modeller. The SUSTCH subroutine searches for saturated layers in the soil profile. It organizes the saturated layers into JTOT sets of ‘continuous’ saturation, each consisting of JJTOT(J) layers. Each set is sandwiched between two unsaturated layers, or between an unsaturated layer and the top or bottom of the profile, or between the top and bottom of the profile. SUSTCH provides the indexes of saturated layers in the array INXSAT. The only inputs are the water content and number of the layers. The SUSTFL subroutine calculates a tentative flux through each set of saturated layers, as allowed by the total hydraulic head jump over the set and the saturated conductivities of each layer. This is achieved by solving a system of JJTOT(J) + 1 linear equations for set J of JJTOT(J) layers. These consist of JJTOT(J) flux equations and one summation of hydraulic head intervals to the total jump (DHH). The solution is obtained by invoking the matrix decomposition subroutine SUSTMD and subsequently solving it with SUSTMS. The output array FLXSTT contains the computed saturated fluxes for SUSEFL. These values are still tentative, since gradients and conductivities in neighboring unsaturated layers must permit these rates to occur. The SUSTHH subroutine identifies the difference in total hydraulic head 177

(DHH) between top and bottom of that particular set for each continuous set of saturated layers. If a saturated set is sandwiched between unsaturated layers, the matric suctions are defined as 0.0 at the saturated-unsaturated boundaries. If one or both boundaries are not free, the matric suction is replaced by the pressure heads imposed at the relevant interfaces. This occurs when a saturated set extends up to the surface and water is ponded (WL0 positive), or with a true groundwater level in the profile (HPBP positive). Other inputs are the array INXSAT indicating the saturated JTOT sets of JJTOT layers, number of layers, gravitational heads at the bottom and top of the layer, and water standing on the surface. The SUSTMD subroutine decomposes the matrix A into upper and lower triangles to allow a rapid solution by the SUSTMS subroutine. The initial elements of A are the coefficients of the original set of linear equations defined in SUSTFL, but are replaced by transformed coefficients for SUSTMS. (Source: LUDCMP subroutine by Press et al. (1986) p. 35-36.) The SUSTMS subroutine solves the set of N linear equations Ax = B. Here A is input, not as the matrix A, but as its upper-lower triangular decomposition determined by the SUSTMD subroutine. INDX is input as the permutation vector returned by SUSTMD. B is input as the right-hand side vector B, and returns with the solution vector x. A, N and INDX are not modified by this routine and can be left in place for successive calls with different right-hand sides B. This routine takes into account the possibility that B will begin with many zero elements (LUBKSB subroutine by Press et al. (1986) p. 37.). The complex SUUNST subroutine calculates tentative fluxes into and out of unsaturated layers. The calculation is based on the matric flux potential concept. Pressure is assumed to be 0.0 at the interface of an unsaturated and a saturated layer (see also SUSTFL) and only the matric flux potential of the unsaturated compartment is required. Fluxes over the interface between two layers are determined by both gravity and matric terms. Inputs are: the choice of equations constituting the k(h) relation (SWICH3 = 1 or = 2); the indication that suction is to be computed from soil water content, or vice versa (SWICH4 = 1 or = 2); the water content of layers; thickness and number of layers; the level of ponded water; and the hydraulic head. Outputs are the fluxes between unsaturated compartments and the moisture suction of layers. If the difference in matric suction and matric flux potential across a layer have the same sign, the product is set to zero to suppress rounding errors in the Gauss integration. The SUWCMS subroutine calculates the matric suction for a layer from its volumetric water content with Equation 2 (for its limitations, see Subsection 5.1.3) if SWICH4 = 1. Otherwise, the reverse calculation is performed. The SUZECA subroutine determines by how much the depth of the evaporation front (ZEQT) changes in a day. This front sinks into the soil when the soil dries, and rises when capillary rise provides more water than the amount lost by evaporation. Its value is reset to 0.0 when rain exceeds potential evap178

oration. Inputs are: the soil water contents of the upper layer, the potential evaporation rate, precipitation, the net fluxes into and out of the first compartment, the amount of ponded water, the time period in the crop module (DELT) and three soil constants.

5.4.6

Two examples using SAHEL and SAWAH

An example using SAHEL is from an evaluation of the potential benefit of soya bean as a dry season crop after rice (Pandey, IRRI, personal communication). Experiments over the last few years show that soya bean can grow well on stored moisture plus incidental rainfall in Los Baños, the Philippines, and responds favourably to supplemental irrigation. To judge whether the crop could be introduced to local farmers, two agronomical questions were raised. What average yield (grain, fodder) may be expected and what is the yield variability due to erratic rainfall? Do soya beans respond well to irrigation, and if so, what is the optimal time or soil condition? To draw conclusions for an upland site, SAHEL was applied (combine L1D + L2C + L2SU + data + T12) and run for 23 years of actual weather data (sowing data and initial soil moisture were fixed for this example). Figure 62 shows that a mean grain yield of 1.1 t ha-1 can be expected. The CV is as high as 30% (assuming a random distribution of yields) due to erratic rainfall. With full irrigation the mean goes up to 3.5 t ha-1 and, more importantly, the CV decreases to 10%. Irrigation increases fodder yield from 3.7 to 4.4 t ha-1 and the CV decreases from 9 to 8%. Irrigation almost exclusively benefits the grain yield. (Economic evaluation should follow these results before recommending any such procedure to farmers.)

Figure 62. The cumulative frequency of pod + grain yields (filled symbols) and leaves + stems (open symbols) of a dry season soya bean crop in Los Baños, the Philippines, without irrigation (circles) and with irrigation (squares).

179

The second example uses SAWAH for simulating the contribution of water from a soil water table to the root zone. Capillary rise is assumed to be important in many subhumid areas in the dry season, e.g., in large areas of northeastern Thailand where rice and peanuts are produced in these conditions. The model consisted of the modules L1D + L2C + L2SS + data + T12. The soil and climatic data in this example apply to a large part of the Korat plateau near Khon Kaen. Its landscape undulates smoothly; a laterite pan at variable depths impedes water flow, allowing a shallow groundwater reservoir to develop during the wet season. In the dry season, water is supplied by capillary rise from this reservoir. The soil texture is relatively coarse, so the distance between the root zone and the soil water table is crucial to crop performance. The simulation starts November 1 with soil at field capacity, or wetter when the water table is less than 1.0 m deep. The soil profile chosen is typical for the area: a loamy fine sand on top (0.2 m) and a subsoil of a sandy clay loam (0.2-2.0) (Suraphol, Khon Kaen University, personal communication). The physical data corresponding with these soil types are given in Table 26. Different, but constant depths for the groundwater table are imposed. Weather data from the Khon Kaen station for 1986 show that there was no rain except on two days in November. Results are shown in Figures 63 and 64. The grain yield is almost 7000 kg ha-1 for a water table at the surface, drops quickly to 4000 kg ha-1 for a water table depth of about 0.5 m and decreases gradually to almost zero at water table depths of 3.0 m and more (Figure 63). This reflects the requirement of ample water for lowland rice. Rooting depth increases to 0.7 m when the water table decreases below 0.9 m. The yield – water table response curve is not smooth because of the interaction of rooting depth, different rates of evaporation and transpiration and the effect of water stress on uptake. The high value at 2.0 m occurs when the water table is exactly at the bottom of the simulated soil profile, which causes a deviation from the trend. No water from the groundwater reservoir reaches the root zone if the depth exceeds 3.0 m. A gradual drying of the topsoil is associated with reduced evaporation, but substantially less so for shallow, than for deep water tables. The water flux at the bottom of the profile can be substantial and varies with the water table (Figure 64). No hard evidence is available to confirm these results; indeed, they are very hard to obtain other than by simulation. The range of simulated rice yields agrees with observations in the region on well-fertilized rice fields with minimum disease and pest levels.

180

Figure 63. The simulated response of the grain yield (paddy rice, zero moisture) of a dry season rice crop in northeastern Thailand to a constant depth of the ground water table on a sandy clay loam below a loamy fine sand top soil. Almost no rain is received during the growing season.

5.5

Exercises

Use the data presented in the introduction to the excercises of Chapter 2.

5.5.1 Soil characteristics T1. Classify the texture of the following soils: 38% sand, 20% silt and 42% clay; 10% sand, 70% silt and 20% clay; 60% sand, 30% silt and 10% clay. T2. Draw the pF-curve for medium-fine sand and for clay loam. Use data from Table 26. T3. How much is the pressure head that corresponds with a suction of + 100 mb, +2 kPa and +1 bar? T4. Estimate the amount of available water that can be stored between wilting point and field capacity in a profile 1 m thick of (a) fine sand, and (b) clay loam. Use Figure 57 and assume field capacity to be at pF = 2. Which soil has highest amount of water available if field capacity is at pF = 2.5? T5. Draw the equilibrium volumetric water content versus depth, for a clay 181

Figure 64. The capillary rise of water into the soil profile to a dry season rice crop in northeastern Thailand, as a function of soil water table depth of 0.0-3.0 m.

loam with the water table at 1.0 m and at 2.5 m. Do the same for a fine sandy loam. In which soil is the largest amount of water stored above the groundwater table? (Remember in equilibrium the suction gradient is 1 cm/cm and the suction is zero at the groundwater level.)

5.5.2 Free-draining soil T6. Make a relational diagram of the processes and variables which determine the water balance of free-draining soil. Indicate the interactions between crop and soil. T7. Write CSMP statements to make the runoff fraction dependent on rainfall intensity (mm d-l ) and on volumetric water content in the first layer. Assume that an amount of 10 mm d-1 can infiltrate at field capacity and beyond. This amount increases linearly with decreasing relative water content to 50 mm d-1 when the soil is air dry. T8. Calculate the cumulative actual evaporation of a bare soil for 10 consecutive days after heavy rain. Plot the results versus time and versus the square root of time. What do you observe? Assume the potential soil evaporation to be constant at 5 mm d0-1 and use the proportionality factor as in Listing 8. 182

T9. What is the effect of a larger coefficient for distributing evaporation over soil layers? In what type of soils should this coefficient be relatively high and in what soil types should it be small? S1. Run L1D + L2C + L2SU for upland rice in the wet season on a loamy soil. Compute total transpiration, evaporation and deep drainage. What is the yield loss due to water stress? Notice how water infiltrates into the soil on a rainy day. What is the water use coefficient in these cases? S2. Rerun L1D + L2C + L2SU for upland rice on a loamy soil in the dry season. What is the yield, and what is the yield loss, due to stress? How much would the yield improve if it were possible to lower the internal/external fraction in stomata to 0.4? What is the penalty of no regulation in this case? What is the water use coefficient in these cases? S3. Rerun L1D + L2C + L2SU for upland rice for the wet season on a sandy, a loamy and a clay soil. How much water is lost by evaporation for each soil type during the season? Explain the difference in yield. How much more water is retained in soils that remain fallow? S4. To the program created in exercise S1, add that roots preferentially draw water from the top layer; all other conditions being equal: 1 cm root in the top layer absorbs 1.2x the average, in the middle 1.0x the average and in the bottom layer 0.8x the average? How do yields change? How do the water contents of the upper and lower layers change? S5. Although the SAHEL submodel was designed for well drained soils, it may be adapted to simulate in a rough but simple way a situation with impeded drainage. This may be useful if the information required by SAWAH is not available. Assume to this purpose that the interface of layers 1 and 2 limits percolation to a maximum of 2 mm d-1. This causes waterlogging in the upper layer and water that exceeds the storage capacity of this layer (i.e., filling till the layer is saturated) runs off the field. Evaporation continues at the potential rate as long as the water content of the upper layer exceeds 90% of the saturated content. By how much is soya bean yield in the wet season on a loamy soil reduced, compared to a soil without this low-percolation layer? Why? How much water runs off or drains out of the soil profile? 5.5.3

Soil with impeded drainage

T10. HOW does a relational diagram of L2SS differ from a relational diagram of L2SU? T11. Calculate the flow rate between an upper layer with WCL(1) = 0.25 cm3cm-3 and a lower layer with WCL(2) = 0.05 cm 3 cm-3. Assume for simplicity that the total amount of water in each layer remains contant, and also that conductivity k is independent of moisture content and has a value of 5 cm d-1 (hypothetical!). The thickness of both layers is 0.1 m (case la). Use the pF curve of coarse sand to derive suctions. What would be the flow rate if WCL(1) = 0.05 cm3 cm-3 and WCL(2) = 0.25 183

cm3 cm-3 (case lb)? And what if the layer thickness were 0.2 m (cases 2a, 2b)? S6. Consider the program SWD (Listing 6). Run it for a loess loam soil with groundwater at 0.8 m, starting with a moisture content of 0.08 cm3 cm-3 throughout the profile. Choose the top boundary condition such that equilibrium with groundwater will finally be reached. Depict the moisture profile (WCL vs depth) for each day and note capillary rise. Why does water enter from the top and bottom of the profile? After how long is the equilibrium situation reached? How do you recognize this state? Why does the water content in the top layer exceed field capacity (pF = 2). Repeat this for light clay (soil Type 17, Table 26) and for fine sand (soil Type 4) with DELT = 0.001. Compare the equilibrium profile on loam with one computed from Figure 57. S7. Rerun the SWD program as in exercise S6, but suppose that the soil profile does not permit water to enter at the top layer. How do results change? Does it take longer to reach equilibrium when starting with a soil at 99% of saturation? Plot WCL(1) and WCL(l0) versus time, and also the fluxes at the bottom of the first and last layers, FLX(2) and FLX(11). Plot WCL versus depth every day (manually) and the final matric suction (MS) versus depth. Repeat this exercise for impeded drainage: set the flux at the bottom (FLX (11)) to zero (remove the old FLX(NL + 1) statement). Use an initial moisture content of 50% of saturation. What happens? S8. Use a simple approach to include transpiration (water uptake) of potentially 2 mm d-1 to each of the top three layers of the SWD model (ground water at 0.8 m, MSBT = 0., MSTP = 80., FLX(1) = 0.0). To mimick stress, introduce a reduced uptake proportional to the relative water content in that layer. Be cautious with dimensions and units. What changes are due to water uptake? S9. Run L1D + L2C + L2SS and data for lowland rice, wet and dry season, on a loamy soil. Use the equations to quantify water stress of Subsection 4.2.2. Assume the water table to be at 0.4 m, and a dense soil layer at 0.3 m that roots cannot penetrate. Total the relevant water fluxes. First use 10 layers, and then rerun it for 5 layers (NL = 5, TKL(1 – 5) = 5 * 0.2). What is the cause of the difference? Is it significant? S10. Determine the grain yield and the capillary rise to the root system using the program from Exercise S9 for a constant water table at depth of 3.0, 1.8, 1.2, 0.8, 0.5, 0.3 and -0.01m. What is the difference when the dense soil layer is removed and roots grow to a depth of 0.7 m? Are simulation results sensitive to the soil water table input? S11. Use the program of Exercise 9 with a water table at 0.2 m at day 50., 0.5 m at day 100 and 1.0 m at day 150. What are the yields for transplanting dates 52, 62, 72 and 82? How much water is supplied from the ground water table and by how much is the soil water depleted? Is the rooting depth the same on both soil types?

184

5.5.4 Soil water balance subprograms T12. Determine the tree of subroutines and functions used for L2SS and L2SU. Which subroutines and functions can be deleted if L2SU is used? S12. Run both water balance simulation modules with soya bean in the wet season for a sandy, a loamy and a clay soil, with the water table at 3.0 m and all layers at field capacity at the start. Are results of the modules identical? Should they be? Does yield increase if the soil water table is at 1.5 m on these soils? 5.6

Answers to exercises

5.6.1 Soil characteristics T1. The soil textures are those of light clay, silty loam and sandy loam. T2. See Figure 57. T3. The pressure head is –100. cm, –20. cm, and –1000. cm, respectively. T4. In a 1.0 m deep soil of medium fine sand or clay loam, the maximum available water (mm) amounts to: WCWP pF 4.2 Medium Fine sand Clay loam

0.011 0.276

WCFC pF 2.0 pF 2.5 0.161 0.399

0.104 0.376

Available water pF 2.0 pF 2.5 150. 123.

93. 100.

If field capacity is at pF = 2, the medium fine sand soil has most water available for a crop. When field capacity is at pF = 2.5, the clay loam can store more water (derived with the equation in Subsection 5.1.3). T5. In both situations the clay loam has more water in the profile above the groundwater table. Close to the groundwater table, volumetric water content in the fine sandy loam is a little higher, but this is insufficient to compensate for the lower water content near the surface.

5.6.2 Free-draining soil T6. Carefully distinguish the types of variables and relations involved. Use the symbols of Figure 5 to draw the diagram. See Figure 6. T7. Equations can be used for linear relationships: RUNOF = RAIN * FRUNOF FRUNOF = AMAX1(0., (RAIN – MAXINF) / RAIN) MAXINF = 50. – (50. – 10.) * RWCLl RWCLl = AMIN1(1., (WCL1– WCAD1) / (WCFC1 – WCAD1))

185

A more flexible solution is to use the TWOVAR function (IBM, 1975). It allows linear interpolation in two dimensions to make the runoff fraction a function of rainfall intensity (the independent variable in the functions) and relative water content of the upper layer (RWCL1, the value after the function name). For instance: RUNOF = RAIN * FRUNOF FRUNOF = TWOVAR(RUNOFT, RAIN, RWCL1) RWCLl = AMIN1(1., (WCL1 – WCAD1) / (WCFC1 – WCAD1)) FUNCTION RUNOFT,0. = 0. ,0., 10.,0., 20.,0., 30.,0.,... 40.,0., 50.,0., 60.,.17, 70.,.29, 80.,.38,... 90.,.44, 100.,0.5, 1000.,.95 FUNCTION RUNOFT,.5 = 0.,0., 10.,0., 20.,0., 30.,0,... 40.,.25, 50.,.40, 60.,50, 70.,.57, 80.,.63,... 90.,.67, 100.,.7, 1000.,.97 FUNCTION RUNOFT,l. = 0. ,0., 10.,0., 20.,.50, 30.,.67,... 40.,.75, 50.,.80, 60.,.83, 70.,.86, 80.,.88,... 90.,.89, 100.,.90, 1000.,.99 T8. On the first day actual evaporation equals potential evaporation. Cumulative evaporation has a linear relation with the square root of time since the last rain. After 10 days, it equals 5 + 0.6· 5· 3 = 14 mm. T9. With a larger partitioning coefficient, the contribution of deeper layers to evaporation decreases. In soils with a fast decrease of hydraulic conductivity with increasing suction the deeper layers tend to contribute less to evaporation. In general, coarse textured soils such as sands have a high partitioning coefficient; fine textured soils with a high silt content have a low coefficient. S1. The totals of transpiration, evaporation and deep drainage are 332.24, 169.38 and 486.15 mm, respectively. The grain yield is 6186.0 kg ha-1, which is only 324.6 kg less than with no water stress. The water use coefficient in these two cases is 95.5 (stressed) and 95.3 kg H2O kg -1 CO2. S2. The yield without irrigation is 949.80 kg ha-1 and the crop dies early; with plenty of water, the yield is 7691.4 kg ha-1. A crop with a lower internal/ external ratio (0.4) produces 1457.2 kg ha-1; a crop without stomatal regulation dies earlier and only 223.88 kg ha-1 is formed. The water use coefficient in these cases is 95.5 (95.3 no stress), 72.7 and 141.9 kg H2O kg -1 CO2. S3. The yields on sandy, loamy and clay soils are 5615.5, 6186.0 and 5385.1 kg ha-1, respectively. Cumulative soil evaporation is 179.45, 169.38 and 183.61 mm, respectively. The small difference in evaporation does not explain the different yields. The crop yields least from the soil with the smallest difference between field capacity and wilting point. Simulate a fallow soil by breaking the interactions between crop and soil (the crop related statements can then be removed): redefine TRW to be equal to TRC and ZRT = 0., replace ALV by 0.0 in FUWRED and in EVSC = .... There is 139.01, 288.23 and 309.45 mm of water in the soil profile on sand, 186

loam and clay at harvest time. Without a crop, there is more water left at the same date: 193.66, 343.42 and 363.25 mm, respectively. S4. This can be simulated by modifying four statements: TRRM = TRC / (ZRT1 * 1.2 + ZRT2 + ZRT3 * 0.8 + 1.E–10) TRWLl = TRRM * 1.2 * ZRTl * WSEl TRWL2 = TRRM * ZRT2 * WSE2 TRWL3 = TRRM * 0.8 * ZRT3 * WSE3 The yield decreases by 165.3 kg ha -1 ; 0.0083 cm3 cm-3 more water is left in the lower layer and 0.0202 less in the upper layer. S5. The waterlogged crop yields 1681.2 kg ha -1 and the crop in free-draining soil 2477.0 kg ha -1 of storage organ. The total runoff in the first case is 600.5 mm and drainage is 8.2 mm, while runoff in the free-draining soil is zero and drainage is 509.7 mm. Therefore, more water is available and less stress occurs in the upper layer in the second case.

5.6.3 Soil with impeded drainage T10. More layers are distinguished in L2SS and there are two additional state variables: WL0QT and ZEQT. The soil water table is an input. Runoff and infiltration are computed and are no longer a constant fraction of precipitation. Soil evaporation is computed in a different way. T11. The different cases are: 1a: WCL(1) = 0.25;WCL(2) = 0.05;TKL(l) = 10 cm;TKL(2) = 10 cm 1b: WCL(1) = 0.05;WCL(2) = 0.25;TKL(l) = 10 cm;TKL(2) = 10 cm 2a: WCL(1) = 0.25;WCL(2) = 0.05;TKL(1) = 20 cm;TKL(2) = 20 cm 2b: WCL(1) = 0.05;WCL(2) = 0.25;TKL(l) = 20 cm;TKL(2) = 20 cm From Figure 57 read suctions of 10 cm and 100 cm at WCL = 0.25 and WCL = 0.05, respectively. Using Equation 4, the flow rates q are calculated as: 1a.: q = (+90 / 10)· 5 + 5 = +50 cm d -1 1b.: q = (-90/ 10)· 5 + 5 = -40cm d -1 2a.: q = (+90 / 20)· 5 + 5 = +27.5 cm d -1 2b.: q = (-90/ 20)· 5 + 5 = -17.5 cm d -1 (negative flows are directed upward) S6. This model cannot deal with saturated soil; with 10 layers, layer thickness must be adjusted to 0.08 m. At the groundwater, MSBT = 0. ; equilibrium can only be reached if boundary condition FLX(1) = 0. or MSTP = 80. is imposed. Choose the latter. Water initially enters from the top and bottom of the profile because hydraulic head decreases from the top downward and from the bottom upward. Water contents differ less than 1% from their equilibrium 3 values (0.38826 cm cm-3 in Layer 4) after three days; Layer 4 is the last to 187

reach equilibrium. Water contents exceed field capacity according to the pF = 2 definition (0.36 cm3 cm-3 , see Table 27) because the water table is less than 1.0m deep. This near-equilibrium is reached after seven days on clay (in Layer 4, 0.39610 cm3 cm-3 ) and after only two days on sand (0.23638 cm3 cm-3, Layer 4). This state is recognized by almost zero fluxes and zero hydraulic head gradients. The equilibrium profile for loam, reading Figure 57, starts with WCL = 0.365 cm3 cm-3 in the upper layer for an average pressure head of -76 cm, and WCL = 0.390 cm3 cm-3 in Layer 5. The difference with the numeric solution is small. S7. This situation is obtained by replacing Listing 6 Line 18 by FLX(1) = 0. Water flows into the soil only by capillary rise. It takes longer to reach equilibrium: after 11 days on loam, 19 on clay and 4 days on sand, the top layer is always the last. Equilibrium is reached faster from wet than from dry situations because the conductivity of wetter soil is higher: in 3 days on loam, 2 days on clay and 1 day on sand. The implicit assumption in the L2SU module (that after a soil is saturated, most water held above field capacity drains within approximately one day) appears to be a fair approximation. Impeded drainage leads to redistribution of water but no water enters or leaves the soil. On loam and clay there is no redistribution; on sand it takes about 10 days. Redistribution is slow because conductivities are low. S8. In the computation of the rate of change DWCLDT(I) for the top three layers add a sink term equal to 0.2 cm d-l / TKL(I) cm * WCL(I) / WCST(I). An actual equilibrium will not be reached, since the roots continue to extract water. To approach the equilibrium state, it takes longer with than without water uptake: 12 days on loam and 4 on sand. The message ‘logarithm of negative argument’ warns when the stress function is not properly designed and as a result non realistic water contents develop. S9. Simulation stops at Julian date 290 when the crop is mature and the storage organs weigh 5341.6 kg ha-1. Rooting depth is 0.3 m. Total transpiration is 262.4 mm, total evaporation is 258.3 mm, total drainage is 33.8 mm, and while total rainfall is 916.0 mm, runoff amounted to 361.9 mm. The same case simulated with five layers yields 5157.5 kg ha-1 of storage organ, rooting depth 0.23 m, and 254.0 mm tranpiration, 264.7 mm evaporation, 38.9 mm drainage and runoff is 358.7 mm. The lesser rooting depth is the result of the above-threshold water content in the second of the five layers. The summed fluxes are different because gradients of pressure head are taken over larger distances, leading to a loss in accuracy. These differences are usually insignificant and in the same order of magnitude as differences due to inaccuracies in input data. S10. At harvest time, the results for panicle weight (WSO), leaf area (ALV), total transpiration (TRWT, mm) and total evaporation (EVSWT) and total drainage (DRSLT) are:

188

ALV 0.51 1.28 2.07 3.68 4.53 5.37 8.38

TRWT 28.4 75.1 140.2 234.7 274.7 312.3 439.2

EVSWT 43.3 89.9 282.3 257.4 246.3 246.8 201.8

DRSLT 30.0 -105.4 -421.0 -487.9 -504.8 -546.1 -641.2

The wet season results are: DATEB = 197 WSO ALV ZW = 3.0 1034.9 0.72 ZW = 1.8 1708.3 1.17 2924.0 ZW = 1.2 2.00 3924.0 ZW = 0.8 2.84 ZW = 0.5 4421.6 3.44 ZW = 0.3 4902.3 4.10 ZW = -0.01 6509.0 6.33

TRWT 56.4 88.2 140.8 190.1 219.3 249.6 351.2

EVSWT 326.7 320.2 352.3 310.1 285.0 264.5 219.3

DRSLT 377.0 304.1 157.6 137.7 131.4 66.2 -301.7

DATEB = 52 ZW = 3.0 ZW = 1.8 ZW = 1.2 ZW = 0.8 ZW = 0.5 ZW = 0.3 ZW = -0.01

WSO 209.0 826.4 2738.0 4721.8 5371.9 5927.3 7688.3

Without the dense soil layer, the values at the highest and lowest soil water tables are the same or quite similar. With the water table at 1.2 and 1.8 m, roots have access to more water and yields are higher by 1000-1500 kg ha-1. Water table depth is an important input variable in both cases and simulation is consequently sensitive to this. (The yield with the water table at 0.4 m (Exercise S9) is higher than that at 0.3 or 0.5 m in this exercise. The difference develops largely between days 72 and 80. The water content of layer 3 is then not fully saturated (which it should be) for the 0.3 m water table depth, while it is for the 0.4 m depth. This is a small error due to the small effective number of layers (3) combined with the constant values of the water tables.) S11. Results are as follows: loam

sand

DATEB 52 62 72 82 52 62 72 82

WSO 5010.7 5217.0 5175.1 4550.3 4747.6 5007.2 5041.5 4667.5

dWCUM -89.95 -97.74 -91.70 -81.72 -76.07 -60.71 -57.62 -62.32

DRSLT -454.29 -442.23 -437.40 -410.10 -429.45 -480.63 -462.82 -429.83

ZRT 0.42 0.58 0.62 0.72 0.61 0.63 0.62 0.71

WSO is grain yield, dWCUM is the change in available soil water since the beginning, DRSLT is total drainage (negative values indicate capillary rise into the profile) and ZRT is rooting depth. 189

5.6.4 Soil water balance subprograms T12. L2SU + L1D + L2C uses FUCCHK, FURSC, FUVP, FUWCHK, FUWRED, FUWS, FUPHOT (calls SUERRM, SUASTC) or SUPHOL (calls SUERRM, SUASTC), SUEVTR (calls SUERRM), SUASTR (calls SUASTC). L2SS + L1D + L2C uses the same functions and subroutines as above, as well as the SAWAH subroutine assembly. It is invoked by calling SUSAWA, which employs the subroutines SUCONV, SUGRHD, SUSLIN (calls SUERRM), SUSTCH, SUSTHH, SUSTFL (calls SUSTMD, SUSTMS), SUUNST (calls SUWCMS (calls SUERRM), SUMFLP (calls SUMSKM), SUMSKM (calls SUERRM)), SUSEFL, SUINTG (calls SUWCMS (calls SUERRM)), and SUZECA. S12. The results from both modules are not identical, but similar, as expected. Yields according to the L2SU module for loam, sand and clay are 3452.0, 3426.5 and 3046.6 kg ha-1 , respectively, and the available water in the profile at the end of the season is 210.4, 62.6 and 244.2 mm. For L2SS, the results are: 3309.9, 3157.6 and 2956.1 kg ha-1 and 205.3, 60.8 and 239.0 mm, respectively. With a water table at 1.5 m, yields in L2SS increase slightly to 3359.2, 3212.7 and 2938.9, and the available water increases to 329.6, 177.2 and 337.0 mm, respectively.

190

6 Weather data

Section 6.1 discusses the nature and precision of weather variables required for crop growth modelling. These variables are: – at Production Level 1: daily values of solar radiation, maximum and minimum temperature and air humidity; – at Production Level 2: in addition to variables at Production Level 1: daily precipitation values, air humidity and windspeed. Section 6.1 includes some hints on pitfalls of data collection. A few derived weather variables are computed in Section 6.2: radiation from clear and overcast skies, net radiation, daylength and potential evapotranspiration. Some are needed in the simulations, others provide reference values for comparison of growth conditions. Generating data can be helpful when sufficient original observations are unavailable. This is discussed briefly. 6.1

6.1.1

Historic weather data

Introduction

Solar radiation is by far the most important weather variable for crop growth simulation at Production Level 1, but air temperature can also be crucial. Air humidity is important in very dry weather for some crops. For simulation at Production Level 2, precipitation is an essential input; solar radiation, temperature, and air humidity are also important, but windspeed has little impact on transpiration. Three full sets of historical weather data are shown in Figure 65 as examples of different climatic types. An example of how such data can actually be used in the form of CSMP tables is shown in Listing 11. Weather data can be obtained from national meteorological services, from the Food and Agricultural Organization (FAO) (Frère, 1987) and from the institutes within the Consultative Group on International Agricultural Research (CGIAR) (e.g., Oldeman et al., 1987). Inspect all data carefully for definitions and units of variables. If data have been obtained locally or from small stations, also inspect the measuring conditions. Information about interpreting weather data and environmental physics can be found in Rose (1966), Monteith (1973), Campbell (1977), Doorenbos & Kassam (1979), and Oldeman & Frère (1982). The World Meteorological Organization recently produced a hardware-software-training package for a data base management system for climatological data (CLICOM). Weather data for modelling are often difficult to obtain, particularly if sets for 5 to 25 years of historical data for all six variables without missing values are required. 191

Figure 65. Patterns of weather variables in 1984 in a temperate climate (Wageningen, the Netherlands, left), a humid tropical climate (Los Baños, the Philippines, middle), and a semi-arid tropical climate (Hyderabad, India). 1: observed total global radiation (circles) and clear sky radiation (MJ m-2 d -1); 2: maximum and minimum temperatures (°C); 3: average windspeed (m s-l x 10) and cumulative precipitation (cm); 4: dew point temperature (line, °C) and the evaporation rate of a standard grass sward (mm d-1).

The one-day time period used in crop growth simulation corresponds well with the frequency with which weather data are often recorded. Average values can be used if the basic data available are only weekly or monthly means. Obviously, the impact of short deviations from the mean can then not be evaluated. The use of averages for radiation, temperature, windspeed and humidity is appropriate for many purposes. Values of precipitation per day, either 192

observed or generated, are essential for simulating water-limited production (van Keulen & Wolf, 1986). Daily values of radiation are required if sensitive phases (such as tillering and panicle initiation in cereals) fall in periods of variable cloudiness. When the simulation time period is about six hours (as in module L1Q) intermediate values of weather parameters are estimated from the daily values. These may be replaced by actual data when available. Weather data are given in the form of CSMP tables: a set of 365 values per variable. It is practical to always supply a full year of data, rather than only a growing season, so that crop growth or derived weather variables can easily be obtained outside the main season. Two other input parameters characterize a site: latitude and elevation above sea level. Latitude affects the maximum amount of solar radiation and daylength, and elevation affects the CO2 concentration of the air.

6.1.2 Solar radiation Solar radiation is a key meteorological variable and its values should be obtained as accurately as possible. Daily values of the ‘total global radiation’ should be obtained, if possible, from a properly calibrated RIMCO pyranometer, or a Gunn Bellani integrator (Oldeman et al., 1987), or similar instruments. ‘Total’ refers to the sum of visible and near infrared radiation (400-1300 nm) and ‘global’ refers to radiation coming from all directions. Readings given in sunshine hours (Campbell-Stokes method) must be converted into J m -2 d -1 (see van Keulen & Wolf (1986) p. 64 for how to do this). Radiation on fully overcast days is, by convention, 20% of the value on fully clear days, though in reality, days with even less radiation occur. The radiation unit conversion factor in the modules L1D and L1Q (Listing 3 Line 103, Listing 4 Line 131) maintains proper calculating units. Solar radiation can be partitioned in two ways: according to wavelength in photosynthetically active radiation (PAR, 400-700 nm wavelength) and near infrared (700-1300 nm), and according to direction in direct (from a point source) and diffuse radiation. PAR is always about 50% of the total solar radiation and this fraction varies little with radiation intensity (Monteith, 1973). But the fraction diffuse of the total radiation depends strongly on the daily total. The relation between the fraction diffuse and the daily total radiation relative to the extraterrestrial radiation at that location and date appears to be constant in temperate regions (Figure 66; Spitters et al., 1986). The relation is built into the SUASTC subroutine (Appendix B). The difference between direct and diffuse radiation is important for canopy photosynthesis and SUASTC always accounts for this. Recordings in a humid climate showed almost the same relation, though with much scatter and higher values on clear days (Figure 66). A higher minimum of the fraction diffuse radiation in the humid tropics can be included by replacing 0.23 by 0.35 in the SUASTC sub193

Figure 66. The ratio of diffuse radiation over total global radiation as a function of the ratio of total global radiation over its normal maximum at sea level (0.75 times extraterrestrial radiation). The dotted line is from Spitters et al. (1986), the dashed line from subroutine SUASTR. The drawn line and the standard deviations at some points are based on five years of recording by the Philippine-German Solar Energy Project (a joint project of the Republic of the Philippines and the Federal Republic of Germany, implemented by the Office of Energy Affairs and the Deutsche Geselschaft für Technische Zusammenarbeit).

routine (Appendix B). In the L1Q module with quarter-day time periods, radiation is supposed to be partitioned equally over morning and afternoon, and is zero at night.

6.1.3

Minimum and maximum temperatures

Temperature affects the rate of most physiological processes. Maximum and minimum temperatures are not used as such in simulation, but are replaced by an effective temperature which is calculated from them. How this is done depends on the process and the time period. The effective temperature for processes that continue during the complete 24-hour time period is the average of the maximum and minimum temperatures. The effective temperature for photosynthesis is assumed to be the average day temperature, calculated as the 194

mean of the 24 hour average and the maximum temperature (Figure 67, Listing 3 Lines 105, 106). Only in the quarter-day time period module (L1Q) is the effective temperature for each time period and for all processes calculated from a fixed, asymmetric pattern over the whole day, by using the nearest maximum and minimum temperatures (Listing 4 Line 133 plus the FUTP function, Appendix B). The four fractions in the call of the FUTP function were determined for Wageningen and also approximate a humid tropical location (Oldeman & Frère, 1982). However, the pattern could be different at other locations. Maximum and minimum temperatures (or values at 14.00 h and sunrise, respectively) can easily be accurately measured. Values used here are based on observations at the standard screen height of 1.5 m above the soil surface. The temperature near the soil surface can differ from the air temperature at screen height. Small plants may have a higher temperature during the day and a lower one at night. This is not accounted for here, but it may be worthwhile to

Figure 67. A graphical representation of the procedures to determine the effective temperature in the modules L1D (dashed line: 24-hour average and daytime average) and L1Q (fine dashes: effective temperatures in 6-hour periods). Circles indicate input data.

195

explicitly consider near surface temperatures when studying crop emergence. Leaf and growing point temperatures can also deviate several degrees from the air temperature (Subsection 4.3.2); the higher the transpiration rate the lower the leaf temperature relative to air temperature. Such deviations mean that using standard meteorological temperatures is not always totally satisfactory. However, as yet they are the best available. 6.1.4

Precipitation

Rain is a particularly important driving variable in the semi-arid and subhumid tropics, but is also important in temperate zones during in dry periods. Its value can change more from day to day than any other meteorological variable. On wet days its value is often 2-20 mm, but can reach 100 mm or more in intensive tropical storms. Determinating its value deserves proper attention and this is more difficult than may appear. Measurements generally underestimate the real value (TNO, 1977). An accuracy of more than 5-10% for absolute values of precipitation is difficult to attain. The wind profile around the rain gauge is very important, for the height of the rain gauge above the surface and its exposure have repercussions of 10-20% and more on the amount of rain caught in the gauge. The standard rain gauge at a meteorological station in the Netherlands has an opening of 200 cm2, with its rim 40 cm from the soil surface; it underestimates reality by a few percent (TNO, 1977). The imprecision in precipitation measurements is significant for simulation. The spatial variability of precipitation is also quite large. Its value must be determined at the field for which the study is undertaken whenever precipitation is a key variable for simulation. Though the quality of data of the nearest official meteorological station may be better, less accurate data from the field for which the simulation is performed can be more relevant. Rainfall generally occurs over periods much shorter than 24 hours, the time period used in simulation. Observations of rainfall intensity with recording rain gauges are rarely made routinely. This is a serious handicap for runoff calculations. In the module L2SS for soils with impeded drainage, it is assumed that all precipitation from a single day is received in the first soil water balance time period. Interception of precipitation by leaves, stems and fruits is 10-20% of the fresh weight of leaves, or almost equivalent to their dry weight. These amounts are equivalent to a water layer of less than 1 mm per occasion, which is disregarded here. Interception should be taken into account in environments receiving a large number of small rain showers. Dew rarely amounts to more than 0.1 mm d-1 and is difficult to measure. Dew is partly condensed soil evaporation. Though the amounts of dew are small and negligible for water balance studies, wetness of the surface can be

196

crucial for other processes. The leaf-wet period is important in simulating crop damage by pathogens, because diseases develop quicker on wet than on dry surfaces.

6.1.5

Humidity

Air humidity affects transpiration and evaporation, and reduces photosynthesis in some crops when its value is very low. Good air humidity measurements are not easy to obtain, but are not of overriding importance in crop simulation. Air humidity can be measured in several ways and expressed in different units. The absolute concentration, expressed as the water vapour pressure in kPa (1 kPa= 10 mbar), is preferred (its value generally changes little during the day, so that the time at which the reading is taken is less important). Humidity expressed in other units can be converted to kPa by equations shown in Table 29 in Subsection 5.4.5. Relative humidity changes a lot during the day and should be avoided as a basic measurement of humidity. If reliable values for air humidity are unavailable, they may be approximated by assuming that the air is saturated with vapour at dawn when the daily minimum temperature is reached. The vapour pressure can then be obtained by calling the FUVP function for this temperature. This is a good approximation when there is dew, but it provides values which are too high for the dry season in arid and semi-arid environments.

Table 29. Equations to convert air humidity data (HUAA) into vapour pressure (VPA, in kPa). TPA is air temperature (°C) at the time the wet bulb temperature or relative humidity was taken, or average day temperature; FUVP(temp) is: 0.611 * e (17.47 * temp/(temp+239) .)

If HUAA is dewpoint temperature: VPA = FUVP(HUAA) If HUAA is wet bulb temperature: VPA = FUVP(HUAA) – 0.0623 * (TPA – HUAA) If HUAA is in mbar: VPA = 0.10 * HUAA If HUAA is in mm Hg: VPA = 1.33 * HUAA If HUAA is in percent relative humidity: VPA = 0.01 * HUAA * FUVP(TPA)

197

6.1.6

Wind speed

Canopy transpiration is only sensitive to windspeeds up to 1-2 m s-1 (see Figure 49 in Subsection 4.1.3). Plant lodging due to high wind speeds and gusts is not considered here. Windspeed is measured directly as a rate and averaged over 24 hours, or obtained as a daily windrun and expressed in m s-1 . It is often measured at 2 m over a low grass sward at a standard meteorological station. This reference height is input to the FURSC function (Listing 7 Line 26 and Appendix B). Observations made at other heights must be converted (cf., van Keulen & Wolf (1986) p. 210). Windspeed during the day is generally higher than at night and the effective windspeed is taken as 1.33 times the average value (Listing 7 Line 52); this factor is somewhat arbitrarily chosen and may need adjustment in specific situations. Thermal air instability inside the canopy occurs spontaneously during the day when there is almost no wind and enhances gas exchange. This is accounted for by limiting the windspeed to a lower value of 0.2 m s-1 (Line 51).

6.1.7

Carbon dioxide

The CO2 concentration was about 340 vppm (cm3 m-3 ) at sea level in 1986, and its value rises steadily (Goudriaan, 1987). The CO2 concentration fluctuates very little during the year and usually does not change significantly inside the canopy. Its value at sea level is specified with a parameter. The volumetric CO2 concentration decreases by 12% per 1000 m elevation (Listing 7 Line 59). 6.2

6.2.1

Derived and generated weather data

Clear sky and overcast radiation

The maximum amount of daily total global radiation can be computed accurately for any day and latitude (Figure 68) using the SUASTR and SUASTC subroutines of module T12 (Appendix B). These were constructed on the basis of van Keulen et al. (1982) and Spitters et al. (1986). The starting point is the solar constant (about 1400 J m-2 s -1 ), i.e., the intensity of solar radiation measured outside the atmosphere and perpendicularly to the solar rays. Radiation at sea level on a perfectly clear day is about 25% lower than the solar constant due to absorption and reflection by water vapour and dust in the atmosphere. Clear sky radiation at sea level can be used as a yardstick for monitored radiation (Subsection 6.1.2) and their ratio is the relative amount of radiation received. Observations taken at sea level are usually between 0.15 and 0.75 times the value for extraterrestrial radiation at the same location and the same date, but are less on very heavily overcast days and up to 0.9 times extraterrestrial radiation under extremely clear skies. 198

6.2.2 Daylength Astronomical daylength is input for the photosynthesis computation in the FUPHOT function and SUPHOL subroutine and in calculating evaporation in the SUEVTR subroutine. Daylength provides the basis for splitting each 24hour period into day and night fractions (L1Q, Listing 4 Line 144). Its value is accurately obtained from a set of mathematical equations in the SUASTR subroutine, using latitude and date as inputs. To compute daylength for photoperiod-sensitive species, it must be realized that, even when the sun is still below the horizon the light level is high enough to trigger the photoperiodicity mechanism. Implied in the SUASTC subroutine is assumption that daylength for photoperiodism is the time per day that the sun is at inclinations higher than –4 degrees. Photoperiodic daylength is longer than astronomical daylength by about 0.5 h at the equator and by about 0.8 h or more in temperate zones, depending on the date in the year. The light level to which photoperiodism is sensitive is quite low and not well quantified. Vergara & Chang (1985) determined it to be 1.5-15 mW m-2 for rice crops; Salisbury (1981) determined the level to be higher. As a compromise, a value of 50 mW m-2 is used here, which corresponds with a sun angle of about –4 degrees. Because calculating the photoperiodic daylength makes no sense when the sun is continuously at higher inclinations, the SUASTR subroutine is limited to –66.5 + 4 and 66.5 – 4 degrees of latitude. Solar height at noon (SUNH, degrees) can be computed by adding to the SUASTC subroutine: SUNH = ASIN(COSLD + SINLD) / RAD

6.2.3 Net radiation Net radiation (all wavelenghts included) is the balance of incoming short wave radiation (wavelength 400-1400 nm) minus its reflection and outgoing thermal radiation (>3000 nm), plus incoming thermal radiation (about 12.000 nm). Its calculation is part of the computation of the energy balance for evapotranspiration (SUEVTR subroutine, Appendix B, cf., van Keulen & Wolf (1986) p. 67). Reflection of short wave radiation is about 0.2-0.3 for crops. Reflection from a soil surface (i.e., its albedo) is 0.1-0.4 (see Table 28 Subsection 5.2.4). Reflection increases strongly at low inclinations of the sun (Menenti, 1984), but this is unimportant as the light level is then low. Net thermal radiation is computed according to the Brunt equation from surface temperature (as an indicator of the outgoing long wave radiation) and from cloudiness and air humidity (as indicators of the intensity of incoming long wave radiation from the sky). Values for net radiation can be obtained as output from the program by adding its name to the list of output variables of SUEVTR. Figure 68 in Subsection 6.2.1 provides an example of the range of values of 199

Figure 68. The measured and observed values of daily total net radiation on each third day for a full year in Wageningen, the Netherlands; the line indicates a 1:1 ratio (Data source: Department of Physics and Meteorology of the Agricultural University, Wageningen).

net radiation during a full year in the Netherlands. In this example observations and computations agree closely; significant differences only occur in winter. 6.2.4

Potential evapotranspiration

Simulation of the transpiration rate of a well-watered canopy was described in Section 4.1. Computed rates of 0-5 mm d-1 for a grass sward during a growing season in the Netherlands (see Figure 45 in Subsection 4.1.2) compared fairly well with measured rates and observations by the Department of Physics and Meteorology of the Agricultural University in Wageningen. Similar simulations were carried out for a healthy rice crop with a closed 200

canopy in the Philippines in the dry and wet seasons when the evapotranspiration rates were 4.5-9 and 3-5.5 mm d-l, respectively (unpublished observations by the Climate Unit of the International Rice Research Institute). Because the canopy was closed, almost all evapotranspiration was transpiration. In the rice crop the 5-day average of simulated and measured values corresponded well (Figure 4.9, but values per individual day differed as much as 15% on average. The discrepancies of day-to-day values are probably largely due to difficulties in measuring this variable over a 24-hour period with the small field lysimeters used, because measured values do not show a consistent relation to weather variables. Soil evaporation was also included in this measurement, but its value must have been negligible. It therefore seems that potential rates of canopy transpiration can be obtained more easily, and at least as well, by simulation, rather than by measurement. Potential evapotranspiration is a useful variable when characterizing a climate. At meteorological stations it is sometimes determined as the rate of transpiration of a standard grass sward well supplied with water and nutrients. As many variables as possible are then fixed. However, it seems easier, and for many purposes at least as good, to compute such rates rather than to measure them (cf., van Keulen & Wolf, 1986 p. 74). This rate is referred to here as the standard simulation of evaporation of grass (EVG, mm d-1). It can be computed as canopy transpiration, separating the effects of radiation (EVGR) and drying power (EVGD), but with more variables constant: the reflection coefficient is 0.24, no night-time transpiration, leaf resistance is 150 s m-1 in the upper 2.5 m2 m-2 of leaves, boundary layer resistance is 12 s m-1 and canopy resistance RSTP equals 132. / (1. + 0.54 • WDSAD) (GELGAM, 1984): EVG = EVGR + EVGD * 2.5 EVGR,EVGD = SUEVTR (RDTC, RDTM, 0.24, DLA / 24., TPAD,... VPA, 150., 12., RSTP) EVG can vary considerably during the year and between sites, as illustrated in Figure 69 (also Figure 65 Subsection 6.1.1). Evaporation from a free water surface can be similarly approximated by setting the reflection coefficient at 0.07, permitting night-time evaporation, using the average day temperature, putting leaf resistance equal to 0.0, and assuming the boundary layer and turbulence resistances to be the same as for the grass sward: EVW = EVWR + EVWD EVWR, EVWD = SUEVTR (RDTC, RDTM, 0.07, 1.00, TPAV,... VPA, 0.0, 12.0, RSTP) The calculations of EVG and EVW yield only approximations and these can be different from the approximations by Doorenbos & Kassam (1979). One source of differences is that the temperature of the evaporating surface (e.g., an evaporation pan) is one or two degrees below the average air temperature 201

Figure 69. The simulated cumulative transpiration of a well-watered, standard grass sward in Wageningen, The Netherlands, Los Baños, the Philippines and Hyderabad, India. (For weather data, see Figure 65).

in a relatively dry environment (Tamisin, IRRI, personal communication), leading to rates 10-15% lower than those computed with the preceding lines. Still another evapotranspiration rate (EVP, in mm d -1), based exclusively on meteorological and physical inputs, is often used for climatic characterization and site comparison. It is determined with a correlation Penman method, described by Doorenbos & Kassam (1979, p. 17). It may be helpful to calculate EVP to perform more extensive comparisons with evaporation data locally obtained. For an average site the correlation can be translated into: EVP = 1.0 * (0.75 * (RDTN / 2.47E6) + (1.0 – 0.75) * ... 0.27 * (1. + WDSAV * 0.864) * (FUVP(TPAV) – VPA) * 10.) The equation makes use of variables already in the program; RDTN is to be calculated as in SUEVTR (Appendix B) with FRD = 1.0 and RF = 0.0. The average site is at sea level, 25 °C, with conditions not very dry or windy. The original paper should be consulted for constants for meteorologically different sites. According to the authors, this value of EVP equals evaporation by a Class A pan multiplied by a factor of about 0.7 (0.5 to 0.8, depending on weather conditions). 202

6.2.5

Generating weather data

It is not often that weather data for more than a few full years is available from a single meteorological station. This is insufficient to test the stability of crop yields over a 10-25 year period using simulation. The next best alternative to a large set of historical weather data, is a large set of weather data generated from observations taken over a few years. These can be generated by using information contained in the historical data: the relations between values of variables on successive days and between the values of all variables on individual days. Richardson investigated these relations for precipitation and temperature in the USA, and his computer programs were modified and extended by Geng et al. (1985a,b). Several programs to generate daily weather variables were moulded into an easy-to-use program (Supit, 1986). With this program, daily values for solar radiation, precipitation, maximum and minimum temperature and windspeed are generated for any number of years from as little as two years of historical data. Humidity is not included. When generating new weather data, twelve monthly precipitation totals plus the number of wet days for each month appear to be just as good as taking 365 daily values (Geng et al., 1986). This requires at least 10 times less data than taking daily values and means that sufficient rainfall data to support crop simulation can be collected from remote areas without frequent measurements. This not only spreads resources, but is particularly helpful because rain varies more than any other meteorological variable over short distances and requires a denser recording network than other weather characteristics.

203

7

Listings of modules

This chapter contains listings of modules discussed in the previous chapters. Listings 1-11 are written in the simulation language Continuous System Modelling Program (CSMP, IBM, 1975; Subsection 1.4.1). Abbeviations are explained in Listing 12. Copies of the listings and the crop data on a floppy disk can be obtained from the authors through PUDOC (P.O. Box 4, Wageningen, 6700 AA, The Netherlands) by sending one high capacity 5.25 inch floppy disk with the note ‘MACROS modules and data’. For a copy of a PCSMP version for IBM PC/ AT compatibles plus a short manual, send two high capacity floppy disks with the note ‘PCSMP’.

Listing 1. A CSMP program to compute growth efficiency characteristics.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

TITLE PROGRAM TO COMPUTE CRG, CPG AND FC (JUNE 1986) STORAGE COMP(6) CRG=FCARB*(1.211+0.064) + FPROT*(l.793+0.094+(0.852+0.045)*LEG)+ ... FFAT *(3.030+0.159) + FLIGN*(2.119+0.112) +... FOA *(0.906+0.048) + FMIN *(0.000+0.120) **CARBOHYDRATE REQUIREMENT GROWTH IN G GLUCOSE PER G PRODUCT CPG=FCARB*(0.123+0.093) + FPROT*(0.679+0.138+(1.250+0.066)*LEG)+ ... FFAT *(1.606+0.234) + FLIGN*(0.576+0.164) + ... FOA *(-0.045+0.070)+ FMIN *(0.000+0.176) **C02 PRODUCTION DURING GROWTH IN G C0 2 PER G PRODUCT FC =FCARB*0.451 + FPROT*0.532 + FFAT*0.774 +... FLIGN*0.690 + FOA*0.375 + FMIN*0.000 **FRACTION CARBON IN G C PER G PRODUCT EC =FCARB*17.3 + FPROT*22.7 + FFAT*37.7 + ... FLIGN*29.9 + FOA*13.9 + FMIN*0. **ENERGY CONTENT PRODUCT, IN KJ PER G

ECAF=EC/(l,·FMIN) **ENERGY CONTENT ASH FREE MATERIAL ENEFF=EC/(CRG*15.6) **ENERGY EFFICIENCY CONVERSION ENEFFA=ECAF/(CRG*15.6) **ENERGY EFFICIENCY CONVERSION, EXPRESSED ON ASH FREE BASIS CAEFDM=FC/(CRG*0.400) **CARBON EFFICIENCY, FRACTION, FOR TOTAL DRY MATTER FCARB=COMP(1) FPROT=COMP(2) FFAT =COMP(3) FLIGN=COMP(4) FOA =COMP(5) FMIN =COMP(6) TOTAL=COMP(1)+COMP(2)+COMP(3)+COMP(4)+COMP(5)+COMP(6)

205

32 33 34 35 36 37 38 39 40 41 42 43 44 45 46

BALANS-CRG*0.400 - CPG*0.273 - FC*1.000 **TOTAL MUST EQUAL 1.000, BALANS MUST REMAIN 0.000 TIMER DELT=l., FINTIM=l., PRDEl=l. PRINT CRG, CPG, FC, TOTAL, BALANS, EC, ECAF, ENEFF, CAEFDM **DATA TITLE RICE LEAVES PARAM LEG=0. TABLE COMP(1-6)=0.53, 0.20, 0.04, 0.04, 0.04, 0.15 END TITLE GROUNDNUT, SEED + POD TABLE COMP(l-6)-0.14, 0.27, 0.39, 0.14, 0.03, 0.03 PARAM LEG=1. END STOP ENDJOB

Listing 2. Module TIL to simulate development of tillers, florets and grains in rice. (To be inserted in module L1Q, Listing 4.) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

206

TITLE TIL Tillers, florets and grains (JANUARY 1989) **To replace line 36 in L1Q.CSM *To be included in INITIAL NTII =WLVI/WTI *To be included in DYNAMIC NTI =INTGRL(NTII,(GNTI-LNTI)*FADL) GNTI =DSTF*AMAXl(0.,(NTIP-NTI)/TCFT) LNTI =DSTD*AMAX1(0.,(NTI-NTIP)/TCDT) NTIP =CAGCR/CNTI DSTF =NOR(DST1-DS,DS-DST2) DSTD =NOR(DST1-DS,DS-(DST2+0.15)) CNTI =AFGEN(CNTIT,DS) NTIPL =NTI/NTII NFL GNFL NFLP CNFL NFLMX DSFL

=INTGRL(0.,GNFL*FADL) =DSFL*AMIN1(NFLMX-NFL,NFLP-NFL)/TCFF =CAGCR/CNFL =0.7*GGRMN =NFLMXT*NTI =NOR(DSF1-DS,DS-DSF2)

NGR GNGR NGRP NGRMX DSGR GGRMN GFP GGRMX WGR

=INTGRL(0. ,GNGR*FADL) =DSGR*AMAX1(0.,AMINl(NGRP-NGR,NGRMX-NGR)/TCFG) =CAGCR/GGRMN =NFL =NOR(DSG1-DS,DS-DSG2) =WGRMX/GFP =1./(1.33*DRR) =GGRMN*2. =WSO/(AMAX1(NGR,l000.))

GSOM

=NGR*GGRMX*AFGEN(GGRT,TPAA)

PARAM DSTl =0.3, DSFl =0.7, DSGl =0.95 PARAM DST2 =0.75,DSF2 =0.95,DSG2 =1.15 PARAM TCFT =15., TCFF =7., TCFG =3.,TCDT =10. PARAM WTI =1.0E-5, NFLMXT =100., WGRMX =23.5E-6 FUNCTION GGRT = 10.,0.0, 15.,0.0, 18.,0.75, ... 23.,1.0,27.,0.9,40.,0.0

40 41 42

FUNCTION CNTIT = 0.0,5.E-6, 0.3,5.E-6, 0.75,25.E-6, 1.0,75.E-6, 2.1,75.E-6 FINISH WGR =WGRMX, DS =2.0, CELVN-3.0

...

Listing 3. Basic module for crop growth simulation (L1D).

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

TITLE L1D (JULY 1987) FIXED IDATE,I,NL STORAGE RDTMT(365),TPHT(365),TPLT(365),RAINT(365), ... HUAAT(365),WDST(365),TKL(lO),TYL(lO) INITIAL WRTI =WLVI ALVI =WLVI/(SLC*AFGEN(SLT,DSI)) CPEW =1. DREW =1. PCEW =1. DYNAMIC **WEIGHTS OF CROP COMPONENTS **Explanation in sections 3.2, 2.2, 3.4 WLV =INTGRL(WLVI,GLV-LLV) WST =INTGRL(WSTI,GST*(l.-FSTR)) WIR =INTGRL(O.,GST*(FSTR*(FCST/O.444))-LSTR) WS0 =INTGRL(WSOI,GSO) WEPSO =WSO*FEPSO WRT =INTGRL(WRTI,GRT-LRT) WSS =WLV+WST+WSO+WIR WCR =WSS+WRT WLVD =INTGRL(0.,LLV) WRTD =INTGRL( 0. , LRT) **GROWTH RATES AND LOSS RATES **Explanation in sections 2.4, 3.2, 2.2 GLV =CAGLV/CRGLV GST =CAGST/CRGST GRT =CAGRT/CRGRT GSO =CAGSO/CRGSO LLV LRT LSTR

=WLV*AFGEN (LLVT , DS) =WRT*AFGEN(LRTT,DS) =INSW(AFGEN(CASTT,DS)-O.Ol,WIR*O.l,O.)

**CARBOHYDRATE AVAILABLE FOR GROWTH, EXPORT **Explanation in sections 3.2, 2.4, 2.3, 2.2 CAGCR =PCGW*0.682-RMCR*0.682+LSTR*1.111*0.947 CAGSS =CAGCR*AFGEN(CASST,DS)*CPEW CAGRT =CAGCR-CAGSS CAGLV =CAGSS*AFGEN(CALVT,DS) CAGST =CAGSS*AFGEN(CASTT,DS) CAGSO =CAGSS-CAGLV-CAGST CELV CELVN

=PCGW-(RMLV+RMST+O.5*RMMA) =INTGRL(0.,INSW(CELV,l.,-CELVN/DELT))

**PHOTOSYNTHESIS, GROSS AND NET **Explanation in sections 2.1, 3.3, 3.4 PCGW =PCGC*PCEW PCGC =FUPHOT(PLMX,PLEA,ALV,RDTM,DATE,LAT)

207

53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110

208

PLMX PLEA PCGT RCRT PCNT

=PLMXP*AFGEN(PLMTT,TPAD)*LIMIT(200.,600.,SLA)/SLC =PLEI*AFGEN(PLETT,TPAD) =INTGRL(0.,PCGW) =INTGRL(0.,RMCR+RGCR) =INTGRL(0.,PCGW-(RMCR+RGCR))

*RESPIRATION **Explanation in sections 2.4, 2.3 RMCT =INTGRL(0.,RMCR) RMCR =RMLV+RMST+RMSO+RMRT+RMMA RMLV =WLV*RMCLV*TPEM*0.75 RMST =WST*0.010*TPEM+WIR*0.0 RMRT =WRT*0.015*TPEM RMSO =AMIN1(1000.,WSO)*0.015*TPEM TPEM =Ql0**((TPAV-TPR)/10.) RMMA

=0.20*PCGW*0.5

RGCR RGLV RGST RGSO RGRT RLSR

=RGLV+RGST+RGSO+RGRT+RLSR =GLV*CPGLV =GST*CPGST =GSO*CPGSO =GRT*CPGRT =LSTR*l.lll*0.053*1.467.

**CARBON BALANCE CHECK **Explanation in section 3.4 CKCRD =FUCCHK(CKCIN,CKCFL,TIME) CKCIN =(WLV-WLVI)*FCLV+(WST-WSTI)*FCST+... (WSO-WSOI)*FCSO+(WRT-WRTI)*FCRT+WIR*O.444 CKCFL =PCNT*O.2727-(WLVD*FCLV+WRTD*FCRT) **LEAF AREA **Explanation in section 3.3 ALV =INTGRL(ALVI,GLA-LLA+GSA) GLA =GLV/SLN LLA =LLV/SLA GSA =0.5*GST/SSC SLN =SLC*AFGEN( SLT, DS) SLA =WLV+0.5*WST*(SLC/SSC))/ALV **PHENOLOGICAL DEVELOPMENT OF THE CROP **Explanation in section 3.1 DS =INTGRL(DSI,INSW(DS-l.,DRV,DRR)) DRV =DRCV*DRED*DREW*AFGEN(DRVTT,TPAV) DRED =AFGEN(DRDT,DLP) DRR =DRCR*AFGEN(DRRTT,TPAV) **WEATHER DATA AND TIME **Explanation in chapter 6 and section 3.4 RDTM =RDTMT(IDATE)*RDUCF RDTC,DLA,DLP=SUASTR(DATE,LAT) TPAV =TPLT(IDATE)+TPHT(IDATE))/2. TPAD =(TPHT(IDATE)+TPAV)/2. DATE =AMOD(DATEB+TIME+364.,365.)+1. IDATE =DATE

111 112 113 114 115 116 117 118 119 120 121 122 123 124 125

**RUN CONTROL AND OUTPUT METHOD RECT TIMER DELT=1., TIME=0., FINTIM=1000., PRDEG=10., OUTDEG=10. FINISH DS =2., CELVN =3. PRINT DATE, WLV, WST, WIR, WSO, WRT, GLV, GST, GSO, GRT,... SLA, PLMX, ALV, DS, TPAV, RDTM, PCGT, RCRT, RMCT PRTPLOT WLV, WLVT, WLVST, WLVSO PAGE GROUP WLVT =WLV+WLVD WLVST =WLVT+WST+WIR WLVSO =WLVST+WSO HI =WSO/WSS RSH =RMLV+RMST+RMSO+RMMA+RGLV+RGST+RGSO+RLSR WSTR =WST+WIR

Listing 4. Crop growth module with quarter-day time periods (L1Q). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

TITLE L1Q (AUGUST 1988) FIXED IDATE, MIN0, MAXO STORAGE RDTMT(365), TPHT(365), TPLT(365), RAINT(365), HUAAT(365), WDST(365) INITIAL WRTI WARI ALVI

=WLVI =WLVI*0.05 =WLVI/(SLC*AFGEN(SLT,DSI))

DYNAMIC **WEIGHT CROP COMPONENTS **Explanation in sections 3.2, 2.4, 2.2, 3.4 WLV =INTGRL(WLVI,(GLV-LLV)*FADL) WAR =INTGRL(WARI,(GAR-CUGCR)*FADL) WARR =WAR/(WLV+l.E-l0) WST =INTGRL(WSTI,GST*FADL) WSR =INTGRL(0.,(GSR-LSR)*FADL) WSO =INTGRL(WSOI,GSO*FADL) WEPSO =WSO*FEPSO WRT =INTGRL(WRTI, (GRT-LRT)*FADL) WSS =WLV+WST+WSO+WAR+WSR WCR =WSS+WRT WIR =INTGRL(0.,((GAR-CUGCR)*0.900+(GSR-LSR))*FADL) WLVD =INTGRL(0.,(0.5*LLV)*FADL) WRTD =INTGRL(0.,LRT*FADL) *GROWTH RATES AND LOSS RATES **Explanation in sections 2.4, 2.2, 3.2, 3.4 GAR =PCGD*0.682-RMCR*0.682+... LSR*1.111*0.947+0.5*LLV*(FCLV/0.400)*0.947 GLV =CAGLV/CRGLV GST =CAGST/CRGST*... (CRGST*(l.0-FSTR)/(FSTR*(l.111/0.947-CRGST)+CRGST)) GRT =CAGRT/CRGRT GSO =AMIN1(CAGSO/CRGSO,GSOM) GSOM =(WSO+10.)*GSORM GSR =INSW((WST+WSR)*(FSTR+0.10)-WSR,0.,GSRP) GSRP =(CAGSS-GLV*CRGLV-GST*CRGST-GSO*CRGSO)*0.947/1.111 LLV

=WLV*0.15*MCLV

209

41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98

210

LRT LSR MCLV MCRT MCSR GSOAVM GSOAV

=WRT*0.l5*MCRT =WSR*0.20*MCSR =INSW(GS0AVM*0.8-GSOAV-l0.,0.,l.) =MCLV =INSW(GSOAVM*1.0-GSOAV-10.,0.,1.) =INTGRL(0.,AMAX1(0.,GSOAV-GSOAVM)*FADL) =INTGRL(0.,((GSO-GSOAV)/2.)*FADL)

**CARBOHYDRATE AVAILABLE AND CONSUMED FOR GROWTH, EXPORT **Explanation in sections 3.2, 2.4, 2.2, 3.4 CUGCR =CUGLV+CUGST+CUGSO+CUGRT+CUGSR CUGLV =GLV*CRGLV CUGST =GST*CRGST CUGSO =GSO*CRGSO CUGRT =GRT*CRGRT CUGSR =GSR*1.111/0.947 CAGCR CAGSS CAGRT CAGLV CAGST CAGSO

=LIMIT(0.,(WAR-0.05*WLV)/(DELT*FADL),(WAR-0.05*WLV)*1.5) =CAGCR*AFGEN(CASST,DS) =CAGCR-CAGSS =CAGSS*AFGEN(CALVT,DS) =CAGSS*AFGEN(CASTT,DS) =CAGSS-CAGLV-CAGST

CELV

=INTGRL(10.,( PCGD-(RMLV+RMST+0,5*RMMA))*FADLINSW(DT1ME-0.0l,CELV/DELT,0.)) CELVN =INTGRL(0.,INSW(CELV,l.,-CELVN/DELT)) **PHOTOSYNTHESIS, GROSS AND NET **Explanation in sections 2.1, 3.3, 3.4 PCGD =PCGC/(DLA/24.) PCGC =FUPHOT(PLMX,PLEA,ALV,RDTM,DATE,LAT) PLMXT =PLMXP*LIMIT(200.,600.,SIA)/SLC PLMX =PLMXT*AFGEN(PLMTT,TPAA)*(l.-ELV/8000.)*... AFGEN(PLMHT,0.75*VPD)*INSW(WARR-0.3,1.,0.3) PLEA =PLEI*AFGEN(PLETT,TPAA) PCGT RCRT PCNT

=INTGRL(0.,PCGD*FADL) =INTGRL(0.,(RMCR+RGCR)*FADL) -(RMCR+RGCR))*FADL) =INTGRL(0.,(PCGD

**RESPIRATION **Explanation in sections 2.4, 2.3 RMCT =INTGRL(0.,RMCR*FADL) RMCR =RMLV+RMST+RMSO+RMRT +RMMA RMLV =INSW(0.5-NIGHT,RMLVN,RMLVD) RMLVN =WLV*RMCLV*TPEM RMLVD =WLV*RMCLV*TPEM*0.5 RMST =WST*0.010*TPEM+WSR*0.0+WAR*0.0 RMRT =WRT*0.015*TPEM RMSO =AMIN1(1000.,WS0)*0.015*TPEM TPEM =Q10**((TPAA-TPR)/l0.) RMMA =0.20*PCGDV*0.5 PCGDV =INTGRL(0.,(PCGD-PCGDV)/l.*FADL) RGCR RGLV

=RGLV+RGST+RGSO+RGRT+RGSR+RLSR+RLLV =GLV*CPGLV

...

99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156

RGST RGSO RGRT RGSR RLSR RLLV

=GST*CPGST = GSO*CPGSO =GRT*CPGRT =GSR/0.947*1.111*0.053*1.467 =LSR*l.lll*0.053*1.467 =(LLV*0.5)*(FCLV/0.400)*0.053*1.467

*CARBON BALANCE CHECK **Explanation in section 3.4 CKCRD =FUCCHK(CKCIN,CKCFL,TIME) CKCIN =(WLV-WLVI)*FCLV+(WST-WSTI)*FCST+... (WSO-WSOI)*FCSO+(WRT-WRTI)*FCRT+WIR *0.444 CKCFL =PCNT*0.2727-(WLVD*FCLV+WRTD*FCRT) **AREA OF LEAVES **Explanation in section 3.3 ALV =INTGRL(ALVI,(GLA-LLA+GSA)*FADL) GLA =GLV/SLN LLA =LLV/SLA GSA =0.5*GST/SSC SLN =SLC*AFGEN(SLT,DS) SLA =(WLV+0.5*WST*(SLC/SSC))/ALV **PHENOLOGICAL DEVELOPMENT **Explanation in section 3.1 DS =INTGRL(DSI,INSW(DS-l.,DRV,DRR)*FADL) DRV =DRCV*DRED*AFGEN(DRVTT,TPAA) DRED =AFGEN(DRDT,DLP) DRR =DRCR*AFGEN(DRRTT,TPAA) *WEATHER DATA, TIME AND DATE **Explanation in chapter 6 and sections 1.4,3.4 RDTM =INSW(0.5-NIGHT,0.,RDTMT(IDATE)*RDUCF) RDTC,DLA,DLP =SUASTR(DATE,LAT) TPAA =FUTP(IDATE,DTIME,TPHT,TPLT,0.15,0.45,0.90,0.60) VPD =AMAXl(0.,FUVP(TPAA)-HUAAT(IDATE)) DATE IDATE DTIME NIGHT

=AMOD(DATEB+TIME+364.,365.)+1. =DATE =TIME-AINT(T1ME) =INSW(AND(DTIME-0.l,0.6-DTIME)-0.1,l.,0.)

**RUN CONTROL AND OUTPUT METHOD RECT TIMER DELT=0.25,TIME=0.,FINTIM=1000.,PRDEL10.,OUTDEL10. FADL =INSW(0.5-NIGHT,2.-DLA/12.,DLA/12.) FINISH DS =2., CELVN= 3. PRINT DATE, WLV ,WST ,WSO,WRT,WAR,WSR,WARR, GLV,GST,GSO, . . . GRT,GSRP,GSR,SLA,PLMX,ALV,DS, TPAA,PCGT,RCRT,RMCT,LLV PRTPLOT WLV, WLVT, WLVST, WLVSO PAGE GROUP WLVT =WLV+WLVD+WAR WLVST =WLVT+WST+WSR WLVSO =WLVST+WSO WSTR =WST+WSR HI =WSO/WSS PRTPLOT PCGD, RSH, PCNSH, WARR, TPAA

211

157 158 159

PAGE GROUP-3 RSH =RMLV+RMST+RMSO+RMMA +RGLV+RGST+RGSO+RGSR+RLSR+RLLV PCNSH =PCGD-RSH

Listing 5. Crop data for rice (variety IR36) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49

212

TITLE OSIR36.DAT ORYZA SATIVA, RICE, CV IR36 **PHOTOSYNTHESIS AND RESPIRATION; TABLES 3,4,5,8,11,23 PARAMETER PLMXP =47., PLEI =0.50 FUNCTION PLMTT = 0.0,0.0, 10.,0.0, 25.,1.00, 30 ., 1.00, ... 42.,0.0, 45.,0.0 FUNCTION PLMHT = 0.0,1.00, l.0,l.0, 2.0,0.99, 3.0,0.86, ... 4.0,0.71 FUNCTION PLETT = 0.0.1.0, 15.,1.0, 25.,0.90, 35 ., 0.60, ... 45.,0.2, 50.,0.01 PARAMETER CRGLV=1.326, CRGST=1.326, CRGSO=1.462, CRGRT=1.326 PARAMETER CPGLV=0.408, CPGST=0.365, CPGSO=0.357, CPGRT=0.365 PARAMETER FCLV =0.419, FCST =0.431, FCSO =0.487, FCRT =0.431 PARAMETER RMCLV=0.02 , TPR =25., Q10 =2. **CONSISTENCY CHECK: 12/30*CRGLV=l,O*FCLV+12/44*CPGLV **BIOMASS PARTITIONING AND AGING; TABLES 7,17 FUNCTION CALVT = 0.0,0.51, 0.5,0.51, 0.6,0.47, 0.7,0.32, ... 0.8,0.26, 1.0,0.00, 1.1,0.00, 2.5,0.00 FUNCTION CASTT = 0.0,0.49, 0.5,0.49, 0.6,0.53, 0.7,0.68, ... 0.8,0.74, 1.0,1.00, 1.1,0.27, 1.2,0.00, ... 2.1,0.0 FUNCTION CASST = 0.0,0.86, 0.5,0.86, 0.6,0.86, 0.7,0.95, ... 0.8,0.94, 1.0,0.89, 1.1,1.00, 2.5,l.00 PARAMETER FSTR =0.25, FEPSO =0.8, GSORM =0.5 FUNCTION LLVT = 0.0,0.0, 1.0,0.0, 1.3,0.007, 1.8,0.012, ... 2.5,0.012 FUNCTION LRTT = 0.0,0.0, 1.0,0.0, 1.3,0.011, 1.8,0.010, ... 2.5,0.0l0 **PHENOLOGICAL DEVELOPMENT; TABLES 12,13,14,15,16,19,20,21 PARAMETER DRCV =0.013, DRCR =0.028 FUNCTION DRVTT = 10.,0.10, 19.,0.80, 25.,1.00, 27 ., 1.10, ... 32.,1.20, 40.,1.00, 45.,1.00 FUNCTION DRRTT = 10.,0.45, 19.,0.75, 25.,0.90, 28 ., 1.00, ... 30.,1.10, 40.,1.10, 45.,1.10 FUNCTION DRDT = 0.0,1.0, 24.,1.0 FUNCTION DRWT = 0.0,1.0, 1.,1.0 PARAMETER SLC = 370., SSC =1000., WDLV =0.015 FUNCTION SLT = 0.0,0.82, 0.6,1.0, 2.1,l.0 FUNCTION PLHTT = 0.0,0.0 , 1.0,1.0, 2.1,l.0 **WATER RELATIONS AND ROOT GROWTH; TABLES 22,24,25 PARAMETER WSSC =0.5, WFSC =1.0, FIEC =0.65 PARAMETER ZRTMC =0.7, GZRTC =0.03 **INITIALIZATION PARAMETER DATEB =197. PARAMETER WLVI =6.8, WSTI =6.8, WSOI =0.0 PARAMETER DSI =0.18, ZRTI =0.20 FINISH DS =2.0, CELVN =3.0, TPAV =3.0

Listing 6. A program to simulate soil water balance dynamics for homogeneous unsaturated soils 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55

(SWD).

TITLE SWD SOIL WATER DYNAMICS (MARCH 1988) STORAGE MS(10),MFLP(10),TKL(10),K(10),KAV(11),DZ(11),FLX(11) FIXED I,NL WCL =INTGRL(WCLI,DWCLDT,10) **WATER CONTENTS OF 10 LAYERS PROCEDURE DWCLDT=PRDW(FLX) **CALCULATION OF RATE OF CHANGE OF WATER CONTENT DO 1 I=1,NL DWCLDT(I)-(FLX(I)-FLX(I+l))/(TKL(I)*CONV) 1 CONTINUE ENDPROCEDURE PROCEDURE FLX=PRFLX((K,DZ,MFLP,MFLPTP,MFLPBT) **CALCULATION OF FLUXES BETWEEN LAYERS (DOWNWARDS POSITIVE) KAV(1) =KSAT*EXP (-KMSA*MSTP) FLX(1) =(MFLPTP-MFLP(l))/DZ(l)+KAV(l) DO 2 I=2 ,NL KAV(1) =SQRT(K(I)*K(I-1)) FLX(1) =(MFLP(I-1)-MFLP(I))/DZ(I)+KAV(I) 2 CONTINUE KAV(NL+l) =KSAT*EXP(-KMSA*MSBT) FLX(NL+l) =(MFLP(NL)-MFLPBT)/DZ(NL+l)+KAV(NL+l) ENDPROCEDURE PROCEDURE MFLP,K,MFLPTP,MFLPBT=PRMFLP(WCL) **CALCULATION MATRIC SUCTION AND HYDRAULIC CONDUCTIVITY MFLPTP=–(KSAT/KMSA)*(l.–EXP(–KMSA*MSTP)) DO 3 I=1,NL MS(I) =EXP(SQRT(–l.*ALOG(WCL(I)/WCST)/MSWCA))-1. K(I) =KSAT*EXP(-KMSA*MS(I)) IF (K(I).LE.1.E-10)K(I)=0. MFLP(I) =-(KSAT/KMSA)*(l.-EXP(-KMSA*MS(I))) 3 CONTINUE MFLPBT=-(KSAT/KMSA)*(l.-EXP(-KMSA*MSBT)) ENDPROCEDURE PROCEDURE DZ=PRDZ(TKL) **CALCULATE DZ DZ(1) =0.5*CONV*TKL(1) DO 4 I=2 , NL DZ(1) =0.5*CONV*(TKL(I)+TKL(I-1)) 4 CONTINUE DZ(NL+l) =0.5*CONV*TKL(NL) ENDPROCEDURE **RUN CONTROL, OUTPUT METHOD RECT TIMER FINTIM=20., PRDEL=2.5, DELT=0.01 PRINT WCL(l-10),MFLP(1) PARAM TABLE PARAM

MSTP=1000., MSBT=0., CONV=100. TKL(1-10)=10*0.10 WCLI(l-10)=10*0.05, MSWCA=0.0164, KSAT=5.0, KMSA=0.0231, WCST=0.503

213

56 57 58 59 60 61 62 63 64 65 66 67 68 69

PARAM NL=l0 END TITLE CLAY (TYPE 17) TABLE WCLI(1-10)=10*0.05 PARAM MSWCA-0.0088, KSAT=3.5, KMSA=0.0174, WCST=0.453 END TITLE SAND (TYPE 4) TABLE WCLI(1-10)=10*0.05 PARAM MSWCA=0.0255, KSAT=50.0, KMSA=0.0500, WCST=0.364 TIMER DELT=0.002 END STOP ENDJOB

Listing 7. Module for canopy transpiration at Production Level 2 (L2C).

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

214

TITLE L2C (JULY 1987) *To be included in INITIAL: TPSI =(TPLT(IDATE)+TPHT(IDATE))/2. IDATE =DATEB *To be included in DYNAMIC: **EFFECTS OF WATER SHORTAGE **Explanations in sections 4.2, 4.3 CPEW =AMIN1(1.,0.5+TRW/(TRC+l.E-10)) DREW =AFGEN(DRWT,TRW/(TRC+l.E-10)) PCEW =TRW/(TRC+l.E-l0) PLEH =AFGEN(PLMHT,VPDC) **POTENTIAL TRANSPIRATION AND DIFFUSION RESISTANCES CANOPY **Explanation in sections 4.1, 4.4 TRC =TRCPR*(l.-EXP(-0.5*ALV))+TRCPD*AMIN1(2.5,ALV) TRCPR,TRCPD=SUEVTR(RDTC,RDTM,0.25,DLA/24 ., TPAD,VPA, ... RSLL,RSBL,RSTL) RSLL

=LIMIT(RSLLM,2000.,(CO2E-C02I)/(PLNA+1.E-10)*... (68.4*24.0/1.6)-RSBL-RSTL) C02I =COPE*FIEC RSLLM =(C02E-C02I)/(PLMX*0.9+1.E-10)*(68.4/1,6)-10. PLNA =(PCGC/(DLA/24.)-RMLV*0.33)/(AMIN1(2.5,ALV+l.E-10)) RSBL =0.5*172.*SQRT(WDLV/(WDSAD*0.6)) RSTL =FURSC(WDSAD,AMINl(2,5,ALV),PLHT,2.)

TRRM

=TRC/(ZRT+l.E-l0)

*ROOTED DEPTH AND CROP HEIGHT **Explanation in section 4.2 ZRT =INTGRL(ZRTI,GZRT*AND(ZRTM-ZRT,l.0-DS)) ZRTM =AMINl(ZRTMC,ZRTMS,TKLT) GZRT =GZRTC*WSERT*TERT PLHT

=AFGEN(PLHTT,DS)

**POTENTIAL EVAPORATION SOIL **Explanation in section 5.1 EVSC =EVSPR*EXP(-0.5*ALV)+EVSPD EVSPR,EVSPD=SUEVTR(RDTC,RDTM,RFS,l.00,TPAV,VPA,...

42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63

0.00,RSBS,RSTS) RFS RSBS WDSS RSTS

=RFSD*(l.-0.5*WCLl/WCSTl) =172.*SQRT(WDCL/WDSS) =FUWRED(WDLV,ALV,PLHT,WDSAV) =FIJRSC(WDSAV,l.,0.l*PLHT,0.63*PLHT)

**EXTRA WEATHER DATA **Explanation in section 6.1, 5.1 WDSAV =AMAX1(0.2,WDST(IDATE)) WDSAD =1.33*WDSAV VPA =AMIN1(FWP(TPAD),HUAAT(IDATE)) RAIN =RAINT(IDATE) VPDC DSIR C02E TPS

=(FUVP(TPAD)-VPA)*AMIN1(1.,30./RSTL) =INTGRL(l.,... INSW(RAINT(IDATE+1)-0.5,1.,1.00001-DSLR)/DELT) =340.*0.88**(ELV/lOOO.) =INTGRL(TPSI,(TPAV-TPS)/5.)

WUPC WUPT

=TRC *l.E4/(PCGC+l.E-10) =TRWT*l.E4/(PCGT+l.E-10)

Listing 8. Module to simulate water movement in free draining soils (Production Level 2, permanently unsaturated soils: L2SU). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

TITLE L2SU water balance SAHEL (JULY 1987) *To be included in INITIAL: WLlI =WCLIl*TKLl*l.E4 WL2I =WCLI2*TKL2*1. E4 WL3I =WCLI3*TKL3*1.E4 TKLT =TKLl+TKL2+TKL3 *To be included in DYNAMIC: **ACTUAL TRANSPIRATION (WATER UPTAKE) **Explanation in section 5.2 TRW =TRWLl+TRWL2+TRWL3 TRWLl =TRRM*WSEl*ZRTl TRWL2 =TRRM*WSE2*ZRT2 TRWL3 =TRRM*WSE3*ZRT3 TRWT =INTGRL(0.,TRW) WSE1 WSE2 WSE3

=FUWS(TRC,ALV,WCLl,WSSC,WFSC,WCWPl,WCFCl,WCSTl) =FUWS(TRC,ALV,WCL2,WSSC,WFSC,WCWP2,WCFC2,WCST2) =FUWS(TRC,ALV,WCL3,WSSC,WFSC,WCWP3,WCFC3,WCST3)

ZRTl ZRT2 ZRT3

=LIMIT(0.,TKLl,ZRT) =LIMIT(0.,TKL2,ZRT-TKLl) =LIMIT(0.,TKL3,ZRT-TKLl-TKL2)

**GROWTH ROOTED DEPTH **Explanation in subsection 4.2.3 WSERT =1NSW(ZRT-TKLl,WSEl,INSW(ZRT-TKLl-TKL2, WSE2,WSE3)) TERT =AFGEN(PLMTT,TPS) **AVAILABLE AND TOTAL SOIL WATER **Explanation in sections 5.2, 5.4 WCLl =WLl/(TKLl*l.E4)

215

33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74

WCL2 WCL3

=WL2/(TKL2*l.E4) =WL3/(TKL3*1.E4)

WL1 WL2 WL3 WCUM WLFLl WLFL2 WLFL3 WLFL4

=INTGRL(WL1I,(WLFL1-WLFL2-EVSWl-TRWLl)*l0.0) =INTGRL(WLPI,(WLFL2-WLFL3-EVSW2-TRWL2)*10.0) =1NTGRL(WL3I,(WLFL3-WLFL4-EVSW3-TRWL3)*10.0) =(WLl+WL2+WL3)/10.0 =RAIN*(l.0-FRNOF) =AMAX1(0.,WLFL1-(WCFC1*TKL1*1000.-WLl*0.10)/DELT) =AMAX1(0.,WLFL2-(WCFC2*TKL2*1000.-WL2*0.10)/DELT) =AMAX1(0.,WLFL3-(WCFC3*TKL3*1000.-WL3*0.10)/DELT)

**EVAPORATION **Explanation in section 5.2 EVSW =INSW(DSLR-l.l,EVSH,EVSD) EVSH =AMINl(EVSC, ... (WL1*0.0001-WCAD1*TKL1)*1000./DELT+WLFL1) EVSD =AMINl(EVSC, ... 0.6*EVSC*(SQRT(DSLR)-SQRT(DSLR-l.))+WLFLl) EVSWl =EVSW*(FEVLl/FEVLT) EVSW2 =EVSW*(FEVLZ/FEVLT) EVSW3 =EVSW*(FEVL3/FEVLT) FEVLl =AMAXl(WLl-WCADl*TKLl*l.E4,0.)* ... EXP(-EES*(0.25*TKLl)) FEVL2 =AMAXl(WL2-WCAD2*TKL2*l.E4,0.)*... EXP(-EES*(TKLl+(O,25*TKL2))) FEVL3 =AMAXl(WL3-WCAD3*TKL3*1.E4,0.)* ... EXP(-EES*(TKLl+TKL2+(0.25*TKL3))) FEVLT =FEVLl+FEVL2+FEVL3 **WATER BALANCE CHECK **Explanation in section 5.4 CKWRD =FUWCHK(CKWFL,CKWIN,TIME) CKWFL =INTGRL(0.,(WLFLl-EVSW-TRW-WLFL4)*10.) CKWIN =WLl-WLlI+WL2-WL21+WL3-WL31 **OUTPUT PRTPLOT TRC,TRW,EVSC,EVCW PAGE GROUP PRTPLOT WCLl,WCL2,WCL3,RAIN PAGE GROUP=3 PRTPLOT ZRT,WSEl,WUPC

Listing 9. Module to simulate water movement in soils with impeded drainage (Production Level 2, temporarily saturated soils: L2SS). 1 2 3 4 5 6 7 8 9 10 11 12

216

TITLE L2SS water balance SAWAH includes subroutines (AUGUST 1988) *before INITIAL FIXED ITYL STORAGE TRWL(10),WCLEQI(10),WCLMQI(10),KMSA1T(20),KMSA2T(20) STORAGE KMSMXT(20),KSTT(20),MSWCAT(20),WCSTT(20) / COMMON /SLDPTH/ ZL(10) / COMMON /VOLWAT/ WCAD(l0),WCFC(l0),WCST(10),WCWP(l0) / COMMON /HYDCON/ KMSMX(l0),KMSA1(10),KMSA2(10),KST(10) / COMMON /PFCURV/ MSWCA(10) *To be included in INITIAL: **INITIALIZATION OF LAYER SOIL WATER CONTENTS

13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70

**Explanation in Section 5.1, 5.3, Subsection 5.4.5 PROCEDURE WCLQT1,WCUMI=PRWCLI(WCLEQI,WCLMQI,WCLISC) WCUMI =0. DO 1 I=1,NL WCLQTI(1) =INSW(WCLISC,WCLEQI(I),WCLMQI(I)) WCUMI =WCUMI+WCLQTI(I)*TKL(I)*1000. 1 CONTINUE ENDPROCEDURE WCLISC =AND(-WCLIS,ZWI-TKLT)-0.5 ZWI =AFGEN(ZWTB,DATEB) PROCEDURE WCLEQI,TKLT=PRWCLE(NL,TKL,TYL,ZWI) DO 2 I=1,NL ITYL =TYL(I) KST(I) =KSTT (ITYL) KMSMX(I)=KMSMXT(ITYL) KMSAl(I)=KMSAlT(ITYL) KMSAZ(I)=KMSA2T(ITYL) MSWCA(I)=MSWCAT(ITYL) WCST(I) =WCSTT (ITYL) WCFC(I) =FUWCMS(I,l00.0) WCWP(I) =FUWCMS(I,1.6E4) WCAD(I) =FUWCMS(I,l.0E7) 2 CONTINUE WCLCH,WLOCH,WCLEQI,EVSW,RUNOF,DRSL,WCUMCH,ZECH,TKLT... =SUSAWA(l,WCLQT,WL0QT,NL,TRWL,EVSC,RAIN,ZWI,TKL,... TYL,l.0,DTMIN,DTMXl,DTFX,WLOMX,ZEQT,CSA,CSB,CSC2) ENDPROCEDURE ZEQTI =0.02*(1.-WCLl/WCSTl) WCLl =WCLQTI(l) WCSTl =WCST(l) *To be included in DYNAMIC: **SOIL WATER, PONDED WATER, DEPTH EVAPORATION FRONT **Explanation Sections 5.3, 5.4.4, 5.4.5 WCLQT =INTGRL(WCLQTI,WCLCH,l0) WL0QT =INTGRL(WL0QTI,WL0CH) WCUM =INTGRL(WCUMI ,WCUMCH*1000.) ZEQT =INTGRL(ZEQTI ,ZECH) WCLCH,WL0CH,WCLEQI,EVSW,RUNOF,DRSL,WCUMCH,ZECH,TKLT=... SUSAWA(2, WCLQT, WL0QT ,NL,TRWL, EVSC , RAIN, ZW ,TKL,TYL,... 1.O,DTMIN,DTMXl,DTFX,WLOMX,ZEQT,CSA,CSB,CSC2) ZW =AFGEN(ZWTB,DATE) WCLl =WCLQT(l) **ACTUAL TRANSPIRATION AND EFFECT WATER STRESS **Explanation Subsections 4.2.2, 5.3.7 TRWT =INTGRL(0.0,TRW) PROCEDURE TRW,TRWL,WSERT=PRTRAN(TRC,ALV,WCLQT,ZRT,TRRM) TRW =0.0 ZLL =0.0 WSERT =0.0 DO 3 I=1,NL WSE=FUWS(TRC,ALV,WCLQT(I),WSSC,WFSC,WCWP(I),WCFC(I),WCST(I)) ZRTL =AMINl(TKL(I),AMAXl(ZRT-ZLL,0.)) WLA =AMAX1(0.,(WCLQT(I)-WCWP(I))*TKL(I)*l000.)

217

71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91

TRWL(I)=AMINl(WSE*ZRTL*TRRM,WLA/DELT) TRW =TRW+TRWL(I) WSEE =INSW(AND(WCLQT(I)+0.05-WCST(I),ZRT–0.2)–0.5,WSE,0.) IF(ZRT.LT.(ZLL+TKL(I)).AND.(ZRT.GE.ZLL)) WSERT =WSEE ZLL =ZLL+TKL(I) 3 CONTINUE ENDPROCEDURE TERT =AFGEN(PLMTT,TPS) **WATER BALANCE CHECK **Explanation Section 5.4.4 CKWIN =INTGRL(0.,(WCUMCH+WL0CH)*1000.) CKWFL =INTGRL(0.,RAIN–RUNOF–EVSW–TRW–DRSL) CKWRD =FUWCHK(CKWFL,CKWIN,TIME) *OUTPUT PRTPLOT EVSW,TRW,DRSL,RAIN,RUNOF PAGE GROUP=3 PRTPLOT WCLQT(l),WCLQT(3),WCLQT(5),WCLQT(7),WCUM PAGE GROUP=4 PRTPLOT ZRT,ZEQT,ZW,WL0QT

Listing 10. Data characterizing a loamy soil for L2SU and L2SS. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

218

TITLE STANDARD DATA LOAMY SOIL, DEEP GROUNDWATER *DATA FOR MODULE L2SU PARAMETER TKLl =0.2, TKL2 =0.3, TKL3 =0.5 PARAMETER WCFCl =0.36, WCWP1 =0.11, WCAD1 =0.01, PARAMETER WCFC2 =0.36, WCWP2 =0.11, WCAD2 =0.01, PARAMETER WCFC3 =0.36, WCWP3 =0.11, WCAD3 =0.01, INCON WCLI1 =0.36, WCLI2 =0.36, WCLI3 =0.36

TABLE (APRIL 1988)

WCST1 =0.50 WCST2 =0.50 WCST3 =0.50

*DATA FOR MODULE L2SS PARAMETER NL =10 TABLE TKL(1–10)=10*0.10, TYL(1–10)=10*13., WCLMQI(1-10)=10*0.36 PARAMETER WCLIS =-1., WL0MX =0.02 FUNCTION ZWTB =0.,3.0, 366.,3.0 INCON WL0QTI=0.0 PARAMETER DTMIN =0.001, DTMXl =0.1, DTFX =0.03 **SURFACE AND OTHER SOIL CHARACTERISTICS PARAMETER FRNOF =0.0, RFSD =0.2, WDCL =0.05, ZRTMS =0.9 PARAMETER EES =20., CSC2 =0.1, CSA =0.15, CSB =10. **CHARACTERISTICS SOIL TYPES 1-20 TABLE KMSA1T(1-20)= ... .1960,.1385,.0821,.0500,.0269,.0562,.0378,.0395,.0750,.0490,... .0240,.0200,.0231,.0353,.0237,.0248,.0274,.0480,.0380,.1045 TABLE KMSA2T(1-20)= ... .08, .63, 3.30,10.90,15.00,5.26, 2.10,16.40, .24,22.60,... 26.50,47.30,14.40,33.60,3.60,1.69,2.77,28.20,4.86,6.82 TABLE KMSMXT(1-20)= ... 80.0, 90.0,125.0,175.0,165.0,100.0,135.0,200.0,150.0,130.0,... 300.0,300.0,300.0,200.0,300.0,300.0,300.0, 50.0, 80.0, 50.0 TABLE KSST(1-20)=... 1120.00,300.00,110.00, 50.00, 1.00, 2.30, .36, 26.50,... 16.50, 14.50, 12.00, 6.50, 5.00, 23.50, 1.50, .98,... 3.50, 1.30, .22, 5.30

35 36 37 38 39 40

TABLE MSWCAT(1-20)= ... .0853,.0450,.0366,.0255,.0135,.0153,.0243,.0299,.0251,.0156,... .0186,.0165,.0164,.0101,.0108,.0051,.0085,.0059,.0043,.0108 TABLE WCSTT(1-20) -... .3950,.3650,.3500,.3640,.4700,.3940,.3010,.4390,,4650,.4550,... .5040,.5090,.5030,.4320,.4750,.4450,.4530,.5070,.5400,.8630

Listing 11. Daily total global radiation in MJ m-2 d-1.

(The CSMP table is formatted to quickly locate any calendar date or Julian date.) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

TITLE METEO LOS BANOS (IRRI), PHILIPPINES, 1984 PARAMETER LAT =14.17, ELV =l0., RUDCF =1.E6 TABLE RDTMT(1-365) = ... 9.85,12.98, 8.88, 4.57,10.52,13.59, 9.63,14.80,12.52,11.66, ... 9.91,17.65,17.33,13.19,15.76,13.51,12.05,11.95,14.80,11.84, ... 18.33,16.08, 5.00,10.02, 7.39, 8.17,10.77, 7.46,16.01,11.31, ... 14.02, ... 14.58,12.27,16.87,20.89,10.27,13.91,11.95, 9.77,12.87, ... 16.72,19.50,16.51,17.97,21.96,17.33,20.25,20.93,13.98,16.58, ... 20.00,17.37,15.44,11.31,21.35,23.03,13.73,21.29,15.30, ... 17.58, ... 21.18,22.04,18.47,13.91, 8.56,14.09,20.54,21.75,22.14,14.69, .... 20.75,15.94,20.64,22.71,21.18,24.35,25.56,18.36,23.92,25.35, ... 25.42,24.53,16.12,23.28,15.37,20.22,12.77,14.44,22.32,21.07, ... 23.57,25.38,17.58,21.71,25.46,20.25,26.06,25.49,23.21,23.96, . . . 23.78,20.29,24.28,24.17,24.28,23.46,21.75,15.73,17.51,20.18, ... 20.18,20.18,23.89,12.63,22.60,20.57,19.50,16.05,16.79,16.15, ... 11.41,19.40,10.70,18.54,11.88,17.11,16.87,20.68,17.79,14.05, ... 21.14,25.46,22.89,24.53,17.61,15.08.19.65,21.75,18.72,17.15, ... 14.41,16.19,12.77,20.97,21.82,21.03,21.93,23.67,24.35,16.15, . . . 19.57, ... 17.40,22.60,19.90,19.72,13.62,18.40,13.23,18.29,21.86, ... 16.33,13.16,18.93,21.86,19.79,11.20, 9.17,15.05, 6.49,10.41, ... 11.45, 5.25, 8.99, 6.85,13.37, 8.88, 7.49, 8.99,22.36,15.58, ... 5.25, ... 7.56,12.52,19.93,18.90,10.45,11.98,17.65,17.08,15.48, ... 18.26,19.00,23.60,23.17,17.40,20.68,24.28,19.25,16.90,21.64, ... 21.78,16.83,22.75,20.43,14.44,17.11,17.29,20.22,23.42,22.36, ... 13.66,14.84, ... 15.12,12.70,21.03,17.83, 9.70, 6.60,13.09, 7.53, ... 16.19,18.50,12.13, 9.02, 5.39, 3.25, 9.13,16.37,10.84,14.26, ... 17.08,19.97,23.07, 6.67, 5.67,11.56,10.73,11.02, 9.74, 9.10, ... 8.99,12.13,16.15, ... 18.11,18.22,11.16,22.89,14.90,12.48,24.71, ... 21.35,23.10,20.07,19.04,23.10,24.60,19.08,20.22,20.75,18.08, ... 21.29,16.87, 9.31,10.84,13.55,17.93,11.27, 9.99,10.06,16.19, ... 16.65,22.42,20.04, ... 15.83,20.89,12.73, 8.31,13.37,11.88,10.81, ... 8.92, 7.96, 7.24,20.25,19.04, 9.74,18.93,15.73,21.64,21.96, ... 11.98, 3.64, 6.82, 1.65, 6.07,12.13,10.20,11.52,20.75,10.13, ... 14.48, 7.92,11.41, 3.68, ... 12.05,16.26,20.25,12.87, 5.28,18.47, ... 17.11,18.86,16.55,11.24,12.44,20.11,17.26,16.01,19.04,17.54, ... 16.65,16.58,18.40, 8.20,16.44,14.02,11.84, 8.06,17.86, 1.08, . . . 7.49, 7.78,13.51, 6.60, ... 5.28,15.16,14.16,10.27,12.27,13.69, . . . 9.85, 9.88,19.00,18.54,15.30,11.16, 7.10,10.09,13.55,11.06, ... 11.52,12.34,13.94,11.45,13.84,10.56, 9.56, 7.74, 9.53,11.70, ...

219

49 50 51 52 53 54 55 56

14.41,14.69, 8.17,15.26, 8.42 TABLE TABLE TABLE TABLE TABLE **Each

RAINT(1-365) = ............................................. TPLT(1-365) = ............................................. TPHT(1-365) = ............................................. HUAAT(1-365) = ............................................. WDST(1-365) = ............................................. table should contain 365 elements for one year simulation

Listing 12. Abbreviations in the modules of Listings 1-11. (For an explanation of CSMP functions and labels, see IBM (1975) or Basstanie & van Laar (1982).) Abbreviation Explanation

AFGEN ALV(1) AMAX1 AMINl AMOD AND

CSMP area CSMP CSMP CSMP CSMP

function leaves (initial) function function function and FORTRAN function

CAG(CR,LV,RT,SO,SS,ST) carbohydrates (glucose) available for growth of total crop (CR), leaves (LV), roots (RT), storage organs (SO), shoot plus storage organs (SS) and stems (ST) CALVT relation of fraction CAGLV/CAGSS to DS CASST relation of fraction CAGSS/CAGCR to DS CASTT relation of fraction CAGST/CAGSS to DS CELV carbohydrate export (glucose, 24 h total) from leaves plus stems, excluding remobilization CELVN number of days that CELV is negative CKCFL sum of integrated carbon fluxes into and out of the crop CKCIN carbon in the crop accumulated since simulation started CKCRD difference between carbon added to the crop since initialization and the net total of integrated carbon fluxes, relative to their sum CKWFL sum of integrated water fluxes into and out of soil compartments CKWIN change in total soil water content since initialization CKWRD difference between water added to the soil since initiation and the sum of integrated water fluxes, relative to this sum CNFL carbohydrates needed to initiate and maintain 1 floret CNTI carbohydrates needed to initiate and maintain 1 tiller CNTIT relation of CNTI to DS COMMON FORTRAN label CO2E CO 2 concentration ambient air CO2I CO 2 concentration in stomatal cavity CPEW effect of water stress on carbohydrate partitioning CPG(LV,RT,SO,ST) weight of CO 2 produced during formation (=growth) of of dry matter of leaves (LV), roots (RT), storage organs (SO) and stem (ST) CRG(LV,RT,SO,ST) weight of carbohydrates required for growth of leaves (LV), roots (RT), storage organs (SO), stems (ST) CSA soil evaporation constant ( A in Eq. 17)

220

Dimension

– ha ha -1

– – – –

kg ha -l d -1

– – – kg ha -l d -1 d kg ha -1 kg ha -1

– mm mm

– kg ha -l d -1 kg ha -l d -1

vppm vppm

— kg kg -1 kg kg -1 cm 2 d -l

CSB soil evaporation constant ( B in Eq. 17) CSC2 soil evaporation constant (c2 in Eq. 12) CUG(CR,LV,RT,SO,SR,ST) weight of carbohydrates used for growth of the whole crop (CR), leaves (LV) roots (RT), storage organs (SO), shielded reserves (SR) and stems (ST) DATE(B)Julian date (at beginning of simulation) DELT CSMP time period for integration DLA daylength, astronomical DLP daylength effective for photoperiodism DR(R,V)development rate crop in vegetative and reproductive phase DRC(R,V)development rate constant in the vegetative (V) and reproductive (R) phase DRDT relation of DRED to daylength DRED effect of daylength in DRV DREW effect of water stress in DRV DRRTT relation of DRR to temperature DRSL water drained from deepest soil layer (equals WLFL4) DRVTT relation of DRV to temperature DRWT relation of DREW to level of water stress DS(I) phenologial development stage crop (initial) DSF1,2 DS when floret formation starts, ends (module TIL) DSFL variable with value 1.0 during floret formation, else 0.0 DSG1,2 DS when grain formation starts, ends (module TIL) DSGR variable with value 1.0 during grain formation, else 0.0 DST1,2 DS when tiller formation starts, ends (module TIL) DSTI variable with value 1.0 during tiller formation, else 0.0 DSLR number of days since last rain DTFX fixed timestep for SAWAH DTIME time in current day DTMIN minimum time period for integration in SAWAH DTMXl maximum time period for integration in SAWAH DWCLDT rate of change of soil water content (program SWD) DYNAM(IC) CSMP label DZ distance between compartment centres (program SWD) EES extinction coefficient for evaporation in soil ELV elevation of growth site above sea level END CSMP label ENDJOB CSMP label ENDPRO(CEDURE) CSMP label EVSC potential soil evaporation rate for current weather conditions and crop EVSD evaporation rate soil on dry days (i.e. almost no rain) EVSH evaporation rate soil on humid days EVSPD potential evaporation soil due to drying power air EVSPR potential evaporation soil due to radiation EVSW evaporation rate from the soil (actual value; e in Eq.12) EVSW1-3 EVSW for individual soil compartments FADL fraction to adapt time period to account for daylength FC(LV,RT,SO,ST) fraction weight carbon of total dry weight in leaves (LV), roots (RT), storage organs (SO) and stems (ST) FEPSO fraction economic product in storage organs (dry weights) FEVL1-3/FEVLT fraction of EVSW from soil compartments 1-3 F1,2(1-5) leaf area fraction in 0-30 and 30-60 degree leaf angle classes for layers 1 to 5

– cm 2 d-1 kg ha-l d-1

d h h d-l d-l

-

mm d-l

-

-

-

d d d d d-l m m-1 m

mm mm mm mm mm mm mm

d -l d-l d -l d-l d-l d-1 d-l

kg kg -1 kg kg -1

-

221

FIEC ratio of CO2I vs CO2E FINISH CSMP function FINTIM CSMP function (finish time simulation) FIXED CSMP function FLX water flux density (program SWD; q in Eq. 4) FRNOF fraction of precipitation that runs off field FSTR fraction stem weight at flowering that is remobilizable FUCCHK user defined function for carbon balance check FUNCTION CSMP or FORTRAN function FUPHOT user defined function for canopy photosynthesis FURSC user defined function for canopy resistance FUTP user defined function for temperature FUVP user defined function for vapour pressure FUWCHK user defined function for water balance check FUWRED user defined function for windspeed reduction FUWS user defined function for water stress

d cm d -l kg kg -1 -

GAR growth rate of available reserves (glucose) G(CR,LV,RT,SO,SR,ST) growth rate (dry matter) of the whole crop (CR), leaves (LV), roots (RT), storage organs (SO), shielded reserves (SR, starch) and stems (ST) GFP grain filling period GGRMN minimal growth rate of one grain GGRMX maximal growth rate of one grain GGRT relation of temperature to growth rate of grains GLA growth rate leaf area GN(FL,GR,TI) growth of number of florets, grains, tillers GSA growth rate photosynthetically active stem area GSOAV running average of GSO GSOAVM maximum value of GSOAV GSOM maximum growth rate storage organs GSORM maximum relative growth rate storage organs GSRP potential rate of GSR GZRT growth rate rooting depth GZRTC maximum value of GZRT

kg ha -ld-1

kg ha-1d-1 d kg d -1 kg d -1 ha ha-1d-1 ha-1d-1 ha ha-1d-1 kg ha-1d-1 kg ha-ld-1 kg ha-ld-1 kg kg-ld-1 kg ha-ld-1 m d -l m d -l

HI HUAAT

kg kg -1 kPa

harvest index (based on above ground dry matter) table of values of VPA during year

I index in DO-loops and dimensioned variables IDATE integer value of DATE INIT(IAL) CSMP label INSW CSMP function INTGRL CSMP function ITYL integer value soil Type number K KAV KMSA KMSAl KMSAlT KMSA2 KMSA2T KMSMX KMSMXT KSAT KST KSTT

222

d -

cm d -l hydraulic conductivity (program SWD; k in Eq.3) cm d -l average hydraulic conductivity of adjacent layers (SWD) cm-1 parameter in exponent of unsaturated conductivity (SWD) soil characteristic ( a in Eq.3) table of characteristic of soil types soil characteristic ( a in Eq.3) table of characteristic of soil types soil characteristic (lhl max in Eq.3) table of characteristic of soil types saturated hydraulic conductivity (program SWD;k s in Eq.3: I cmd -l cm d -l saturated hydraulic conductivity ( ks in Eq.3) table of characteristic of soil types

LAT LIMIT LLA LLV LLVT LNTI LRT LRTT LSR LSTR

latitude (south of equator negative values) CSMP funcion rate of loss of leaf area rate of loss of leaf weight (dry matter) relation of relative loss rate of leaves to DS loss of number of tillers rate of loss of root weight (dry matter) relation of relative loss rate due to aging to DS rate of loss of shielded reserves to WAR loss rate of stem reserves (starch)

MCLV MCRT MCSR

trigger for mobilization of carbohydrates from leaves trigger for mobilization of carbohydrates from roots trigger for mobilization of carbohydrates from shielded reserves CSMP label matric flux potential (program SWD; F in Eq.5) MFLP at bottom of the profile (program SWD) MFLP at top of the profile (program SWD) matric suction (program SWD; |h| in Eq.2) MS at the bottom of the profile (program SWD) MS at the top of the profile (program SWD) soil characteristic (gamma in Eq.2) table of characteristic of soil types

METHOD MFLP MFLPBT MFLPTP MS MSBT MSTP MSWCA MSWCAT

NFL(MX,P,MXT) number of florets (maximum, potential, maximum per tiller) (module TIL) NGR(MX,P) number of grains (maximum, potential) (module TIL) NIGHT variable to indicate day part: night (1) or day (0) NL number of soil compartments simulated in L2SS NOR CSMP and FORTRAN function NTI number of tillers, including number of main stems (NTII) NTII initial number of 'tillers', i.e. the number main stems NTIP potential number of tillers (limited by carbohydrates) OUTDEL CSMP function (output interval)

degree ha ha-l d-l kg ha-1 d -l ha -1 d -1 kg ha-l d -1 kg ha -l d -1 kg ha -l d -1

cm2 d -1 cm2 d -1 cm2 d -1 cm cm cm

ha ha

ha ha ha

-1 -1

-1 -1 -1

d

PAGE CSMP label PARAM(ETER) CSMP label PCEW effect of water stress on PCGC PCGC photosynthesis canopy, gross, in current weather and physiological state (level l), as CO 2 per daytime period kg ha -l d -1 PCGD PCGC expressed per 24 h (equal to PCGC for 1 d time steps)kg ha -l d -l PCGDV running average of PCGD kg ha -1 d -1 PCGT PCGC totaled since start of simulation kg ha -l PCGW photosynthesis canopy, gross, reduced by water shortage (level 2), as CO 2 kg ha -1 d -1 PCNSH net photosynthesis above ground part crop kg ha -l d -l PCNT net canopy photosynthesis totaled since start simulation kg ha -1 PLEA PLEI at actual temperature kg CO 2 ha -l h-l /(J m -2 s-1 ) PLEH direct effect air humidity on PLMX PLEI initial efficiency use absorbed light by individual leaves,as PLEA PLETT relation of PLEI to temperature PLHT plant height m PLHTT relation of PLHT to DS PLMHT relation of PLMXP to air humidity

223

relation of PWP to temperature maximum rate of photosynthesis of single leaves (CO 2) in current conditions kg PWXP PLMX for standard SLC and optimal conditions kg PLMXT PLMX adjusted for leaf thickness kg PLNA daytime average of leaf net photosynthesis per unit area kg PRDEL CSMP function (print interval) d PRINT CSMP function PROCED(URE) CSMP label PRTPLOT CSMP function PLMTT PLMX

Q10

ha -l h -1 ha -l h -1 ha -l h-l ha -l d -1

Ql0 of maintenance respiration sensitivity to temperature -

RAIN RAINT RCRT RDTC

precipitation table of daily precipitation values during a year respiration crop, totaled (for CO 2) radiation daily total global above atmosphere (400-1400 nm) RDTM radiation, daily total global, measured (400-1400 nm) RDTMT table of measured daily total global radiation during year RDUCF radiation units conversion factor RFS reflection coefficient soil for RDTM RFSD RFS for dry soil RG(CR,LV,RT,SO,SR,ST) respiration (in CO 2) due to growth of the whole crop (CR), leaves (LV), roots (RT), storage organs (SO), shielded reserves (SR) and stems (ST) RLLV respiration caused by remobilization from dying leaves RLSR respiration caused by remobilization (loss) of shielded reserves RMCLV standard coefficient for leaf maintenance respiration (CO2) RM(CR,LV,RT,SO,ST) maintenance respiration (CO 2) of whole crop (CR), leaves (LV), roots (RT), storage organs (SO), stems (ST) RMCT RMCR, totaled since initialization RMLV(D,N) RMLV in daytime (D) and nighttime (N) RMMA maintenance respiration due to metabolic activity RSB(L,S) boundary layer resistance for water vapour diffusion from average leaf (L) or soil (S) RSH respiration rate of shoot (growth plus maintenance resp) RSLL leaf resistance for water vapour diffusion in average leaf RSLLM minimum value of RSLL RST(L,S) resistance to diffusion for water vapour, CO 2 and heat due to turbulence in canopy from average leaf (L) or soil (S) RUNOF water flowing from surface to other fields specific leaf weight, actual value (eventually corrected for contribution stem area) SLC specific leaf weight constant SLN SLA for new leaves SLT relation of SLA to DS SQRT CSMP function SSC specific stem weight constant (SLC analogy) STOP CSMP label STORAGE CSMP label

mm d -l mm d -l kg ha -1

-

J m -2 d -1 J m -2 d-1

variable

kg ha -l d -1 kg ha -l d-l kg ha -l d -1 kg kg -l d -1

kg kg kg kg

ha -l d -1 ha -1 ha -l d-1 ha -l d -1

s m -1 kg ha -l d -1 s m -1 s m -1

s m -1 mm d -l

SLA

224

kg ha -1 kg ha -1 kg ha -1

kg ha -1

-

SUASTR user defined subroutine for astronomical variables SUEVTR user defined subroutine for evapotranspiration SUPHOL user defined subroutine for canopy photosynthesis SUSAWA user defined subroutine for soil water balance TABLE CSMP label TCD time constant for dying of tillers TCF time constant for formation of plant organs TERT effect of temperature on root growth rate TIMER CSMP label TITLE CSMP label TKL(I)1-3 thickness soil compartment I, 1-3 TKLT thickness of combined soil compartments TPA(A,D,V) actual air temperature at each DTIME (A), in daytime (D) and 24h average (V) TPEM temperature effect on maintenance respiration TPHT table of maximum day temperatures during a year TPLT table of minimum night temperatures during a year TPR reference temperature for maintenance respiration TPS(1) temperature of the soil (initial) TRC transpiration rate canopy, potential value for current weather and crop (level 1) TRCP(D,R) potential transpiration canopy due to drying power air (D) and absorbed radiation (R) TRRM potential transpiration rate per unit rooted length TRW transpiration rate canopy, actual value with water stress (level 2) TRWL1-3,(1-NL) TRW from individual compartments 1-3 or 1-NL TRWT TRW totaled since start of simulation TYL(1-NL) number indicating soil type of compartment VPA VPD VPDC

humidity of the air, early morning value vapour pressure difference vapour pressure deficit in canopy, daytime

average

d d

m m °C °C °C °C °C mm d-l mm d-l mm d-l m-1 mm d -l mm d -l

mm kPa kPa kPa

WAR(I) available carbohydrate (glucose) in leaves (initial) kg ha -1 WARR WAR relative to WLV WC(AD,FC,ST,WP) volumetric water content of soil when air dry (AD), at wilting point (WP), field capacity (FC) and saturation (ST, equals relative total pore space); these variables are indexed 1-NL in L2SS, and numbered 1-3 in m3 m -3 L2SU WCL(I)1-3 relative soil water content per layer in L2SS (initial) ( 0 in Eq.2) m3 m-3 m3 m-3d-l WCLCH(1-NL) rate of change of WCLQT WCLEQI(1-NL) initial value of WCLQT in equilibrium situation m3 m -3 WCLIS switch parameter for soil water initialization WCLISC switch water soil layers initial (see Section 5.4.4) WCLMQI(1-NL) initial value of WCLQT from observations m3 m -3 WCLQT(I)(l-NL) same as WCL(I)1-3 in module with impeded drainage m3 m -3 WCR weight crop, including roots kg ha -1 WCST volumetric water content at saturation ( 0 in Eq.2) m3 m -3 WCSTT table of water content at saturation for soil types WCUM(I)total water in soil profile (initial) mm WCUMCH rate of change of WCUM mm d -1 WDCL width of soil cloth (WDLV analogue) m WDLV width of leaves m WDS(AD,AV,S) wind speed, daytime average (AD), 24 h average (AV), and near the soil surface (S) m s-1

225

WDST table of daily values observed wind speeds during a year WEPSO weight economic part of storage organs WFSC flooding stress sensitivity coefficient WGR(MX)average weight of grains, filled plus unfilled (maximum) WIR weight increment reserves (starch) since start simulation WLA water available to the crop in a layer WLFL1-4 fluxes of water into layers 1-3 and out of layer 3 WL1-3(I) volumetric soil water content per compartment (initial) WL0QT(I) water standing above soil surface (initial) WL0CH rate of change of WL0QT WL0MX maximum level of water on the surface (bund height) WLV(I) weight leaves (initial) WLVD weight dead leaves WLVSO total above ground dry weight WLVST sum of WLV, WLVD and WST WLVT sum of WLV and WLVD WRT(I) weight roots (initial) WRTD weight dead roots WSE(1-3 ) effect of water stress on water uptake in layers 1-3 WSERT effect of water stress on root water uptake WSO(I) weight storage organs (initial) WSR(I) weight of shielded reserves (starch) in stem (initial) WSS weight shoot plus storage organs WSSC water stress sensitivity coefficient WST(I) weight stems (initial) minus WSR or WIR contained in it WSTR stem weight (WST+WSR or WST+WIR, depending upon module) WUPC water use efficiency, current, relative to net photosynthesis leaves (water transpired per kg CO 2 fixed, net, daytime) WUPT WUPC of total net photosynthesis and transpiration ZEQT(I)depth evaporation front in upper soil layer ( zE in Eq.12) ZECH rate of change of ZEQT ZL(1-NL) depth of upper boundary of each soil compartment ZLL depth upper boundary compartment (L2SS) ZREF reference height windspeed observations ZRT(I ) rooting depth (initial) ZRT1-3 ZRT differentiated per soil compartment ZRTL rooting depth in individual layers (L2SS) ZRTM maximum for ZRT ZRTM(C,S) maximum rooting depth for crop (C) and soil (S) ZW (I) depth of free water table (initial) ZWTB table with observed ZW versus time

226

m s -1 kg ha -1 kg kg ha -1 m3 ha -1 mm d -l m ha m m d -1 m kg ha -1 kg ha -1 kg ha -1 kg ha -1 kg ha -1 kg ha -1 kg ha -1 kg ha -1 kg ha -1 kg ha -1 kg ha -1 kg ha -1

kg kg -1 kg kg -1 m m d -l m m m m m m m m

8

References

Aggarwal, P.K. & Penning de Vries, F.W.T., 1989. Potential production of wheat in southeast Asia. Agricultural Systems 30(1):49-69. Alberda, Th., 1969. The effect of low temperature on dry matter production, chlorophyll concentration and photosynthesis of maize plants of different ages. Acta Botanica Neerlandica 18(1):39-49. Alberda, Th. et al., 1977. Crop photosynthesis: methods and compilation of data obtained with mobile field equipment. Agricultural Research Reports, 865. Pudoc, Wageningen. Alberda, Th., 1984. Production and water use of several food and fodder crops under irrigation in the desert of southwestern Peru. Agricultural Research Reports, 928. Pudoc, Wageningen. Amthor, J.S., 1984. The role of maintenance respiration in plant growth. Plant, cell and environment 7:561-569. Angus, J.F. & Zandstra, H.G., 1980. Climatic factors and the modeling of rice growth and yield. In: Agroclimatology of the rice crop. International Rice Research Institute, Los Baños, pp. 189-199. Angus, J.F., Mackenzie, D.H., Morton, R. & Schafer, C.A., 1981. Phasic development in field crops. II. Thermal and photoperiodic responses of spring wheat. Field Crops Research 4:269-283. Azcon-Bieto, J., Lambers, H. & Day, D.A., 1983. Effect of photosynthesis and carbohydrate status on respiratory rates and the involvement of the alternative pathway in leaf respiration. Plant Physiology 72:598-603. Bakema, A.H. & Jansen, D.M., 1987. Using a simulation model to evaluate weather effects. In: Impact of weather parameters on the growth and yield of rice. International Rice Research Institute, Los Baños, pp. 283-290. Baker, C.H. & Curry, R.B., 1976. Structure of agricultural simulations: a philosophic view. Agricultural Systems 1: 201-218. Bassett, D.M., Stockton, J.R. & Dickens, W.L., 1970. Root growth of cotton as measured by P32 uptake. Agronomy Journal 62:200-203. Basstanie, L.J.M. & Laar, H.H. van, 1982. Introduction of CSMP by an elementary simulation program. In: F.W.T. Penning de Vries and H.H. van Laar (Eds.): Simulation of plant growth and crop production. Simulation Monographs. Pudoc, Wageningen, pp. 50-65. Bell, C.J., 1982. A model of stomatal control. Photosynthetica 16(4):486-495. Benschop, M., 1986. TUCROS, een simulatiemodel voor de tulpecultivar “Apeldoorn”. Simulation Report CABO-TT No. 6. Centre for Agrobiological Research, Wageningen. Berge, H.F.M. ten, 1986. Heat and water transfer at the bare soil surface. 227

Aspects affecting thermal imagery. Ph.D. Thesis, Agricultural University, Wageningen. Berge, H.F.M. ten, Metselaar, K. & Stroosnijder, L., 1987. Measurement of matric flux potential: a simple procedure for the hydraulic characterization of soils. Netherlands Journal of Agricultural Science, 35:371-384. Berge, H.F.M. ten & Jansen, D.M., 1989. Water balance module SAWAH: users guide and outline. Simulation Report CABO-TT. Centre for Agrobiological Research, Wageningen (forthcoming). Berry, J.A. & Downton, W.J.S., 1982. Environmental regulation of photosynthesis. In: Govindjee (Ed.): Photosynthesis, Vol II. Academic Press, New York, pp. 263-343. Beuerlein, J.E. & Pendleton, J.W., 1971. Photosynthetic rates and light saturation curves of individual soybean leaves under field conditions. Crop Science 11:217-219. Bhagsari, A.S. & Brown, R.H., 1976. Photosynthesis in peanut (Arachis) genotypes. Journal Peanut Science 3:1-5. Bhagsari, A.S., Brown, R.H. & Schepers, J.S., 1976. Effect of moisture stress on photosynthesis and some related physiological characteristics in peanut. Crop Science 16: 712-715. Bhagsari, A.S. & Brown, R.H., 1976. Relationship of net photosynthesis to carbon dioxide concentration and leaf characteristics in selected peanut (Arachis) genotypes. Journal Peanut Science 3:10-14. Bierhuizen, J.F. & Slatyer, R.O., 1964. Photosynthesis of cotton leaves under a range of environmental conditions in relation to internal and external diffusive resistances. Australian Journal of Biological Sciences 17:348-359. Bidwell, R.G.S., 1983. Carbon nutrition of plants: photosynthesis and respiration. In: Steward, F.C. & Bidwell, R.G.S. (Eds.), 1983. Plant Physiology, Vol VII: Energy and Carbon Metabolism. Academic Press, New York, pp. 287-458. Bjorkman, O., 1981. Responses to different quantum flux densities. Encyclopedia Plant Physiology, New Series, Vol 12a. Springer Verlag, Berlin, pp. 57-107. Bodlaender, K.B.A., 1986. Effects of drought on water use, photosynthesis and transpiration of potatoes. 1. Drought resistance and water use. In: Potato research of tomorrow. Pudoc, Wageningen, pp. 36-43. Bouma, J., 1985. Morphology of lowland soils and flow of water and gases. In: Proceedings of the International Workshop on Physical Aspects of Soil Management in Rice-based Cropping Systems (PASMIRCS), International Rice Research Institute, Los Baños, 1985, pp. 199-215. Breman, H., Cisse, A.M., Djiteye, M.A. & Elberse, W.Th., 1982. Le potentiel botanique des paturages. In: F.W.T. Penning de Vries & M.A. Djiteye (Eds.): La productivite des paturages saheliens. Agricultural Research Reports 918. Pudoc, Wageningen. Brockington, N.R., 1979. Computer modelling in Agriculture. Oxford Uni228

versity Press, Oxford. Brown, K.F. & Biscoe, P.V., 1985. Fibrous root growth and water use of sugar beet. Journal of Agricultural Science, 105:679-691. Brunt, D., 1932. Notes on radiation in the atmosphere. I. Quarterly Journal of the Royal Meteorological Society 58:389-420. Bull, T.A., 1969. Photosynthetic efficiencies and photorespiration in Calvin cycle and C4 dicarboxylic acid plants. Crop Science, 9:726-729. Campbell, G.S., 1977. An introduction to environmental biophysics. Springer Verlag, New York. Charles-Edwards, D.A., 1981. The mathematics of photosynthesis and productivity. Academic Press, London. Cock, J.H. & Yoshida, S., 1973. Changing sink and source relations in rice (Oryza sativa L.) using carbon dioxide enrichment in the field. Soil Science and Plant Nutrition 19(3):229-234. Cock, J.H., Franklin, D., Sandoval, G., & Juri, P., 1979. The ideal cassava plant for maximum yield. Crop Science 19:271-279. Cook, M.G. & Evans, L.T. 1983. Some physiological aspects of the domestication and improvement of rice (Oryza spp.) Field Crops Research 6:219238. Constable, G.A. & Rawson, H.M., 1980. Effect of leaf position, expansion and age on photosynthesis, transpiration and water use efficiency of cotton. Australian Journal of Plant Physiology 7:89-100. Dagley, S. & Nicholson, D.E., 1970. An introduction to metabolic pathways. Blackwell Scientific Publications, Oxford. Dantuma, G., 1973. Rates of photosynthesis in leaves of wheat and barley varieties. Netherlands Journal of Agricultural Science 21:181-187. Davidoff, B. & Selim, H.M., 1986. Goodness of fit for eight water infiltration models. Soil Science Society of America Journal 50:759-764. Day, W., Legg, B.J., French, B.K., Johnston, A.E., Lawlor, D.W. & Jeffers, W. de C., 1978. A drough experiment using mobile shelters: the effect of drought on barley yield, water use and nutrient uptake. Journal Agricultural Science, Cambridge 91:599-623. Dayan, E., Keulen, H. van & Dovrat, A., 1981. Tiller dynamics and growth of Rhodes grass after defoliation: a model named TILDYN. Agro-ecosystems 7:101-112. Dent , J .B. & Blackie, M. J . ,1979. Systems Simulation in Agriculture. Applied Science Publishers, London. Dingkuhn, M., 1985. Physiologische Faktoren der Trockenheitsresistenz bei Reis. Ph.D. Thesis, University of Hamburg, West Germany. Doorenbos, J., Kassam, A.H., Bentvelder, C. & Uittenboogaard, G., 1978. Yield response to water. U.N. Economic Commission to West Asia, Rome. Doorenbos, J. & Kassam, A.H., 1979. Yield response to water. FAO irrigation and drainage paper, No.33. FAO, Rome. Dornhoff, G.M. & Shibles, R.M., 1970. Varietal differences in net photosyn229

thesis of soybean leaves. Crop Science 10:42-45. Driessen, P.M., 1986a. The waterbalance of soil. In: H. van Keulen & J. Wolf (Eds.): Modelling of agricultural production: weather, soils and crops. Simulation Monographs. Pudoc, Wageningen, pp. 76-116. Driessen, P.M., 1986b. Soil physical data. In: H. van Keulen & J. Wolf (Eds.): Modelling of agricultural production: weather, soils and crops. Simulation Monographs. Pudoc, Wageningen, pp. 212-223. Eastin, J.D., 1983. Sorghum. In: Potential productivity of field crops under different environments. International Rice Research Institute, Los Baños, pp.181-204. Elderen, E. van, 1987. Scheduling farm operations, a simulation model. Simulation Monographs. Pudoc, Wageningen. Ehleringer, J., 1976. Leaf absorptance and photosynthesis as affected by pubescence in the genus Encelia. Annual Report Director Department Plant Biology, Carnegie Institution, pp. 413-418. Ehleringer, J. & Bjorkman, O., 1976. Carbon dioxide and temperature dependence of the quantum yield for CO2 uptake in C3 and C4 plants. Annual Report of the Director Department of Plant Biology, Carnegie Institution, pp. 418-421. El-Sharkawy, M.A. & Hesketh, J.D., 1964a. Effect of temperature and H2O deficit on leaf photosynthetic rate of different species. Crop Science 4:514518. El-Sharkawy, M.A. & Hesketh, J.D., 1964b. Effect of stomatal differences among species on leaf photosynthesis. Crop Science 4:619-621. El-Sharkawy, M.A., Hesketh J.D. & Muramoto, H. 1965. Leaf photosynthesis rates and other growth characteristics among 26 species of gossypium. Crop Science 5:173-175. El-Sharkawy, M.A., Cock, J.H. & de Cadena, G., 1984a. Influence of differences in leaf anatomy on net photosynthetic rates of some cultivars of cassava. Photosynthesis Research 5:235-242. El-Sharkawy, M.A., Cock, J.H. & Held, K.A.A., 1984b. Water use efficiency of cassava II. Differing sensitivity of stomata to air humidity in cassava and other warm-climate species. Crop Science 24:503-507. Encyclopedia of Plant Physiology, New Series, Vol. 1-19, 1975-1986. Springer Verlag, Heidelberg. Eswaran, H., 1985. Interpreting physical aspects of wetland soil management from soil taxonomy. In: Proceedings of the International Workshop on Physical Aspects of Soil Management in Rice-based Cropping Systems (PASMIRCS). International Rice Research Institute, Los Baños, pp. 1730. Evans, L.T. & Wardlaw, I.F. , 1976. Aspects of the comparative physiology of grain yield in cereals. Advances in Agronomy 28:301-359. Farquhar, G.D. & von Caemmerer, S., 1982. Modelling of photosynthetic response to environmental conditions. Encyclopedia Plant Physiology , New 230

Series 12b. Springer Verlag, Berlin, pp. 549-587. Feddes, R.A., Kowalik, P.J., & Zaradny, H., 1978. Simulation of field water use and crop yield. Simulation Monographs. Pudoc, Wageningen. Forrester, J.W., 1961. Industrial dynamics. MIT Press, Boston. Forward, D.F., 1983. Respiration, a holistic approach to metabolism. In: Steward & Bidwell, pp. 458-510. Frère, M., 1987. The FAO agroclimatological data base: its use in studying rice-weather relationships. In: The impact of weather parameters on the growth and yield of rice. International Rice Research Institute, Los Baños, pp. 41-46. Fukai, S., Kuroda, E. & Yamagishi, T. , 1985. Leaf gas exchange of upland and lowland rice cultivars. Photosynthesis Research 7:127-135. Gardner, W.R., 1958. Some steady state solutions of the unsaturated moisture flow equation with application to evaporation from a water table. Soil Science 85:228-232. Gates, D.M., Strain, B.R. &Weber, J.A., 1983. Ecophysiological effects of changing atmospheric CO2 concentration. Encyclopedia Plant Physiology, New Series, Vol 12D. Springer Verlag, Berlin, pp. 503-526. GELGAM, 1984. Herziening van de berekening van de gewasverdamping in het hydrologisch model GELGAM. Rapport van de ad hoc groep Verdamping. Dienst Waterbeheer, Provincie Gelderland, Arnhem. Geng S., Penning de Vries, F.W.T. & Supit, I. 1985a. Analysis and Simulation of weather variables: rain and wind in Wageningen. Simulation Reports CABO-TT No 4. Centre for Agrobiological Research, Wageningen. Geng S., Penning de Vries, F.W.T. & Supit, I., 1985b. Analysis and Simulation of weather variables: temperature and solar radiation. Simulation Reports CABO-TT No 5. Centre for Agrobiological Research, Wageningen . Geng S., Penning de Vries, F.W.T. & Supit, I., 1986. A simple method for generating daily rainfall data. Agricultural and Forest Meteorology 36:363376. Gifford, R.M. & Evans, L.T., 1981. Photosynthesis, carbon partitioning and yield. Annual Review Plant Physiology 32:485-509. Gifford, R.M. & Jenkins, C.L.D., 1982. Prospects of applying knowledge of photosynthesis toward improving crop production. In: Govindjee (Ed.): Photosynthesis, Vol II. Academic Press, London, pp. 419-457. Giraldez, J.V. & Sposito, G., 1985. Infiltration in swelling soils. Water Resources Research 21:33-44. Goudriaan, J., 1977. Crop micrometeorology: a simulation study. Simulation Monographs. Pudoc, Wageningen. Goudriaan, J., 1982a. Some techniques in dynamic simulation. In: F.W.T. Penning de Vries & H.H. van Laar (Eds.): Simulation of plant growth and crop production. Simulation Monographs. Pudoc, Wageningen, pp. 98-113. Goudriaan, J., 1982b. Potential production processes. In: F. W.T. Penning de 231

Vries and H.H. van Laar (Eds.): Simulation of plant growth and crop production. Simulation Monographs. Pudoc, Wageningen, pp. 98-113. Goudriaan, J., 1986. A simple and fast numerical method for the computation of daily totals of crop photosynthesis. Agricultural and Forest Meteorology 38:249-254. Goudriaan, J., 1987. The biosphere as a driving force in the global carbon cycle. Netherlands Journal of Agricultural Science 35:177-187. Goudriaan, J., 1988. The bare bones of leaf angle distribution in radiation models for canopy photosynthesis and radiation exchange. Agricultural and Forest Meteorology 43:155-169. Goudriaan, J., 1989. Modelling of ageing, development, delays and dispersion. In: Rabbinge et al. (Eds.): Simulation and systems management in crop protection. Simulation Monographs, Pudoc, Wageningen. Goudriaan, J. & Laar, H.H. van, 1978a. Calculation of daily totals of the gross CO2 assimilation of leaf canopies. Netherlands Journal of Agricultural Science 26:373-382. Goudriaan, J. & Laar, H.H. van, 1978b. Relations between leaf resistance, CO2-concentrations and CO2-assimilation in maize, beans, lalang grass and sunflower. Photosynthetica 12(3):241-249. Goudriaan, J. & Ruiter, H.E. de, 1983. Plant growth in response to CO2 enrichment, at two levels of nitrogen and phosphorus supply. 1. Dry matter, leaf area and development. Netherlands Journal of Agricultural Science 31: 157-169. Goudriaan, J., Laar, H.H. van, Keulen, H. van & Louwerse, W., 1984. Simulation of the effect of increased atmospheric CO2 on assimilation and transpiration of a closed crop canopy. Wissenschafliche Zeitschrift Humbolt Universitaet Berlin, Math.-Nat. R. 33(4):352-356. Graham, D. & Chapman, E.A., 1979. Interactions between photosynthesis and respiration in higher plants. Encyclopedia Plant Physiology 6:150-162. Grashoff, C., Klein Hulze, J.A., & Smid, G.H., 1987. Opbrengstinstabiliteit en oogstzekerheid bij veldbonen en erwten. CABO Reports. Centre for Agrobiological Research, Wageningen. Gregory, P.J., McGowan, M., Biscoe, P.V. & Hunter, B., 1978. Water relations of winter wheat. 1. The growth of the root system. Journal of Agricultural Science, Cambridge 91:91-102. Gregory, P.J. & Reddy, M.S., 1982. Root growth in an intercrop of pearl millet/groundnut. Field Crops Research 5:241-252. Gulmon, S.L. & Chu, C.C., 1981. The effects of light and nitrogen on photosynthesis, leaf characteristics and dry matter allocation in the chaparral shrub Diplacus aurantiacus. Oecologia 49:207-212. Gupta, S.C., Larson, W.E. & Allmaras, R.R., 1984. Predicting soil temperature and soil heat flux under different tillage-surface residue conditions. Soil Science Society of America Journal 48(2):223-232. Gupta, P.C. & O‘Toole, J.C., 1986. Upland rice, a global perspective. In232

ternational Rice Research Institute, Los Baños. Hahn, S.K. & Hozyo, Y., 1983. Sweet potato and yam. In: Potential productivity of field crops under different environments. International Rice Research Institute, Los Baños, pp. 319-340. Halevy, A.H., 1985. Handbook of Flowering, Volumes I to V. CRC Press Inc, Boca Raton, Florida. Hall, A.E., 1982. Mathematical models of plant water loss and plant water relations. In: O.L. Lange et al. (Eds.): Encyclopedia of Plant Physiology Vol.12B. Springer Verlag, Berlin, pp. 231-261. Hall, A.E., & Loomis, R.S. 1972. Photosynthesis and respiration by healthy and beet yellow virus-infected sugar beets (Beta vulgaris L.). Crop Science 12:566-572. Hansen, S. & Aslyng, H.C., 1984. Nitrogen balance in crop production, simulation model NITCROS. Hydrotechnical Laboratory, Royal Veterinary and Agricultural University, Copenhagen. Hanway, J.J. & Weber, C.R., 1971. N, P and K percentages in soybean (Glycine max (L.) Merrill) plant parts. Agronomy Journal 63:286-290. Haverman, P.P., 1986. A characterization of the growth of the cowpea crop in the Sahelian zone. MSc thesis Department of Theoretical Production Ecology, Agricultural University, Wageningen. Heemst, H.D.J. van, 1986. The distribution of dry matter during growth of a potato crop. Potato Research 29:55-66. Hesketh, J.D. & Jones, J.W. (Eds.): 1980. Predicting photosynthesis for ecosystem models, I, II. CRC Press, Boca Raton, Florida 33431. Hodges, T., Kanemasu, E.T. & Teare, I.D., 1979. Modeling dry matter accumulation and yield of grain sorghum. Canadian Journal of Plant Science 59:803-818. Hofstra, G. & Hesketh, J.D., 1969. Effects of temperature on the gas exchange of leaves in the light and dark. Planta (Berl.) 85:228-237. Hoogmoed, W.B. & Bouma, J., 1980. A simulation model for predicting infiltration into cracked clay soils. Soil Science Society of America Journal 44:458-461. Horie, T., 1987. Dynamic models for evaluating and predicting the growth and yield of rice and sunflower crops from weather. Food and Fertilizer Technology Center, Technical Bulletin 99. Taipei, Taiwan. Horie, T., Wit, C.T. de, Goudriaan, J. & Bensink, J., 1979. A formal template for the development of cucumber in its vegetative stage. I, II and III. Proceedings Nederlandse Academie Wetenschappen Serie. C. 82(4):433-479. Horie, T., Nakagawa, H. & Kira, T., 1986. Studies on modelling and prediction of rice development processes. Japanese Journal of Crop Science 55(1):214-215. Hozyo, Y., Shimotsubo, K., & Kato, S. 1982. The interrelationship between photosynthetic activity and tuberous root growth in sweet potato grafts. 5th International Symposium on Tropical Root and Tuber Crops. International 233

Rice Research Institute, Los Baños, pp. 105-113. Huda, A.K.S., Sivakumar, M.V.K., Virmani, S.M., Seetharama, N., Singh Sardar & Sekaran, J.G., 1984. Modeling the effect of environmental factors on sorghum growth and development. In: Virmani (Ed.): Agrometeorology of sorghum and millet in the semi-arid tropics. International Crop Research Institute for the Semiarid Tropics, Patancheru, India, pp. 277-287. Huffaker, R.C. & Miller, B.L., 1978. Reutilization of Ribulose Biphosphate Carboxylase. In: Siegelman and Hind (Eds.): Photosynthetic carbon assimilation, Plenum Publishing Corporation, pp. 139-152. Huke, R.E., 1985. FARIDPUR: a computer-assisted instruction model for rainfed lowland rice. IRRI Research Paper Series 104, International Rice Research Institute, Los Baños, 14 pp. IBM, 1975. Continuous system modeling program III. General system information manual (GH19-7000) and users manual (SH19-7001-2). IBM Data Processing Divison, White Plains, New York. IBM/Microsoft, 1984. Fortran Compiler Version 2, IBM Personal Computer, Computer Language Series. Portsmouth, England. ICAR (Indian Council for Agricultural Research), 1977. Crop planning in drought and flood prone areas. New Delhi. Idso, S.B., Reginato, R. J., Jackson, R.D. & Pinter, P.J., 1981. Foliage and air temperatures: evidence for a dynamic ‘equivalence point’. Agricultural Meteorology 24:223-226. IIASA, 1980. Beware of pitfalls; a short guide to avoiding common errors in systems analysis. Executive Report, International Institute for Application of Systems Analysis (IIASA). IIASA, A-2361, Laxemburg, Austria. Ingram, K.T. & McCloud, D.E., 1984. Simulation of potato crop growth and development. Crop Science 24:21-27. IRRI, 1977. Annual Report for 1976. The International Rice Research Institute, Los Baiios. IRRI, 1984. Terminology for rice growing environments. International Rice Research Institute, Los Baños. IRRI, 1985. Soil physics and rice. In: Proceedings of the International Workshop on Physical Aspects of Soil Management in Rice-based Cropping Systems (PASMIRCS). International Rice Research Institute, Los Baiios. Jackson, M.B. &Drew, M.C., 1984. Effects of flooding on growth and metabolism of herbaceous plants. In: T.T. Kozlowski (Ed.): Flooding and plant growth. Academic Press, London, pp. 47-128. Jackson, B.S., Arkin, G.F. & Douglas, W.A., 1983. Anevaluation of selected time steps in a crop growth simulation model. Extended Abstracts, 16th Conference of Agriculture and Forest Meteorology. American Meteorological Society, Boston, Mass., pp. 110-112. Jansen, D.M. & Gosseye, P., 1986. Simulation of growth of millet (Penniseturn americanum) as influenced by water stress. Simulation Reports CABOTT No. 10. Centre for Agrobiological Research, Wageningen. 234

Jarvis, P.G., 1981. Stomatal conductance, gaseous exchange and transpiration. In: Plants and their atmospheric environment. Blackwell Scientific Publications, Oxford, pp. 175-214. Johnson, I.R. & Thornley, J.H.M., 1983. Vegetative crop growth model incorporating leaf area expansion and senescence, and applied to grass. Plant, Cell and Environment 6:721-729. Jolliffe, P.A. & Tregunna, E.B., 1968. Effect of temperature, CO2 concentration, and light intensity on oxygen inhibition of photosynthesis in wheat leaves. Plant Physiology 43:902-906. Jones, L.H., 1971. Adaptive responses to temperature in dwarf french beans, Phaseolus vulgaris. Annals of Botany 35:581-596. Kaigama, B.K., Teare, I.D., Stone, L.R. & Powers, W.L., 1977. Root and top growth of irrigated and nonirrigated grain sorghum. Crop Science, 17:555-559. Keulen, H. van, 1975. Simulation of water use and herbage growth in arid regions. Simulation Monographs. Pudoc, Wageningen. Keulen, H. van, 1976. A calculation method for potential rice production. Contributions of the Central Research Institute for Agriculture No. 21. Bogor, Indonesia, 26 pp. Keulen, H. van, 1977. Simulation of water use and herbage growth in arid regions. Simulation Monographs. Pudoc, Wageningen. Keulen, H. van, 1982. Crop production under semi-arid conditions, as determined by nitrogen and moisture availability. In: F. W.T. Penning de Vries & H.H. van Laar, (Eds.), Simulation of plant growth and crop production. Simulation Monographs. Pudoc, Wageningen, pp. 234-249. Keulen, H. van & De Milliano, W.A.J., 1984. Potential wheat yields in Zambia: a simulation approach. Agricultural Systems 14:171-192. Keulen, H. van, Laar, H.H. van, Louwerse, W. & Goudriaan, J., 1980. Physiological aspects of increased CO2 concentration. Experientia 36:786-792. Keulen, H. van, Penning de Vries, F.W.T. & Drees, E.M., 1982. A summary model for crop growth. In: F.W.T. Penning de Vries and H.H. van Laar (Eds.): Simulation of plant growth and crop production. Simulation Monographs. Pudoc, Wageningen, pp. 87-97. Keulen, H. van & Heemst, H.D.J. van, 1986. Meteorological data. In: H. van Keulen & J. Wolf (Eds.): Modelling of agricultural production: weather, soils and crops. Simulation Monographs. Pudoc, Wageningen, pp. 208-211. Keulen, H. van & Wolf, J., 1986 (Eds.): Modelling of agricultural production: weather, soils and crops. Simulation Monographs. Pudoc, Wageningen, 464 pp. Keulen, H. van, & Seligman, N.G., 1987. Simulation of water use, nitrogen nutrition and growth of a spring wheat crop. Simulation Monographs. Pudoc, Wageningen. Khan, M.A. & Tsunoda, S., 1970. Evolutionary trends in leaf photosynthesis and related leaf characteristics among cultivated wheat species and its wild 235

relatives. Japanese Journal Plant Breeding 20(3):133-140. Kingwell, R.S. & Pannell, D.J., 1987. A bioeconomic model of dryland farming. Simulation Monographs. Pudoc, Wageningen. Klute, A., 1952. A numerical method for solving the flow equation for water in unsaturated materials. Soil Science 73: 105-117. Koerner, Ch. & May, R., 1979. Stomatal behaviour in alpine plant communities between 600 and 2600 m above sea level. In: J. Grace, E.D. Ford & P.G. Jarris (Eds.): Plants and their atmospheric environment. Blackwell Scientific Publisher, Oxford, U.K., pp. 205-218. Koller, D. & Hadas, A., 1982. Environmental aspects of the germination of seeds. In: O.L. Lange et al. (Eds.): Encyclopedia of Plant Physiology Vol. 12B: 401-431. Springer Verlag, Berlin. Koorevaar, P., Menelik, G. & Dirksen, C., 1983. Elements of Soil Physics. Developments in Soil Science 13. Elsevier Scientific Publisher, Amsterdam, 228 pp . Kropff, M.J., 1987. Physiological effects of sulphur dioxide. I. The effect of SO2 on photosynthesis and stomatal regulation of Vicia Faba. Plant, Cell and Environment 10:753-760. Kwon, Y.W., 1984. Fundamental physiological constraints in breeding for higher yielding cold tolerant rice variety. International Crop Science Symposium on Potential Productivity and Yield Constraints of Rice in East Asia. Kyushu University, Fukuoka, Japan. Lambers, J.Th., 1979. Energy metabolism in higher plants in different environments. Ph.D. thesis, University of Groningen. Lambert, J.R., Baker, D.N. & Phene, C.J., 1976. Dynamic simulation of processes in the soil under growing row crops: RHIZOS. In: Computers applied to management of large scale agricultural enterprises. Proceedings US-USSR seminar, Moscow Riga Kishinev. Landivar, J.A., 1979. The application of cotton simulation model GOSSYM in genetic feasibility studies. MSc thesis, Mississippi State University, Mississippi. Lantinga, E.A., 1985. Productivity of grasslands under continuous and rotational grazing. Ph.D. thesis, Agricultural University, Wageningen. Leffelaar, P.A. & Ferrari, T.J., 1989. Some elements of dynamic simulation. In: Rabbinge et al.(Eds.): Simulation and systems management in crop protection. Simulation Monographs. Pudoc, Wageningen. Lin, Z.F. & Ehleringer, J., 1982. Changes in spectral properties of leaves as related to chlorophyll content and age of Papaya. Photosynthetica 16(4) :520-525. Linneman, H., DeHoogh, J., Keyzer, M.A. & Heemst, H.D.J. van, 1979. MOIRA, model of international relations in agriculture. North Holland Publishing Company, Amsterdam. Lof, H., 1976. Water use efficiency and competition between arid zone annuals, especially the grasses Phalaris minor and Hordeum murinum. Agri236

cultural Research Reports 853. Pudoc, Wageningen. Loomis, R.S., Rabbinge, R. & Ng, E., 1979. Explanatory models in crop physiology. Annual Review Plant Physiology 30: 339-367. Louwerse, N. & Zweerde, W. van de, 1977. Photosynthesis transpiration and leaf morphology of Phaseolus vulgaris and Zea mays grown at different irradiances in artificial and sunlight. Photosynthetica 11(1): 11-21. Louwerse, W., 1980. Effects of CO2 concentration and irradiance on the stomatal behaviour of maize, barley and sunflower plants in the field. Plant, cell and environment 3:391-398. Louwerse, W. & Eikhoudt, J.W., 1975. A mobile laboratory for measuring photosynthesis, respiration and transpiration of field crops. Photosynthetica 9:31-34 Louwerse, W. & Oorschot, J.L.P. van, 1969. An assembly for routine measurements of photosynthesis, respiration and transpiration of intact plants under controlled conditions. Photosynthetica 3:305-315. Ludwig, L.J., Saeki, T. & Evans, L.T., 1965. Photosynthesis in artificial communities of cotton plants in relation to leaf area. Australian Journal of Biological Sciences 18:1103-1118. MacKerron, D.K.L., 1984. Potato spout emergence in soil as a function of temperature and time of planting. In: F.A. Winiger, and A. Stockli (Eds): Conference papers 9th Biennial Conference of the European Association for Potato Research. Interlaken, Switzerland, pp. 364-365. Makino, A., Mae, T. & Ohira, K., 1984. Changes in photosynthetic capacity in rice leaves from emergence through senescence. Analysis from Ribulose –1,5–biphosphate carboxylase and leaf conductance. Plant and Cell Physiology 25(3):511-521. Marletto, V. & Keulen, H. van, 1984. Winter wheat experiments in the Netherlands and Italy analyzed by the SUCROS model. Simulation Reports CABO-TT No. 3. Centre for Agrobiological Research, Wageningen. McCree, K.J., 1974. Equations for the rate of dark respiration of white clover and grain sorghum, as functions of dry weight, photosynthetic rate and temperature. Crop Science 14:509-514. McCree, K.J., 1982. Photosynthetically active radiation. Encyclopedia Plant Physiology 12a. Springer Verlag, Berlin, pp. 41-55. McCree, K.J., 1983. Carbon balance as a function of plant size in sorghum plants. Crop Science 23: 1173-1177. McCree, K.J. & Kresovich, S., 1978. Growth and maintenance requirements of white clover as a function of daylength. Crop Science 18:22-25. McDermitt, D.K. & Loomis, R.S., 1981. Elemental composition of biomass and its relation to energy content, growth efficiency, and growth yield. Annals of Botany 48:275-290. McKeague, J.A., Eilers, R.G., Thomasson, A.J., Reeve, M.J., Bouma, J., Grossman, R.B., Favrot, J.C., Renger, M. & Strebel, O., 1984. Tentative assessment of soil survey approaches to the characterization and interprets237

tion of air-water properties of soils. Geoderma 34:69-100. McKinion, J.M., 1980. Dynamic simulation: a positive feedback mechanism for experimental research in the biological sciences. Agricultural Systems 5:239-250. Menenti, M., 1984. Physical aspects and determination of evaporation in deserts applying remote sensing techniques. Ph.D. thesis, Institute for Land and Water Management, Wageningen. Miedema, P., 1982. The effects of low temperature on Zea mays. Advances in Agronomy 35:93-128. Minchin, F.R., Summerfield, R.J., Hadley, P., Roberts, E.H. & Rawsthorne, S., 1981. Carbon and nitrogen nutrition of nodulated roots of grain legumes. Plant, Cell and Environment 4:5-26. Monteith, J., 1973. Principles of environmental physics. Edward Arnold, London. Moorby, J., 1981. Transport systems in plants. Longman, London and New York. Moorman, F.R. & Breemen, N. van, 1978. Rice: soil, water, land. International Rice Research Institute, Los Baños, 185 pp. Morris, R.A., 1987. Characterizing rainfall distributions at IRRI cropping systems research sites in the Philipines. In: Impact of weather parameters on the growth and yield of rice. International Rice Research Institute, Los Baños, pp. 189-214. Muramoto, H., Hesketh, J.D. & El-Sharkawy, M.A., 1965. Relationships among the rate of leaf area development, photosynthesis rate and rate of dry matter production among American cultivated cottons and other species. Crop Science 5:163-166. Nevins, D.J. & Loomis, R.S., 1970. A method for determining net photosynthesis and transpiration of plant leaves. Crop Science 10:3-6. Ng, N. & Loomis, R.S., 1984. Simulation of growth and yield of the potato crop. Simulation Monographs. Pudoc, Wageningen. Oldeman, L.R. & Frère, M., 1982. A study of the agroclimatology of the humid tropics of Southeast Asia. FAO, Rome. Oldeman, L.R., Seshu, D.V. & Cady, F.B., 1987. Response of rice to weather variables. In: Impact of weather parameters on the growth and yield of rice. International Rice Research Institute, Los Baños, pp. 5-40. O’Toole, J.C. & Cruz, R.T., 1980. Response of leaf water potential, stomatal resistance and leaf rolling to water stress. Plant Physiol. 65: 428-432. Pallas, J.E. & Samish, Y.B., 1984. Photosynthetic response of peanut. Crop Science 14:478-482. Parlange, J.-Y., 1971. Theory of water movement in soils: 2. One-dimensional infiltration. Soil Science 111:170-174. Passioura, J.B., 1977. Grain yield, harvest index, and water use of wheat. The Journal of the Australian Institute of Agricultural Science 43 (3-4):117-120. Penman, H.L., 1956. Evaporation: an introductory survey. Netherlands Jour238

nal of Agricultural Science 4:9-29. Penning de Vries, F.W.T., 1975. The cost of maintenance processes in plant cells. Annals of Botany 39:77-92. Penning de Vries, F.W.T., 1977. Evaluation of simulation models in agriculture and biology: conclusions of a workshop. Agricultural Systems 2:99-107. Penning de Vries, F.W.T., 1982. Le potentiel physiologique des pâturages et des cultures agricoles; La production potentielle des pâturages naturelles. In: F.W.T. Penning de Vries and M.A. Djiteye (Eds.): La productivité des pâturages saheliens. Agricultural Research Reports 918. Pudoc, Wageningen, pp. 87-98, 165-181. Penning de Vries, F.W.T., 1983. Modeling of growth and production. Encyclopedia of Plant Physiology, New Series, Vol. 12D. Springer Verlag, Berlin, pp. 117-150. Penning de Vries, F.W.T., 1987. Some opportunities for using crop models in rice. In: Impact of weather parameters on the growth and yield of rice. International Rice Research Institute, Los Baños, pp. 291-300. Penning de Vries, F.W.T., Brunsting, A.B. & Laar, H.H. van, 1974. Products, requirements and efficiency of biological synthesis, a quantitative approach. Journal Theoretical Biology 45:339-377. Penning de Vries, F.W.T. & Laar, H.H. van, 1976. Substrate utilization in germinating seeds. In: Landsberg, J.J. & C.V. Cutting (Eds.): Environmental effects on crop physiology. Academic Press, London, New York. Penning de Vries, F.W.T., Witlage, J.M. & Kremer, D., 1979. Rates of respiration and of increase in structural dry matter in young wheat, ryegrass and maize plants in relation to temperature, to water stress and to their sugar content. Annals of Botany 44:595-609. Penning de Vries, F.W.T. & Laar, H.H. van, (Eds.) 1982: Simulation of plant growth and crop production. Simulation Monographs. Pudoc, Wageningen. Penning de Vries, F.W.T., Laar, H.H. van & Chardon, M.C.M., 1983. Bioenergetics of growth of seeds, fruits and storage organs. In: I. Yoshida (Ed.): Potential productivity of field crops under different environments. International Rice Research Institute, Los Baños, pp. 38-59. Penning de Vries, F. W.T., Rabbinge, R., Jansen, D.M. & Bakema, A., 1988. Transfer of systems analysis and simulation in agriculture to developing countries. Agricultural Administration and Extension 29: 85-96. Press, W. et al., 1986. The art of scientific computing. ISBN O-521-30811-9. In: Numerical recipes. Cambridge University Press, U.S.A. 700 pp. Raats, P.A.C., 1970. Steady infiltration from line sources and furrows. Soil Science Society of America: Proceedings 34:709-714. Rabbinge, R., 1982. Pests, diseases and crop production. In: F.W.T. Penning de Vries and H.H. van Laar (Eds.): Simulation of plant growth and crop production. Simulation Monographs, Pudoc, Wageningen, pp. 253-265. Rabbinge, R., Ward, S.A., & Laar, H.H. van, 1989. Simulation and systems management in crop protection. Simulation Monographs. Pudoc, Wagen239

ingen 1989. Rawls, W.J., Brakensiek, D.L., & Saxton, K.E., 1982. Estimation of soil water properties. 1982 Transactions of the ASEA. Paper No. 81-2510. Rawson, H.M. & Constable, G.A., 1980. Carbon production of sunflower cultivars in field and controlled environments. I. Photosynthesis and transpiration of leaves, stems and heads. Australian Journal of Plant Physiology 7:555-573. Rawson, H.M., Begg, J.E. & Woodward, R.G., 1977. The effect of atmospheric humidity on photosynthesis, transpiration and water use efficiency of leaves of several plant species. Planta 134:5-10. Ray, T.B. & Black, C.C., 1979. The C4 pathway and its regulation. In: M. Gibbs and E. Latzko (Eds.), Encyclopedia of Plant Physiology, New Series, Vol 6. Springer Verlag, Berlin, pp. 77-101. Rasche, K., 1979. Movements of stomata. In: Encyclopedia of Plant Physiology, New Series, Vol 7. Springer Verlag, Berlin, pp. 383-441. Robertson, G.W., 1968. A biometeorological time scale for a cereal crop involving day and night temperature and photoperiod. International Journal of Biometeorology 12:191-223. Rose, C.W., 1966. Agricultural physics. Pergamon Press, New York. Ryle, G.J.A., Brockington, N.R., Powell, C.E., & Cross, B., 1973. The measurement and prediction of organ growth in uniculm barley. Annals of Botany 37:233-246. Rijtema, 1969. Soil moisture forecasting. Bulletin 513. Institute of Land and Water Management Research (ICW), Wageningen, 17 pp. Salisbury, F.B., 1981. Responses to Photoperiod. Encyclopedia Plant Physiology, New Series, Vol. 12A. Springer Verlag, Berlin, pp. 135-168. Saxena, N.P., Natarajan, M. & Reddy, M.S. Chickpea, pigeonpea and groundnut. In: Potential productivity of field crops under different environments. International Rice Research Institute, Los Baños, pp. 281-305. Saxton, K.E., Romberger, J.S. & Rawls, W.J., 1986. Estimating generalized soil water characteristics by texture. Soil Science Society of America Journal 50:1031-1036. Schapendonk, A.H.C.M. & Bouma, P., 1984. First growth of cucumber in relation to assimilate supply and sink capacity. Scientia Horticulture 23:2133, Schuh, W.M. & Bauder, J.W., 1986. Effect of soil properties on hydraulic conductivity-moisture relationships. Soil Science Society of America Journal 50:848-855. Schulze, E.-D. & Hall, A.E., 1982. Stomatal responses, water loss and CO2 assimilation rates of plants in contrasting environments. In: O.L. Lange, P.S. Nobel, C.B. Osmond & H. Ziegler (Eds.): Encyclopedia of Plant Physiology New Series, Volume 12B. Springer Verlag, Berlin, pp. 181-230. Shaykewich, C.F. & Stroosnijder, L., 1977. The concept of matric flux potential applied to simulation of evaporation from soil. Netherlands Journal of 240

Agricultural Science 25:63-82. Sharma, P.K., De Datta, S.K. & Redulla, C.A., 1987. Root growth and yield response of rainfed lowland rice to planting method. Experimental Agriculture 23:305-313. Sibma, L., 1987. Ontwikkeling en groei van mais (Zea mays L.) onder Nederlandse omstandigheden. Pudoc, Wageningen. Sinclair, T.R., Hoffer, R.M. & Schreiber, M.M., 1971. Reflectance and internal structure of leaves from several crops during a growing season. Agronomy Journal 63:864-868. Sinclair, T.R. & Wit, C.T. de, 1976. Analysis of carbon and nitrogen limitations to soybean yield. Agronomy Journal 68:319-324. Sivakumar, M.V.K., Seetharama, N., Singh, S. & Bidinger, F.R., 1979. Water relations, growth and dry matter accumulation of sorghum under postrainy season conditions. Agronomy Journal 71:843-847. Slatyer, R.O., 1967. Plant water Relationships. Academic Press, London. Smith, R.E. & Parlange, J.-Y., 1978. A parameter efficient hydrologic infiltration model. Water resources research 14:533-538. Spitters, C.J.T., 1984. A simple simulation model for crop-weed competition. 7th International Symposium on Weed Biology, Ecology and Systematics. COLUMA-EWRS, Paris, pp. 355-366. Spitters, C.J.T., 1986. Separating the diffuse and direct component of global radiation and its implications for modeling canopy photosynthesis. Part II. Agricultural and Forest Meteorology 38:231-242. Spitters, C.J.T., Toussaint, H.A.J.M. & Goudriaan, J., 1986. Separating the diffuse and direct component of global radiation and its implications for modeling canopy photosynthesis. Part I. Agricultural and Forest Meteorology 38:217-229. Spitters, C.J.T., Keulen, H. van & Kraalingen, D.W.G. van, 1989. A simple and universal crop growth simulation: SUCROS87. In: R. Rabbinge, S.A. Ward & H.H. van Laar (Eds): Simulation and systems management in crop protection. Simulation Monograph 32. Pudoc, Wageningen. Stansel, J.W., 1975. The rice plant — its development and yield. Six decades of rice research in Texas. Research Monograph 4. Texas Agricultural Experiment Station, pp. 9-21. Steinhorst, R.K., Hunt, H. W., Innis, G.S. & Haydock, K.P., 1978. Sensitivity analysis of the ELM model. In: G.S. Innis (Ed.): Grassland simulation model. Ecological Studies 26, Springer Verlag, New York. Steward, F.C. & Bidwell, R.G.S. (Eds.), 1983. Plant Physiology, Vol VII: Energy and Carbon Metabolism. Academic Press, New York. Steyeart, L.T., le Duc, S.K., & McQuigg, J.D., 1978. Atmospheric pressure and wheat yield modeling. Agricultural Meteorology 18:23-24. Stone, L.R., Teare, I.D., Nickell, C.D. & Mayaki, W.C., 1976. Soybean root development and soil water depletion. Agronomy Journal 68:677-680. Stroosnijder, L., 1982. Simulation of the soil water balance. In: F.W.T. Pen241

Stroosnijder, L., 1982. Simulation of the soil water balance. In: F.W.T. Penning de Vries and H.H. van Laar (Eds.): Simulation of plant growth and crop production. Simulation Monographs. Pudoc, Wageningen, pp. 175193. Summerfield, R.J., Minchin, F.R., Roberts E.H. and Hadley, P., 1983. In: Potential productivity of field crops under different environments. International Rice Research Institute, Los Baños, pp. 249-280. Summerfield, R.J. & Roberts, E.H. 1987. Effects of illuminance on flowering in long and short day grain legumes: a reappraisal and unifying model. In: J.G. Atherton (Eds.): Manipulation of flowering. Butterworth, London, pp. 203-223. Supit, I., 1986. Manual for generation of daily weather data. Simulation Reports CABO-TT No. 7. Centre for Agrobiological Research, Wageningen, 46 pp. Tanner, C.B. & Sinclair, T.R., 1983. Efficient use of water in crop production: research or re-search. In: H.M. Taylor, W.R. Jordan & T.R. Sinclair (Eds.): Limitations to efficient water use in crop production, ASA-CSSASSSA, Madison, pp, 1-27. Taylor, H.M. & Klepper, B., 1973. Rooting density and water extraction patterns for corn. Agronomy Journal 65:965-968. Taylor, H.M. & Klepper, B., 1974. Water relations and cotton. 1. Root growth and water use as related to top growth and soil water content. Agronomy Journal 66:584-588. Taylor, H.M. & Klepper, B., 1978. The role of rooting characteristics in the supply of water to plants. Advances in Agronomy 30:99-128. Teubner, F., 1985. Messung der Photosynthese und Transpiration an Weizen, Kartoffel und Sonnenblume. Diplomarbeit University Bayreuth, West Germany. TNO, 1977. Precipitation and measurements of precipitation. Proceedings and Informations No. 23. Verslagen en Mededelingen van de Commissie Hydrologies Onderzoek, Central Organization for Applied Scientific Research, the Hague. USDA Soil Survey Staff, 1975. Soil taxonomy: A basic system of soil classification for making and interpreting soil surveys. Agricultural Handbook 436. U.S. Government Printers Office, Washington, DC. Vanderlip, R.L. & Arkin, G.F., 1977. Simulating accumulation and distribution of dry matter in grain sorghum. Agronomy Journal 69:917-923. Vauclin, M., Haverkamp, R. & Vachaud, G. 1979. Resolution numerique d‘une equation de diffusion non-lineaire. Presse universitaire de Grenoble, France, 183 pp. Veen, B.W. & Boone, F.R., 1981. The influence of mechanical resistance and phosphate supply on morphology and function of corn roots. Plant and Soil 63:77-81. Veihmeyer, F.J. & Hendrickson, A.H., 1949. Methods of measuring field ca242

pacity and wilting percentages of soils. Soil Science 68:75-94. Veldkamp, H. J., 1986. Physiological causes of yield variation in cassava (Manihot esculenta Crantz). Agricultural University Wageningen Papers, 85-6, Agricultural University, Wageningen, Netherlands. Vergara, B.S. & Chang, T.T., 1985. The flowering response of the rice plant to photoperiod. International Rice Research Institute, Los Baños. Versteeg, M.N., 1985. Factors influencing the productivity of irrigated crops in southern Peru, in relation to prediction by simulation models. Ph.D. thesis, Agricultural University, Wageningen. Vertregt, N. & Penning de Vries, F.W.T., 1987. A rapid method for determining the efficiency of biosynthesis of plant biomass. Journal of Theoretical Biology 128:109-119. Visser, A.J.C. de, 1984. Interactions between energy and nitrogen metabolism in Pisum sativum. Ph. D. thesis. University of Groningen, Groningen. Vos, J., Drees, E.M. & Penning de Vries, F.W.T., 1982. Modelling of postfloral growth of wheat. In: F.W.T. Penning de Vries and H.H. van Laar (Eds.): Simulation of plant growth and crop production. Simulation Monographs. Pudoc, Wageningen, pp. 144-151. Vos, J. & Groenwold, J., 1983. Estimation of root densities by observation tubes and endoscope. Plant and Soil 74:295-300. Vos, J. & Groenwold, J., 1986. Root growth of potato crops on a marine-clay soil. Plant and Soil 94:17-33. Warrington, I.J. & Kanemasu, E.T., 1983. Corn growth response to temperature and photoperiod. I. Seedling emergence, tassel initiation and anthesis. Agronomy Journal 75:749-754. Werk, K.S., Ehleringer, J., Forseth, I.N. & Cook, C.S., 1983. Photosynthetic characteristics of Sonoran desert winter annuals. Oecologia (Berl.) 59(1) :101-105. Whisler, F.D., 1983. Sensitivity tests of the environmental variables in IRRIMOD. IRRI Research Paper Series 94. International Rice Research Institute, Los Baños. Whisler, F.D., Acock, B., Baker, D.N., Fye, R.E., Hodges, H.F., Lambert, J.R., Lemon, H.E., Mckinion, J.M. & Reddy, V.R., 1986. Crop simulation models in agronomic systems. Advances in Agronomy 40: 141-208. Wilson, D., 1982. Response to selection for dark respiration rate of mature leaves of Lolium perenne and its effects on growth of young plants and simulated swards. Annals of Botany 49:303-312. Wit, C.T. de, 1958. Transpiration and crop yields. Agricultural Research Reports 64.6. Pudoc, Wageningen, 88 pp. Wit, C.T. de, 1965. Photosynthesis of leaf canopies. Agricultural Research Reports 663. Pudoc, Wageningen. Wit, C.T. de, 1967. Photosynthesis: its relationship to overpopulation. In: Harvesting the sun. Academic Press, New York. Wit, C.T. de, 1968. Plant production. Miscellaneous Papers 3. Agricultural 243

University, Wageningen, pp. 25-50. Wit, C.T. de, 1970. Dynamic concepts in biology. In: I. Setlik (Ed.): Prediction and measurement of photosynthetic productivity. Proceedings International Biological Program/Plant Production Technical Meeting Trebon. Pudoc, Wageningen, pp. 17-23. Wit, C.T. de, Brouwer, R. & Penning de Vries, F.W.T., 1970. The simulation of photosynthetic systems. In: I. Setlik (Ed.): Prediction and measurement of photosynthetic productivity. Proceedings International Biological Program/Plant Production Technical Meeting Trebon. Pudoc, Wageningen, pp. 47-70. Wit, C.T. de & Keulen, H. van, 1972. Simulation of transport processes in soils. Simulation Monographs. Pudoc, Wageningen. Wit, C.T. de et al., 1978. Simulation of assimilation, respiration and transpiration of crops. Simulation Monographs. Pudoc, Wageningen. Wit, C.T. de & Goudriaan, J., 1978. Simulation of ecological processes. Simulation Monographs, Pudoc, Wageningen. Wit, C.T. de, Laar, H.H. van & Keulen, H. van, 1979. Physiological potential of crop production. In: J. Sneep and A.J.T. Hendiksen (Eds.): Plant breeding perspectives. Pudoc, Wageningen, pp. 47-82. Wit, C.T. de & Penning de Vries, F.W.T., 1982. La synthese et la simulation des systemes de production primaire. In: Penning de Vries, F.W.T. & M.A. Djiteye (Eds): La productivite des paturages saheliens. Agricultural Research Reports 918. Pudoc, Wageningen. Wit, C.T. de & Penning de Vries, F.W.T., 1983. Crop growth models without hormones. Netherlands Journal of Agricultural Science 31:313-323. Wit, C.T. de, Keulen, H. van, Seligman, N.G. & Sfarim, I., 1988. Application of interactive multiple goal linear programming to agricultural development. Agricultural Systems 26:211-230. Wittenback, V.A., Lin, W. & Hebert, R.R., 1982. Vacuolar localization of proteases and degradation of chloroplasts in mesophyl protoplasts from senescing primary wheat leaves. Plant Physiology 69:98-102. Woldendorp, J.W., 1978. The rhizosphere as part of the plant-soil system. In: A.H.J. Frijsen & J.W. Woldendorp (Eds.): Structure and functioning of plant populations. North Holland Publishing Company, Amsterdam, pp. 237-267. Wösten, J.H.M., Bannink, M.H. & Beuving, J., 1986. Waterretentie- en doorlatendheidskarakteristieken van boven-en ondergronden in Nederland: de Staringreeks. Institute for Land and Water Management, rapport 18. Wageningen. Yoshida, S., 1981. Fundamentals of rice crop science. International Rice Research Institute, Los Baños. Yoshida, S. & Hasegawa, S., 1982. The rice root system: its development and function. In: Drought resistance and crops with emphasis on rice. International Rice Research Institute, Los Baños, pp. 97-114. 244

Zadoks, J.C., Rijsdijk, F.H. & Rabbinge, R., 1984. EPIPRE, a systems approach to supervised control of pests and diseases in wheat in the Netherlands. In: G.R. Conway (Ed.): Pest and pathogen control: Strategy, tactical and policy models. International series on applied systems analysis. Wiley and Sons, pp. 34-351.

245

9 Index

air dry 148, 153 altitude 130 anaerobic conditions 135, 137, 138, 170 available water 160

BACROS 14 barley 32, 35, 47, 52, 76, 77, 78, 82, 83, 89, 100, 102, 125, 127, 137 bean 135 bulk density 148

CSMP iv, 19, 22, 23, 105, 142, 173 day length 73, 75, 81, 106, 108, 193, 199 descriptive model 1 deterministic model 161 development stage 73, 99, 141 dew 196 discrete processes 22 drainage 157 effect of temperature 136, 137

C3 crop 28, 30, 129, 130, 160 C4 crop 28, 30, 129, 130, 160 calibration 25, 136 canopy photosynthesis 14, 21, 28, 36, 38, 95, 97, 108, 109 canopy transpiration 119 capillarity 150 capillary rise 147, 160, 164, 180 carbohydrate accumulation 42 carbohydrate allocation patterns 87 carbon balance check 96, 107 carbon dioxide 198 carbon use efficiency 59 cassava 32, 64, 76, 77, 78, 84, 88, 92, 93, 96, 98, 100, 101, 127, 130, 132, 135 chickpea 83, 125 clay 148 comprehensive model 5 cotton 32, 35, 47, 52, 76, 77, 78, 81, 82, 84, 89, 100, 102, 125, 127, 132, 135, 137 cowpea 32, 64, 76, 77, 78, 82, 84, 89, 100, 108, 109, 125, 127, 132, 135, 137 crop height 124 246

ELCROS 14 evaluation 24 evaporation 118, 158, 167, 175 evaporation front 168, 178 evapotranspiration 118, 199 excretion of carbon 96 explanatory model 1, 14, 74 extinction coefficient 36, 119, 158, 159 faba bean 17, 27, 32, 47, 52, 64, 76, 77, 78, 82, 83, 84, 89, 100, 102, 121, 125, 127, 132, 137, 140, 141 field bean 52, 64, 84 field capacity 134, 139, 148, 152, 155 floret formation 94 flux density 162 function iv gas pressure 150 gravitational potential 150 groundnut 32, 35, 64, 76, 77, 78, 82, 83, 84, 89, 100, 102, 125, 127, 135 groundwater 166 hormones 95, 140

humidity 45, 118, 121, 130, 171, 191, 197 hydraulic conductivity 148, 153, 164 hydraulic head 165, 177 infiltration 164, 166 information sciences 12 integration period 22, 161 interception 196 internal/external ratio of CO2 125 irrigation 155, 166, 179 L1D 171 L2C 171 L2SS 171, 173 L2SU 171, 172 leaf angle distribution 44 leaf resistance 201 leaf rolling 142 leguminous crops 62 levels of plant production 6 light response curve 29 listing iv lowland rice 160 MACROS iv maize 2, 3,27, 31, 32, 33, 35, 37, 47, 52, 64, 76, 77, 78, 82, 83, 84, 88, 89, 98, 99, 100, 101, 102, 103, 108, 110, 125, 127, 128, 135, 137 matric flux potential 162, 178 matric potential 150 micrometeorological terms 118 millet 20, 32, 35, 47, 52, 64,76,77, 78, 82, 83, 84, 89, 100,102,125, 127, 128, 135, 137 mimick iv model iv module iv moisture characteristic 151 net radiation 199 nitrogen 7, 9, 130 nitrogen reduction 62

objectives 13 organ formation 94 PAR 30 parametric 161 partitioning 141 pea 135 perched water table 160 permanent wilting point 134, 152 pF curve 151 Phaseolus vulgaris 35 phosphorus 7, 11 photoperiodicity mechanism 199 photorespiration 28, 30 photosynthesis 28-46, 53-54, 128-130 piezometer 170 pigeon pea 64, 84 potato 19, 21, 32, 34, 35, 42, 47, 48, 59, 64, 76, 77, 78, 79, 81, 82, 83, 84, 85, 87, 88, 89, 93, 100, 102, 125, 126, 127, 128, 130, 131, 132, 135, 137 potential growth 6 precipitation 17, 157, 166, 179, 191, 192, 196,203 pressure head 150 production level 2 171 program iv rainfed lowland 135, 138 reflectivity 158 reserves 56 respiration 14, 49, 51,65, 122, 143 rhizobia-bacteria 62 rice 14, 15, 17, 19, 32, 35, 44, 47, 52, 62, 63, 64, 73, 74, 76, 77, 78, 79, 80, 82, 83, 84, 85, 86, 88, 89, 93, 95, 99, 100, 102, 125, 126, 127, 132, 135, 137, 138, 140, 142, 180, 181, 182, 199, 200 rooted depth 133,134, 136, 138 running average 23,107,140 runoff 157, 166 247

SAHEL iv, 155, 172, 179 sand 148 saturation 139 SAWAH iv, 161, 166, 173, 180 senescence 43, 48, 95, 134 sensitivity analysis 149 shielded reserves 92 silt 148 simulation 4, 11, 18 sink size 42, 58, 93 soil texture 148 soil water potential 150 solar constant 198 solar height 199 solar radiation 30, 118, 191, 193, 198 sorghum 32, 35, 47, 52, 64, 76, 77, 78, 82, 83, 84, 89, 100, 102, 125, 127, 132, 135, 137 soya bean 19, 32, 35, 47, 64, 76, 77, 78, 82, 83, 84, 85, 86, 89, 100, 102, 125, 127, 132, 135, 137, 139, 179, 183 spatial variability 147, 196 specific leaf weight 96, 98 specific stem weight 104 state variable approach 5 stomatal resistance 14, 118, 121, 126, 130 structural dry matter 56, 96 subroutine iv SUCROS 14 suction 150 sugar-beet 32, 35, 64, 76, 77, 78, 84, 89, 100, 102, 125, 127, 137, 141 sugar-cane 32, 33, 47, 56, 64, 76, 77, 78, 80, 84, 87, 93, 100, 127, 135, 140 summary type model 15

sunflower 32, 35, 43, 47, 52, 64, 76, 77, 78, 82, 83, 84, 89, 100, 102, 125, 126, 127, 131, 132, 135, 139 surface roughness 158 surface storage capacity 166 SUSAWA 161, 176 sweet potato 32, 35, 47, 64, 76, 77, 78, 83, 84, 89, 125, 127, 135 system iv, 4 temperature 44, 141 tensiometer pressure 150 tiller formation 94 time coefficient 3, 17, 19, 23, 51, 54, 94, 140 tomato 64 topsoil drying 168 transpiration 22, 118 transpiration coefficient 128 tulip 16, 32, 47, 49, 76, 77, 78, 82, 83, 87, 100, 137 vertical cracks 153 vertisols 148 water balance 22, 117, 147, 176 water stress 140, 160, 171, 175 water use 117 water use efficiency 128 water use coefficient 128 wheat 15, 16, 17, 32, 33, 35, 47, 52, 57, 64, 76, 77, 78, 79, 80, 82, 83, 84, 88, 89, 94, 100, 101, 102, 125, 126, 127, 131, 132, 135, 137, 139, 142 wilting point 148 wind speed 118, 120, 198 yam 64

248

Appendix A. Error messages from SUBROUTINES and FUNCTIONS

Message Name Condition number subroutine 1.1

FUPHOT

ALV < 0 or ALV > 25

1.2

FUPHOT

PLMX < 0 or PLMX > 100

1.3

FUPHOT

PLEA < 0 or PLEA > 0.7

2.1 2.2 2.3

SUASTC SUASTC SUASTC

DATE < 0. or > 365. AOB < -1.0 or > 1.0 RDTM < 0. or > RDTC

3.1 3.2 3.3

SUEVTR SUEVTR SUEVTR

FRD < 0. or > 1. VPAS < 0. or > 12.55 VPA < 0. or > VPAS

4.1 5.1

SUMSKM SUPHOL

MS < 0. or > 1.E8 ALVL(I2) < 0 or > 25

5.2

SUPHOL

ALVDL(I2) < 0 or > 25

5.3 5.4 6.1 6.2

SUPHOL SUPHOL SUSLIN SUSLIN

7.1

SUWCMS

7.2

SUWCMS

PLMXL(I2) < 0 or > 100 PLEAL(I2) < 0 or > 0.7 TYL(I) < 1. or > 20. WCLQTI(I) < WCAD(I) or > WCST(I) WCLQT < WCAD(I) or > WCST(I) MS < 0. or > 1.E8

Possible cause

Wrong initialization of leaf area or weight; too much leaf death Wrong initialization of PLMXP; reduction of PLMXP excessive Wrong initialization of PLEI; reduction of PLEI excessive Wrong calculation of Julian date LAT < -66.5 or > 66.5 degrees Wrong conversion of measured radiation into J m -1 d-1 Wrong calculation of fraction Wrong calculation of TPAD Wrong conversion of measured air humidity into kPa Wrong calculation of MS Wrong calculation of green area of a layer Wrong calculation of dead area of a layer Wrong calculation of PLMX(I) Wrong calculation of PLEA(I) Wrong number in TABLE TYL Wrong initialization water contents; check measurements, units Wrong calculation of water content in a soil compartment Wrong calculation of MS

249

Appendix B. Listing of module T12 with SUBROUTINES and FUNCTIONS used in Listings 1-11

TITLE T12 .CSM, MAY 88 END STOP FUNCTION FUCCHK (CKCIN,CKCFL,TIME) C check on crop carbon balance. used in LlD, L1Q. 03/07 FUCCHK-2.0*(CKCIN-CKCFL)/(CKCIN+CKCFL+l.E-10) IF(ABS(FUCCHK).GT.0.0l) WRITE (6,l0) FUCCHK, CKCIN, CKCFL, TIME 10 FORMAT(/'* * *error in carbon balance, please check* * *',/,'CKCRD $=',F6.3,' CKCIN=',F8.2,' CKCFL=’,F8.2,' AT TIME=',F6.1) RETURN END FUNCTION FUPHOT (PLMX,PLEA,ALV,RDTM,DATE,LAT) C computes canopy photosynthesis. used in LlD, L1Q. 8/87 IMPLICIT REAL (A-Z) INTEGER IT,I DATA KDIF/0.7155/,PI/3.1415926/,SCV/0.200/,GAUSR/0.3872893/ CALL SUERRM(l.l,ALV, 0., 25.,6.) CALL SUERRM(1.2,PLMX,0., l00.,6.) CALL SUERRM(1.3,PLEA,0., 0.7,6.) CALL SUASTC(DATE, LAT, RDTM, RDTC, FRDIF, COSLD, SINLD, DSINBE, SOLC, DLA) GDFG =0. IF(PLMX*PLEA*ALV.LE.0.0) GOTO 50 ALVL -AMINl(l0.,ALV) -(l.-SQRT(l.O-SCV))/(l.+SQRT(l.-SCV)) REFH DO 40 IT-1,3 HOUR =12.0+DLA*O.5*(0.5+(IT-2)*GAUSR) SINB =AMAX1(0.,SINLD+COSLD*COS(2.*PI*(HOUR+12.)/24.)) REFS =REFH*2./(1.+1.6*SINB) PAR =0.5*RDTM*SINB*(1.0+0,4*SINB)/DSINBE PARDIF=AMINl(PAR,SINB*FRDIF*(RDTM/RDTC)*O.5*SOLC) PARDIR=PAR-PARDIF KDIRBG=(0.5/SINB)*KDIF/(O.8*SQRT(1.-SCV)) KDIRT =KDIRBL*SQRT(l.-SCV) FGROS =0. DO 30 I=1,3 ALVC =0.5*ALVL+GAUSR*(I-2)*ALVL VISDF =(1.-REFS)*PARDIF*KDIF*EXP(-KDIF*ALVC) VIST =(1.-REFS)*PARDIR*KDIRT*EXP(-KDIRT*ALVC) VISD =(1.-SCV)*PARDIR*KDIRBL*EXP(-KDIRBL*ALVC) VISSHD =VISDF+VIST-VISD FGRSH =PLMX*(l.-EXP(-VISSHD*PLEA/PLMX)) VISPP =(I..-SCV)*PARDIR/SINB IF (VISPP.LE.0.) GO TO 10 FGRSUN=PLMX*(l.-(PLMX-FGRSH)*(l.-EXP(VISPP*PLEA/PLMX))/ $ (PLEA*VISPP)) GO TO 20 10 FGRSUN=FGRSH 20 CONTINUE FSSLA =FXP(-KDIRBL*ALVC)

251

30

40 50

FGL =FSSIA*FGRSUN+(l.-FSSLA)*FGRSH IF(I.EQ.2) FGL =FGL*1.6 FGROS =FGROS+FGL CONTINUE FGROS =FGROS*ALVL/3.6 IF(IT.EQ.2) FGROS =FGROS*1.6 GDFG =GDFG+FGROS CONTINUE FUPHOT=GDFG*DLA/3.6 RETURN END FUNCTION FURSC(WDS,ALV,PLHT,ZREF)

C calculates canopy resistance upper layers.

in L2C. 4/87

ZR =AMAXl(ZREF,PLHT+l.) D =AMAX1(0.1,0.63*PLHT) ZNOT =AMAX1(0.05,0.1*PLHT) ALVX =AMAX1 (1.,ALV) WDSX =AMAX1(0.2,WDS) FURSC =0.74*(ALOG((ZR-D)/ZNOT))**2/(0.16*WDSX)*ALVX RETURN END FUNCTION FUTP (IDATE,DTIME,TPHT,TPLT,FA,FB,FC,FD) C approximates daily course of air temperature. in L1Q. 9/85 DIMENSION TPHT(365),TPLT(365) IF(IDATE.EQ.366) IDATE=365 FUTP =FA*TPHT(MAX0(l,IDATE-1))+(1.-FA)*TPLT(IDATE) IF(DTIME.GT.0.2) FUTP =FB*TPHT(IDATE)+(l.-FB)*TPLT(IDATE) IF(DTIME.GT.0.4) FUTP =FC*TPHT(IDATE)+(l.-FC)*TPLT(IDATE) IF(DTIME.GT.0.6) FUTP =FD*TPHT(IDATE)+ (l.-FD)*TPLT(MIN0(365,IDATE+l)) $ RETURN END FUNCTION FUVP (TP) C vapour pressure (kPa) relation to temperature. in LlQ, L2C. 9/85 FUVP =0.100*6.1l*EXP(17.47*TP/(TP+239.)) RETURN END FUNCTION FUWCHK (CKWFL,CKWIN,TIME) C check on soil water balance. used in L2SU, L2SS. 3/87 FUWCHK=2.0*(CKWIN-CKWFL)/(CKWIN+CKWFL+l.E-10) IF(ABS(FUWCHK).GT.0.0l.AND.ABS(CKWIN).GT.0.2) $WRITE (6,10) FUWCHK,CKWIN,CKWFL,TIME 10 FORMAT(/‘* * *error in water balance, please check***‘,/,' $=’,F6.3,’ CKWIN=’,F8.2,’ CKWFL=',F8.2,’ AT TIME=',F6.1) RETURN END FUNCTION FVWCMS (I,MS) C converts moisture suction into water contents. in L2SS. 9/87 REAL MS CALL SUWCMS(I,2,WCL,MS) FUWCMS=WCL RETURN END

252

CKWRD

FUNCTION FUWRED (WDLV,ALV,PLHT,WDS) C calculates windspeed near soil surface. in L2C. 9/87 IMPLICIT REAL (A-Z) PLHTX =AMAX1(0.05,PLHT) ALVX =AMAXl(0.0l,ALV) MIXL =SQRT(l.2732*AMAXl(0.005,WDLV)/(ALVX/PLHTX)) A =SQRT(0.2*ALVX*PLHTX/(2.*MIXL*0.5)) FUWRED =AMAX1(0.2,WDS)*EXP(-A*(l.0-0.05/PLHTX)) RETURN END FUNCTION FUWS (TRC,ALV,WCL,WSSC,WFSC,WCWP,WCFC,WCST) C computes reduction of water uptake, used in L2SU, L2SS. 5/87 DATA A,B,ALVMAX/0.76,0.15.2./ IF(WCL .LE. WCFC) THEN SDPF =1./(A+B*ALVMAX*TRC/(ALV+l.E-l0))-(l.-WSSC)*0.4 IF(WSSC.LT.0.6) THEN SDPF =SDPF+0.025*AMIN1(0.,ALVMAX*TRC/(ALV+1.E-10)-6.)/ (1.+5.*WSSC+4.*WSSC*WSSC) $ ENDIF WCX =WCWP+(WCFC-WCWP)*(l.00-AMINl(l.,AMAX1(0.,SDPF))) FWSX =(WCL-WCWP)/(WCX-WCWP+l.E-10) ELSE FUWSX =1.-(1.-WFSC)*(WCL-WCFC)/(WCST-WCFC+l.E-10) ENDIF FUWS =AMINl(l.,AMAXl(0.,FUWSX)) RETURN END SUBROUTINE SUASTC(DATE,LAT,RDTM,RDTC,FRDIF,COSLD, SINLD,DSINBE,SOLC,DLA) $ C astronomical standard computations. used in LID, L1Q. 5/87 IMPLICIT REAL (A-Z) DATA PI/3.1415926/,RAD/0.0174533/ DEC =-ASIN(SIN(23.45*RAD)*COS(2.*PI*(DATE+10.)/365.)) COSLD =COS(DEC)*COS(LAT*RAD) SINLD =SIN(DEC)*SIN(LAT*RAD) AOB =SINLD/COSLD CALL SUERRM(2.l,DATE.0.,365.,6.) CALL SUERRM(2.2,AOB,-l.0,1.0,6.) DLA =12.*(1.+2.0*ASIN(AOB)/PI) DSINBE =3600.*(DLA*(SINLD+0.4*(SINLD*SINLD+COSLD*COSLD*0.5))+ 12.O*COSLD*(2.0+3.0*0.4*SINLD)*SQRT(1.-AOB*AOB)/PI) $ DSINB =3600.*(SINLD*DLA+24./PI*COSLD*SQRT(l.-AOB**2)) SOLC =1370.*(1.0+0.033*COS(2.*PI*DATE/365.)) RDTC =SOLC*DSINB CALL SUERRM(2.3,RDTM,O.,RDTC.6.) ATMTR =RDTM/RDTC IF(ATMTR.GT.0.75) FRDIF =0.23 IF(ATMTR.LE.0.75.AND.ATMTR.GT.0.35) FRDIF =1.33-1.46*ATMTR IF(ATMTR.LE.0.35.AND.ATMTR.GT.0.07) FRDIF =1.-2.3*(ATMTR-0.07)**2 IF(ATMTR.LE.0.07) FRDIF =1.00 RETURN END SUBROUTINE SUASTR (DATE,LAT,RDTC,DLA,DLP) C computes daylength, daily total radiation clear. in LlD, L1Q. 5/87

253

IMPLICIT REAL (A-Z) DATA INSP/-4.0/,PI/3.1415926/,RAD/0.0174533/ CALL SUASTC(DATE,LAT,RDTM,RDTC,FRDIF,COSLD,SINLD,DSINBE,SOLC,DLA) DLP =12.*(PI+2.*ASIN((-SIN(INSP*RAD)+SINLD)/COSLD))/PI RETURN END SUBROUTINE

SUCONV(SWICH2,TKL,ZL,ZLT,RAIN,ZW,WL0,WL0MX,

$ RUNOF,TRWL,EVSW,EVSC,DRSL,NL) C converts units between main program and subroutines. IMPLICIT REAL (A-Z) INTEGER I,NL,SWICH2 DIMENSION TKL(10),ZL(10),TRWL(10) F1 =0.01 F2 =10.0 IF(SWICHZ.EQ.1) F1=1./F1 IF(SWICH2.EQ.l) F2=1./F2 DO 10 I=l, NL =TKL(I) *F1 TKL(I) =ZL(I) *F1 ZL(I) TRWL(I) =TRWL(I)*F2 10 CONTINUE ZLT =2LT *F1 ZW =2W *F1 WL0 =WL0 *F1 WL0MX =WL0MX*Fl RAIN =RAIN *F2 EVSW =EVSW *F2 EVSC =EVSC *F2 RUNOF =RUNOF*F2 DRSL =DRSL *F2 RETURN END

in L2SS. 4/88

SUBROUTINE SUERRM(MNR,X,XMIN,XMAX,NUNIT) C checks whether X is between limits. in FUNCTIONS, SUBROUTINES. 8/87 IMPLICIT REAL (A-Z) INTEGER IUNIT IF((X.LT.XMIN*0.99).AND.(XMIN.NE.-99.)) GOTO 10 IF((X.GT.XMAX*l.0l),AND.(XMAX.NE.-99.)) GOTO 10 RETURN 10 IUNIT =IFIX(NUNIT) WRITE(IUNIT,20) MNR,X,XMIN,XMAX STOP 20 FORMAT(//,' ***fatal error in variable or parameter value ***', $/,' message number, value, minimum and maximum: ';/,10X,F4.1, $3(3X,E10.3)) END SUBROUTINE SUEVTR(RDTC,RDTM,RF,FRD,TPAD,VPA,RSL,RSB,RST,EVPR,EVPD) C potential evapotranspiration rates crop, soil. used in L2C. 7/87 CALL SUERRM(3.1,FRD,0.,1.,6.) VPAS =FIJVP(TPAD) CALL SUERRM(3.2,VPAS,0.0,12.55,6.) CALL SUERRM(3.3,VPA,0.0,VPAS,6.) SLOPE =4158.6*10.*VPAS/(TPAD+239.)**2 APSCH =0.67*(RSB+RST+RSL)/(RSB/0.93+RST) RLWI =4.8972E-3*(TPAD+273.)**4*(0.618+0.0365*SQRT(10.*VPA))

254

RLWO RDTN EVPR DRYP EVPD RETURN END

=4.8972E-3*1.00*(TPAD+273.)**4 =RDTM*(1.-RF)-(RLWO-RLWI)*(RDTM/(0.75*RDTC))*FRD =0.001*RDTN*SLOPE/((SLOPE+APSCH)*2390.) =(VPAS-VPA)*l0.*1200./(RSB+RST) *FRD =86400.*0.00l*DRYP/((SLOPE+APSCH)*2390.)

SUBROUTINE SUGRHD(TKL,NL,HGT,HGB) C calculates gravitational head at interfaces. in L2SS. 9/87 DIMENSION TKL(l0),HGT(l0),HGS(l0) HGT(1) =0. HGB(1) =-TKL(1) DO 10 I-2,NL HGT(1) =HGT(I-l)-TKL(I-l) HGB(1) =HGB(I-1)-TKL(I) 10 CONTINUE RETURN END SUBROUTINE SUINTG(HPBP,SWICH4,SWICH5,HPP,FLX,TKL,WCL, $ MS,DTMIN,DTMXl,DTMXZ,DTFX,INXSAT,DHH, $ JTOT,FLXSQ2,WL0,WL0MX,NL,DELT,DT) C calculates SAWAH-timestep, integrates rates during day;in L2SS;O8/88 IMPLICIT REAL (A-Z) INTEGER NL,I,IX,ITEL1,ITEL2,SWICH4,SWICHS,INXSAT,JTOT COMMON /VOLWAT/ WCAD(10),DUMMY1(10),WCST(l0),DUMMY2(10) DIMENSION WCL(10),HPP(11),INXSAT(10,10) DIMENSION WCLRCH(10),FLX(l1),TKL(10),MS(l0),DHH(10) DATA TINY/l.0E-l0/ HPTP =WL0 IF(HPTP.LE.0..OR.FLX(l).LE.0.) THEN DTSRF =DELT ELSE DTSRF =AMAXl(TINY,HPTP/(FLX(l))) ENDIF DT =AMINl(DELT,DTSRF,DTMXl,DTMX2) DO 10 I-1,NL WCLRCH(1)=(FLX(I)-FLX(I+l))/TKL(I) IF(WCLRCH(I).LT.-TINY)SATTIM=-(WCL(I)-WCAD(I)-TINY)/WCLRCH(I) IF(WCLRCH(I).GT. T1NY)SATTIM= (WCST(1)-WCL(I)-TINY)/WCLRCH(I) IF(ABS(WCLRCH(I).LT.TINY)SATTIM=DELT DT =AMINl(SATTIM,DT) 10 CONTINUE IF(SWICH5.EQ.2) THEN DT =AMINl(DT,DTFX) IF(DHH(JT0T).LT.-TINY.OR.JTOT.EQ.0) THEN CONTINUE ELSE IX =INXSAT(JTOT,1)-1 IF(IX.LT.l) THEN CONTINUE ELSEIF(FLX(IX+l).GT.-0.1) THEN CONTINUE ELSEIF(FLXSQ2.LE.0.) THEN CONTINUE ELSE MSAL =TKL(IX)/Z.

255

20 30

40 50 60

256

CALL SUWCMS(IX,0,WIX2,MSAL) MSACT =MS(IX) CALL SUWCMS(IX,0,WIXl,MSACT) IF(WIX2.LE.WIX1) THEN CONTINUE ELSE DLIM=-(WIX2-WIXl)*TKL(IX)/FLX(IX+1) DT=AMIN1(DT,AMAX1(DLIM,TINY)) ENDIF ENDIF ENDIF GOTO 100 ELSE CONTINUE ENDIF GOTO 30 DT =DT/2. IF(DT.LT.DTMIN) GOTO 100 IF(ABS(WCLRCH(l)).LT.TINY) GOTO 40 IF(HPTP.GT.TINY) THEN MST1 =-(HPTP-DT*(FLX(l))) ELSE GOTO 40 ENDIF WCT2 =WCL(l)+DT*FLX(l)/TKL(l) IF((WCT2-WCAD(l)).LE.-TINY) GOTO 20 IF(WCT2.GE.WCST(1)) THEN MSTZ =WCT2-WCST(l) ELSE CALL SUWCMS(l,SWICH4,WCTL,MST2) ENDIF DELZ =0.5*TKL(l) DH =-MST2+MST1-DELZ CH =-DH*FLX(l) IF(CH.LT.-TINY) GOTO 20 DO 70 I=2,NL GOTO 60 DT =DT/2. ITEL1 =0 ITEL2 =0 IF(DT.LT.DTMIN) GOTO 100 IF(ABS(WCLRCH(I-l)).LT.TINY.AND.ABS(WCLRCH(I)).LT.TINY) IF(ABS(WCLRCH(I-l)).LT.TINY) THEN IF(ABS(WCL(I-1)-WCST(I-l)).LT.TINY) THEN DELZ =0.5*TKL(I) MSTl =0. ITEL2 =1 ELSE DELZ =0.5*(TKL(I-l)+TKL(I)) MSTl =MS(I-1) ENDIF ELSE DELZ =0.5*(TKL(I-l)+TKL(I)) WCTl =WCL(I-l)-DT*FLX(I)/TKL(I-l) IF((WCT1-WCAD(I-l)).LE.-TINY) GOTO 50 IF(WCTl.GE.WCST(I-1)) THEN MSTl =WCTl-WCST(I-I) ELSE

GOTO 70

70 80 90

100

CALL SUWCMS(I-l,SWICH4,WCTl,MSTl) ENDIF ENDIF IF(ABS(WCLRCH(I)).LT.TINY) THEN IF(ABS(WCL(1)-WCST(I)).LT.TINY) THEN DELZ =0.5*TKL(I-l) MST2 =0. IF(HPP(I).GT.TINY) THEN IF(HPP(I).GE.TKL(I-1)) ITELl =1 DELZ =0.5*TKL(I-l) MST2 =0. ENDIF ELSE DEL2 =0.5*(TKL(I-l)+TKL(I)) MST2 =MS(I) ENDIF ELSE DELZ =0.5*(TKL(I-l)+TKL(I)) IF(ITEL2.EQ.l) DELZ =0.5*TKL(I) WCT2 =WCL(I)+DT*FIX(I)/TKL(I) IF((WCT2=WCAD(I)).LE.-TINY) GOTO 50 IF(WCT2.GE.WCST(I)) THEN MST2 =WCT2-WCST(I) ELSE CALL SWCMS(I,SWICH4,WCT2,MST2) ENDIF ENDIF DH =MST2+MSTl-DELZ CH =DH*FIX(I) IF(ITELl.EQ.1) CH =+1. IF(CH.LT.-TINY) GOTO 50 CONTINUE GOTO 90 DT =DT/2. IF(DT.LT.DTMIN) GOTO 100 IF(ABS(WCLRCH(NL)).LT.TINY) GOT0 100 ITELl =0 ITEL2 =0 WCTl =WCL(NL)-DT*FLX(NL+l)/TKL(NL) IF((WCT1=WCAD(NL)).LE.-TINY) GOTO 80 IF(WCTl.GE.WCST(NL)) THEN MSTl =WCTl-WCST(NL) ELSE CALL SWCMS(NL,SWICH4(,WCTl,MSTl) ENDIF DELZ -0.5*TKL(NL) IF(HPBP.GT.TINY.AND.WCL(NL).LT.WCST(NL)) THEN IF(HPBP.GE.TKL(NL)) ITELl =1 DELZ =TKL(NL) MST2 =-HPBP ENDIF DH =MST2+MSTl-DELZ CH =DH*FLX(NL+l) IF(ITEL1.EQ.1) CH =+1 IF(CH.LT.-TINY) GOTO 80 WL0 =AMAXl(0.,WL0-DT*FLX(l)) DO 110 I=1,NL WCL(1) =WCL(I)+DT*WCLRCH(I)

257

110

CONTINUE RETURN END

SUBROUTINE SUMFLP(SWICH3,I,MS,MFLP) C calculates matrix flux potential. in L2SS. 8/87 IMPLICIT REAL (A-Z) INTEGER I,IG,IX,SWICH3 COMMON /HYDCON/DUMMY1(10),KMSA1(10),DUMMY2(10),KST(10) DIMENSION MSI(8),XGAUS(3),WGAUS(3) DATA MSI/0.,10.,50.,250.,750.,1500.,5000.,10000./ DATA XGAUS/0.112702,0.5,0.887298/ DATA WGAUS/0.277778,0.444444,0.277778/,TINY/1.E-10/ MFLP =0. IF(MS.GT.TINY) THEN IF(SWICH3.EQ.l) THEN MFLP =(KST(I)/KMSAl(I))*(EXP(-KMSAl(I)*MS)-1.) ELSE DO 30 IX=2,7 DMFLP =0. IF(MS.GT.MSI(IX-1)) THEN MSX =AMIN1(MSI(IX),MS) IF(IX.EQ.2) THEN DO 10 IG=1,3 X =MSX*XGAUS(IG) CALL SUMSKM(SWICH3,I,X,KMSX) DMFLP =DMFLP+KMSX*WGAUS(IG) 10 CONTINUE MFLP =MFLP-DMFLP*(MSX-MSI(1X-1)) ELSE DO 20 IG=1,3 X =MSX*(MSI(IX-l)/MSX)**XGAUS(IG) CALL SUMSKM(SWICH3,I,X,KMSX) DMFLP =DMFLP+X*KMSX*WGAUS(IG) 20 CONTINUE IF(DMFLP.LE.0.0) GOTO 40 MFLP =MFLP-DMFLP*ALOG(MSX/MSI(IX-1)) ENDIF ENDIF 30 CONTINUE ENDIF ENDIF 40 CONTINUE RETURN END SUBROUTINE SUMSKM(SWICH3,I,MS,KMS) C calculates hydraulic conductivity from suction. in L2SS. 8/87 IMPLICIT REAL (A-Z) INTEGER SWICH3,I COMMON /HYDCON/ KMSMX(10),KMSA1(10),KMSA2(10),KST(10) DATA TINY,MSAD/l.E-10,l.E7/ CALL SUERRM(4.1,MS,0.,l.E8,6.) THEN IF(MS.LT.MSAD-TINY) IF((SWICH3.EQ.2).AND.(MS.GT.KMSMX(I))) THEN KMS =KMSA2(I)*(MS**(-1.4)) ELSE KMS =KST(I)*EXP(-KMSAl(I)*MS)

258

ENDIF IF(KMS.LT.TINY) KMS ELSE KMS =0. ENDIF RETURN END

=0.

SUBROUTINE SUPHOL (IN,PLMX,PLEA,ALVL,ALVDL,Fl,F2,RDTM, $ DATE,LAT,PCGC,PCGCL) C computes canopy photosynthesis for 1-5 layers. for L1D,L1Q. 8/87 IMPLICIT REAL (A-Z) INTEGER I1,I2,I3,I4,IN DIMENSION PLMX(5),PLEA(5),ALVL(5),ALVDL(5),PCGCL(5) DIMENSION F1(5),F2(5),F3(5),PHL(5,3) DATA PI/3.1415926/,SCV/0.200/,GAUSR/0.3872983/ CALL SUASTC(DATE,LAT,RDTM,RDTC,FRDIF,COSLD,SINLD,DSINBE,SOLC,DLA) KBLTOP=0.97*F1(1)+0.85*F2(1)+0.65*(1.00-F1(1)-F2(1)) KDFTOP=KBLTOP*SQRT(l.0-SCV) DO 20 I2=1,5 PCGCL(I2)=0.0 DO 10 I1=1,3 PHL(I2,I1)-0.0 10 CONTINUE CALL SUERRM(5.1,ALVL(I2), 0.,25.0,6.) CALL SUERRM(5.2,ALVDL(I2),0.,25.0,6.) CALL SUERRM(5.3,PLMX(I2), 0.,100.,6.) CALL SUERRM(5.4,PLEA(I2), 0., 0.7,6.) 20 CONTINUE DO 60 Il=-l,l HOUR =12.0+DLA*0.5*(0.5+1l*GAUSR) SINB =AMAXl(0. ,SINLD+COSLD*COS(2.*PI*(HOUR+12.)/24.)) 015 =AMAX1(0.16,0.966*SINB) 045 =AMAX1(0.46,0.707*SINB) 075 =1.0-0.268*015-0.732*045 0T0P =Fl(l)*015+F2(1)*045+(1.00-Fl(l)-F2(1))*075 REFH =(1.-SQRT(l.0-SCV))/(l.+SQRT(l.-SCV)) REFV =REFH*2.0*0T0P/(0T0P+KDFTOP*SINB) PAR =0.5*RDTM*SINB*(1.0+0.4*SINB)/DSINBE PARDIF=AMINl(PAR,SINB*FRDIF*(RDTM/RDTC)*0.5*SOLC) PARDIR=AMAXl(0.,PAR-PARDIF) RTDIF =(l.0-REFV)*PARDIF RTDIRT=(1.0-REFV)*PARDIR RTDIRD=(1.0-SCV )*PARDIR SUNPER=RTDIRD/SINB FSSL =1.0 DO 50 I2=1,IN IF ((ALVL(I2)*PLEA(I2)*PLMX(I2)).LE.0.) GOTO 50 F3(12)=l.00-F2(I2)-Fl(I2) 0 =Fl(I2)*015+F2(I2)*045+F3(I2)*075 T2DS =F1(I2)*0.034+F2(I2)*0.25+F3(I2)*0.47+ $ SINB*SINB*(F1(I2)*0.90+F2(I2)*0.25-F3(I2)*0.42) RANGET=SQRT(12.0*AMAXl(0.,T2DS-0*0)) KBL =0.97*F1(I2)+0.85*F2(I2)+0.65*F3(I2) KDIF =KBL*SQRT(l.0-SCV) KDIRBL=0/SINB KDIRT =KDIRBL*SQRT(l.-SCV) FGROS =0.

259

30

40

50 60

70

DO 40 I3=-1,l ALVC =(0.5+GAUSR*I3)*(ALVL(I2)+ALVDL(I2)) VISDF =RTDIF*KDIF*EXP(-KDIF*ALVC) VIST =RTDIRT*KDIRT*EXP(-KDIRT*ALVC) VISD =RTDIRD*KDIRBL*EXP(-KDIRBL*ALVC) VISSHD =VISDF+VIST-VISD FGRSH =PLMX(I2)*(1.-EXP(-VISSHD*PLEA(I2)/PL(I2))) FGRSUN =0.0 DO 30 I4=-1,1 SN =0+RANGET*I4*GAUSR VISSUN=VISSHD+SN*SUNPER FGRS =PLMX(I2)*(1.0-EXP(-VISSUN*PLEA(I2)/PLMX(I2))) IF (14.EQ.0) FGRS =FGRS*1.6 FGRSUN=FGRSUN+FGRS CONTINUE FGRSUN=FGRSUN/3.6 FSSLA =FSSL*EXP(-KDIRBLK*ALVC) FGL =FSSLA*FGRSUN+(l.-FSSLA)*FGRSH IF(13.EQ.0) FGL =FGL*1.6 FGROS =FGROS+FGL CONTINUE FGROS =FGROS*ALVL(I2)/3.6 IF(I1.EQ.0) FGROS =FGROS*1.6 PHL(I2,I1+2) =FGROS*DLA/3.6 RTDIF =RTDIF *EXP(-KDIF *(ALVL(I2)+ALVDL(I2))) RTDIRT =RTDIRT*EXP(-KDIRT *(ALVL(I2)+ALVDL(I2))) RTDIRD =RTDIRD*EXP(-KDIRBL*(ALVL(12)+ALVDL(I2))) FSSL =FSSL *EXP(-KDIRBL*(ALVL(I2)+ALVDL(I2))) CONTINUE CONTINUE DO 70 I2-l,IN PCGCL(I2)-PHL(I2,1)+PHL(I2,2)+PHL(I2,3) CONTINUE PCGC =PCGCL(l)+PCGCL(2)+PCGCL(3)+PCGCL(4)+PCGCL(5) RETURN END

SUBROUTINE SUSAWA(SWICHl,WCLQT,WL0QT,NL,TRWL,EVSC, $ RAIN,ZW,TKL,TYL,DELT,DTMIN,DTMX1,DTFX, $ WL0MX,ZEQT,CSA,CSB,CSCZ,WCLCH,WL0CH, $ WCLEQI,EVSW,RUNOF,DRSL,WCUMCH,ZECH,ZLT) C calculates the soil water balance for 24 h. in L2SS. 04/88 IMPLICIT REAL (A-Z) INTEGER SWICHl,SWICH2,SWICH3,SWICH4,SWICH5 INTEGER NL,I,INXSAT,JJTOT,JTOT,J COMMON /SLDPTH/ ZL(10) COMMON /VOLWAT/ WCAD(10),WCFC(10),WCST(10),WCWP(l0) COMMON /HYDCON/ KMSMX(10),KMSA1(10),KMSA2(10),KST(10) COMMON /PFCURV/ MSWCA(10) DIMENSION INXSAT(10,10),JJTOT(10),WCL(10),TKL(10),MS(10) DIMENSION FLXSTT(11),FLXUNT(11),FLX(11),FLXINT(11) DIMENSION HGT(10),HGB(10),HPP(11),DHH(10),TRWL(10),TYL(10) DIMENSION WCLQT(10),WCLCH(10),WCLEQI(10) DATA TINY1,TINY2,SWICH3,SWICH4,SWICH5/1.E-3,1.E-4,1,1,2/ DO 10 I=1,NL IF(SWICH1.EQ.1) ZL(I)=0. WCLCH(1) =0. WCL(1) =WCLQT(I)

260

10

CONTINUE WLO =WLOQT ZE =ZEQT*100. EVSW =0. DRSL =0. RUNOF =0. CALL SUCONV(l,TKL,ZL,ZLT,RAIN,ZW,WL0,WL0MX,RUNOF,TRWL,EVSW, $

20

30

40

50

EVSC,DRSL,NL) IF(SWICHl.EQ.1) THEN CALL SUGRHD(TKL,NL,HGT,HGB) CALL SUSLIN(TYL,NL,TKL,ZW,ZLT,WCL) DO 20 I=1,NL WCLEQI(I)=WCL(I) CONTINUE ELSE TIMTOT=0 . DO 30 I=1,NL WCL(I) =WCL(I)-DELT*TRWL(I)/TKL(I) FLXINT(I)=0. CONTINUE FLXINT(NL+l)=0. DTMX2 =1. WLO =WLO+RAIN*DELT HPBP =ZLT-ZW IF(WL0.GE.EVSC*DELT) THEN EVSW =EVSC EVSWX =0. WLO =WL0-EVSC*DELT ELSEIF(WL0.GT.0.) THEN EVSWl =WLO/DELT EVSW2 =AMINl(EVSC-EVSWl,(WCL(l)-WCAD(l))*TKL(l)/DELT) EVSW =EVSWl+EVSWZ EVSWX =EVSW2 WCL(l)=WCL(l)-EVSWZ*DELT/TKL(l) WLO =0. ELSE EVSWl =EVSC EVSW2 =(WCL(l)-WCAD(l))*TKL(l)/DELT EVSW3 =CSCZ/(ZE+TINYl) EVSW =AMINl(EVSWl,EVSW2,EVSW3) EVSWX =EVSW WCL(l)=WCL(l)-EVSW*DELT/TKL(l) ENDIF JTOT =0 DO 50 J=1,NL JJTOT(J) =0 CONTINUE CALL SUSTCH(WCL,NL,INXSAT,JTOT,JJTOT) IF(JTOT.NE.0) THEN CALL SUSTHH(INXSAT,JTOT,JJTOT,NL,HGT,HGB,WL0,WL0MX,HPBP, HPP, DHH) $ CALL SUSTFL(NL,INXSAT,JTOT,JJTOT,TKL,DHH,FLXSTT,FLXSQl,FLXSQZ) ELSE CONTINUE ENDIF IF(JTOT.EQ.l.AND.JJTOT(l).EQ.NL) THEN FLXlRTT(l)-0. ELSE

261

CALL SUUNST(SWICH3,SWICH4,WCL,TKL,NL,WL0,WL0MX,HPBP, MS , FLXUNT) ENDIF CALL SUSEFL(INXSAT,NL,JTOT,JJTOT,FLXSTT,FLXUNT,WL0,WL0MX, $ FLXSQl,FLXSQ2,FLX) CALL SUINTG(HPBP,SWICH4,SWICH5,HPP,FLX,TKL,WCL,MS,DTMIN,DTMXl, DTMX2,DTFX(,INXSAT,DHH,JTOT,FLXSQ2,WL0,WL0MX,NL,DELT,DT) $ DO 60 I=l,NL+l FLXINT(1)=FLXINT(I)+DT*FLX(I) CONTINUE TIMTOT=TIMTOT+DT DTMX2 =DELT-TIMTOT IF(DTMX2.GT.TINY2) GOTO 40 IF(WL0.GT.WL0MX) THEN RUNOF =WL0 - WL0MX WL0 =WL0MX ENDI F DRSL =FLXINT(NL+l)/DELT CALL SUZECA(WCLQT(l),EVSC,RAIN,-EVSWX,FLXINT(2), $ WL0,DELT,ZE,CSA,CSB,CSC2) ENDIF CALL SUCONV(2,TKL,ZL,ZLT,RAIN,ZW,WL0,WL0MX,RUNOF,TRWL,EVSW, $ EVSC,DRSL,NL) ZECH =ZE/l00.-ZEQT WL0CH =WL0-WL0QT WCUMCH=0 . DO 70 I=1,NL WCLCH(I)=WCL(I)-WCLQT(1) WCUMCH =WCUMCH+WCLCH(I)*TKL(I) CONTINUE RETURN END $

60

70

SUBROUTINE SUSEFL (INXSAT,NL,JTOT,JJTOT,FLXSTT,FLXUNT, WL0,WL0MX,FLXSQl,FLXSQZ,FLX) $ C selects between saturated and unsaturated fluxes; in L2SS; 04/88 IMPLICIT REAL (A-Z) INTEGER INXSAT,JJTOT,JTOT,J,JJ,I,IIN,IOUT,NL DIMENSION INXSAT(10,10),JJTOT(10),FIXSTT(11),FLXUNT(11),FLX(11) DATA TINY/0.001/ HPTP-AMINl(WLO,WLOMX) DO 10 I=l,NL+1 FLX(1) =FLXUNT(I) 10 CONTINUE DO 70 J=1,JTOT IF(FLXSTT(J).GT.TINY) THEN IOUT =INXSAT(J,JJTOT(J))+l IIN =INXSAT(J,l) FLX(IOUT)=AMINl(FLXSTT(J),FLXUNT(IOUT)) IF(IOUT.EQ.NL+l) FLX(IOUT)=FLXSTT(J) IF (JJTOT(J).GT.l) THEN DO 20 JJ=2,JJTOT(J) FLX(INXSAT(J,JJ))=FLX(I0UT) 20 CONTINUE ENDIF FIX(IIN)=AMIN1(FLX(IOUT),FLXUNT(IIN)) IF(IIN.EQ.1.AND.HPTP.GT.TINY) FLX(IIN)=FLX(IOUT) ELSEIF(FLXSTT(J).LT.-TINY) THEN

262

30

40

50

60

70

80

IOUT =INXSAT (J ,1) IIN =INXSAT(J,JJTOT(J))+l IF(FJXSQ2.LE.0.) THEN FLX(I0UT)=FLXSTT(J) ELSE THEN IF(FLXUNT(IOUT).LT.-TINY) FLX(IOUT)=FLXUNT(I0UT) ELSEIF(FLXUNT(IOUT).GT.TINY) THEN FLX(IOUT)=AMAXl(FLXSQ2,FLXUNT(IOUT)) ELSE FLX(IOUT)=0. ENDI F ENDIF IF(IOUT.EQ.1) FLX(IOUT)=FLXSTT(J) IF(JJTOT(J).GT.l) THEN DO 30 JJ=2,JJTOT(J) FLX(INXSAT(J,JJ))=AMAXl(FLX(IOUT),FLXSTT(J)) CONTINUE ENDIF FLX(IIN)=AMAXl(FLX(IOUT),FLXSTT(J)) ELSE IIN =INXSAT(J, 1) IOUT =NL+1 IF(FLXUNT(IIN).LE.-TINY) THEN FLX(IIN)=FLXUNT(IIN) IF(JJTOT(J).GT.l) THEN DO 40 JJ=2,JJTOT(J) FLX(INXSAT(J,JJ))=AMAXl(FLXSQl,FLX(IIN)) CONTINUE ENDIF FLX(IOUT)=AMAXl(FLXSQl,FLX(IIN)) ELSEIF(FLXUNT(IIN).GE.TINY) THEN FLX(IIN)=AMINl(FLXUNT(IIN),FLXSQ2) IF(JJTOT(J).GT.l) THEN DO 50 JJ=2 ,JJTOT(J) FLX(INXSAT(J,JJ))=FLX(IIN) CONTINUE ENDIF FLX(IOUT)=FLX(IIN) ELSE FLX(IIN)=0. IF(JJTOT(J).GT.l) THEN DO 60 JJ-2,JJTOT(J) FLX(INXSAT(J,JJ))=0. CONTINUE ENDIF FLX(IOUT)=0. ENDIF ENDIF CONTINUE DO 80 I=l,NL+l FLXUNT( I)=0. FLXSTT(I)=0. CONTINUE RETURN END SUBROUTINE

SUSLIN (TYL,NL,TKL,ZW,ZLT,WCL)

263

C derives soil parameters. in L2SS. 03/88 IMPLICIT REAL (A-Z) INTEGER NL,I,IX COMMON /VOLWAT/ WCAD(10),WCFC(10),WCST(10),WCWP(10) COMMON /SLDPTH/ ZL(10) DIMENSION TYL(l0),TKL(l0),WCL(l0) DATA TINY/l.E-l0/ WRITE(6,20) DO 10 I=1,NL CALL SUERRM(6.1,TYL(I),1.,20.,6.) WRITE(6,30)I,TYL(I),TKL(I)*.01,WCAD(I),WCWP(I),WCFC(I),WCST(I) IF( I. EQ. 1) THEN ZL(I)=0. ELSE ZL(I)=ZL(I-l)+TKL(I-l) ENDIF MS =AMAX1(0.,ZW-ZL(I)-0.5*TKL(I)) CALL SWCMS(I,2,WCL(I),MS) CALL SUERRM(6.2,WCL(I),WCAD(I),WCST(I),6.) 10 CONTINUE ZLT =ZL(NL)+TKL(NL) 20 FORMAT(' SOIL CHARACTERISTICS PER COMPARTMENT: ’,/, $' COMPARTMENT TYPE NR TKL WCAD WCWP WCFC WCST') 30 FORMAT(3X,I4,8X,F5.1,3X,F5.3,3X,4(F5.4,3X)) RETURN END SUBROUTINE SUSTCH (WCL,NL,INXSAT,JTOT,JJTOT) C checks for presence of saturated layers. in L2SS. 9/87. IMPLICIT REAL (A-Z) INTEGER NL,INXSAT,I,J,JTOT,JJTOT,JJ COMMON /VOLWAT/ DUMMY1(20),WCST(10),DUMMY(10) DIMENSION WCL(lO),INXSAT(10,10),JJTOT(10) DATA TINY/0.001/ J =0 JJ =0 DO 10 I=1,NL IF(ABS(WCL(I)-WCST(I)).LT.TINY) THEN JJ =JJ+1 IF(JJ.NE.l) THEN J =J+0 ELSE J =J+1 JTOT =J ENDIF INXSAT(J,JJ)=I JJTOT(J)=JJ ELSE JJ =0 ENDIF 10 CONTINUE RETURN END SUBROUTINE SUSTFL(NL,INXSAT,JTOT,JJTOT,TKL,DHH, $ FLXSTT,FLXSQl,FLXSQ2) C calculates tentative saturated fluxes. in L2SS. 04/88 IMPLICIT REAL (A-2)

264

10 20

30

40

50

60

70 80

90

100

INTEGER I,J,N,NL,JTOT,JJTOT,INXSAT,INDX,IA,JA COMMON /HYDCON/ DUMMY(30),KST(10) DIMENSION A(11,11),TKL(10),INXSAT(10,10),JJTOT(10),B(11) DIMENSION DHH(10),FLXSTT(11),DIS(10),INDX(11),KS(10) DATA TINY/l.0E-l0/,LARGE/-l00./,DUMMY2/0./,DUMMY1/0./ DO 190 J=1,JTOT DO 20 IA=1 ,NL+1 DO 10 JA=l,NL+1 A(IA,JA)=0. B(JA)=0. CONTINUE CONTINUE IF(ABS(DHH(J)).LT.TINY) THEN FLXSTT(J)=0. ELSE DO 30 IA=l,JJTOT(J) I=INXSAT(J,l)+IA-l KS(IA)=KST(I) DIS(IA)=TKL(I) CONTINUE DO 40 IA=l,JJTOT(J) A(IA,JJTOT(J)+l)=-l. JA-IA A(IA,JA)=-KS(IA)/DIs(IA) CONTINUE DO 50 JA=l,JJTOT(J) A(JJTOT(J)+l,JA)=+l. CONTINUE A(JJTOT(J)+l,JJTOT(J)+l)=0. DO 60 JA=l,JJTOT(J) B(JA)=0. CONTINUE B(JJTOT(J)+l)=DHH(J) N=JJTOT(J)+l CALL SUSTMD(A,N,INDX,D) CALL SUSTMS(A,N,INDX,B) FLXSTT(J)=B(N) ENDIF IF(DHH(J).GT.-TINY) THEN DO 80 IA=l,NL+l DO 70 JA=l,NL+l A(IA,JA)=0. B(JA)=0. CONTINUE CONTINUE IF(JJTOT(J).EQ.l) THEN FLXSQ1=LARGE ELSE DO 90 IA=I,JJTOT(J)-l I=INXSAT(J,2)+IA-1 KS(IA)=KST(I) DIS(IA)=TKL(I) CONTINUE DO 100 IA=1, JJTOT(J) - 1 A(IA,JJTOT(J))=-l. JA=IA A(IA,JA)=-KS(IA)/DIS(IA) CONTINUE

265

110

120

130 140

150

160

170

180

190

DO 110 JA=l,JJTOT(J)-l A(JJTOT(J) ,JA)=+l. CONTINUE A(JJTOT(J) ,JJTOT(J))=0. DO 120 JA=l,JJTOT(J)-l B(JA)=0. CONTINUE B(JJTOT(J))=DHH(J)+TKL(INXSAT(J,l)) N=JJTOT(J) CALL SUSTMD(A,N,INDX,D) CALL SUSTMS(A,N,INDX,B) FLXSQ1=B(N) ENDIF DO 140 IA=l,NL+l DO 130 JA=l,NL+l A(IA,JA)=0. B(JA)=0. CONTINUE CONTINUE IF(INXSAT(J,l).EQ.l) THEN FLXSQ2=DUMMY2 ELSE DO 150 IA=l,JJTOT(J)+l I=(INXSAT(J,l)-l)+IA-l KS(IA)=KST(I) DIS(IA)=TKL(I) CONTINUE DO 160 IA=l,JJTOT(J)+l A(IA,JJTOT(J)+2)=-1. JA=IA A(IA,JA)=-KS(IA)/DIS(IA) CONTINUE DO 170 JA=l,JJTOT(J)+l A(JJTOT(J)+2,JA)=+l. CONTINUE A(JJTOT(J)+2,JJTOT(J)+2)=0. DO 180 JA=l,JJTOT(J)+l B(JA)=0. CONTINUE B(JJTOT(J)+2)=DHH(J)-TKL(INXSAT(J,l)-l) N=JJTOT(J)+2 CALL SUSTMD(A,N,INDX,D) CALL SUSTMS(A,N,INDX,B) FLXSQ2=B(N) ENDIF ENDIF CONTINUE RETURN END

SUBROUTINE SUSTHH(INXSAT,JTOT,JJTOT,NL,HGT,HGB, WL0,WL0MX,HPBP,HPP,DHH) $ C identifies hydraulic head across saturated layers. in L2SS. 04/88 IMPLICIT REAL (A-Z) INTEGER I,NL,INXSAT,J,JTOT,JJTOT DIMENSION INXSAT(10,10),JJTOT(10),HHT(10),HHB(10),HGT(10),HGB(10) DIMENSION HPP(ll),DHH(lO) DATA TINY/l.E-5/

266

10

20

HPTP =AMINl(WL0,WL0MX) DO 10 I=1,ll HPP(1)=0. CONTINUE DO 20 J=1,JTOT EXTRAT=0. IF(INXSAT(J,l).EQ.l) EXTRAT=HPTP EXTRAB=0. IF(INXSAT(J,JJTOT(J)).EQ.NL) EXTRAB=HPBP HHT(J)=EXTRAT+HGT(INXSAT(J,l)) HHB(J)=EXTRAB+HGB(INXSAT(J,JJTOT(J))) DHH(J)=HHB(J)-HHT(J) IF(DHH(J).GT.TINY) HPP(INXSAT(J,l))=DHH(J) IF(ABS(DHH(J)).LT.TINY) DHH(J)=0. CONTINUE RETURN END

SUBROUTINE SUSTMD (A,N,INDX,D) C decomposes matrix A. in LZSS. (PRESS etal, 1986.) IMPLICIT REAL (A-Z) INTEGER I,J,K,N,IMAX,INDX DIMENSION A(ll,ll),INDX(ll),W(ll) DATA TINY/l.0E-l0/ D=1. DO 12 I=l,N AAMAX=0. DO 11 J=l,N IF(ABS(A(I,J)).GT.AAMAX) AAMAX=ABS(A(1,J)) 11 CONTINUE VV(I)=l./AAMAX 12 CONTINUE DO 19 J=l,N IF(J.GT.l) THEN DO 14 I=1,J-1 SUM=A(1.J) IF(I.GT.l) THEN DO 13 K=1,I-1 SUM=SUM-A(I,K)*A(K,J) 13 CONTINUE A(1,J)=SUM ENDIF 14 CONTINUE ENDIF AAMAX=0 . DO 16 I=J,N SUM=A(1,J) IF(J.GT.l.) THEN DO 15 K=1,J-1 SUM=SUM-A(I,K)*A(K,J) 15 CONTINUE A(1,J)=SUM ENDIF DUM=VV(I)*ABS(SUM) IF (DUM.GE.AAMAX) THEN IMAX=I AAMAX=DUM ENDIF

267

16

17

18 19

CONTINUE IF(J.NE.IMAX) THEN DO 17 K=l,N DUM=A(IMAX,K) A(IMAX,K)=A(J,K) A(J,K)=DUM CONTINUE D=-D VV(IMAX)=VV(J) ENDIF INDX(J)=IMAX IF(J.NE.N) THEN IF(ABS(A(J,J)).LT.TINY) A(J,J)=TINY DUM=l./A(J,J) DO 18 I=J+l,N A(1,J)=A(I,J)*DUM CONTINUE ENDIF CONTINUE IF(ABS(A(N,N)).LT.TINY) A(N,N)=TINY RETURN END

SUBROUTINE SUSTMS (A,N,INDX,B) C solves a set of linear equations. in L2SS. (PRESS et al, 1986) IMPLICIT REAL (A-Z) INTEGER I,J,II,LL,N,INDX DIMENSION A(ll,ll),INDX(ll),B(ll) II=0 DO 12 I=l,N LL=INDX(I) SUM=B (LL) B(LL)=B(I) IF(II.NE.0) THEN DO 11 J=II,I-1 SUM=SUM-A(I,J)*B(J) 11 CONTINUE ELSE IF (SUM.NE.0.) THEN II=I ENDIF B(I)=SUM 12 CONTINUE DO 14 I=N,1,-1 SUM=B(I) IF(I.LT.N) THEN DO 13 J=I+l,N SUM=SUM-A(I,J)*B(J) 13 CONTINUE ENDIF B(1)=SUM/A(I,I) 14 CONTINUE RETURN END SUBROUTINE SUUNST (SWICH3,SWICH4,WCL,TKL,NL,WL0,WL0MX, $ HPBP,MS,FLXUNT) C calculates tentative fluxes of unsaturated layers; in L2SS; 04/88 IMPLICIT REAL (A-Z)

268

INTEGER NL, I, SWICH3, SWICH4 COMMON /HYDCON/ DUMMY1(30),KST(10) COMMON /VOLWAT/ WCAD(10),DUMMY2(10),WGST(10),DUMMY3(10) DIMENSION WCL(l0),TKL(l0),DIS(ll),FLXUNT(11),MS(10) DIMENSION KMS(l0),MFLP(l0),MFLPQT(l0) DATA TINY1,TINY2/0.001,0.001/ HPTP =AMINl(WL0,WL0MX) IF(WCST(1)-WCL(l).GT.TINYl) THEN CALL SWCMS(l,SWICH4,WCL(l),MS(l)) CALL SUMFLP(SWICH3,1,MS(l),MFLP(l)) DIS(1)=0.5*TKL(l) IF(HPTP.GT.TINY1) THEN DZDH =DIS(l)/(-MS(1)-HPTP) DMFLP =MFLP(l)-0.-KST(l)*HPTP FLXUNT(l)=(DZDH-1.)*DMFLP/DIS(l) ELSE FLXUNT(l)=0. ENDIF ELSE FLXUNT(l)=0. END1 F DO 10 I=2 ,NL IF(ABS(WCL(I)-WCST(I)).LT.TINY2) THEN IF(ABS(WCL(I-l)-WCST(I-l)).GT.TINY2) THEN DIS(I)=0.5*TKL(I-l) DZDH =DIS(I)/(0.+MS(I-1)) DMFLP =0.-MFLP(I-l) FLXUNT(1)=(DZDH-l.)*DMFLP/DIS(I) ELSE CONTINUE ENDIF ELSE CALL SUWCMS(I,SWICH4,WCL(I),MS(I)) CALL SUMFLP(SWICH3,1,MS(I),MFLP(I)) IF(ABS(WCL(I-l)-WCST(I-l)).LT.TINY2) THEN DIS(I)=O.5*TKL(I) DZDH =DIS(I)/(-MS(I)-0.) DMFLP =MFLP(I)-0. FLXUNT(1)=(DZDH-l.)*DMFLP/DIS(I) ELSE CALL SUMFLP(SWICH3,I-1,MS(I),MFLPQT(I)) CALL SUMFLP(SWICH3,I,MS(I-l),MFLPQT(I-l)) DIS(I)=0.5*(TKL(I)+TKL(I-l)) DMFLP1=MFLP(I)-MFLPQT(I-I) IF((MS(I)-MS(I-1))*(MFLP(I)-MFLPQT(I-1)).GT.-TINY1) THEN DMFLP1=0. ENDIF DMFLP2=MFLPQT(I)-MFLP(I-l) IF((MS(I)-MS(I-1))*(MFLPQT(I)-MFLP(I-1)).GT.-TINY1) THEN DMFLP2=0. ENDIF IF(ABS(DMFLPl).LT.TINY2.0R.ABS(DMFLP2).LT.TINY2) THEN DMFLP -0. ELSE SIGN =DMFLPl/ABS(DMFLPl) DMFLP =SIGN*SQRT(DMFLPl*DMFLP2) ENDIF IF(MS(I).NE.MS(I-1)) THEN

269

10

DZDH =DIS(I)/(-MS(I)+MS(I-1)) FLXUNT(I)=(DZDH-l.)*DMFLP/DIS(I) ELSE CALL SUMSKM(SWICH3,I,MS(I),KMS(I)) CALL SUMSKM(SWICH3,I-l,MS(I-l),KMS(I-l)) KAV =SQRT(KMS(I)*KMS(I-1)) FLXUNT(I)=+KAV ENDIF ENDIF ENDIF CONTINUE IF(ABS(WCL(NL)-WCST(NL)).LT,TINY2) THEN CONTINUE ELSE DIS(NL+l)=0.5*TKL(NL) CALL SUMFLP(SWICH3,NL,MS(NL),MFLP(NL)) IF(HPBP.GT.TINY1) THEN DZDH =DIS(NL+l)/(HPBP+MS(NL)) DMFLP =0.-MFLP(NL)+KST(NL)*(HPBP-0.) FLXUNT(NL+l)=(DZDH-l.)*DMFLP/DIS(NL+l) ELSE MSB =-HPBP CALL SUMFLP(SWICH3,NL,MSB,MFLPB) IF(HPBP.NE.-MS(NL)) THEN DZDH =DIS(NL+l)/(HPBP+MS(NL)) DMFLP =MFLPB-MFLP(NL) IF((MSB-MS(NL))*(MFLPB-MFLP(NL)).GT.-TINY1) DMFLP =0. FLXUNT(NL+1)=(DZDH-l.)*DMFLP/DIS(NL+l) ELSE CALL SUMSKM(SWICH3,NL,MS(NL),KMS(NL)) FLXUNT(NL+l)=+KMS(NL) ENDIF ENDIF ENDIF RETURN END

SUBROUTINE SUWCMS (I,SWICH4,WCL,MS) C relates volumetric water content and suction; in L2SS, 03/87 IMPLICIT REAL (A-Z) INTEGER I,SWICH4 COMMON /VOLWAT/ WCAD(10),DUMMY1(10),WCST(10),DUMMY2(10) COMMON /PFCURV/ MSWCA(10) DATA TINY/0.001/ IF(SWICH4.EQ.l) THEN CALL SUERRM(7.1,WCL,WCAD(I),WCST(I),6.) MS =EXP(SQRT(-ALOG(AMAX1(WCAD(I),WCL)/WCST(I))/MSWCA(I)))-1. ELSE CALL SUERRM(7.2,MS,O.,l.E8,6.) WCL =AMAX1(TINY,WCST(I)*EXP(-MSWCA(I)*((ALOG(MS+1.))**2))) ENDIF RETURN END SUBROUTINE SUZECA (WCLQT1,EVSC,RAIN,FLX1,FLX2,WL0, $ DELT,ZE,AEXP,BEXP,C2) C calculates the depth of the evaporation front; in L2SS; 04/88 IMPLICIT REAL (A-Z)

270

COMMON /VOLWAT/ WCAD(10),DUMMY1(10),WCST(10),DUMMY2(10) DATA TINYl,LARCE/l.E-3,10./ IF(RAIN.GT.EVSC) THEN ZE =0. ELSEIF(FLX2.LT.-TINYl.AND.FLX2.LT.FIXl) THEN ZE =ZE-(FLXl-FLXZ)/(0.5*WCST(1)) IF(ZE.LE.TINY1) THEN ZE =0. ENDIF ELSEIF(RAIN.CT.0.05) THEN CONTINUE ELSE WI =WCLQTl/WCST(l) WTH =WI-WCAD(l)/WCST(l) IF(WTH.LE.TINY1) THEN ZE =LARGE ELSEIF(WL0.GT.TINYl.OR.ABS(WCLQT1-WCST(1)).LT.TINY1) THEN ZE =0. ELSE C1 =AEXP*(EXP(AMAX1(0.,(W1-0.5)*BEXP))-l.) C3 =C2/(WTH+C1/C2) ZE =SQRT(ZE*ZE+2.*C3*DELT) ENDIF ENDIF RETURN END ENDJOB

271