The Evolution of the System of Rice Intensification as a Socio-technical Phenomenon:

A Report to the Bill & Melinda Gates Foundation

February 2011

Wageningen University and Research Centre Technology and Agrarian Development Group Development Economics Group

This report was written by Dr. Ezra Berkhout and Dr. Dominic Glover with the guidance of Prof. Herman van Keulen, Dr. Rob Schipper and Dr. Harro Maat.

Contact information: Ezra Berkhout, Development Economics Group, Dominic Glover, Technology and Agrarian Development Group, Social Sciences Department Wageningen University and Research Centre PO Box 8130 6700 EW Wageningen The Netherlands

Email addresses: ezra.berkhout wur.nl dominic.glover wur.nl

Suggested citation: Berkhout, E. and Glover, D. (2011) The Evolution of the System of Rice Intensification as a socio-technical phenomenon: A report to the Bill & Melinda Gates Foundation, Wageningen, NL: Wageningen University and Research Centre.

2

Acknowledgements and thanks Many individuals and organizations helped us to carry out our work and we appreciate their generous support in time, information, advice and resources. In particular, we would like to offer special thanks to the many farmers, agricultural labourers and extensionists in Madagascar and Tamil Nadu who so generously welcomed us and shared their knowledge and opinions on rice farming and SRI. We also express our appreciation to all of the participants in our workshops in Antananarivo and Hosur, whose names are listed in Appendices A and B. A large number of other people provided advice and guidance, organizational and logistical assistance, or otherwise extended their support to our project in various intangible but important ways. We would like to thank the following people in particular: Rames Abhukara, Bruno Andrianaivo, Hein ten Berge, David Bergvinson, Prem Bindraban, Bas Bouman, P. M. Boopathi, Erwin Bulte, Ollie Burch, Robert Chambers, S. Chellamuthu, Toon Defoer, John Duxbury, Lucy Fisher, Winifred Fitzgerald, Philippe Grandjean, Biksham Gujja, Peter Hofs, Joelibarison, Marian Jonker, Bhaskar Joshi, Herman van Keulen, Murali Krishnamurthy, Tim Krupnik, Thom Kuyper, Julie Lauren, Melissa Leach, Cees Leeuwis, Ingrid Lefeber, Glenn Lines, Liu Xiaona, Michael Loevinsohn, Harro Maat, R. Mahender Kumar, Baharul I. Majumder, Alfred Mokuwa, G. Nammalwar, Florent Okry, B. J. Pandian, C. Shambu Prasad, Jürgen Pütz, Raj Rajendran, Jacqueline Rakotoarisoa, Irène Rakotoniaina, Justin Léonard Rabenandrasana, Myriam Ramahandridona, Danièle Ramiaramanana, Lucile Ramilison, Tsimba Randriamiarintsoa, Andry Randrianarivelo, Fetrasoa Ernesto Ratovoarimanana, Aimé Lala Razafinjara, Paul Richards, Mr. Rivo, Inge Ruisch, T. Sampathkumar, Rob Schipper, Ian Scoones, Michel Siméon, Pushpalatha Sivasubramanian, Siva Sivasubramanian, Martin Smith, Mr. Solo, Willem Stoop, Paul Struik, Jim Sumberg, M. S. Swaminathan, Béla Teeken, Janice Thies, T. M. Thiyagarajan, John Thompson, Norman Uphoff, Rajendra Uprety, Olivia Vent, S. Vijayabaskaran, Zéphyrin Zanarison and Zhu Defeng. Last but certainly not least, we extend our thanks to the Bill & Melinda Gates Foundation for providing a grant to fund this project.

Ezra Berkhout, Dominic Glover Wageningen, NL February 2011

3 Executive Summary

This document reports the findings and conclusions of a short exploratory study into the emergence, spread and impacts of the System of Rice Intensification (SRI). SRI is a crop management system that has been depicted as a more productive and more ecologically sustainable method for cultivating rice. It is said to be particularly appropriate, accessible and beneficial for poor and marginal farmers because it can achieve substantial increases in productivity and grain yield without the need for improved seeds or chemical inputs. It has also been reported that SRI methods can produce much higher yields while consuming much less water. These claims have provoked controversy among some scientists. With the goal of moving the debates about SRI forward, this research project was designed to evaluate the current state of knowledge about the origins, spread and impacts of SRI, to examine the mechanisms and processes that have helped to spread the system internationally, and to identify key knowledge gaps. The research approach was designed to examine the SRI phenomenon from technical, socio- economic and institutional perspectives. The project involved a review of academic literature and other documents on SRI, and two short field visits to Madagasar and Tamil Nadu, India, to learn about SRI at field level. SRI consists of a suite of practices, often summarized as six in number, involving particular procedures for crop establishment, irrigation management, weed control and fertilization. Several of these procedures deviate from currently recommended practices for rice cultivation. They are said to be based on the innate physiological characteristics of rice, though SRI advocates stress that the system should be adapted to suit specific local rice varieties and agro-ecological settings. The project has made several useful advances. First, the report shows that SRI was shaped not only by close observation of rice plants, but also by the resource- constrained setting of Madagascar during the 1970s and 1980s, where the system was originally developed. In particular, it was designed to suit the capacities of poor and marginal Malagasy farmers and their agro-ecological and institutional context. SRI is said to have been discovered or invented by a French Jesuit priest, Father Henri de Laulanié, but in fact it was compiled by him from various existing sources and each of the individual elements has precedents in rice cultivation practices from different times and places. This indicates, on one hand, that SRI is not as original or radically novel as it has been portrayed, and on the other hand that it rests on a reasonably firm foundation of knowledge and practice in rice cultivation. Second, our review of literature on the biophysical mechanisms involved in SRI and studies on the adoption and impacts of the system has made progress identifying areas of knowledge that are fairly well established and the many areas where gaps in knowledge remain. It is fairly well established that SRI cultivation methods cause changes in the physiology and morphology of rice plants, which can lead to improved productivity and grain yield under favourable conditions. Analyses of the

4 contributions made by individual SRI practices produce a mixed picture, but it seems fairly well established that transplanting younger rice seedlings with optimal spacing and the use of water-saving irrigation techniques can combine to produce a good yield and increase the productivity of seeds and water inputs. Substantially higher yields have also been reported, but it is not clear whether these are attributable to increased fertilizer or labour inputs, more skilful farmers, better soil fertility on plots chosen for SRI management, or synergetic interactions among the SRI practices. Third, there is a dearth of detailed and reliable information on the international spread of SRI. A handful of small-scale adoption studies have observed that relatively few farmers adopt all the practices of SRI and that adopters often do not practice the SRI techniques on all of their rice plots. There are indications that wealthier farmers are more likely to be (early) adopters of SRI methods and that SRI methods may be allocated preferentially to the most fertile and well- irrigated plots. Some disadoption has also been observed. Fourth, it is clear that the adoption of SRI methods is associated with quite substantial changes in allocations of inputs, especially labour, water and fertilizer. These include changes to the temporal distribution of labour demand and the gender division of labour. The available literature allows few firm conclusions to be drawn regarding the impacts of these changes, which theoretically may be positive or negative for different households or groups. The nature of operations such as weeding may also change with SRI adoption, but these effects have not been studied. Finally, substantial variations have been found in the ways that SRI is specified in different locations and in the scientific literature compared to grey literature. Further research will be required to make sense of these patterns, but they suggest that SRI methods change as they move from one setting to another, and that there may be a difference between the ways scientists, on one hand, and SRI promoters and extensionists, on the other, approach rice intensification issues and problems. The international spread of SRI has been catalysed by charismatic individuals at international and national levels. The range of actors involved in SRI today is wide, encompassing large and small non-governmental organizations, government agencies, universities and research institutes, extension services, funding bodies, and other actors. Taking into account the origins of SRI and the mechanisms by which it has spread, we conclude that SRI is not merely a set of crop management principles but the product of a distinctive socio-technical system that has operated, at least partly, outside the mainstream circuits of international agricultural research. This means that SRI raises important questions about the connections between agricultural research and agricultural development, the mobilization of social and professional networks, the exploitation of scientific knowledge, communication networks, and learning processes. The report concludes by identifying a range of questions and proposing an integrated, interdisciplinary research approach for making further progress in the understanding of this important phenomenon.

5 Table of Contents

Executive Summary ...... 4 Table of Contents ...... 6 List of Abbreviations ...... 8 List of Tables ...... 11 List of Figures ...... 11

Part 1: Introduction and research approach ...... 12 1. Introduction ...... 13 1.1 Background to the present study ...... 14 1.2 Conceptual approach: Economics and the Sociology of Science and Technology ...... 15 1.3 Aims and goals ...... 17 1.4 Design of the project ...... 17 1.5 Structure of the report ...... 17 2. Implementation of the project ...... 19 2.1 Workshop in Brighton and visit to Cornell University ...... 19 2.2 Literature search and review ...... 20 2.2.1 Madagascar...... 20 2.2.2 India...... 21 2.2.3 China...... 21 2.3 Identification of actors and networks ...... 22 2.4 Field visits ...... 23 2.4.1 Madagascar ...... 23 2.4.2 Tamil Nadu...... 24 2.4.3 Nepal ...... 25 2.5 Project outputs ...... 25 2.5.1 This report...... 25 2.5.2 A database of documents related to SRI...... 25 2.5.3 A set of interactive timelines ...... 26 2.5.4 Journal articles...... 26 2.6 Flexibility in implementation ...... 27

Part 2: An overview of the SRI phenomenon ...... 29 3. Origins and theory of SRI ...... 30 3.1 The origins of SRI ...... 30 3.2 Elaboration and refinement of SRI ...... 33 3.3 A working definition of SRI methods ...... 35 3.4 The agronomic theory underlying SRI ...... 36 3.4.1 Healthy seedlings: nursery management, seedling age and transplanting ...... 37 3.4.2 Optimal plant density: number of seedlings per hill and spacing of hills ...... 38 3.4.3 Soil aeration: promoting healthy roots and soil microbial activity ...... 38 3.4.4 Synergies ...... 39 4. An overview of agronomic issues in SRI cultivation ...... 41 4.1 The ‘Rice Wars’ ...... 42 4.2 General studies on SRI cultivation methods, physiology and morphology ...... 46 4.3 Crop establishment: Young seedlings, seedlings per hill, spacing distances ...... 48 4.4 Irrigation management ...... 51 4.5 Weed control, soil biology and soil nutrient management ...... 53 4.6 Synergies ...... 54 5. Adoption and impact studies on SRI ...... 56 5.1 Impressionistic indications of the spread of SRI ...... 56 5.2 Studies on SRI adoption and adoption processes ...... 61 5.2.1 What an SRI adopter adopts ...... 62

6 5.2.2 Who adopts SRI components, and why? ...... 63 5.2.3 Who disadopts, and why? ...... 65 5.2.4 Role of extension approaches and methods ...... 66 5.2.5 Summary ...... 67 5.3 Measuring impact: evidence from farm-level studies ...... 67 5.3.1 Mechanisms underlying impact...... 68 5.3.2 Observable and unobservable characteristics ...... 70 5.3.3 Characteristics of SRI impact studies ...... 71 5.3.4 Changes in input productivity ...... 76 5.3.5 Changes in marginal productivities and effects ...... 82 5.3.6 Changes in production variance/risk ...... 83 5.3.7 Summary ...... 85

Part 3: The spread, evolution and variation of SRI ...... 86 6. SRI ‘on the ground’: Insights from Madagascar, Tamil Nadu and Nepal ...... 87 6.1 Insights into SRI in Madagascar ...... 87 6.1.1 Historical overview ...... 87 6.1.2 Key observations from our field visit ...... 90 6.2 Insights into SRI in Tamil Nadu ...... 94 6.2.1 Historical overview ...... 94 6.2.2 Key observations from our field visit ...... 98 6.3 Impressions of SRI in Nepal ...... 104 6.3.1 Variations in SRI in three local contexts ...... 104 7. Mechanisms that shaped the development and spread of SRI ...... 106 7.1 SRI was shaped by the institutional context in which it was developed ...... 106 7.2 The key role of charismatic individuals ...... 108 7.3 Institutional positions may have influenced the scientific reception of SRI ...... 109 7.4 Networks and information flows ...... 110 8. Diversity in SRI in principle and practice ...... 114 8.1 Variation in SRI specifications ...... 116 8.2 Variations on a theme: SRI-like rice cultivation systems ...... 122 8.2.1 Rice cultivation systems inspired by SRI ...... 122 8.2.2 Independent contemporary and historical parallels to SRI methods ...... 125 8.3 Systems of Crop Intensification? ...... 128 9. Ways forward ...... 129 9.1 What we know and do not know about SRI ...... 129 9.1.1 Biophysical mechanisms, agronomic and crop management changes ...... 131 9.1.2 Spread and impact of SRI ...... 134 9.2 Proposals for further investigation ...... 137 9.2.1 Biophysical mechanisms ...... 137 9.2.2 Historical and contemporary analogues for SRI and/or SRI components ...... 138 9.2.3 The level of spread, patterns of adoption and impact of SRI methods ...... 138 9.2.4 Adoption processes and mechanisms leading to variation in SRI practice ...... 139 10. References ...... 141

Appendices and supplementary material ...... 155 Appendix A: Madagascar trip report ...... 156 Appendix B: Tamil Nadu trip report ...... 175 Appendix C: Timeline ...... 199 Appendix D: Abstracts of articles developed during the project ...... 201

7 List of Abbreviations

ADRA The Adventist Development Relief Agency, Silver Spring, MD, USA AME The Agriculture, Man, Environment Foundation, Bengaluru, Karnataka, India ANGRAU Acharya N. G. Ranga Agricultural University, Hyderabad, India AROPA Appui au Renforcement des Organisations Professionnelles et aux Services Agricoles , Support for the Reinforcement of Professional Organizations and Agricultural Services, Fianarantsoa, Madagascar ATS L’Association Tefy Saina , The Forge the Spirit Association, Madagascar AWD alternate wetting and drying irrigation BMGF The Bill and Melinda Gates Foundation, Seattle, WA, USA BMPs Best management practice(s) BUF The Better U Foundation, Los Angeles, CA, USA CEDAC Le Centre d'Étude et de Développement Agricole Cambodgien , The Cambodian Agricultural Study and Development Centre, Phnom Penh, Cambodia CIIFAD The Cornell International Institute for Food, Agriculture and Development, Ithaca, NY, USA CIRAD Le Centre de Coopération Internationale en Recherche Agronomique pour le Développement , The Centre for International Cooperation in Agronomic Research for Development, Montpellier, France CPR Le Centre de Promotion Rural , The Rural Promotion Centre, Ambositra, Madagascar ESSA L’Ecole Supérieure des Sciences Agronomiques , The Faculty of Agronomic Science, University of Antananarivo, Madagascar F3E Fonds pour la promotion des études préalables, études transversales, evaluations , Fund for the promotion of preliminary studies, transverse studies and evaluations, Paris, France FAO Food and Agricultural Organization of the United Nations, Rome, Italy FERT L’Organisation Professionnelle Agricole Française de Coopération Internationale pour le Développement Rural, The French Agricultural Professional Organization for International Cooperation for Rural Development, Paris, France FFS farmer field school

8 FOFIFA Foibe Fikarohana ampiharina amin’ny Fampandrosoana ny eny Ambanovohitra , The National Centre for Applied Research for Development, Antananarivo, Madagascar GCD Le Groupe Conseil Développement , The Development Advice Group, Antananarivo, Madagascar GSRI Le Groupement SRI , The SRI Group/Grouping, Antananarivo, Madagascar IAMWARM The Tamil Nadu Irrigated Agriculture Modernization and Waterbodies Restoration and Management Project ICRISAT International Crop Research Institute for the Semi-Arid Tropics, Patancheru, Hyderabad, Andhra Pradesh, India IDS The Institute of Development Studies at the University of Sussex, Brighton, UK IFAD The International Fund for Agricultural Development, Rome, Italy INGO international non-governmental organization IPE International Political Economy IRRI The International Rice Research Institute, Los Baños, the Philippines IWMI The International Water Management Institute, Colombo, Sri Lanka LAI leaf area index LEISA Low External Input Sustainable Agriculture MAFF Mitsitsy Ambioka sy Fomba Fiasa , ‘saving seeds and cultural practices’, a rice cultivation system derived from SRI (Madagascar) MAP Madagascar Action Plan NABARD National Bank for Agriculture and Rural Development, Mumbai, India NGO non-governmental organization NREGA the National Rural Employment Guarantee Act (India) PRADAN Professional Assistance for Development Action, Delhi, India PTD participatory technology development RMPs recommended management practices SRA Le Système de Riziculture Améliorée , the Improved System of Rice Cultivation (Madagascar) SRI The System of Rice Intensification / Le Système de Riziculture Intensive SRIA a rice cultivation system derived from SRI and SRA (Madagascar) SRT Le Système de Riziculture Traditionnelle , traditional rice cultivation methods in Madagascar

9 STS Science and Technology Studies or Science, Technology and Society Studies TNAU Tamil Nadu Agricultural University, Coimbatore, Tamil Nadu, India USAID The United States Agency for International Development, Washington, DC, USA WARDA The Africa Rice Centre, formerly the West African Rice Development Association, Cotonou, Benin, formerly based at Bouaké, Côte d’Ívoire WASSAN Watershed Support Services and Activities Network, Hyderabad, Andhra Pradesh, India WUR Wageningen University and Research Centre, Wageningen, the Netherlands WWF formerly known as the Worldwide Fund for Nature and the World Wildlife Fund XIMB Xavier Institute of Management, Bhubaneswar, Orissa, India

10

List of Tables

Table 5.1: Indicative figures on the international spread of SRI ...... 57

Table 5.2: Adoption studies ...... 61

Table 5.3: Studies on input productivities and technology effects ...... 72

Table 5.4: Reported changes in land productivity ...... 80

Table 6.1: Comparing agriculture and rice production between Tamil Nadu and Madagascar 102

Table 8.1: Documents providing a definition of SRI components ...... 116

Table 8.2: Specifications of SRI components ...... 118

Table 8.3: Specifications of seedling age ...... 118

Table 8.4: Specifications of seedling spacing distance ...... 118

Table 8.5: Definition of SRI compared between countries ...... 120

List of Figures

Figure 7.1: South–South exchanges in SRI 111

11

Part 1: Introduction and research approach

12

1. Introduction

The System of Rice Intensification (SRI) is a method of rice cultivation that has been claimed to greatly enhance the vegetative growth and grain yield of rice in the context of smallholder rice farming, while consuming substantially less water, without the necessity to adopt expensive modern technologies or external inputs. The system was first reported in a technical journal in 1993 by Father Henri de Laulanié, (de Laulanié 1993). De Laulanié was a French Jesuit missionary and trained agronomist, who had developed SRI while working with rice farmers in Madagascar over a period of several decades (Glover forthcoming; Rafaralahy 2002). The system came to the attention of development workers and academics working in Madagascar during the second half of the 1990s and was subsequently described and discussed in various publications (e.g. Association Tefy Saina & Uphoff 2001; Stoop et al. 2002; Uphoff 1999; 2002; Uphoff & Randriamiharisoa 2002). A conference held in Sanya, China in April 2002 drew further international attention to SRI (Uphoff et al. 2002). Since the late 1990s, SRI is reported to have spread from Madagascar to have some impact on rice cultivation practices in nearly 50 countries across Asia, Africa and South America. 1 Although reliable data on actual levels of adoption and practice of SRI methods is scarce (see Chapter 5), the reported levels of SRI activity in some locations nevertheless appear to be substantial. For example, the Indian state of Tamil Nadu is implementing a multi-million dollar, multi-year, World Bank-funded agricultural development and irrigation management programme that includes SRI promotion as a major component, potentially affecting the cultivation practices of many thousands of rice farmers (see Chapter 6). Also in India, WWF and the International Crop Research Institute for the Semi-Arid Tropics (ICRISAT, Hyderabad, India) are involved in a partnership to promote SRI in several Indian states (see Chapter 6). In Indonesia, a loose network of actors including the Asian Development Bank, the president, government ministries, universities, companies and non-governmental organizations (NGOs) has helped to stimulate a rather robust expansion of SRI activities, as reflected in various projects, programmes, conferences and workshops. 2 Other countries where substantial SRI activity has been reported include major rice-producers such as China, 3 the Philippines 4 and Vietnam. 5 Notwithstanding this rather rapid spread and the support of some heavyweight development organizations, the remarkable claims made on behalf of SRI provoked a

1 See http://sri.ciifad.cornell.edu/countries/index.html (accessed 19 January 2011). 2 See http://sri.ciifad.cornell.edu/countries/indonesia/index.html (accessed 19 January 2011). 3 See http://sri.ciifad.cornell.edu/countries/china/index.html (accessed 19 January 2011). 4 See http://sri.ciifad.cornell.edu/countries/philippines/index.html (accessed 19 January 2011). 5 See http://sri.ciifad.cornell.edu/countries/vietnam/index.html (accessed 19 January 2011).

13 lively and sometimes acrimonious debate among rice scientists (see Surridge 2004). In particular, certain claims about very high rice yields achieved using SRI methods were strongly disputed by scientists who argued that such productivity would be beyond the physiological yield potential of rice (Dobermann 2004; Sheehy et al. 2004). Other scholars argued that there was no consistent evidence that SRI methods could out-yield conventional best management practices, except perhaps in atypical contexts like the regions of Madagascar where the system was developed (McDonald et al. 2006). In turn, these critiques were themselves strongly contested by other scholars and defended by their authors (McDonald et al. 2008; Sheehy et al. 2005; Stoop & Kassam 2005; Uphoff et al. 2008). Some scientists dismissed SRI altogether, urging that no resources should be wasted investigating it (Sinclair & Cassman 2004). Attempts at dialogue between scholars on either side of the debate confirmed the lack of common ground between them (Sheehy et al. 2005; Sinclair 2004; Stoop & Kassam 2005; Uphoff 2004b). A modest amount of agronomic research has now been carried out to test the claims made on behalf of SRI and investigate the biophysical mechanisms that may underlie observed changes in the growth and yield of rice under SRI management (see Chapter 3). However, many important questions remain unanswered and it may be that the scientific controversy itself has had some inhibiting effect on research in this field. Many researchers from all sides of the debate have felt that high quality research was necessary to address key knowledge gaps and help to resolve the SRI controversy, in order to promote the interests of rice farmers and sustainable rice cultivation internationally. 1.1 Background to the present study During 2008, a proposal was developed to carry out a substantial international and interdisciplinary collaborative research project that would have attempted to achieve those goals. The project was designed to resolve key agronomic questions while also examining the socio-economic impacts and sociological dynamics of the SRI phenomenon. The project aimed to move ‘towards ecologically sustainable rice intensification’ (TESRI) by developing a new set of recommended principles or guidelines for rice cultivation that could command a wide consensus among agronomists and agricultural development experts. The TESRI project was designed to bridge institutional divides and would have integrated agronomic and social scientific expertise from Cornell University (Ithaca, NY, USA), the International Rice Research Institute (IRRI, Los Baños, the Philippines) and Wageningen University and Research Centre (WUR, Wageningen, the Netherlands). The request for financial support was submitted to the Bill & Melinda Gates Foundation (BMGF, Seattle, WA, USA). Unfortunately, this ambitious project was not funded. Coincidentally, while the TESRI proposal was being reviewed by the BMGF, two post-doctoral social scientists at WUR had developed an independent interest in pursuing complementary research on SRI. Together with the guidance of senior WUR researchers who were involved in the TESRI proposal, the two post-doctoral

14 researchers designed a proposal to investigate the spread and effects of SRI in India, in collaboration with four Indian PhD students and two senior researchers. The proposal was submitted to the Dutch Organization for Scientific Research (NWO) and was awarded funds in December 2009 to commence work in the summer of 2010. Following the disappointing outcome of the TESRI application, researchers at WUR remained convinced that research on SRI remained timely and useful. Drawing on the momentum generated during the preparations of the NWO proposal, WUR researchers decided to seek modest funds from the BMGF for a small exploratory study, involving primarily desk-based research, to review the current state of knowledge on the SRI phenomenon, identify important knowledge gaps, and explore its international spread. The BMGF agreed that such a preliminary study would be valuable, and work commenced in November 2009. This report provides an account of that work. 1.2 Conceptual approach: Economics and the Sociology of Science and Technology Prior to the beginning of this research project, the academic discussion about SRI had been dominated by agronomic analyses, alongside a small number of adoption and impact studies carried out by agricultural economists (see Chapter 5 for a review). It was evident that the controversy played out in scientific journals had not prevented SRI ideas and practices from continuing to spread among development practitioners and rice farmers in various countries. This could be regarded as a paradoxical situation, especially in light of the fact that agricultural scientists can often be heard to complain about the difficulty of transferring scientifically validated methods and technologies into farmers’ practice. With SRI, it appeared that at least some farmers were embracing practices that did not enjoy the unanimous support of the rice science community. The distinctive approach proposed by the WUR team was to recognize SRI as a real sociological and practical phenomenon and to use an open-ended exploratory method to try and discover what might be going on in this controversial field of agricultural knowledge and practice. This approach was partly a pragmatic one but was also informed by insights from social studies of science, technology and society (STS). In various discussions, the spread of SRI had already been characterized as a kind of ‘social movement’, and this terminology was adopted by the WUR team in the title of our research proposal. However, intellectual terms come with conceptual baggage, and the language of ‘social movements’ immediately invoked the rich sociological literature on the roles played by social movements (and ‘global social movements’ and ‘new social movements’) in political struggle and social change (e.g. Cohen & Rai 2000; Della Porta & Diani 1999). For an exploratory study of the kind envisaged by the WUR team, of an emerging phenomenon like SRI, it would in fact have been inappropriate to assume in advance that a particular set of conceptual and methodological tools would be suitable. Throughout this study, therefore, the idea of

15 SRI as a ‘movement’ has been used in a loose sense and should be understood in the same way. STS scholars argue that any functioning technological system necessarily involves the integration of human, social and institutional features alongside the tools, devices and machines that are commonly thought of as representing ‘technology’. Technologies of all kinds should therefore be understood as ‘socio-technical ensembles’ (Bijker 1997). Drawing on this conceptual resource, the WUR team intended that, if indeed SRI could be understood as some kind of movement of people and ideas, it would also need to be understood as a movement involving techniques, tools and practices as intrinsic, rather than incidental, features. In other words, it should be understood as a ‘socio-technical’ phenomenon rather than a purely social one; and the social and technical dimensions of SRI should be studied in a symmetrical and even-handed way. The WUR team’s conceptual move was therefore intended to be orthogonal to the existing SRI debate. This orthogonal position was designed to create an independent perspective on the contrary concepts, theories and empirical claims that have been invoked in the scientific dispute surrounding SRI, without resorting to an alternative conceptual toolkit that would try to explain SRI and hostility to SRI by invoking sociological or political concepts such as interests or ideologies. The aim was to appreciate the biophysical realities involved in rice farming and scientific experimentation without underestimating the importance of factors such as communication, learning, organization, input prices and resource endowments in determining the success or failure of a technological system like SRI. If this approach could be encapsulated in a few words, it would be the simple idea that the reasons why SRI might work, at least for some farmers in some contexts, may have as much to do with the methods by which the system is taught and learned and the institutions within which it is embedded, as with the technical performance of a particular set of agronomic practices. The WUR research team comprised a set of complementary disciplinary perspectives that could help shed light on the social and technical dimensions of SRI. Two post- doctoral researchers carried out most of the practical work in the project. One (Berkhout) has expertise in economics and econometrics and took the lead in a critical review of the data presented and methods used to assess the agronomic performance and other impacts of SRI to date. The other (Glover) has experience in the fields of STS and International Political Economy (IPE) and took the lead in studying the content of the SRI controversy, including the ways in which SRI has been conceptualized, attacked and defended, and the kinds of social, technical and institutional mechanisms that have helped to spread SRI locally and internationally. Two senior researchers (Schipper and Maat) provided support and guidance in these two respective areas. To ensure that the social scientific approaches used by these researchers remained connected to agronomic science, the project was led and advised by a senior farming systems agronomist (van Keulen).

16 1.3 Aims and goals Based on the conceptual approach outlined above, the overall aim of this project was to explore the emergence and evolution of the SRI phenomenon as a socio-technical movement, conceived in broad terms. The specific aims were to collect and analyse existing literature and materials on SRI and to attempt to map the historical processes and networks involved in the growth of an international SRI network and spread of SRI knowledge and practice. The project also aimed to identify all the significant national and international research institutions, extension agencies, non-governmental organizations (NGOs) and civil society organizations (CSOs) that have been engaged in promoting SRI. The overarching goal was to use these resources to identify the most important knowledge gaps surrounding the SRI phenomenon, encompassing agro-technical issues, socio-economic issues and (dis)adoption behaviour, and to formulate hypotheses and propose appropriate approaches and methods for further examination of SRI. 1.4 Design of the project In order to fulfil the aims and goals described above, the project was designed primarily as a desk-based study involving a literature review and identification and analysis of the network of actors involved in SRI activities. To ensure that the desk-based work was linked to ground-level realities, two field surveys involving focus-group discussions were planned, to Madagascar and Tamil Nadu (India). These visits were intended to include a field survey of SRI practitioners and promoters, focus-group discussions with SRI and non-SRI farmers, and workshops in both locations. The field visits were successful in uncovering a wealth of useful information about the dynamics of the SRI ‘movement’ in both locations, even though it proved impossible to perform all of the planned activities exactly as envisaged (see section 2.4). 1.5 Structure of the report Chapter 2 of this report describes in detail the activities undertaken within the scope of this project and the outputs produced. Chapter 3 provides a broad and detailed historical and conceptual overview of the SRI phenomenon. It describes the system’s origins in Madagascar and discusses the empirical observations and theoretical reasoning on which it was based. The chapter reviews the claims made on behalf of SRI, provides a working definition of the system’s basic principles, discusses the agronomic arguments arising around its individual components and gives an impression of the system’s international spread. This general overview provides the necessary foundation for the more detailed analyses presented in the remaining chapters. Chapter 4 contains a review of the existing scientific literature on technical and biophysical aspects of SRI cultivation, while Chapter 5 reviews agronomic and economic studies on the adoption and impacts of SRI methods. Chapter 6 then turns

17 to accounts of the SRI scene in Madagascar, Tamil Nadu and Nepal, based on field visits by members of the WUR project team. Chapter 7 discusses the actors, mechanisms and channels through which SRI has spread internationally, while Chapter 8 discusses the adaptations and variations in SRI specifications and practices that have arisen in different contexts. The structure of the report is such that the insights and conclusions of the report are brought together in Chapter 9, which identifies questions raised by SRI and gaps in our knowledge about the SRI phenomenon, and proposes an integrated methodology for further research and analysis. A number of appendices are attached.

18

2. Implementation of the project

Work commenced on this project at the end of November 2009 and was originally planned to be completed within 12 months. Difficulties in delivering the ambitious goals of the project, combined with some personal challenges, led us to request a two- month extension, which was kindly granted by the BMGF. This chapter describes the activities undertaken and describes the outputs produced during the course of the project. The chapter concludes with a discussion of some of the unanticipated challenges involved in carrying out research in the contested field of SRI. 2.1 Workshop in Brighton and visit to Cornell University Professor Norman Uphoff was among the first international scholars to encounter SRI in Madagascar in the mid-1990s (see Chapter 3). Since the late 1990s, Uphoff has sought to draw attention to SRI among scholars, development agencies and governments. In this role, he has given numerous presentations, published many papers, and participated in many international and national conferences, workshops and meetings to do with SRI (see Appendix C: Timeline). Having performed this role for over a decade, Uphoff can reasonably claim to be one of the best informed people in the world with an intimate knowledge of the SRI movement. In the course of this work, Uphoff has mobilized resources at his institution, the Cornell International Institute for Food, Agriculture and Development (CIIFAD), to build and maintain a website on SRI that performs the role of a portal and repository for SRI information online. 6 The CIIFAD SRI website and Norman Uphoff’s personal archive clearly represented a vital resource for this project, which Uphoff has generously made available to the WUR team. In December 2009, Uphoff was visiting the UK to take part in a conference and had also arranged to visit staff at the Institute of Development Studies at the University of Sussex (IDS, Brighton, UK). On Uphoff’s initiative and with the enthusiastic support of IDS researchers, an impromptu workshop was organized at IDS on 17 and 18 December 2009. In addition to Uphoff and representatives of the WUR research team and IDS staff, Dr. Biksham Gujja and Dr. Willem Stoop were able to attend at short notice. This group of researchers interested in SRI came together to share information and discuss and comment on the WUR team’s plans. These informal discussions were very useful in helping the WUR representatives to approach their work. Norman Uphoff also extended an invitation to the WUR team to visit CIIFAD at Cornell University (Ithaca, NY, USA) for further discussions, an opportunity to meet other CIIFAD staff involved in managing the website, and to explore Uphoff’s

6 http://sri.ciifad.cornell.edu/ (accessed 21 January 2011).

19 archive of documents on SRI. A member of the WUR team (Glover) was able to take up this invitation while travelling in the USA in February 2010. The visit created an opportunity for further discussions with Uphoff and his CIIFAD and Cornell colleagues, an informal seminar, and the chance to explore the CIIFAD SRI archives and website together with Ms. Lucy Fisher, who built and manages the website. These preliminary discussions and explorations formed a valuable foundation for the remainder of the project, especially in so far as they provided helpful insights into the types of information held on the SRI website, the SRI-related documents and personal communications that have not been incorporated into the website, and other background information. It was also very helpful to meet some of the Cornell University agronomists and soil scientists and learn about their views on and questions about SRI. During the course of the project, we have also kept in touch with TESRI collaborators from IRRI, especially Bas Bouman. One of the senior supervisors, Harro Maat, visited institutions involved in research activities on SRI in China during 2010. We also received numerous emails with information, expressions of interest in and comments on our work from other individuals and organizations involved in the SRI arena during the course of the project. Their names are listed in the Acknowledgements section. We are very grateful for their support and advice. 2.2 Literature search and review The literature search and review formed the major element of the work carried out under this project. The WUR research team cast its net wide in order to collect not only scientific literature but also various types of grey literature including draft scientific papers, consultants’ reports, project reports, working papers, memos, etc. The search strategy was open-ended and included a review and downloading of materials available on the CIIFAD SRI website, searches in academic databases such as ISI Web of Knowledge, OvidSP and EBSCOhost and web searches using Google and Google Scholar. Additional materials were identified by consulting the bibliographies of documents collected. We are also grateful to our network of contacts in the academic and SRI communities for sending numerous documents to us by email from time to time. The above search strategies were particularly effective for identifying documents in the international, English-language literature on SRI. To build a picture of the SRI scenes in Madagascar and India, we used additional strategies. 2.2.1 Madagascar. The CIIFAD SRI archive was especially useful as a source of unique unpublished documents relating to the early development of SRI in Madagascar, especially some annotated typescripts written by Henri de Laulanié. Also useful were several Masters and Doctoral dissertations produced by students from the University of Antananarivo (Antananarivo, Madagascar), which are available on the CIIFAD SRI website and/or in Norman Uphoff’s personal archive. Another invaluable source was de Laulanié’ s

20 book, Rice in Madagascar (Le Riz à Madagascar ) (de Laulanié 2003), which was published posthumously and could be obtained from an online bookseller. However, our efforts to collect additional literature during our visit to Madagascar indicated that relatively little scientific or grey literature on SRI is readily available or routinely sought on the island itself, 7 and we did not invest further resources in additional searches. 2.2.2 India. Scientific papers produced by Indian scholars are often published in international English-language journals and were easy to obtain from the Netherlands. Grey SRI literature from Indian sources is generally easy to obtain from the worldwide web and we are also grateful to various Indian contacts for sending materials to us. However, some Indian academic publications, despite being published in English, are difficult to obtain outside India. We therefore commissioned a researcher at Tamil Nadu Agricultural University (TNAU, Coimbatore, TN, India) to carry out a search of the Indian scientific literature for papers on SRI and/or its component practices. This search yielded a total of 28 papers, some of which were later judged not to be directly relevant to SRI and were excluded from our literature review. As a supplementary strategy to tap into information produced by the Indian SRI community, WUR team-members joined the SRI India Google group, which is a busy list server. 2.2.3 China. China was not one of the countries to be visited within the scope of this project. However, it became evident during the course of the research, as well as from advice received from various individuals, that a substantial amount of research and experimentation on SRI methods has been carried out by Chinese researchers and organizations. As is the case with India, some proportion of this work appears in international English-language journals, but much of it is accessible only to Chinese speakers within China. In order not to omit a substantial body of work that seems directly relevant to understanding the progress of scientific research on SRI, we commissioned a Chinese researcher from the China National Rice Research Institute to carry out a search of the Chinese scientific literature on SRI and, where necessary, to translate the abstracts from Chinese into English. This search yielded a total of 60 documents published in 28 Chinese journals between 2001 and 2010. All the collected documents were stored and categorized in a database. 8 In addition to recording basic bibliographic information, the database was used to give a systematic

7 Henri de Laulanié’s struggles to obtain scientific literature on rice cultivation indicates the difficulties of sourcing high quality scientific information in Madagascar during his lifetime (de Laulanié, 2003). 8 The Chinese documents and English-language abstracts were delivered in mid-January and have not yet been incorporated in the database.

21 structure to the literature review. Thus, the database was used to record, where applicable: (1) the definition or specification of SRI that was used in documents; (2) the control treatment or reference practice to which SRI was being compared, if any; (3) whether statistical methods were applied and if so, how this had been done; and (4) whether the statistical methods had been corrected for possible differences in inputs between SRI and non-SRI treatments (e.g. labour use, nutrient application or soil quality), and/or potential selection effects. The first of these classifications allowed us to assess the diversity of different implementations of SRI encountered in scientific experiments and field studies (see Chapter 8). The remainder were designed to enable critical analysis of the impacts of SRI methods on yield and productivity, as well as insights into the spread of SRI and the associated processes of adoption and adaptation at farm level. These procedural steps made statistical meta-analysis possible and ensured the work was done methodically. However, many documents were also read or scanned more qualitatively, as the content warranted. The outcomes of this review are described and discussed in Chapters 4 and 5. 2.3 Identification of actors and networks Individuals and organizations associated with published documents and grey literature on SRI are identifiable from the database. A second strategy for identifying actors in the SRI movement and discovering the links between them was to conduct surveys of SRI organizations and individuals. However, plans for a formal survey in each of two field locations proved to be impractical, although many useful interviews and group meetings were accomplished. Meanwhile, an online survey of organizations involved in SRI, which was designed to elicit information about the extent of SRI practice and the sources and channels through which SRI knowledge and practice has spread, had a very low response rate and was also abandoned. Partly as a result of these disappointments, additional effort was invested in developing ‘SRI timelines’ which depict the emergence and international spread of SRI knowledge and practice through a historical sequence of events (see Appendix C). Particular attention was paid to events associated with the spread of SRI in Madagascar and India, the activities of Norman Uphoff in raising awareness and promoting SRI, and a series of significant conferences, workshops and meetings that have played a role in disseminating SRI knowledge and stimulating practice internationally. Norman Uphoff has painstakingly recorded his work in the field of SRI over many years, in the form of detailed trip reports that provide valuable information about his encounters with individuals and organizations including many that have become

22 involved in SRI research, promotion or practice. These documents, as well as a series of overviews and summaries written by Uphoff and generously shared with the WUR team and others, provide an invaluable source of information that makes it possible, at least in principle, to trace the processes and channels through which SRI information has been disseminated, shared and possibly taken up. However, tracing these linkages in this manner rather than, for example, through a survey that can be tabulated and analysed quantitatively, is a particularly slow and laborious process. It also relies exclusively on the perspective of a single individual. For these reasons, the timelines remain a work in progress (see below). 2.4 Field visits This study was originally conceived as a desk-based study that would comb the scientific and grey literature and mine the world wide web for a broad, multi- dimensional picture of the SRI movement. However, with good reason, some interlocutors felt that a desk-based study alone would be insufficient to analyse a phenomenon whose origins and effects are intimately connected to field-level agronomy and farmers’ practice (see Chapter 3). The BMGF therefore encouraged the WUR team to include some field visits that would ensure that the desk-based research was grounded in field-level realities. Since the SRI movement seems to have evolved idiosyncratically in different places, it was evident that more than one field visit was appropriate. Yet, limitations of time, if not travel budgets, would make it impossible to spend much time in the field. In the light of these considerations, the decision was made to arrange short field visits to two locations which were felt likely to shed useful light on the origins and recent dynamics of SRI in contrasting agro-ecological and institutional contexts – Madagascar and Tamil Nadu, India. The choice of these two locations was also influenced by the fact that the WUR team felt able to mobilize networks of existing contacts in both places. In addition, most of the WUR team- members have a functional command of French, which is widely spoken in Madagascar, and all are fluent in English, which is widely spoken in India. 2.4.1 Madagascar Madagascar was chosen because the island is reputed to be the centre of origin of SRI. The two post-doctoral members of the WUR project team visited Madagascar from 14 April to 11 May 2010. They were joined by Prof. Herman van Keulen from 25 April onwards. The trip was divided into three parts. The first few days were spent in and around Antananarivo, the national capital, meeting some key stakeholders in the Madagascar SRI ‘movement’, visiting SRI sites within striking distance of the city, and planning and organizing the itinerary for the remainder of the visit. For the following ten days, Berkhout and Glover went on a tour of sites to the south of Antananarivo in order to meet farmers, extension agents and NGO representatives in and near the towns of Ambositra, Antsirabe and Fianarantsoa, in the central highlands of the country. Van Keulen followed an independent itinerary until the two post-docs returned to Antananarivo.

23 The final part of the field visit was devoted to a stakeholder workshop that took place in Antananarivo on 4 and 5 May. Almost 70 participants attended the workshop, including farmers and representatives from farmers’ organizations, government departments, NGOs, development cooperation agencies and research organizations. In addition to the WUR team, several other foreign visitors took part in the workshop and a field day that was organized for the following day: Dr. T. M. Thiyagarajan (TNAU, retired; consultant), Dr. C. Shambu Prasad (Xavier Institute of Management, Bhubaneswar, XIMB, Orissa, India), Dr. Julie Lauren (Dept. of Crop and Soil Sciences, Cornell University, Ithaca, NY, USA) and Mr. Tim Krupnik (PhD candidate, Dept. of Environmental Studies, University of California, Santa Cruz, CA, USA, representing the BMGF). A representative from IRRI was invited but none was able to attend. The participation of the Indian delegates in the Madagascar workshop was designed to ensure that there would be a degree of South–South exchange between our field sites. The Madagascar field visit is discussed in detail in section 6.1 and Appendix A. The field visit to Madagascar and the workshop were facilitated by Dr. Yvonne Rabenantoandro, Research Director, and her colleagues at the National Centre for Applied Research for Development (FOFIFA), by kind permission of FOFIFA’s Director General, Dr. Aimé Lala Razafinjara. We also received valuable help and advice from Mr. Peter Hofs, a development consultant then based in Antananarivo. Research assistance was ably provided by Mr. Fetrasoa Ernesto Ratovoarimanana, a student of the Department of Agronomic Sciences at Antananarivo University. 2.4.2 Tamil Nadu Tamil Nadu, India, was selected because SRI was understood to have spread widely and quickly, with the benefit of strong institutional and financial support. The two WUR post-docs visited Tamil Nadu from 25 July to 15 August 2010. They were joined by Dr. Rob Schipper on 6 August. The trip was divided into two parts. For the first two weeks, the WUR delegation travelled along a long, open-jawed loop, visiting numerous field sites and TNAU research stations across Tamil Nadu and Puducherry. Beginning in Coimbatore in the west of Tamil Nadu, the itinerary took them east and south to Pattukotai and Tiruchirapalli (Trichy); north-eastwards to Villupuram, Puducherry and Chennai (where Schipper joined the group); then westwards to Vellore, Krishnagiri and Hosur. A stakeholder workshop was held in Hosur on 11 and 12 August 2010. There were 50 participants in the workshop, including farmers and representatives of NGOs, funding organizations, agricultural extension services and TNAU. As was the case for the Madagascar visit, some overseas visitors were invited to take part in the workshop and a field day that was arranged for the following day: Mr. Tsimba Randriamiarintsoa (TOM, Behenjy, Antananarivo, Madagascar) and Mr. Tim Krupnik (representing the BMGF). A local IRRI representative affiliated to TNAU

24 also took part. Representatives from Cornell University Department of Crop and Soil Sciences were invited but none was able to attend. The Tamil Nadu field visit and workshop are discussed in detail in section 6.2 and Appendix B. The field visit and workshop in Tamil Nadu were facilitated by Dr. B. J. Pandian of the Water Technology Centre, TNAU and numerous colleagues, with the kind permission of Prof. P. M. Boopathi, Vice Chancellor, TNAU and Dr. S. Chellamuthu, Director of the Water Technology Centre. 2.4.3 Nepal In addition to the planned field visits to Madagascar and Tamil Nadu, this report also draws on a visit by WUR team-member Dominic Glover to three locations in eastern Nepal in November 2009. This field visit was undertaken in support of research by a WUR PhD student, Mr. Rajendra Uprety, who is currently studying the adoption and impacts of SRI in Nepal. The visit took place before the beginning of the BMGF project but involved observations of SRI practice and discussions with SRI farmers and agricultural extension officers, which provide useful insights to supplement the research carried out within this project. These insights are discussed in section 6.3 and a journal article (Glover in press). 2.5 Project outputs The substantive outputs of the project comprise the following: 2.5.1 This report. The report describes and explains the work carried out, discusses the emergence and socio-technical characteristics of the SRI phenomenon; reviews the available scientific and grey literature on the agronomy, adoption and impacts of SRI; and provides an account of the SRI scenarios in Madagascar, Tamil Nadu and eastern Nepal. Based on these reviews, the report discusses the extent of and mechanisms underlying the spread of SRI; considers the diversity in SRI practice; identifies important knowledge gaps in our understanding of SRI; and proposes methods suitable for future integrated, interdisciplinary research on the social, institutional and technical aspects of SRI. 2.5.2 A database of documents related to SRI. The database comprises 277 documents and includes peer reviewed journal articles, draft scientific papers, consultants’ reports, working papers, unpublished memos, a few official documents and some newspaper and magazine articles. The database contains basic bibliographic data, links to web sources, and document abstracts or summaries, where available. The documents have also been categorized and annotated. A small number of the documents, primarily peer reviewed journal articles, have short critical reviews attached. The database also lists organizations that are involved in some kind of SRI activities and some SRI-related websites.

25 The database is not comprehensive, nor does it contain a random sample of SRI documents. Some documents that we cite in this report have not been added to the database, primarily because uploading is a time-consuming process; at some point, we had to focus our attention on analysing the material rather than merely gathering and cataloguing it. So, the database remains a work in progress. To date, it has been used to assist our review of SRI literature and to evaluate the diversity of ways SRI has been defined and the degree of variation in individual component specifications that have been tested in field trials and on-farm studies. The database has also been designed to allow for future expansion and additional or alternative types of analysis, including network analysis. The latter type of analysis was not carried out in this project, primarily because of the failure of the online survey, which had been designed to provide insights into the linkages and communication channels between individuals and organizations involved in SRI promotion and practice. The database uses a Microsoft Access format which is also compatible with spreadsheet software such as Microsoft Excel and statistical tools such as SPSS. The database is currently accessible to WUR team-members only via a Microsoft SharePoint ‘Teamsite’ on the WUR intranet. Options for making the database more widely accessible will be explored with the BMGF. 2.5.3 A set of interactive timelines The timelines document key events in the evolution and spread of SRI. They were set up as a tool for visualising the historical sequence of events that have helped to spread SRI knowledge and practice beyond Madagascar. Like the database, the timelines are a work in progress and could be expanded or augmented in future. The timelines have been set up using the online facility TimeGlider . The timelines are scrollable, scalable and searchable and there is the possibility to attach notes, web links and images to each event. The user interface for viewing or editing the timelines is simple. Timelines can also be viewed as a list of events and can be down- or uploaded for editing using spreadsheet software. Access to view and edit the timelines is currently restricted, but they can be made more widely and even publicly accessible. TimeGlider is currently a free service in beta, but will become a fee- charging service with modest cost in future. We believe that the data from the existing timelines could be adapted for an alternative system if desired. 2.5.4 Journal articles. Drafting articles for submission to peer-reviewed journals was found to be an effective way to give focus to the literature review undertaken in this project, especially for elaborating the conceptual analysis of SRI and its origins. The development of these articles also helped to refine and explain the distinctive, orthogonal approach to understanding SRI that has been adopted by the WUR team (see section 1.2). Three articles have already been submitted to journals during the course of this project, of which one has been published and another accepted for publication. The titles, abstracts and publication status of the three articles are

26 presented in Appendix D. A fourth article, envisaged in the project proposal and currently under development, will present our review of the methodologies and major findings of the published scientific literature on SRI. 2.6 Flexibility in implementation The aims and goals of this project were ambitious (see section 1.3). As the sections above have indicated, not all of the specific goals envisaged at the start of the project could be achieved, for two key reasons. First, the exploratory nature of the study meant that it was important to be flexible and adapt to circumstances. The analysis and outcomes of the project were strongly influenced by the types and quality of the data that could be collected in a limited time. On one hand, the volume of documents of different kinds, the number of SRI projects and programmes and the diverse range of actors involved in SRI activities in different locations proved to be overwhelming. It was important not to allow the sheer amount of information to swamp the analytical work required. On the other hand, some of the efforts made to generate reliable information about the extent of SRI knowledge and practice, and the mechanisms by which these have spread, did not succeed. The online survey of organizations involved in SRI is the prime example. As with other setbacks, that situation was partially recovered by investing more effort in alternative strategies, such as the timelines. The second reason for not achieving all of the goals envisaged in the project proposal has to do with the nature of the SRI phenomenon, and amounts to a key learning outcome in its own right. A contested field like SRI is subject to the keen attention of many stakeholders with diverse and sometimes conflicting interests. In such a context, significant time and energy needs to be invested in communicating and explaining the aims, goals, approaches, methods and evolving insights of a research project like the present one. Many interested parties were understandably keen to share their views and inform our research with their knowledge and insights. These inputs were entirely welcome and helpful, but the WUR team had seriously underestimated how much time would be taken up in engaging with the constructive and often challenging points raised from various quarters. Significant time was usefully invested in refining, clarifying and explaining the WUR approach. This activity contributed directly to improving our understanding of the SRI phenomenon and how to examine it in an interdisciplinary way and contributed to the drafting of three journal articles (as listed above). Nevertheless, these activities undoubtedly diverted time and attention away from carrying out more procedural work, such as data gathering, building a database and reviewing large volumes of literature. One area in which the effects can be seen is in the design and implementation of the field visits. As described above, various people involved in designing the project and commenting on it felt that it was necessary to include a field research component. Evidently, a purely desk-based study would have lacked credibility among some of the people and organizations taking an interest in the SRI debates. The field visits did add a valuable dimension to the project, ensuring – as intended – that the desk-based

27 research was connected to ground-level realities. However, incorporating the field visits entailed significant adjustments to the initial project concept, with associated costs in time, energy and focus. The field visits were generally perceived as case studies, which implied a deeper and more extensive investigation than we had originally conceived. Planning, organizing and managing the field visits and workshops, and providing reports to involved parties afterwards, consumed a significant amount of time and energy, especially during critical early stages of the project. One result of investing more effort in understanding the individual dynamics of SRI in Madagascar and Tamil Nadu was that we were able to give less attention to reviewing experiences in other countries where SRI seems to have had some impact, such as Indonesia and Vietnam. This report reflects that imbalance. The field visits also widened the circle of stakeholders directly implicated in the research, which created an additional demand for explanation, communication and negotiation of terms. These imperatives brought home to us the distance that separates different perspectives on SRI. Everyone involved in the field of rice agriculture seems to have a view on SRI, some positive and encouraging further work, others negative and disparaging any time spent on SRI as a waste of effort. One particularly interesting insight from the field visits was that few people on the ground in Madagascar or Tamil Nadu, particularly but not exclusively those within the SRI movement, expressed any interest in the SRI controversy that has been played out in the pages of scientific journals (see Chapter 4). Very few farmers were even aware of the controversy. SRI-promoting organizations typically regarded the scientific dispute as largely or entirely irrelevant to their work, preferring to base their assessment of SRI methods on their own field-level experiences. This factor apparently influenced some people’s perceptions of our own efforts to review the status of the SRI phenomenon from an academic perspective. It certainly helps to account for some significant difficulties we experienced in organizing the field visits to Madagascar and Tamil Nadu, as well as the very low response to our online survey. Interestingly, even some scientists who took a negative stance towards SRI – of whom there were some in Madagascar, for example – expressed little interest in the arguments that had been rehearsed in the peer-reviewed scientific literature. Instead, they rooted their views in their own professional judgement and/or the alternative strategic priorities being pursued by their organizations. For the reasons discussed here, the outputs listed above are not precisely the same as the ones envisaged in the project proposal. Nevertheless, they represent concrete steps forward in our understanding of SRI as a real socio-technical phenomenon. Part 2 of the report discusses the nature of the SRI phenomenon in detail.

28

Part 2: An overview of the SRI phenomenon

29

3. Origins and theory of SRI

The System of Rice Intensification is typically summarized as a list of technical principles or practices for rice cultivation (see below). When presented in this way, SRI resembles a technological ‘black box’ (Latour 1987). The approach taken in this research has been to try and open up that black box and peer inside it, in order to explore how this technological system has been built and examine its social and technical components. This chapter therefore introduces the SRI phenomenon by putting it into a historical perspective. It discusses the origins of SRI methods, explores the development of its individual components, examines the theoretical arguments that have been mobilized to explain it, and begins to consider how widely the SRI phenomenon has spread internationally. This discussion thus provides a foundation for the more detailed analysis in the following chapters. 3.1 The origins of SRI The discovery or invention of SRI is typically attributed to Father Henri de Laulanié, a French Jesuit missionary who worked in Madagascar for 34 years. The celebrated ‘father of SRI’ was sent to Madagascar in 1961, when he was 41 years old, and spent the rest of his life living and working on the island. As well as being a Roman Catholic priest, de Laulanié had trained as an agronomist at the National Agronomic Institute in Paris, France. He had been in Madagascar for 20 years when, in 1982, he succeeded in establishing a rural training centre near the town of Antsirabe. There, the following November, occurred the fateful events that led to the ‘accidental discovery’ of SRI. The late arrival of seasonal rains caused a delay in seeding the rice nursery at the training centre. With a shorter rice season in prospect, de Laulanié and his students needed to produce rice seedlings in a hurry. If they were to generate enough seedlings to transplant into the paddy fields they had prepared, their small rice nursery would have to be used twice within 30 days (de Laulanié 1992; 2003). The consequences of this accidental experiment took de Laulanié by surprise. The tiny rice seedlings, transplanted after 15 days instead of the 25–30 days that were typical for the area at that time, grew much more profusely than expected. The plants displayed a prodigious capacity to produce the potentially grain-bearing stems, or tillers, on which the eventual yield of grain logically depended. The rice yield indeed increased. De Laulanié decided immediately that 15 days should thenceforth be considered the latest date for rice transplanting. To this thunderclap of inspiration, de Laulanié attributed his discovery of le Système de Riziculture Intensive or the System of Intensive Rice Cultivation, which is commonly translated into English as the System of Rice Intensification, thus keeping the French initials SRI (de Laulanié 1992; Rafaralahy 2002).

30 It is a dramatic and inspiring story, but one with a dose of artistic licence. The transplanting of very young seedlings was indeed one of the most distinctive features of de Laulanié’s novel rice cultivation system, but it was not the only one. The others had nothing in particular to do with the happy accident that occurred in November 1983. They included transplanting seedlings one by one instead of in clumps. This was a practice that de Laulanié had seen used by some small farmers in certain districts of Madagascar as early as 1965, which he had adopted in his own practice before 1975 (Elyah 2006; Rafaralahy 2002). According to some sources, the practice of planting between one and three seedlings per hill was part of a package of improved rice cultivation practices promoted in Madagascar through a Green Revolution-style programme after 1966 (Elyah 2006).9 Another characteristic of de Laulanié’s system was the transplanting of seedlings in carefully spaced rows or, preferably, square or rectangular grids. These recommendations were associated with the use of mechanical rotary hoes or weeders. De Laulanié attributed the basic principle to IRRI and pointed out that these practices were already being promoted by agricultural extension services in Madagascar (see Chapter 6). De Laulanié’s twist, which has been perceived as controversial, was to recommend spacing distances of 25x25 cm in the Madagascar highlands (which might be regarded as fairly conventional today) and as wide as 40x40 cm in the higher temperatures to be found at lower altitudes near the island’s coasts (de Laulanié 1992; 2003; Glover forthcoming). In fact, de Laulanié’s upper and lower ranges for spacing distance were only a little higher than the 20 cm to 35 cm guidelines that had been promoted in the Madagascar highlands after 1966 (Elyah 2006). One of the more striking features of de Laulanié’s system concerned irrigation. De Laulanié affirmed that his system was optimized for irrigated rice, but in fact he placed more emphasis on good drainage than on irrigation as such. He was convinced that rice is not an aquatic plant, merely one that can tolerate submergence, albeit with adverse consequences for its productivity. De Laulanié believed, instead, that it was important to promote aerobic soil conditions in rice plots, which would favour healthy roots and beneficial soil micro-organisms. Again, he was drawing on observations of existing farmer practice in Madagascar, where some farmers achieved good results by allowing their rice fields to dry out from time to time during the vegetative growth period. De Laulanié’s two key recommendations for promoting aerobic growing conditions were to irrigate with a minimum of water and to devise irrigation systems that would aerate the water delivered to the fields (de Laulanié 1992; 1993; 2003; Glover forthcoming; Rafaralahy 2002). Again, the emphasis on improved irrigation management, including drainage, was not entirely unprecedented for Madagascar (Elyah 2006).

9 It may be that de Laulanié’s recollection of having seen some farmers transplanting single seedlings in 1965 was out by a year or two. However, note that Andriankaja (no date) disagrees with Elyah (2006), stating that the guideline in question was ‘about 3–5’ seedlings per hill. Both authors cite sources for their statements, although only Elyah, whose account is more detailed, provides a full bibliographic reference.

31 De Laulanié recommended producing rice seedlings in a ‘garden-like’ nursery close to the rice fields, in order to allow the seedlings to be transplanted quickly and without drying out. In certain respects, de Laulanié’s system resembles the dapog system, in which nursery-raised rice seedlings are transplanted when still very young. De Laulanié was aware of the dapog system, which was originally developed in the Philippines and had been trialled in Madagascar as early as 1964 by the agronomist J. P. Dobelmann. It was in use by a seed centre in the country’s Lac Alaotra region in the early 1990s (de Laulanié 1992; Glover forthcoming). Remarkably, de Laulanié had relatively little to say about adding fertilizer to rice fields. 10 He described rice as ‘one of the least demanding crops that could exist’ (de Laulanié 2003: 92) and believed that incorporating rice crop residues, in the quantities that he expected to be produced using his improved management methods, would be sufficient to maintain normal levels of humus in the soil. Although he did not object to chemical fertilizers in principle, he had stopped recommending them to the farmers he worked with because of rising costs and declining availability. As an alternative strategy for resource-constrained farmers, he suggested that poor soils could be improved over time by planting well-adapted local rice varieties that could grow in less fertile patches, improving the soil through the action of their root systems (de Laulanié 2003; Glover forthcoming). It is important to note that de Laulanié also had a lot of other things to say about rice cultivation, among many other topics linked to peasant farming, human capacity building and rural development in Madagascar. With regard to rice in particular, he drew attention, for example, to the importance of land levelling and soil preparation, the careful handling and shallow transplanting of rice seedlings (a consequence of using very young seedlings) and the importance of timely harvesting (de Laulanié 1992; 2003). De Laulanié portrayed SRI as ‘a system based on the physiology of rice’ (de Laulanié 2003: 59). He liked to say that he had learned everything he knew about rice from having studied the plant itself. Faced with a dearth of reliable information, he decided to ‘go down into the rice field and pull up the rice plants in order to look at them up close’ (de Laulanié 1993: 73). Elsewhere, he wrote: In rice cultivation, rice alone is the master of the game and the rice farmer is its servant or, if you like, its disciple. The environment is important of course, but the soil, fertilization, nutrition, treatments etc… [sic] are secondary: what is essential is the entity, rice: it is the supreme judge[…] Rice is the only one to be able to teach rice cultivation[...]: “Observe and listen to your rice: it is the only one that knows the truth and can tell it to you” (de Laulanié 1992: 17).

10 This contrasts markedly with the attention he paid to fertiliser in relation to non-rice cultivation, to which 14 pages of his posthumously published book are devoted (de Laulanié, 2003). Needless to say, fertiliser applied to off-season crops may provide benefits to rice grown in the same plots during the rice season.

32 It is true that de Laulanié’s scientific method was almost entirely inductive, based on working closely with rice plants in farmers’ fields. However, it would be an exaggeration to say that de Laulanié based his conclusions solely on his close observation of rice plants. His cultivation system was also developed ‘in dialogue with the peasants’ (de Laulanié 2003) and designed explicitly to suit the capacities of the poor and marginal Malagasy farmers with whom he worked. He made some specific choices in his design of the system that took into account not only the physiological needs of rice as he perceived them, but also his estimation of the capacities of poor and marginal farmers and farm labourers. For instance, he rejected the idea of using direct seeding as a crop establishment method, even though he thought it could be beneficial for the rice plants, because he thought the techniques required were too onerous for manual labourers (Glover forthcoming). He also drew on farmers’ practices and the few sources of scientific information that were available to him in the Madagascar of the 1960s, 70s and 80s (see Chapter 7). It should be evident from this summary that de Laulanié’s rice cultivation system was significantly more complex than just transplanting very young seedlings, as his own account of the system’s ‘accidental discovery’ might have implied. It combined several distinct elements, few of which were entirely unprecedented or unique to SRI. Indeed, de Laulanié considered most of his guiding principles to be statements of accepted good agronomic practice for rice; and he embedded his observations within a much wider philosophical and strategic view of agricultural and rural development for Malagasy smallholders. Nevertheless, he did believe that he had stumbled upon two distinct new insights, namely the youthful vigour of very young seedlings and the desirability of aerobic rather than saturated soil conditions. However, even these two practices had precedents in farmers’ practice or existing systems of improved rice cultivation, of which de Laulanié was certainly aware. It appears that de Laulanié had already shortened the nursery period to 21 days in the mid-1970s (Elyah 2006). The nature of de Laulanié’s innovation, therefore, lay in the combination and refinement of several disparate observations, over many years, into a coherent system that was intended to improve the cultivation practices of poor rice farmers in the communities where he worked (Glover forthcoming). 3.2 Elaboration and refinement of SRI Henri de Laulanié died in Antananarivo in June 1995. A comparison of recent descriptions of SRI methods with de Laulanié’s own writings reveals how his ideas on rice cultivation have been developed and refined by his collaborators and successors since his death. Some subtle shifts of emphasis can be identified, as well as areas where de Laulanié’s broad-ranging discussion of good rice agronomy has evolved into a set of more crisply and concisely specified principles or guidelines. Chapter 8 contains a detailed, quantitative analysis of the range of variation that exists in the ways SRI is defined in documents contained in our database. In this section, we highlight a number of examples chosen to illustrate some interesting changes in the way SRI has been described over time.

33 De Laulanié’s immediate heirs were his colleagues from l’Association Tefy Saina (ATS), an organization he established in July 1990 to develop the capacities and self- reliance of Malagasy rural people. Mr. Sebastien Rafaralahy of ATS has apparently stuck rather faithfully over the years to de Laulanié’s view that SRI consisted of just two essential principles: first, irrigation using a minimum of water and second, the transplanting of very young seedlings, singly and with ‘optimum’ spacing. However, when it comes to describing the most important ‘complementary techniques’ to go alongside these essential principles, the two men’s accounts differ in an interesting way. De Laulanié’s 1993 journal article on SRI mentioned three points, namely the ‘garden-like’ nursery, careful transport of seedlings from the nursery to the main field with soil still attached, and an early first weeding after transplantation. Rafaralahy’s conference paper from 2002 agrees on the matter of early weeding but his second ‘additional element’ was fertilization, specifically recommending the use of organic compost or animal manure to promote soil biological activity. Rafaralahy attributed this insight to de Laulanié himself but, as discussed above, de Laulanié had not – at least in his writings – placed any significant emphasis on fertilization for rice, and had expressed a preference for organic fertilizer on pragmatic grounds of cost rather than an argument about soil biology (de Laulanié 1993; Rafaralahy 2002). Unsurprisingly, the earliest individuals and organizations to take up SRI, apart from ATS, were also based in Madagascar. Drawing directly from de Laulanié’s work, an Antananarivo-based NGO published a 13-page English-language handbook on SRI in August 1997. 11 This manual referred to SRI as ‘the Malagasy Early Rice Planting System’, but otherwise discussed the new method in a manner that was strikingly faithful to de Laulanié’s own. However, the document also illustrated the way de Laulanié’s ideas were being collated and summarized by others for communication to a wider audience, producing a rather more concise and accessible – yet perhaps also rather truncated – introduction to SRI than could be found in any of de Laulanié’s own writings (Vallois 1997).12 This process of distilling and summarising de Laulanié’s original ideas has since been taken even further. Nowadays, SRI is typically described as a list of five or six key practices or principles. For example, Norman Uphoff’s first journal article on SRI organized his description into five sections: (1) younger seedlings (ideally 8–12 days and not more than 15 days old); (2) widely spaced, single seedlings planted in grid patterns (typically 25x25 cm and possibly wider); (3) water management to promote moist, aerated soil conditions, sometimes including dry periods of 3–6 days; (4) up to four weedings, carried out either by hand or using a mechanical rotary weeder; and (5) fertilization (Uphoff 1999: 300–3). Uphoff’s (1999) definition is essentially similar to one provided by Stoop, Uphoff and Kassam (2002: 252), except that they include ‘raising seedlings in a carefully

11 The booklet was evidently translated from a French original. 12 Patrick Vallois subsequently developed an alternative, ‘softer’ version of SRI which has been promoted by CIRAD in Madagascar (see Chapters 6 and 8).

34 managed, garden-like nursery’ and thus arrive at a list of six components. The planting of single, widely spaced seedlings is sometimes decomposed into two independent practices, which was done by Stoop, Adam and Kassam (2009) and Krupnik (2008), who nevertheless also produced a list of six basic principles because they omitted nursery management as a core component in its own right. Statements about the appropriate fertilizer regime for SRI typically emphasise that the application of organic compost or animal manure is a highly desirable, optional practice, conditioned by the availability of local resources (e.g. Stoop et al. 2002; Uphoff 2007). Chemical fertilization is not ruled out, especially in the presence of specific nutrient deficits in local soils, though organic fertilizers are strongly advocated because they are thought to promote positive microbial activity in the rhizosphere (Uphoff et al. 2009). In this aspect one can see a distinct evolution in the way SRI is formulated, compared to de Laulanié’s original ideas. Weed control is recognized as a critical issue in SRI because the wide transplanting of very young seedlings creates an ideal opportunity for weeds to emerge and compete with the tiny rice plants. The use of mechanical rotary weeders is commonly recommended because eradicating weeds by hand is recognized as a strenuous and time-consuming task. However, mechanical weeding is often also discussed as a technique for churning and thus aerating the soil, as well as destroying weeds. In this respect too, SRI methods have been refined since de Laulanié died. De Laulanié’s principal recommendations for achieving aerobic soil conditions were good drainage, minimal irrigation and aerating the water supply, rather than churning the soil as such – although he also acknowledged a possible contribution of the rotary weeder in this regard (de Laulanié 1992; 2003). One aspect that de Laulanié considered vital but which is rarely included as a principal practice in its own right is the gentle handling, quick transplanting and shallow placing of rice seedlings. De Laulanié thought that these practices were essential in order to minimize the ‘transplanting shock’ incurred by rice seedlings, which could retard their re-establishment and renewed growth in the main rice field. In contemporary accounts of SRI, this feature is sometimes discussed at length, particularly by certain authors, but very rarely mentioned on the whole (see Chapter 8). 3.3 A working definition of SRI methods Our purpose in exploring a selection of alternative definitions of SRI in the previous section was to gain some insight into the process by which de Laulanié’s intricate and broad-ranging discussion of principles for improved rice agronomy has been distilled and organized into a more concise list of practices. Chapter 8 contains a more comprehensive discussion of the variation in the ways SRI is defined in the documents contained in the project database, including a quantitative analysis. One might argue that the differences between de Laulanié’s two key principles (underpinned by a range of complementary practices) and another person’s list of six

35 principles (or three, or five, etc.) are not ones of substance. In both cases, a lot of detail remains to be explained once the headline principles have been mentioned (Glover forthcoming). However, it does seem important to pay attention to the subtle differences, as well as the areas of broad agreement, among the alternative definitions that have been put forward by different authors. It may make a substantial difference on the ground if features like a particular type of nursery management or a specific fertilizer regime are specified as fundamental principles instead of optional or supplementary features. Variations in the way SRI has been described might therefore be related to substantive changes in the nature of the system. On the other hand, differences between alternative definitions of SRI may also indicate that there are differences in the way the same basic concepts have been perceived by different actors, at different times or in different settings. Variations in the way SRI has been summarized may also tell us something about the practical challenge of communicating and explaining a rather complex and integrated system in a concise and accessible manner. To illustrate the latter point, it is worth noting that certain authors have described SRI in slightly different ways at different times (see section 3.2 and above). The differences between their formulations on different occasions do not necessarily indicate that they have changed their basic understanding of SRI (though that is possible, of course). The differences may simply reflect an on- going effort to describe and explain SRI in clear and convincing ways for different audiences. A broad underlying agreement does seem to unite most of the definitions we have found. For the purpose of discussion, we propose to adopt, as a kind of working definition of SRI, the three overarching principles proposed by Uphoff (2003: 39): Healthy seedlings , which accounts for the nursery management component as well as early, quick and gentle transplanting; optimal plant density , which explains the use of wider spacing and single seedlings; and soil aeration , which underpins the irrigation practices and soil disturbance. This working definition provides a framework for a discussion in the next section on the agronomic theory underlying SRI. 3.4 The agronomic theory underlying SRI Uphoff’s (2003) concise distillation of SRI’s basic elements is based on his understanding of the system’s theoretical foundations. As described in section 3.1, Henri de Laulanié’s insights into the physiology and growth patterns of rice were derived by induction from empirical observation rather than theoretical literature. In fact, de Laulanié lacked a theoretical explanation for the prolific tillering of young seedlings until early 1988, when he came across a French-language technical publication that described a growth model for rice plants, based on work by a Japanese plant scientist. Subsequently, de Laulanié also came across work by two other scientists, Puard and Angladette, whose work seemed to explain why rice plants could grow well in aerobic soils. From these resources, de Laulanié built a theoretical foundation for SRI retrospectively (Glover forthcoming).

36 This work having been done by the mid-1990s, Norman Uphoff’s introduction to SRI was able to take place through theory, as well as discussions and other interactions with de Laulanié’s ATS colleagues (Uphoff 1999). In the ensuing years, Uphoff has published numerous papers and made many presentations explaining SRI theory. His distillation of the system into three basic principles or pillars is concise, reasonably clear and accounts for all the signature practices of SRI (Uphoff 2003). We recognize that some stakeholders in the SRI community might dispute certain aspects and nuances in our working definition, but we believe that it expresses the contours of the underlying theory of SRI. It forms the foundation for our discussion of the agronomic literature on SRI in the next chapter. 3.4.1 Healthy seedlings: nursery management, seedling age and transplanting

Rice seedlings maintain greater potential for tiller and root growth (a) if transplanted while still very young, and (b) if transplanted very quickly and carefully to avoid desiccation and traumatization of the plant, particularly protecting the roots (Uphoff 2003: 39). De Laulanié’s rationale for planting younger seedlings was based, in the first instance, on his empirical observation of the luxuriant growth of seedlings when they were transplanted much younger than had previously been tried (see section 3.1). After the first ‘accidental experiment’ with 15-day-old seedlings in 1983, de Laulanié and his students experimented with seedlings as young as seven or eight days. Later, de Laulanié came across a model of rice tillering that had been developed by a Japanese researcher, T. Katayama, which described the growth patterns of rice and analysed the relationship of descent between one generation of tillers and the next. Katayama’s model broke down the growth of rice plants into phyllochrons , the time intervals between the emergence of successive leaves on the same tiller. De Laulanié’s extrapolation of Katayama’s model suggested that transplanting the rice seedlings before they reached the fourth phyllochron, i.e. when they were no more than about 15 days old (depending on the climatic conditions) and still had just two leaves, would allow the resilient young plants to recover from transplanting shock before they entered their most prolific growth phase. They would then produce abundant tillers and roots, and retain their potential to produce many panicles (Glover forthcoming; Stoop et al. 2002; Uphoff 2003). In de Laulanié’s analysis, the avoidance of transplanting shock thus became a key aim. His recommendations for nursery management and transplanting technique followed directly from this goal: a garden-like nursery, close to the main rice fields, would allow healthy seedlings to be produced not far from where they were to be transplanted; they should be pulled up gently and transported quickly, with the seed still attached and mud adhering to the roots, to avoid drying out; and transplanted delicately and at a shallow depth, using a specific technique designed to ensure the root could resume its downward growth as quickly as possible (de Laulanié 1992; 1993; Stoop et al. 2002; Uphoff 2003).

37 3.4.2 Optimal plant density: number of seedlings per hill and spacing of hills

Rice plants can better realize their potential for tiller and root growth, and for subsequent grain filling, if spaced widely rather than densely. Yield depends on the number and size of fertile tillers per m 2 rather than per plant, but total plant performance can be enhanced with optimum spacing rather than crowding (Uphoff 2003: 39). De Laulanié seems to have used practical reason rather than sophisticated theory to explain and justify the wide separation of individual rice plants in the main field. With more profuse tillering and vigorous root growth, sufficient space would be needed to prevent the rice plants from crowding one another. Also, crowding would logically result in competition between the rice plants for soil nutrients, light and water. The unifying logic of de Laulanié’s theory was to enable rice plants to achieve their full potential for growth, so it must have seemed logical to provide as much space as possible for the plants to expand and exploit the available resources in the surrounding soil. In general terms, de Laulanié thought that the growth and vigour of above- and below-ground plant structures were directly correlated, so that achieving the potential yield of rice depended on creating favourable conditions for large and healthy root growth. In particular, he argued that a rice plant grown under SRI conditions would have more functioning leaves and roots at the time of panicle formation and grain filling than one grown under conventional management (de Laulanié 1993). This aspect of SRI also seems to attract less commentary from later authors; evidently its benefits are thought to speak largely for themselves (e.g. Stoop et al. 2002; Uphoff 2003). Uphoff (1999), however, points out that rice plants with sufficient space to grow can produce robust roots and sturdy shoots that resist lodging. De Laulanié recommended a spacing of 25x25 cm in the central highlands of Madagascar and as wide as 40x40 cm at sea level. The difference hinged primarily on temperature but also soil fertility. He suggested that spacing could be widened even further as farmers became more adept in SRI methods and, over time, as soil conditions improved (de Laulanié 1993). ATS has reported farmers achieving spectacular yields with spacing distances as wide as 50x50 cm, even at higher altitudes (at Soatanana) (Rafaralahy 2002). 3.4.3 Soil aeration: promoting healthy roots and soil microbial activity

Rice plants have better root and tiller growth if the soil is kept intermittently aerated during the vegetative growth period, prior to panicle initiation. This minimizes root degeneration and plant senescence that occur with hypoxic soil conditions and at the same time supports aerobic as well as anaerobic microbial populations in the soil (Uphoff 2003: 39). As described in section 3.1, de Laulanié adopted the principle of maintaining moist rather than submerged soil conditions, and even occasionally allowing rice fields to dry out for short periods, after having seen some Malagasy farmers apply such methods with good results. It was not until 1991 that he came across a theoretical

38 rationale to explain why this practice could favour rice growth and productivity. He relied on scientific research by Puard and Angladette to make an argument that structural changes occur in rice roots as they adapt to submergence, which enable the plant to convey oxygen from the leaves to the root tips but inhibit the transport of nutrients from the roots to the above-ground parts of the plant. The structural changes involve the formation of air pockets in the roots, called aerenchyma . According to this theory, aerobic soil conditions could enable the rice plants to obtain sufficient oxygen from the root zone, rather than having to adapt their roots in order to transport it from above (de Laulanié 1993). Regarding soil micro-organisms and their contribution to soil fertility and rice growth, de Laulanié had relatively little to say. He realized that the activity of both aerobic and anaerobic microbes would help to determine the levels of soil nutrients available to feed the rice crop, and he recommended certain precautions to protect microbial life in the soil from cold and moisture loss during the winter. However, he lacked concrete and detailed information about the micro-organisms he was dealing with or the respective roles and contributions of aerobic and anaerobic microbes, although he assumed that the two types would be complementary. Accordingly, he seems to have aimed at a pragmatic balance by maintaining moist, aerobic conditions as much as possible (de Laulanié 1992; 2003). In both of the above dimensions, the theoretical arguments for maintaining aerobic soil conditions have been developed substantially by other scholars since de Laulanié’s death. In particular, Norman Uphoff, Willem Stoop and Amir Kassam have explored the scientific literature extensively to see whether existing research might shed light on these aspects of the SRI method. Although they have found a good deal of helpful research in that regard, they also point out that there are still large areas of ignorance and uncertainty regarding the interactions between roots and soil microbial communities, under aerobic or anaerobic conditions (Stoop et al. 2002; Uphoff 1999). 3.4.4 Synergies One could imagine a rice cultivation system that took up the theoretical principles and related practices described in sections 3.4.1, 3.4.2 or 3.4.3 independently of one another: the principle of aerobic soil conditions, for example, without the young seedlings and widely spaced, individual transplants; or using very young seedlings but transplanting them densely and applying conventional irrigation management techniques. Advocates of SRI methods go a step further by arguing that there are synergies among the system’s individual components, so that the benefits of the complete system are greater than the sum of the benefits of its parts. The claim that synergy is an intrinsic, emergent property of SRI is highly significant and has proved to be controversial. In its most potent sense, the idea of synergy suggests that the individual component practices of SRI are not just complementary or mutually supportive but that they interact and multiply their individual effects,

39 producing a powerful, emergent dynamic that leads to dramatic increases in yield and/or factor productivity. This is another area in which the theoretical arguments underpinning SRI appear to have developed substantially since de Laulanié’s death. The idea of synergy is no more than implicit in de Laulanié’s own writings on rice cultivation. He came closest to it when he began to extrapolate from the combined insights that he had drawn from the work of Katayama, Puard, Angladette and other scientists, and discerned the outlines of a general theory of rice physiology that would encompass the mutually sustaining relationships among plant structures above and below ground, the importance of promoting healthy, early and sustained tillering, and the potential importance of interactions between soils and roots in the rhizosphere (Glover forthcoming). The theory that there may be synergies among the SRI principles has been developed much further by several other authors. It was first raised in the scientific literature by Uphoff (1999), having previously raised the topic with researchers at IRRI. Synergetic relationships were later described by Stoop, Uphoff and Kassam (2002: 267) as ‘the critical element of SRI’. They argued that these synergies could ‘contribute to a shift in what have been considered agronomic “yield ceilings”’ (Stoop et al. 2002: 250), thus expanding the physiological potential of rice through changes to crop management methods. Uphoff (2003) has developed this idea to make the argument that SRI methods are capable of producing new and improved phenotypes from unchanged genotypes. These claims about synergies and an expanded yield potential are perhaps the most controversial ones made about SRI, which helped to spark the scientific disputes that are discussed further in Chapter 4. More recently, Uphoff has emphasized the idea that SRI methods foster and exploit symbiotic relationships between plants and soil micro-organisms (Uphoff 2007; Uphoff et al. 2009).

40

4. An overview of agronomic issues in SRI cultivation

This chapter presents a review of scientific literature and other relevant documents on biophysical and agronomic issues arising from SRI cultivation methods, including key arguments raised during the SRI controversy, the physiology and morphology of rice plants, crop establishment methods, irrigation management, weed control and soil nutrient management. The chapter concludes with a discussion on the issue of synergy. Although we are not agronomists, our attempt to study SRI as a socio-technical phenomenon would hardly be credible unless we based our analysis on a grasp of some basic agronomic principles. Our approach has therefore been to root our study firmly in relevant agronomic literature, supplemented by the helpful advice and insights of the numerous agronomists, farmers and agricultural extension workers whom we met during the course of the project. However, we do not presume to usurp the expertise of trained agronomists. In this section, we sketch the broad outlines and contours of the scientific debates surrounding SRI, taking our cues from the arguments and discussions of scientists and following their lead when they refer to wider literature on rice cultivation, plant physiology and crop management. We do not attempt to insert our own critical analysis of the methods or arguments used by experts, but take their arguments and conclusions at face value. In other words, we try to provide an accurate picture of how key scientific issues have been discussed, how arguments have been constructed and how knowledge has been made in this field, without any pretence to resolve disputes or settle arguments ourselves. One of the things we learned about the agronomy of SRI is the difficulty involved in delineating the focus of study. As discussed in Chapter 3, SRI is not an entirely discrete system of agricultural practice, since it has some features in common with other systems and practices that have been used by farmers or promoted by extension and development agencies in various locations and at different points in time (see also Chapter 8). In addition, as we discussed in Chapter 3, the three essential principles of healthy seedlings, optimum plant density and aerobic soil conditions were derived from three different theoretical and empirical foundations. Thus, the theories and principles of SRI impinge upon numerous issues within the broad field of agronomy. In addition, it is important to be aware that literature that addresses SRI, as such, comes predominantly from those sections of the scientific community that have chosen to explore the system. Relevant knowledge is also to be found in the work of other scientists who may not have positioned their work explicitly within the field of SRI. This category encompasses relevant studies on soil biology, tillering, weed control, water management, and so on, that are not about SRI as such. As we learned more about these topics, we became aware of the enormously wide range of scientific

41 and practical issues that were potentially implicated in understanding the principles of SRI from an agronomic perspective, ranging across the fields of plant physiology, soil biology and soil chemistry to the management of nurseries, irrigation systems, weeds and pests. This steadily widening horizon contributed to the overwhelming amount of material we came across that could be relevant to our research (see Chapter 1). Given the constraints of our project, we do not pretend that our review of agronomic questions is comprehensive and we realise full well that scientific experts might take issue with our summary. Nevertheless, we believe that scientific experts as well as policy makers, SRI practitioners and others may find our brief summaries of salient issues useful. We hope that our effort to link social and economic studies to biophysical studies can be the basis for further integrated research on the social and technical dynamics of SRI. 4.1 The ‘Rice Wars’ The ‘Rice Wars’ is a term given to the fierce dispute that erupted after SRI first arrived in the global scientific arena (Shambu Prasad 2009). Before publishing his first journal article on SRI, Norman Uphoff collaborated with some colleagues based in Madagascar to draft a scientific paper on SRI (possibly Holiarison et al. 1998), which they sent to scientists at IRRI for comment. IRRI’s response argued that five of the six basic SRI practices had previously been investigated by IRRI or other rice scientists; the exception was the transplanting of single seedlings instead of clumps of a few seedlings. 13 Uphoff was disappointed that the IRRI response did not take up the question of synergies (Uphoff 1999). Between 1999 and 2003, Uphoff and a number of colleagues published a slew of papers on SRI, which included journal articles, book chapters and a collection of papers from a conference (Stoop et al. 2002; Uphoff 1999; 2002; 2003; Uphoff et al. 2002; Uphoff & Randriamiharisoa 2002). Several of these represented efforts to describe and explain SRI in general terms, as a novel and distinctive approach to rice cultivation. In terms of discussing and elaborating the theoretical and conceptual issues raised by SRI, and also because of the calibre of the journal where it was published, one of the most prominent of these publications was a paper by Stoop, Uphoff and Kassam (2002). Their paper outlined the historical origins of SRI in Madagascar; described the basic components of the system, summarized as a list of six practices; reviewed some key scientific issues and principles concerning the physiology of rice and biophysical mechanisms involved in rice growth and cultivation; and discussed some knowledge gaps and their implications for future research and agricultural development strategies.

13 This is an interesting observation in view of the fact that a recommendation to use between one and three seedlings per hill had apparently been promoted in Madagascar under Operation Rice Productivity in 1966 (Elyah, 2006). Later, Sheehy et al. reported that the effect of the number of seedlings planted per hill had been ‘studied for nearly 100 years … and there is no evidence that transplanting a single seedling per hill would increase yield’ (Sheehy et al., 2004: 6, citations deleted).

42 In the process, the authors positioned SRI as a low external-input alternative to dominant Green Revolution approaches to plant improvement and crop management. They also argued that SRI management practices could mobilize synergetic interactions that could ‘contribute to a shift in what have been considered agronomic “yield ceilings”’ (Stoop et al. 2002: 250), thus positioning SRI crop management methods as an alternative to the plant breeding and genetic modification strategies currently being employed to expand the yield potential of rice plants. Although Stoop, Uphoff and Kassam (2002) called only for a diversification of rice R&D strategies to accommodate SRI research, their arguments could also be understood as a strong criticism of mainstream thinking on how to improve rice agriculture. This particular aspect of their argument provoked a robust response from some of the scientists associated with those mainstream views. For example, Achim Dobermann, who now serves as IRRI’s deputy director-general for research, 14 responded with an article in which he argued that SRI methods were not necessary to produce rice yields close to their maximum potential, since other, more cost-efficient methods were known to lead to very high yields. He argued that SRI methods therefore had little to offer to rice farmers in areas with good soils and irrigation. At the same time, however, he conceded that SRI methods could be suitable for poor and marginal farmers coping with poor soils (Dobermann 2004). Dobermann thus seems to have interpreted Stoop, Uphoff and Kassam’s (2002) arguments as a claim that SRI methods alone could achieve the best possible yields in rice. In fact, their paper had also made arguments about the long-term ecological sustainability of high external- input methods and proposed SRI as a system that was particularly suitable for resource-poor smallholders. What seems to have provoked Dobermann (2004) and some other rice scientists are some reports of SRI yields in excess of 20 tonnes per hectare, which were cited in a conference paper by Sebastian Rafaralahy (2002). This short paper was not a scientific publication but a short account of SRI’s origins and the experiences of farmers working with ATS in Madagascar. Dobermann argued that these claimed yields significantly exceeded the theoretical yield potential of rice as well as the best yields achieved in research experiments. Rafaralahy’s yield figures provoked similar reactions from other scientists including Sheehy et al. (2004), Sinclair and Cassman (2004) and McDonald et al. (2006) Sheehy et al. (2004) reported some field trials conducted at three Chinese sites to compare SRI methods with ‘conventional management’. The paper concluded that SRI methods had no yield advantage over conventional practice, at a cost of additional organic fertilizer and labour. The paper also presented a theoretical model of rice yield which led the authors to conclude that the ‘fantastic’ yields reported by

14 In 2004, Dobermann was working at the University of Nebraska–Lincoln (Lincoln, Nebraska, USA), having previously worked at IRRI between 1992 and 2000. See http://irri.org/about-irri/our-people/staff-profiles/irri-management/achim-dobermann (accessed 2 February 2011).

43 Rafaralahy (2002) exceeded the photosynthetic efficiency of rice and were therefore ‘probably the consequence of some form of measurement error’ (Sheehy et al. 2004: 7). Sinclair and Cassman (2004) added a commentary to Sheehy et al.’s article, in which they labelled SRI as an ‘agronomic UFO’: an ‘unconfirmed field observation’. In March 2004, a news feature in Nature reported on the conflicting views on SRI (Surridge 2004). A few months later, the IRRI magazine Rice Today published a pair of opinion articles by Uphoff (2004b) and Sinclair (2004). Sinclair’s opening words were, ‘Discussion of the system of rice intensification (SRI) is unfortunate because it implies SRI merits serious consideration. SRI does not deserve such attention’ (Sinclair 2004). The following year, Field Crops Research published another exchange of views between Sheehy, Sinclair and Cassman (2005) and Stoop and Kassam (Stoop & Kassam 2005). In these commentaries, Sheehy and colleagues dismissed SRI as a form of ‘nonsense’ or ‘non-science’, while Stoop and Kassam fiercely criticized both the methodology used in the earlier paper by Sheehy et al. (2004) and the ‘inappropriate and unjustified’ commentary by Sinclair and Cassman (2004). McDonald et al. (2006) compared reported yields of rice under SRI management and best management practices (BMPs) from a dataset of 40 site-years of observations. They concluded that, while SRI out-yielded BMPs in some comparisons, there was no evidence of a consistent, systematic advantage to SRI methods compared with BMPs, except in Madagascar. Following Dobermann (2004), they noted that this finding might be explained by local agro-climatic conditions, especially iron toxicity in acid Madagascar soils (a problem associated with continuously flooded irrigation). McDonald et al.’s (2006) analysis drew attention to the importance of clearly identifying the relevant comparator against which to assess SRI methods. They noted that it was reasonable to think that SRI methods could improve rice yields compared to existing farmers’ practices, though the same could be said for any intensive crop management system. They concluded that ‘there is no empirical evidence that SRI unlocks a previously unexploited yield potential in rice’ (McDonald et al. 2006: 35). Uphoff, Kassam and Stoop (2008) responded with a critique of McDonald et al.’s (2006) methodology and data. They argued that the study had omitted key data, especially evidence from field trials in China, India and Indonesia, and that their analysis had been preoccupied with avoiding a false positive conclusion without placing the same weight on avoiding a false negative conclusion. They argued that there was still insufficient evidence for a definitive conclusion about the superiority of SRI or BMPs and proposed some considerations to be taken into account in the design of future agronomic studies. McDonald et al. (2008) disputed the criticisms of their methodology and affirmed their original conclusions. Again, they emphasized that they found no evidence to show that SRI methods could raise the yield ceiling for rice and argued that, therefore, ‘the global promotion of SRI rests on a false premise’ (McDonald et al. 2008: 190). However, they also reaffirmed that SRI methods might have beneficial impacts compared to farmers’ existing practices, although ‘factor

44 productivity and yield advantages will be observed with almost any management innovation including the very basic steps of ensuring timely and uniform transplanting, which many farmers still do not practice’ (McDonald et al. 2008: 190). The Achilles’ heel of the arguments for taking SRI seriously, at least until recent years, has been the dearth of high-quality, peer-reviewed studies that could provide sufficient empirical evidence to convince the sceptics. Some scholars argue that this fact reflects the intrinsic character of SRI as a distinctive system developed ‘experientially and empirically’ in farmers’ fields in Madagascar (Stoop & Kassam 2005). Indeed, from the beginning, SRI advocates have acknowledged the lack of detailed understanding and called for further scientific research (Stoop et al. 2002; Uphoff 1999). Both sides in the ‘Rice Wars’ have argued partly from theoretical and conceptual standpoints, drawing upon existing literature, modelling approaches and informal reports from field studies to make their arguments. In the absence of compelling empirical evidence to decisively confirm or confound these conceptual arguments, neither side has succeeded in landing a killer blow. Until fairly recently, empirical verification has been relatively scarce. Many of the early reports about SRI were informal or anecdotal or from grey literature rather than peer-reviewed scientific sources. The first systematic research on SRI agronomy was carried out by students belonging to the Faculty of Agriculture at the University of Antananarivo (e.g. Andriankaja no date; Barison 1997; Holiarison et al. 1998; Rajaonarison 1999). It is perhaps not surprising that they did not carry as much weight as peer reviewed journal articles by senior researchers would have. Similarly, the report of the Madagascar SRI Consortium provided a relatively terse and impressionistic account of the project’s research results (The SRI Consortium 2003). An international conference on SRI that was held in Sanya, China in 2002 generated a body of papers from 14 different countries (Uphoff et al. 2002). The majority of these took the form of summaries of experiences and discussions of future prospects for applying SRI in the countries concerned. Nine short research reports were included, however, of which five were from China, three from Madagascar and one from India. Seven of these research reports were studies on the biophysical mechanisms and physiological effects of SRI cultivation; two were agronomic evaluations of SRI-type cultivation systems. The conference proceedings also included the paper by Rafaralahy (2002) in which yields of more than 20 t/ha. were reported, that provoked disbelief among some rice scientists (see above). In more recent years, a greater quantity of peer-reviewed literature has become available. In the following sub-sections, we discuss in general terms what the agronomic literature has had to say about the individual component practices of SRI and their value. One remarkable observation is that the SRI debate has encouraged some scholars to search more widely in the literature, uncovering and re-assessing some old studies in order to understand why and how changed crop management methods might produce different yield outcomes.

45 One important area of SRI literature that we have only been able to review superficially is the output of articles in Chinese-language scientific journals. We commissioned a Chinese researcher to carry out a literature search and translate the abstracts of Chinese scientific papers on SRI. The search yielded 60 documents, indicating that Chinese scientists have carried quite a wide range of studies on SRI and its components. Articles about SRI first appeared in Chinese scientific journals beginning in 2001, with the earliest of them introducing SRI methods, describing the system’s origins and explaining key concepts (e.g. LIN et al. 2004; MAO 2003; TANG 2002; YUAN 2001). Since then, Chinese SRI research has apparently expanded and progressed quite rapidly, as recent studies have included some quite complex multi-factorial experiments assessing SRI cultivation methods alongside or in comparison to a diverse range of specifications of planting densities, crop establishment methods, transplanting patterns, fertilization regimes and tillage systems (including no tillage) (e.g. FAN et al. 2007; GUO et al. 2007; WANG et al. 2006). The collection of Chinese abstracts was only delivered to us on 21 January 2011, and therefore we have not uploaded the documents to our database. It is important to note that our interpretation of these studies has been very rapid and relies on the translated abstracts only. 4.2 General studies on SRI cultivation methods, physiology and morphology General support for the principle underlying SRI – namely, that substantial improvements in yield and productivity can be achieved through changes to crop management practices rather than improved rice varieties – has come from an analysis published by Horie et al. (2005). They have argued rather persuasively that record- breaking rice yields produced by Japanese farmers during the 1950s and 1960s, which outclass average yields in contemporary Japan and China by around 100 per cent, were achieved largely through good cultural practices that included the incorporation of large volumes of manure and intermittent irrigation with well-drained soils. Horie et al. (2005) also discussed the potential of some distinctive SRI management practices, especially the use of young seedlings planted at low density (see below). Chinese studies have provided some general confirmation of basic concepts of SRI, demonstrating that the system’s methods produce vigorous vegetative growth that can lead to improved yield (e.g LU et al. 2005; Wang et al. 2004). Chen et al. (2006) found that SRI methods did not produce more tillers/m 2 than conventional methods but did produce larger spikelets, more grains per panicle and had a better seed filling rate. Lu et al. (2006) confirmed that rice plants under SRI management had higher root activity throughout the growing stages, higher nitrogen (N) content and grain filling rate in the late growth stage, higher chlorophyll content during ripening, and delayed leaf senescence, leading to faster grain setting and higher grain yield. Xu et al. (2005) identified an improvement in the grain quality of rice under SRI management, characterized by an improved milling rate and less chalkiness, which may have been associated with an earlier or longer heading stage.

46 Experiments in India have also tended to confirm that SRI methods produced physiological and morphological changes in rice plants that could lead to improved yields and higher factor productivity. For example, Vijayakumar et al. (2006) found that SRI methods produced a strong performance in most of the yield attributes of rice, including panicle length, the number of panicles per hill, the 1,000-grain weight, the number of grains per panicle, and the number of filled grains per panicle. Although the number of productive tillers per square metre was not quite as high as in conventional cultivation, grain yield was substantially better with SRI. The combination of 14-day-old dapog seedlings, 25x25 cm spacing, water-saving irrigation and mechanical rotary weeding produced a yield of 7 t/ha. in the wet season and 5.6 t/ha. in the dry season. This compares with yields of 5.6 t/ha. and 4.4 t/ha., respectively, that were produced using a conventional specification of 21-day-old seedlings, 15x10 cm spacing, conventional irrigation and weed control using herbicides and hand weeding. Satyanarayana et al. (2007) also reported a quality improvement in rice grown under SRI management, leading to a higher milling rate that justified higher prices for the farmers. Thakur et al. (2010b) carried out experiments to compare a ‘modified SRI’ specification with local recommended management practices (RMPs). The modified SRI practice included 10-day-old, single seedlings, with no continuous flooding and use of mechanical weeders. Hill-to-hill spacing was just 20x20 cm, but this was still wider than the local RMP. Modified SRI methods produced 637 g/m 2 of grain compared with 449 g/m 2 for the RMP treatment. These positive findings in China and India contrasted with the results of studies by Latif et al. (2009; 2005), who concluded that rice under SRI management (5.95 t/ha.) did not yield as much as rice managed according to local BMPs (6.88 t/ha.). SRI methods did produce a better yield than local farmers’ practices (5.05 t/ha.), but at a higher cost, so that the net returns for SRI and farmers’ methods were similar. The key reason was a higher demand for labour with SRI, especially for weeding. Latif and colleagues concluded that SRI was unlikely to be a very attractive method for Bangladeshi rice farmers. It is worth noting that yields reported in these studies, for both SRI and local best management, are in the range 4.5–7 t/ha. This range is fairly typical of the controlled field trial studies we have seen, sometimes extending to around 8 or 9 t/ha. or even a little higher in very productive experiments. SRI yields on farmers’ fields can be as low as 1–2 t/ha., though this may still be higher than yields achieved with conventional methods in the same location (see section 5.3.4). We have found very few reports of extremely high rice yields and none of the magnitude reported by Rafaralahy (2002). However, improvements in rice cultivation are about productivity increases as well as higher yields. In the discussion that follows, we do not focus strongly on absolute grain yields. Instead, we concentrate on describing differences that have been reported between the outcomes of SRI and other cultivation methods, including changes in plant physiology, morphology and yield attributes.

47 4.3 Crop establishment: Young seedlings, seedlings per hill, spacing distances The early studies in Madagascar generally confirmed that transplanting younger seedlings of about 8–12 days of age produced better yields (Andriankaja no date; Barison 1997; Rajaonarison 1999; The SRI Consortium 2003). In experiments carried out in Indonesia, Makarim et al. (2002) reported better yields from 15-day-old seedlings than 21-day-old seedlings. Mishra and Salokhe (2008) found that seedlings transplanted at 12 days of age performed better than 30-day-old seedlings, with greater uptake of both N and manganese (Mn). Menete et al. (2008) reported that 10- day-old seedlings matured faster and produced more panicles per plant and more grain yield than 20- or 30-day-old seedlings, although the magnitude of the effect varied considerably according to the rice variety used. Manjunatha et al. (2010) found that seedlings transplanted nine or 12 days after nursery sowing produced significantly higher yields than older seedlings, at both 20x10 cm and 25x25 cm spacing distances. However, Thiyagarajan et al. (2002), Latif et al. (2009; 2005) and Long et al. (2005) found that seedling age had no effect on yields, for studies carried out in Tamil Nadu, Bangladesh and Sichuan Province, China, respectively. Thiyagarajan et al. (2002) also found that 14-day-old seedlings contributed to slightly lower water productivity compared to 23-day-old seedlings. Pasuquin et al. (2008) designed an experiment specifically to compare the effects of different nursery methods and seedling ages on rice yield and other physiological characteristics. They found that grain yield was consistently higher for younger seedlings, but they identified signs that transplanting shock was not a very serious problem for older seedlings, which recovered quickly after transplanting. They suggested that the drag on the yield of the older seedlings was more likely to be due to crowding in the nursery rather than transplanting shock, as such. In another study that could be relevant to the understanding the phenomenon of transplanting shock, Latif et al. (2009) found that quick transplanting (as few as 15 minutes but up to 3 hours) 15 and careful placement of seedlings had no effect on grain yield in the winter season. They also found that planting one or two seedlings per hill had no effect on the yield of rice grain during the winter season in Bangladesh. A study by San-oh et al. (2004) added complexity to the discussion on the respective merits of single seedling transplanting and the spacing of hills. Their study showed that a planting pattern of single plants per hill at a relatively high density of 51.3 hills/m 2 produced a better yield than a planting pattern of three plants per hill at only 17.5 hills/m 2. Their study also showed that direct-sown rice plants produced more straw and grain than transplanted rice, apparently confirming the adverse effect on rice plants of transplanting shock.

15 A three-hour delay is longer than recommended in some SRI guidelines and by NGOs and extension officers we met in Madagascar and Tamil Nadu.

48 This is an interesting finding because transplanting shock is one reason why direct seeding is today an increasingly used and researched practice for rice establishment in smallholder farming systems. Recent attention to direct seeding has also been motivated by concerns about the labour costs and organizational difficulties created by the transplanting process (Pandey & Velasco 2002; Tajuddin & Rajendran 2002). Henri de Laulanié himself recognized that direct seeding could be better for rice plants than transplanting, regardless of the age of the seedlings, but he rejected the practice because he thought that it presented too many practical difficulties for the Malagasy farmers he was working with. One obstacle he perceived was the lack of suitable seeding machines in Madagascar in the 1980s (Glover forthcoming). Drum seeders are now relatively commonplace in some regions, however, and in recent years direct seeding has been tried experimentally alongside SRI methods in China and Thailand, with promising results (Sanjeewanie Ginigaddara & Ranamukhaarachchi 2009; Zhou et al. 2005). However, drum seeding is not precise enough at present to ensure that the rice seeds are placed singly, one-by-one. Thus, there may be a trade-off between the attractive properties of direct seeding and the low-density planting advocated in SRI. It remains to be seen how this dynamic may work itself out, but the combination of economic and technical incentives to move towards direct seeding, at the expense of ignoring recommendations to place plants singly, would seem to be fairly compelling for those small farmers who face a labour constraint and can afford to buy or rent a drum seeder. It should be noted that Makarim et al. (2002) produced contrary findings in their research in Indonesia, where pregerminated seed planted with a dibble stick was reported to have produced a lower yield than transplanted 15-day-old seedlings. Early studies carried out in Madagascar confirmed some of the expected changes in tillering, rooting, panicle formation and grain filling through the use of SRI methods, and reported yields that were much higher than farmers’ existing practices and national averages. However, spacing distances wider than 25x25 cm were not found to be beneficial (Andriankaja no date; Barison 1997; Rajaonarison 1999). Latif et al. (2009; 2005) and Thakur, Rath et al. (2010a) carried out experiments specifically designed to evaluate different spacing distances. Latif et al. (2009; 2005) found that they obtained the best yields with relatively narrow spacing of 25x15 cm and the lowest yields at a wide spacing of 40x40 cm. Similarly, Thakur, Rath et al. (2010a) found that they achieved the best yields with a spacing distance of 20x20 cm and the lowest at 30x30 cm spacing, in spite of a higher productivity per plant in the less dense field. Menete et al. (2008) reported very similar findings, although the size of the negative effect of wider spacing on yield was not consistent across two seasons. Mishra and Salokhe (2010) also reported that single seedlings at 20x20 cm spacing outperformed single seedlings at 30x30 cm, as well as 3–4 seedlings at 20x20 cm spacing. Ceesay et al. (2006) reported that 20 cm spacing produced better yield than 30 cm or 40 cm spacing, in experiments carried out on alluvial soils in Gambia.

49 In a theoretical and historical review article, Horie et al. (2005) discussed the evidence of a benefit to transplanting seedlings singly in widely spaced hills. Their review of Japanese scientific research suggested that rice grain yield was likely to remain roughly the same across a wide range of planting densities between ten and 100 hills/m 2, because of compensating mechanisms regulating the number of panicles, spikelets per panicle and grain filling. They noted that younger seedlings, if planted at low density and with abundant N, could produce excessive vegetative growth leading to fewer, smaller productive tillers and the risk of lodging. However, they also discussed research which suggested that sparsely planted rice could produce larger roots and absorb more N, especially during heading. They called for more research into the relationships between spacing density and yield. Thakur et al. (2010b) reported on a study in Orissa, eastern India using slightly modified SRI practices that included spacing distances of only 20x20 cm. This spacing distance was considered optimum on the basis of previous studies in the area. The modified SRI treatment was compared to local RMPs, which included spacing of 20x10 cm (see above). The modified SRI management methods were reported to out- yield RMPs by 42 per cent. Positive physiological and architectural changes were observed that could explain the improvement in yield, including larger roots, a higher leaf area index (LAI) and a more open plant structure, allowing the plants to intercept more sunlight. In this case, the slightly lower-density planting was compensated for by a higher grain yield per plant. Experiments in Sichuan Province, China confirmed that very low planting density (4x10 4 hills/ha.) led to profuse tillering but found that the highest grain yields were achieved with a medium-range planting density (9x104 hills/ha.) (BAO et al. 2008). Similarly, Wu et al. (2005) found that very low density and high density transplanting were less beneficial than a mid-range density of 12x10 4 hills/ha. The results of these studies place a question mark over the very wide spacing distances that have been reported in Madagascar, and help to explain the scepticism of critics who calculated that very widely spaced plants would not be able to capture sufficient solar radiation to achieve the very high yields that had been claimed (Rafaralahy 2002; Sheehy et al. 2004). However, people who have observed the situation of the farmer responsible for the best yield reported by Rafaralahy (Rafaralahy) insist that his achievement was real and even claim that he has now reached 27 t/ha. (Uphoff 2008). It is worth noting that spacing distances as wide as 35 cm were apparently recommended in Madagascar in the mid-1960s (Elyah 2006). It may be that the specific agro-ecological context of Madagascar could explain a difference in optimal spacing on the island. It is important to consider these findings on transplanting density in the light of indications that a saving in the seed rate is one of the key factors that has been reported to make SRI practice economically attractive for farmers (see sections 5.2 and 5.3 and Chapter 6).

50 4.4 Irrigation management It seems to be in the area of irrigation management in particular that SRI overlaps with a much broader research effort. IRRI researchers among others have carried out quite extensive research on rice cultivation systems that could save water while increasing water productivity (e.g. Belder et al. 2004; Belder et al. 2007; Belder et al. 2005; Bouman et al. 2006). In 2002, a workshop on water-saving rice cultivation methods was held at IRRI in Los Baños, in which a number of SRI-related papers were presented (Bouman et al. 2002). A broad generalization from the body of work on water-saving irrigation is that more efficient irrigation methods can sustain rice yields while also achieving substantial water savings compared to conventional, flooded irrigation. Among SRI studies, however, modest and occasionally even large yield increases are sometimes reported. Cao (2002) and Thakur et al. (in press) have also reported some positive morphological and physiological changes associated with water-saving irrigation management, in an experimental setting and an SRI field trial, respectively. Zhang et al. (2007) found significant water savings using SRI with aerobic soil conditions, along with an improvement in water productivity of 34 per cent, larger panicles and a seven per cent increase in yield. Experiments carried out in Senegal indicated that SRI methods could conserve water without adversely affecting yield, thus increasing the productivity of water. Realising these gains depended on effective eradication of weeds (Krupnik et al. 2010). Reddy et al. (2005) found that SRI methods did not increase the net returns to farmers but did save water in Andhra Pradesh. Senthilkumar et al. (2008) carried out experiments on a modified rice cultivation system that had some features in common with SRI. The study found that the method could enable savings in both water and seed, as well as modest increases in yields. The highest yields, however, were achieved when SRI techniques were combined with conventional irrigation. Chapagain and Yamaji (2010) reported very similar findings from their experiments, while also reporting that AWD irrigation methods reduced the incidence of pests and diseases and shortened the crop cycle. Some researchers have reported substantially higher yields with water-saving irrigation management in conjunction with other SRI cultivation methods. For instance, Zhao et al. (2009b) reported that SRI methods produced about 21–22 per cent higher yields and a higher harvest index than traditionally flooded treatments with a water saving of about 43 per cent, increasing water use efficiency significantly. Yield and water productivity improvements of similar magnitudes were also reported by Vijayakumar et al. (2006) and Thakur et al. (in press) for their field experiments in Tamil Nadu and Orissa, respectively. In addition to their findings on yields and water productivity, Zhao et al. (2009b) also reported that SRI irrigation was associated with a more efficient use of N at lower N application rates. Ramasamy et al. (1997) also found an interaction between N use

51 and reduced irrigation, showing that rice showed a better response to N applications under well-drained soils, and that late-season rice crops in poorly drained soils suffered a reduced yield at higher rates of N application. This accords with Belder et al.’s (2005) observation that optimal N management is as important as water-saving irrigation in raising water productivity. Occasionally, very dramatic yield and water productivity increases are reported with SRI management. In Panama, Turmel et al. (in press) found that water consumption could fall by as much as 86 per cent compared to existing practices in the first year of SRI adoption, accompanied by yield increases of 47 per cent on average. SRI methods outyielded conventional practice on eight out of 10 sites; on the other two sites, SRI practices produced slightly less than conventional methods but still with a considerable water saving, which was valued by the farmers. The minimum and maximum yields with SRI methods were 1.21 and 8.98 t/ha., respectively, compared with 0.61 and 7.48 t/ha. using conventional methods at the same locations. Sato and Uphoff (2007) reported that they had found yield increases of 78 per cent alongside water savings of 40 per cent and a reduction in fertilizer use of 50 per cent, compared to conventional methods. Ceesay et al. (2006) reported that AWD irrigation methods produced yields of 7.3 t/ha. compared to just 2.5 t/ha. under conventional flooded irrigation, in on-station trials carried out in Gambia. The authors suggested that AWD irrigation methods created an intensive pathway of N mineralization and improved N availability in the soil. Mishra and Salokhe (2010) reported that they had found an interaction effect between plant density and water regime, which suggested that dense conventional planting was able to produce better yields under continuous flooding, whereas wider spacing of one seedling at 20x20 cm spacing produced better yields when an intermittent flooding regime was applied during the vegetative growth stage. Very similar results were reported by Menete et al. (2008), who also found that intermittent irrigation had a large, negative effect on yield attributes and grain yield on rice grown in a moderately saline soil. By contrast, it is known that water-saving irrigation can be beneficial when growing rice on acid soils with potential iron (Fe) toxicity (Dobermann 2004; Gani et al. 2002). A very common observation in the literature on water-saving irrigation and SRI is that effective control over the water supply is a critical factor in order to be able to adopt water-saving irrigation methods (e.g. McHugh et al. 2002) (see Chapter 6). However, Mushtaq et al. (2009) found that having access to a reliable tank irrigation source was not enough by itself to encourage some Chinese farmers to adopt water-saving irrigation methods such as AWD. They concluded that AWD adoption was more likely to be influenced by institutional arrangements and farmer training, and a desire to mitigate risk in the face of water scarcity.

52 4.5 Weed control, soil biology and soil nutrient management The principle that suppressing weeds is important for preventing yield losses is uncontroversial, regardless of the cultivation system being applied. With SRI management, the necessity of weeding is made more acute by lower density planting patterns, the use of very young seedlings – which are more vulnerable to competition from weeds – and non-flooded irrigation. Experiments carried out in Senegal indicated that weeds were a greater threat to rice under SRI management than rice under conventional BMPs, so that effective weed management was vital in order to attain the potential advantages of the other SRI practices (Krupnik et al. 2010). Yadao and Zamora (2007) carried out on-station experiments comparing SRI methods with conventional cultivation and failed to find any of the performance advantages reported elsewhere, which they said could possibly be attributed to weed competition, as well as poor soils and the inexperience of farmers. Mechanical rotary weeders are proposed as a way to make the increased weeding requirement under SRI cultivation more feasible for farmers. In some variants of SRI, however, stress is also placed on the role played by rotary weeders in churning and aerating the soil. In this perspective, the weeders are understood to contribute not only to the suppression of competition from weeds but also the promotion of an oxygenated soil environment that favours aerobic bacterial life and perhaps also mycorrhizal fungi. This is argued to benefit rice plants by fostering symbiotic interactions between rice roots and soil micro-organisms that can help provide nutrients to the plant (Uphoff 2007; Uphoff et al. 2009). We found and reviewed relatively little research on this aspect of SRI management, but the topic stimulated a considerable amount of discussion among scientists and extensionists during our field visits (see Chapter 6). Early studies carried out by researchers from the University of Antananarivo’s Faculty of Agronomy under the auspices of the Madagascar SRI Consortium reported huge increases in the number of N-fixing Azospirillium bacteria in the roots of plants under SRI management, though not in the rhizosphere. On better soils, the number increased by about 17 times under SRI management without soil nutrient amendments, and by a factor of more than 21 if organic compost was added. By contrast, amendments with NPK multiplied the number of Azospirillium by about 6.9. On poor, loamy soils, the addition of compost had a much more dramatic effect, with a population of root-dwelling Azospirillium rising from 75x10 3 with no amendments to 2,000x10 3 with compost (Randriamiharisoa 2002; The SRI Consortium 2003). This could suggest that the factor productivity of SRI techniques alone may have been fairly modest on poor soils, compared with fertilization. Tsujimoto et al. (2009) concluded that the exceptional results achieved with SRI methods in Madagascar could be attributed to very high levels of soil organic carbon and a high N-supplying ability of the soil in particular plots, resulting from many years of organic compost amendments, combined with the beneficial effect of deep ploughing. Turmel et al. (in press) also found that SRI yields were positively correlated with soil available

53 potassium (K), but in addition some evidence that SRI methods enabled the rice plants to exploit other soil nutrients more effectively than under conventional management (see also section 5.3). Nonetheless, Randriamiharisoa and colleagues also reported that relatively modest additions of compost of around 1–2 t/ha. could stimulate a good tillering response, which they suggested was an indication of an ‘incitement threshold’ for stimulating soil bacteria (Randriamiharisoa 2002; The SRI Consortium 2003). Improvements in soil biology under SRI management were also found by Zhao et al. (2009a). He and Ma (2005) compared different fertilization regimes and found that, for the same level of NPK delivered to the soil, organic fertilizer improved the vegetative growth and grain yield of rice while also improving the physical and chemical status of the soil. As we saw in Chapter 3, Henri de Laulanié believed that rice was a very undemanding plant that did not require large amounts of soil fertilization. In view of the resource constraints faced by poor and marginal farmers, he thought it was better to improve basic, low external-input crop management practices first, before recommending the use of chemical fertilizers (de Laulanié 2003). Norman Uphoff has followed his lead, arguing that large improvements in yield can be achieved by applying SRI methods with no or relatively small quantities of organic compost and/or manure (Uphoff 2003). However, Uphoff (1999) and ATS have also recognized the risk of soil nutrient mining and noted the likelihood that particular inorganic amendments may be desirable in particular soils. We have not found or reviewed studies that helped us to understand the relative, positive or negative, contributions to plant health and productivity of aerobic and anaerobic soil bacteria, or mycorrhiza, or possible interactions between them. One soil scientist we consulted, who is an expert on mycorrhiza and knowledgeable about SRI, pointed out that the particular mix of bacteria in a given soil is likely to be highly localized and very dynamic, and that specific aerobic and anaerobic bacteria could have beneficial or adverse effects on plants. He also noted that mycorrhiza tend to be negatively rather than positively affected by soil disturbance (Thom Kuyper, pers. comm.). The dose and timing of nutrient amendments is also commonly mentioned as an important factor. It is known that excessive N could be damaging to very young rice seedlings (Uphoff 2003), as can organic manure that is not properly rotted. 4.6 Synergies Norman Uphoff (1999) first raised the idea of synergy in the scientific literature in 1999, although in our reading Uphoff’s account seemed to depict the SRI components as mutually dependent or mutually coherent, rather than interacting together to produce a multiplier effect. Some researchers have reported that synergetic interactions have been found among SRI practices (e.g. Mishra & Salokhe 2010; Thiyagarajan et al. 2002; Turmel et al. in press), but we are aware of relatively few studies that have investigated these interactions systematically.

54 Two studies that have attempted to do so have been published by Randriamiharisoa and Uphoff (2002) and Menete et al. (2008). Randriamiharisoa and Uphoff’s (2002) experiments evaluated the four SRI practices of young seedlings, aerobic soil conditions (moist but unsaturated), a single seedling per hill and organic fertilization, at both 25x25 cm and 30x30 cm spacings (both of which were considered to be within the range of SRI treatments). The data presented, at a high level of aggregation, indicated that the highest yields were achieved when all four of the tested SRI practices were combined, as well as a progressive increase in yields as the proportion of SRI practices increased. A more detailed investigation of the possible synergetic interactions among SRI practices was carried out by Menete et al. (2008) on moderately salt-affected soils. They found that there were very few such interactions, some of which were actually negative. The key factors that dragged yield down in their study were the wider spacing of hills (30x30 cm) and intermittent irrigation, which may have had adverse consequences on saline soils. However, compost applications had a beneficial effect on saline soils and a negative one on the non-saline control. Menete et al. concluded by suggesting that the concept of synergy was less relevant than ‘Liebig’s Law of the Minimum, i.e., one management component limited the growth and yield potential associated with other management practices, water management being the main limiting factor in this study’ (Menete et al. 2008: 42). In section 4.4, we discussed reports of substantial yield increases having been achieved when applying water-saving irrigation methods. An important point to note in this regard is the difficulty involved in ascribing proportions of a given yield increase to one particular cultivation practice or set of cultivation practices rather than another. As noted above, ‘mainstream’ research on water-saving irrigation systems has tended to show that yields can be sustained but not increased by managing water resources more efficiently. SRI studies that have found yield increases therefore imply that the combination of reduced irrigation with SRI crop establishment, weed management and fertilization methods adds additional value in terms of yield potential for a given input of water. The mainstream approaches to studying water- saving irrigation have recently been criticized in very strong terms by Stoop et al. (2009), who argued that conventional ‘reductionist’ experimental designs prevent researchers from appreciating the interaction among different cultivation practices or the complexity of local conditions. Uphoff et al. (2008) have similarly criticized a study comparing SRI methods with conventional BMPs for failing to appreciate the supposed synergetic relationships among SRI practices. These arguments seek to raise fundamental questions about the appropriate design of experiments and trials to investigate integrated crop management systems, such as SRI. In Chapter 9 we will return to the question of what is known and what is not known about SRI, including its biophysical aspects. We now turn to examine the available evidence on adoption patterns and the household-level impacts of SRI practice.

55

5. Adoption and impact studies on SRI

SRI consists of a suite or package of practices that could, in principle, be implemented independently of one another or in various different combinations (see Chapter 3). In practice, diverse specifications of SRI methods can be found in documents and observed in practice (see Chapter 8). This section examines what the available literature reveals about patterns in the adoption, adaptation and disadoption of SRI components by farmers. 5.1 Impressionistic indications of the spread of SRI As mentioned in the Introduction, SRI is reported to have spread from Madagascar since the 1990s to nearly 50 countries around the world. 16 Anecdotes attesting to this spread are abundant and it is clear that substantial investments have been made and large SRI-promotion programmes established. However, the levels of SRI activity reported in different countries vary quite widely. By contrast with pace-setting countries (and major rice producers) like China, India, Indonesia, the Philippines and Vietnam (see above), levels of SRI activity reported in countries like Afghanistan, 17 Costa Rica, 18 Guyana, 19 Laos, 20 Panama 21 and Rwanda 22 can fairly be described as somewhere between modest and very low at present. Some important multilateral donor agencies, such as the World Bank (see above), the Asian Development Bank (see above) and the International Fund for Agricultural Development (IFAD), as well as large international non-governmental organizations (INGOs) such as WWF (see above) and Africare, have provided financial and technical support to SRI promotional activities in various locations. An alternative impressionistic indicator of the spread of SRI activity and interest is the expansion of scientific research interest. Scientific studies investigating SRI methods or their impacts, which have been reported in scientific journals, have been conducted in various countries including Bangladesh (e.g. Latif et al. 2005), China (e.g. Zhao et al. 2009b), the Gambia (e.g. Ceesay et al. 2006), India (e.g. Thakur et al. 2010b), Madagascar (e.g. Moser & Barrett 2003), Mali (e.g. Styger in press),

16 See http://sri.ciifad.cornell.edu/countries/index.html (accessed 19 January 2011). 17 See http://sri.ciifad.cornell.edu/countries/afghanistan/index.html (accessed 19 January 2011). 18 See http://sri.ciifad.cornell.edu/countries/costarica/index.html (accessed 19 January 2011). 19 See http://sri.ciifad.cornell.edu/countries/guyana/index.html (accessed 19 January 2011). 20 See http://sri.ciifad.cornell.edu/countries/laos/LaosArchives.html (accessed 19 January 2011). 21 See http://sri.ciifad.cornell.edu/countries/panama/index.html (accessed 19 January 2011). 22 See http://sri.ciifad.cornell.edu/countries/rwanda/index.html (accessed 19 January 2011).

56 Mozambique (e.g. Menete et al. 2008), Panama (e.g. Turmel et al. in press), the Philippines (e.g. Pasuquin et al. 2008) and Thailand (e.g. Mishra & Salokhe 2008). Table 5.1: Indicative figures on the international spread of SRI

Figures indicating SRI adoption Location Source 110,000 farmers using SRI on 60,000 ha. Cambodia Uphoff (2010a) (2009) SRI methods used on 285,867 ha. (2009) Sichuan Province, China Uphoff (2010a) 688,000 ha. under SRI management (2009) Zhejiang Province, China Uphoff (2010a) 65,000 households having been introduced to 12 states of eastern and Uphoff (2010a) SRI methods (2010) northern India Around 13,000 farmers using SRI, while Uttarakhand and Himachal PSI, 2010 23 another 1,200 farmers applying SRI Pradesh, India techniques to other crops (2010) SRI practices extended to ‘perhaps over Tamil Nadu, Tripura, Uphoff (2010a) 500,000 households’ by state government Orissa and Andhra extension services (2010) Pradesh, India SRI programmes run by diverse NGOs ‘may India Uphoff (2010a) have reached more than 100,000 households’ (2010) ‘about 600,000 farmers are growing rice with India WWF-ICRISAT all or most of the recommended SRI crop project 24 management practices on about 1 million ha. distributed across 300 districts of the country’ (no date) 12,133 farmers using recommended SRI West Nusa Tenggara, Sato and Uphoff practices (2007) South and Central (2007) Sulawesi, Gorontalo, Indonesia ‘As many as 200,000 farmers may be using Madagascar Uphoff (2010a) some or all of the recommended methods’ (2010) ‘The number of farmers using SRI reached Myanmar Kabir & Uphoff over 20,000 by the start of 2005. (…) reaching (2007) to ~29,000 by the end of 2006’ 250 ha. under SRI management (2009) North Korea Uphoff (2010a) 800,000 farmers using SRI methods, of which Vietnam Uphoff (2010a) 20 per cent using ‘full SRI’ and 80 per cent ‘partial SRI’ (2010)

All of these qualitative indicators show that the SRI phenomenon is real and substantial, but it is difficult to evaluate the spread of SRI knowledge and practice in quantitative terms. A lot of claims have been made in this regard. Table 5.1 lists a

23 http://www.peoplesscienceinstitute.com/activities/nrm/sri_intro.html (accessed on 9 February 2011). 24 http://www.sri-india.net/html/wwf_project.html (accessed 30 January 2011).

57 selection of claims from various sources. Several of them have been taken from a helpful summary recently presented by Uphoff (2010a) in the form of a personal retrospective on SRI experience internationally. Uphoff can legitimately claim to be one of the best-placed individuals to provide such an overview of SRI at the international level, and his information sources are exhaustively documented on the CIIFAD SRI website. 25 It is significant that so well-informed a source as Uphoff (2010a) uses circumspect language such as ‘perhaps,’ ‘may have reached’ and ‘may be using’. Within the scope of this project, based as it was on a review of literature and web materials and visits to just two of the many countries where SRI is reported to have been taken up, it has been impossible to verify independently the growth and spread of SRI internationally. However, our research has provided several reasons for treating the quantitative claims described in Table 5.1 with caution. One reason for taking a cautious approach to broad headline numbers on SRI adoption is that such figures may obscure important details about exactly what is adopted and how extensively SRI practice is evident on the ground. Typically, indications of spread are expressed either in terms of numbers of farmers adopting SRI or the rice area under SRI management. Both terms are subject to interpretive problems unless they are specified fairly precisely: • Where numbers of farmers or households are said to have adopted SRI practices, it may not be clear whether they are practising the entire suite of recommended methods or a sub-set of them. • Where SRI practices are said to have been taken up on an area of land, it may also be the case that the farmers concerned practice SRI or SRI components on all of the land under their control or only part of it, or not in all rice seasons during a given year. • Where the numbers of reported adopters or the numbers of hectares under SRI management are not expressed in proportion to the total number of rice farmers or rice hectares in a given region, it is difficult to assess the size or importance of the phenomenon. In and of itself, the fact that farmers may not have adopted all of the recommended SRI practices should not be used to dismiss such numbers entirely, since it is said to be an important characteristic of SRI that it is a set of flexible principles that need to be adapted for particular settings. One should therefore expect to find diverse implementations of SRI methods in different settings (Glover in press). The fact remains, however, that it is exceedingly difficult to gauge levels of adoption and practice unless adoption is fairly precisely specified, and unless estimates are based on empirical evidence rather than inferred from project goals or projected from proxy indicators such as the number of farmer training meetings conducted or demonstration plots established.

25 http://sri.ciifad.cornell.edu/index.html (accessed 23 January 2011).

58 We draw attention to these issues based on several observations made during the course of our research. For example: • In Tamil Nadu, a target was set to achieve SRI adoption on 750,000 ha. in 2010 (see Chapter 6). 26 However, no comprehensive impact assessment has yet been carried out to measure the effectiveness of the SRI promotion activities undertaken in Tamil Nadu to date. We understand that a project evaluation is now under way; it remains to be seen what it will discover. However, it seems likely that the extent of adoption may turn out to be substantially less than the target of 750,000 ha. would suggest, for several reasons. First, the design of the project is such that field officers are both accountable for achieving targets and responsible for reporting their achievements. Such a design creates both opportunities and incentives for achievements to be exaggerated. Second, an independent study conducted by a researcher from the International Water Management Institute (IWMI) has sought to evaluate SRI adoption in districts of Tamil Nadu that had been targeted for SRI promotion (Palanisami unpublished). The study’s author sought to construct a sample of SRI ‘adopters’ and requested a list from agriculture department records. He was surprised to discover that some farmers designated as SRI adopters denied that they were SRI adopters (Palanisami, pers. comm., August 2010). In fact, the study found that the great majority of the farmers in the sample actually were applying some SRI methods, though most of them were applying only one or two of the core SRI practices and very few could be designated ‘full adopters’ (Palanisami unpublished) (see section 5.2). The fact that the farmers in question were actually practising some SRI methods could be a legitimate reason for a researcher to label them as ‘partial SRI adopters’, if their practices could be shown to have been influenced by exposure to SRI information or training. However, the fact that some of the farmers in the IWMI sample did not recognize themselves as SRI adopters – which is not surprising in view of the limited degree to which SRI methods were actually being applied – also represents an important reminder that researchers should provide a good justification for applying such a label to them. • A short field visit to eastern districts of Nepal in November 2009 provided several examples of farmers who were labelled ‘SRI adopters’ by their extension officer and recognized the label themselves, yet who, for various good reasons, were actually unable to practise SRI in strict accordance with the training they had received. The example illustrates the value of gathering detailed information, and being clear about what one means by ‘adoption’, in order to be able to make clear and reliable statements about levels of ‘SRI adoption’ (see section 6.3 and Glover in press).

26 Note that this figure differs markedly from the one cited by Norman Uphoff (2010a) (see Table 5.1).

59 • The state of Tripura is celebrated within the Indian SRI community as one of the pioneers of SRI practice in India. SRI methods were reported to have been taken up by just over 162,000 Tripura farmers on 32,500 ha. in the 2007–08 rice season (Shambu Prasad 2009).27 However, we learned that these estimates were based not on a survey or monitoring but on the numbers of farmers claiming a state government subsidy that was offered to SRI farmers after 2006 (Majumder, pers. comm.). This appears to be an unreliable basis for making an estimate of SRI practice. • As noted in Table 5.1, SRI practice in Madagascar has been estimated at up to 200,000 farmers. However, one of the remarkable observations we made during our field visit to Madagascar was the lack of keen interest, on the part of some of the organizations promoting SRI in the country, in impact monitoring and evaluation. This attitude was associated with the general philosophy of some organizations. For instance, representatives of ATS took the view that their priority should be to offer SRI training and information to any farmers who were interested in receiving it; it was then entirely up to the farmers to decide whether they wanted to follow the new methods or not. Similarly, the Adventist Development Relief Agency (ADRA) had begun providing training in SRI methods because farmers had asked for it; the NGO did not see it as a priority for the organization to verify whether this responsive service was having an impact on farmers’ practice. New support institutions are being established in Madagascar to promote SRI in the country and these may begin to collect and collate more systematic data on SRI practice. For the time being, it is extremely difficult to quantify the level of SRI practice in Madagascar (see section 6.1). Before moving on, we would like to emphasize that seeking clarity and precision in evaluating claims about SRI adoption and practice does not imply in any way that we wish to impose rigid norms of our own about what should constitute ‘adoption’ of SRI or indeed ‘non-adoption’ or ‘disadoption’. We have pointed out elsewhere that such terms are inherently problematic in relation to smallholder farmers and agricultural technologies, especially if they are treated dogmatically (Glover in press). Our point is simply that we think anyone attempting to quantify SRI practice should themselves describe, explain and clearly justify the criteria they intend to use and devise their data-collection strategy and measurements accordingly. There is no reason why the chosen criteria should not accommodate diversity in SRI practice or be sensitive to context. Such a procedure makes the analytical strategy transparent and amenable to critical analysis. When studies fall short of such standards, an opportunity has been missed to make progress in quantifying the spread and impacts of this distinctive rice cultivation system. We shall return to our examination of the spread of SRI in Chapter 7, which will explore the mechanisms and channels through which knowledge about and practice of

27 See http://ciifad.cornell.edu/sri/countries/india/ (accessed 23 January 2011).

60 SRI appears to have spread beyond Madagascar. First, building on the principles outlined in the preceding paragraph, the following sections present a critical discussion and evaluation of the academic studies that have been carried out on the agronomy, adoption dynamics and socio-economic impacts of SRI to date. 5.2 Studies on SRI adoption and adoption processes We identified only six studies in our database that compare on-farm adoption patterns of SRI in relation to an alternative practice in a relatively detailed way (see Table 5.2). The most detailed of these studies were carried out in the early 2000s (2000– 2005) and are geographically restricted to Madagascar, Sri Lanka and Cambodia. Of these, Moser and Barrett’s (2003; 2006) studies are also the oldest, concerning adoption processes that took place in the late 1990s. Two, more recent studies from Tamil Nadu examine adoption patterns on the Indian subcontinent. Although relatively substantial SRI (research) activity has been reported from countries such as China, Indonesia, the Philippines and Vietnam, we have not identified any detailed studies that examine adoption processes or patterns in these locations.

Table 5.2: Adoption studies Study Location Extension organization Number of farmers Moser and Five villages in the central Association Tefy Saina 937 rice producing Barrett ; highlands of Madagascar: households 2003, 2006 Manandona, Anjazaforsy, Ambatovaky, Iambara, Torotosy Namara et Kalthota Irrigation System & Ceylon Electricity 60 SRI adopters, al., 2004 Kurunegala district Sri Lanka Board & Ministry 60 non-SRI responsible for poverty adopters. alleviation Anthofer, Provinces of Kampong Thom, CEDAC (NGO) 400 SRI adopters, 2004 Kandal Prey Veng, Takeo and 100 non-SRI Kampot adopters. Sita Devi Cuddalore District, Tamil Nadu Tamil Nadu 50 SRI adopters, and Department of 50 non-SRI Ponnarasi, Agriculture adopters 2009 Palanisami, 10 districts in Tamil Nadu: Tamil Nadu 500 SRI adopters, unpublished Cuddalore, Kancheepuram, Department of 100 non-adopters. Pudukkottai, Tanjavur, Agriculture Thiruvallur, Thiruvarur, Thiruvannamalai, Tirunelveri, Tiruchirapalli, Villupuram

Second, the studies listed in Table 5.2 surveyed farmers in a relatively small number of locations. Moser and Barrett’s (2003; 2006) study involved the largest sample of farmers, but they were located in just five villages. The geographical range of SRI in Tamil Nadu seems relatively well captured by the recent study conducted by Palanisami (unpublished) but, as discussed below, the study has some limitations that make it difficult to interpret the results.

61 Third, several of the studies purposely constructed samples of adopters and non- adopters. Consequently, these figures do not provide any insight into the overall ratio of adopters to non-adopters in their respective locations. In addition, such a quota- based sampling strategy raises the question of how scholars have distinguished adopters from non-adopters (see Section 5.2.2). The studies listed in Table 5.2 are the only studies we have encountered which analyse (partial) adoption processes in statistical detail. In the discussion below, however, we also incorporate information from a number of studies that did not specifically investigate adoption patterns. These include a study by Resurreccion and Sajor (2008) on gender relations and SRI adoption, a study by Senthilkumar et al. (2008), presenting results from a survey on farmer experiences with SRI, a study by Minten and Barrett (2008) on rice productivity in Madagascar, and finally an evaluation by Sato and Uphoff (2007) on SRI extension in Indonesia. 5.2.1 What an SRI adopter adopts Anthofer (2004) and Resurreccion and Sajor (2008) provide information about the specific SRI practices adopted by some Kampuchean farmers. Anthofer (2004) reports that the average SRI adopter transplants seedlings that are 16.8 days old at a rate of 1.3 seedlings per hill. Most farmers also planted seedlings in rows and applied AWD irrigation, while very few farmers weeded more than once. Surprisingly, this characterization is rather different from that by Resurreccion and Sajor (2008), who surveyed SRI adopters in contact with the same NGO. They found that only a few farmers practised AWD, square planting or transplanting of young seedlings, while the majority adopted intensified weeding practices, followed by transplanting of widely spaced, single seedlings. Palanisami (unpublished) describes the adoption of SRI components in more detail for the case of Tamil Nadu. He shows that the majority of farmers in his sample (around 90 %) adopt just one or two SRI components out of five promoted. The most widely adopted components were reducing the seed rate and an improved nursery method, which is part of the SRI package being promoted in Tamil Nadu. Adoption of other key SRI components was very low, ranging from just eight per cent who followed an intensive weeding strategy, 18 per cent who planted seedlings in squares and 25 per cent who transplanted young seedlings at around 14 days of age. A study by Senthilkumar et al. (2008) sheds some light on the findings by Palanisami (unpublished). They present the results from a survey carried out in Tamil Nadu, in which farmers’ views on SRI were collected. Many of the farmers in their sample complained about high rates of mortality of young seedlings, and many farmers reported difficulties with, and increased labour demands for, transplanting seedlings in squares. In one location, many farmers complained about the ‘increased drudgery’ associated with using the mechanical rotary weeder. This result may have been connected to local soil conditions.

62 The adoption rates reported by Palanisami (unpublished) are very low compared with observations made by Satyanarayana et al. (2007), also in Tamil Nadu. The latter observed that 36 out of 100 farmers adopted all of the SRI practices, and that all farmers in the sample adopted 14-day-old seedlings and 20x20 cm spacing. Satyanarayana et al. (2007) however did not base these findings on a randomly constructed sample. Kabir and Uphoff (2007) carried out a survey of a sample of farmers attending farmer field schools in Myanmar. They reported that most of the farmers planted young, single and widely spaced seedlings, but followed other recommendations less strictly. However, the exact variation in the data was not presented. Sato and Uphoff (2007) reported findings on the impact of SRI extension activities in Indonesia. Their report claimed that all 12,133 farmers reached by the extension programme were following all of the SRI components (8–12-day-old seedlings, one seedling per hill, 30x30 cm square planting, using a rotary weeder 2–3 times), except for the recommended fertilizer strategy. The definition of SRI practice differs across these studies, thereby limiting the potential to make generalizations from these findings. However, this brief review suggests that adoption of SRI is partial rather than complete in many regions, while even within regions there may be wide variations in the specific components actually adopted. This important observation raises the key question of who qualifies as an adopter of SRI. Several of the authors concerned apply a numerical principle, that a farmer should have adopted at least a certain number of SRI component practices in order to be considered an SRI adopter. Anthofer (2004) and Palanisami (unpublished) applied a definition of an SRI farmer as somebody who practices at least one of the SRI components, whereas Namara et al. (2004) defined an SRI farmer as any farmer who stated that he had tried SRI methods (and potentially disadopted them), without specifying the actual number of components that had been tried or abandoned. Some of the studies did not explain how SRI adopters were initially distinguished from non- adopters at all (Moser and Barrett, 2003; 2006; Sita Devi and Ponnarasi, 2009). 5.2.2 Who adopts SRI components, and why? The literature on technology adoption in smallholder agriculture confirms that, in most cases, better-endowed farmers and wealthier farm households are more likely to adopt new technologies or management practices, and/or to do so sooner than less well-off farmers. This is because well-endowed farmers are generally in a better position to take the initial risks involved in experimenting with new production techniques or technologies, and/or because such farmers are often better connected to extension agents, making them more likely to learn about new technologies earlier than other farmers. On the other hand, a technology may display increasing returns to scale, making it more likely that larger and better endowed farmers adopt initially. In the case of SRI this may for example apply to investment costs related to the rotary

63 weeder, or investment in training of labourers, both of which may be relatively less costly on larger farms. These, and other, characteristics of early adopters have been described extensively in literature on adoption of agricultural innovations (e.g. Feder et al., 2004). SRI is said to be a cultivation system that is peculiarly appropriate for poor, resource- constrained farmers and households, because clear benefits can be gained through changes to management practices, without having to adopt expensive external inputs such as new seeds or chemical fertilizers (see Chapters 3 and 4). It is therefore very interesting to examine the adoption patterns in SRI to see whether poorer farmers are able and willing to adopt SRI practices to a similar or greater extent than richer farmers. Although they did not precisely define SRI adopters and non-adopters, the most detailed studies on SRI adoption are two papers by Moser and Barrett (2003, 2006), who studied the dynamics of SRI adoption in five villages in the Madagascar highlands. They found that the initial adopters were wealthier and more capable farmers. An important explanation put forward for this is that these farmers are able to cope with increased labour demands posed by the new technology, for example through hiring in wage labour. In another Madagascar study, Barison and Uphoff (in press) observed that the yields of both SRI and conventionally managed rice fields in their survey were substantially higher than the national average, which suggests that these SRI adopters may have been better off or more skilful to begin with. Minten and Barrett (2008), using the results from a nationwide survey carried out in Madagascar in 2001, showed that SRI adoption was determined largely by access to improved irrigation infrastructure and land with secure title, in villages where the average education level was higher. The latter factor may be associated with a wealth effect, but could equally suggest that better-educated farmers were more likely to adopt SRI methods. Minten and Barrett also showed that SRI adoption was largely concentrated in the Madagascar highlands and strongly correlated to the presence of an extension officer nearby. Overall, they found that SRI methods were practised in just 0.46 per cent of villages in Madagascar (see also Chapter 6). The data presented by Minten and Barrett does not shed light on the degree to which individual SRI components were being practised. A number of studies have been published that report the activities and impact of a project in Cambodia which is run by CEDAC, a local NGO (Anthofer, 2004; Resurreccion and Sajor, 2008). Anthofer (2004) compared the performance of rice production under SRI and conventional practice at farm household level. Although the sample of farmers consisted of SRI-adopters only, Anthofer (2004) also specified the characteristics of adopters compared to non-adopters. His analysis suggests that the SRI adopters were better endowed than non-adopters with regard to household size, livestock and education level. Resurreccion and Sajor (2008), analysing the impact of SRI on gender relations, cited a more recent evaluation report by CEDAC,

64 which stated that non-adopters of SRI typically depend more heavily on off-farm activities, which also implies that they may have been better-off households. Curiously, in their detailed analysis of SRI adoption patterns in Sri Lanka, Namara et al. (2003) found that both the wealthiest and poorest households in their sample were more likely to adopt. Levels of adoption were lower in the middle range of households. Namara and colleagues attributed this finding to the fact that the poorest farmers may have sufficient labour to enable them to cope with labour-intensive SRI practice, while larger farmers are less risk-averse and more willing to try a new crop management practice. This suggests that the adoption dynamics, or types of costs and benefits at the household level, may be different for differently endowed households. However, Namara et al.’s (2003) explanation also raises the question why the poorest farmers, who are typically most risk-averse, would nevertheless adopt SRI practices. It seems plausible that richer and poorer households may not be adopting the same SRI components, but unfortunately the study does not describe the specific practices adopted by individual farmers. Finally, Namara et al. (2003) observed that very few farmers adopted SRI on more than one plot. This pattern might also be explained by specific characteristics of an individual farmer’s operating environment, or it could indicate a difficulty or unwillingness to scale up SRI practices. Sita Devi and Ponnarasi (2009) agreed that wealthier farmers are more likely to adopt SRI, but unfortunately their paper does not specify the differences in crop management between SRI adopters and conventional farmers. Since the authors constructed their sample using a list of SRI adopters obtained from the local extension agency, their results may merely signal that wealthier farmers have better contacts with the extension agency. The available literature that allows us to examine patterns in SRI adoption thus conforms to the typical pattern in technology adoption, which is that better endowed households are more likely to adopt new practices. This is perhaps a disappointing finding for a cultivation system that has been portrayed as a peculiarly suitable one for poor and marginal farmers. However, it would not be appropriate to draw firm conclusions about the findings of just a few studies, especially when so little detail is available about the specific SRI components adopted (or not adopted) by different types of farmers or on different plots. On the other hand, it would be appropriate to examine whether poorer households are merely later adopters, who catch up with their richer neighbours over time, but this is not possible with the studies available at present. However, there are some studies that shed a little light on the dynamics of disadoption, which may also tell us something about the capacities of poor and marginal farmers to take advantage of SRI methods. The next section considers these. 5.2.3 Who disadopts, and why? In both Madagascar (Moser and Barrett, 2003; 2006, SRI Consortium Madagascar, 2003) and Sri Lanka (Namara et al. 2004), some farmers have been observed to disadopt SRI practices after having tried them. In the case of Sri Lanka, Namara et al. (2004) noted that disadoption levels were highest among the poorest adopters and, not

65 surprisingly, among those farmers who had experienced disappointing yields with SRI. However, why and in what ways the yields were disappointing was not further analysed. In Madagascar, disadoption seems to have been strongly related to the withdrawal of extension support. This could indicate that it takes some time before farmers feel confident enough to continue practising SRI without technical support. One explanation analysed in detail by Moser and Barrett (2006) relates to a phenomenon that they call conformity, whereby farmers conform to the advice of the extension agent, who is a person with high status and influence. This pressure to conform fades once extension support is withdrawn or reduced. Similarly, farmers may conform to the behaviour of their neighbours, meaning a trend towards adoption or disadoption may be amplified as farmers conform to the behaviour of early adopters or disadopters. Moser and Barrett’s (2006) regression analysis suggests that such conformity effects were indeed present. 5.2.4 Role of extension approaches and methods Moser and Barrett’s (2006) exploration of conformity effects concentrated on the notion that adoption and disadoption patterns might be particularly influenced by the authority, status or charisma of an extension officer. Our approach to SRI as an integrated socio-technical system (see Chapter 1) prompts us to consider the influence of social relationships and institutional frameworks more broadly. One issue that may be relevant is the possible influence of the extension approaches and training methods that are used. SRI is said to be a distinct kind of technology (or socio-technical system) that is knowledge-intensive, since it is based on altering management practices and ‘changing mind-sets’ rather than changing particular inputs, so that a distinct approach to agricultural extension is required (de Laulanié 1993; Uphoff 2004a; Vallois 1997). This idea was articulated to us on numerous occasions by extension workers and NGO staff during our field visits to Madagascar and Tamil Nadu. This suggestion implies that different approaches to and/or methods of extending SRI would differ in their rates of success. The SRI literature indicates that certain specific approaches have been employed in particular locations, including participatory technology development (PTD) in Afghanistan (Thomas & Mohammad Ramzi in press), community-based evaluation in Mali (Styger in press) and farmer field schools (FFS) in Myanmar, Senegal and Vietnam (Africare et al. 2010; Kabir & Uphoff 2007; Krupnik 2008). The more conventional method of setting up demonstration plots is used, for example, by the IAMWARM project in Tamil Nadu (see Chapter 6). Alongside the methods of teaching and learning that are employed, another possibility is that the types of social relationships and institutional frameworks that are built or reinforced through different approaches to SRI promotion may make a difference to patterns of adoption. Resurreccion and Sajor (2008) observed that labour-sharing

66 agreements between households are common in Cambodia, which we also observed in Madagascar (see Chapter 6). The transaction costs of maintaining such cooperative agreements may already be high when many or diverse households are involved (Gerichhausen et al., 2008). The additional transaction costs of negotiating a change in farming practices may be excessive, especially when returns to SRI are uneven across the members of a labour-sharing group. This problem arises equally in wage- labour markets, whereby the costs to labourers of costs of learning new methods may not be fully offset by increased employment opportunities or wages. This may explain why SRI promoters in Tamil Nadu have organized training for wage labourers themselves, in order to facilitate SRI adoption. Another institutional arrangement that seems to have shaped the spread of SRI is the teams of specialist SRI transplanters, usually women, who travel from place to place applying SRI methods and sharing their skills with local labourers (see Chapter 6). Although these are all intriguing and important research questions in relation to SRI, especially for policy-makers who need to allocate scarce financial resources in the most effective way, we did not any encounter studies that have analysed them in detail. 5.2.5 Summary Only a few detailed studies on SRI adoption have been published so far. These few papers suggest fairly robustly that the best-endowed farmers are more likely to adopt SRI practices, at least initially. There are also clear signs of differences in the patterns of adoption of individual SRI components, which implies that some SRI components may fit better with particular types of farmers, households, rice plots or other specific characteristics. This would further imply that the benefits of SRI are heterogeneous across households, but this also remains to be investigated. To conclude, in spite of the relatively large presence of SRI extension projects and programmes in many rice producing areas of the world, very little is known about the patterns and dynamics of SRI adoption, non-adoption and disadoption. As a result major questions, such as the actual levels of adoption of SRI or its components, variations in these levels across time and space, and/or the effectiveness of various extension systems, remain unanswered. 5.3 Measuring impact: evidence from farm-level studies In this section we review the literature that has investigated the question of impact at farm or household level. First, in order to provide a framework for our assessment, we present a bit of necessary theoretical background on technical change (Section 5.3.1) and the problem of separating impact of a technology from observable and unobservable farmer and plot characteristics (Section 5.3.2). Section 5.3.3 describes the SRI impact studies that were included in the literature review, which is presented in Sections 5.3.4, 5.3.5 and 5.3.6. Section 5.3.7 summarizes the discussion.

67 5.3.1 Mechanisms underlying impact The methodological challenges involved in identifying and measuring the impact of new agricultural practices and technologies at farm and household level are complex. Various sources of heterogeneity need to be taken into account, if they are not to obscure the true impact of the technology or practice. Various sources of endogenous and exogenous variation need to be considered, including diversity in biophysical conditions and agro-ecological settings; macro- and micro-economic factors; and institutional contexts. A fourth source of complexity stems from the diversity that exists in farm-level implementations of SRI practices (see Chapter 8). This factor makes it necessary to assess the impact of various forms of partial SRI adoption, including different combinations of individual SRI components. Under such conditions, sophisticated statistical methods are required in order to disentangle SRI practices from the many other confounding variables present. Farmers may also differ in ways that could affect yield outcomes, independently of the practice or technology being investigated. For example, differences in yields between adopters and non-adopters of SRI practices may arise from natural variations in input use, such as labour or pesticide, instead of the new practices themselves. Some other yield-influencing characteristics may be observable – such as age, education level and experience, or plot characteristics such as soil fertility levels – but could also be unobserved, intangible factors such as skill. The latter can be especially problematic. For example, if more skilful farmers are more likely to adopt SRI, but are also more likely to produce higher yields irrespective of the cropping system applied, then a comparison of yields between groups of adopters and non-adopters should try to take the differences of skill into account. In addition, statistical methods are required in order to identify whether a difference in yields results from a change in one SRI management practice, such as an increase in organic fertilization, or another practice, such as the use of younger seedlings – or a combination of both. Evaluating the possible synergies that may exist among the individual practices requires particular care. The conventional economics literature offers useful insights on the methods and techniques required to distinguish production increases that result from increasing inputs, and production increases that result from technological change. It is a common insight in agricultural production economics that crop yields can feasibly be increased by an increase in inputs (such as labour, water or fertilizer). However, it is not always economically efficient to do so. An economically rational farmer is only likely to choose to invest more if his or her increased costs are offset by a sufficient increase in benefits. Therefore, economically optimal production strategies do not necessarily equate to the maximum theoretical production level. Consequently, it is important to calculate the likelihood that a farmer will make an increased investment in terms of the marginal costs and benefits of the investment. Marginal costs and benefits are likely to differ across farm households and so will the optimal input quantities, depending on differences in plot fertility, access to credit, and so on. The consequence

68 of these theoretical considerations is that individual farmers are likely to differ in the level of inputs they provide. The implication of a change in production technology – which in our analysis may include the adoption of new crop management methods such as SRI – is that production or yield may be changed without altering input levels. If the new technological system is found to be an improvement, this implies that the input productivities (the ratio of output to input) have increased. Estimating such changes is made more complicated by the fact that the new technology very probably changes the economically optimal level of inputs, so that farmers may respond by adjusting their input levels. For example, if, as a result of new technology, the marginal productivity of labour has changed, then a farmer may increase or decrease his or her labour supply to the point where the marginal benefits of the increased or decreased investments equal marginal costs once more. This applies equally to other inputs. A further complication arises from the fact that the specific adjustments made by farmers are a function of marginal costs, which in most cases relate directly to local prices. Hence, when the relative prices of inputs differ across sites or households, this factor is likely to lead to differences in the input adjustments made by different farmers as a result of adopting the same new technology. In spite of this complexity, two facts can still be established. For a technology to be superior compared to an alternative method, it should hold, first, that at least one of the input productivities has increased, and second, that the total input costs required for any given level of output must have decreased, as a result of which farm profit increases. 28 In principle, this can be assessed using cost–benefit analyses and/or adoption studies. However, as discussed above, a major pitfall can arise with such analyses if independent variables that simultaneously contribute to differences in productivity, whether observed or unobserved, are not accounted for in the analysis. We discuss this point in the next section. Before moving to that discussion, however, we draw attention to a specific issue that arises when trying to identify and measure the possible productivity-enhancing technological effect of a technological system like SRI in particular. As described in Chapter 3 and section 4.4, the use or increase of organic fertilization is often recommended as a highly desirable component of SRI. Fertilizer is an input that would be expected to have a beneficial effect on crop production across a wide range of cases, especially if fertilizer use has previously been minimal. Accordingly, some critics have argued that the claimed benefits of SRI should instead be attributed to straightforward improvements in fertilization (e.g. McDonald et al., 2006, see also Chapter 4). It therefore seems important to try and disentangle the discrete effects of

28 In theory, the production curve for the new technology may be better than the conventional production curve only in relation to a subdomain of input levels. For instance, the technology may work better only for certain levels of labour use.

69 fertilizer from the other management practices involved in SRI, such as the use of younger seedlings, wider spacing of hills, soil aeration or intermittent irrigation. 5.3.2 Observable and unobservable characteristics Farmers differ from one another in multiple characteristics, and these differences can give rise to differences in farm productivity. Farmers differ in terms of their access to and use of inputs, as well as their access to information and knowledge. There are also likely to be significant differences between the soil fertility levels of farmers’ plots. Finally, farmers may differ in skills, experience and motivation. While some of these variables are readily observed by the impact assessor (e.g. input levels, education levels, soil fertility levels), other variables are not directly observable (e.g. skills and motivation), or are not observed because the assessor has not taken them into account. Barrett et al. (2004) and Anthofer (2004) illustrate the importance of such observed and unobserved variables for the case of SRI adopters and non-adopters in Madagascar and Cambodia respectively. They show that yields achieved in conventional and SRI yields belonging to the same household are highly correlated, which indicates that farmer-specific variables explain a substantial portion of any differences in outputs that are found. For example, a well-endowed and well- motivated farmer, owning some plots rich in fertility, is likely to achieve higher yields than a neighbour who does not enjoy these advantages, irrespective of the management practices followed by the two farmers. If the better-resourced farmers are found to be more likely to adopt a new crop management system such as SRI, then an observed difference in crop yield between an adopter and a non-adopter may reflect these pre-existing differences and not necessarily the impact of SRI itself. Such a difference can provide neither a measure of the impact of SRI in the adopting household, nor an estimate of the potential impact that could be produced if non- adopting households were to adopt SRI practices. Even though this is a well- established insight from the impact-assessment literature (Duflo et al., 2007), this problem afflicts some of the SRI impact studies we reviewed. In order to assess the impact of a new technology, all the observable and unobservable characteristics that may influence both production levels and the likelihood of adoption need to be included, and corrected for. One of the best approaches to achieve this is to use randomized controlled trials, but these have not yet been used in assessing impact of SRI. Another relatively straightforward methodology, as applied in some of the studies discussed below, is to compare SRI management and a reference treatment managed by farm households that practised both methods at the same time, on plots that are as similar as possible in relevant characteristics such as soil fertility, water control or distance from the home. In such a controlled evaluation, all the potentially confounding variables that help to determine yield are cancelled out, and the impact assessor is left with a true measure of the impact of the technology. Nevertheless, as shown in the previous section, adoption patterns are usually not random and the observed impact in this approach cannot be generalized to non-adopters.

70 Even so, an additional source of complexity can complicate such an assessment. This is the possibility that a new technology may alter the opportunity costs of a farmer’s own or family labour in relation to the two different production methods. The farmer or household might respond by investing more labour in plots under SRI management and decreasing the labour supplied to the conventionally managed plots. If so, a direct comparison between labour used under SRI and conventional systems may overstate the actual difference in labour used, or conversely the change in labour productivity. This point has been raised with regard to the allocation of limited supplies of manure between SRI and non-SRI plots, whereby the observed effect of SRI may be inflated due to a reduction in manure use on non-SRI plots (Tsujimoto et al. 2009). Taking these considerations into account, we now turn to our review the studies that have presented evidence on the farm- or household-level impacts of SRI. First, we describe the main characteristics of these studies, including the control variables used. 5.3.3 Characteristics of SRI impact studies For this part of our analysis, we selected 21 studies from our database (see Table 5.3). The criteria for selection were as follows: First, the studies presented primary empirical evidence and analysis. Second, the authors explicitly aimed to compare changes in productivity between SRI and a reference treatment, in most cases ‘conventional practice’. Third, they provided a sufficient level of detail in the description of SRI and the reference treatment, allowing for analysis by third parties. Finally, the studies were carried out in on-farm settings, providing insights based on actual farmer behaviour. Note that these selection criteria led us to exclude a number of on-station trials that presented findings on the productivity of land under SRI and conventional treatments. Since these trials were primarily aimed at analysing agronomic questions about SRI methods, we have discussed them in Section 4.1.

71 Table 5.3: Studies on input productivities and technology effects Study and Type of study Reference Controlled for: Not-controlled for: location 29 treatment Andr ianaivo On-farm Traditional - The cost-benefit analysis compares inputs of - Data on both systems are (2002), evaluation practice conventional practice and SRI practice. collected in different years and Madagascar potentially in different locations under different circumstances for different farmers. Yamah (2002), On-farm Farmers’ - Unobserved farmer characteristics by comparing - Changes in labour use not Sierra Leone evaluation techniques yield differences between farmers reported contemporaneously cultivating SRI and - Yields are not corrected for conventional practice. plot level characteristics Ceesay (2002), On-station Farmers’ - Unobserved farmer characteristics by comparing - Changes in inputs (fertilizer, The Gambia and on-farm practice yield differences between farmers manure, labour) not reported evaluation contemporaneously cultivating SRI and - Yields are not corrected for conventional practice. plot level characteristics Anthofer (2004), On-farm Conventional - Unobserved farmer characteristics by comparing - Changes in labour use and Cambodia evaluation practice yield differences between farmers some inputs such green contemporaneously cultivating SRI and manure not reported conventional practice; - Plot characteristics, although the author observes that SRI is often cultivated on more fertile plots. The author does not try to link this to yield differences. Barrett et al. On-farm Conventional - Unobserved farmer characteristics by comparing (2004), evaluation practice yield differences between farmers Madagascar contemporaneously cultivating SRI and conventional practice; - Differences in plot level characteristics.

29 Studies are sorted by author name and year.

72 Study and Type of study Reference Controlled for: Not-controlled for: location 29 treatment Namara et al. On-farm Conventional - Unobserved farmer characteristics may explain - Yields are not corrected for (2004), Sri Lanka evaluation practice large difference in yield between adopters and plot level characteristics non-adopters; - The authors note that labour use increases substantially, but do not try to link this to yield differences. Schiller (2004), On-farm No clear - - Study does not present details Laos evaluations comparison on comparison provided Latif et al. On-station BMP and - Unobserved farmer characteristics; - Yields are not corrected for (2005), and on-farm farmers’ practice - Cost benefit analysis accounts for differences in plot level characteristics Bangladesh trials all used inputs. Dhakal (2005), On-farm Traditional - Not clear how SRI and Nepal trials methods traditional methods are compared Kabir and On-farm Practices before - Unobserved farmer characteristics. - Yields are not corrected for Uphoff (2007), evaluation and after SRI plot level characteristic Myanmar extension - Changes in labour use. Sato and Uphoff On-farm Conventional - SRI is more often cultivated with certified seeds, - SRI-yields compared with non- (2007), evaluation practice but the authors do not try to link this to yield SRI yields for potentially Indonesia differences. different farmers, with different input and soil fertility levels.

Kumar Sinha On-farm Conventional - Control for farmer characteristics: Labour use, - No control for soil fertility or and Talati evaluation practice and yields, are compared at households other inputs used. (2007), West contemporaneously practicing SRI in a non- Bengal, India random sample. Feuer (2008), On-farm Rice yield is - Yields are related to cultivation practices and - Regression analysis does not Cambodia evaluation explained in adoption of SRI technology in regression analysis account for many variables regression from describing input use. various exogenous variables

73 Study and Type of study Reference Controlled for: Not-controlled for: location 29 treatment Sita Devi and On-farm Conventional - Cost benefit analysis shows differences in use of - SRI-yields compared with and Ponnarasi evaluation practice seeds, labour, machines and irrigation water. non-SRI yields at different (2009), Tamil farmers, with different soil Nadu, India fertility levels. Barah (2009), On-farm SRI compared to - Differences in input levels are reportedly included - No control for unobserved Tamil Nadu, evaluation non-SRI farmers in statistical analysis, but these data are not farmer and plot characteristics India described in detail. Thomas and On-farm Conventional - Unobserved farmer characteristics by comparing - No control for soil fertility or Ramzi (in evaluation practice yield differences between farmers input use. press), contemporaneously cultivating SRI and Afghanistan conventional practice. Styger et al. ( in On-farm Control and - Unobserved farmer characteristics by comparing - (Opportunity) costs of manure press), Mali evaluation farmers practice yield differences between farmers collection and application not contemporaneously cultivating SRI and included conventional practice; - Use of labour and nutrients double under SRI, but the authors do not link this yield differences. Turmel et al. ( in On-farm Farmers practice - Unobserved farmer characteristics by comparing - No control for changes in press), Panama evaluation yield differences between farmers labour use. contemporaneously cultivating SRI and conventional practice. Adusumilli and On-farm Conventional - Total use of inputs listed for both SRI and non- - No control for unobserved Baggy Lama (in evaluations practice SRI farmers. farmer and plot characteristics press), Andhra Pradesh, India Barison and On-station Conventional - Unobserved farmer characteristics by comparing - No control for differences in Uphoff (in and on-farm practice yield differences between farmers use of labour and fertilizer press), evaluations contemporaneously cultivating SRI and Madagascar conventional practice. Yokoyama and On-farm Conventional - Yields controlled for various cultivation practices, - No control for unobserved Zackari (no evaluation practice as well as increases in use of organic matter and farmer and plot characteristics date), Indonesia application of micro-organisms. - No control for differences in labour use.

74 Some of the studies listed in Table 5.3 compared SRI with an alternative practice within the same household, thereby enabling them to cancel out some unobserved farmer characteristics. Some studies also controlled for differences in plot-level soil fertility (e.g. Anthofer, 2004; Barrett et al., 2004; Styger et al., in press; Barison and Uphoff, in press). Anthofer (2004) reported on-farm rice yields under SRI and non-SRI management for similar plots in the same households in different years. In this way Anthofer (2004) was able to control for unobserved farmer characteristics such as skills and access to inputs. He found that SRI yields were around 30 per cent higher than yields under conventional management. However, he also observed that SRI was generally applied on fertile plots, which were also closer to the homestead. As a result it is not clear whether the SRI yield increase should be attributed to differences in inputs (e.g. labour) or soil fertility levels, changes in crop management practices, or a combination of both. In addition, Anthofer’s study included only SRI adopters. This is an important qualification because, for the same areas in Cambodia, Resurreccion and Sajor (2008) suggested that SRI adopters may have been better off to begin with. Barrett et al. (2004) have presented one of the most detailed analyses that corrects for observed and unobserved farmer differences. They found a yield increase for SRI management of almost 2,500 kg/ha. But, they estimated that about half of this increase could be attributed to SRI practices, while the remaining half reflected pre-existing differences between farmers and soil fertility. Unfortunately, a mistake was later found in the arithmetic of the procedure used by Barrett et al. (2004), which casts doubt on the validity of these findings (Chen and Yen, 2006). Other adoption studies and/or on-farm evaluations only partially controlled for farmer and plot differences. For example, Kabir and Uphoff (2007) in Myanmar, Namara et al. (2004) in Sri Lanka, and Thomas and Ramzi (in press) in Afghanistan all controlled for farmer characteristics but not for plot characteristics. For a sample of farmers in Madagascar, Barison and Uphoff (in press) controlled for farmer differences, and stated that plot characteristics of SRI and traditionally managed fields were largely similar. However, they did not specify differences in fertilizer use across plots. Meanwhile, Sato and Uphoff (2007), Sita Devi and Ponnarasi (2009) and Palanisami (unpublished) did not control differences in crop yields for any observed or unobserved farm household characteristics. In addition, Sato and Uphoff (2007) mentioned that more SRI farmers used certified seeds, but they did not further specify whether or how this factor might explain yield differences. Finally, Kumar Sinha and Talati (2007) investigated changes in labour productivity in detail for their sample of farmers in West Bengal, India, but did not control for differences that may have stemmed from any other observed or unobserved farm household characteristics. While some studies carefully accounted for farmer- and plot-level differences, few reported or took account of changes in other key inputs. Most notably, few studies presented information about changes in labour use, or in some cases detailed information about differences in fertilizer use (e.g. Anthofer, 2004; Kabir and Uphoff, 2007; Thomas and Ramzi, in press). This makes it difficult to determine if any input productivities increased.

75 The on-farm analysis reported by Styger et al. (in press) highlights a particular difficulty involved in separating the effect of a technological change from a change in inputs. This arises in some exceptional cases when a change in technology is perfectly correlated with a change in input use. Recall that SRI adoption is sometimes defined as involving an increase in a key input – organic fertilizer – alongside a set of other management practices. In such circumstances, the shift to practising the new technology is inextricable from a change in a certain input, which affects all the farmers in the community of adopters. In the case studied by Styger et al. (in press), there was an increase in the use of organic manure that affected all the participating households. In such cases, it is practically impossible to determine the separate effects of the increase in inputs and an increase in productivity caused by a change in a management practice such as transplanting. At least in theory, the novel management practices might even have a negative effect on crop yields, masked by the increased use of fertilizers. Thus, while Styger et al.’s (in press) study indicates that there were substantial benefits to SRI adoption, it remains unclear which SRI component or combination of components contributed to those benefits and to what degrees, or whether this simply results from increased use of organic fertilizers. 5.3.4 Changes in input productivity An important indicator of impact of a technology is farm profit. But, as we discuss in the various subsections on productivity changes below, very few studies document all production costs and often changes in use of labour, fertilizer and manure are not, or only partially reported. As a result it is difficult to assess unambiguously if farm profits increase with SRI adoption. We therefore primarily base our review in this section on the impact of SRI on changes in four types of reported productivity, namely labour productivity, fertilizer productivity, land productivity and, very briefly, seed productivity. We begin with labour productivity. The adoption of SRI methods requires changes in multiple crop management techniques and thus changes in labour supply. Two issues have attracted much attention in the literature as well as wider debates about SRI: transplanting and weeding. Transplanting tiny young seedlings, carefully spaced and arranged in rows or squares has been thought likely to require additional care, skill and time. On the other hand, it has also been argued that much less dense transplanting implies a lower labour requirement. Non-flooded irrigation, increased spacing between plants and the use of very young seedlings is generally agreed to lead to a rise in weed growth and an increased threat to the seedlings, compared to continuously flooded management and traditional random (dense) transplanting. This is expected to require additional time for weeding; typically, four weedings at 10–12- day intervals are recommended, ideally using a mechanical rotary weeder. However, it is also argued that using a mechanical weeder may actually reduce the time needed for weeding. Two other SRI practices may lead to a higher demand for labour: first, the increased collection, processing and application of organic manure or compost; and, somewhat trivially, if production increases, an increase in time spent on crop harvest and post-harvest processes. Changes in labour productivity are therefore likely to play an important role in shaping the dynamics and patterns of SRI adoption. The first study that looked in detail at labour use in SRI was presented by Moser and Barrett (2003), who found that labour use increased with

76 SRI methods, mostly as a result of increased weeding requirements. The authors hypothesized that for many of the poorest farmers the opportunity costs of labour were higher than the potential gains, explaining why these farmers generally would not adopt SRI. In a different sample of SRI adopters in Madagascar, Barrett et al. (2004) noted that labour productivity under SRI increased for some farmers, but also that it decreased for a significant portion of farmers in the sample. A frequently cited study by Kumar Sinha and Talati (2007) analysed labour use under SRI management in West Bengal, India. For a non-random sample of SRI adopters, they reported that labour use on SRI plots was lower than on conventional plots. It is difficult to evaluate this conclusion as the authors did not specify the procedure employed to calculate labour use for their sample, in which not all farmers had adopted all the SRI components. In addition, they did not examine the actual variation in labour use between farmers or test the differences statistically. Some of their other findings are counterintuitive, such as the indication that labour use in harvesting operations decreased even while crop yield increased. This finding casts doubt on the procedure used. Meanwhile, some data presented in tables did not correspond with observations made by the authors in the text. For example, the authors stated that ‘If we consider the total time spent on weeding and hoeing throughout the crop cycle, SRI farmers had to invest more time compared to conventional farmers’ (Kumar Sinha and Talati, 2007: p.58). However this does not correspond with their calculations on labour use overall, in which weeding under SRI appears to be lower than conventional practice (Kumar Sinha and Talati, 2007: Table 5). Other studies have also reported changes in labour use and labour productivity, but again with different findings. Kabir and Uphoff (2007) observed that in Myanmar farmers’ overall production costs did not change, after graduation from a farmer field school in which SRI methods were extended. Unfortunately, the authors did not break these costs down, though they did specify that costs of purchasing and applying manure and fertiliser were not included. In on-farm trials in Bangladesh, Latif et al. (2005; 2009) found higher labour costs for SRI practice, mostly due to increased time spent on transplanting and weeding. Sato and Uphoff (2007) also found slightly higher labour costs. They did not include labour used for manure applications, though they also noted that only a few farmers used manure. Thomas and Ramzi (in press) report that farmers considered labour issues a constraint to SRI adoption, although they also showed that more time spent on weeding led to an increase in crop production. Styger et al. (in press) reported a substantial increase in labour use for SRI adopters. Their figures are an underestimate, since the costs of collecting and applying the substantial amounts of manure reported in the study were not included in the estimation. Few other on- farm evaluations reported differences in labour input at all. Both increases and decreases in labour use may be rational for individual farmers, depending on specific temporal and spatial differences. For instance, taking into account local input prices (or price ratios), it could make perfect sense for farmers in Madagascar to increase

77 labour use, while farmers in India decrease labour use but, for example, substitute labour for relatively cheaper fertilizers. The findings in these studies also point to issues surrounding temporal allocations of labour over the course of a season or year. This point was highlighted for example by Dhakal (2005), who reported from on-farm evaluations in Nepal that labour use for weeding increased substantially, but this was offset by a reduction in the supply of labour to transplanting and irrigation, so that overall the labour requirement did not change. Similarly Moser and Barrett (2003) reported that the shift in demand for labour was especially problematic for the poorest farmers, because they needed to engage in wage labour in order to buy food during the hunger gap, at a point in the season when SRI methods created an increased weeding requirement. These observations suggest that the opportunity costs of labour (and perhaps other inputs) vary not only across households but also across the cropping calendar, and that changes in labour supply imposed by SRI methods may be more feasible for some households than for others. Differences in opportunity costs, as well as perceived comparative advantages in the performance of farming operations, also shape the division of labour within households. In many regions, rice transplanting and weeding operations have typically been carried out by women, while men have usually carried out more physically demanding tasks, such as land preparation. The adoption of SRI methods, especially weeding with a mechanical rotary hoe, often triggers a change in these task allocations between men and women. In Tamil Nadu, for example, we found that many men have taken over weeding operations (see Chapter 6). The nature of tasks and the social institutions built around them have also changed. An informant based in Orissa reported that women used to sing during transplanting, but have stopped, at least temporarily, while they get to grips with the new transplanting technique (Sabarmatee, pers. comm.). Similarly, we observed in Tamil Nadu that hand weeding operations carried out by groups of women were rather social occasions, accompanied by gossip and laughter, whereas weeding with the rotary weeder was carried out by one or two or three men, working with their heads down in silent concentration (see Chapter 6). Resureccion and Sajor (2008) noted in Cambodia that the gender division of labour had not changed significantly following SRI adoption, although time spent on various tasks had. As a result, women were able to devote more time to non-farm employment, but men less. These kinds of changes may have various impacts within households, for example on power relations between household members, household income, and local wage rates for men and women. As far as we know, the impact of SRI adoption on gender relations has not been thoroughly investigated elsewhere. It has been suggested that, while SRI methods may increase labour requirements at first, a learning curve means that farmers can save time after a few seasons, as they become more skilful and confident in the use of the new methods (Adusumilli and Bhagya Laxmi, in press). Such a trend is indeed observed by Barrett et al. (2004), who reported that the median of labour productivity increased over time, after SRI adoption. Nevertheless, the authors noted that this observation could also result from an attrition bias, caused by farmers with the lowest rates of labour productivity disadopting SRI and dropping out of the sample.

78 Turning our attention to fertilizer productivity, we found that we were not able to draw any definitive conclusions. Major limiting factors are that total amounts of supplied nutrients are not always described in detail, while in some other cases nutrient use covaried perfectly with changes in technology (see above). For instance, Styger et al. (in press) found that supplying substantially more nutrients increased rice production, but this would rather suggest a movement along the conventional production curve instead of a shift towards a new, more productive, SRI production curve. By contrast, Barison and Uphoff (in press) reported substantial increases in rice production in a sample where only six farmers out of 109 farmers were using compost. Apparently, differences in compost use alone could not explain the production increases that they observed. The authors did not specify whether there were any differences between farmers in terms of chemical fertilizer use. Some authors have observed relationships between the soil fertility status of rice plots and reported SRI yields. In on-farm evaluations of SRI in Panama, Turmel et al. (in press) found that the difference in yields between SRI and conventional practice increased on plots that were richer in calcium (Ca), magnesium (Mg) and manganese (Mn). Anthofer (2004) also observed that the difference in yields between conventional and SRI plots widened on plots with better soil fertility levels. Other authors have argued that the yield increases attributed to SRI methods could largely be an artefact of soil fertility differences. For example, Tsujimoto et al. (2009) found that high SRI yields were primarily related to soil fertility effects, in particular a greater nutrient- supplying ability. Similarly, Schiller (2004) states that in Laos SRI is only suitable under very fertile conditions, or when large quantities of nutrients are applied. The findings reported by Turmel et al. (in press) suggest that the marginal productivities of soil nutrient inputs change when switching the production method to SRI. This suggests that SRI methods may be associated with an upward shift in the production curve in highly fertile plots only. Unfortunately, however, they did not specify whether the use of other inputs (such as labour) also changed. Before we turn the discussion towards a comparison of reported land productivities, we note that most of the studies in Table 5.3 report increases in seed productivity that are sometimes substantial. One of the attractive selling points of SRI is the potential saving in seed, because of the significant decrease in planting density. Single-seedling transplanting at a spacing distance of 25x25 cm requires just 16 plants per square metre, as compared with 200 seedlings per square meter when planting two seedlings at a distance of 10x10cm. The saving in seed contributes directly to food security in the case of saved seed, or cash in the case of purchased seed. The saving may be particularly attractive in the case of expensive improved varieties and hybrids, where these are used. Needless to say, the reduction in the seed rate should be accompanied by a dramatic increase in seed productivity, otherwise the yield will shrink disastrously.

79 Table 5.4: Reported changes in land productivity Study and location 30 Yield from Standard Yield from Standard reference deviation of SRI deviation in practice yield in SRI practice 1 reference practice 1 t/ha t/ha

Andrianaivo (2002), 2,0 - 6,0-20,0 - Madagascar Yamah (2002), 2,5 Range: 5,3 Range: (4,9- Sierra Leone (1,9-3,2) 7,4) Ceesay (2002), 2,5 - 7,4 - The Gambia Anthofer (2004), 1,6 - 2,3 - Cambodia Barrett et al. (2004), 3,4 SD: 0.5 6,3 SD: 1.8 Madagascar Namara et al. (2004), 3,8 - 5,5 - Sri Lanka Schiller (2004), - - - - Laos Dhakal (2005), Various data - Various data - Nepal Latif et al. (2005), 6,8 (BMP); - 5,9 - Bangladesh 5,0 (Farmers’ Practice) Kabir and Uphoff (2007), 2,1 SD: 0,5 6,5 SD: 2,6 Myanmar Kumar Sinha and Talati 4,0 - 5,3 - (2007), West Bengal, India Sato and Uphoff (2007), 4,3 - 7,6 - Indonesia Feuer (2008), 1,8 - 3,1 (Good - Cambodia SRI fields) 2,7 (Partial SRI fields) Barah (2009), - Range: 4,0- - Range: Tamil Nadu, India 6,2 5,1-7,0 Sita Devi and Ponnarasi - - - - (2009), Tamil Nadu, India Adusumilli and Bhagya 4,6 SD: 0,6 5,4 SD: 1,1 Laxmi (in press), Andhra Pradesh, India Barison and Uphoff (in 3,4 SD: 0,5 6,4 SD: 1,8 press), Madagascar Styger et al. (in press), 5,5 (control) SD: 0,3 9,1 SD: 0,2 Mali 4,9 (farmers (control) practice) SD: 0,2 (farmers practice) 1Column lists standard deviation (S.D.) of crop yield, or the range of crop yields observed (Range).

30 Studies are sorted by author name and year

80 Study and location Yield from Standard Yield from Standard reference deviation of SRI deviation in practice yield in SRI practice reference practice t/ha t/ha

Thomas and Ramzi (in 5,6 SD: 1,5 9,3 SD: 3,4 press), Afghanistan Range: 2,0- Range: 4,0- 9,0 20,0 Turmel et al. (in press), - Range: - Range: Panama 0,6 – 7,5 1,2 -9,0 Yokoyama and Zakari (no 6,3 SD: 1,3 8,1 SD: 1,7 date), Indonesia

The vast majority of the studies discussed in this section also reported increases in land productivity (see Table 5.4). Of the studies listed in this Table, only the on-farm trials by Latif et al. (2005) in Bangladesh found that the reference practice recorded a higher land productivity than SRI. In this case, the reference practice was a locally recommended BMP rather than farmers’ practice. In the case of the rest of the studies, which report higher land productivity for SRI than the respective reference practice, the magnitude of the differences varies from modest, e.g. the study by Adusumilli and Bhagya Laxmi (in press), with a difference of around twenty per cent, to huge differences such as the range of SRI yields presented by Andrianaivo (2002). Again, we note that direct comparisons between SRI and non-SRI yields may not highlight a causal effect of SRI, unless artefacts caused by unobserved or observed farmer and plot characteristics are properly suppressed. The results of these studies should also be interpreted carefully for two specific reasons. First, we note that many of the studies on land productivity involved very small sample sizes, where sampling error could have relatively large distorting effects. This applies to the studies of Andrianaivo (2002), Ceesay (2002), Yamah (2002), Dhakal (2005), Latif et al. (2005), Thomas and Ramzi (in press) and Turmel et al. (in press), all of which concern samples of fewer than 25 SRI farmers. Second, we note that in this same set of papers, plus the study by Kumar Sinha and Talati (2007), the samples were not constructed randomly. Although most of the studies discussed in this section (except Latif et al., 2005; 2009) found that yields were higher with SRI than with conventional practices or best management practices, the degree to which inferences may be drawn from these studies is limited because of the many observable and unobservable farmer characteristics were not taken into account. We conclude that the existing literature does not allow for a conclusive test for the presence of technological change, i.e. an increase in the productivity of at least one of the inputs or a decrease in overall input costs for a given output. Before moving on, we note that the data presented in Table 5.4 suggests not only that land productivities are higher under SRI but also that the variation in land productivity is wider for SRI plots/farmers. This could create a source of increased risk for farmers. We will return to this issue in Section 5.3.6. We now move on to discuss changes in productivity and impacts at the margins.

81 5.3.5 Changes in marginal productivities and effects In this section, we discuss a handful of papers whose authors have tried to assess the impact of SRI using different methods. Barrett et al. (2004) used some advanced statistical techniques to analyse whether the marginal productivities of inputs change when switching to SRI; Palanisami (unpublished) tried to disentangle the relative contributions of the various SRI components; and finally Barah (2009) investigated whether practising SRI affects the (allocative) efficiency of input use. In their study on productivity changes among Malagasy SRI farmers, Barrett et al. (2004) used a decomposition approach to identify possible changes in the marginal productivities of inputs when switching from conventional practice to SRI. The results from their estimation suggested that under SRI the production curve is scaled upward with a constant fraction. Albeit statistically insignificant, the authors found that the marginal productivities of manure and labour, in particular, increased under SRI. This would imply that there is a change in the curvature of the SRI production function, which would suggest that a profit-optimising, SRI- adopting farmer would increase his or her application of these inputs. The authors suggest that their sample size was probably too small and/or affected by too much statistical noise to find statistically significant effects. As mentioned above, the methodology used in this study may have provided incorrect results (Chen and Yen, 2006). A different approach was followed by Kabir and Uphoff (2007) and Palanisami (unpublished). These authors aimed to disentangle the various SRI components – in samples of SRI adopters in Myanmar and Tamil Nadu, respectively – in order to identify the respective marginal contributions of these individual components to crop yields. The findings presented by Kabir and Uphoff (2007) suggest that SRI, i.e. the complete package, has a larger marginal effect on yield than adopting improved seeds or varieties. However, the authors did not further decompose SRI into its individual components. The presented results were based on very small sample sizes and not accompanied by statistical tests. Palanisami (unpublished) followed a similar decomposition approach, with similar methodological limitations. In both studies the caveat about controlling for observed and unobserved farmer characteristics applies. Nevertheless, the approaches used by these authors are very promising and could be used to discover not only the marginal effects of the different components of SRI, but also whether returns differ across regions or farmer types. The latter information could help to explain patterns of partial adoption. In addition to the decomposition approach, Palanisami (unpublished) estimated a rice production function in which he included SRI components as explanatory variables. Surprisingly, his results show that increasing seedling age and increasing the number of seedlings per hill both had a positive effect on overall rice crop yield, while the farmer’s cumulative experience with SRI methods had a negative effect. The latter finding seems counter-intuitive, while the former two certainly contradict theories about the biophysical mechanisms that are supposed to underlie SRI methods. If validated, they would, however, offer a clear explanation of why so few farmers in Palanisami’s (unpublished) study used younger seedlings or planted one seedling per hill.

82 Yokoyama and Zakari (no date) made similar observations in their regression analysis of a sample of Indonesian SRI adopters. They observed that rice yield was not significantly correlated to the use of SRI components, only to differences in input use and plot characteristics. Feuer (2008) did not find an effect of using SRI methods in his regression on a sample of Cambodian farmers. Once again, however, caution is required in interpreting these results, since Palanisami (unpublished), Yokoyama and Zakari (no date) and Feuer (2008) did not fully control for observed and unobserved differences in farmer characteristics. Therefore it is not clear if these findings reflect actual causal relationships. Barah (2009) investigated whether cultivation of SRI leads to a change in farmer efficiency levels. He used Data Envelopment Analysis to assess technical, allocative and economic efficiency for SRI adopters and non-adopters. Barah (2009) observed that yield under SRI is higher than conventional practice for different wealth groups, and more importantly that SRI farmers are more efficient, both technically and economically, than non-adopters. Unfortunately, the study provides very few insights in the exact methods used and variables included. As such it is difficult to establish whether SRI adoption has an impact on efficiency, or whether SRI farmers were more efficient to begin with. 5.3.6 Changes in production variance/risk Many smallholder farmers are risk-averse, especially with respect to downside risks, i.e. they are keen to reduce the likelihood of failing to meet a minimum target, such as producing enough rice to satisfy household needs (e.g. Di Falco and Chavas, 2009; Antle, 2010). For this reason, many smallholder farmers are as much if not more concerned about the degree of variation in crop yields as they are interested in achieving a high yield. The practice of transplanting very young, single seedlings, in particular, has been criticized for potentially exposing smallholder farmers to excessive risk. This was confirmed during our field visit to Tamil Nadu, where some farmers informed us that they would not transplant very young seedlings or single seedlings because of the risk of salt damage, especially in certain parts of the Cauvery (Kaveri) delta near the coast. One development worker in Madagascar, who works with lowland rice farmers in the north-western part of the island, rejected single seedling transplanting as far too severe a risk for poor and marginal farmers (Toon Defoer, pers. comm.). The MAFF system developed by Patrick Vallois in Madagascar, which he designed explicitly to minimize risk, does not emphasize especially young seedlings and is presented to farmers as a method for producing ‘at least enough’ rice at reduced cost, rather than achieving very high yields (see Chapters 3 and 8). Irrespective of the technology or management practice followed, crop yields are likely to vary across farm households. The degree of variation is affected by a range of observable and unobservable factors. Technology may have positive and negative impacts on the variability of crop yields. For example, SRI trainers sometimes encourage farmers to transplant only the most healthy seedlings, which is likely to minimize the risk that the young seedlings die and thus potentially narrows the range of yield outcomes. On the other hand, transplanting a single seedling per hill may increase the risk of yield losses if some seedlings die, thus widening the range of possible yield outcomes. Also, changes in plant or field architecture

83 may have a bearing on the vulnerability of crops to pest or disease pressure. Hence, the overall effect of SRI on the variability in crop yields is ambiguous. Information on the distribution of crop yields under SRI as compared with conventional or other improved practices is extremely rare. The most detailed analysis on changes in yield risk under SRI was carried out by Barrett et al. (2004). They showed that the adoption of SRI is associated with a base increase in rice yield variability, while no variables, including experience with the method, increased or decreased the yield risk under SRI. Some of the studies listed in Table 5.4 included variances, standard deviations, or in some cases the yield range, although without decomposing it in detail. Yamah (2002) noted an increase in the range of crop yields under SRI in on-station trials in Sierra Leone. Anthofer (2004) did not provide specific information on the distribution of SRI yields, but included a scatter diagram that does not suggest a large difference in variation between conventional and SRI yields. Kabir and Uphoff (2007) observed that yields are higher among SRI farmer field school graduates, but the variation in crop yields for these graduates is also substantially wider. In a recent special issue on SRI published in the journal Paddy and Water Environment , nearly all the contributors observed an increase in crop yield variation. For example, in Andhra Pradesh, Adusumilli and Bhagya Laxmi (in press) observed that the standard deviation of yield was higher among SRI adopters while, interestingly, the standard deviation of expenditure amongst this group was lower. However, no further analysis was provided to investigate a possible relationship between these variables. Styger et al. (in press) showed a small increase in variation with SRI compared to farmers practice, although slightly lower variance compared to a control treatment. Barison and Uphoff (in press) and Thomas and Ramzi (in press) found that the range of SRI yields was considerably wider than that for conventional yields. In some of the above cases, the ranges of SRI and conventional yields overlap. In the case of Thomas and Ramzi (in press), SRI yields encompass conventional yields on both the lower and upper sides. However, in other cases the SRI yield range, which in some cases should be understood as the yield range among SRI adopters, appears well to the right-hand side of the conventional yield range. To our knowledge, no study has further analysed the properties of the yield distribution that are specifically important for assessing changes in the downside production risks, such as the skewness and kurtosis of the yield distribution. An increase in crop variation under SRI is compatible with the observation that generally well-endowed, potentially less risk-averse, households are more likely to adopt SRI. Moreover, when the adopter has fully mastered the new technology the returns to the new technology could be expected to become more stable over time. Even though Barrett et al. (2004) did not observe such a trend in Madagascar it could well be present in other regions. On the other hand we reiterate that the finding of a more variable yield with SRI methods could be an artefact of selection effects in the samples compared. Which of these effects are in play needs to be uncovered in further research.

84 5.3.7 Summary From our review of the literature discussed in this section, we conclude that the existing evidence on patterns of adoption and impact of SRI remains inconclusive in multiple respects. Nevertheless, the studies carried out to date have provided a number of important leads and inspiration for further and more focused research. We will discuss our conclusions and proposals for further analysis in Chapter 9. The next chapter turns to an account of what we learned from our field visits to Madagascar and Tamil Nadu.

85

Part 3: The spread, evolution and variation of SRI

86

6. SRI ‘on the ground’: Insights from Madagascar, Tamil Nadu and Nepal

In order to ensure that the insights from our historical research and literature review were firmly rooted in ground-level realities, field visits were undertaken to Madagascar and Tamil Nadu, India. The choice of these sites and the goals, design and organization of these field visits are described in section 2.4 and the trip reports are available in Appendix A and B. This chapter reports the insights gained from these visits as well as insights from a separate, short visit to Nepal. 6.1 Insights into SRI in Madagascar

6.1.1 Historical overview The Green Revolution in rice first impinged upon Madagascar’s Hautes Terres region (the highlands between Antananarivo and Fianarantsoa, where Henri de Laulanié worked) under a European-funded project, Operation Rice Productivity, which commenced in 1966. The improved rice cultivation methods extended within this programme included not only the use of chemical fertilizers, but also some management techniques that to some extent prefigured SRI. They included shortening the duration of rice nurseries to about 30–35 days, line planting with spacing distances between 20 cm and 35 cm, 1–3 seedlings per hill, the use of mechanical rotary weeders and improved water management including, ‘to a modest extent, drainage’. These methods formed the original core of the set of practices and technologies known in Madagascar as the Système de Riziculture Améliorée (Improved System of Rice Cultivation, SRA) (Elyah 2006: 31–32).31 The origins of SRI in Madagascar are discussed in Chapter 3. As described there, the Association Tefy Saina (ATS) was created in 1990 to build the capacity of Malagasy rural people and, in particular, to provide training in SRI methods to rice farmers. In May 1994, ATS began to promote SRI methods to farmers in an area close to Ranomafana National Park, under the aegis of a USAID-funded project coordinated by CIIFAD. The collaboration continued for the next four years, by which time ATS was working with 275 farmers (Rabenandrasana 2002). It was in connection with this project that Norman Uphoff, then serving as CIIFAD’s director, first learned about SRI (see Timeline). ATS was later involved in another USAID-funded project, Landscape Development Interventions (LDI) (Rabenandrasana 2002). Glenn Lines, a CIIFAD alumnus and Peace Corps volunteer who had been involved in the earlier project, later worked for LDI (Lines, pers. comm., February 2010). Norman Uphoff visited some SRI farmers under an LDI project in the Lac Alaotra region in 2001 (Uphoff 2001).

31 Elyah (2006) notes that improved rice varieties were not incorporated into the technology package until sometime after Operation Rice Productivity. The dominance of chemical fertilisers in the original package may be explained by the interests of aid donor countries, which wanted to enlarge the market for agricultural chemicals manufactured by their companies.

87 During the second half of the 1990s, ATS expanded its SRI-promotion work to other areas of Madagascar. The organization received funds from the French development agency to promote SRI methods in two sites in the central highlands (near the town of Antsirabe and the capital city, Antananarivo) and on the west coast of the island (around Morondava), in collaboration with the Madagascar Department of Agriculture and a French NGO, respectively. Other funds were received from the Swiss development agency and Malagasy foundations (Elyah 2006; Rabenandrasana 2002). Between 1999 and 2003, the Rockefeller Foundation funded a consortium involving four partners: CIIFAD, ATS, the applied research agency FOFIFA and the Faculty of Agriculture at the University of Antananarivo. The purpose of this collaboration was to investigate obstacles to SRI adoption; to study the effects of SRI methods under farm conditions; and to carry out some agronomic field trials under controlled conditions (The SRI Consortium 2003). The study generated findings that were reported during the Sanya Conference in April 2002, which are discussed in Chapter 4. One of the first people to take up de Laulanié’s ideas was Patrick Vallois, whose NGO published a handbook on ‘the Madagascar early rice transplanting system’ in 1997 (Vallois 1997) (see Chapter 3). Later, Vallois was responsible for devising an adapted version of SRI, called MAFF 32 in Malagasy, which was first promoted in Madagascar’s Lac Alaotra region in 2004. Vallois positioned MAFF as a training method for the improvement of rice cultivation rather than a set of rice cultivation practices as such, based on a more flexible and less demanding version of SRI (Vallois 2005) (see Chapter 8).33 Willem Stoop made an independent study visit to Madagascar in March 2003, facilitated by ATS and ESSA. Stoop had already co-written a theoretical article on SRI (Stoop et al. 2002) and carried out some experiments on SRI methods in West Africa. His visit to Madagascar allowed him to compare the experiences of ATS and Malagasy farmers with his own, less encouraging experiments, and convinced him that ‘SRI has indeed the potential to produce extraordinary grain yields (even way above 10 t/ha), provided the farmer has adequately mastered the techniques, and in particular the calendar of operations’ (Stoop 2003: 11). The expansion of SRI practice is perceived by local stakeholders to have stagnated in Madagascar during the mid-2000s. At some point, internal disagreements led to a split within ATS, with different fractions continuing to use the ATS name. In 2006, the then president of Madagascar, Marc Ravolamanana, agreed to allow an SRI demonstration plot to be established in the grounds of the presidential palace. The initiative was organized with the support of the US Embassy in Madagascar, which had been encouraged to get involved by an intern, who had previously promoted SRI in the Lac Alaotra region, (probably under the aegis of LDI) (Uphoff 2008). In February 2007, the Hollywood film star Jim Carrey invited Norman Uphoff to Los Angeles in order discuss SRI. Subsequently, Carrey’s philanthropic organization, the Better

32 Mitsitsy Ambioka sy Fomba Fiasa , ‘saving seeds and cultural practices’ (Vallois, 2005). 33 See http://www.erails.net/MG/divers/la-methode-maff/contexte-general/ (accessed 29 January 2011.

88 U Foundation (BUF), hired two development professionals, Rames Abhukara and Winifred Fitzgerald, to make preparations for an African SRI Summit which had been planned to take place in Madagascar in 2009. Abhukara and Fitzgerald contracted a local consultancy organization, the Groupe Conseil Développement (GCD), to assist them (Uphoff 2008). Jim Carrey visited Madagascar in August 2008, where he met with President Ravolamanana and visited some SRI sites in the president’s helicopter. During the visit, the plan for an African SRI Summit was dropped 34 and it was agreed that the BUF would provide financial support for SRI promotion within the president’s flagship Madagascar Action Plan (MAP), an integrated approach to rural development, being implemented in various model villages. A national MAP planning summit was held in Antananarivo a few weeks later, during which two days were devoted to SRI. Not long after this meeting, the Groupement SRI Madagascar (Madagascar SRI Group) was established, with a secretariat run by the GCD (Razafimanantsoa 2009; Uphoff 2008). President Ravolamanana was deposed in a coup in early 2009 and the MAP was largely abandoned. However, BUF support for SRI promotion in Madagascar continued. During 2010, the Groupement has worked with other Madagascar SRI organizations to launch the first of a network of regional ‘SRI platforms’ they hope to establish across the country. In a separate development, Operation SRI Madagascar35 was established in 2008 by two francophone consultants based in Réunion. Operation SRI Madagascar provides technical support and raises funds for one of the ATS fractions, while also ‘supporting the association to defend its rights in the face of generalized piracy [turning] organic SRI into chemical SRA’ and resisting the depradations of the ‘[n]umerous [people] trying to control Tefy Saina in order to better grab SRI while keeping the association [ATS] at a distance from the large- scale SRI-promotion programmes in Madagascar and around the world’ (Chauvet & Guérin 2010: 18–19). There is a dearth of reliable statistical information about rice practices being applied in Madagascar. The figures widely considered to be most reliable are estimates by the Ministry of Agriculture, indicating that SRI occupies only about 3,000 ha. out of a total of about 1.3 Mha. of rice in the country. For comparison, traditional methods occupy about 760,000 ha., direct seeding about 130,000 ha. and SRA about 90,000 ha. 36 A widely cited estimate from a CIRAD/FAO report prepared in 2002 suggests that SRI occupies only 0.34 per cent of the rice area. A commune-level census carried out in 2001 suggested a similar low adoption rate, with SRI methods being practised in 0.47 per cent of villages in Madagascar, concentrated in the highlands (Minten and Barrett, 2008). During our field visit, many local stakeholders insisted that these figures must be considerable underestimates; but we have not been able to

34 Observers in Madagascar believe that President Ravolamanana dropped his plans for an African SRI Summit because he was concerned that Madagascar, despite having been the birthplace of SRI, actually had very little to show to other African countries in terms of progress. See Razafimanantsoa (2009). 35 See http://www.srimadagascar.org/ (accessed 29 January 2011). 36 Note that these figures provided by the Ministry do not add up.

89 find more recent figures to show how many farmers are now practising SRI or its components on how many hectares of land. 6.1.2 Key observations from our field visit 37 Perhaps the most important observation from our visit to Madagascar is that SRI is a quite firmly established part of the rice scene in the country. We met farmers, for example in Soatanana and Talata Ampano, who had been using SRI methods for more than two decades, as well as numerous others who were well aware of the system and had some experience with it. Nonetheless, we also met some others who have only been practising SRI for a couple of years. The Malagasy rice community in general applies a three-fold classification of systems of rice cultivation: SRI, SRA and SRT ( le système de riziculture traditionelle , traditional rice cultivation). SRT in principle describes farmers’ conventional practices, though it may be little more than a residual, catch-all label for whatever is not SRA or SRI. The relationship between SRA and SRI is not universally agreed. SRI post-dates SRA, which may explain how its name came into being. Some stakeholders see the two systems as diametrically opposed, with SRA representing a Green Revolution-style high-input system and SRI a low-external input alternative. Others suggest that SRA is effectively a stepping stone towards SRI. Many of the farmers we met evidently regarded SRA and SRI (and indeed SRT) as complementary alternatives that could be suitable choices for different fields and under different circumstances. In other words, the three systems co-exist and will probably continue to do so. In the following paragraphs, we highlight some of the most important and interesting issues that arose out of our discussions with farmers, extension officers, programme managers, academics, NGO staff and others during our visit. SRI specifications and rice cultivation practices According to some informants, certain organizations formerly advocated a strict guideline of transplanting 8-day-old seedlings; more typical nowadays is a slightly more flexible guideline of 8–15 days. ATS 38 now commonly recommends transplanting at the two-leaf stage, which they regard as a more universal guideline that can be applied in different regions. There was some disagreement about the interaction between soil fertility and plant spacing distance. Should seedlings be more widely spaced as soil fertility is more favourable, or more narrowly? The biophysical mechanisms thought to be at work would be different in each case. Is it a question of wider spacing allowing plants to spread and grow in fertile soils, or is it that more fertile soils mean that plants can be placed closer together without adverse competition for nutrients among them? Most informants felt that spacing should increase when soil fertility increases. (This conforms with Henri de Laulanié’s (1993) view). We only

37 The background to our field visits to Madagascar and Tamil Nadu is described in section 2.4 and in the trip report in Appendix A. 38 Interview, Bernard Rakotonirina, 27 April 2010

90 encountered one field, belonging to a farm household close to Fianarantsoa, 39 where seedlings had been transplanted very widely apart. The field was located in a valley bottom with good water control and drainage. The plants were spaced about 40 to 50 cm apart in a square pattern. Much more common was a spacing of about 25 cm. University of Antananarivo agronomist Bruno Andrianaivo acknowledged that early guidelines had advocated an increase in spacing up to 50x50 cm on very rich soils. Based on more recent research and experience, he now suggests that a maximum of 33x33 cm is more suitable. Finally, most informants stressed the need of square planting, but Bernard Rakotonirina, regional coordinator of ATS in Fianarantsoa, pointed out that square planting is only required if the farmer will use a mechanical weeder. Most organizations working in agriculture in Madagascar stress the importance of improving soil fertility. Many informants described soils in Madagascar as badly degraded and compacted due to the overuse of chemical fertilizers. Some NGOs working on SRI promotion, such as ATS in Fianarantsoa 38 and the Centre for Rural Promotion (CPR) in Ambositra encourage the use of compost and manure in order to rehabilitate soils over time. Other SRI-promoting organizations recommend the use of biological fertilizers, such as bat droppings, which are supplied by the Malagasy fertilizer company Guanomad (e.g. the Lutheran development organization FLM/FAFAFI 40 and extensionists working in the former MAP village, Antsahabe) . Many SRI promoters, in Madagascar as elsewhere, believe that a key strength of SRI is that farmers can achieve significant benefits without having to adopt new varieties. However, a few informants, including Bruno Andrianaivo, believed strongly that it was very important to recommend improved varieties for use with SRI. His reason was that new varieties were necessary in order to deliver the substantially improved results that were vital in order to convince farmers to adopt and maintain SRI practices. This would appear to be a strategic (or perhaps tactical) judgement, rather than one that is dictated by technical factors. Moreover, Andrianaivo is convinced that synergetic interactions explain a substantial proportion of the benefit of SRI cultivation and that the best results are achieved when all SRI components are adopted jointly. Other informants took a more pragmatic stance, including John Ravelomanantsoa, 41 director of a development project run by ADRA. He felt that each of the individual components of SRI independently represented an improvement on conventional methods. Therefore, if farmers adopted any or a few of the SRI practices, it would represent a desirable step forward and would deliver an associated increase in production. He portrayed this as a pragmatic stance, based on an acceptance that many farmers would not be able to practise all the SRI components fully. The single issue raised most often by stakeholders as a key condition of farmers’ ability to adopt SRI was the capacity to manage water, which is a function of both a dependable water supply and good drainage. In Madagascar’s highlands there are only limited possibilities to increase farm sizes, and it appears that few farmers have access to, or security of tenure over,

39 Farmer meeting Fianarantsoa, 26 April 2010 40 Interview, Alfred Rasamimanana (FLM/FAFAFI), 26 April 2010 41 Interview, John Ravelomanantsoa (ADRA), 29 April 2010

91 plots with good access to water. This accords with the findings of Minten and Barrett (2008), who found that SRI techniques were mostly adopted by wealthier households with access to (titled) land and improved irrigation infrastructure. Several of the NGO representatives we met drew attention to their work in promoting vegetable cultivation on rice plots during the off season. This activity is intended to help farmers diversify their cultivation systems, produce food for household consumption and possibly generate income, but also to ensure good soil management: compost and/or manure is chiefly applied to vegetable crops in the off season, rather than to rice itself. This can have beneficial effects on rice planted the following season, not only from the added nutrients but also from the crops themselves, either through breaking up the soil (e.g. potatoes) or fixing nitrogen (e.g. green peas). Off-season cropping options are particularly relevant to farmers in areas with degraded water delivery systems, who cannot rely on irrigation during the dry season. In some areas, green manures (e.g. Tephrosia ) are grown. The attention given to off- season crops and the management of soils in the winter months echoes the recommendations of de Laulanié (2003). Views on the costs, benefits and risks of SRI cultivation It was widely accepted, though not universally, that SRI is a more expensive practice, primarily because of higher labour demand for transplanting and weeding. However, many informants argued that savings on seeds and fertilizers outweighed the higher labour costs, so that total investments in SRI are lower. In either case, SRI was widely regarded as worth adopting, if possible, because it was more productive and more profitable. The feasibility of practising SRI was not always clearly defined, but depended heavily on water control (see above). Some informants agreed that SRI yields were more variable than with conventional methods, creating the possibility of heightened risk. Some informants were concerned about the risk involved in planting single seedlings (see Chapter 5). In one location, Ambalavao, a farmer noted that the incidence of stemborer seemed to be reduced in rice under SRI management. However, very few of the farmers we met in the central highlands mentioned explicit concerns about increased production risks. It may be that such concerns are implicitly reflected in their choice to apply SRI, SRA or SRT on a particular plot, shaped by their perception of the plot’s suitability (or riskiness) in relation to the three different methods. We discussed in Chapter 5 the possibility that SRI is practised preferentially on better rice plots, and such a preference could represent a way for farmers to manage risk. Institutional frameworks and support networks promoting SRI in Madagascar Some informants claimed that cultural factors and group dynamics played a key role in determining SRI adoption and practice. It was suggested that farmers liked to conform with their community, which could be a constraint on SRI adoption. Some informants even suggested that some farmers would not want to adopt SRI precisely because it is more productive. This was attributed to a strong social or cultural pressure against becoming too successful or wealthy compared with the rest of the community. Group dynamics seemed to

92 have a more visible influence in some villages, where local institutions for collective labour management ( Tamby ) exist. A key process in the adoption of SRI is the learning curve involved in getting accustomed to the novel methods. In particular, transplanting young and single seedlings was regarded as difficult at first, but informants felt that the learning curve was rapid. We learned that some women have become specialists in SRI transplanting, moving from one community to another at transplanting time to lead the transplanting activities and guide local transplanters. We found similar specialized labour teams in Tamil Nadu. We came across several anecdotal accounts of knowledge exchanges among farmers about SRI. In particular instances, knowledge about SRI seems to have been carried by migrant farmers moving back and forth between communities, or through visits to relatives in other parts of the island. It is difficult to assess how important such mechanisms are in spreading SRI knowledge and practice. It has been found that SRI adoption is strongly correlated to the presence of extension agents and NGOs (Moser and Barrett, 2006; Minten and Barrett, 2008) and is mostly restricted to the highlands were transport costs for extension services and NGOs are relatively low.42 Many informants, especially farmers, attested that sustained technical support was vital in ensuring that SRI became embedded. Such sustained support and follow-up appeared to be less likely in cases where SRI information and training was offered to farmers in response to their requests, rather than as part of a targeted promotion strategy, which was the case with ADRA 41 and AROPA/FERT. 43 These organizations pointed out that they had not made a strategic decision to promote SRI, but offered SRI training in response to demand. Farmer organizations may play a role in articulating such demand. ATS in Fianarantsoa described an interesting approach and funding model for SRI promotion, whereby they did not charge farmers for SRI training but were repaid by receiving a portion of the increase in rice production. At the level of organizations promoting SRI practices, we found few indications of strong or sustained linkages or exchanges among SRI organizations until the recent past. Until about 2007, the various small NGOs, training organizations and projects working on SRI were operating largely independently of one another. This situation has changed recently with the creation of the Groupement SRI (see above). This loose network of SRI organizations currently has about 70 members and is in the process of establishing regional SRI platforms across the country. At the time of writing, such platforms have been established in Antananarivo, Antsirabe, Ambositra and Fianarantsoa, among others. This initiative may foster increased sharing and exchange of information and perhaps also greater coordination and mutual learning, but this remains to be seen. Some informants were sceptical about the prospects for this effort, however, and there is some mutual mistrust and disagreement among SRI-promoting organizations and other rice stakeholders in Madagascar.

42 Interview with Phillipe Martel, director Fonds Régional de Développment Agricole (FRDA), Fianarantsoa, 27 April 2010 43 Interview with staff at AROPA/FERT, 28 April 2010

93 Connections between SRI-promoting organizations and the Madagascar research community appeared to be weak. This was partly a question of supply but also of demand. Few of the SRI-promoting organizations actively sought scientific input, advice or research on the agronomy of SRI. However, field-level technicians and trainers evidently have a large reservoir of practical, qualitative knowledge on the characteristics, pros and cons of all the local systems of rice cultivation. Bearing in mind the fact Madagascar is the birthplace of SRI, we were surprised to find very little evidence of connections and exchanges between the Malagasy SRI community and SRI organizations and practitioners in other countries. We found some awareness that SRI was being practised in other countries, but hardly anyone who had significant knowledge of what was going on outside Madagascar. Few people received, or were aware of how to find, information on SRI from outside the country. This even seemed to apply to some international organizations that promote or provide training on SRI in Madagascar and also work in other countries. Some stakeholders expressed the view that it was disappointing and embarrassing that a cultivation system that was perceived as a Malagasy innovation had progressed relatively weakly in its country of origin. Some local observers have formed the view that an important factor that has retarded the spread of SRI within Madagascar has been the dominance over national agricultural development programmes of foreign aid agencies, which are perceived to be interested only in exporting high-input agricultural technologies to Madagascar (Elyah 2006; Razafimanantsoa 2009). Perhaps unsurprisingly, this argument did not come up during our workshop. 6.2 Insights into SRI in Tamil Nadu 6.2.1 Historical overview SRI first took root in India primarily in the southern states of Tamil Nadu and Andhra Pradesh. Knowledge about SRI arrived in these locations through at least three different channels, in chronological order: • It seems likely that the first occasion SRI knowledge arrived anywhere in India was at the experimental internationalist commune, Auroville, which is located on the border between Tamil Nadu and Puducherry, 44 in 1999. Puducherry is a former French colony, which is now part of a Union Territory of India. Information about SRI seems to have arrived in the form of a French-language pamphlet that was carried by a francophone visitor to Auroville, which described SRI methods in Madagascar. Following the descriptions provided in that document, Annapoorna organic farm in Auroville carried out some small trials with SRI methods between 1999 and 2003. The system was observed to stimulate vigorous growth in rice plants without producing a particular improvement in yield. From Auroville, knowledge of SRI spread via organic farming networks to various individual farmers and the staff of some small NGOs, who started trying out the system for themselves. One person who came across SRI via this route was Dr. G. Nammalwar, who is celebrated by some

44 Puducherry is also known by its former name, Pondicherry.

94 people in Tamil Nadu and Puducherry as a prominent ‘guru’ of organic agriculture (Shambu Prasad 2006).45 • In 2000, a research project on water-conserving rice cultivation was initiated by Wageningen University researchers, led by Dr. Hein ten Berge and Dr. Prem Bindraban. They collaborated with Dr. T. M. Thiyagarajan, who was then serving as the Director of the Department of Soil and Crop Management Studies at TNAU. The project was stimulated by the Dutch researchers’ curiosity about water-saving methods and technologies and the effects of soil aeration in rice cultivation systems. Ten Berge and Bindraban had previously heard about SRI methods and discussed them with Norman Uphoff. Although they were somewhat sceptical about SRI as a complete system, they thought that there could be advantages associated with intermittent irrigation methods in soils where iron (Fe) toxicity was a risk, which is the case in the Madagascar highlands and some parts of Tamil Nadu (see Chapter 4). The Wageningen-led project was designed to investigate the effects of water-saving irrigation methods on rice cultivation in Tamil Nadu (Shambu Prasad 2006).46 Experiments carried out under the project suggested that modified SRI methods could enable savings in both water and seed, as well as modest increases in yields. (The highest yields, however, were achieved with SRI planting and soil aeration methods combined with conventional irrigation) (Senthilkumar et al. 2008) (see Chapter 4). • Norman Uphoff visited India to hold meetings with agricultural officials about SRI in May and November 2002. These initial contacts did not stimulate as much interest as he had hoped. However, in January 2003, Uphoff’s institution, CIIFAD, coordinated and sponsored a visit by two Indian scientists to see SRI in Sri Lanka. One of these scientists was the Director of Extension at the Acharya N. G. Ranga Agricultural University (ANGRAU, Andhra Pradesh, India), Dr. Alapati Satyanarayana. Satyanarayana was initially very sceptical but returned to India having been persuaded that SRI was something worth investigating. He commissioned 200 SRI field trials to be conducted in different agro-ecological settings around Andhra Pradesh. The trials convinced Satyanarayana that SRI was a productive and efficient cultivation method whose effects could be explained scientifically (Satyanarayana 2004; Satyanarayana et al. 2007; Shambu Prasad 2006). In other parts of India, SRI seems to have entered by yet other channels. For instance, SRI first came to the attention of a state agriculture department official in the north-eastern state of Tripura, Baharul Majumder, via a friend who had heard about SRI while studying at Cornell University (Majumder, pers. comm.). Shambu Prasad, who has attempted to trace the history of SRI in India and helped to create a ‘learning alliance’ on SRI in Orissa, first heard about the system from an Indian colleague, who had heard about SRI from Norman Uphoff, when she returned from a visit to the Philippines (Shambu Prasad 2006).

45 Interview, Dr. G. Nammalwar, Puducherry, 4 August 2010. 46 Interviews: Hein ten Berge, Wageningen, 14 September 2010; Prem Bindraban, Wageningen, 1 October 2010.

95 In just a few years, SRI has become an established part of the rice scene in India. SRI is reported to have spread to the states of Kerala, Karnataka, Tamil Nadu, Puducherry, Andhra Pradesh, Maharashtra, Chhattisgarh, Orissa, West Bengal, Tripura, Punjab, Uttarakhand and Himachal Pradesh (Shambu Prasad 2006).47 SRI is now being promoted within the framework of the central government’s National Food Security Mission, 48 through the work of the national Agricultural Technology Management Agency and in the financial grants of the National Bank for Agriculture and Rural Development (NABARD, Mumbai, India). In December 2009, World Bank president Robert Zoellick visited India and published an article in The Hindustan Times in which he noted the potential of SRI, which he said had helped farmers in Tamil Nadu to increase their rice yields by between 30 and 80 per cent. 49 According to Zoellick and other sources, SRI has caught the personal interest of Indian prime minister Manmohan Singh. In May 2010, Mr. Bill Gates, co-founder of the BMGF, visited a village in Bihar where the NGO Professional Assistance for Development Action (PRADAN, Delhi, India) is promoting SRI as part of a NABARD-supported watershed development programme. 50 SRI seems to have emerged and spread through various channels and involved diverse interactions among NGOs and civil society organizations (CSOs), farmers and farmers’ groups, agricultural extension agencies, national and international agricultural research organizations, policy makers and funding bodies. SRI methods gained some early credibility when the M. S. Swaminathan Research Foundation (MSSRF, Chennai, Tamil Nadu) carried out some early experiments on SRI and promoted the system through its ‘biovillage’ concept (Shambu Prasad 2006). The spread of SRI in India has also been boosted by the Agriculture, Man, Ecology (AME) Foundation (Bangalore, Karnataka), which is the publisher of LEISA India magazine and a key member of the international Low External Input and Sustainable Agriculture (LEISA) network. National symposia on SRI in India were held in Hyderabad, Andhra Pradesh in 2006, Agartala, Tripura in 2007 and Coimbatore, Tamil Nadu in 2008. 51 Various non-governmental projects and programmes are promoting SRI in India. A substantial collaboration between WWF, ICRISAT and a number of research agencies and NGOs works to evaluate and promote SRI methods among farmers in states across India, with the benefit of some high-level political support and financial backing (Shambu Prasad 2006). SRI is also being promoted by a prominent Indian philanthropic organization, the Sir Dorabji Tata Trust (Mumbai, Maharashtra). 52 In recent years, Indian researchers have made significant interventions in international scientific debates about SRI, both by evaluating theoretical and conceptual arguments in

47 See http://www.sri-india.net/ (accessed 10 July 2010). 48 See http://nfsm.gov.in/Rice.aspx (accessed 20 August 2010). 49 ‘India could be a new pole of global growth: World Bank president’. Hindustan Times , 2 December 2009. http://www.hindustantimes.com/News-Feed/india/India-could-be-a-new-pole-of-global-growth- WB-President/Article1-482145.aspx (accessed 29 Jan 2010). 50 http://www.pradan.net/index.php?option=com_content&task=view&id=159&Itemid=106 (accessed 30 January 2011). 51 See http://ciifad.cornell.edu/sri/countries/india/index.html#workshops (accessed 20 August 2010). 52 See http://www.dorabjitatatrust.org/NGO_Grants/sri.aspx (accessed 20 August 2010).

96 favour of taking SRI methods seriously (e.g. Mishra et al. 2006; Satyanarayana et al. 2007) and by reporting the results of experiments and field trials carried out in India (e.g. Kumar Sinha & Talati 2007; Senthilkumar et al. 2008; Thakur et al. 2009; Thakur et al. 2010b). Some Indian scientists have directly confronted SRI’s scientific critics, insisting that the method must be taken seriously if it has been evaluated positively by farmers in their own fields (Satyanarayana 2004; Thakur 2010). In Tamil Nadu and Puducherry, SRI promotion has been taken up by individuals such as Dr. Nammalwar and a number of small NGOs and farmers’ associations, such as Ekoventure in Puducherry; Palmyra, based in Auroville; and Thumbal farmers’ association in Salem District, Tamil Nadu. However, the story of SRI in Tamil Nadu is dominated overwhelmingly by a single, major actor: the Tamil Nadu Irrigated Agriculture Modernization and Waterbodies Restoration and Management (IAMWARM) Project. The IAMWARM Project is a large, state government-led programme financed by a World Bank loan. The goals of this US $500 m project are to improve agriculture and water management in 63 sub-basins of the state, covering more than 600,000 ha. The agricultural components of the project are largely implemented by TNAU, and include the promotion of SRI as a major element, in addition to other water-saving agricultural technologies, such as drip fertigation, agricultural diversification schemes, and market development. The project was launched in 2007 and is expected to continue until 2013. 53 SRI seems to be regarded by various actors as a promising solution to a diverse range of current technical and policy challenges in Indian agriculture. In Tamil Nadu in particular, SRI is seen as a potential solution to shortages of both water and rural labour, which are expected to intensify in the coming years. Tamil Nadu is currently embroiled in a dispute with the neighbouring state of Karnataka concerning access to water from the Cauvery River, which is heavily in demand in both states for irrigation and other purposes. Meanwhile, SRI is regarded as a key means of boosting national rice production under the National Food Security Mission. A coalition of international NGO projects has depicted SRI as a method that can simultaneously increase rice yields, improve smallholder productivity and reduce water consumption while cutting methane emissions and nitrogen pollution from rice farming (Africare et al. 2010). SRI is being explored as a promising technology for adapting to climate change and possibly mitigating its effects in TNAU’s ClimaRice project, which is financed by the Norwegian government. 54 SRI is also regarded by some stakeholders as a better alternative to Green Revolution methods and even a solution to problems it created. Some NGOs and farmers regard SRI as an intrinsically organic method, although major initiatives like the TN–IAMWARM project promote the use of chemical fertilizers in SRI because manure and compost can be difficult to obtain and there are questions about whether organic fertilizers alone can provide sufficient quantities of the right kinds of nutrients for particular soils (Shambu Prasad 2006). A group of NGOs has criticized attempts to compare SRI with conventional best management

53 See http://tniamwarmtnau.org/index.html and http://iamwarm.gov.in/tnau.asp (accessed 30 January 2011). 54 See http://www.tnau.ac.in/climarice/index.html (accessed 20 August 2010).

97 practices, since they argue that high-input methods are typically out of reach of small farmers, whereas SRI methods are ‘fundamentally “pro-poor”’ (Africare et al. 2010: 13). Recently, it has come to light that methods resembling SRI in several respects were in use in the Madras Presidency (now part of Tamil Nadu) for an unknown period of years during the early twentieth century (Gujja & Thiyagarajan 2009; Thiyagarajan & Gujja 2009) (see Chapter 8). Some of the farmers we met in Tamil Nadu had dim recollections of a ‘single seedling method’ applied by previous generations. Thus, it seems plausible that SRI might represent a reinvigoration or reorientation of earlier methods rather than something distinctively new, in Tamil Nadu. Another component of SRI that has already been promoted in the more recent past is the practice of row planting in conjunction with the use of mechanical cono or rotary weeders. What is sometimes known as the ‘Japanese method’ was introduced to India by government extension programmes in the 1950s, but evidently did not have a major or lasting impact (ICAR 1960) (see Chapter 8). One farmer we met in Tamil Nadu remembered his father and uncle having used such weeders several decades ago. He felt that the practice had not caught on because Green Revolution varieties such as IR8 and chemical fertilizers had produced such a dramatic boost in rice yield that farmers had no particular incentive to achieve the additional precision involved in row planting. 55 He showed us a badly decayed rotary weeder, made of metal and wooden parts, that had belonged to his father and been abandoned in an outhouse for many years. 6.2.2 Key observations from our field visit Our most important observation concerning SRI in contemporary Tamil Nadu was that the general context for rice agriculture shows several signs of having entered a period of rapid and profound change, linked to the wider changes taking place in Tamil Nadu’s and India’s society and economy. Economic development is associated with increasing rates of urbanization and permanent or temporary migration out of rural areas in pursuit of work and income-generating opportunities. Land prices near towns and cities are rising and some land is being taken out of agricultural production. Wage rates have increased, partly because of an increased labour demand from industry and the service sector, and partly because of the National Rural Employment Guarantee Act (NREGA), which aims to provide stable income for the landless rural population. The urbanization trend has increased demand for agricultural products including maize, vegetables, fruits and even flowers. Land is being converted from rice cultivation to meet these rising demands. In addition, rising wage rates, labour scarcity and the rising opportunity costs of family labour have encouraged some landowners to switch their land from rice plots to less labour-intensive, high-value plantation crops such as oil palm and Casuarina , a timber crop. Rice acreages in Tamil Nadu are decreasing, but productivity has increased (Thiyagarajan et al., 2000). The remaining rice farmers search for ways to intensify rice production. SRI is promoted as one such opportunity, but more generally momentum is building behind a trend towards

55 Farmer meeting, Virinjipuram Krishi Vigyan Kendra , Vellore District, Tamil Nadu, 7 August 2010.

98 mechanization. Both public and private sectors are addressing this demand, the former primarily by research and development, the latter apparently primarily through the import of small-scale rice cultivation machinery from Japan, South Korea and China. Combined harvesters are already common, which travel around the countryside at harvest time on a hired basis. Mechanical transplanters are already common on larger farms, while motorized weeders are modestly priced and within the reach of medium- and even relatively small-scale farmers. Such machines are still out of reach of the poorest farmers, however, who still rely on manual labour. There are also many poor and landless rural people who rely heavily on agricultural labour for their incomes. The other important observation, which makes Tamil Nadu a distinct contrast compared with Madagascar, is the dominance of strong and well-organized institutions, TNAU and the IAMWARM project. TNAU manages the IAMWARM project, of which SRI promotion is a dominant element. TNAU and the IAMWARM project are not concerned exclusively with SRI promotion, however, but SRI methods have been incorporated into a framework of other methods and technologies developed or promoted by the university for improving and intensifying rice cultivation. These include machinery, herbicides, fertilizers and improved rice varieties and hybrids. In the field of machinery, TNAU carries out research and development and/or trials on mechanical weeders, mechanized transplanters, drum seeders and machines for automating the sowing of seed-tray nurseries, up to sophisticated and very expensive laser land-levelling equipment (which facilitates more precise irrigation management on rice fields). The university also promotes the development of technologies for horticultural and plantation crops, such as drip fertigation, polytunnels and others. SRI specifications and rice cultivation practices Based on initial trials carried out in the early 2000s (associated with the WUR–TNAU project mentioned above), TNAU has developed a relatively standardized SRI package adapted to the conditions of Tamil Nadu. This package is based on the SRI principles originally developed in Madagascar and is referred to as ‘modified SRI’: • Seedlings transplanted after 14–16 days; • Planting of seedlings in a grid 25x25 cm; • Single seedling per hill, where possible; • Regular weeding with a rotary weeder, four times during the growing period at a 10- day intervals; • AWD irrigation; • Integrated nutrient management, combining chemical with organic fertilizers, where available. Some informants noted that the above specifications should not be interpreted as a blueprint, rather, a considerable flexibility should be maintained. In view of this, we were surprised to find that the spacing distance of 25x25cm was regarded as a suitable guideline across all soil types in the state. Nonetheless, we were informed by scientists at the Tamil Nadu Rice

99 Research Institute in Aduthurai 56 that some farmers do deviate from this practice, including one who is said to have increased his spacing distances considerably on very fertile plots. As if to reinforce the demand for flexibility, we met some farmers who complained that the recommended spacing was too wide, others who said that it was too narrow. In the former case, the problem was that the rice plant canopy would not close completely and bare soil would remain visible up till maturity, which may allow weeds to grow and suggests an inefficient use of solar radiation. We observed such plots in a number of areas, for example around Pattukottai.57 Some farmers and researchers also mentioned, however, that wider spacing had a beneficial effect early in the season, since it discouraged rats from entering the fields and damaging the rice plants. In the latter case, the problem was that the canopy closed sometime after the second or third weeding, so that it was difficult or impossible to carry out intercultivation (soil disturbance with the mechanical weeder) four times, as recommended. This raises questions about the interaction between different SRI recommendations concerning spacing distance, weed suppression and mechanical soil aeration, including the relative importance of different components. As was the case in Madagascar, we were told that a key process in the adoption of SRI is the learning curve involved in getting accustomed to the novel methods. In particular, transplanting young and single seedlings was regarded as difficult by most labourers at first, but informants felt that the learning curve was rapid. Specialized teams of women transplanters have been given training in the new transplanting methods, a skill that sometimes allowed them to seek employment outside their usual working environment. As in Madagascar, good control over water was generally considered a prerequisite for practising SRI methods. Broadly speaking, Tamil Nadu can be divided into command areas where irrigation water is sourced from the Cauvery River and areas where most irrigation comes, increasingly, from privately owned borewells. In principle, the IAMWARM project, and the promotion of SRI methods within the project, are intended to conserve water, but farmers in Tamil Nadu are not charged for their water consumption and electricity for pumping is provided free of charge to farmers (between certain hours). Farmers therefore had weak incentives to save water, and it is unknown what influence this factor may have on the adoption or non-adoption of water-saving irrigation techniques. We believe that Tamil Nadu could present an interesting opportunity to compare how farmers’ decisions to adopt AWD or other water-saving irrigation techniques are shaped by the type of irrigation source they use. Prof. T. M. Thiyagarajan and many of his colleagues at TNAU are convinced that there is a substantial beneficial effect from using the rotary weeder and that this relates not to soil aeration but to pruning the lateral roots of the rice plants (Thiyagarajan, pers. comm.). It is argued that pruning the lateral roots encourages the rice plants to develop a deeper root system that can access water and some nutrients from lower soil layers. We had not heard this

56 Interview with various scientists at TNRR, Aduthurai, 31 July 2010 57 Various field visits and farmer meetings around Pattukottai, 28 July 2010

100 argument elsewhere, and we do not know whether this observation has been confirmed in other studies. Views on the costs, benefits and risks of SRI cultivation A major selling point of SRI in Tamil Nadu is considered to be the substantial reduction in the seed rate. This factor seemed relatively more important to Tamil Nadu farmers compared with those we met in Madagascar, and this may be due to the wider prevalence of more- expensive improved and hybrid rice varieties in India. Opinions differed on the question whether variety choice was important in SRI practice. Dr. Velu, a TNAU soil scientist, noted 58 that the proportional increase in yield under SRI management was generally higher with traditional varieties than improved varieties. Along with the reduction in the seed rate, many farmers reported a reduction in labour costs for transplanting and weeding, leading to lower production costs overall. This dynamic is clearly different from Madagascar, where most informants felt that labour costs were considerably higher, even if the overall returns were better. Lowering labour costs is an attractive proposition for farmers, for the reasons discussed above, and it seems likely that production methods with even lower costs, such as direct seeding, may replace the more labour-intensive components of SRI in the future. However, the impact of SRI methods on rural labourers is obscure and has apparently not been examined. Some farmers reported that they encountered considerable resistance from labourers to the new methods of planting, which was usually attributed to the labourers’ unwillingness or inability to learn and adapt to new methods. Instead, the resistance of some labourers may stem from the overall reduction in demand for labour. Reflecting the findings of Moser and Barrett (2003; 2006) in Madagascar, farmers in Poonglam village argued that SRI methods were hard for smaller-scale and/or poorer farmers to adopt, since they often relied more on off-farm income and could not manage their own rice crop intensively. Some TNAU staff argued that, on the contrary, larger farmers were less likely to adopt SRI methods because of labour scarcity at key times, whereas farmers with small rice plots could take care of them using their own labour. The trend towards mechanization suggests that larger farmers do indeed find labour savings attractive. Some Tamil Nadu farmers, like those in Madagascar, told us that SRI methods worked best on the most fertile plots, 59 which may indicate that SRI is practised selectively on the most fertile soils. Farmers are advised not to practice SRI methods on fields with high levels of salinity and acidity, which may adversely affect the development of young seedlings. 60 In areas with moderately saline soils, e.g. Krishnagiri District, farmers plant two or three seedlings instead of one, to mitigate the risk of seedlings dying. Farmers and extension workers in another area, Tiruchirapalli District, reported that gap filling may be required to compensate for up to ten per cent of seedlings lost in the early stages of growth. This concern may also explain why transplanting seedlings younger than 14 days old is not recommended.

58 Interview, 26 July 2010 59 Meeting with SRI-farmers at the Department of Agriculture, Tiruchirapalli, 29 July 2010 60 Meeting with staff at Deparment of Agriculture, Chennai, 6 August 2010

101 Very young seedlings were also considered to be vulnerable to heavy rain and local flooding. When mature, however, rice plants under SRI management were considered to be more resilient to storms and cyclones. Like the soils in Madagascar’s highlands, some areas of Tamil Nadu have soils where there are problems with iron toxicity, where AWD irrigation and mechanical soil aeration are considered to be beneficial. Dr. Velu, 61 observed in experiments that using the rotary weeder in aerated rice plots also increases zinc and nitrogen availability; zinc is an important limiting nutrient in soils in Tamil Nadu. Institutional frameworks and support networks promoting SRI in Tamil Nadu TNAU possesses considerable research capacity for investigating and testing rice cultivation methods, tools and systems. However, relatively little of the research carried out by TNAU scientists has been published in scientific journals, either locally or internationally. TNAU staff are oriented towards, and incentivized to provide, practical extension services rather than producing peer-reviewed scientific articles. This has the positive effect that their work is strongly oriented towards practical impacts for the improvement of agriculture in the state, but it also means that the expertise of TNAU scientists is difficult for SRI practitioners or scientists from other Indian states or other countries to access. SRI extension in Tamil Nadu has some of the characteristics of a classic top-down, target- driven, centrally managed extension programme, in which ‘modified SRI’ has the appearance of a relatively fixed package of BMPs. The methods used by the IAMWARM team involve the familiar suite of demonstration plots, which are often established by the roadside; working with prominent farmers in the expectation that others will be inspired to follow their example; and providing subsidies for seeds, chemicals and equipment, such as rotary weeders. TNAU also enjoys strong linkages with government extension services and policy- makers, This represents a contrast with the situation in Madagascar, where the public extension service is weak and the institutional void has been partially filled by many different actors, often small- or medium-sized NGOs or programmes with modest capacity. The range of different SRI specifications and extension methods was therefore, not surprisingly, larger in Madagascar. There is a dearth of reliable statistical information about the scale of SRI adoption or practice in Tamil Nadu or its impacts on households and the environment. The IAMWARM project managers maintain quite detailed records concerning the project activities and outputs (e.g. IAMWARM, 2008, 2009), but considerably less information has been collected regarding the project outcomes. Table 6.1: Comparing agriculture and rice production in Tamil Nadu and Madagascar Tamil Nadu Madagascar Agricultural sector: Infrastructure - High literacy rates; - Low rural literacy rates; - Good local infrastructure and - Poor rural infrastructure

61 Interview, 26 July 2010.

102 facilities. (roads, electricity) and facilities. Trends in - Increasing wages due to - Heavy reliance on manual factor urbanization and government labour in agriculture. markets: policies; - Trend towards mechanization (transplanting, drum seeding, weeding, harvesting, laser- levelling). Market - Relatively prosperous farmers, - Relatively poor, often very orientation with high degree of isolated; commercialization - Subsistence production for - Shift towards agriculture where most farmers; labour is more productive: - Production of market crops in - plantation agriculture; less isolated areas. - fruits, vegetables, horticulture. Natural - Lowland irrigated areas; - Erratic water supply, often due Resource - Decreasing availability of water to poorly maintained irrigation Management: due to dropping water tables and infrastructure; conflicts with neighbouring states; - Soil degradation due to - Major rehabilitation program of erosion and overuse of local irrigation facilities; chemical fertilizers. - Soil degradation due to overuse of chemical fertilizers. Institutional - Effective state-wide institutions for - Weak government institutions Environment SRI extension; for extension, as well as a - Strong support for SRI promotion limited presence in many by government; areas; - Top-down, target-oriented - Rural development activities, implementation of extension including SRI promotion, by activities; diverse scene of NGOs; - Very little NGO activity in SRI - Information-sharing network of promotion; SRI-extending organizations; - Strong linkages between research - Some NGOs working on and extension. responsive and farmer demand driven basis; - Weak linkages between research and extension. Characteristics of rice production and SRI: Average rice 2,930 kg/ha (2010) 1 2,750 kg /ha (2003) 2 yields: SRI - No recent data on adoption - Formerly very low rates of prevalence: (evaluation under way); adoption, concentrated in - SRI promotion in all rice-growing highlands (early 2000s); districts. - No recent data is available. - New institution to improve collaboration and coordination among SRI- promoting organizations. Table summarizes the key insights from the field visits in Madagascar and Tamil Nadu. 1 Data from Palanisami (unpublished) 2 Data from Minten et al. (2007)

The scale of activities of other NGOs is small. The Centre for Ecology and Research is in contact with around 500 farmers in the Thanjavur district, of which around 100 farmers have adopted SRI components. Ekoventure estimates that 60 per cent of farmers in Puducherry

103 have adopted SRI methods, including AWD irrigation and planting young single seedlings in rows. The AME Foundation reports that 105 farmers in 30 different villages were practising SRI methods in 2008, although few farmers followed the recommendation to transplant 15- day-old seedlings (AME Foundation, 2008). Some disadoption of SRI practices was also reported, including one farmer from Auroville who stated that disadoption of SRI practices in this location was nearly complete. Table 6.1. highlights a few of the key differences between the situations in Tamil Nadu and Madagascar, as discussed above. The remainder of this chapter presents supplementary insights on SRI in Nepal. 6.3 Impressions of SRI in Nepal As a source of additional insights into the way SRI is implemented on the ground, this section presents some information gathered during a short field visit by Dominic Glover to eastern Nepal in November 2009. The visit was undertaken in support of PhD research by Mr. Rajendra Uprety, an agricultural extension officer with the Nepalese Department of Agriculture. The trip involved visits to meet rice farmers in three districts: Indrapur and Jhorahat, near the small city of Biratnagar in the Terai plains of eastern Nepal; and Dhankuta in the hills north of Biratnagar. The text in the next section is adapted from Glover (in press). A detailed review of SRI experience in Nepal was beyond the scope of this report. An overview can be found on the CIIFAD SRI website. 62 6.3.1 Variations in SRI in three local contexts In Indrapur district, I met a rice farmer who was considered to be ‘an SRI farmer’ and evidently acknowledged that label himself. The field where we met had been used at the beginning of the growing season as the location for a training and demonstration exercise on SRI transplanting methods. The rice harvest had been completed a few days before I arrived, so the stubble of cut rice plants was clearly visible, spread out across the field like the bristles of a worn-out scrubbing brush. Where we stood near the field edge, the fat tufts were widely spaced and arranged in a neat square pattern. It looked like a textbook example of SRI planting. Within a short distance, however, the neat, straight lines had become wavy and the stumps much closer together. A few metres away, there was no longer any sign that an SRI planting technique had been used. Was this still an ‘SRI field’? Would it be correct to label this farmer an ‘SRI farmer’? A short walk further on, my companion and I met another ‘SRI farmer’. Her fields, also recently harvested, were choked with weeds. There was no evidence of careful weed management or mechanical soil aeration here. I learned that the farmer had good reason to neglect these activities: she was the sole breadwinner in a household with school-age children, so she was too busy to keep on top of farming operations on her own fields. She had no money to pay for hired help. If and when she had time and money to do so, she tried to tackle the weed problem using a selective herbicide. Did her use of herbicides – instead of a

62 http://sri.ciifad.cornell.edu/countries/nepal/index.html (accessed 8 February 2011).

104 mechanical weeder – make her a ‘conventional’ farmer? Or was her planting technique enough to make her an ‘SRI farmer’? A few kilometres away, farmers in Jhorahat district had been having a more difficult time during the rice season. Their rice crop was mostly still standing in the fields, although the harvest was under way. The harvest was delayed primarily because the rains had been late at planting time. As a result, very few of the local farmers were applying SRI principles. The late rains meant that they had been unable to transplant their young seedlings in suitably ‘puddled’ fields. Still, we met several people who identified themselves as ‘SRI farmers’. They thought the system was a good one, having been satisfied with its performance the previous season. All being well, they intended to apply SRI methods the following season. In the meantime, they had coped with the late rains by double transplanting some of their rice seedlings. By transplanting young seedlings in clumps of 10–12 plants and a short while later re-transplanting them in clumps of 4–5 seedlings per hill, they felt they had obtained a satisfactory harvest. But double transplanting could be said to infringe, or at least significantly modify, two of the key principles of SRI, namely avoiding the trauma involved in transplanting older rice plants and planting single seedlings. Bearing all these considerations in mind, would it be correct to label these farmers ‘SRI adopters’? The following day, my companion and I travelled into the hills near Dhankuta. There, we met some more ‘SRI farmers,’ but these upland rice farmers were following practices that differed in interesting ways from those applied by farmers in Indrapur and Jhorahat. The leading farmer in a women’s self-help group had received training in SRI as it was practised in the plain. Returning home, she and her neighbours quickly realized that some of the SRI principles would have to be modified to suit the small terraces and steep hillsides where they grew their rice. Abandoning the idea of trying to plant their seedlings in straight lines or a grid pattern, their rows followed the contours of the hillside. The distance between the rows, which was generally a little wider than on conventional rice terraces but judged by eye rather than precisely measured, fluctuated gently in harmony with the tapering shapes of the terraces. These farmers also used little compost on their fields. Hiking up and down the steep paths that ran between the rice terraces, I was impressed by the physical exertion that would be involved in delivering loads of compost (or chemical fertilizer) to the fields. These observations provoke many interesting questions about how to make sense of such disparate experiences, all of which can lay claim to the label ‘SRI’ to some degree. Evidently, the identification of ‘adopters’, ‘non-adopters’ or ‘disadopters’ of SRI is not as easy as it might seem. Nevertheless, the mere fact that different practices could be observed in three different places, not too far from one another, does not necessarily mean that the farmers concerned could not be said to be practising versions of the same system. Clearly, however, the precise sets of practices being applied by the farmers I met were shaped not only by the farmers’ intentions but also by their negotiations with, or accommodations of, the agro- ecological, institutional and socio-economic circumstances they faced. In the next two chapters, we consider the mechanisms that have helped to spread SRI internationally, while at the same time shaping it in different ways.

105

7. Mechanisms that shaped the development and spread of SRI

In Chapter 5 we discussed some claims that have been made about the spread of SRI in various locations around the world. We highlighted some reasons for treating such claims with caution and discussed the inherent methodological difficulties involved in trying to measure how widely and deeply SRI knowledge and practice have taken root. In the same chapter, we critically analysed the relatively small number of academic studies that have examined SRI adoption patterns and impacts, and found that they raised many questions about levels of adoption and partial adoption. In Chapter 2, we pointed out that our attempt to gather more reliable data on the spread of SRI knowledge and activity through a survey was not successful. We conclude that it is at present extremely difficult to quantify the spread of SRI practice – let alone the spread of SRI knowledge or information, which should be considered equally important for a knowledge-based methodological innovation like SRI. Accurately measuring and evaluating the spread of intangible factors such as knowledge or skill is inevitably a difficult task, and this would apply equally to many agricultural innovations and changes of farming practice, not just SRI. Nevertheless, it is evident that activities of various scales, scopes and kinds – projects, agronomic trials, seminars, conferences, field schools etc. – that are labelled as ‘SRI’ are actually quite abundant. Scientific studies on SRI have been conducted in at least 10 countries (see Chapter 5). The 229 organizations listed in the database are involved in SRI activities of some kind in 21 countries. Chapter 6 made clear that SRI is a relatively well- established part of the rice farming scene in Madagascar and Tamil Nadu. The SRI-India Google Group has 436 members and the group has generated messages on nearly 1,750 topic threads since October 2007. 63 In the absence of reliable quantitative data, we explored the spread of SRI knowledge and practice qualitatively, by means of our literature review as well as the numerous meetings and discussions that took place during our field visits to Madagascar and Tamil Nadu. To help make sense of the unfolding history of SRI, we constructed a Timeline that serves as a helpful visualization tool (see Appendix C). This chapter discusses the themes and observations that emerged from our exploration. 7.1 SRI was shaped by the institutional context in which it was developed As we showed in Chapter 3, SRI was shaped by the particular scientific and institutional context in which it emerged, which was strongly constrained by modest scientific capacities, scarce information resources and weak governance structures. Henri de Laulanié’s inductive scientific method was forced upon him by the lack of alternatives. He fell back on his agronomic training, which was not in rice, and resorted to close observation of rice plants and farmers’ practices because he did not have access to good quality scientific information to

63 http://groups.google.com/group/sriindia?hl=en (accessed 24 January 2011).

106 guide him. He may have realized that, as he did so, he ran the risk of reinventing the wheel, but he felt he had no other choice. In the writings he left behind, his acute sense of isolation from the world of scientific agronomy is often palpable: After twenty five years of work for development, many questions remain without an answer. Some demand theoretical research which it is impossible to carry out in the peasant context, where the advisor is not in control of the techniques used and cannot check the application of precise research protocols. The answers to these questions nonetheless limit to a great degree the possibility of continuing to make progress in the implementation of a rational subsistence agriculture in the peasant context. The second part of the book … makes an inventory of problems solved and questions posed to research in tropical agronomy. These questions are often parallel to ones that have been studied and resolved, at least in part, for temperate agriculture. Perhaps there already exist some answers to some of them? In the current poor state of diffusion of research results to under-developed countries (translated into many international languages), the lack of good quality textbooks and extension manuals on tropical subsistence agriculture blocks the flow of information (de Laulanié 2003: 5– 6). De Laulanié made significant efforts to seek out scientific knowledge that might help him to make sense of his empirical observations, and seized upon the few pieces of research information that came his way. For example, he tried to get in touch with the Japanese scientist Katayama, whose model of rice growth patterns seemed to provide an explanation for the profuse tillering de Laulanié had seen in farmers’ fields. De Laulanié had access to Katayama’s model because it had been translated into French and somehow reached him in Madagascar. De Laulanié also relied heavily on the work of two francophone scientists, Puard and Angladette, one of whom had presented his work at a seminar attended by de Laulanié in Antananarivo in December 1991. From these meagre resources, as well as a few IRRI manuals, de Laulanié assembled his rice cultivation system (de Laulanié 1993; Glover forthcoming; Rafaralahy 2002). A second set of factors helped to shape SRI and to ensure that it could take root in the land of its origins. Harsh economic conditions affected Madagascar in the late 1970s, caused by the oil crisis and the economic policies of the socialist government that was installed in 1975. The rise in the price of oil led to an increase in the prices of agricultural inputs, which was the factor that led Henri de Laulanié to stop recommending chemical fertilizers to poor and marginal farmers. These factors were combined with a shift in agricultural development strategy towards large, donor-funded infrastructure projects rather than extension services for small-scale farmers. Against this background, the work of religious organizations, funded by overseas congregations, filled an institutional void in agricultural and rural development (Elyah 2006). As Rafaralahy (2002: 18) has noted, the early and uncertain development of SRI methods depended heavily on the commitment of ‘religious congregations of women who were willing to persevere with this system’. To a significant degree, the emergence of SRI and the scientific controversy that was sparked when it entered the international scientific arena in the late 1990s can both be explained by de Laulanié’s isolation from the international agricultural research system. SRI emerged as a kind of bottom-up innovation, based on inductive scientific practice, which was rooted in and

107 shaped by the particular setting where de Laulanié worked. It is important to remember that SRI was not designed solely around the physiology of rice but was also intended to suit the specific capacities and constraints of poor and marginal farmers and farm labourers, who carried out most farming operations by hand (see Chapter 3). It is not surprising, therefore, that de Laulanié’s rice cultivation system differed to some extent from recommended practices developed in other settings. On the other hand, it is also unsurprising, since SRI had been based on a close study of rice, that it also had some elements in common with other rice cultivation systems (see Chapter 8). In the light of these two factors, it is, finally, not surprising that de Laulanié’s claim to have developed a superior rice cultivation method provoked surprise and incredulity among experienced researchers from well-resourced and well-connected research institutes. 7.2 The key role of charismatic individuals It is hard to spend much time in the SRI arena without being struck by the dominant presence of two key individuals: Henri de Laulanié and Norman Uphoff. The influence and effects of these two people is impossible to quantify but it is beyond reasonable doubt that they have played critically important roles in the emergence and spread of SRI. Henri de Laulanié is routinely credited with having invented, or discovered, this distinctive rice cultivation system. In fact, de Laulanié’s innovation was not principally an invention or a discovery as such, but consisted primarily in the compilation of a set of different techniques into a coherent package, with an original name (see Chapter 3). De Laulanié’s role was nonetheless crucial. He was evidently a charismatic personality in his own right, but it is perhaps even more important that he was a powerful figure in the community where he worked, having the status and influence of both a priest and an expert in agronomy. Rafaralahy (2002) notes that de Laulanié might not have been able to convince Malagasy farmers to transplant 15-day-old seedlings in November 1983 if they had not been young students at his training centre. We noted above that SRI survived in the early days because women from religious congregations were willing to persist with the methods advocated by one of their priests. Norman Uphoff has played a similarly pivotal role in the SRI story. The website he established, his prolific correspondence, his wide-ranging travels and dozens of meetings, presentations and talks have helped to spread knowledge and information about SRI around the world and across scientific, policy and practice arenas (see Timeline). His numerous publications have also helped to develop and refine the theoretical concepts underlying SRI (see Chapter 3). It matters that Uphoff was serving as director of CIIFAD at the time he first encountered SRI and for several years afterwards. This status undoubtedly gave him considerable credibility with certain communities, and helped open doors to many of the individuals and organizations that have paid attention to SRI. Other influential individuals have also helped to promote the spread of SRI within different countries. In China, it almost certainly made a significant difference to the reception of SRI that one of the first articles discussing the concept was written by Prof. Yuan Longping, the esteemed and celebrated ‘father of hybrid rice’ in China (YUAN 2001). The credibility of SRI in India was undoubtedly boosted when the research institute led by Prof. M. S.

108 Swaminathan, India’s ‘father of the Green Revolution’, carried out some early trials on SRI. Another charismatic person, who has helped to spread SRI through organic farming networks in southern India, is Dr. G. Nammalwar. Norman Uphoff regards the initiative of numerous individuals as a key mechanism in the spread of SRI around the world (Uphoff 2010a). In many cases, key individuals have been in a position to take an independent initiative to carry out experiments and trials, even if in some cases they have not been able to mobilize the organizations they have been working for. One example is Alapati Satyanarayana, who was able to set up field trials in his position as Director of Extension for ANGRAU, the agricultural university in Andhra Pradesh (see Chapter 6). Satyanarayana was eventually successful in mobilizing other actors, at least for a while. Another example is Rajendra Uprety who, as a district-level agricultural extension officer in Nepal, was able to carry out some SRI trials with farmers without having to convince his entire department first (Uprety 2006). Numerous other examples are cited by Uphoff in his trip reports and accounts of how SRI spread to different regions and countries. 64 For India and south Asia generally, Shambu Prasad (2006) and Basu (2009) cite further examples. 7.3 Institutional positions may have influenced the scientific reception of SRI Although the charisma of Henri de Laulanié and Norman Uphoff certainly helped in the development and spread of SRI among certain communities, their status as outsiders to the scientific community also shaped the SRI story in a different sense. De Laulanié’s combined status as a religious priest, a field-level agronomist and a development worker were unlikely to give his ideas and claims any particular credibility among rice scientists. Indeed, de Laulanié himself recognized that he was stuck on the outer periphery of science and evidently suspected that valuable research on rice cultivation might be out there in the international agricultural research system, if only he could get his hands on it. Meanwhile, Uphoff’s background in political science has undermined his credibility among some rice scientists (e.g. Sinclair 2004). The way that SRI was perceived by critics such as Dobermann (2004), Sheehy et al. (2004) and Sinclair and Cassman (2004) can be contrasted with the reaction of Willem Stoop, a scientist who has made key contributions to the development of SRI theory and the international spread of the system (Stoop & Kassam 2005; Stoop et al. 2002). Stoop is not a rice specialist, having trained originally as a soils scientist. Later, however, he became an unwitting pioneer of Farming Systems Research, during a period spent working for ICRISAT in West Africa. There, he learned to work with farmers and to appreciate their farming strategies. This experience seems to have inclined him to be sympathetic to the inductive, bottom-up style of science that de Laulanié had used. Stoop was first prompted to learn about SRI by the Nigerian Director of WARDA (the West African Rice Development Association, now the Africa Rice Centre), Kanayo Nwanze, when he later served as interim Director of Research there. Stoop took SRI seriously because it helped him to make sense of his earlier observations of West African farming systems. He struggled to get other WARDA staff, who

64 http://sri.ciifad.cornell.edu/publications/TripReportsNTU.html (accessed 5 February 2011).

109 were pre-occupied by ‘conventional Green Revolution thinking’, to pay any attention to it. However, Stoop’s successor at WARDA, Amir Kassam, showed more interest and the two men have since worked together on several SRI articles (Stoop 2003; 2010; Stoop & Kassam 2005; Stoop et al. 2002). Stoop’s, Nwanze’s and Kassam’s inclinations to take a positive interest in SRI, contrasting with the attitudes of some other scientists, suggests that the scientific disagreements over SRI have been shaped not only by disagreements about empirical observations and theoretical concepts, but also by alternative disciplinary approaches and scientific paradigms, mediated by institutional frameworks and relationships. Both Stoop and Norman Uphoff have suggested that the criticisms of SRI made by scientists such as Sheehy, Sinclair and Cassman may have been motivated by their ‘considerable financial stakes in different and very ambitious types of rice research’ (Stoop, pers. comm., 18 January 2011), and that they may have attacked SRI because they perceived it as a potential ‘competitor for research funding’ (Uphoff, pers. comm., 13 January 2011). They point to the fact that John Sheehy, for example, is head of IRRI’s C4 Rice Project, which aims to ‘supercharge’ rice photosynthesis by radically improving the efficiency of rice carbon fixation. 65 In his criticisms of SRI, Sheehy has argued that the ‘failure to understand the fundamental nature of what governs yield and what constitutes a yield barrier will seriously impede efforts to design the radically new plant types and management systems needed to meet the future demand for rice’ (Sheehy et al. 2004: 7). Uphoff and Stoop point to the fact that Sinclair (2004) and Sinclair and Cassman (2004) have also urged development donors not to waste valuable resources on SRI research. 7.4 Networks and information flows Within the scope of this project, it was not possible to conduct a quantitative network analysis. Nonetheless, a qualitative review of SRI documents and websites and the construction of the timelines demonstrate that information and knowledge about SRI has spread through a variety of different channels and networks. A very important network of actors has centred upon Norman Uphoff and CIIFAD. Uphoff himself was introduced to SRI by Glenn Lines, who was then a CIIFAD employee working in Madagascar on a USAID- funded project. A CIIFAD alumnus from Madagascar, Joelibarison, subsequently introduced SRI practices to Sri Lanka. CIIFAD then sponsored a visit by Alapati Satyanarayana and another Indian colleague to see SRI in Sri Lanka, which convinced Satyanarayana that SRI methods could be worth investigating. A CIIFAD alumnus was responsible for introducing SRI to Baharul Majumder in Tripura (Majumder, pers. comm.). Another CIIFAD alumnus helped to arrange Uphoff’s first meetings with government officials in India (Basu 2009). Erick Fernandes is a former Cornell University staff-member and one of Norman Uphoff’s co-authors on SRI publications; he has since moved to the World Bank Institute, which has established a website on SRI (see below). Erika Styger, who has been promoting SRI in Mali in recent years, received her PhD from Cornell University and now works as Director of Programmes for SRI Rice, a new BUF-funded programme for the promotion of SRI, which is

65 http://irri.org/our-partners/networks/c4-rice/about-us (accessed 2 February 2011).

110 hosted by CIIFAD. 66 The timeline lists numerous other links connecting the spread of SRI to CIIFAD or Uphoff personally (see Appendix C). The fact that CIIFAD has been involved in many of the steps by which SRI knowledge and information has spread internationally does not mean that the SRI network is centralized or even that CIIFAD is necessarily the dominant node in the network. Figure 7.1 illustrates the fact that there have been quite a few bilateral, South–South exchanges that have played a role in spreading SRI. However, the journeys depicted on the diagram were identified from information compiled by Norman Uphoff, and include trips that were enabled or facilitated by Uphoff or CIIFAD, as described above. Figure 7.1: South–South exchanges in SRI

Note: Each arrow represents at least one visit by at least one person from one country to another. NB these movements are not sequential. Source: Compiled from various sources. One of the major channels that Uphoff and others have used to raise the profile of SRI is through publications of various kinds. These have included scientific journal articles and other academic publications, beginning with the early papers published by Uphoff (1999; 2002; 2003) and Stoop et al. (Stoop et al. 2002). Other articles and papers have been published in technical and practical journals such as LEISA Magazine (e.g. Rajukkannu et al. 2007; Uphoff & Fernandes 2002). The first such article was published in December 1999; it was written by Justin Rabenandrasana of ATS in Madagascar and is often cited by others as the first occasion they heard about SRI (Rabenandrasana 1999). In December 2006, LEISA Magazine published four SRI-related articles together, as part of a special issue about

66 http://sri.ciifad.cornell.edu/aboutsri/aboutus/index.html#people (accessed 5 February 2011).

111 ecological processes in farming (Stoop 2006; Thiyagarajan 2006; Uphoff 2006; Uprety 2006). Meanwhile, the IRRI magazine Rice Today was the forum for an exchange of views between Norman Uphoff and Thomas Sinclair that also drew attention to SRI (Sinclair 2004; Uphoff 2004b). Norman Uphoff has also made numerous presentations on SRI, not only at various international conferences and workshops but also in staff seminars and private meetings with senior staff within organizations. A list compiled by Uphoff indicates that he delivered presentations, presented posters or participated in panel discussions on approximately 181 different occasions, in 49 different countries, between October 1997 and December 2009 (Uphoff 2010b). Websites would appear to be an important means of sharing information about SRI internationally. In addition to the main CIIFAD website, the CIIFAD SRI Rice group maintains a blog, 67 a Slideshare site, 68 and an online photo gallery. 69 Other SRI websites for aimed at international audiences include an online SRI toolkit managed by the World Bank’s knowledge management arm, the World Bank Institute 70 and a Slideshare site on SRI run by IFAD.71 However, it should be noted that we found that very few of the people we met in Madagascar or Tamil Nadu used these information sources themselves or were even aware of them. This was especially true in rural villages, for obvious reasons. Nevertheless, it is clear that information and materials used by extensionists and NGOs was originally derived from information available on the web. At national level, publications and websites also exist, although it is difficult to gauge how important they are in terms of spreading awareness of SRI. Within India, LEISA India magazine has published a small number of articles about SRI. The SRI-India Google group has been an important and busy communication channel for its members, who include academics, project staff, experimenting farmers, NGO workers, etc. However, at field level, farmers seem to get their information on SRI primarily from NGOs, extension officers, local newspapers, farmer meetings, and sometimes radio broadcasts (Basu 2009). In Orissa, a ‘Learning Alliance’ has been set up to facilitate communication and other exchanges among a network of people and organizations working with SRI methods (Shambu Prasad et al. 2007). A similar attempt is under way in Madagascar, where the BUF and the GSRI are trying to build a loose network of SRI platforms across the country (see Chapter 6). Another method that has been used to foster information exchange and build relationships among different types of actors in the SRI field has been to hold periodic workshops, such as the three national symposia that have been held in India to date (see Chapter 6). Similar institutional frameworks exist in other countries where SRI is reported to have spread. Overall, the range of actors, types of organizational arrangements, communication channels and relationships are diverse, dynamic and context-specific. The promotion and extension

67 http://srinewsandviews.blogspot.com/ (accessed 3 February 2011). 68 http://www.slideshare.net/SRI.CORNELL (accessed 3 February 2011). 69 http://picasaweb.google.com/sri.cornell/ (accessed 3 February 2011). 70 http://info.worldbank.org/etools/docs/library/245848/index.html (accessed 29 Jan 2010). 71 http://www.slideshare.net/ifad/the-system-of-rice-intensification-sri (accessed 3 February 2011).

112 approaches adopted by different organizations vary quite widely too. The IAMWARM project seems to epitomise a rather classical type of large-scale, top-down, target-driven extension programme that relies heavily on demonstration plots, farmer meetings and close contacts between agricultural extension officers and ‘leading farmers’. Similar approaches are taken by government and university extension services in other locations. By contrast, small, local NGOs such as Ekoventure in Puducherry and Sambhav in Orissa, India, embed their SRI activities within wider work on community and social development, human empowerment, environmental conservation and other activities. This integrated approach is shared by otherwise dissimilar organizations such as ADRA in Madagascar and the Fianarantsoa branch of ATS in Madagascar. These organizations also aim to promote rural development in general and provide training in SRI methods to farmers who seek it, without adopting a programmatic, target-driven approach. In the light of this diversity, which we have not managed to analyse quantitatively or in fine- grained ethnographic detail, we find it impossible to make generalizations about whether particular institutional arrangements have been more or less common, prominent, or successful. We draw the conclusion that this reflects the loosely coupled, decentralized and dynamic nature of the social movement that has formed around SRI. In the next chapter, we explore how these characteristics are reflected in the diversity that exists in the ways SRI has been implemented or adapted in different locations.

113

8. Diversity in SRI in principle and practice

The System of Rice Intensification is said to be a peculiarly adaptable and flexible set of principles for improved rice cultivation (see Chapter 3). In this chapter, we examine the degree to which flexible interpretations and implementations of SRI can be observed in practice, and consider what these variations of the basic system might signify. The purpose of this analysis is to provide a foundation for some ideas that we will develop in the final chapter, where we will pose questions about the factors or processes that might drive and shape variations in the ways a rice cultivation system is applied in practice. Rice production systems are heterogeneous between and within the countries in which SRI is being extended. First, diversity arises from major differences in biophysical conditions, as the various documents reviewed in this report cover areas from the mountainous plateaus in Madagascar, with low control over water and degraded soils, to lowland irrigated areas in southern India. Second, macro- and micro-economic as well as institutional conditions differ greatly across these countries, which in turn affect rates of urbanization, local wage rates and the opportunity costs of cultivating crops other than rice. Villages and households across these countries and continents differ greatly in microeconomic, institutional and socio- cultural aspects, with differential access to resources and technical and institutional differences in rice production systems. Third, the institutional arrangements with respect to the extension of new agricultural technologies differ across locations. Whereas many official extension systems have but all disappeared in some African countries, government extension services are still major contact points for farmers in some Asian countries. These three types of heterogeneity are likely to lead to considerable differences in the way in which SRI is extended and adapted or adopted by farmers or farm households. A fourth layer of complexity arises from the conceptualization of SRI itself, which is said to be a flexible set of principles for rice cultivation, rather than a fixed set of practices – ‘a menu not a recipe ’ (Uphoff et al. in press, emphasis in original). Norman Uphoff, in particular, has insisted on many occasions that SRI should not be thought of as a ‘technology’, since he interprets that term to imply a fixed package of practices or ‘a typical component technology’, akin to the much-derided cultivation systems promoted during the Green Revolution. He proposes terms such as ‘strategy’, ‘methodology’, ‘philosophy’ or even ‘paradigm shift’ instead (e.g. Uphoff 1999; 2003; Uphoff et al. in press). We share Uphoff’s misgivings about the everyday interpretation of the term ‘technology’, which indeed implies that technology consists solely of technical components such as gadgets, devices, chemical additives, machines and so on. Nevertheless, we think that Uphoff’s objection to the use of the term in relation to SRI is misplaced. Based on our understanding of technologies as socio-technical ensembles (see Chapter 1), we take the view that it is more appropriate and useful to conceptualize technology as a ‘process of making’ or ‘the human capacity to make’. This perspective focuses attention on the human uses of

114 ‘skills, tools, knowledge and techniques to accomplish certain ends’ (Jansen & Vellema in press).72 By this definition, examining the intentions, goals, knowledge and skills of human beings is just as important as studying the tools they use. By such a definition, the term ‘technology’ can certainly be applied to SRI or indeed any other crop cultivation system. Any agricultural technology (according to our definition) necessarily undergoes a process of translation or transformation as it passes from the laboratory or research station into practice, or from one region or farmer to another (Glover in press; Maat in press). Accordingly, we do not see any real meaning in the distinctions Uphoff has tried to draw between ‘technology’, ‘methodology’, ‘set of principles’, ‘philosophy’ or other alternative terms. 73 Uphoff’s apparent purpose in objecting to the term ‘technology’ (according to his definition) is to make the argument that SRI is not susceptible to conventional methods of evaluation, since the ‘adoption’ of SRI cannot be measured in the same way that the use of a particular improved seed or chemical can be (Uphoff et al. in press). It is true that SRI is not an agricultural system defined by the use of particular inputs. However, we think that the same could be said about many aspects of ‘Green Revolution packages’, which involve recommended crop management methods as well as new crop varieties and agricultural chemicals. It is clear that SRI methods involve a combination of particular practices, even if the individual components are defined as desirable rather than mandatory, and their specifications are defined as flexible guidelines or principles rather than precise rules. Whether SRI is more ‘knowledge-intensive’ than a conventional Green Revolution-style intervention is really a question about the manner in which the system is packaged and the methods by which it is extended to farmers, rather than an intrinsic feature of the technical practices involved. As far as evaluation is concerned, one can certainly go into farmers’ fields and determine which specific practices are being used, and how. In fact, it seems to us that this is the only way to give due weight to the idea that SRI adoption should not be thought of as a question of exactly replicating the rules laid down in a training manual, but one of flexible, context- specific adaptation. We began to explore this question in Chapter 5, where we tried to look beyond the numbers of farmers who were said to have ‘adopted SRI’ in order to try and see which particular elements of the system were being practised by which kinds of farmers and in which types of contexts. In this chapter, we take a different approach to try and evaluate the degree of diversity in SRI practice across sites and over time. This could be the first step to determine whether there are distinct patterns in the ways SRI is being practised or promoted and, for example, whether such patterns may be related to differences in agro-ecological, socio- economic or institutional environments.

72 Drawing on francophone scholarship, technology can be understood as the study of technique in the same way that musicology is the study of music-making and sociology is the study of social processes and relationships (Schlanger, 2006). 73 To argue that SRI represents a ‘paradigm shift’ is another matter altogether, on which we make no comment here.

115 8.1 Variation in SRI specifications To get a sense of the range of different implementations of SRI that have been investigated, promoted or practised in different locations, we examined the SRI definitions given in the documents in our database. For the analysis presented in this section, we included only 259 documents whose purpose made it appropriate to give a detailed definition of SRI methods. These included scientific studies investigating SRI practices as well as documents such as training manuals and some promotional literature. In some of these documents, more than one definition of SRI was provided, for example in studies that included an on-station trial of recommended SRI practices as well as field studies based on actual farmer practices. In such cases we have opted to record the farmer practice, in order to try and capture the breadth of diversity in SRI specifications. Table 8.1 displays the number of documents that provided a detailed definition of SRI practice, sub-divided into two groups according to the intended audience. The data reveal some interesting patterns within and between the two categories. 74 Documents in the scientific category were much more likely (p<0.1) to provide a detailed definition of SRI, including the component practices. Nevertheless, the chance of an SRI component not being specified in a scientific document was still quite high, around 30 per cent for some of the core practices of SRI (water management, seedling age, number of seedlings per hill and spacing distance) and considerably higher for weeding or the fertilizer regime. Still within the scientific documents category, we observed that documents reporting on-station trials (mainly agronomic studies) are significantly more likely to define the SRI components in detail, compared with farmer field studies. This is perhaps not surprising, since on-station trials typically compare very well-defined treatments.

Table 8.1: Documents providing a definition of SRI components Scientific audience (n=100) Non-scientific audience (n=159)

Specified SRI Yes (%) No (%) Yes (%) No (%) component: Water Management 64 36 35 65 Seedling age 73 27 47 53 Seedlings per hill 62 38 45 55 Spacing 69 31 42 58 Weeding 46 54 31 69 Fertilizer 51 49 32 68 Nursery 15 85 8 92 Quick and gentle 9 91 10 90 transplanting

74 Findings in this section are based on cross-tabulations of the variables described. Significance findings imply a significant result (p<0.1) in a chi-squared test in a cross tabulation using an exact correction method. All calculations were carried out in PASW Statistics.

116 The ranking of practices most likely to be defined is similar across the two types of documents, with the first four practices most likely to be defined and seedling age the most commonly mentioned in both categories. However, among the scientific documents, spacing distance was the second most commonly described practice, followed by water management. The number of seedlings per hill was in fourth place. Among documents aimed at non- scientific audiences, the number of seedlings per hill was the second most commonly defined practice, which suggests that NGOs and practitioners place greater emphasis on transplanting single seedlings than scientists. Spacing distance and water management followed, in that order. The emphasis placed on these four practices by both categories of documents clearly indicates their significance as the signature practices that most clearly distinguish SRI from conventional rice cultivation systems. Among the scientific documents, specifying the weeding and fertilization regimes for SRI came some way behind the first four practices, whereas, in the non-scientific documents, these two principles were apparently not far behind water management in importance. One particularly interesting observation emerging from Table 8.1 is the relative lack of attention given to nursery management and the quick and gentle transplanting of seedlings, across both categories of documents. This is surprising in the light of the emphasis that Henri de Laulanié placed on these two components. The reasons why both scientific and non- scientific documents tend to neglect these aspects merit further investigation. A final observation is the fact that the number of publications intended for non-scientific audiences has increased in recent years. This category contains a broad range of documents from training manuals to project reports to newspaper articles, and probably signals the growing popularity of SRI over time. In our database, this observation is directly related to an increase in Indian documents promoting SRI. Tables 8.2, 8.3 and 8.4 provide further information on the SRI definitions given in the documents. Table 8.2 indicates how individual SRI components were specified in those documents that provided a specification. The frequencies shown in Table 8.3 indicate that most of the studies that provide a specification of a particular SRI component do so in a detailed way. Few studies used ambiguous terms such as ‘wider spacing’ (8 %) or ‘younger seedlings’ (7%). More importantly, Table 8.3 shows that some SRI components vary more than others. For instance, very large majorities of documents define SRI water management as Alternate Wetting and Drying irrigation (88%) and specify weed suppression using a mechanical weeder (86%). On the other hand, there is considerably more diversity in the way fertilizer regimes are defined, with some studies specifying the use of organic fertilizers only (28%), while a majority advocate or describe mixtures of organic and inorganic fertilizers (57%)

117 Table 8.2: Specifications of SRI components SRI Component: % reported Water management (n=137) - Continuous flooding 1 - Alternate wetting and drying 88 - Rainfed 3 - Aerobic 5 - Multiple water management strategies compared 4 Seedling ag e (n=137) - “young seedlings” 7 - multiple ages compared 9 - days specified exactly 75 - Leaf stage specified (only) 6 Seedlings per hill (n=144) - 1 seedling/hill 81 - 1-2 seedlings per hill 13 - more than 1-2 seedlings per hill 4 - multiple seedling densities compared 2 Plant spacing (n=152) - wider spacing 8 - multiple spacings compared 9 - centimetres specified 83 Weed ing regime (n=104) - multiple compared/various 5 - hand 7 - herbicide 39 - mechanical 86 Fert il ize r re gime (n=113) - no fertilizer 1 - organic 28 - chemical 13 - chemical and/or organic 57 Nursery (n=32) - conventional 12 - dapog 31 - others 56

Table 8.3: Specifications of seedling age Range of seedling age: lower bound (%) Range of seedling age: upper bound (%) 7–8 days 40 7-8 days 7 10–12 days 23 10-12 days 44 13-14 days 18 13-14 days 19 15-18 days 7 15-18 days 26 no minimum 12 >18 days 5

Table 8.4: Specifications of seedling spacing distance Range of plant spacing: lower bound (%) Range of plant spacing: upper bound (%) Less than 25 cm 22 Less than 25 cm 14 25 cm - 29 cm 63 25-29 cm 30 30 cm - 39 cm 13 30-39 cm 25 more than 40 cm 2 40-49 cm 18 no maximum "at least... 14

118

Table 8.3 and 8.4 show considerable variation in the specifications for the age and spacing of seedlings. Table 8.3 shows the lower and upper bounds of the seedling age range mentioned in the studies in the database, where a specification of 10–15-day-old seedlings has a lower bound of 10 days and an upper bound of 15 days. The table shows that a considerable proportion of studies recommends transplanting with a minimum age of 7-8 days (40 %), while only five per cent allow seedlings older than 18 days. However, the table also shows that there are considerable differences between documents. For example, 18 per cent of the documents specify a minimum seedling age of 13–14 days, which is already older than the maximum seedling age of 10–12 days specified by 44 per cent of the documents. One interesting feature of Tables 8.2 and 8.3 is the fact that very few documents specified the age of seedlings at transplanting in terms of the number of leaves. This is surprising because, according to SRI theory, the key time for transplanting is not primarily a matter of chronological time (e.g. a certain number of days) but of phyllochronological time (e.g. a certain number of intervals in a plant’s physiological development). Phyllochronological time may vary in chronological time according to climatic and weather conditions (see Chapter 3). According to SRI theory, it is highly desirable to transplant before the start of the fourth phyllochron, when the seedling has no more than two leaves. The start of the fourth phyllochron is likely to occur around 16 days after sowing in the nursery, but the precise timing is likely to vary from place to place, depending on local temperature and seasonal weather conditions. It would therefore seem useful to specify the ‘two-leaf stage’ as an objective, observable and adaptable measure that can be applied in different settings. Such a guideline might serve as a way to facilitate the flexible adaptation of SRI methods in different settings, which is supposed to be a key feature of the system. However, as shown in Table 8.2, very few documents specified seedling age in terms of a number of leaves, and a few of these were older than the two-leaf stage. Table 8.4 displays the range of specifications for spacing distances between hills. A large majority of the documents in the database (63%) specified a minimum SRI planting distance of 25x25 cm, whereas 14 per cent specified a maximum spacing narrower than 25 cm. There is considerably more diversity in the specifications of maximum spacing distances compared to minimum spacing distances. It is unlikely that differences in the specifications of SRI components are a result of random variation. They are more likely to reflect different experiences with and opinions about best SRI practice. The specifications may vary in patterned ways, between individuals and organizations, between agro-ecological or institutional settings or over time. We carried out some analytical tests to explore whether the variation in SRI components could be correlated to countries, year of publication, and types of publication / intended audiences.

119 Table 8.5: Specifications of SRI compared between countries Madagascar India China Seedling Age - minimum n=22 n=52 n=10 - 7–8 days 73% 29% 30% - 10–12 days 4% 25% 20% - 13–14 days 0% 31% 20% - 15–18 days 0% 10% 20% - no minimum 23% 6% 10% p-value Fisher ’s exact test : p =0.000 Seedling Age - maximum n=22 n=52 n=10 - 7–8 days 32% 2% 0% - 10–12 days 32% 40% 40% - 13–14 days 0% 38% 10% - 15–18 days 27% 17% 40% - >18 days and no maximum 9% 2% 10% p-value Fisher ’s exact test : p =0.000 Altern ate wetting and drying n=18 n=49 n=10 - Alternate wetting and drying 72% 92% 100% - rainfed 0% 2% 0% - aerobic 17% 0% 0% - not directly applicable 0% 2% 0% - multiple compared 11% 4% 0% p-value Fisher ’s exact test : p =0.086 Seedlings per hill n=24 n=46 n=11 - 1 seedling/hill 87% 80% 82% - 1–2 seedlings per hill 8% 11% 9% - more than 1–2 seedlings per hill 4% 9% 0% - multiple seedling densities 0% 0% 9% compared p-value Fisher ’s exact test : p =0.554 Spacing - minimum n=20 n=51 n=12 - Less than 25 cm 5% 43% 8% - 25 cm – 29 cm 95% 51% 58% - 30 cm – 39 cm 0% 6% 17% - more than 40 cm 0% 0% 17% p-value Fisher ’s exact test : p =0.000 Spacing – maximum n=20 n=51 n=12 - Less than 25 cm 0% 28% 0% - 25 –29 cm 10% 51% 8% - 30 –39 cm 35% 14% 25% - 40 –49 cm 40% 0% 42% - no maximum "at least.. .” 15% 8% 25% p-value Fisher ’s exact test : p =0.000 Weeding n=13 n=49 n=3 - multiple compared/various 8% 0% 0% - hand 0% 0% 33% - herbicide 0% 41% 0% - mechanical 92% 96% 67% p-value Fisher ’s exact test : p =0.049 Fertil iza tion strategy n=12 n=40 n=9 - organic 67% 17% 22% - chemical 0% 20% 22% - chemical and/or organic 33% 62% 56% p-value Fisher ’s exact test : p =0.019 The table gives the average scores per category within each SRI component. For example 73 % of studies from Madagascar report a minimum seedling age of 7–8 days, while only 29 % of studies from India report a similar minimum age. In addition, we report for each component the p-value of the Fisher exact test statistic, which tests whether differences between countries are statistically significant. A p-value smaller than 0.1 would thereby lead to the rejection of the null-hypothesis that SRI components are not defined differently across countries.

120

First, we analysed whether SRI specifications differed significantly across the three most frequently occurring countries in the database, Madagascar, India and China (see Table 8.5). For each component, we included only those documents that provided a specification of the component in question. Interestingly, there is a statistically significant difference in the specifications of most of the components (a p-value <0.1 in the Fisher exact test) across the three countries. The only exception is the number of seedlings per hill. For this component, the majority of documents specified one seedling per hill as SRI practice, a smaller number allowed up to two seedlings, but very few specifications allowed more than two seedlings per hill. All the other components differ significantly across countries, which is most apparent when comparing Madagascar to India. In Madagascar, seedling age is generally younger, spacing between plants is generally wider, and SRI is more frequently specified as involving the use of organic fertilizers only, when compared with Indian SRI specifications. On the other hand, Indian SRI specifications more frequently allow the use of herbicides and typically involve the combined use of organic and chemical fertilizers. Typical SRI practice in China closely resembles the Indian variant, although differences may be masked by the relatively small number of Chinese studies included in the database. Separately, we also compared the documents from other countries in Asia, Africa and the Americas, but this comparison did not highlight any major differences between SRI specifications in these countries. Again, real differences could be concealed by the small number of documents from these locations. We also examined the correlations between the year of publication and the specification of SRI components. We found that SRI specifications in more recent documents tend to include a slightly older seedling age and less frequently specify that SRI incorporates organic fertilizers only. These findings could be related to the relative prevalence of documents concerning SRI in Madagascar in earlier years and the increasing number of publications from India more recently. We did not find any significant differences in the specifications of spacing distances, water management practices or weeding regimes over time. Finally, we investigated whether the specifications of SRI components differs across the type of publication or intended audience. Only in the case of seedling age was a significant difference observed, where documents intended for scientific audiences tend to specify younger seedlings than those intended primarily for non-scientific audiences. This is a very interesting observation in view of the stress placed on young seedlings in SRI theory (see Chapter 3) and the possible advantages of transplanting young seedlings in terms of rice growth and yield (see Chapter 4). It may relate to the observation that few documents specify the seedling age in terms of the number of leaves (see above), and the findings of various adoption studies which showed that few farmers actually adopt the practice of using very young seedlings (see Chapter 5). It may signal that the use of very young seedlings is discussed primarily within the scientific domain, whereas documents that are used for SRI promotion and extension more frequently recommend slightly older seedlings. This would make sense if, for example, the concerns expressed by some farmers, extensionists and

121 project workers about using young seedlings – e.g. that they pose an elevated risk for poor farmers – are widely shared (see Chapters 3, 5 and 6). It is important to note, however, that this observation could again be an artefact of the increase of SRI activity in India, where seedling ages are typically older than in Madagascar. Overall, the analysis in this section suggests that SRI has, in general, taken a different shape in India compared to Madagascar. The question remains, what might have caused such differences to emerge in the way SRI is conceptualized, promoted or practised? The causes could be associated with agro-ecological, socio-economic or institutional factors, including soil fertility, temperature, day length, input prices, wage rates, and so on. These remain to be studied in detail. In the next section, we explore some specific variants and adaptations in SRI practice that have emerged in different countries. 8.2 Variations on a theme: SRI-like rice cultivation systems During the course of our research, a number of SRI-like rice cultivation systems came to our attention. These include not only some adapted systems that were directly inspired by SRI, but also a number of other systems that seem to share some features with SRI without having any particular connection to the ideas of Henri de Laulanié. The latter category includes some fascinating historical precedents that seem to prefigure SRI to a remarkable degree. We discuss these systems here because they can shed light on the SRI phenomenon from several different directions. First, they make it possible to assess the basis for the claim that SRI is an intrinsically flexible and adaptable system. Second, they may shed light on the origins of SRI itself, including Henri de Laulanié’s claim that he had devised a system that was firmly based on the physiology of rice itself. Third, they can help to expose which elements of SRI are found most challenging or problematic in practice, leading to efforts to make adjustments in response. Finally, the emergence of alternative systems perhaps indicates the attraction of transforming a set of rice cultivation guidelines into a kind of package, with a distinctive label. 8.2.1 Rice cultivation systems inspired by SRI Mitsitsy Ambioka sy Fomba Fiasa (MAFF) is the rice cultivation system developed and extended by Patrick Vallois, the Madagascar-based development consultant who was among the earliest people to take up the ideas of Henri de Laulanié (see Chapters 3 and 6). Between 1997 and 2004, Vallois was involved in various projects to improve rice cultivation practices, based on SRI methods. Drawing from these experiences, he felt a need to adapt SRI in several key ways. In particular, he conceived MAFF as a less risky system than SRI, and therefore a more suitable one for poor and marginal farmers. An important part of the MAFF approach was that emphasis was placed on ‘investing less and producing at least enough ’, rather than dramatically improving yields (Vallois 2005: 6, emphasis in original). Vallois defined MAFF as follows: A new method for training in the improvement of rice cultivation, retaining many points from the System of intensive rice cultivation (SRI), but oriented towards economising on seeds more than extremely young seedlings and the promise of extraordinary yields, with an explication that is analytical and explanatory rather than

122 synthetic and dogmatic, and introducing flexibility and understanding of the system rather than demanding perfection. 75 MAFF was thus a more relaxed version of SRI in some key respects. To begin with, the emphasis was placed on the general principle of a low transplanting density, rather than a strict guideline about spacing distances. Second, the recommended seedling age at transplanting was 14–20 days compared to 8–12 or 8–15 days in other Madagascar SRI variants. This was still a good deal younger than the 25–40 days in conventional practice, but removed the necessity to handle extremely young seedlings and reduced the risk of seedlings dying. De Laulanié’s emphasis on gentle handling of seedlings, quick and shallow transplanting and placing seedlings singly were retained but, finally, row or square planting was only recommended to those farmers who already used mechanical weeders. Another SRI variant that has emerged in Madagascar has been labelled SRIA, which appears to have been derived from a combination of SRA and SRI (Rasoanaivo et al. 2006). This system has been promoted in a project supported by Inter Aide and F3E 76 and is described as ‘a bit less “strict” for the farmers than the soft version of SRI (MAFF) currently diffused by Patrick Vallois’ team in PC 15 (Lac Alaotra)’ (Rasoanaivo et al. 2006). However, the key recommendations under this system are scarcely distinguishable from SRI: seedlings 10–20 days old; 1–2 seedlings per hill; and square planting at 25x25 cm spacing. The project also stresses precise levelling of rice plots and the use of mechanical weeders. It is hard to see these prescriptions as a ‘less strict version’ of MAFF. The blending of SRA and SRI is not particularly unusual in Madagascar (see Chapter 6). Another example of a system that is said to be a modified version of SRI, but which actually seems to have few substantive differences, is the Sustainable System of Irrigated Agriculture (SSIA), developed in the Philippines. SSIA is practically indistinguishable from Henri de Laulanié’s original version of SRI, with the small exception that a dapog nursery is used (Kikuchi & Xie 2008). (The use of a dapog nursery with SRI is actually not very unusual – see Table 8.2.) Our picture of the SRI scene in China is far from complete, but there are indications that certain variations in SRI practice have emerged there too. According to Cai et al. (2008), these were prompted by concerns among some rice experts after SRI methods were first introduced in China, which lead to the elaboration of a variety of alternative intensification methods suited to local conditions. For example, Zheng et al. (2004) reported that SRI methods had improved rice yield attributes and grain yield but that the sparse planting had been found problematic in the low solar-radiation conditions of Sichuan province. In addition, the weeding and water-management techniques had been found laborious and difficult. Accordingly, Zheng and colleagues investigated various possible adaptations of SRI, including triangle and oblong planting; they reported having achieved yields of 12 t/ha. using their adapted system. Later, Wang et al. (2006) reported on trials with the ‘Large-

75 http://www.erails.net/MG/divers/la-methode-maff/contexte-general/ (accessed 29 January 2011). 76 Fonds pour la promotion des études préalables, études transversales, évaluations (Fund for the promotion of preliminary studies, transverse studies and evaluations).

123 triangle intensification system’, which integrated SRI methods with another method, called ‘three dimensional cultivation strengthening’, that had been developed at Sichuan University. Chinese scientists have evidently also developed a rice cultivation system that combines principles from SRI with no-tillage practices (GUO et al. 2007; ZHENG et al. 2006). This is particularly interesting in the light of the stress placed on the importance of active soil aeration in SRI by some practitioners and scientists (see Chapters 4 and 6). Chinese SRI research has also investigated the role of varietal choice in SRI, combining SRI management methods with elite rice lines to achieve ‘super-high yield’ (SUI & Wang 2008). The use of elite rice varieties is not at all incompatible with SRI cultivation methods, although it does represent a change of emphasis compared to the views of de Laulanié and others, who stressed the value of changing cultivation practices without having to adopt new seeds. Presumably, the Chinese ‘super high yield’ experiments reflect the wide availability of improved rice varieties, which were not readily available in Madagascar during the 1970s and 1980s. In Chapter 3, we discussed the fact that de Laulanié designed his rice cultivation system not only around the physiology of rice but also based on his assessment of the capacities of poor and marginal farmers. This means that SRI theories about the growth patterns and physiology of rice could equally have been used to justify an alternative cultivation system based on large-scale mechanized farming instead of small-scale subsistence agriculture. Agribusiness companies in Pakistan 77 and Nepal 78 are developing systems they call ‘mechanized SRI’, which have adopted some SRI techniques but incorporated them into a very different technical approach from the one developed by de Laulanié. Motorized or tractor-drawn implements are used for constructing raised beds, transplanting, weeding and harvesting (Glover forthcoming). In Tamil Nadu, we also saw indications of an emerging trend towards mechanization in rice cultivation. Combine harvesters are already common. Farmers who can afford to do so are also using mechanized transplanters or direct seeders to plant their rice. Some weeding is now done using motorized weeders. On the largest farms, sophisticated laser land levellers are appearing (see Chapter 6). These trends have implications for the type of SRI that can be applied. Drum seeders and mechanized transplanters are not currently precise enough to plant just one seed or seedling at a time and it is open to question whether they will ever be refined enough to do so. In addition, machines such as transplanters, drum seeders and weeders are generally built in standard widths that allow only a limited range of adjustment in the spacing distances between rows. Moreover, with transplanting machines and drum seeders it is difficult to achieve a regular spacing distance between plants within rows as well as between rows. These constraints have implications for the idea that plant spacing distances should be carefully optimized for local soil and climatic conditions.

77 See http://www.farmalltechnology.com/index.php?option=com_content&view=article&id=18&Itemid=16 (accessed on 1 July 2010). 78 Project implemented by the Indian firm Mahindra & Mahindra, observed by Dominic Glover near Biratnagar, Nepal, November 2009.

124 8.2.2 Independent contemporary and historical parallels to SRI methods It has recently come to light that some rice cultivation systems strikingly similar to SRI were being practised in different locations at earlier times. For example, Indian researchers recently uncovered evidence which showed that a ‘single-seedling method’ was being practised in the Madras Presidency of British India (modern Tamil Nadu), with official support from colonial agriculture officials, in the early years of the twentieth century. The system involved transplanting single seedlings in widely spaced rows or squares and was reported to produce good yields (Gujja & Thiyagarajan 2009; Thiyagarajan & Gujja 2009). Miyazato et al. (2010) have referred to an intensified rice cultivation system that was being practised in the Philippines as early as 1954, known as the Margate System of Rice Production, which involved the transplanting of young seedlings, row planting, and intermittent irrigation. It is possible that the Margate system is the same as, or a version of, the dapog nursery-based system, which was originally practised in the Philippines and, like SRI, involves the transplanting of young seedlings. As we saw in Chapter 3, Henri de Laulanié was aware of the dapog system because it had been tried experimentally in Madagascar as early as 1964. De Laulanié thought that the dapog nursery was decidedly different from the ‘garden-like’ nursery he advocated, but in practice dapog nurseries are frequently combined with SRI in both scientific experiments and NGO recommendations (see Table 8.2). In Chapter 4, direct seeding was discussed as an alternative crop establishment method, which some people have suggested could be combined with other SRI techniques in order to avoid transplanting stress. Henri de Laulanié thought that direct seeding was too challenging for poor farmers, even though he believed that it could be beneficial for rice plants (Glover forthcoming) (see Chapter 3). In fact, however, a direct-seeding method known as gogorancah was used by smallholder farmers for rainfed rice cultivation in Indonesia before 1905 (Fagi & Kartaatmadja 2002). Even more remarkably, Henri de Laulanié himself was aware of some farmers in Madagascar who practised a form of direct seeding that he called ‘seed transplanting’, because the seeds were carefully placed one by one into the main rice field. As with SRI, the seeds were carefully spaced to allow room for the plants to grow. Although this was obviously a laborious practice, de Laulanié acknowledged that it would save time that would otherwise be spent on nursery management and transplanting seedlings (de Laulanié 2003). Sital (2002) has reported that Hindustani and Javanese immigrants to Suriname in the last part of the 19 th century brought with them rice cultivation practices that included transplanting two or three rice seedlings into the main rice field using a kind of dibble stick, at spacing distances between 20 and 30 cm. A book published by the Indian Council of Agricultural Research in 1960 shows that the rice cultivation practices prevailing in India at that time included not only row planting and weed suppression using mechanical rotary weeders – modern practices that were being introduced to India during the 1950s – but even a type of intermittent irrigation that was being practised in an area of modern Bihar (ICAR 1960).

125 Norman Uphoff has reported that he came across an SRI-like rice cultivation system on a trip to China in February 2004. Known as the ‘3S’ system, it was developed independently by Prof. Jin Xueyong for application in the cold climate of Heilongjiong, China’s northernmost province. Seedlings are first planted in greenhouses at the end of the winter and transplanted when they are about 45 days old. They are transplanted in clumps of 1–2 seedlings per hill and in a rectangular pattern of about 15x40 cm. Rice plants cultivated with this system reportedly displayed vigorous tillering and root growth, a high quality grain and a higher yield compared to conventional local methods. 79 It is important to point out that mainstream textbooks and manuals on rice and rice cultivation also acknowledge a range of known practices that resemble SRI in certain respects. For instance, the sixth edition of D. H. Grist’s Rice (1986, first edition 1959) describes the following transplanting practices from various locations. First, a set of practices that closely resemble de Laulanié’s recommendations about gentle handling and careful transplanting of seedlings: In Hong Kong and parts of China … seedlings are removed from the nursery with care by means of a sharp, flat hoe of special design (Herklots 1948). The blade of the hoe is pushed into the bed so as to lift a patch of seedlings together with the soil and fertilizer in the immediate vicinity of the seedlings. It is to be noted that in contrast to the system obtaining elsewhere, seedlings with adhering soil are planted in the field. Sometimes seedlings for a second crop are removed by hand in Hong Kong. Rough treatment of seedlings for the first crop is avoided because the seedlings are smaller, having been in the nursery for about twenty-five days. It is also possibly intended to retain the fertilizer applied to the nursery in close contact with the roots (Grist 1986: 162). Grist goes on to describe a number of different transplanting practices relating to seedling age and transplanting density. His summary of conventional practices in Asia is not very close to SRI recommendations: A bunch of seedlings is held in the left hand, from two to six are transferred to the right hand and thrust into the mud, the distance between ‘hills’ or plants being from 10 to 15 cm, depending on the variety, local conditions and custom (Grist 1986: 162). However, he continues by giving an account of several regional variations of these practices, as well as experiments which illustrate the kinds of physiological principles and cultural methods that had been investigated many years earlier. Some of Grist’s observations resemble elements of Henri de Laulanié’s (2003) thinking on rice cultivation remarkably closely, touching on topics such as transplanting density, depth of standing water, variety choice, and labour requirements, including the interactions between alternative practices: Experiments in India (Hedayetullah et al. 1947) showed that the number of fertile tillers was greater with three to four seedlings per hill and decreased with closer planting; length of panicle and yields increased with wider spacing. On first-class land in Malaysia, a spacing of 37 to 45 cm is usual, using a seven months’ variety of paddy. Jack (1923) estimates that where the water in the field is deeper than 28 to 37 cm is becomes necessary to space the plants somewhat closer than in shallow water,

79 http://sri.ciifad.cornell.edu/countries/china/cn3ssys.html (accessed 3 February 2011).

126 because deep water inhibits tillering considerably. In Burma, from one to four plants per hill are spaced from 10 to 15 cm apart. The optimum spacing in Sri Lanka is 10 to 15 cm, one seedling only being placed at each point, except where there is liability to crab damage, when two seedlings are planted (Paul 1945). In view of this very close planting, it is not surprising that it required 35 to 40 women to plant 1 ha. per day, as compared with 25 to 30 in other countries where wider spacing is usual and the number of seedlings per hill greater. In the Philippines 5 to 7 seedlings at the age of 10 to 15 days are transplanted per hill: in most other countries two or more seedlings are planted per hill at the age of 30 to 40 days. In Japan, the so-called triangular or hexagonal spacing is usually adopted, in which the interval between hills in the row is narrowed. The object is to give the plants sufficient sunlight despite close planting and to allow passage for a weeding implement in three directions (Grist 1986: 162–3). Grist continues with some more observations on the proper transplanting technique – which closely resemble de Laulanié’s recommendations – as well as the interactions between variety, spacing distance and fertilization: Shallow transplanting is recommended, for if seedlings are planted deeply the roots fail to develop normally and a new system of roots must develop from the upper nodes. This delay in root formation retards plant growth. Roots near the surface are subject to higher temperatures during the day and lower temperatures at night and this encourages tillering. Furthermore, as has been pointed out in the Philippines (IRRI 1964a), with high nitrogen applications, tall heavy-tillering rice varieties respond much more to nitrogen, especially during the rainy season, when they are widely spaced, but medium- to low-tillering varieties of shorter stature yield better under closer planting, even when nitrogen is applied (Grist 1986: 164). Grist’s book also discusses the topics of irrigation and drainage in great detail. Although he is firm that ‘it has now been proved that nutrition of the plant is in large measure assured by inundation’ (Grist 1986: 41), listing the several reasons for drawing this conclusion, he also notes that ‘Drawbacks of inundation are depletion of free and combined oxygen from the subsoil, and accumulation of various organic acids which retard root development, inhibit nutrient absorption and normal aerobic respiration and cause root rot’ (Grist 1986: 41). Finally, Grist also discusses practices for weed management, including Japanese practices from the early 1950s whereby a rotary hoe was used for both weed removal and intercultivation, three or four times, beginning 10 or 12 days after transplanting (Grist 1986: 164). It is significant that four of the five sources cited by Grist in the passages reproduced here predate de Laulanié’s work in Madagascar by many years: 1923, 1945, 1947 and 1948. The fifth was published just three years after de Laulanié arrived on the island. The first edition of Grist’s book was already in print before de Laulanié departed from France. The independent development of several similar rice cultivation systems gives some weight to de Laulanié’s view that he had stumbled across a set of practices that suited the intrinsic properties of rice itself; it also raises doubts about the suggestion – which anyway was not entirely de Laulanié’s own claim – that SRI practices were a radical novelty in rice cultivation. Bearing in mind the 10,000 year history of human rice cultivation, it seems eminently plausible that the same features that de Laulanié found attractive could also have

127 been stumbled upon by other people or communities or scientists over the course of history. This suggests to us that a more thorough survey of historical and contemporary practices in rice cultivation would be very valuable to our understanding of what the SRI phenomenon represents. 8.3 Systems of Crop Intensification? Since SRI is regarded by many people and organizations fundamentally as a set of general agro-ecological principles rather than a fixed system for growing rice, some of them have begun to explore the possibility of applying similar kinds of principles to other cropping systems. Such systems have been reported for wheat, sugarcane, finger millets (ragi), teff and maize, and even for pulses, mustard, aubergine (brinjal, eggplant), onions, potatoes and carrots. We have not investigated these cropping systems thoroughly and details on the principles being applied are very sketchy. In the case of the grassy grain crops, the basic concept adopted from SRI seems to be the planting of single and widely spaced seeds or seedlings. In the cases of the vegetable crops, it is hard to imagine what similarity these crop intensification systems have to SRI on a technical level. 80 This suggests that what unites these cropping systems has as much to do with linkages at institutional levels and a general philosophical approach, as with specific technical practices. This observation reinforces our view that it is important to appreciate the social and institutional dynamics of the SRI phenomenon alongside its technical features. We now move to the final chapter, where we will draw together the insights and conclusions from the report as a whole, and propose ways forward for a deeper understanding of SRI.

80 http://sri.ciifad.cornell.edu/aboutsri/othercrops/index.html (accessed 3 February 2011).

128

9. Ways forward

This chapter reviews the findings of our exploratory study. It summarises what can (and cannot) be said with confidence about SRI, discusses the important questions raised, suggests some hypotheses that could form a foundation for further investigation, and makes proposals for a conceptual and methodological framework for integrated research on the socio- economic, institutional and technical dimensions of the SRI phenomenon. 9.1 What we know and do not know about SRI

Perhaps the clearest of all the lessons to emerge from our work is that it is a mistake to think of SRI as a radical invention or discovery, in the sense of a unique, original and unprecedented novelty in rice cultivation. As we discussed in Chapter 3, Henri de Laulanié drew on farmers’ practices, IRRI manuals, agronomic research and contemporary extension service recommendations, as well as his empirical observation of rice plants, to develop his system. Every individual element of SRI practice has precedents, and there are even cases where a combination of practices produced a cultivation system strikingly similar to SRI (see Chapter 8). De Laulanié’s innovation was the real but modest one of combining different practices together and promoting them as an integrated system. Even then, it is clear that the interactions that de Laulanié was grappling with – among spacing distance, water management, tillering, variety, fertilization regime and weed control, etc. – had been fairly well understood for many years (see Chapter 8). These observations bring mixed blessings for the promoters and advocates of SRI. On one hand, the existence of multiple precedents for SRI practices, and even some SRI-like systems, demonstrates that SRI methods rest on a reasonably firm foundation, buttressed by farmer practice and agronomic science. On the other hand, the existence of these other systems rather undermines the claim, which has always been implicit if not explicit, that SRI represents a completely new and definitively improved system of rice cultivation. A related notion, which also has not withstood close scrutiny, is that SRI represents the particular combination of practices that is peculiarly appropriate for the specific physiological characteristics of rice. As we discussed in Chapter 3, de Laulanié designed SRI not only for rice but also for the poor and marginal Malagasy farmers he was trying to help. His insights into the physiology and growth of rice plants could have been used equally well to create other kinds of rice cultivation system. Indeed, this has been done by others since de Laulanié’s death (see Chapter 8). In other words, there is more than one appropriate way to cultivate rice, where appropriateness depends not only on the biophysical characteristics of rice and the agro-ecological setting, but also the socio-economic and institutional context in which the cultivating is done. SRI was a cultivation system designed in and for a particular time and place. The fact that it has subsequently been taken up by farmers, NGOs and other organizations working in later times and other places suggests that those actors see something of value in SRI that could be applicable to their circumstances as well.

129 This represents the second key insight of our study, namely that SRI was shaped by the constrained circumstances and relative isolation in which Henri de Laulanié and his collaborators worked. SRI’s development reflected the degree to which de Laulanié, a field- level agronomist working in a poor developing country, was disconnected from the world of scientific agronomy. He was forced to fall back on his limited resources because he found it hard to access scientific information that could have helped him in his work. The fact that de Laulanié ended up by developing a system whose elements resembled other rice cultivation systems is not surprising, and suggests indeed that he did not do a bad job. But it did have an impact on the next phase of the SRI story, because the claim that a new, dramatically improved rice cultivation system had emerged from a seminary in the highlands of Madagascar was bound to be provocative to mainstream rice scientists, especially when they looked at this new system and found that it had features they thought they recognized. This leads us to the third key insight of our study, which is that the Rice Wars controversy was not merely a disagreement about scientific methods, theory or empirical observations. It also had an important institutional dynamic, in which the protagonists framed their arguments in terms of rather different perspectives and priorities. On the surface, the argument was about whether the record yields that had been claimed were actually achievable. Beneath the surface, the Rice Wars represented a confrontation between two styles of science, one an inductive, applied, field-level science and the other a rarefied and sophisticated, deductive, high technology science. The implied disagreement between them was about the appropriate role of science – whether it should be about raising the yield ceiling of rice, using plant breeding and cutting-edge biotechnological approaches, or tackling the yield gap that separates farmers’ yields from the much higher yields achieved on research stations, by devising technologies and support strategies that would be appropriate and accessible for poor and marginal farmers (Glover forthcoming). This is the reason why one set of protagonists in the debate celebrated, rather than apologising for, the ‘empirical’ and ‘experiential’ nature of SRI (Stoop & Kassam 2005), while the other argued that the optimal methods for rice cultivation were already known, that the constraints to further advances had already been identified, and that genetic improvement was the only way to increase the physiological yield potential of rice (Sheehy et al. 2004). The two sides were arguing about very different strategies for international agricultural development, based on rather different philosophies concerning the appropriate relationship between agronomic science and its application in farmers’ fields (Glover forthcoming). The fact that SRI has grown and spread from Madagascar implies that it is seen by some people and organizations to answer a demand for economically and ecologically sustainable agricultural technologies that are suitable and appropriate for smallholder agriculture in the developing world. We suggest that SRI would merit further study for that reason by itself, though not only for that reason. During our study, we also found that the technical practices of SRI cultivation seem to merit further investigation in their own right; and that the incorporation of these practices into farming systems and communities raises numerous questions that invite further research.

130 Overall, our review of the available scientific and grey literature indicates that SRI methods may indeed increase productivity and raise yields, at least for particular types of farmers and in certain contexts. In particular, there is evidence to indicate that SRI methods have the potential to produce good yields while dramatically lowering the seed rate and reducing water consumption. SRI practices could therefore make a valuable contribution to development, increasing some farmers’ incomes and enhancing their household food security while boosting global rice production and conserving water supplies. Such potential benefits are obviously very attractive. However, the magnitude of the potential benefits, the balance between benefits and costs, the underlying biophysical principles, and the interactions between different elements of the system – including all the various social and technical components – remain rather obscure. Nevertheless, our exploratory study gives us the confidence to reach some general conclusions and identify some important knowledge gaps. 9.1.1 Biophysical mechanisms, agronomic and crop management changes

Among the individual component practices of SRI, as we discussed in Chapter 4, there is some evidence that the transplanting of younger seedlings can lead to more vigorous tillering and ultimately a higher grain yield. However, there have also been exceptions to this general rule. Explaining these incongruous results seems to require further analysis or experimentation. It is important to note, in addition, that profuse tillering will not necessarily lead to a higher yield (see below). Studies designed to investigate the specific practice of single-seedling transplanting are relatively scarce, and those studies that have tried to measure the impact of this practice have not found a strong effect. In practice, some farmers and agriculture experts express concern about the risk of seedling death and it seems that transplanting two or three seedlings is regarded by some SRI-promoting organizations as an appropriate relaxation of SRI guidelines. This is particularly the case on saline or acid soils. Evidence to support the very wide spacing of hills is weak, as discussed in Chapter 4. Fairly consistently, the best results with SRI management have been achieved with mid-range plant spacings, about 20x20 cm or 25x25 cm. Wider and narrower spacings have, in general, been found to have an adverse effect on yield, which suggests that yields and spacing are related in an inverse U pattern. However, it should be noted that spacing distances of 30x30 cm and 35x35 cm are not unprecedented in rice agriculture, so it may be that wider spacing can be successful on very fertile plots. This may explain the good results reported for very wide spacing in Madagascar. It seems to be now fairly well established that, in many cases, water use in rice farming can be reduced significantly without having a large adverse effect on production levels. In some cases, modest or substantial yield increases are also possible, compared to continuously flooded irrigation. In such cases, the enhanced yield may be due to beneficial effects of soil aeration, but it is difficult to disentangle the effects of reduced irrigation from the increase in fertilizer that is also recommended for SRI.

131 The mechanisms contributing to plant health status, growth and productivity under aerobic soil management remain rather obscure to us. We are aware that the interactions among aerobic and anaerobic soil bacteria, soil fungi and plant roots are likely to be complex, dynamic and site-specific, and their individual and combined effects may be both positive and negative for plants. Certainly, aerobic bacteria are more likely to be abundant in non- saturated soils and it may be that maintaining moist soil conditions is a pragmatic way to maintain a mixture of aerobic and anaerobic micro-organisms. Research has shown that certain soil bacteria may be much more abundant in moist, non-saturated soils, but the precise mechanisms by which this phenomenon may contribute to crop growth remain to be investigated in further detail. The possible contribution of soil disturbance to the promotion of aerobic soil conditions is not clear. The mechanical rotary hoe was originally introduced as a weed-management tool and the importance of suppressing weed competition is unquestionably important, especially with wider spacing distances and younger, more vulnerable seedlings. Whether the mechanical weeder, as it is typically used, measurably increases available oxygen in the root zone is unclear. It has also been suggested that weeders may encourage longer and deeper root systems by pruning the lateral roots, although this has not been demonstrated conclusively. Deeper root systems may also result when water-saving irrigation is applied on well-drained soils. The benefits of deeper roots have been questioned because most soil nutrients are thought to be concentrated in the top layers of the soil, though that would seem to depend partly on the nature and quantity of fertilizer or crop residues that are being added to the soil surface. In short, we perceive that substantial further research is required in order to determine what mechanisms may be at work in terms of both soil aeration and the effects of aerobic soil conditions, as well as their relative advantages and disadvantages in conjunction with different soils, fertilizer applications and irrigation regimes. It is logical that the addition of organic compost or manure will stimulate soil micro- organisms. This may have beneficial effects on soil fertility and the interactions between soil biota and plant roots, though the actual effects will also depend on the water regime being used. The availability of high-quality organic matter is perceived as a key constraint in many locations. Access to chemical fertilizers may equally be a problem for some farmers, for different reasons. Even if organic fertilizers are available, depending on the existing condition of soils, it is widely accepted that applications of inorganic fertilizer may still be desirable in particular locations, in order to provide specific nutrients. Some research has suggested that SRI methods improve the availability of existing nutrients in soils. Concerns remain about the long-term sustainability of boosting rice production without replenishing soil nutrients. These concerns seem to have been stimulated partly by the reports that high rice yields have been achieved on poor soils with minimal soil amendments. Such concerns may be less urgent if it is confirmed that the increased yields are partly attributable to the higher fertility of plots that have actually received large quantities of fertilizer (see Chapters 4 and 5). Nonetheless, it has also been suggested that the addition of quite modest quantities of organic fertilizer can ‘incite’ vigorous microbial life in the soil, helping to feed the rice plants even on poor soils. Further research may be necessary in order to reconcile these alternative explanations.

132 Overall, as the discussion in Chapter 4 showed, it has been confirmed that rice under SRI management displays distinctive physiological and morphological characteristics compared to rice grown under conventional systems. These changes include a more open plant architecture that can capture more solar energy; larger roots that are less affected by root senescence and can more firmly anchor the plant; differences in metabolic processes during the growing season; more tillers; larger leaves; more and larger panicles; and more grains per panicle. The proportion of fertile grains per panicle may be lower than with conventional methods, however, and the number of productive tillers per m 2 may also be lower because of the lower-density planting pattern. Achieving a higher yield with SRI methods then depends on whether improved yield attributes such as a higher 1,000-grain weight and larger panicles can compensate for a reduced plant density. Identifying and selecting the optimal planting density therefore seems to be a key issue for SRI. It is evident from the preceding paragraphs that considerable complexity arises from the relationships and possible interactions between different SRI practices. This brings us into the contested field of possible synergies. Some positive but also some negative interaction effects have been identified between pairs of SRI practices. Synergetic relationships encompassing the full suite of SRI practices have also been found, but these have not been confirmed by other studies. Our reading of the literature on this topic leads us to raise a question about the concept of synergy. Our perception is that the definition or meaning of synergy has not been adequately elucidated. We believe that, in its richest sense, synergy implies an emergent, multiplier effect through which additional value is added to a system by means of the interaction among the constituent parts. However, the term synergy is often used in a much simpler sense, to describe rather straightforward relationships of dependence or interdependence. A dependent relationship simply means that the full potential of a given element can only be achieved in the presence of a supporting or enabling element. Or, to put it another way, that two or more elements are mutually coherent and make sense together: to say that an electric plug and an electric socket are synergetic would not be very meaningful, whereas it is true that their individual functions are mutually interdependent and that each artefact only makes sense in conjunction with the other. It stands to reason that a farmer’s choices among alternative seedling ages, densities per hill, spacing distances and fertilizer regime should fit with one another and be conditioned by his or her soil quality and water management system. Does it add something to say that the relationships among these elements are synergetic? We are not sure. Claims about synergies were one of the chief causes of the Rice Wars controversy, because the idea of synergy was used to underpin the claim that SRI methods could unlock a hitherto unsuspected yield potential in the rice genome (see Chapter 4). We think that some of the disagreements over SRI could be clarified (though not necessarily resolved) if the scientists involved were to specify more precisely what kind of mechanism they mean when they discuss the possible presence of synergetic interactions among the SRI practices. Then, the discussion could focus on the validity of the proposed relationships, and experiments could

133 be designed and arguments refined in order to address whether such relationships are present or producing a given effect. Working out which individual practices or interactions are responsible for which beneficial effects, and the magnitude of these effects, is a vital task for agronomic science if it is to provide reliable and relevant guidance to farmers. Farmers need to know which practices or combinations of practices are likely to achieve what kinds of effects in their specific contexts. This is especially important for poor and marginal farmers who need to optimize their cultivation systems in seriously resource-constrained settings. Judging by our review of the scientific literature on the biophysical aspects of SRI cultivation, much work remains to be done in this area. However, agronomic research could and should be informed by the experiences of farmers and field-level extension workers, who have tried and adapted the methods. This brings us to our insights and conclusions from our analysis of what is currently known about the spread, practice and impacts of SRI in different locations. 9.1.2 Spread and impact of SRI

As we saw in Chapter 5, reliable data on the spread of SRI methods is very scarce. Such information as does exist is fragmented, sketchy and hard to use for meaningful comparisons across sites. The small number of methodical adoption studies that have been carried out are mostly very small-scale and do not allow an assessment to be made of the spread of SRI relative to conventional practices. They have also been concentrated in just four of the nearly 50 countries where SRI is said to have been adopted – Cambodia, India, Madagascar and Sri Lanka. Very little is known about the patterns of adoption. The adoption studies we have reviewed suggest that levels of ‘full adoption’ of SRI practices are low in proportion to the total number of farmers who have tried SRI methods. Some disadoption has been observed. Very little concrete information is available to evaluate or account for patterns of partial adoption or disadoption. In particular, it is not yet fully understood how such patterns are influenced by biophysical, socio-economic or institutional characteristics. The studies carried out in Madagascar suggest that SRI is more likely to be adopted (or sustained) by better endowed farmers, and/or on more fertile rice plots with better access to a dependable irrigation source. These findings corresponded with the views of farmers we met in Madagascar, who stressed that water control was a decisive factor and did not necessarily apply SRI methods on all the land under their control. This may be disappointing, in light of the argument that SRI should be especially suitable for poor and marginal farmers because it does not depend on capital investments. However, it should be stated that whereas poor and marginal farmers may not be among the early adopters, SRI methods may nevertheless be beneficial to them if they can manage to adopt them. The available evidence does indicate that SRI practices can substantially improve the productivity of seed, land and water, which could indicate a ‘technology effect’. The increases in seed and water productivity follow logically from reductions in the seed rate and irrigation, which can evidently be achieved without adverse effects on output, and sometimes with an increase in yield. It should be noted that water savings and improvements in water

134 productivity can be achieved by adopting water-saving management methods, without necessarily adopting the whole set of SRI practices. The nature and exact cause of the increase in land productivity are less clear. Differences in land productivity between SRI plots and non-SRI plots (or farmers) could be a cause or an effect of improved results with SRI methods. Or, such differences could be linked to other, observable or unobservable, plot or farmer characteristics. In other words, it is possible that higher yields under SRI management may be attributed to a preferential allocation of SRI to more fertile plots, and/or to a preferential allocation of fertilizer and labour to SRI plots. On the other hand, it has also been argued that SRI methods tend to improve soils over time, which could be another, longer-term effect of SRI. Consequently, it is difficult to determine whether the productivity of labour and fertilizer (chemical and organic) also increase with SRI management. There is no firm consensus on the labour-increasing or -decreasing effects of SRI methods, although it is clear that the adoption of SRI transplanting and weeding methods leads to significant changes in the organization of tasks and in the temporal distribution of labour demand, including the possibility of an increased labour requirement at harvest time. There is a widely held view that SRI methods may increase labour demand in the short term but that the labour requirement can be reduced once the new methods have been mastered, and this is by no means an unreasonable proposition. We have not found any studies that have explored this dynamic in detail. Assessing possible changes in the productivity of fertilizer is difficult because of the fact that improvements in the fertilizer regime are typically part and parcel of SRI promotion. This makes it hard to determine whether the improvements in the productivity of seed, land and water are the consequence of a technology effect (such as a synergetic interaction) or merely an increase in fertilizer inputs. Some evidence exists to suggest that SRI practice enables rice plants to exploit soil nutrients more effectively, which is in accordance with the theory that soil aeration promotes favourable conditions for soil bacteria, which in turn interact with rice roots to feed the rice plant. The impacts of changes in labour demand, including its temporal distribution, are likely to be different for different households or groups within a community. For example, an overall reduction in labour demand may be a positive effect for larger farmers but negative for people who rely on selling their labour. Alternatively, a peak in labour demand for transplanting may improve income-earning opportunities for rural labourers, while possibly inhibiting their ability to apply SRI methods on their own fields, if they have any. Changes in the gender division of labour, the organization of task groups and the nature of tasks such as weeding may also have far-reaching effects on households, communities and labour markets, but these effects have not been investigated in detail. Some studies have indicated that the variability of yields increases under SRI management compared to conventional practice. This represents a possible source of elevated risk, especially for the most vulnerable farmers. Another possible hazard is the loss of seedlings when they are very small, due to pests, flooding or soil salinity. These factors, together with the physical difficulty involved in handling tiny seedlings, may explain why some farmers

135 and SRI-promoting organizations have opted not to use extremely young seedlings. Counterbalancing these concerns, SRI may well be a superior production method in areas where iron toxicity is a problem. Our exploratory review has only enabled us to gain a relatively superficial impression of the SRI phenomenon on an institutional level, because the landscape of individuals and organizations involved in SRI activity around the world is evidently very broad and diverse. It would therefore be difficult and unwise to make generalizations about the causes and influences that have helped to spread SRI in particular countries, regions or at local level. At international level, however, it is very clear that Norman Uphoff and CIIFAD have played a very significant role, which appears to have had a catalytic effect on many other organizations and individuals. The role of other charismatic individuals at international, national and regional levels seems also to have been important, and it may be that this pattern is repeated at local level too. Overall, it is clear that the spread of SRI activity has been shaped by interactions among various different kinds and sizes of actors and organizations, including small and large NGOs, universities and research institutes, religious organizations, extension agencies, organic farming networks and funding bodies. Communications and information exchanges among networks of actors have played a role, although the nature and importance of these mechanisms have not yet been studied in great detail. At local level, patterns of SRI adoption are probably shaped by local institutions of labour organization, communication networks and information flows, social capital, trust and other factors. The style, content and sustained commitment of training and extension support also appears to be very important, and may take different forms, leading to different outcomes. Taking into account the origins of SRI, through to the distinctive ways in which it has spread internationally, we conclude that SRI should be recognized and understood not merely as a set of crop management principles but as the fruit of a distinctive socio-technical system of knowledge and innovation that has operated, at least partly, outside the mainstream circuits of the international agricultural research system. Evidently, it is perceived by an eclectic mixture of NGOs, scientists, extension agencies and other actors to fill a gap in existing systems for generating and extending improved farming techniques to small and marginal farmers. These people and organizations view SRI as a potential solution to various pressing challenges in agriculture. In particular, it is seen by some as an accessible, low external-input alternative to conventional types of modern agricultural intensification, which exploits agro-ecological processes to produce sufficient and perhaps increased quantities of rice while also contributing to farm livelihoods and conserving natural resources. It is important not to indulge in simplistic optimism about the accessibility and appropriateness of low external- input farming methods (Giller et al. 2009; Tripp 2006). Nevertheless, if the potential benefits of SRI can be realized, and if it really is or can be made more accessible to poor and marginal farmers, it could be of direct benefit to them and indirectly beneficial to a much wider group of stakeholders. We conclude that SRI poses important and intriguing questions, not only about the rice and the techniques of rice cultivation but also about agricultural research and development, communication, the mobilization of social and professional networks, the exploitation of

136 scientific knowledge, learning processes, and other topics. Answers to these questions would be useful and relevant for international agricultural development policy and practice and could help to improve the mobilization of science and technology to help poor and marginal farmers in the developing world. In the next section, we consider these questions and propose some general hypotheses that could form a framework for further analysis of SRI. 9.2 Proposals for further investigation

Our exploration of the SRI phenomenon indicates the degree to which the optimization of rice cultivation practices requires numerous adjustments to suit not only agro-ecological but also socio-economic and institutional settings. Rice cultivation is highly diverse and growing conditions are often region-, village-, household- and even plot-specific. It is inappropriate to consider such a system from a technical perspective alone. The multi-dimensional nature of the SRI phenomenon calls for an integrated, multi-disciplinary research approach, encompassing quantitative and qualitative analyses, biophysical studies, agronomic research and socio-economic perspectives. Such research should also evaluate SRI at different levels of analysis, i.e. field, household, regional and, in some dimensions, national and international levels. In this section, we offer a non-exhaustive, non-exclusive list of suggestions about the important questions we see arising from the SRI phenomenon and how to investigate them. We hope that these ideas can stimulate discussion and inspire others to come forward with critical comments and their own ideas. 9.2.1 Biophysical mechanisms

We leave it to professional agronomists to make definitive judgements about the biophysical mechanisms at work in SRI. Based on our understanding of the issues, however, we suggest that a number of key questions would repay further research. We think that further experiments and farmer field trials are needed in order to investigate the separate and collective effects of SRI practices. This is worth doing because there are intriguing signs that at least some of the practices are responsible for changes in the physiology, morphology and growth rate and yield attributes of rice that could be harnessed to achieve higher levels of productivity. Improved grain yield may also be achieved compared to existing farmer practices. In particular, we believe that the mechanisms and interactions that are stimulated by cultivating rice in aerobic soil conditions need substantial further analysis. We think that this will require more work on the interactions between roots, micro-organisms and soil nutrients in the root zone, to determine what positive or negative effects these interactions can produce under different conditions. The aim of these studies should, we suggest, be aimed at understanding the dynamics of these relationships and helping farmers to influence them positively in different soils, rather than trying to determine an idealized optimum at which farmers should aim. Among the distinctive crop establishment practices of SRI cultivation, we suggest that the vigour of young seedlings is now firmly established, and that this behaviour can form a

137 platform for a good yield under favourable conditions and with good crop management. We suggest, therefore, that applied research should take (or continue to take) two directions: (1) further investigation into the potential of direct seeding as a way to avoid transplanting shock, its relative benefits, disadvantages and trade-offs compared to transplanting, and the possibility of developing methods and tools for direct seeding that are accessible for poor and marginal farmers; (2) studies designed to elucidate the interactions among the number of seedlings per hill, spacing distances between hills and seedling age, in relation to different soils and, especially, water management regimes. In this aspect, it seems to us necessary to properly specify, and then investigate, the possibility that there are synergetic interactions among the SRI practices. The aim of such studies should be to develop heuristics that could help farmers judge which combination is most likely to be appropriate for their conditions and capabilities, rather than identifying an ideal combination. The dissemination of such types of heuristics would require suitable styles and methods of extension. 9.2.2 Historical and contemporary analogues for SRI and/or SRI components

An important step towards understanding and explaining the emergence of SRI in the relative scientific isolation of Madagascar in the 1970s and 1980s is the existence of historical and contemporary parallels in other rice cultivation systems. Henri de Laulanié’s inductive scientific method seems to have enabled him to identify characteristics that had been recognized by others in the past and in other countries. The existence of SRI and these other systems is intriguing and it would be very useful to know when and where they emerged and why they are not more prevalent today. Have such systems always subsisted in the background, in particular niches or in isolated pockets? Have they emerged repeatedly and been selected against? It seems possible that the selection environment for such systems was affected by the advent of Green Revolution rice varieties and methods, and that the selection environment has improved in recent times, as farmers and environmental campaigners have begun to express concern about high costs, declining returns and adverse environmental impacts of excessive fertilizer and pesticide use. 9.2.3 The level of spread, patterns of adoption and impact of SRI methods

We have identified a vital need to generate more detailed and reliable information about the spread and levels of adoption of SRI methods. In particular, we think it is very important to study the patterns of adoption and partial adoption, including the range and degrees of variation in the ways SRI is specified and practised by farmers. We emphasize that the absolute levels of adoption are less interesting and significant than the patterns of adoption, because of the insight that such patterns can provide into the mechanisms driving the spread and shaping the practice of SRI methods. There is a need to further investigate the reasons why different farmers, or farmers belonging to different communities or regions, are more or less likely to adopt particular SRI methods, or to apply them on certain fields or in particular seasons. There is a wide range of factors that could shape these processes, including agro- ecological characteristics (e.g. soil type, soil fertility, irrigation source, climate), socio- economic factors (e.g. wealth, household size, income sources) and institutional aspects (e.g.

138 land ownership, gender divisions of labour, extension services, information sources, access to markets). Identifying causal relationships between these factors and the adoption, non-adoption or partial adoption of SRI methods could reveal when, why and how SRI spreads through communities or across landscapes. We suggest that collecting this kind of data should not be limited to household surveys and interviews but should also include landscape surveys and field measurements (e.g. observing the prevalence of grid planting, measuring spacing distances, counting numbers of seedlings per hill, observing weeding processes). Assessing levels of participation in farmer training activities and studying subsequent behaviour, or mapping social networks and information flows, would also shed light on the mechanisms involved in spreading SRI practice. Also useful would be a study of micro-level institutions within villages or communities, such as trust and conformity effects. The adoption, or partial adoption, of SRI methods by farmers will lead to changes in input productivity and potentially to increased output levels. As a consequence, different types of farmers are likely to re-allocate scarce production factors across off-farm and non-farm activities in different ways. In addition, labour demand changes in rice production may have a profound impact on the income and well-being of landless labourers. In all these cases SRI methods will have a measurable impact on household income, food security status and/or health of household members. The magnitude of this impact, and how it varies across different regions, types of households and labourers, has not yet been established categorically, but can be addressed through the implementation of well-designed randomized impact assessments. 9.2.4 Adoption processes and mechanisms leading to variation in SRI practice

SRI has been portrayed as a type of cultivation system which is intrinsically adaptable and flexible. Our review of the biophysical literature (Chapter 4) indicates the need to fine-tune a cultivation system to suit local conditions and resources. Diverse specifications and implementations of SRI are indeed evident in SRI documents, farmers’ practice, and variants of SRI that have been developed by researchers and project workers (see Chapter 8). But does this diversity in the specifications of SRI stem from the inherent technical characteristics of the system itself, or from the diversity intrinsic to smallholder farming? Any formalized system of agricultural practice, including SRI methods or Green Revolution packages, necessarily undergoes a process of transformation as it is carried from one setting to another. This is true when cultivation systems move from the laboratory or research station to farmers’ fields, or from one country, region or farm to another (Glover in press; Maat in press). Our analysis of SRI documents and discussions with SRI stakeholders in Madagascar, Tamil Nadu and Nepal suggests that the manner in which SRI is promoted may be just as formulaic or dogmatic as the promotion of any other system. We do not perceive any intrinsic properties of SRI techniques that would necessarily ensure that they are conveyed to farmers in ways that encourage them to experiment and adapt the system, though they may well do so anyway. Whether such flexibility is actively encouraged would seem to be primarily a matter of the style and method of training and support, not an intrinsic feature of the cultivation system itself.

139 Based on these insights, we see a need to investigate the processes and mechanisms leading to adaptation and variation in SRI practice. Suitable studies could include investigations into the alternative training methods that have been used to extend SRI, including the learning processes involved. It would also be useful to investigate the decision-making processes of farmers and, in particular, to discover which kinds of theorising, experimentation and analysis are involved. Particular areas of interest include: (1) the extent to which the farmer’s choice of spacing distance is shaped by the recommendations of a trainer, the width of available weeders, soil analysis, variety choice, trial and error, or other influences; (2) the extent to which seedling age (planting date) is determined primarily by the farmer’s preferences or by local labour availability; (3) the degree to which spacing distance is determined by the preferences of the farmer or the willingness, haste, skill or care of the transplanter. We suggest that, if progress can be made in the areas identified above, useful, practical progress can be made in developing rice cultivation systems that answer the felt needs of poor and marginal farmers. In the process, policy makers, aid agencies, research managers and extension workers can perhaps also learn how to improve the connections between agricultural science and the fields where scientific knowledge should be applied.

140

10. References

Adusumilli, R., Bhagya Laxmi, S., in press. Potential of the system of rice intensification for systemic improvement in rice production and water use: the case of Andhra Pradesh, India. Paddy Water Environment , forthcoming. Africare, Oxfam America & WWF-ICRISAT Project 2010, 'More Rice for People, More Water for the Planet', Hyderabad, AP, India: WWF-ICRISAT Project. Andrianaivo, B., 2002. Evaluations of the System of Rice Intensification in Fianarantsoa Province of Madagascar. Abstracts of papers, International Conference on assessments of the system of rice intensification (SRI) Sanya, China, April 1-4, 2002, edited by Uphoff, N., Fernandes, E., Longping, Y., Jiming, P., Rafaralahy, S., Rabenandrasana, J. Andriankaja, A. H. no date, Mise en evidence des opportunités de développement de la riziculture par adoption du SRI, et évaluation de la fixation biologique de l'azote [Study of the opportunities for the development of rice cultivation through the adoption of SRI, and evaluation of biological nitrogen fixation], University of Antananarivo. Anthofer, J., 2004. Potential of the System of Rice Intensification (SRI) for Cambodia. Consultancy Report for GTZ, Food Security and Nutrition Policy Support Project (FSNPSP), Rural Development Program Kampot & Kampong Thom (RDP), Phnom Penh, Cambodia / Spaichingen, Germany Antle, J.M., 2010. Asymmetry, partial moments and production risk. American Journal of Agricultural Economics 92(5): 1294-1309. Association Tefy Saina & Uphoff, N. 2001, 'More rice with less water', Appropriate Technology , 28 (3): 8-11. BAO, L., LIN, G., ZHAO, D., LI, Y. & HE, B. 2008, 'SRI Study on Culturing Density of Hybrid Rice Yixiang 1577 in the South of Sichuan Province', Southwest China Journal of Agricultural Sciences , 21 (3): 590-2. Barah, B.C., 2009. Economic and ecological benefits of System of Rice Intensification in Tamil Nadu. Agricultural economics research review 22: 209-214 Barison, J. 1997, Perspective de développement de la région de Ranomafana: Les mecanismes physiologiques du riz sur sols de basfonds, Cas du Système de Riziculture Intensive [Development perspective for the Ranomafana region: The physiological mechanisms of rice on irrigated soils, The case of the System of Rice Intensification]. University of Antananarivo.

141 Barison, J., Uphoff, N., in press. Rice yield and its relation to root growth and nutrient-use efficiency under SRI and conventional cultivation: an evaluation in Madagascar. Paddy Water Environment , forthcoming. Barrett, C.B., Moser, C.M., McHugh, O.V., Barison, J., 2004. Better technology, better plots, or better farmers? Identifying changes in productivity and risk among Malagasy farmers farming systems for resource-poor farmers. American journal of agricultural economics 86 (4): 869-888. Basu, S. 2009, "In the name of SRI": Retrospection on the Spread of SRI in Andhra Pradesh, India, Wageningen University. Belder, P., Bouman, B. A. M., Cabangon, R., Guoan, L., Quilang, E. J. P., Li Yuanhua, Spiertz, J. H. J. & Tuong, T. P. 2004, 'Effect of water-saving irrigation on rice yield and water use in typical lowland conditions in Asia', Agricultural Water Management , 65 (3): 193-210. Belder, P., Bouman, B. A. M. & Spiertz, J. H. J. 2007, 'Exploring options for water savings in lowland rice using a modelling approach', Agricultural Systems , 92 (1-3): 91-114. Belder, P., Spiertz, J. H. J., Bouman, B. A. M., Lu, G. & Tuong, T. P. 2005, 'Nitrogen economy and water productivity of lowland rice under water-saving irrigation', Field Crops Research , 93 (2-3): 169-85. Bijker, W. E. 1997, Of Bicycles, Bakelites and Bulbs: Toward a Theory of Sociotechnical Change , Boston, MA, USA: MIT Press. Bouman, B. A. M., Hengsdijk, H., Hardy, B., Bindraban, P. S., Tuong, T. P. & Ladha, J. K., eds. 2002, Water-wise Rice Production. Proceedings of the International Workshop on Water-wise Rice Production, 8-11 April 2002, Los Baños, the Philippines , Los Baños, the Philippines: International Rice Research Institute. Bouman, B. A. M., Humphreys, E., Tuong, T. P. & Barker, R. 2006, 'Rice and water', Advances in Agronomy , 92: 187-237. CAI, Y., TAO, W., ZHANG, Y. & ZHANG, X. 2008, 'Effect of Planting Density on the Production and N Utilization of Rice Under Large-triangle Intensification System', Chinese Journal of Eco-Agriculture , 16 (6): 1603-5. Cao, W. 2002, 'Physiological characterization of rice grown under different water management systems', in Water-wise Rice Production , eds. B.A.M. Bouman, H. Hengsdijk, B. Hardy, P.S. Bindraban, T.P. Tuong and J.K. Ladha, Los Baños, the Philippines: IRRI. Ceesay, M., Reid, W. S., Fernandes, E. C. M. & Uphoff, N. T. 2006, 'The effects of repeated soil wetting and drying on lowland rice yield with System of Rice Intensification (SRI) methods', International Journal of Agricultural Sustainability , 4 (1): 5-14. Ceesay, M. M., 2002. Experiments with the System of Rice Intensification in the Gambia. Abstracts of papers, International Conference on assessments of the system of rice

142 intensification (SRI) Sanya, China, April 1-4, 2002, edited by Uphoff, N., Fernandes, E., Longping, Y., Jiming, P., Rafaralahy, S., Rabenandrasana, J. Chapagain, T. & Yamaji, E. 2010, 'The effects of irrigation method, age of seedling and spacing on crop performance, productivity and water-wise rice production in Japan', Paddy and Water Environment , 8 (1): 81-90. Chauvet, C. & Guérin, F. 2010, ' Quand riziculture rime avec autosuffisance alimentaire [When rice cultivation rhymes with food self-sufficiency]', in BIOForum . CHEN, H., ZHU, D., RAO, L., LIN, X. & ZHANG, Y. 2006, 'Effects of SRI Technique on Population Quality after Heading Stage and Yield Formation in Rice', Journal of Huazhong Agricultural University , 25 (5): 483-7. Chen, Z., Yen, S.T., 2006. Better technology, better plots, or better farmers? Identifying changes in productivity and risk among malagasy farmers : a comment. American journal of agricultural economics 88 (3): 763 -766. Cohen, R. & Rai, S. M., eds. 2000, Global Social Movements , London, UK / New York, NY, USA: The Athlone Press. de Laulanié, H. 1992, 'Présentation technique du Système de Riziculture Intensive (SRI) basé sur le modèle de tallage de Katayama [Technical presentation of the System of Rice Intensification (SRI) based on Katayama's tillering model, typescript with manual corrections and annotations by the author]', in CIIFAD SRI collection , Ithaca, NY, USA. de Laulanié, H. 1993, 'Le système de riziculture intensive Malgache [The Malagasy system of intensive riziculture]', Tropicultura : 1-19. de Laulanié, H. 2003, Le Riz à Madagascar: Un développment en dialogue avec les paysans , Antananarivo, MG: Editions Ambozontany/Editions Karthala. Della Porta, D. & Diani, M. 1999, Social Movements: An Introduction , Oxford, UK / Maldon, MA, USA: Blackwell. Dhakal, M.P., 2005. Farmers’ evaluation of the System of Rice Intensification in the middle mountains of Nepal . Consultant’s report People and Resource Dynamics in Mountain Watersheds of the HKH (PARDYP) /International Centre for Integrated Mountain Development (ICIMOD), Kathmandu, Nepal. Di Falco and Chavas, 2009. On crop biodiversity, risk exposure, and food security in the highlands of Ethiopia, American Journal of Agricultural Economics 91(3): 599-611. Dobermann, A. 2004, 'A critical assessment of the system of rice intensification (SRI)', Agricultural Systems , 79 (3): 261–81. Duflo, E., Glennerster, R., Kremer, M.,2007. Using randomization in development economics research: a toolkit . Chapter 61 in Handbook of Development Economics , Vol 4, edited by Schultz, T.P. and Strauss, J.A. Elsevier, Amsterdam, The Netherlands.

143 Elyah, A. 2006, Sur la diffusion du SRI à Madagascar: Un goût d'inachevé? [On the diffusion of SRI in Madagascar: A taste of failure?], Institut Catholique de Madagascar. Fagi, A. M. & Kartaatmadja, S. 2002, ' Gogorancah rice in Indonesia: a traditional method in the modern era', in Direct seeding: research issues and opportunities. Proceedings of an international workshop on direct seeding in Asian rice systems: Strategic research issues and opportunities , eds. S. Pandey, M. Mortimer, L. Wade, T.P. Tuong, K. Lopez and B. Hardy, Los Baños, Philippines: IRRI. FAN, Y., LU, A., LI, Q., LU, H., YUAN, M. & YIN, L. 2007, 'Preliminary Study on Influence of the System of Rice Intensification on Main Economic Characteristics and Yield of IIIYou 98', Journal of Anhui Agricultural Sciences , 35 (21): 6398-401. Feder, G., Just, R.E., Zilberman, D.,1985 Adoption of agricultural innovations in developing countries: a survey. Economic Development and Cultural Change 33(2): 255-298. Feuer, H.N., 2008. Sustainable Agricultural Techniques and Performance Oriented Empowerment: An Actor-Network Theory Approach to CEDAC Agricultural and Empowerment Programmes in Cambodia. M.Phil thesis, University of Oxford, United Kingdom. Gani, A., Rahman, A., Dahono, R. & Hengsdijk, H. 2002, 'Synopsis of water management experiments in Indonesia', in Water-wise Rice Production , eds. B.A.M. Bouman, H. Hengsdijk, B. Hardy, P.S. Bindraban, T.P. Tuong and J.K. Ladha, Los Baños, the Philippines: IRRI. Gerichhausen, M., Berkhout, E.D., Hamers, H.J.M., Manyong, V.M., 2009. A quantitative framework to analyse cooperation between rural households. Agricultural Systems 101 (3): 173-185. Giller, K. E., Witter, E., Corbeels, M. & Tittonell, P. 2009, 'Conservation agriculture and smallholder farming in Africa: The heretics' view', Field Crops Research , 114 (1): 23- 34. Glover, D. forthcoming, 'A system designed for rice? Materiality and the invention/discovery of the System of Rice Intensification', Accepted for publication in the Journal of East Asian Science, Technology and Society (EASTS) . Glover, D. in press, 'The System of Rice Intensification: Time for an empirical turn', NJAS- Wageningen Journal of Life Sciences . Grist, D. H. 1986, Rice , 6th edn., Harlow, UK: Longman. Gujja, B. & Thiyagarajan, T. M. 2009, 'New Hope for Indian Food Security? The System of Rice Intensification', in IIED Gatekeeper Series , London, UK: International Institute for Environment and Development. GUO, L., LIANG, T., TANG, M., HE, L., LI, R., YANG, W., XIA, Y. & JIANG, L. 2007, 'Plant Growth and Nitrogen Utilization of the System of Rice Intensification Under

144 Different Fertilization Patterns and Tillage Ways', Chinese Agricultural Science Bulletin , 23 (1): 185-8. HE, Y. & MA, J. 2005, 'Effect of Different Kinds of Fertilizer on the Yield and Physical and Chemical Status of Soil in SRI', Southwest China Journal of Agricultural Sciences , 18 (3): 315-20. Holiarison, W., Barison, J., Lines, G., Naivohanitriniaina, L., Rabenandrasana, J., Rafaralahy, S., Rataminjanahary, E. & Uphoff, N. 1998, 'Evoking underutilized genetic potentials in rice by modifying the growing environment to change plant structure and increase yields': Ithaca, NY, USA: CIIFAD. Horie, T., Shiraiwa, T., Homma, K., Katsura, K., Maeda, S. & Yoshida, H. 2005, 'Can yields of lowland rice resume the increases that they showed in the 1980s?', Plant Production Science , 8 (3): 259-74. ICAR 1960, Rice in India , revised edn., Delhi, India: Indian Council of Agricultural Research. Jansen, K. & Vellema, S. in press, 'What is technography?', NJAS Wageningen Journal of Life Sciences . Kabir, H. & Uphoff, N. 2007, 'Results of disseminating the System of Rice Intensification with Farmer Field School methods in Northern Myanmar', Experimental Agriculture , 43 (4): 463-76. Kikuchi, K. & Xie, M. 2008, 'Sustainable System of Irrigated Agriculture (SSIA): An Application of System of Rice Intensification (SRI) in the Philippines. How does it improve productivity?', in CIIFAD SRI collection , Ithaca, NY, USA: Cornell International Institute for Agricultural Development. Krupnik, T. J. 2008, 'Current Research and Evaluation Efforts of the System of Rice Intensification in the Senegal River Valley', Santa Cruz, CA, USA: Department of Environmental Studies (Agroecology), University of California, Santa Cruz Krupnik, T. J., Rodenburg, J., Shennan, C., Mbaye, D. & Haden, V. R. 2010, 'Risky trade- offs: Comparing yield, water productivity, and weed competition under the System of Rice Intensification and Best Management Practices in the Sahel', in Africa Rice Congress , Bamako, Mali. Kumar Sinha, S. & Talati, J. 2007, 'Productivity impacts of the system of rice intensification (SRI): A case study in West Bengal, India', Agricultural Water Management , 87 (1): 55-60. Latif, M. A., Ali, M. Y., Islam, M. R., Badshah, M. A. & Hasan, M. S. 2009, 'Evaluation of management principles and performance of the System of Rice Intensification (SRI) in Bangladesh', Field Crops Research , 114 (2): 255-62. Latif, M. A., Islam, M. R., Ali, M. Y. & Saleque, M. A. 2005, 'Validation of the system of rice intensification (SRI) in Bangladesh', Field Crops Research , 93 (2-3): 281-92. Latour, B. 1987, Science in Action , Cambridge, Mass.: Harvard University Press.

145 LIN, X., ZHU, D. & ZHANG, Y. 2004, 'Origins of System of Rice Intensification (SRI) and Its Cultivation System', China Rural Science & Technology , (1): 35-6. LONG, X., MA, J., XU, F., WANG, H., HUANG, Q. & YUAN, Z. 2005, 'Study on the Seedling-age and Planting Density in SRI', Journal of Sichuan Agricultural University , 23 (3): 368-73. LU, X., HUANG, Q. & LIU, H. 2006, 'Research of some Physiological Characteristics Under the System of Rice Intensification', Journal of South China Agricultural University , 27 (2): 5-7. LU, Y., LI, J., WANG, J., TANG, Y. & YU, G. 2005, 'Effects of SRI on Dry Matter Production and Grain Yield of Yuyou 11', Southwest China Journal of Agricultural Sciences , 18 (6): 79-83. Maat, H. in press, 'The history and future of agricultural experiments', NJAS Wageningen Journal of Life Sciences . Makarim, A. K., Balasubramanian, V., Zaini, Z., Syamsiah, I., Diratmadja, I. G. P. A., Handoko, Arafah, Wardana, I. P. & Gani, A. 2002, 'System of Rice Intensification (SRI): evaluation of seedling age and selected components in Indonesia', in Water- wise Rice Production , eds. B.A.M. Bouman, H. Hengsdijk, B. Hardy, P.S. Bindraban, T.P. Tuong and J.K. Ladha, Los Baños, the Philippines: IRRI. Manjunatha, B. N., Basavarajappa, R. & Pujari, B. T. 2010, 'Effect of age of seedlings on growth, yield and water requirement by different system of rice intensification', Karnataka Journal of Agricultural Sciences , 23 (2): 231-4. MAO, G. 2003, 'System of Rice Intensification (SRI) and Its Prospect in Zhejiang', System of Rice Intensification (SRI) and Its Prospect in Zhejiang , (4): 10-1. McDonald, A. J., Hobbs, P. R. & Riha, S. J. 2006, 'Does the system of rice intensification outperform conventional best management? A synopsis of the empirical record', Field Crops Research , 96 (1): 31-6. McDonald, A. J., Hobbs, P. R. & Riha, S. J. 2008, 'Stubborn facts: Still no evidence that the System of Rice Intensification out-yields best management practices (BMPs) beyond Madagascar', Field Crops Research , 108 (2): 188-91. McHugh, O., Steenhuis, T. S., Barison, J., Fernandes, E. & Uphoff, N. 2002, 'Farmer implementation of alternate wet-dry and nonflooded irrigation practices in the System of Rice Intensification (SRI)', in Water-wise Rice Production , eds. B.A.M. Bouman, H. Hengsdijk, B. Hardy, P.S. Bindraban, T.P. Tuong and J.K. Ladha, Los Baños, the Philippines: IRRI. Menete, M. Z. L., van Es, H. M., Brito, R. M. L., DeGloria, S. D. & Famba, S. 2008, 'Evaluation of system of rice intensification (SRI) component practices and their synergies on salt-affected soils', Field Crops Research , 109 (1-3): 34-44. Minten, B., Barrett, C.B., 2008.Agricultural technology, productivity and poverty in Madagascar. World Development 36(5): 797-822.

146 Minten, B., Randrianarisoa, J., Barrett, C.B., 2007. Productivity in Malagasy rice systems: wealth-differentiated constraints and priorities. Agricultural Economics 37 (S1): 225- 237. Mishra, A. & Salokhe, V. M. 2008, 'Seedling characteristics and the early growth of transplanted rice under different water regimes', Experimental Agriculture , 44 (3): 365-83. Mishra, A. & Salokhe, V. M. 2010, 'Flooding stress: The effects of planting pattern and water regime on root morphology, physiology and grain yield of rice', Journal of Agronomy and Crop Science , 196 (5): 368-78. Mishra, A., Whitten, M., Ketelaar, J. W. & Salokhe, V. M. 2006, 'The System of Rice Intensification (SRI): a challenge for science, and an opportunity for farmer empowerment towards sustainable agriculture', International Journal of Agricultural Sustainability , 4 (3): 193-212. Miyazato, T., Mohammed, R. A. & Lazaro, R. C. 2010, 'Irrigation management transfer (IMT) and system of rice intensification (SRI) practice in the Philippines', Paddy and Water Environment , 8 (1): 91-7. Moser, C. M. & Barrett, C. B. 2003, 'The disappointing adoption dynamics of a yield- increasing, low external-input technology: case of SRI in Madagascar', Agricultural Systems , 76 (3): 1085-100. Moser, C. M. & Barrett, C. B. 2006, 'The Complex Dynamics of Smallholder Technology Adoption: The Case of SRI in Madagascar', Agricultural Economics , 35 (3): 373-88. Mushtaq, S., Khan, S., Hafeez, M. & Hanjra, M. A. 2009, 'Does reliability of water resources matter in the adoption of water-saving irrigation practices? A case study in the Zhanghe irrigation system, China', Water Policy , 11 (6): 661-79. Namara, R.E., Weligamage, P., Barker, R., 2004. Prospects for Adopting System of Rice Intensification in Sri Lanka: A Socioeconomic Assessment . Research Report 75, International Water Management Institute (IWMI), Colombo, Sri Lanka Palanisami, K. unpublished, 'Adoption of SRI under different irrigation sources and farm size categories in Tamil Nadu, India': International Water Management Institute (IWMI). Pandey, S. & Velasco, L. 2002, 'Economics of direct seeding in Asia: patterns of adoption and research priorities', in Direct seeding: research issues and opportunities. Proceedings of an international workshop on direct seeding in Asian rice systems: Strategic research issues and opportunities , eds. S. Pandey, M. Mortimer, L. Wade, T.P. Tuong, K. Lopez and B. Hardy, Los Baños, Philippines: IRRI. Pasuquin, E., Lafarge, T. & Tubana, B. 2008, 'Transplanting young seedlings in irrigated rice fields: Early and high tiller production enhanced grain yield', Field Crops Research , 105 (2): 141-55. Rabenandrasana, J. 1999, 'Revolution in rice intensification in Madagascar', ILEIA Newsletter , 15 (3&4): 48-9.

147 Rabenandrasana, J. 2002, 'SRI Experience of Association Tefy Saina in Madagascar', in Assessments of the System of Rice Intensification (SRI). Proceedings of an International Conference held in Sanya, China, April 1-4, 2002 , eds. N. Uphoff, E. Fernandes, Y. Longping, P. Jiming, S. Rafaralahy and J. Rabenandrasana, Ithaca, NY, USA: Cornell International Institute for Food, Agriculture and Development. Rafaralahy, S. 2002, 'An NGO Perspective on SRI and its Origins in Madagascar', in Assessments of the System of Rice Intensification (SRI). Proceedings of an International Conference held in Sanya, China, April 1-4, 2002 , eds. N. Uphoff, E. Fernandes, Y. Longping, P. Jiming, S. Rafaralahy and J. Rabenandrasana, Ithaca, NY, USA: Cornell International Institute for Food, Agriculture and Development. Rajaonarison, J. d. D. 1999, Contribution à l'amélioration des rendements de 2ème saison de la double riziculture par SRI sous exprimentations multifactorielles (Cas des sols sableux de Morondava) [The contribution of SRI to the improvement of yield in the second season of double rice cultivation under multifactorial experiments (Case of the sandy soils of Morondava] University of Antananarivo. Rajukkannu, K., Ramadass, P. & Jecitha, J. 2007, 'Keeping the brown plant hopper at bay with SRI', LEISA Magazine , 23 (4): 30-1. Ramasamy, S., ten Berge, H. F. M. & Purushothaman, S. 1997, 'Yield formation in rice in response to drainage and nitrogen application', Field Crops Research , 51: 65-82. Randriamiharisoa, R. 2002, 'Research Results on Biological Nitrogen Fixation with the System of Rice Intensification', in Assessments of the System of Rice Intensification (SRI). Proceedings of an International Conference held in Sanya, China, April 1-4, 2002 , eds. N. Uphoff, E. Fernandes, Y. Longping, P. Jiming, S. Rafaralahy and J. Rabenandrasana, Ithaca, NY, USA: Cornell International Institute for Food, Agriculture and Development. Randriamiharisoa, R. & Uphoff, N. 2002, 'Factorial Trials Evaluating the Separate and Combined Effects of SRI Practices', in Assessments of the System of Rice Intensification (SRI). Proceedings of an International Conference held in Sanya, China, April 1-4, 2002 , eds. N. Uphoff, E. Fernandes, Y. Longping, P. Jiming, S. Rafaralahy and J. Rabenandrasana, Ithaca, NY, USA: Cornell International Institute for Food, Agriculture and Development. Rasoanaivo, S., Randrianarisoa, J. D. & Beauval, V. 2006, ' Evaluation du projet d'irrigation et développement rural dans l'arrière pays de Manakara - Madagascar. Rapport Final [Evalution of the irrigation and rural development project in the back country of Manakara - Madagascar. Final Report]', Paris / Versailles, FR: Fonds pour la promotion des études préalables, études transversales, évaluations (F3E) and Inter Aide. Razafimanantsoa, R. 2009, Analyse de l'échec et de la diffusion du Système de Riziculture Intensive à Madagascar [Analysis of the failure and the diffusion of the System of Rice Intensification in Madagascar], University of Antananarivo.

148 Reddy, V. R., Reddy, P. P., Reddy, M. S. & Raju, D. S. R. 2005, 'Water use efficiency: a study of system of rice intensification (SRI) adoption in Andhra Pradesh', Indian Journal of Agricultural Economics , 60 (3): 458-72. Resurreccion, B.P., Sajor, E.E., 2008. Gender Dimensions of the Adoption of the System of Rice Intensification (SRI) in Cambodia . Research Report, Asian Institute of Technology and Royal University of Phnom Penh, Phnom Penh, Cambodia. San-oh, Y., Mano, Y., Ookawa, T. & Hirasawa, T. 2004, 'Comparison of dry matter production and associated characteristics between direct-sown and transplanted rice plants in a submerged paddy field and relationships to planting patterns', Field Crops Research , 87 (1): 43-58. Sanjeewanie Ginigaddara, G. A. & Ranamukhaarachchi, S. L. 2009, 'Effect of conventional, SRI and modified water management on growth, yield and water productivity of direct-seeded and transplanted rice in central Thailand', Australian Journal of Crop Science , 3 (5): 278-86. Sato, S. & Uphoff, N. 2007, 'A review of on-farm evaluations of system of rice intensification methods in Eastern Indonesia', in CAB Reviews: Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources . Satyanarayana, A. 2004, 'Rice, research and real life in the field', Nature , 429 (6994): 804. Satyanarayana, A., Thiyagarajan, T. M. & Uphoff, N. 2007, 'Opportunities for water saving with higher yield from the system of rice intensification', Irrigation Science , 25 (2): 99-115. Schiller, J.M., 2004. System of rice intensification -SRI- suitability for lowland rice production in the Lao PDR . Consultant’s report, Food and Agricultural Organisation (FAO), Lao DPR. Schlanger, N. 2006, 'Introduction. Technological commitments: Marcel Mauss and the study of techniques in the French social sciences', in Marcel Mauss. Techniques, Technology and Civilisation , ed. N. Schlanger, NewYork, NY, USA: Durkheim Press / Berghahn Books. Senthilkumar, K., Bindraban, P. S., Thiyagarajan, T. M., De Ridder, N. & Giller, K. E. 2008, 'Modified rice cultivation in Tamil Nadu, India: Yield gains and farmers’ (lack of) acceptance', Agricultural Systems , 98 (2): 82-94. Shambu Prasad, C. 2006, 'System of Rice Intensification in India: Innovation History and Institutional Challenges', Patencheru, Hyderabad, India: WWF International - ICRISAT Dialogue Project. Shambu Prasad, C. 2009, 'Conversations on knowledge and democracy: fables from SRI ', Seminar , (597): 2-6. Shambu Prasad, C., Beumer, K. & Mohanty, D. 2007, 'Towards a Learning Alliance SRI in Orissa', Hyderabad, Andhra Pradesh / Bhubaneswar, Orissa, India: WWF Project on Water, Food and Environment / Xavier Institute of Management.

149 Sheehy, J. E., Peng, S., Dobermann, A., Mitchell, P. L., Ferrer, A., Jianchang Yang, Yingbin Zou, Xuhua Zhong & Huang, J. 2004, 'Fantastic yields in the system of rice intensification: fact or fallacy?', Field Crops Research , 88 (1): 1-8. Sheehy, J. E., Sinclair, T. R. & Cassman, K. G. 2005, 'Curiosities, nonsense, non-science and SRI', Field Crops Research , 91 (2-3): 355–6. Sinclair, T. R. 2004, 'Agronomic UFOs waste valuable scientific resources', Rice Today , 3 (3): 43. Sinclair, T. R. & Cassman, K. G. 2004, 'Agronomic UFOs', Field Crops Research , 88 (1): 9- 10. Sita Devi, K.,Ponnarasi, T., 2009. An Economic Analysis of Modern Rice Production Technology and its Adoption Behaviour in Tamil Nadu. Agricultural Economics Research Review 22: 341-347 Sital, J. T. 2002, How Suriname came to its own green revolution: A study of change in smallholder farming in the Nickerie District, 1900-1986 , Wageningen, NL: Wageningen University. Stoop, W. 2003, 'The system of rice intensification (SRI) from Madagascar: Myth or Missed Opportunity? Report on a study visit to the “Hauts Plateaux” region of Madagascar (3–15 March 2003)', Driebergen, Utrecht, NL: Stoop Consult. Stoop, W. 2006, 'SRI in context: lessons from the field', LEISA Magazine , 22 (4): 30-1. Stoop, W. 2010, 'Van bodemkundige, naar agronoom, naar systeemonderzoeker [From soil scientist, to agronomist, to systems researcher]', in Profiel van de Nederlandse Bodemkunde: 75 jaar Nederlandse Bodemkundige Vereniging (1935-2010) [Profile of Dutch Soil Science: 75 years of the Dutch Soil Sciences Association] , eds. J. Bouma, A.E. Hartemink, H.W.F. Jellema, J.J.M. van Grinsven and E.C. Verbauwen, Amsterdam, NL: Nederlandse Bodemkundige Vereniging. Stoop, W., Adam, A. & Kassam, A. 2009, 'Comparing rice production systems: A challenge for agronomic research and for the dissemination of knowledge-intensive farming practices', Agricultural Water Management , 96 (11): 1491–501. Stoop, W. & Kassam, A. 2005, 'The SRI controversy: a response', Field Crops Research , 91 (2-3): 357–60. Stoop, W., Uphoff, N. & Kassam, A. 2002, 'A review of agricultural research issues raised by the system of rice intensification (SRI) from Madagascar: opportunities for improving farming systems for resource-poor farmers', Agricultural Systems , 71 (3): 249-74. Styger, E. in press, 'Application of system of rice intensification practices in the arid environment of the Timbuktu region in Mali', Paddy and Water Environment : xx-xx. SUI, Y.-y. & Wang, Y. 2008, 'Super-high Yield Technique System of Rice Intensification', Journal of Hebei Agricultural Sciences , 12 (11): 9-10. Surridge, C. 2004, 'Feast or famine?', Nature , 428 (6981): 360-1.

150 Tajuddin, A. & Rajendran, P. 2002, 'Devices for inter-cropping green manure in wet seeded rice', Agricultural Mechanisation in Asia, Africa and Latin America , 33 (3): 9-10. TANG, S. 2002, 'Madagascar’s system of rice intensification', World Agriculture , (9): 44. Thakur, A. K. 2010, 'Critiquing SRI criticism: beyond scepticism with empiricism', Current Science , 98 (10): 1294-9. Thakur, A. K., Chaudhari, S. K., Singh, R. & Kumar, A. 2009, 'Performance of rice varieties at different spacing grown by the system of rice intensification in eastern India', Indian Journal of Agricultural Sciences , 79 (6): 443-7. Thakur, A. K., Rath, S., Patil, D. U. & Kumar, A. in press, 'Effects on rice plant morphology and physiology of water and associated management practices of the system of rice intensification and their implications for crop performance', Paddy and Water Environment . Thakur, A. K., Rath, S., Roychowdhury, S. & Uphoff, N. 2010a, 'Comparative Performance of Rice with System of Rice Intensification (SRI) and Conventional Management using Different Plant Spacings', Journal of Agronomy and Crop Science , 196 (2): 146-59. Thakur, A. K., Uphoff, N. & Antony, E. 2010b, 'An assessment of physiological effects of System of Rice Intensification (SRI) practices compared with recommended rice cultivation practices in India', Experimental Agriculture , 46 (1): 77-98. The SRI Consortium 2003, 'Report to the Rockefeller Foundation', Ithaca, NY, USA: Cornell International Institute for Agricultural Development. Thiyagarajan, T. M. 2006, 'Adapting SRI in Tamil Nadu, India', LEISA Magazine , 22 (4): 28- 9. Thiyagarajan, T. M. & Gujja, B. 2009, 'Single-seedling planting and the Gaja planting system', in SRI Newsletter , Patancheru, Hyderabad, India: ICRISAT-WWF SRI Project. Thiyagarajan, T. M., Ranganathan, C. R., Bhaskaran, A., Mathan, K. K., Karivaradaraaju, T. V., 2000. Trends in rice area, production and productivity in the different agro- climactic zones of Tamil Nadu. Madras Agricultural Journal 87(4-6): 287-290. Thiyagarajan, T. M., Velu, V., Ramasamy, S., Durgadevi, D., Govindarajan, K., Priyadarshini, R., Sudhalakshmi, C., Senthilkumar, K., Nisha, P. T., Gayathry, G., Hengsdijk, H. & Bindraban, P. S. 2002, 'Effects of SRI practices on hybrid rice performance in Tamil Nadu, India', in Water-wise Rice Production , eds. B.A.M. Bouman, H. Hengsdijk, B. Hardy, P.S. Bindraban, T.P. Tuong and J.K. Ladha, Los Baños, the Philippines: IRRI. Thomas, V. & Mohammad Ramzi, A. in press, 'SRI contributions to rice production dealing with water management constraints in northeastern Afghanistan', Paddy and Water Environment .

151 Tripp, R. 2006, 'Is low external input technology contributing to sustainable agricultural development?', in ODI Natural Resource Perspectives , London: Overseas Development Institute. Tsujimoto, Y., Horie, T., Randriamihary, H., Shiraiwa, T. & Homma, K. 2009, 'Soil management: The key factors for higher productivity in the fields utilizing the system of rice intensification (SRI) in the central highland of Madagascar', Agricultural Systems , 100 (1-3): 61–71. Turmel, M.-S., Espinosa, J., Franco, L., Pérez, C., Hernández, H., González, E., Fernández, G., Rojas, C., Sánchez, D., Fernández, N., Barrios, M., Whalen, J. K. & Turner, B. L. in press, 'On-farm evaluation of a low-input rice production system in Panama', Paddy and Water Environment : xx-xx. Uphoff, N. 1999, 'Agroecological implications of the System of Rice Intensification (SRI) in Madagascar', Environment, Development and Sustainability , 1 (3-4): 297-313. Uphoff, N. 2001, 'Report from field visit to farmers practicing SRI around Lac Alaotra, Madagascar - June 6, 2001', Ithaca, NY, USA: CIIFAD. Uphoff, N. 2002, 'Opportunites for Raising Yields by Changing Management Practices: The System of Rice Intensification in Madagascar', in Agroecological innovations: , ed. N. Uphoff, London, UK: Earthscan. Uphoff, N. 2003, 'Higher yields with fewer external inputs? The system of rice intensification and potential contributions to agricultural sustainability', International Journal of Agricultural Sustainability , 1 (1): 38-50. Uphoff, N. 2004a, 'The System of Rice Intensification (SRI): Capitalizing on existing yield potentials by changing management practices to increase rice productivity with fewer inputs and more profitability', in World Rice Research Conference , Tokyo-Tsukuba, Japan. Uphoff, N. 2004b, 'System of rice intensification responds to 21st century needs', Rice Today , 3 (3): 42. Uphoff, N. 2006, 'The System of Rice Intensification and its implications for agriculture ', LEISA Magazine , 22 (4): 6-8. Uphoff, N. 2007, 'Agroecological Alternatives: Capitalising on Existing Genetic Potentials', Journal of Development Studies , 43 (1): 218-36. Uphoff, N. 2008, 'Madagascar: Report on a visit to review progress with the System of Rice Intensification (SRI), September 25-October 3, 2008', Ithaca, NY, USA: CIIFAD. Uphoff, N. 2010a, 'A decade's perspective on the use and spread of the System of Rice Intensification (SRI)', in National Workshop on SRI: A Decade in India - Lessons, Experiences and Future Directions in the Context of Food Security and Climate Change , ICRISAT, Patancheru, Hyderabad, Andhra Pradesh, India. Uphoff, N. 2010b, 'Review of Presentations and Initiatives on SRI: 1998-Present': Cornell International Institute for Food, Agriculture and Development.

152 Uphoff, N., Anas, I., Rupela, O. P., Thakur, A. K. & Thiyagarajan, T. M. 2009, 'Learning about positive plant-microbial interactions from the System of Rice Intensification (SRI)', Aspects of Applied Biology , 98: 29-53. Uphoff, N. & Fernandes, E. 2002, 'System of rice intensification gains momentum', LEISA Magazine , 18 (3): 24-9. Uphoff, N., Fernandez, E., Yuan, L.-P., Jiming, P., Sebastien, R. & Rabenandrasana, J., eds. 2002, Assessment of the system for rice intensification (SRI). Proceedings of an International Conference, Sanya, China, April 1-4, 2002 , Ithaca, NY, USA: Cornell International Institute for Food, Agriculture and Development (CIIFAD). Uphoff, N., Kassam, A. & Harwood, R. in press, 'SRI as a methodology for raising crop and water productivity: productive adaptations in rice agronomy and irrigation water management', Paddy and Water Environment : xx-xx. Uphoff, N., Kassam, A. & Stoop, W. 2008, 'A critical assessment of a desk study comparing crop production systems: The example of the 'system of rice intensification' versus 'best management practice'', Field Crops Research , 108 (1): 109-14. Uphoff, N. & Randriamiharisoa, R. 2002, 'Reducing water use in irrigated rice production and the Madagascar System of Rice Intensification', in Water-wise Rice Production , eds. B.A.M. Bouman, H. Hengsdijk, B. Hardy, P.S. Bindraban, T.P. Tuong and J.K. Ladha, Los Baños, the Philippines: IRRI. Uprety, R. 2006, 'SRI takes root in Nepal', LEISA Magazine , 22 (4): 25-7. Vallois, P. 1997, 'Malagasy Early Rice Planting System for a Ton Increase per Acre in Developing Countries', Tananarive (Antananarivo), MG and Brussels, BE: Institut de la Promotion de la Nouvelle Riziculture, Frères de Saint Gabriel & Entraide et Fraternité. Vallois, P. 2005, ' MAFF (Mitsitsy Ambioka sy Fomba Fiasa): Intensification de la riziculture répiquée par l'amélioration des façons culturales, PC 15 - Vallée Marianina, Rapport de Campagne 2004-2005 [Intensification of transplanted rice cultivation through the improvement of cultural practices, PC 15 - Marianina Valley, Campaign Report 2004- 2005] ', Ambatondrazaka, MG: CIRAD, Madagascar; Ministry of Agriculture, Livestock and Fisheries (MALEP), Madagascar; Agence France de Développement, Madagascar. Vijayakumar, M., Ramesh, S., Chandrasekaran, B. & Thiyagarajan, G. 2006, 'Effect of System of Rice Intensification (SRI) Practices on Yield Attributes, Yield and Water Productivity of Rice ( Oryza Sativa l. )', Research Journal of Agriculture and Biological Sciences , 2 (6): 236-42. WANG, R., MA, J., TONG, P., ZHANG, R., LI, Y., FU, T., WU, H. & LIU, Z. 2006, 'Effects of Planting Method of Triangle of System of Rice Intensification （TSRI ） on Photosynthetic Characteristics and Formation of Grain Yield', Hybrid Rice , 21 (6): 60-5.

153 Wang, Y., Min, Z., Hou, S., Wang, Y., Yu, G., Gao, R. & Zhang, X. 2004, 'Discussions on System of Rice Intensification Techniques', Reclaiming and Rice Cultivation , (5): 16- 7. WU, C., LIN, H., LAI, L. & LIN, H. 2005, 'Discussion about Fitting Density on Rice Intensification', Reclaiming and Rice Cultivation , (1): 29-31. Xu, F., Ma, J., Wang, H., Liu, H., Huang, Q., Ma, W. & Ming, D. 2005, 'Rice quality under the cultivation of SRI', Acta Agronomica Sinica , 31 (5): 577-82. Yadao, A. C. & Zamora, O. B. 2007, 'Responses of rice under the system of rice intensification and conventional production systems in Ilocos Norte, Philippines', Philippine Journal of Crop Science , 32 (2): 99-107. Yamah, A., 2002. The practice of the System of Rice Intensification in Sierra Leone . Abstracts of papers, International Conference on assessments of the system of rice intensification (SRI) Sanya, China, April 1-4, 2002, edited by Uphoff, N., Fernandes, E., Longping, Y., Jiming, P., Rafaralahy, S., Rabenandrasana, J. YUAN, L. 2001, 'The System of Rice Intensification', Hybrid Rice , 16 (4): 1-3. Yokoyama, S., Zakari, A.K. (no date). Technological characteristics of Organic-SRI (System of Rice Intensification): Case in West Java, Indonesia . Draft scientific paper. ZHANG, Y., ZHU, D., LIN, X., CHEN, H. & YANG, Y. 2007, 'Effects of Dry and Wet Rice Under System Irrigation on the Growth of Rice Intensification', Agricultural Research in the Arid Areas , 25 (5): 109-13. ZHAO, L.-m., WU, L.-h., LI, Y.-s. & WU, M.-y. 2009a, 'Effects of the System of Rice Intensification on Soil Biological Properties', Acta Pedologica Sinica , 46 (2): 321-5. Zhao, L., Wu, L., Li, Y., Lu, X., Zhu, D. & Uphoff, N. 2009b, 'Influence of the system of rice intensification on rice yield and nitrogen and water use efficiency with different N application rates', Experimental Agriculture , 45 (3): 275-86. Zheng, J., Lu, X., Jiang, X. & Li, C. 2004, 'Testing modifications for the system of rice intensification (SRI) to achieve super-high-yield in the Sichuan basin', Southwest China Journal of Agricultural Sciences , 17 (2): 169-73. ZHENG, X., SUN, Y. & YANG, W. 2006, 'Effect of different application methods of nitrogen fertilizer on the yield and quality of late rice by intensified cultivation with no-tillage', Guangxi Agricultural Sciences , 37 (3): 298-302. Zhou, L., Zeng, X., Liu, Y., Yang, X., Zhang, X., Xie, J. & Zhang, Y. 2005, 'Demonstrative performance of sparse direct seeding SRI (system of rice intensification) of Liangyou Peijiu', Hybrid Rice , 20 (4): 46-8.

154

Appendices and supplementary material

155

Appendix A: Madagascar trip report

156 The Emergence of the System of Rice Intensification as a Social Movement

Project funded by the Bill & Melinda Gates Foundation

Trip report

Field visit to Madagascar, April–May 2010

Contents

Background ...... 158

Key observations ...... 159

Some knowledge gaps and areas for further research...... 166

Appendices: ...... 168

A1. Itinerary Madagascar field visits ...... 168

A2. Organizations visited in Madagascar ...... 169

A3. Visits to villages and meetings with farmers in Madagascar: ...... 170

A4. List of workshop participants ...... 171

A5. Workshop report ...... 174

Report prepared by:

Dr. Ezra Berkhout, Postdoc, Development Economics Group Dr. Dominic Glover, Postdoc, Technology and Agrarian Development Group Prof. dr. ir. Herman van Keulen, Professor of Plant Production Systems/Senior Research Scientist, Agrosystems Research Business Unit, Plant Research International

157 Background This document provides a brief report of a field visit to Madagascar carried out by researchers from Wageningen University, NL between 14 April and 11 May 2010. The visit took place in the context of an exploratory research project entitled “The emergence of the System of Rice Intensification (SRI) as a social movement”, funded by the Bill and Melinda Gates Foundation (BMGF) (project no. OPP1002894).

Aims This field visit complements the other activities undertaken in this project. The major part ofthe project consists of desk-based research, inventorizing the available official and grey literature on SRI as a basis for documenting the current state of knowledge on the emergence and spread of SRI and the networks of people and organisations involved in promoting and spreading SRI nationally and internationally. Two field visits supplement and complement the desk-based research, with the aim of ensuring that the insights derived from literature reviews and studies of the SRI network are grounded with reference to the experiences and insights of practitioners and researchers working at ground level. This research mission to Madagascar is the first of the two field visits. The second will take place in Tamil Nadu, India, later in 2010.

The field visit had two components: • A programme of field visits to observe rice cultivation and carry out meetings and discussions with farmers and other stakeholders. • A 2-day stakeholder workshop to discuss the social dynamics of SRI in Madagascar. A list of locations visited and the people and organisations met is provided in the appendices.

Specific objectives • Consult local stakeholders about the particular dynamics of the SRI-movement in Madagascar, known as the birthplace of SRI. Assess the current state of SRI practice in Madagascar, including the roles of key promoters and organisations, projects, etc. Learning from farmers and other stakeholders about their actual rice cultivation practices, the inter-relationships and coherence of cultivation systems, farmers’

158 motivations and rationales, etc. Exploring the factors that drive and constrain adoption and disadoption of SRI and its component practices.

Local support The field visit was organised with the help of FOFIFA (the National Centre of Applied Research for Rural Development) and the kind cooperation of the Groupement SRI (SRI Grouping) and the Better U Foundation (BUF). The itinerary for field visits is shown in the Appendix.

Constraints In the limited time available for this field visit, it was possible to visit areas along the North– South axis of the Route Nationale 7 between Antananarivo and Fianarantsoa. This axis includes the area where SRI was first developed and is understood to be a region where SRI practice is concentrated, having been promoted by various organisations. Although the area visited includes different soil types in the northern and southern zones, our visit was limited to the central highlands and we saw a limited range of agro-ecological contexts. However, we succeeded in meeting a varied sample of different organisations, including government agencies, non-governmental and civil society organisations (NGOs and CSOs), farmers’s organisations, prominent farmers, the private sector and academic researchers. A list of the people and organisations we met is provided in the Appendix.

Purpose of this report This report is to inform interested parties about the field visit to Madagascar and to make some preliminary observations about what we learned from the visit. It is important to mention that this report is not a comprehensive review of our progress with the project as a whole, nor an interim statement about our findings on SRI as a social movement. The report focuses specifically on our observations in Madagascar and how these inform our understanding of how SRI works, how it spreads and the mechanisms by which it has been adapted and changed as it spreads in practice.

Key observations We present our observations under three themes: 1) SRI agronomy and components, 2) SRI spread at institutional level and 3) SRI spread at village and farm level.

159 1. SRI agronomy and components: SRI: a fixed package, or a flexible suite of principles? This is a well-known theme in debates and arguments about SRI, an issue on which there are diverse and conflicting views among stakeholders. We met some people who clearly regarded the six principles of SRI 81 as something very like a fixed package, but also others who were very relaxed about farmers implementing only some of the recommended practices or who celebrated the adaptations made by farmers in practice.

Relationships between rice systems The Malagasy rice community in general applies a three-fold classification of systems of rice cultivation: SRI, SRA ( le système de riziculture amelioré , the improved system of rice cultivation) and SRT ( le système de riziculture traditionelle , traditional rice cultivation). SRA may be understood as an improved, ‘Green Revolution’ package, which was promoted by IRRI in the 1970s. SRT in principle describes farmers’ conventional practices, though it may be little more than a residual, catch-all label for whatever is not SRA or SRI. SRI post-dates SRA, which may explain how its name came about. The relationship between SRA and SRI is not universally agreed. Some stakeholders see the two systems as diametrically opposed, with SRA representing a GR-style high-input system and SRI a low-external input alternative. Others suggest that SRA is effectively a stepping stone towards SRI. Others – notably some of the farmers themselves – evidently regarded SRA and SRI (and indeed SRT) as complementary alternatives that could be suitable choices for different fields and under different circumstances. In other words, the three systems co-exist and will probably continue to do so for the foreseeable future.

Switching to SRI: a gradual process or a quick transition? Related to the above point, we found differing views on the desirable methods for making transitions from existing methods to improved practices. • Some informants advocated an ‘all-or-nothing’ approach. This appeared to be for one or both of two reasons: (1) SRI should be regarded as an integrated package, whose full benefits can only be realized through the synergistic effects of the six practices in concert, and therefore any deviation from ‘orthodox recommendations’ was undesirable on principle; (2) SRI would only produce benefits substantial enough to

81 Laulanié, H. de (1993) ‘Le système de riziculture intensive malgache’, Tropicultura (Brussels), 11(1), 110–114.

160 showcase its benefits and convince farmers of its merits, if it was implemented as a whole package. If the complete package was compromised, even if the resulting system still performed better than SRT, there was a risk that the benefits would not be substantial to convince farmers that it was worth doing. • Other informants felt that each of the individual components of SRI independently represented an improvement on conventional methods. Therefore, for farmers to adopt any or a few of the SRI practices represented a desirable step forward. This was portrayed as a pragmatic stance, rather than taking a dogmatic view of what farmers ought to be doing for optimal results.

Water management is key to SRI practice Again, this is a familiar point of discussion in SRI. In Madagascar, the single most important issue raised by almost all stakeholders as a key constraint/determinant in farmers’ ability to adopt SRI was the capacity to manage water. This is a separate question from the availability of water.

The learning curve A key process in the adoption of SRI is the learning curve involved in getting accustomed to the novel methods. In particular, transplanting young and single seedlings was regarded as difficult at first, but informants felt that the learning curve was rapid. We learned that some women transplanters ( répiqeuses ) have become specialists in SRI transplanting, moving from one community to another at transplanting time to lead the transplanting activities and guide local transplanters.

Adaptations in transplanting guidelines Original guidelines recommended transplanting 8-day seedlings; many local organisations typically now recommend a slightly more flexible guideline of 8–15 days. However, others now recommend transplanting at the two-leaf stage, a physiological rather than temporal guideline, which they regard as a more universal rule that can be applied in different regions (according to faster or slower seedling growth). This appears to represent a divergence of views.

Seedling spacing and soil fertility

161 There appears to be some unclarity/disagreement over the relationship between seedling spacing and soil fertility. Should seedlings be more widely spaced as soil fertility is more favourable, or more narrowly? The biophysical mechanisms thought to be at work would be different in each case. Is it a question of wider spacing allowing plants to spread and grow in fertile soils, or is it that more fertile soils mean that plants can be transplanted closer together without adverse competition for nutrients among them? This would appear to be a question for agronomic research.

Costs It was widely accepted (although not universally) that SRI is a more expensive practice, primarily because of higher labour demand. However, some people emphasise that savings on seeds and fertilizers outweigh the higher labour costs, so that total investments in SRI are lower. In either case, SRI was widely regarded as being worth adopting if possible because it was more productive and more profitable. The phrase ‘if possible’ is important, but the precise combination of conditions judged to be favourable is not always clearly explicitly defined. The economics and productivity of SRI cultivation are clearly important areas for further research.

SRI and risk Some informants suggested that SRI might increase production risk. Key concerns surrounded the fragility of young seedlings and the wisdom of transplanting such small seedlings singly and widely spaced. Some informants insisted that this was too great a risk to invite the poorest farmers to take on. Might this factor inhibit SRI practice among some groups?

SRI and variety choice Many SRI promoters, in Madagascar as elsewhere, believe that one of the key strengths of SRI is that farmers can achieve significant benefits without having to adopt new varieties. However, a few informants strongly believed that it was very important to use improved varieties with SRI. The rationale was that new varieties were necessary in order to deliver the substantially improved results these informants felt were vital in order to convince farmers to adopt and maintain SRI practices. This would appear to be a strategic (or perhaps tactical) judgement, rather than one that is dictated by technical factors. However, a good question

162 would be whether particular varieties exist, or could be bred, to produce good results under SRI management.

2. SRI spread at institutional level Two major waves of SRI in Madagascar? A hypothesis On of the most interesting insights gained from investigating the dynamics of SRI in Madagascar is the realisation that, for at least some farmers and organisations, SRI is already a very well-embedded part of the rice farming scene. We met some farmers who had been using SRI methods for more than two decades and numerous others who were fully aware of the system as one among a number of alternatives they might use alongside one another. However, we also met others who have only been practising SRI for a couple of years. We have formed the impression that SRI has spread in two waves in Madagascar. The first wave dates from some years after the system’s discovery/invention in 1983, beginning perhaps in the second half of the 1980s and sustained into the mid- or perhaps late-1990s, approximately. This wave appears to have been led originally by Père de Laulanié himself and latterly by the Association Tefy Saina . Progress seems to have stagnated somewhat between the mid-1990s and early–mid 2000s. The second wave seems to have been stimulated by the involvement of the Better U Foundation and the enrolment of the former president, Marc Ravalomanana, in the mid-2000s. Whether this hypothesis is valid is a question that requires further investigation.

Little international exchange on SRI Bearing in mind the way that SRI appears to have spread internationally from its origins in Madagascar, we were surprised to find extremely little evidence of connections between the Malagasy SRI community and SRI organisations and practitioners in other countries. We found some awareness that SRI was being practised in other countries, but hardly anyone who had significant knowledge of what was going on outside Madagascar. Some stakeholders expressed the view that it was disappointing and embarrassing that what was perceived as a Malagasy innovation, had progressed relatively weakly in its country of origin. Few people received, or were aware of how to find, information on SRI from outside the country. This even seemed to apply to some international organisations that promote or provide training in SRI in Madagascar and also work in other countries.

Little exchange between scientific researchers and practitioners

163 We found very little evidence of knowledge- or information exchange between scientific researchers and practitioners in the Madagascar SRI community. Very few scientific organisations in the country seem to be carrying out agronomic research on SRI. One university agronomist with considerable expertise and experience in SRI appeared to be disconnected from the SRI community and lacked funding to carry out agronomic research on the system. Few, if any, of the SRI-promoting organisations were seeking scientific input, advice or research on the agronomy of SRI. However, field-level technicians and trainers evidently have a large reservoir of practical, qualitative knowledge on the characteristics, pros and cons of all the local systems of rice cultivation.

Exchanges among SRI organisations in Madagascar We found few indications of strong or sustained linkages or exchanges among SRI organisations until the recent past. It seems that up until 2–3 years ago, the various small NGOs, training organisations and projects working on SRI were operating largely independently of one another. This situation has changed recently with the creation of the Groupement SRI . This loose network of SRI organisations at the national level is in the process of establishing regional SRI platforms across the country. This initiative may foster increased sharing and exchange of information and perhaps also greater coordination and mutual learning, but this remains to be seen. A few informants, including one or two involved with the Groupement , seemed sceptical about the network’s prospects, though they were prepared to engage with it and see what would emerge.

Exchanges among farmers There is considerable anecdotal evidence of knowledge exchanges among farmers about SRI. In particular instances, knowledge about SRI seems to have been carried by migrant farmers moving back and forth between communities, or through visits to relatives in other parts of the island. Some organisations support individual farmers to become ‘farmer technicians’ and a few farmers have become nationally famous for their knowledge and experience with SRI; these farmers are visited by others and spend some proportion of their time informing and training others. We do not know how important these exchanges are as a mechanism for spreading knowledge and practice in SRI. Some farmers had also first heard about SRI through radio broadcasts.

Lack of reliable data on SRI’s extent and spread

164 There is a dearth of reliable statistical information about rice practices being applied in Madagascar. The figures widely considered to be most reliable are estimates by the Ministry of Agriculture, which believes that SRI occupies only about 3,000 ha. out of a total of about 1.3 Mha. of rice in the country. SRT occupies about 760,000 ha., direct seeding about 130,000 ha. and SRA about 90,000 ha. (See the workshop report in the appendix; note that the figures given by the Ministry do not add up…). A widely cited estimate from a CIRAD/FAO report prepared in 2000 82 suggests that SRI occupies only 0.34% of the rice area. It is not clear how this figure was derived and it appears not to have been updated. Local stakeholders insist that this figure must be a considerable underestimate; but we have not been able to find more recent dependable figures to show how many farmers are now practising SRI or its components on how many hectares of land.

The role of technical support Many of our informants, especially among farmers themselves, attested that sustained technical support was vital in ensuring that SRI became embedded. This claim would appear to be endorsed by Moser and Barrett’s 83 findings in the villages they studied. However, we also met farmers who had taken considerable initiative to learn about and try to practise the system with relatively little support. It seems that examples exist of both cases. What is the relative importance of unsupervised processes of initiative and experimentation compared with formal training and technical support?

Labour as a key constraint Next to water/irrigation (see above), labour was considered to be a key constraint for SRI practice, especially at transplanting time. Once again, however, it was felt that the extra labour was worth the investment if farmers could afford it. Some villages had made use of, created or formalised local institutions for collective labour management to help with this aspect. 3. SRI spread at village level SRI practice, farmer attitudes and group dynamics

82 Marie-Hélène Dabat, avec l’ appui de Pierre Fabre (2000): Diagnostic et perspectives de la filiére riz à Madagascar. Avec l’ UPDR et Equipe de consultants nationaux. CIRAD/FAO. 83 Moser, C.M., Barrett, C.B. (2003) The disappointing adoption dynamics of a yield-increasing, low external-input technology: the case of SRI in Madagascar. Agricultural Systems 76, 1085–1100.

165 Some informants claimed that group dynamics played a role in SRI adoption and practice. It was suggested that farmers liked to conform with their community; this could be a constraint on SRI practice, because some farmers hesitate to start doing something new and different from their neighbours. It was also suggested that some farmers would not want to adopt SRI precisely because it is more productive. This was attributed to a strong societal or cultural pressure against becoming too successful or wealthy compared with the rest of the community. In one community, an institution had been formalised for collective management of labour and this was rationalised as a way to implement SRI without any individuals doing disproportionately well out of it. Farmers’ attitudes, e.g. conservatism or laziness, were also often blamed for non-adoption of SRI or unwillingness to experiment. We are sceptical about explanations that attribute phenomena such as cultivation choices to cultural values or factors, as if they were independent causal mechanisms. We are interested in learning more about farmers’ own rationales for trying, adopting, not adopting or disadopting new practices.

Farmer organisations and demand for technical support We found indications that farmer organisation may play a role in SRI promotion and adoption. What might the role of farmers’ organisations be? One hypothesis is that organised farmers exert a more coherent and coordinated demand on NGOs and extension organisations. Several of the NGOs we met pointed out that they had not made a strategic decision to promote SRI, but had offered training and support in SRI in response to farmer requests. This observation relates to the question above about how farmers learn about SRI and confirms that, at least in the relatively well-connected areas we were able to visit, SRI is already firmly established as one of the improved rice cultivation systems recognised locally.

Some knowledge gaps and areas for further research • What combinations of cultivation practices are actually implemented by farmers? Are there any systematic patterns? Do farmers’ practices conform to SRI or other recommended systems? If so, how closely? How far has SRI spread? What is the balance between SRI and other systems in individual farmers’ or communities’ practices? • Are there any systematic correlations between SRI practice and agro-ecological factors, such as soil quality, rainfall/irrigation, temperature, slope/topography and rice variety? Are there any patterns discernable through these correlations? Are there any

166 systematic correlations between SRI practice and household socioeconomic factors, such as labour organisation, household size, wealth and non-farm livelihoods? Are there any patterns discernable through these correlations? • Are there any systematic correlations between SRI practice and institutional/community factors, such as the presence or absence of farmer organisations and peasant-technicians, etc? Are there any patterns discernable through these correlations? • Are there any systematic correlations between SRI practice and infrastructural factors, such as the availability of technical support, irrigation infrastructure, distance from the main road, distance of fields from the farm, availability of credit, etc? Are there any patterns discernable through these correlations? • Is labour demand increased under SRI? If so, do/can these increased demands create positive externalities for farm workers through the labour market?

167 Appendices: A1. Itinerary Madagascar field visits

Date Ezra Berkhout and Dominic Glover Herman van Keulen 13 April Wageningen – meeting with Martin Smith, former FAO representative in Madagascar and SRI farmer 15–20 April Arrival / Antananarivo 21 April • Antananarivo / University of Antananarivo, Department of Agronomic Sciences (ESSA) • Ambalavao commune / M. Randrianarison 22 April Ampitatafika commune 23 April • Antsirabe (ville) / DDR • Manandona commune, Andranomafana commune 24 April Fianarantsoa (ville) / FOFIFA 25 April Androy commune / Koloharena Arrival / Antananarivo 26 April Fianarantsoa (ville) / DRDR, DDR, FRDA, FAFAFI 27 April Maroharona commune, Soatanana commune Fianarantsoa (ville) / Association Tefy Saina 28 April Talata Ampano / FERT 29 April • Ambositra / ADRA • Return to Antananarivo 30 April Antsahabe commune (‘MAP village’) 1–3 May Antananarivo – workshop planning 4–5 May 2-day workshop, Antananarivo 6 May Field visit, Antsahabe (‘MAP village’) 7–9 May Antananarivo – mission completion

168 A2. Organizations visited in Madagascar

Government agencies Ministry of Agriculture M. Philibert, Secretary-General Direction regionale du developpement Mme Joséane Voahangy Rakotondranaivo rurale (DRDR), Antsirabe, (Regional Director), Mme Modestine Ratsimbazafy Direction du Developpement Rurale Mme Simona Pierrette Rasoarivelo (DDR) (Regional Director, Haute Matsiatra) DRDR Fianarantsoa [Regional Director] and Mr. Benoît Rakotomiandrisoa (field agronomist) FRDA Pilot Project, Fianarantsoa Mr. Philippe Martel Research organizations Université d’Antananarivo, École Prof. Bruno Andrianaivo supérieure des sciences agricoles (ESSA) FOFIFA Antananarivo Dr. Aimé Lala Razafinajara (Director General), Mme Yvonne Rabenantoandro (Scientific Director), Mme Jacqueline Rakotoarisoa (agronomist), Mme Irene Rakotoniaina (Director of Communications) FOFIFA Fianarantsoa Mme Danielle Ramiaramanana (Regional Director), Mr. Solo (field assistant) International donors Japan International Cooperation Agency (JICA) NGO’s Association Tefy Saina (Fianarantsoa) Mr. H. Bernard Rakotonirina Adventist Development Relief Mr. John Ravelomanantsoa, Mr. Association (ADRA) (Ambositra) Mahatiatra, Randrianasolo, Mr. Jean- Michel Ralaivao Centre de Promotion Rurale Frère Joany and M. Jean Denis (Ambositra) Groupement SRI / Groupe Conseil Chair GSRI, Prof. Michel Simeon; Mr. Developpement (GCD) Joeli Barison, + 1 Better U Foundation Mme Winifred Fitzgerald, Mr. Rames Abhukara. Aga Khan Foundation / ICRAF Dr. Toon Defoer SAF/FJKM Mr. Alfred Rasamimanana AROPA/FERT M. Andry TOM M. Justin Léonard Rabenandrasana Farmers organizations:

169 Koloharena M. Jules Randrianarivelo (president) M. Jean Louis (communal president) Groupement de Paysans Semenciers M. Ratsangana, M. Ignace Private Sector: SDMad (Semis Direct de Madagascar – ??? Direct Seeds of Madagascar)

A3. Visits to villages and meetings with farmers in Madagascar:

Visits to villages / meetings with farmers: Near Antananarivo Ampitatafika M. Jean Rabefaritra Mme. Hanitra Mme. Randriamiharisoa Ambotovotsy Ankazobe Antsahabe (MAP village / GCD supported) Behenjy Prof. Bruno TOM / Justin

Near Antsirabe Manandona M. Dada Andranomafana Near Fianarantsoa Ambatovaky Ampapane Maroharona M. Ratsangana, M. Ignace Soatanana M. Ralalason Talata Ampano M. Raleva, M. Michel Ralaivao

170 A4. List of workshop participants ORGANISMES Noms et prénoms FONCTIONS 1. Dominic Bruce Andrew GLOVER Post Doctorant/Economiste 2. Ezra David BERKHOUT Post Doctorant/Agro Economiste UNIV. WAGENINGEN/PAYS BAS 3. HERMAN Van Keulen Chef du Département Systèmes de Cultures Végétales CORNELL UNIVERSITY/USA 4. Julie Gabrielle LAUREN Depart.Crops and Soil Sciences UNIV. DE CALIFORNIE/USA 5. Tim KRUPNIK Depart.of Environnemental Studies TAMIL NADU AGRICULTURAL UNIVERSITY, INDE 6. TM THIYAGARAJAN INSTITUT DE MANAGEMENT XAVIER 7. C Shambu PRASAD Associate Professeur BHUBANESWAR . INDE ESSA, UNIVERSITE D’ANTANANARIVO 8. Andrianaivo Bruno Enseignant chercheur LAULANIE GREEN UNIV . 9. Rakotonirina Marius Technicien MINAGRI 10. Raoelinirina Harisoa Chargée d’études 11. Ramaherintsoa Claudine Assistante/DGA MINISTERE DE l’AGRICULTURE 12. Rakotoarisoa Mina Tsiriarijao Coordinatrice des Organismes Rattachés 13. Ranoromalala Olivia Représentante/DPA FOFIFA 14. Razafinjara Aimé Lala Directeur Général 15. Ramilison Lucile Directeur.Administratif et Financier FOFIFA 16. Rabeson Raymond Chef du Département Recherches Rizicoles 17. Rakotondrasata Martin Fidèle Chercheur/DRD

171 18. Ramiaramanana Danièle Chef du Centre Régional de Recherches/HPS 19. Rakotoarisoa Jacqueline Coordinatrice URP/SCRID GROUPE CONSEIL DEVELOPPEMENT

20. Razafimanantsoa Rijaharilala SP GSRI GCD /SP GSRI 21. JOELIBARISON 22. Randrianarivelo Andry GSRI ORGANISMES D’APPUI AABCV–Antsirabe 23. Razafimahefa Heriniaina ADRA-USAID / Salohy Ambositra 24. Ralaivao Jean Michel CRS 25. Rakotoarimanga Njara Tahiana Responsable Suivi & Evaluation SAF FJKM 26. Ramanankatsoana Laurence Louisa BHC 27. Peter HOFS Directeur 28. WINIFRED Fitdzgerald Consultante Internat.Dév. BUF – Better U foundation 29. RAMES ABHUKARA Consultant Internat.Dév 30. TSUKII YOSHIFUMI JICA 31. Nakamura Hiroraka Expert technique/JICA - MinAgri 32. Junichi Yamaguchi JICA - MinAgri ASSOC/CONFED/PAYSANS ASSOCIATION TOM 33. Randriamiarintsoa Tsimba Conseiller Technique 34. Randriamahala Télesphore Technicien FEKRITAMA 35. Rabarimanana Rafamantanantsoa Technicien CEDAM 36. MERISON Serge Spécialiste/développement, Coordinateur

172 FOKONTANY Antsahabe 37. Rakotoarimanga Jean Claude Président ; Paysan ASSOCIATION TONTOLOMIAINA 38. Rabenandrasana Justin Léonard Président ASSOCIATION TAOEZAKA 39. FRERE HUBERT Conseiller Technique TRANOBEN’NY TANTSAHA / Haute Matsiatra 40. Rasamimoratsiahiana Henri ASSOCIATION TEFISAINA 41. Rafaralahy Sébastien 42. Ravoavy Michel Vice Président CPM Itasy 43. Razafiarison Clément Paysan ASSOCIATION KOLOHARENA 44. Andrianjafimahery Haingovola CDR Tanà Sud 45. Randriamiharisoa Mamy Harison DI R. REGIONAL .DE DEV / DDR .

DRDR ANALAMANGA 46. Raharifara Modestine C. SRAPV CIRAGRI 47. Randriamanana Laza Responsable/Tana Sud DRDR Vakinankaratra/ANTSIRABE 48. Ratsimbazafy Modestine C.SRAgri DRDR AMBOSITRA (Amoron’i Mania) 49. Rajaonarison Jean Désiré Responsable PAPV DRDR HAUTE MATSIATRA 50. Rakotomiandrisoa Benoît SRAGRI DRDR SOFIA / Fondation Agakhan 51. HOEREAU Marcelline Directeur de programme DR PAYSANS Comm. Vatomandry 52. Ramananjatovo Herman SG FITAFA / Paysan producteur Comm .Mahabo/Morondava 53. Remi Modeste SAMBO Paysan producteur Comm.Ambararatabe/Tsir/didy 54. Raharinjaka Herison Paysan producteur Comm.Ampitatafika 55. Rabefaritra Jean Paysan producteur COORDINATION VISITES/TERRAIN

173 ESSA 63. Ratovoarimanana Fetra Etudiant stagiaire FACILITATEURS 64. Rasoanaivo Faly BOSS CORPORATION 65. HASINA 66. Ramahandridona Myriam Interprète INTERPRETARIAT 67. Zanarison Zéphyrin Interprète

A5. Workshop report A detailed report of the workshop organised in Antananarivo on 4-5 May 2010 is available on request from the authors of this report

174

Appendix B: Tamil Nadu trip report

175 The Emergence of the System of Rice Intensification as a Social Movement Project funded by the Bill & Melinda Gates Foundation Trip report Field visit to Tamil Nadu, July – August 2010

Table of Contents

1 Background ...... 177

2 Key observations ...... 179 2.1 SRI agronomy and components: ...... 179 2.2 SRI spread at institutional level ...... 183 2.3 SRI spread at village level ...... 187

3 Some knowledge gaps and areas for further research ...... 189

A1. Field visit itinerary Tamil Nadu ...... 191

A2. Map of locations visited in Tamil Nadu ...... 193

A3. Organizations visited in Tamil Nadu ...... 194

A4. Visits to villages and meetings with farmers in Tamil Nadu: ...... 195

A5. List of workshop participants in Hosur (11-12 August 2010) ...... 196

A6. Workshop report ...... 198

Report prepared by: Dr. Ezra Berkhout, Development Economics Group, Wageningen University Dr. Dominic Glover, Technology and Agrarian Development Group, Wageningen University Dr. ir. Rob Schipper, Development Economics Group, Wageningen University

176 1 Background

This document provides a brief report of a field visit to Tamil Nadu, India carried out by researchers from Wageningen University, NL between 25 July and 15 August 2010. The visit took place in the context of an exploratory research project entitled “The emergence of the System of Rice Intensification (SRI) as a social movement”, funded by the Bill and Melinda Gates Foundation (BMGF) (project no. OPP1002894).

Aims This field visit complements the other activities undertaken in this project. The major part of the project consists of desk-based research, making an inventory of the available official and grey literature on SRI as a basis for documenting the current state of knowledge on the emergence and spread of SRI and the networks of people and organisations involved in promoting and spreading SRI nationally and internationally. Two field visits supplement and complement the desk-based research, with the aim of ensuring that the insights derived from literature reviews and studies of the SRI network are grounded with reference to the experiences and insights of practitioners and researchers working at ground level. This research mission to Tamil Nadu is the last of two field visits. The first took place in Madagascar in April/May 2010.

The field visit had two components: • A programme of field visits to observe rice cultivation and carry out meetings and discussions with farmers and other stakeholders; • A 2-day stakeholder workshop to discuss the social dynamics of SRI in Tamil Nadu. The itinerary for and map of field visits, including a list of locations visited and the people and organisations met, as well as attendees to the 2-day workshop are provided in the appendices.

Specific objectives • Consult local stakeholders about the particular dynamics of the SRI-movement in Tamil Nadu;

177 • Assess the current state of SRI adoption and practice in Tamil Nadu, including the roles of key promoters and organisations, projects, etc.; • Learning from farmers and other stakeholders about their actual rice cultivation practices, the inter-relationships and coherence of cultivation systems, farmers’ motivations and rationales, etc.; • Exploring the factors that drive and constrain adoption and disadoption of SRI and its component practices.

Local support The field visit and workshop were organised with the help of Tamil Nadu Agricultural

University (TNAU), in particular staff associated with the Tamil Nadu Irrigated

Agriculture Modernisation and Water-Bodies Restoration and Management (IAMWARM) - project.

Constraints • Given the time period spent in Tamil Nadu, we have been unable to cover all areas and activities in which SRI is being extended. Nevertheless, our visits did cover different agro-ecological regions as well as the most important institutions that play a role in agriculture and SRI extension in the state. Our visits covered both areas with paddy cultivation in lowland river basins (mainly in the Cauvery river basin) as well as areas more upland in Villupuram, Vellore and Krishnagiri districts. Next to many activities from IAMWARM, the major player in SRI extension, we visited most of the NGOs that are involved in SRI extension in the state; • Given the short time period of the visit and the relatively large number of locations visited, the findings in this report do not give in-depth insights in the local processes of SRI adoption and extension, as this was out of the scope of this study and trip. Nonetheless we do feel that this report gives a representative view of the current state of knowledge pertaining SRI in Tamil Nadu.

Purpose of this report This report is to inform interested parties about the field visit to Tamil Nadu and to make some preliminary observations about what we learned from the visit. It is

178 important to mention that this report is not a comprehensive review of our progress with the project as a whole, nor an interim statement about our findings on SRI as a social movement. The report focuses specifically on our observations in Tamil Nadu and how these inform our understanding of how SRI works, how it spreads and the mechanisms by which it has been adapted and changed as it spreads in practice.

2 Key observations

We present our observations in three themes: 1) SRI agronomy and components, 2) SRI spread at institutional level and 3) SRI spread at village and farm level.

2.1 SRI agronomy and components: Research on SRI agronomy A considerable amount of research on SRI has been carried out by TNAU, while relatively little of this research has found its way to scientific journals, both locally and internationally. Based on a number of such research efforts, mainly in the early 2000s, a standard package of SRI adapted to the specific conditions in Tamil Nadu has been developed. This package is based on the initial SRI principles in Madagascar and is locally sometimes referred to as ‘modified SRI’. Hence, extension of SRI in Tamil Nadu, especially by government agencies, mostly follows this standard package and can be defined as follows: • Seedlings transplanted after 14-16 days; • Use of single seedlings (if possible); • Planting of seedlings in grid of 25x25 cm; • Plots are weeded four times during the growing period at a 10-day interval; • Use of alternate wetting and drying; • Integrated nutrient management, using both chemical and organic fertilizer.

Below, a number of key observations with regard to these components will be discussed.

Adaptations in transplanting guidelines

179 The agricultural sector in Tamil Nadu is characterised by a transition towards mechanisation, as a result of scarcity and a relatively high price of labour. The latter is a result of urbanisation and a government scheme to combat rural unemployment, both of which appear to decrease the supply of labour to agriculture. In the case of transplanting, mechanised transplanting machines do exist; however they are not able to carefully transplant a single young seedling per hill. Moreover, although the use of single seedlings is advocated, it is widely recognized that this is not feasible in all circumstances. Especially in saline soils, seedlings do not establish easily and to minimise risks, usually multiple seedlings per hill are planted on such soils, but sometimes also on other soil types. Some informants felt that the losses in productivity with multiple seedlings per hill may not be severe, although quantitative evidence is scarce. One recent study analysed the costs and benefits of using a mechanised direct seeding method that results in paddy planted in rows, but with little spacing between plants within rows 84 . The results suggest that the loss in production, and increase in seed costs are more than offset by the reduction in labour for transplanting.

Seedling spacing and soil fertility The advocated spacing in Tamil Nadu is based on the field trials in the early 2000s and it is suggested that is appropriate for all soil types in Tamil Nadu. There are however some isolated reports of deviations from this practice. One farmer is said to have increased spacing, on very fertile plots, while some informants told us that farmers complain that the recommended spacing is too wide. The latter case would describe a situation in which the plants would not cover the field completely and bare soil would remain visible up till the maturing. This suggests an inefficiently low number of plants in such fields. Indeed we did observe such plots on a number of cases, yet in many instances farmers and researchers have also suggested that this is beneficial as it reduces incidence of rat damage. Quantifications of such damage, and how it changes when adopting SRI, are not available. Whether such losses outweigh the potential increase in production by reducing the spacing remains to be investigated.

84 Field trials conducted in 2009-2010 at the TNAU Regional Research station in Krishnagiri. Researchers are currently in the process of drafting the research paper/report.

180 In some cases it was observed that farmers are advised not to grow SRI in acidic or saline fields. To this end soil testing kits are sometimes provided to farmers. While this is likely to be an effective way to minimise risks to farmers, it does suggest that SRI is allocated selectively on plots, conditional on local soil fertility conditions. This may substantially invalidate conclusions based on direct comparisons between yields in conventional production and SRI practice, as commonly done by many organisations in the state.

Weeding Within the standard package farmers are recommended to weed four times, in intervals of 10 days. The marginal benefits of each respective weeding are not clearly documented, but the benefits of the third and fourth weeding are likely to be low since many farmers did not follow this recommendation strictly. Many informants strongly believed that the beneficial effect of using the cono- or rotary-weeder is a result of root pruning. However, it is difficult to separate this effect from other possible explanations like soil aeration, incorporating organic matter (from weed growth) in the soil, or loosening of the soil, something that has not yet been investigated in great detail.

Water management Similar to the findings in Madagascar, we find that good control over water is a prerequisite for SRI adoption. Most SRI-farmers that we visited were supplying water from their own bore wells, with most informants confirming that water tables decrease steadily. The provision of water through canal-irrigated areas is often erratic and is likely to refrain SRI-adoption. As such the effectiveness of SRI is likely to vary according to differences in the water control regime. More detailed information on this information is scarce. The promotion of SRI in Tamil Nadu is closely associated with saving water (or increasing water use efficiency), even though it is actively promoted in some areas where water supply is abundant. This focus on reducing inefficiencies in water use was much less apparent in Madagascar. But while this policy objective is straightforward, electricity is supplied free of cost to the rural population, and farmers using electrically powered bore well pumps have no direct incentive to reduce water use. The impact of SRI adoption on the actual use of water, and the associated

181 environmental externalities, are not well documented. Given the policy focus, this clearly is an important area for future research.

SRI and rice varieties Similar to our findings in Madagascar, our informants had conflicting views on the performance of SRI in relation to different rice varieties. One view is that while the gain in yield in absolute terms is highest when using hybrid seed (when switching from a traditional system to SRI), it is highest in relative terms when using local varieties. Again, others stated that SRI performed equally well with all varieties.

Mechanisation Currently, changes in the economic structure of Tamil Nadu have a large impact on the rural sector. Both the development of the industrial sector in urban areas as well as the National Rural Employment Guarantee Act (NREGA), aimed to provide stable income for the landless rural population, create alternative employment opportunities for agricultural labourers and push up rural wages. Hence, farmers respond by mechanising part of their paddy production. Both the public and private sector address this increased demand for mechanisation. The former by research on mechanical weeders and transplanters specifically suited for SRI, the latter through local production and imports of such machinery. While transplanters are ideal for transplanting in rows, it is considerably more difficult to mechanically transplant in grids (next to transplanting single seedlings as mentioned earlier). While current research at TNAU addresses such and other issues, it is not clear whether the yield loss in row (and not grid planted) fields outweighs potentially higher costs of mechanically transplanting in grids. Nevertheless, given the substantial capital investments the current use of mechanised transplanters is still relatively low. A form of mechanisation that starts becoming more common is the use of motorised weeders. While many farmers appear to use the cono- and rotary weeder, some others are using motorized weeders that are imported from Japan/Korea and China. Many of these weeders are adapted to a specific spacing as used in the countries of origin, and as such are likely to dictate the actual distance between rows, which in some cases differs from the present recommendations.

182 The learning curve Similar to our findings in Madagascar, we observe that a key process in the adoption of SRI is the learning curve involved in getting accustomed to the novel methods. In particular, transplanting young and single seedlings was regarded as difficult by most labourers at first, but informants felt that the learning curve was rapid. We learned that labour teams received training in new transplanting methods, a skill that sometimes allowed them to seek employment outside of their usual working environment (see also discussion on impact on p.12).

Costs SRI is widely seen as a more productive package. Many farmers report a significant increase in production, but the improved input productivities mentioned by many informants, are equally a result of important input reductions. The latter is clearly different from Madagascar, where most informants felt that labour costs were considerably higher. An important cost reduction relates the large reduction in seed rate; a cost that is also more pronounced in Tamil Nadu where use of improved and hybrid seeds is much higher. In addition, many farmers reported that costs of labour, transplanting and weeding, were reduced as compared to conventional cultivation methods. The importance of reducing costs is also apparent from the transition to mechanised production methods. Given this trend, it is likely that production methods with even lower costs, such as direct seeding, may replace the more labour-intensive components of SRI in the future.

SRI and risk Many informants perceive that the risk of crop failure is reduced in SRI systems, especially when SRI methods are not used in saline or acidic soils. This relates both to the rat damage mentioned earlier, but also to the ability of plants to withstand the strong winds in the cyclone season better, mostly due to the deeper rooting system.

2.2 SRI spread at institutional level

The origins of SRI in Tamil Nadu

183 SRI has been brought to Tamil Nadu through two distinct pathways. The first is an anecdote 85 about a French traveller that came to Auroville 86 in the late 1990s. This traveller brought with him a book about the SRI method in Madagascar. It was met with some enthusiasm and led to experimentation in the Auroville/Puducherry area, though most farmers have stopped using SRI by now because of disappointing yields. While this path may have contributed to the spread of SRI in Tamil Nadu, in particular through the work of some NGOs in this area, the second and more important pathway came through Tamil Nadu Agricultural University. TNAU and Wageningen University cooperated in projects on rice cultivation in the late 1990s and early 2000s in which SRI was assessed. Scientists from Wageningen University thereby made initial contact with Norman Uphoff and learned about the method in more detail. Similarly, TNAU and the International Institute of Rice Research (IRRI) cooperated in a research project in the early 2000s in which a number of SRI components was tested. Based on these, and other local research, the standard SRI- package was developed and taken up for further extension. Initial adoption and spread remained low according to some government statistics, but figures from the IAMWARM project suggests adoption has taken up with the advent of this project. That said, some of the components in SRI are not new to Tamil Nadu and have been practised in the region well before the 1990s, as expressed by several informants. The rotary weeder for example, and by consequence planting in rows, had been introduced in the region as a result of IRRI activities in the 1970s, but together with row-planting was perceived as labour and drudgery-increasing and never really caught on. One farmer was still able to show a rotary-weeder he had used 30 years earlier, while some others recalled their parents using it. Similarly, although strictly speaking some may not consider is as part of SRI, the currently advocated improved nursery raising techniques, are very similar to the Dapog nurseries introduced by IRRI in the 1960s. Two farmers also stated that they has used the practice of single seedlings before, born out of necessity due to a shortage of seeds, but most other farmers had not encountered this practice before.

85 This is also documented in the study by Shambu Prasad (2006) in “System of Rice Intensification in India: innovation history and institutional challenges”. 86 Auroville is an experimental town in Tamil Nadu, close to the border with Puducherry founded in the late 1960s. Being a wasteland when many travellers from around the world settled here, it has undergone an amazing transformation to a sustainably managed area of farm and forestland.

184

SRI as another BMP? Most of the SRI promotion in Tamil Nadu is done through the IAMWARM project in addition or in cooperation with the government extension services. Given this small number of players most SRI extension takes the shape of promoting a relatively fixed technological package, in fact in a way very similar to the promotion of other Best Management Practices (BMPs) around the world. This situation is clearly different from Madagascar, where SRI extension strongly centred on the adaptation of the SRI principles to local conditions. This large difference in the approach to SRI is likely to stem mainly from differences in the institutional environments. In Madagascar, lacking an effective extension service in many parts of the country as well as endorsement by some government agencies, NGOs play a relatively important role in extension of agricultural practices. In Tamil Nadu the presence of NGOs is much smaller, probably a result of the fairly effective government extension system. Hence, not only did many (but certainly not all) NGOs in Madagascar refrain from considering SRI as a fixed package, the sheer number of different NGOs in Madagascar accounts for much of the diversity in SRI observed. The dominance of government-related institutions in Tamil Nadu, and the development of a statewide recommendation by TNAU, thus explains the relatively homogenous package being extended. However, as discussed below this does not imply adoption of SRI- components is homogenous throughout the state, with farmers adopting and adapting SRI in various ways.

Links between organisations The system of universities and research bodies in India is relatively strongly focused on policy-makers. As such there exist close links between TNAU and policy-makers, as a result of which TNAU advised for the inclusion of SRI in the IAMWARM project when the proposal was being developed. Moreover the university itself is directly involved in extension, for example through the KVK 87 system. We observed relatively little cooperation though between government and non- government organisations. As mentioned only few NGOs extend SRI in Tamil Nadu.

87 Krishi Vigyan Kendra is a network of resource centres that provide information and extension to farmers at local level. In principle each local district has one KVK resource centre.

185 Moreover, other than in Madagascar or for example another Indian state (Orissa), there exists no platform in which SRI experiences are shared amongst different organisations.

(Inter)national linkages Although it is not formalised there exists considerable exchange of information on SRI at a national level. The IAMWARM is regarded as a good example of effective extension of SRI and their input is asked in many other instances in the country. Some exchange is also present in the sector of NGOs, as some work in multiple Indian states. Moreover, given the World Bank funding of the IAMWARM project there is also some international knowledge crossover. Yet, even though most people (including many farmers) know that SRI originated from outside India, there is relatively little contact between local research organisations and those abroad.

Lack of reliable data on SRI’s extent and spread Similar to our observations in Madagascar there is a dearth of reliable statistical information about the scale in which SRI is being taken up and the impact is has on households and the environment. This is somewhat surprising, given the extensive involvement of the government as well as major international institutions such as the World Bank in IAMWARM project. While the IAMWARM keeps good track on the project outputs (i.e. the number of demonstration farmers reached and their practices), they have considerable less information about the project outcomes (i.e. the number of subsequent (partial) adopters). Added to this, the state government sets targets for the number of SRI-adopters in the state, and this is likely to result in an overestimation of the number of adopters as reported by many local government offices. There also appears to be very little information on impact (and heterogeneity thereof) of SRI on households and labourers. Most of the information collected within the IAMWARM project, including field practices, relates to the demonstration farmers and not the subsequent adopters. This is all the more important given the fact that average crop yields as often reported are likely to be confounded by other factors such as differences in soil fertility. That said a number of activities are currently being geared up by IAMWARM to fill this hiatus.

186 2.3 SRI spread at village level

What is an SRI adopter? We have found no agreement amongst the informants as to the definition of an SRI adopter. Some informants went so far as to argue that any farmer who achieves a certain improved yield level may be considered to be an SRI-farmer. We are concerned that such views might greatly distort reliable data collection. SRI adoption cannot easily be captured in simple binary categories. In a recent study 88 in Tamil Nadu, enumerators went to interview SRI-adopters based on a list provided by the extension service, but found that many farmers on the list did not consider themselves as SRI-adopters. Nevertheless, in their survey it often turned out that they were in fact practicing one or more of the SRI-components. Hence, we suggest that to properly understand SRI adoption, it needs to be described in terms of the different components, and/or as a degree of adoption. Such a definition is also more useful in understanding differences in the uptake of SRI components and their impact across different socio-economic/biophysical environments. Finally, such a distinction would also allow to identify the marginal effects of each component, a subject on which some research has been started 5.

Measuring impact Most informants confirmed that the net benefits under SRI are superior to conventional methods due to higher output and reduction in costs. Hence, it is expected that this translates into improved well-being in the households that adopt SRI, or its components. Some informants mentioned the results of a preliminary study, the results of which suggest that larger farmers are less likely to adopt, since they have more difficulties in timely recruiting sufficient labour for all activities. However, this contradicts with the views of other informants, who stated that in fact as a smaller farmer, who commonly relies more on off-farm income, it is more difficult to follow the recommendations on transplanting and weeding closely due to time constraints. Hence, to determine which types of farmers adopt SRI, or its

88 As presented by Dr. Palanisamy during the workshop in Hosur. Also see the workshop report

187 components, and how the impact of SRI differs across farmer types and locations remains fruitful ground for further research. At the same time, farmers in Tamil Nadu depend largely on, (often landless) agricultural labourers, for whom the impact of SRI is also not clear. Even though many labourers confirmed they were able to quickly adapt to e.g. changes in transplanting techniques, a switch to SRI implies an overall reduction in labour demand. While this is an expected trend in the wake of a decrease in labour supply and increase in wages, the impact on the remaining labour teams remains uncertain. In fact many farmers reported that they met considerable resistance from labourers to the new method of planting. This resistance may indeed stem from the overall reduction in the demand for labour, something that may place a negatively burden on the social structure of labour groups. Clearly, the impact of SRI on labourers merits additional research.

2.3.1.1 Extension and knowledge exchange among farmers The government/IAMWARM-project on one side, and the NGOs on the other side, follow opposing approaches towards the extension of SRI. The former relies heavily on demonstration farmers/plots, often along roadsides. Such farmers receive subsidy for inputs such as rotary-weeders and act as local resource persons. Given their position in the village and contact with extension services, these demonstration farmers are likely to be wealthier, and exert a certain local political influence, but are also not likely to be well connected to the poorest groups of farmers. Some of the NGOs actively try to avoid the group of wealthier farmers in order to target the poorest groups of farmers in villages. This is achieved by organizing a number of participatory meetings in which constraints, and potential possibilities for technical advice are discussed. The wealthier farmers, who can usually access such information themselves, are not likely to remain taking part after a few of such meetings. Hence, the training and information is then shared in relatively homogenous groups of resource poor farmers. If and how these different approached impact the effectiveness of communication and learning between different types of farmers is an important area for future research. In either way SRI is often one of the different technologies that is advocated, and SRI is thereby the preferred technology for paddy production, as advocated both by the government projects and NGOs. Both types of extension bodies offer a pallet of

188 activities/advice to farmers. This ranges from drip-irrigation systems for horticulture (IAMWARM), to assistance with crop marketing (both IAMWARM and NGOs), to social health and hygiene activities (e.g. Ekoventure). In addition both NGOs and IAMWARM make active use of radio programs for technology promotion.

3 Some knowledge gaps and areas for further research

Similarly to Madagascar, we find that there are large knowledge gaps in the documentation of spread and adoption of SRI and its impact. Hence, the first five knowledge gaps below, quoted from our trip report to Madagascar, are equally relevant for the case of Tamil Nadu: • What combinations of cultivation practices are actually implemented by farmers? Are there any systematic patterns? Do farmers’ practices conform to SRI or other recommended systems? If so, how closely? How far and in which form has SRI spread? What is the balance between SRI and other systems in individual farmers’ or communities’ practices? • Are there any systematic correlations between SRI practice and agro- ecological factors, such as soil quality, rainfall patterns, irrigation systems, temperature, slope/topography and rice variety? Are there any patterns discernable through these correlations? • Are there any systematic correlations between SRI practice and household socio-economic factors, such as labour organisation, household size, wealth and non-farm livelihoods? Are there any patterns discernable through these correlations? • Are there any systematic correlations between SRI practice and institutional/community factors, such as the presence or absence of farmer organisations and peasant-technicians, etc? Are there any patterns discernable through these correlations? • Are there any systematic correlations between SRI practice and infrastructure factors, such as the availability of technical support, irrigation infrastructure, distance from the main road, distance of fields from the farm, availability of credit, etc.? Are there any patterns discernable through these correlations?

In addition we identified the following knowledge gaps.

189 • What is the impact of SRI adoption on household indicators, such as income, consumption and health, as well as environmental indicators such as water use and soil quality? How does impact differ across different systems and regions? • To which degree is SRI adoption driven by other sector-wide changes, in particular wages and prices of seeds; furthermore, could the spacing be determined by the availability of imported weeders with a preset spacing? • What are the marginal net benefits for a partial adopter of SRI-components to adopt another SRI-component? What are the marginal net benefits of each recommended (cono-/rotary-) weeding operation? What are the marginal net benefits of switching from row planting to grid planting. In the latter case, would this justify the capital-intensive development of transplanters capable of grid planting? [Is this a point about synergies? If so, should we add the word itself to make this clear? If not, perhaps we should add a point about synergies. It definitely came up in the workshop. DG] • There may be different mechanisms, i.e. root pruning, puddling of the soil, incorporating organic matter in the soil, or aerating the soil, that cause an increase in yields when using the cono- or rotary-weeder. What are the relative contributions of these mechanisms to increases in yield in cono- or rotary- weeded fields? • There are a considerable number of aspects of SRI in relation to soil fertility that remain to be investigated. First, to what degree are higher yields in SRI a results of selective allocation (i.e. based on aspects related to acidity, salinity and iron toxicity) of SRI to the best plots? How does soil structure and quality change under SRI over time? What are the long-term effects of SRI on soil fertility status?

190 Appendices:

A1. Field visit itinerary Tamil Nadu

Date Ezra Berkhout and Dominic Glover Rob Schipper 26-7-2010 • Meeting various scientists at Water Technology Centre and Tamil Nadu Agricultural University, Coimbatore 27-7-2010 • Meeting various scientists at Water Technology Centre and Tamil Nadu Agricultural University, Coimbatore 28-7-2010 • Visits of IAMWARM project sites, Pattukottai • Meeting with scientists at Agricultural Research Station, Pattukottai 29-7-2010 • Visit Department of Agriculture, Tiruchirapalli • Visit of SRI-demonstration farms of the Department of Agriculture 30-7-2010 • Visit of Krishi Vigyan Kendra (KVK), Karur District • Meeting with scientists of the ClimaRice research project • Meeting staff and demonstration fields of the Centre for Ecology and Research (NGO) 31-7-2010 • Meeting scientists and visit research farm at Tamil Nadu Rice Research Institute • Meeting SRI farmers at Poonglam 2-8-2010 • Visit IAMWARM project sites Villupuram district • Meeting district collector Villupuram district 4-8-2010 • Meeting with staff of Ekoventure (NGO) at Puducherry • Meeting with Dr. Nammalvar (involved in SRI promotion throughout India) 5-8-2010 • Meeting with chair of Palmyra (NGO) at Auroville • Meeting with Dr. M.S. Swaminathan, MS Swaminathan Research Foundation, Chennai

191 6-8-2010 • Meeting IAMWARM project director • Meeting with staff at Department of Agriculture, Chennai 7-8-2010 • Visit IAMWARM project sites Vellore district 9-8-2010 • Visit IAMWARM project sites Krishnagiri district • Visit horticulture fields in Krishnagiri district 11-8-2010 • Workshop on SRI at Hosur 12-8-2010 • Workshop on SRI at Hosur 13-8-2010 • Visit SRI farmers and SRI fields with workshop participants at 14-8-2010 • Meeting staff at Agriculture Man Ecology (AME) Foundation at Bangalore

192 A2. Map of locations visited in Tamil Nadu

All field locations that the researchers visited in Tamil Nadu have been marked with a GPS. The locations are stored in Google Maps and are marked in the map of eastern Tamil Nadu below:

The locations can also be viewed directly, and in more detail, in Google maps. This can be done by selecting and copying the hyperlink below, and pasting it in the navigation bar of your internet browser: http://maps.google.nl/maps/ms?hl=nl&ie=UTF8&msa=0&msid=1028270745049584 22821.00048e06d235f66694824&t=h&ll=11.765192,79.101563&spn=3.688534,4.63 0737&z=8&iwloc=00048e07550e23278c4ee

193 A3. Organizations visited in Tamil Nadu

Government agencies: • Agricultural Department (JDA) (Tiruchirapalli) • Agricultural Department (Chennai) • Krishi Vigyan Kendra (Karur, Virinjipuram)

Research organizations: • Tamil Nadu Agricultural University (Coimbatore, Pattukottai, Thanjavur, Aduthurai, Villupuram, Chennai, Krishnagiri) • Climarice-project (Karur) • IAMWARM - project sites:(Pattukottai, Tiruchirapalli, Thanjavur, Villupuram) • MS Swaminathan Research Foundation

NGOs: • Centre for Ecology and Research (Soorokottai/Thanjavur) • Palmyra (Auroville) • Ekoventure (Puducherry)

• Agriculture Man Ecology Foundation (Bangalore)

194 A4. Visits to villages and meetings with farmers in Tamil Nadu:

Village Name District in Tamil Nadu Maharasasamuti Thanjavur Ennanivayal Thanjavur Pallathur Thanjavur Various villages around Tiruchirapalli Tolgate Tiruchirapalli Soorokottai Thanjavur poonglam Thanjavur Konakkampattu Villupuram Alampoondi Villupuram Thenpalai Villupuram Chendur Villupuram Vikravandi Villupuram Kirumampakkam Puducherry, Union Territory Ranipet Vellore Velam Vellore Mungilpudhur Krishnagiri Banganatha Krishnagiri Keeranapalli Krishnagiri Pathakotta Krishnagiri Pillekothur Krishnagiri

195 A5. List of workshop participants in Hosur (11-12 August 2010) Name Position & affiliation Dr. Rob Schipper Development Economics Group, Wageningen University Dr. Ezra Berkhout Development Economics Group, Wageningen University Dr. Dominic Glover Technology and Agrarian Development Group, Wageningen University Mr. Tim Krupnik PhD Candidate, Environmental Studies Department, University of California. Representing Bill and Melinda Gates Foundation. Mr. Tsimba TOM/ATS Madagascar Randriamiarintsoa Dr. Shambu Prasad Associate professor, Xavier Institute of Management, Bhubaneswar Mr. Debashish Sen Deputy director, People’s Science Institute, Dehradun Dr. B.C. Barah Principal scientist (economics), National Centre for Agricultural Economics & Policy Research, New Delhi Dr. Palanisamy Director IWMI-TATA Policy Research Program, International Water Management Institute (IWMI), Patancheru Mr. A Biswas Executive (research) SRI Secretariat (SDTT), Bhubaneswar Mr. A. Ravindra Watershed Support Services and Activities Network (WASSAN), Secunderabad Dr. Vinod Goud WWF-ICRISAT Mr. Bhaskar Joshi Central Programme Officer, Agriculture Man Ecology Foundation, Bangalore Ms. Sangeeta Patil Team Leader, Dharwad, Agriculture Man Ecology Foundation Mr. Ravi Kumar Area unit coordinator, Tiruchirapalli, Agriculture Man Ecology Foundation Dr. T.M.Thiyagarajan IAMWARM project advisor, Chennai Dr. Amod Thakur Water Technology Centre, Eastern Region, Bhubaneswar Dr. S. Chellamuthu Director Water Technology Centre, Tamil Nadu Agricultural University, Coimbatore Dr. S. Vijayabaskaran Associate Professor (Agronomy), Tamil Nadu Agricultural University, Kaveripattinam Dr. S. Manickam Associate Professor (Agronomy), Tamil Nadu Agricultural University, Sandhiyur

196 Dr. S. Mohandass Assistant Professor, Coconut Research Station, Veppankulam Dr. S. Anbumani Assistant Professor, Oilseed Research Station, Tindivanam Dr. S. Selvakumar Assistant Professor, Water Technology Centre, Tamil Nadu Agricultural University, Coimbatore Dr. R. Duraisingh Professor (Agronomy), Tamil Nadu Agricultural University, Madurai Dr. Venkatesa Palanisamy Associate Professor (Economics), Tamil Nadu Agricultural University, Coimbatore Dr. A.K. Mani Professor, Tamil Nadu Agricultural University, Kaveripattinam Dr. M.N. Budhar Associate Professor (Agronomy), Tamil Nadu Agricultural University, Kaveripattinam Dr. T. Sampathkumar Research Associate, Water Technology Centre, Tamil Nadu Agricultural University, Coimbatore Dr. K.R. Jahanmohan IAMWARM-project, Chennai Dr. Raj Rajendran CSISA project (CIMMYT/IRRI), Tamil Nadu Rice Research Institute, Aduthurai. Representative of the International Institute of Rice Research (IRRI) Dr. V. Manivannan Assistant Professor, Tamil Nadu Agricultural University Dr. B.J. Pandian Professor, IAMWARM project, Water Technology Centre, Tamil Nadu Agricultural University, Coimbatore Mr. K. Senthilkumar Senior research fellow, Tamil Nadu Agricultural University, Kaveripattanam Mr. B. Sivakumar Senior research fellow, Tamil Nadu Agricultural University, Kaveripattanam Mr. Sivaraj Joint Director of Agriculture, Tiruchirapalli Mr. Vijayakumar Deputy Director of Agriculture, Perambalur G. Muthukoori Assistant Director of Agriculture, Kumbakonam Dr. T. Sundararajan Program coordinator, KVK, Krishnagiri V. Thirumalai Assistant General Manager, National Bank for Agriculture and Rural Development (NABARD), Krishnagiri V. Palaniappan Secretary, Centre for Ecology and Research, Thanjavur Mr. Rajendran Social Welfare of Educational and Environmental Trust, Nagapattinam Th. P. Baskaran President, Thumbal SRI Farmers Association, Attur

197 Taluk, Salem District Th. Manavalan Farmer, Vikkirapandi Th. Rajappa Farmer, Pillaikuthur Th. Srinivasan Farmer, Parthakotta Th. Jayappa Farmer, Pillaikuthur Th. Jeysankar Farmer, Pillaikuthur Mrs. R. Kanakavalli Farmer, Soorakottai Mr. N. Nanjundan Farmer, Keeranapalli Mr. Ganesan Farmer, Moongil Puthur

A6. Workshop report A detailed report of the workshop organised in Hosur on 11-12 August 2010 is available on request from the authors of this report

198

Appendix C: Timeline

199 A set of interactive timelines was developed during this project, using the online platform TimeGlider . The timelines are scrollable, scalable and searchable. Click on the individual events to read more information. Web links or original sources or further information are provided in some cases. Copy and paste the URLs below to see the timelines on the web:

1. Main timeline: All the logged events displayed on a single timeline. http://timeglider.com/app/viewer.php?uid=line_8ac64a13e780806830e3bafc6b50b08 8

2. Activities of Norman Uphoff: A log of the international awareness-raising activities of Norman Uphoff. http://timeglider.com/app/viewer.php?uid=line_7f424c749cc422176e5f07bd276a42a2

3. Publication of some key scientific papers on SRI: http://timeglider.com/app/viewer.php?uid=line_03753044b9ad50c95f6ca81c6fe37b7f

200

Appendix D: Abstracts of articles developed during the project

201 Three journal articles were prepared and submitted as a result of work carried out under this project. The abstracts of these three papers are reproduced below

1. The System of Rice Intensification: Time for an empirical turn Dominic Glover. NJAS – Wageningen Journal of Life Sciences , in press. doi:10.1016/j.njas.2010.11.006 The System of Rice Intensification (SRI) is claimed to be a new, more productive and more sustainable method for cultivating rice. These claims have proved controversial. One dimension of the controversy has centred on the imprecision with which SRI’s component practices have been defined. The supporters of SRI suggest that the system has been designed to satisfy the needs of rice itself, implying that it is a set of integrated, mutually reinforcing practices that need to be implemented as a package in order to obtain the best results. However, they also argue that the system should be understood as a suite of flexible principles to be adapted to particular agro-ecological and socio-economic settings – the antithesis of a fixed package. This poses a conceptual and practical challenge for scientific evaluation of SRI methods. However, this apparent difficulty is chiefly an artefact created by conceptualizing agricultural methods as standardized packages. A process of translation is always necessary to convert theoretical models or norms into farming practices. Because smallholder farming practices are intrinsically constrained and contingent, they rarely conform precisely to abstract norms. As an alternative, the notion of performance offers a useful way to frame a methodological and analytical approach to understanding what is going on in SRI. Such an approach calls for close technographic observation of farming activities and the interaction between farmers and their fields, plants and tools.

2. A system designed for rice? Materiality and the invention/discovery of the system of rice intensification Dominic Glover. Accepted for publication in the Journal of East Asian Science, Technology and Society (EASTS ) special issue on ‘Rice science, rice technology and rice societies: materiality in research, knowledge and practice in Asia’s main food crop’, forthcoming. The system of rice intensification (SRI) is a novel approach to rice cultivation that is claimed to be both more productive and more sustainable than conventional methods. It is said to have been discovered by a French Jesuit missionary working in Madagascar during the 1970s and 80s. The system has been depicted as a set of methods determined by the needs of rice itself. However, a close analysis of its

202 origins indicates that the creation of the system involved elements of invention as well as discovery. In a process where conceptual understanding evolved over a period of years, empirical observation, theoretical analysis, and purposive judgement all contributed to the compilation of a set of cultivation practices making up SRI. In particular, some aspects of the SRI methodology were not dictated by agronomy alone but were designed to suit peasant farmers. The theoretical underpinnings of SRI could equally have been used to justify alternative choices. Key aspects of the scientific controversy that surrounds SRI reflect not only disagreements on scientific questions but also different perspectives on the appropriate roles of agricultural researchers and strategies of agricultural research.

3. Science, practice and the System of Rice Intensification in Indian agriculture Dominic Glover. Submitted to Food Policy August 2010; currently under peer review The System of Rice Intensification (SRI) is claimed to be a novel approach to rice cultivation that is both more productive and more sustainable than conventional methods. Such claims have been challenged or dismissed by many rice scientists, however. Despite the lack of clear and unequivocal endorsement by science, SRI seems to have spread widely and rather quickly to many rice-growing regions, including various areas of India. This paper discusses how and considers why SRI seems to have attracted the support of diverse stakeholders in Indian rice farming. As such, the SRI phenomenon should be taken seriously. Nevertheless, many scientific questions remain to be answered, concerning the biophysical mechanisms involved in SRI and their effects on plant performance and crop yields, the true spread of SRI practices among farmers and the system’s impacts on farm livelihoods, rice production and resource use. Indian enthusiasm for SRI implies a level of dissatisfaction with conventional approaches to rice intensification and a demand for new methods that can address the perceived problems and challenges of agriculture in the future.

203