Comment to ‘The System of Rice Intensification: Time for an empirical turn’, [NJAS - Wageningen Journal of Life Sciences 57 (2011) 217–224]

An article published a year ago in NJAS considered reasons for the controversy amongst agricultural scientists regarding the System of Rice Intensification (SRI), proposing that an integrated, socio-technical perspective on agricultural innovation be employed to gain a better understanding of SRI and other such innovations [1]. While the article offered many useful insights, its being subtitled ‘Time for an empirical turn’ suggested that the controversy derives crucially from a paucity of empiricism regarding SRI. In fact, much more evidence on SRI has been available from the outset than sceptics or critics have been willing to acknowledge. This raises questions not considered in [1] about why such evidence was ignored or dismissed.

It is appropriate to consider this case further now that Glover has entered into the literature one possible explanation for why SRI became so controversial. This explanation focused on the accumulation and interpretation of knowledge, rather than on institutional and other interests. The latter could provide an alternative, possibly more parsimonious explanation. As one of the protagonists, my understanding of events is not complete or conclusive. However, involvement has given me knowledge of events that can be verified and that put the matter in a different light from how it is presented in [1]. Knowledge on these events should be of interest to professionals in many disciplines who want to see agricultural science and practice advanced.

Critics of SRI have pointed out, correctly, that little agronomic research had been done on the new rice methodology to support some of the claims made for it when these were first presented; there were indeed few published articles in the peer-reviewed literature. This, however, glosses over the fact that in the peer-reviewed literature it was argued that SRI should not even be investigated, that spending any resources on evaluating SRI would be a waste [2,3]. Thus, researchers were steered away from studying SRI, and donors were advised against funding such studies. Further, there was a bias in some journals against publishing SRI papers, as noted below. This created a ‘Catch-22’ situation, where funding was needed to do proper evaluations of the innovation, but funding for this was simultaneously discouraged until evaluations had been done.

The history of this controversy should be of interest because rejection of the SRI phenomenon has been contradicted now by substantial pragmatic evidence supporting the earlier claims, coming more from farmers’ results than from scientists’ reports. How is it that the rice science community could get so embroiled and stalled in contention while the world of applied agronomy moved ahead? SRI experience raises the question of what might be done to reduce or discourage Type II errors (false negatives) within agricultural science. This question has real-world consequences when a beneficial innovation is deterred or delayed by controversy.

•	In Vietnam, the Ministry of Agriculture and Rural Development reported in October 2011 that over 1 million farmers are now using the new methods [4]. Four years earlier, when the Ministry designated SRI to be a ‘technical advance’, fewer than 10,000 farmers in Vietnam even knew about SRI [5].

•

In China, Sichuan Provincial Department of Agriculture reports that the use of SRI methods has expanded from 1133 ha in 2004, to over 300,000 ha in 2010. It has calculated an average yield increase of 1.7 t ha−1 from using SRI ideas and methods during this period. This gave farmers 1.6 million tons of additional paddy rice, worth over $300 million, while reducing their water requirements by one-quarter in a province that has growing water constraints [6].

•

In India, the state of Tamil Nadu started promoting SRI methods in 2004, with added support for SRI extension subsequently from a World Bank project in 2006. It reports that one-third of its rice area is now under SRI management. The Minister of Agriculture has credited SRI with giving the state's rice sector substantial buffering against the effects of drought [7].

•

In the state of Bihar, the government set a goal of 350,000 ha of SRI extension in the summer season of 2011, based on good yields in the two preceding seasons, despite drought [8]. Its 2011 yield is now estimated to reach 11.7 million tons with good rains and spread of SRI methods. The government aims to expand SRI use to 40% of its rice-growing area in 2012. Experience in these countries gives support to one of the arguments advanced in [1]: that farmer practice invariably differs from the recommendations of scientists and extension services as farmers ‘mix and match’ new ideas and elements according to their own resource availabilities and understandings [1]. Glover appropriately refers to the concept of technology transfer as “a hoary model” because farmers adopt only practices that are easy for them or that are fully understood. This makes for a patchwork pattern of technical change in agriculture.

•

Only about 20% of the million Vietnamese farmers who changed their production practices in response to SRI extension are using the full set of recommended practices, according to the Ministry, due to constraints such as limited water control. This could account for why SRI yield increases in Vietnam have been less than reported from other countries.

•

However, other considerations besides yield have made SRI recommendations attractive – water saving, cost reduction, higher net income, resistance to pests and diseases, drought tolerance, and resistance to lodging during typhoons. There is still also scope for further yield increase once the new methods are used more fully or better.

•

In Sichuan, farmers and technicians have adapted SRI concepts to permanent raised beds (no till) with plastic mulch, getting higher yields and incomes even in drought seasons [9]. This makes several improvements upon the original set of SRI concepts to deal with local needs and opportunities. Further, Chinese scientists have begun adapting SRI ideas also to wheat production within the rice–wheat cropping system, seeking to lower greenhouse gas emissions from agricultural fields [10].

Glover's characterization of SRI as a phenomenon rather than as a technology is thus quite apt. However, it begs the question of why there was so much resistance to SRI from some quarters of the scientific community. This comment adds information on the history of SRI dissemination and evaluation, starting in Madagascar but giving also international examples. It is not a full history, as that would require much more than a comment; nor does it try to assess the whole controversy, as that would take a whole volume. It points to salient episodes, incidents and facts that illuminate the case and raise questions about how scientific interests and debates interact, focusing on events and publications that are not subjective matters or matters of opinion.

In [1], Glover ascribed disagreements over the efficacy of SRI in particular to a focus on and preoccupation with just the technical factors in agricultural innovation. He called attention to a failure to construct holistic understandings of technological systems and their dynamics, asking for more consideration of social and behavioral factors. The remedy suggested was to engage in more thorough, detailed and empirical studies, particularly by preparing technographies. A review of some of the milestones in SRI history suggests, on the other hand, that the controversy did not stem from a lack of empirical evaluation or inadequate conceptualization. Rather, various factors more associated with political economy appear to have played larger roles than did data, or a lack thereof.

1 Introducing SRI

For readers not acquainted with this innovation in question, SRI is a set of agronomic principles and practices that were proposed originally by civil society actors in Madagascar to improve the production of irrigated rice for poor, resource-limited households. Its concepts and methods have now been extended also to upland, unirrigated rice production; to larger-scale, even mechanized production; and even to other crops like wheat, sugarcane and finger millet. So SRI is a protean phenomenon, not easily understood or evaluated with reductionist thinking, regarding SRI as ‘only’ X, or ‘no more than’ Y.

SRI methods have been reported now in almost 50 countries to give higher yield than is achieved with usual rice-growing practices – by changing the management of plants, soil, water and nutrients, and not requiring either new, higher-yielding seeds or agrochemical inputs (http://sri.ciifad.cornell.edu/). The increases are achieved with reduced inputs of seeds and water, and resulting rice plants are less vulnerable to pests and disease losses and to climatic stresses.

The extent to which such improvements are even possible has been vigorously contested by some rice scientists, however, creating the controversy considered in [1]. By now, the evidence on SRI mechanisms and impacts is rather extensive, although still not as complete or as well-documented as most people, including proponents, would like. This was a case where practice has moved ahead of science, rather than where science leads the way.

Many of the reports have come from persons not trained in the language and methods of agronomic science, so these have been easily dismissed by scientists who have what they consider rigorous standards as to what evidence can be considered. At the same time, however, evidence from well-trained, qualified scientists has been similarly discounted and ignored.

Scientific findings on SRI through early 2011 have been summarized in papers in a special issue of the journal Paddy and Water Environment [11]. Research papers submitted from India, China, Indonesia, Thailand and Madagascar were peer-reviewed, complemented by field reports from countries as varied as Afghanistan, the Gambia, India, Indonesia, Iraq, Kenya, Mali, Pakistan and Panama. More information on SRI going back to 2000, much in peer-reviewed journals, is available at this website: http://sri.ciifad.cornell.edu/research/JournalArticles.html.

Glover argues credibly that disagreements about the validity of SRI methods have arisen, at least in part, from disparate ideas about what constitutes SRI, and generally from differing conceptions about the process of agrarian technological innovation. His main proposition – that agricultural innovation is more than technical and encompasses multiple aspects, economic, sociological, cultural and other – is certainly correct. However, such a focus does not account for the sequence of events in this case, or for the vehemence of the discourse.

The controversy over SRI became a cause célèbre in some scientific circles only after 2003 [2,3,12–16], with the innovation being dismissed by some sceptics as based on ‘unconfirmed field observations’ (UFOs) [2,3] and as an example of ‘voodoo science’ [3]. Whereas such language is not unprecedented in exchanges among scientists, its vehemence is incongruous in a debate ostensibly over the validity of data and reported results.

2 SRI originated from empiricism

The subtitle of [1] implied that the controversy has been mostly a matter of opinion and could be resolved by engaging in more empirical evaluations, specifically the kinds of technographic analysis that have been undertaken by Wageningen researchers. However, it should be noted that before there was any publicity of SRI outside of Madagascar, its country of origin, the merits of SRI practices had been documented by several research studies that were thoroughly empirical.

Between 1998 and 2002, half a dozen theses on SRI were researched and written for the Faculty of Agriculture (ESSA) at the national University of Antananarivo, under the supervision of its director of research, the late Prof. Robert Randriamiharisoa. Although these were not PhD theses (few of these were being done in Madagascar at the time) and were written in French (which limits their readership to some extent), the research was done with appropriate scientific methodology, and the main results were reported internationally, as seen below.

The first SRI study in 1998 [17] documented, amongst other things, that SRI methods lead to significantly greater root growth. In replicated trials, SRI practices were found to increase the measured root-pulling resistance (RPR) per plant by more than five times compared with rice plants conventionally grown. This difference was highly significant statistically, and was confirmed in more extensive trials conducted for a Cornell MSc thesis in crop and soil sciences. In 2000 and 2001, two large sets of factorial trials were done to assess the respective components of SRI practice and their collective effects under contrasting tropical and temperate conditions [18,19].¹

Although the ranges of productivity in the two diverse locations were different, as was expected given soil and climatic differences, the pattern of crop response to the management practices that constitute SRI was essentially the same in both. Paddy yields with SRI practices – much younger seedlings (8 days), just one seedling per hill, wider spacing, no continuous flooding, and organic fertilization – were 2–3 times higher than with the use of standard practices – older seedlings (20 days), 3 per hill, closer spacing, continuous flooding, and inorganic fertilization (NPK 16-11-22).

The factorial research design showed the respective as well as collective impacts of the alternative crop, soil, water and nutrient management practices following the recommendations for SRI. The averages for different combinations of factor values were based on six replications rather than the more standard three; the resulting differences were highly significant statistically (p values ranged from 0.021 to 0.000).

•	Under the first set of tropical conditions, SRI yields from a local variety (riz rouge) were 140% higher than with conventional practice, and with a popular ‘modern’ variety (2798), SRI yields were 183% greater.

•	Under more favourable, temperate conditions on the high plateau, using the local variety known as riz rouge with both sets of management methods, SRI yields were 213% higher on loam soils, and 245% higher on better clay soils.

These results were reported to an international conference in China in 2002, hosted by the China National Hybrid Rice Research and Development Center [20]. The results were published also in the proceedings of an international workshop that followed the Sanya conference, organized by a Wageningen University project and held at IRRI in Los Baños [21]. Evidence confirming and helping to explain the SRI results being reported from farmers’ fields was thus not only available to the international rice science community in general, but it was presented in an IRRI publication before the ‘controversy’ began.

Once Chinese scientists learned about SRI, they began doing their own evaluations, although much of their research was published in Chinese language. The majority of articles appeared in peer-reviewed journals, including several international journals [22]. Evidence on SRI could thus have been easily accessed if there was interest in empirical assessments of the claims and counter-claims surrounding SRI. Several of the studies were done at prominent research institutions like Nanjing Agricultural University [23] and the China National Rice Research Institute [24,25] – indeed, the Director-General of CNRRI was a co-author of [25] – so the findings should not have been obscure.²

3 Confirmations of SRI management practices

While research done in Madagascar, reported in French, and in China, most of it in Chinese, could escape the attention of most rice scientists for some time, the subsequent dismissals of SRI that started in 2004 are hard to explain. By this time, SRI had been evaluated and endorsed by two of the best-known rice scientists: Prof. Yuan Long-ping of China, who is regarded internationally as ‘the father of hybrid rice’, and Dr. M.S. Swaminathan of India, hailed as the ‘the father of India's Green Revolution’, and the Director-General of IRRI from 1982 to 1988.

•

In 2000, Prof. Yuan read a paper on SRI from Cornell that had been shared with IRRI scientists in 1999. Yuan found that with SRI methods he could raise the yields of his hybrid varieties by 1–3 t ha−1, with less water and less cost [29]. In 2002 he hosted a first international conference on SRI, to which he invited some 50 Chinese scientists. Fifty international participants also attended, giving reports on SRI experience and results from 15 countries [30]. The proceedings included detailed evaluations by rice scientists from China, India and Madagascar. Yuan's keynote [31] showed that he had understood SRI principles and was able to make effective use of its associated practices.

•

Dr. Swaminathan, after doing his own evaluations of SRI at his research foundation in India, concluded that the methods have merit and started recommending SRI at state and national levels, e.g., [32]. The foundation's annual report for 2005–06 reported a 30% increase in on-farm yield with SRI methods, with a concomitant 18% reduction in the costs of production. The resulting net income per hectare was calculated to be more than 4 times higher with SRI methods than from comparison plots. Also, with SRI practices, the ratio of seeds planted-to-harvested was more than 20 times higher [33].

Such large increases in economic benefits and seed multiplication could hardly be explained away as due to measurement error, although some sceptics nevertheless dismissed SRI results as due to differences in grain moisture content or as attributable to incorrect sampling methods [12]. One might expect 4-fold or 20-fold differences to elicit interest among rice scientists to learn more about SRI from whomever knew the most about it at the time, but this did not happen. Prof. Yuan and Dr. Swaminathan were exceptions to the general response.³ That scientists of their stature and experience would give a ‘thumbs-up’ to SRI, and would recommend its use to others, added certainly to our confidence in SRI methods.

How is it that even after such well-known scientists had satisfied themselves of SRI's merit, other scientists began dismissing it in 2004, maintaining that there was ‘no scientific basis’ for SRI claims? This raises the question of what other reasons might have been operative for rejection of SRI, beyond the explanations that were suggested in [1].

4 Perceived competition for research funding

As the SRI controversy is a complex matter, with many different actors and many different motivations, it is unlikely that any single explanation will be sufficient or complete. However, statements made in the published literature discussed below indicate that opposition to SRI may have been prompted, at least in part, by considerations other than scientific and empirical issues. Several published, peer-reviewed critiques of SRI were dismissive, stating that:

•	SRI is at most “a niche innovation,” suitable only for certain poor soils and more generally just for smallholders, offering no advantages on better soils or for larger operators [13];

•	“SRI has no major role in improving rice production generally” [14]; and

•	SRI does not give results superior to the best management practices (BMP) of rice scientists and “does not fundamentally change the physiological yield potential of rice” [34].

These conclusions can be challenged with contrary evidence, so they are within the realm of normal scientific discourse. Two other critiques, on the other hand, argued that SRI should not even be evaluated, asserting that there was no need to assess the claims of SRI empirically because they contradicted “known principles” governing rice crop performance [2,3]. Such arguments replace evidence with a priori reasoning. The SRI case becomes more interesting when we see that two of the “known principles” proposed as justifications for not even investigating SRI claims can be challenged on empirical grounds as not valid, at least not applying to the more-productive phenotypes of rice that can be produced from given genotypes when SRI methods are used.

4.1 Introducing ‘UFOs’ into the SRI debate

In 2004, when I asked to report in an IRRI publication Rice Today some SRI evidence that had come in from several countries, its editor proposed that I write instead an opinion column, with the stipulation that a rebuttal be included following it. I readily agreed to this condition, not being averse to disagreement and dispute. Dr. Thomas Sinclair (USDA/Florida) was invited to write the rebuttal. This was titled ‘Agronomic UFOs waste valuable scientific resources’ and it began with these sentences. “Discussion of the System of Rice Intensification (SRI) is unfortunate because it implies SRI merits serious consideration. SRI does not deserve such attention.” [2]

Sinclair asserted that no evaluation of SRI was needed or even warranted because SRI reports were based on “unconfirmed field observations (UFOs).” As seen already, this characterization was wrong because data gathered according to standard agronomic methodologies had been presented, making a strong prima facie case for evaluating SRI. Indeed, they had been published in the proceedings from an IRRI-Wageningen workshop! Nevertheless, this assertion that there was no scientific evidence to support SRI claims became a standard justification used by some rice scientists for not giving SRI “serious consideration,” and for not making any systematic efforts to look into others’ success with SRI methods.⁴

Sinclair justified his dismissal of SRI by citing [14], which was itself hardly conclusive [15]. This dismissal of SRI had the effect, however, of putting it outside of ‘mainstream science’ and made it difficult to get support from governments, donor agencies or foundations to do the kinds of evaluation of SRI that were called for. Based on the sceptics’ own methodology, sample construction and analysis, one can point out how their own evaluations of SRI were based on questionable samples and data analysis. It is noteworthy that peer review did not produce more impregnable evaluations.

A. The only empirical evidence in [14] came from three small on-station trials in China. These compared SRI with best management practices (BMP) as determined by local scientists. The SRI trials, however, did not include some of the basic elements of SRI, even though when evaluating claims it is normal practice to follow carefully the protocols on which the claims were based.

•

In the SRI trials, 180–240 kg ha−1 of N fertilizer was applied, even though SRI practice relies primarily on organic fertilization. Inexplicably, this application of inorganic N was 2.5–3 times more than the 80 kg ha−1 of N that an IRRI-supported evaluation [37] had concluded was optimal for Chinese soils. With such excessive N application, it is not surprising that one of the three SRI trials was reported to have lodged.

•

Further, no mechanical weeding (active soil aeration) was done on the SRI plots that would have stimulated the growth and activity of soil biota. This is a key part of SRI theory and methodology. Instead, herbicides were used for weed control in the trials, which would have had more negative than positive effects on the soil biota.

Even though SRI recommendations were not all used or used as specified for SRI, and although one of the three plots lodged, the average yield from the three SRI plots was actually higher than the average yield from the BMP trial plots, by 0.1 t ha⁻¹. This difference was not statistically significant, to be sure, but somehow, this result was reported in a way that implied that – since the SRI yield was not significantly higher than BMP – the trials had confirmed the superiority of scientists’ BMP, even though no evidence was provided to support this inference.

B. The results of a modelling exercise were also presented in [14], which contended that the highest yields reported from Madagascar were not biologically possible because, given day length and mean temperatures where the yields were obtained, there would not have been enough photosynthesis to support such yields. This modelling was done, however, with coefficients derived from rice plants that had been conventionally grown. Such plants would have had degraded root systems because of the hypoxic soil conditions caused by continuous flooding [38].

SRI rice plants, conversely, maintain their root health and function throughout the crop cycle, not suffocating from hypoxia. Larger SRI root systems that resist senescence contribute to greater growth and productivity in the above-ground organs of the plant (see data from Indian evaluations reported below). When this was pointed out to the article co-author who had done the modelling (Dobermann), his response was that the model had not been constructed to deal with roots but only with the process of photosynthesis. This implies that what goes on in plants’ roots has no bearing on what occurs in their canopies.

The controversy engendered over the highest SRI yields reported deflected attention from the not-disputed reports about average increases in SRI productivity. With SRI methods, farmers in Madagascar who had been getting usual yields of 2 t ha⁻¹ were able to achieve average yields of 8 t ha⁻¹, four times more on the same soil, with the same varieties, and using the same methods of measurement to calculate both sets of yields. Such large increases in average yields have been reported also in Cambodia, Indonesia and India [39].

One might anticipate that such large average increases would evoke scientific interest. But consideration of increases in average yield was eclipsed by contention over admittedly outlying maximum yields that were reported with SRI management. These were cited to indicate some phenotypic potentials of rice, and now have also been seen elsewhere [40]. More attention should have been given to averages as these are most relevant for farmers. The same measurement methods were being used to compare SRI yields with whatever was the ‘control’, farmer practice or recommended management practices. Relative differences (ratios) should have been taken seriously even if there was dispute about absolute differences. However, controversy engendered over the latter distracted attention from the former, which was more important.

C. A major shortcoming in [14], which should have been evident to reviewers and readers, was its ignoring the results of dozens of evaluations of SRI already done by Chinese researchers [22]. These rice scientists had reached conclusions mostly different from [14], confirming positive SRI effects on plant phenotypes and on crop productivity. Although most of these findings were published in Chinese language, they could have been at least mentioned in the discussion section of [14] since half of that article's co-authors were Chinese and they should have known and considered the Chinese literature on SRI. There was abundant ‘confirmed’ evidence about SRI.

4.2 Adding the charge of ‘voodoo science’ to the SRI debate

The arguments of [2] were restated in an editorial published that same year in the journal Field Crops Research [3]. This repeated the characterization of SRI as based on ‘UFOs’ and further described SRI as ‘voodoo science.’ While this is a clever phrase, as an ad hominem attack it proves nothing in scientific terms.⁵ The authors of [3] considered it sufficient to justify their dismissal of SRI simply by citing the reported findings of [14], which as has been argued were less than a thorough assessment of the innovation [15].

In [3], Sinclair and Cassman acknowledged that “scientists need to consider ideas from all sources to meet the challenge of increasing crop productivity.” However, they asserted, “these ideas need to be subjected early to critical evaluation for their consistency with known principles governing plant development, growth and yield” (emphasis added). Why “critical evaluation” should be restricted to considering ‘known principles’ and need not be subjected to empirical assessments was not explained. Relying only on ‘known principles’ assumes that our present scientific knowledge is essentially complete and perfect, so that it is unnecessary to consider results from the field. Such a priori reasoning becomes even less tenable if the principles that are invoked to support an argument are themselves vulnerable to empirical challenge.

A. The first ‘known principle’ that Sinclair proposed as making reported SRI yields impossible was: one cannot get higher yield with sparser plant populations because with fewer plants there will not be enough tillers per unit area. “High plant density enhances light interception, growth and yield. SRI suffers from poor light interception because of low plant densities.” [2].

While this sounds plausible, it is not necessarily true. Rice plant phenotypes grown with the suite of practices recommended for SRI are seen to tiller profusely and also to have larger, heavier panicles. More tillers and panicles per plant, plus longer panicles and heavier grains, can more than compensate for the reduced number of plants per unit area. This has been seen in the research done by Dr. Amod Thakur at the Water Management Centre of the Indian Council for Agricultural Research [41]. His studies compared rice plants of the same variety grown either with SRI practices as downloaded from the SRI website (http://sri.ciifad.cornell.edu/) or recommended management practices (RMP) as posted on the website of India's Central Rice Research Institute (CRRI) (http://crri.nic.in/).

Thakur and colleagues conducted their own replicated factorial trials over three years, with no contact with Cornell or CRRI. The comparison showed that rice plants of the same variety (genotype) when grown with SRI management had many morphological and physiological advantages, including:

•	Greater light interception after panicle initiation for SRI plants, 89% vs. 78% for RMP plants, a 15% advantage for SRI.

•	Higher levels of chlorophyll in the leaves compared with those on plants conventionally grown. The fourth leaves of SRI plants had higher chlorophyll levels than the flag (top) leaves of conventional plants.

•	The rate of decline in the leaf chlorophyll between the flowering stage and late-ripening stage was greater in RMP than in SRI leaves, indicating more rapid senescence in RMP plants.

•

While crop growth rate (CGR) was initially higher on an area basis for RMP plants, up to 50–60 days after germination (DAG), after this the SRI plants had superior CGR, rising more than 150%, from 20 g m−2 d−1 to 53 g m−2 d−1 at 60–70 DAG. This was attributable to SRI plants’ higher chlorophyll and reduced senescence. Beyond 50–60 DAG, CGR in the RMP plants was seen to fall by half, from 36 g to 18 g, as leaf senescence set in. This was apparently related to their roots’ degeneration with continuous flooding.

•	By the early-ripening stage, root depth of SRI plants was 33.5 cm vs. 20.6 cm for RMP roots, and root volume on an area basis (ml m−2) was 40% more for SRI plants [41].

Thus, even though the SRI plant population per square metre in these trials was much lower than with RMP, on an area basis the SRI plants gave 42% more yield than did six times as many BMP plants. The number of tillers per unit area under SRI management was slightly higher (450.1 vs. 442.1), but not significantly higher. But many other differences measured were statistically significant: greater root growth; higher rate of panicle formation (effective tillering), longer panicles with more branching, leading to higher numbers of grains per panicle; accelerating crop growth rate (CGR) after panicle initiation, as noted above; more dry matter accumulation, both above and below ground; larger leaf area with more favourable leaf inclination and canopy angle, which led to more light interception, which interacted with higher chlorophyll levels to give a higher photosynthesis rate; all occurring with delayed senescence of roots and canopy [41].

These factors contributed respectively and collectively to higher crop yield and greater productivity of land, labour, capital and water. Ironically, this research by Thakur was prompted by his reading of [14] and considering its conclusion insufficiently supported (pers. comm.). This led him to undertake his own evaluation of SRI, comparing it with the practices being recommended by Indian rice scientists.

B. Another ‘known principle’ asserted was that rice grown in flooded soils performs better than rice raised in only moist soils. Sinclair wrote: “The physiology and physics of plant water use have been researched for more than 300 years, and the relationship between growth and plant water use is unambiguous. Ample water maximizes rice yields, and flooded paddy fields assure that no water limitations develop.” [2].

This view reflects the long-standing preference for growing rice in flooded fields. A former head of IRRI's agronomy department wrote in his textbook on rice: “Rice… thrives on land that is water saturated, or even submerged, during part or all of its growth cycle… Most rice varieties maintain better growth and produce higher yields when grown in a flooded soil.” [42].

IRRI has been moving away from this position in recent years, however, ever since SRI emerged on the scene, and water scarcity has become a more serious constraint for rice production. It has been evaluating and promoting a water management strategy called ‘alternate wetting and drying’ (AWD) and trying to develop rice varieties more suitable for aerobic cultivation [43]. Few if any rice scientists will any longer affirm this ‘known principle’ which was asserted by Sinclair and agreed to by Cassman as recently as 2004. Chinese researchers have shown clearly that better paddy yields can be achieved with existing varieties by applying less water, thereby raising both water productivity and the productivity of nitrogen fertilizer [26,44].

C. Sinclair also stated categorically that SRI promotes “organic fertilizer to the exclusion of inorganic fertilizer” [2], which is a stereotyped view but incorrect. The message given to farmers is that with SRI management, they do not need to rely on inorganic fertilizer to get higher yields, which is true. The SRI website says that inorganic fertilizer can be used with the other SRI practices, but also that such fertilizer does less to improve the soil's structure and functioning than will organic fertilization, because the latter supports the soil biota better. With organic soil amendments, using rice straw or any other biomass, farmers can get as good or better results or at least lower-cost results. SRI recommendations have not excluded the use of inorganic fertilizer, although some SRI proponents strongly favour organic cultivation.

The factorial trials cited already showed that inorganic fertilizer together with the other SRI practices can raise yields substantially [10,11]. At the same time, they supported the SRI advice to rely as much as possible on organic fertilization. The factorial trials reported in Section 2 showed a large response to inorganic fertilizer when it was used with SRI methods – 68% higher yield than with no fertilizer. But when SRI methods were used with compost instead of inorganic fertilizer, the yield was even higher, by 7.3%. Farmers interested in SRI are told that to the extent they enhance their soil's organic matter, this will result in better crop performance. This is hardly a controversial statement, although it is probably better understood and more widely accepted now than it was in 2004.

A relevant question is: Why would scientists who were sceptical about SRI want to persuade other scientists and donor agencies not to even evaluate SRI effects? This question is more salient if two of the ‘known principles’ cited as reasons for not investigating SRI are not themselves valid, when the full set of SRI practices are used together and achieve more productive rice phenotypes. And why was SRI rejected with such vehemence? Deprecating language such as ‘UFOs’ and ‘voodoo science’ is not unknown in scientific arguments, but this does not make it acceptable or helpful for resolving disagreements.

The arguments made against SRI and against its being evaluated suggest that there was some apprehension that ‘taking SRI seriously’ could lead to reduced funding for research, including possibly for that of the critics. While their wording is carefully couched, the operative argument in both [2] and [3] was that no funds should be allocated even for assessing SRI because based on ‘known principles’ this would be a ‘waste’ of resources.

4.3 Direct competition for research funding?

The most outspoken critic of SRI at IRRI has been, until he retired in 2009, John Sheehy, the project leader for IRRI's initiative to bioengineer rice plants from a C₃ to a C₄ pathway for photosynthesis [14,45,46]. This project to modify the biochemistry of rice plants’ pathway for photosynthesis and their stomatal functioning is a complex one, initially budgeted at $5 million a year, for which the Bill and Melinda Gates Foundation as the major donor is providing three-fourths of the funding.

If the reported benefits from SRI's alternative crop management methods were accepted as real, the justification for substantial donor funding for this project could be considerably diminished. Not only are yield levels increased with SRI – by more than the 30–50%, which is claimed for C₄ transformation – but seed, water and N fertilizer requirements could be reduced, which has environmental and other benefits. In particular, there is some evidence that with SRI management, the rice plants’ internal water use efficiency is increased, a major claim in favour of achieving C₄ photosynthesis.⁶

This and other SRI benefits are directly competitive with the claims made for investing in the transformation of rice from a C₃ to C₄ photosynthetic pathway. There is no way to establish that criticisms of SRI were motivated by intention to protect funding for a C₃–C₄ genetic modification of rice. But I and others know that when I made a seminar presentation on SRI at IRRI in March 2003, Sheehy raised the most aggressive objections, saying that I should not have been allowed to make such a presentation at IRRI, and calling SRI “no better than alchemy”. All we have in writing, however, is the published arguments considered above that SRI claims should not even be evaluated.

5 Discussion

Since the controversy erupted in 2004, the number of countries where SRI merits have been demonstrated has grown from 22 to 46 (http://sri.ciifad.cornell.edu/). Farmers and professionals working with SRI have seen that they can get more productive phenotypical expression of the genetic potential in their rice plants by changing their growing environment.

Governments in India, Indonesia, China, Vietnam and Cambodia, where two-thirds of the world's rice is grown, have begun promoting the spread of SRI methods, perhaps as much for water saving as for yield improvement. Farmers’ interest in adopting the methods is often motivated by opportunities for reducing their costs of production and enhancing income, and often even for saving labour once they have mastered the methods. Environmental quality benefits have not been systematically evaluated, such as net reduction in greenhouse gas emissions, but farmers and countries should be considering all of these aspects of crop performance, including but not limited to yield. Scientific understanding of how these beneficial effects are achieved has been growing substantially [41,47–50].

This comment was written to throw more light on the issue raised by Glover: how to account for the opposition that SRI has encountered from some parts of the rice science community. Some of this may be ‘normal’ resistance to a paradigm shift [51]. Opposition to new ideas and new approaches is not unique to SRI. Certainly part of the explanation could derive from the kind of conflicting conceptualizations that Glover discussed. But such explanations are quite abstract.

Resistance to SRI could be attributed, more personally, to the ways in which SRI has been presented by those like myself who have proposed that it be evaluated, by researchers and by farmers. My own style of speaking and writing has been objected to by some critics. Certainly, I do not communicate like an agronomist, since my formal training is in the social sciences. However, neither Prof. Yuan nor Dr. Swaminathan had difficulty in grasping SRI ideas and proceeding to evaluate them after learning about SRI from written materials and from face-to-face discussions. Arriving at tenable answers to questions concerning resistance to SRI pretty surely needs to reach beyond personal factors and to address structures and incentives that reflect a variety of institutional, financial and other influences.

The science and practice of SRI are still a work in progress, largely promoted and tested by farmers, civil society organizations, and some scientific researchers, although increasingly by extension leaders and government agencies that share an interest in less input-dependent, more climate-resilient agricultural development. The agricultural science community could benefit, I think, from reflecting on the SRI controversy in more detail and from more perspectives. Others who have more detachment than I should study this as a case to consider how agricultural science can progress more rapidly and more accurately, without the friction and obstacles that occur often in the advancement of scientific knowledge and its application, not just with SRI.

Thought should be given to how the scientific profession, which has gotten reasonably good at ferreting out and minimizing Type I errors, i.e., false positives – wrong claims that something is true – can structure incentives and also penalties that minimize Type II errors, i.e., false negatives – incorrect conclusions that something is false. In this case, it can be argued that the controversy has delayed millions of needy families around the world from getting access to new knowledge that would enable them to produce more food and income from the limited resources that they control. If IRRI, Cornell and other scientists had co-operated in the investigation of SRI when it was first introduced outside of Madagascar a dozen years ago, contemporary agricultural history might have been different and better, especially since SRI concepts and methods are now being extended to other crops beyond rice, such as wheat and sugarcane [52,53].

The full extent of benefit from these modified concepts and practices for agricultural cropping is not yet known. It remains possible that SRI will not fulfill the expectations that I and others have articulated, based on our experience and observations and on a growing number of published scientific evaluations. Empirical matters that have potential to benefit humankind and the environment deserve thorough and systematic study, which means that Type II errors should be rigorously challenged and minimized.

Trying to prevent proposed beneficial innovations from being evaluated is hard to justify. Those who are involved in the phenomena of SRI and related agroecological innovations remain ever ready to co-operate with anyone wanting to assemble evidence and make more satisfactory sense out of this controversy. We are seeing in a good many countries that understanding SRI principles and practice can contribute to a better-fed and more prosperous world.

Notes

1 The 2000 trials [N = 288] were done on the west coast of Madagascar at Morandava: sea level, tropical climate, poor sandy soils; the 2001 trials [N = 240] were done on the country's central high plateau at Anjomakely: 1200 m asl, temperate climate, better loamy and clay soils. The trials were done on 2.5 m × 2.5 m plots, laid out with split-block design with the flooded and unflooded plots kept separate (to avoid inter-plot water flow) and then with random block design for the other five factors evaluated. There were three replications of all the combinatorial treatments. In addition to assessing yield effects as the dependent variable to be explained, each plot was sampled for tiller number, panicle number, grains per panicle, root density, and root length, making the evaluations quite thorough.

2 Note that another article on SRI of which the Director-General was a co-author [26] was rejected for publication by the journal Field Crops Research, without any suggestions for revision and resubmission. FCR had previously accepted and published two articles [27,28] by most of the same authors employing the same methods, with trials even done on the same experiment station. The main difference between the articles rejected and accepted was that [26] evaluated SRI practices, while [27] and [28] did not.

3 After informing themselves about SRI from available materials, both sought to get more information on the methods and their results. In December 2000, Prof. Yuan invited me to visit the Sanya experiment station of his China National Hybrid Rice Research and Development Centre on Hainan Island to see his SRI trials. This I was able to do in April 2001, and I then visited his Center headquarters at Changsha in August that year. Dr. Swaminathan invited me to make presentations on SRI at the M.S. Swaminathan Research Foundation in Chennai, India, in May and December 2002. Both listened carefully to what I could tell them and their fellow scientists about SRI. Indeed, Prof. Yuan himself translated both of my presentations to his staff into Chinese language, so that he paid attention to every word. After each of these talks, I asked my hosts if there was anything in my presentation on SRI that I should change or should correct. I told both of them that I knew they had more knowledge of rice science than I did. I anticipated some criticisms and looked forward to their corrections. But each time I was told that the presentation had been fine, and that nothing needed to be changed – although each said that he found the results that I was reporting rather hard to accept – until they tried the methods themselves and saw the effect.

4 When IRRI did its own on-station trials of SRI methods in 2002, the yield from this first-year trial was just 1.4 t ha⁻¹, a fraction of what farmers in the Philippines were getting with SRI methods on their own fields, averaging 6 t ha⁻¹ [35]. How such different results could have been obtained was never explained. The next year, the SRI yield in a set of comparative management trials was just 3 t ha⁻¹. But in these trials, IRR's best management practices yielded only 4 t ha⁻¹ [36], no more than the national Philippine average. Not only in these trials but also in Nepal and India, we have seen experimenters getting lower SRI yields than farmers obtain with these methods on their own farms. This is an anomaly worth investigating.

5 Readers of the editorial did not know that, four years before, one of its authors had been invited to co-operate with us at Cornell in doing empirical evaluations of SRI and had declined. In June 2000, while Cassman was attending a conference on rice at Cornell University, I asked him to help us evaluate and better understand SRI, knowing that he was one of the most knowledgeable rice agronomists in the USA. With SRI methods we were seeing rice yields in Madagascar increase by multiples rather than increments. After learning that SRI management involved six changes in crop management, Cassman responded that SRI has “too many variables” to be evaluated easily in factorial trials with replications. Fortunately for us, agriculture honours students at the University of Antananarivo were not deterred by the large number of plots needed to evaluate all of the combinatorial treatments, using standard agronomic methods of random block design, replications, etc. [10,11]. Their research and results were summarized in Section 2 above.

6 The evaluation research done at the Water Management Centre in India cited above found that the same variety of rice when grown with SRI methods fixed 3.6 μmol of CO₂ per millimol of water transpired, compared with 1.6 μmol of CO₂ fixed (into photosynthate) by rice plants conventionally managed [41]. This represents a doubling in the productivity of water used within plants.