On-farm management of rice genetic diversity: understanding farmers’ knowledge on rice ecosystems and varietal deployment

Introduction

The scientific community has come to realize the complementary nature of ex situ and in situ conservation methods in addressing different aspects of conservation of genetic resources (Brush 2000). As a result, there is an increased interest in understanding the functioning of in situ conservation in relation to agrobiodiversity. In situ conservation for local crop diversity effectively implies on-farm management, which has been defined as “farmers’ continued cultivation and management of a diverse set of crop populations in the agro-ecosystems where the crops evolved” (Bellon et al. 1997). Since in situ conservation of agrobiodiversity on-farm is a relatively new subject, limited experience exists concerning its management, both at national and international levels. There are gaps in understanding of the subject matter and also of the methods, tools and techniques, and institutional settings for conserving and utilizing agrobiodiversity for the benefit of farmers (Wood and Lenne 1997). With the implementation of an IPGRI (now Bioversity International) coordinated global project on ‘Understanding the Scientific Basis of In Situ Conservation of Agrobiodiversity On-farm’ in nine different countries, significant progress has been made in developing tools and techniques for conserving, monitoring and utilizing agrobiodiversity on-farm.

The rationale for studying farmers’ local knowledge of ecosystems and varietal (landraces and modern varieties (MV)) deployment rests on the tenet that agrobiodiversity and local knowledge held by farmers are two faces of the same coin (Rajasekaran and Warren 1994). Therefore, it is vital for researchers to document and analyse farmers’ local knowledge, not only in terms of ‘what’ and ‘how’, but also ‘why’, so that the blending of scientific knowledge and local knowledge is achieved for strengthening farmers’ capacity to continue growing landraces and to contribute to on-farm conservation. The present article deals with the methodological development of a rice (Oryza sativa L.) ecosystem classification and varietal deployment tool, and explains how the exercise contributes to on-farm conservation, varietal deployment and the monitoring of varietal dynamics on-farm over time by practitioners in the field.

Methodology

The methodology adopted for the study comprised Participatory Rural Appraisal (PRA) tools—Focus Group Discussions (FGD), transect walks, and direct observations—that were employed to elicit farmers’ knowledge on rice ecosystems, varieties and their deployment in different ecosystems. The study was conducted in two contrasting ecosites: Begnas village, representing mid-hill (600–1400 masl) conditions characterized by limited intervention from research and extension agencies, with medium level of road and market access; and Kachorwa village, representing plain (<100 masl) conditions, with high intervention from research and extension agencies, and good road and market access.

FGDs conducted at site level were specifically used to identify different criteria used by farmers to delineate rice ecosystems. During FGDs, farmers were asked to list criteria they use to identify different ecosystems for rice in their locality. Once the group agreed with the criteria, then they classified and characterized different ecosystems based on the agreed criteria. After that the farmers discussed ‘how’ and ‘why’ they deployed different landraces and MVs in particular ecosystems, based on their experience and knowledge of the ecosystems and varietal performance. The FGD exercise was followed by a transect walk by a combined team of farmers and researchers to visit rice fields to verify on the ground the deployment of varieties to different ecosystems.

Findings

Farmers indigenous method of characterization of rice ecosystems
Farmers employ multiple criteria to characterize rice ecosystems (Tables 1 and 2). During the study, the farmers identified eight different criteria to classify their rice ecosystems in the village. When characterizing rice ecosystems, the farmers considered physical properties of soil such as texture, colour and presence of pebbles. Ability to retain water (after rain or irrigation) for a longer period is an important criterion. The farmers also considered ease of ploughing and weeding. Finally, they considered attributes that directly contribute to crop yield, such as inherent fertility status of soil, as well as productivity potential influenced by human-managed factors (application of compost/farmyard manure, chemical fertilizers and irrigation).

Soils at the study sites were generally clay for marshy lands, loamy to clay in irrigated lands, sandy loam to clay for rainfed, and sandy loam to loam for upland. The colour varied, with red soil in upland, black in irrigated and marshy lands, and grey or yellowish white in rainfed land. Water holding capacity refers to the duration one can observe water on the plots after rain or irrigation stops, and this trait is very important in distinguishing rice ecosystems. Different ecosystems within and between sites vary greatly for this trait, ranging from a few hours to whole seasons. Upland has the least capacity to hold water due to soil type, presence of pebbles and lack of bunds along the edge of the field, whereas marshy or pond fields hold water beyond the rice growing season. The amount of pebbles present in upland condition explains the better drainage of water, which implies that rice in upland suffers more frequently from moisture stress than in other categories of rice ecosystem. In general, ease of ploughing and ease of intercultural operations are important traits associated with irrigated land. Ploughing is most difficult in marshy land because of heavy (clay) soil with plenty of moisture, but weeding is easy as weeds are generally suppressed due to standing water throughout the rice growing period.

Apart from water holding capacity, fertility status and production potential of plots are the major criteria farmers employ when classifying their rice ecosystems. Across the sites there was agreement that inherent fertility of soil is highest for marshy plots (pond in Kachorwa) and lowest for rainfed plots. In terms of productivity potential, the farmers identified the irrigated ecosystem as having the highest potential, followed by marshy ecosystem in Begnas, but rainfed in Kachorwa. The rainfed ecosystem had the least production potential for rice in Begnas, whereas the deep-water ecosystem was the least productive for Kachorwa. The farmers estimated that about 65 to 70% of the total rice area in Begnas was irrigated (temporary), followed by rainfed lowland (20–28%), while marshy areas and upland accounted for 5 to 7% and 2 to 3%, respectively, of the total rice area. Similarly, for Kachorwa, 50% of the rice fields were irrigated, 25% rainfed, 20% marshy and the remaining 5% being a deep-water ecosystem.

Deployment of varieties across ecosystems
After characterization of ecosystems, the farmers engaged themselves in deploying varieties to different ecosystems. Because of the numerous varieties to deal with, this process took much longer than expected at both ecosites. Moreover, there were several varieties that could be placed in more than one ecosystem, which also complicated the exercise. The list of varieties (landraces and MVs) currently grown at the study sites was first developed. Varieties were then placed in extreme (marginal) ecosystems, and the exercise continued for more favourable ecosystems. The varieties deployed in different ecosystems for Begnas and Kachorwa are presented in Tables 3 and 4 respectively.

Analysis of varieties across ecosystems at Begnas indicated that there was no MV in rainfed upland condition, though numerous landraces existed for the ecosystem. In other words, available MVs were less competitive than the existing landraces in rainfed upland conditions. Under rainfed conditions, there was a dearth of both landraces and MVs. Only three landraces were prevalent in this ecosystem, though it occupied 20–28% of the rice area. In the irrigated ecosystems there were plenty of varieties (options) for the farmers, with over 25 different landraces and 4 MVs available. Similarly, in marshy ecosystems, a number of landraces and MVs compete. As a result, there is competition between varieties for space in favourable ecosystems, whereas in marginal environments the farmers face a lack of varieties.

Findings from Kachorwa also suggest that only a limited number of varieties exist for extreme conditions, whereas plenty of options exist for favourable conditions. There is not a single MV suitable for deep-water ecosystems and only a limited number of landraces exist for such conditions. In contrast, there are many landraces and MVs competing for space in more favourable environments. As in Begnas, farmers have a greater choice of varieties in favourable compared with marginal ecosystems. From both study sites, it could be seen that MVs are confined to more favourable ecosystems, and the number of landraces in marginal conditions is much lower, reflecting the problem of adaptation of the majority of these varieties. Farmers in marginal ecosystems have limited choice of genetic materials at their disposal, which greatly reduces their capacity to manipulate the production systems in such environments.

Discussion

Concept of ecosystems and variety deployment

Key informants in any given community can provide very reliable information on ecosystems for rice. Similarly, farmers can provide detailed features of each ecosystem in terms of soil type, irrigation and drainage, fertility status, ease of cultivation, production potential, cropping patterns, relative coverage of area under different ecosystems and so forth.

Based on analysis of the characteristics of different ecosystems across sites, and the distribution of landraces and MVs within and between ecosystems, an attempt has been made to develop a generalized model for rice ecosystems (Figure 1). In the following subsections, the characteristic features of the ecosystems are explained. Finally, we look at the implications of the exercise for conservation, variety deployment and monitoring of diversity on-farm.

Number and size of ecosystems
The number of ecosystems varies from place to place as conditions differ between sites. Ecosystem size also varies, with more extreme ecosystems being relatively smaller than more favourable ones, and, depending upon the geographical location (high-potential production systems versus marginal growing environments), the size of each ecosystem will vary. For instance, in marginal environments for rice, the extreme ecosystems will be relatively larger than other ecosystems, whereas in favourable environments the productive ecosystems will be relatively larger.
Landrace and MV distribution across ecosystems
Understanding the distribution of landraces and MVs across ecosystems, the features of those ecosystems, and the particular traits of varieties are fundamental to appreciating the complexity of farmers’ strategies to manage plant genetic resources to meet their multiple needs. The analysis clearly indicates that one landrace or MV is ‘best suited’ or most competitive in only one ecosystem, though farmers might grow the same variety in more than one ecosystem. This implies that a variety competes with other varieties from within the ecosystem, and that there is less competition between varieties across ecosystems, except when there is an overlap of varieties. This implies need for situation-specific participatory crop improvement programmes. Overlap signifies the presence of transitional zones between ecosystems, which explains the presence of landraces and MVs in two different but adjacent ecosystems. Within ecosystems, farmers’ socio-economic status, market forces, cultural factors, preference for specific traits, and abiotic and biotic factors explain the area and number of households growing the different landraces and MVs. Depending on the area coverage and number of households growing the variety, it could fall in any of four categories: many households + large area; many households + small area; few households + large area; and few households + small area.
 
Although landraces and MVs may overlap in adjacent ecosystems, no case was found where a landrace or MV was found in more than two ecosystems. This suggests that varieties have specific adaptations. It also reinforces the idea that a variety is most competitive in only one ecosystem. The findings suggest that competition for space between varieties primarily occurs within the ecosystem, and there is a lot of competition in favourable ecosystems compared with marginal ecosystems, owing to the presence of more varieties in the former. This shows that the landraces grown in marginal areas are less prone to genetic erosion than are those grown under favourable conditions. Favourable market prices or inherent quality traits in certain varieties may attract farmers to grow these varieties beyond their ‘best-fit’ ecosystems, but such cases would be few.

Agromorphological characterization work by Bajracharya (2003) on rice landraces collected from the study sites revealed that principal component analysis (PCA) outputs grouped landraces according to ecosystems for Begnas and Kachorwa ecosites. Varieties falling within the same ecosystem are more likely to be similar in their genetic composition than are varieties from dissimilar ecosystems, as landraces have been conditioned over time by their continued cultivation and selection in specific ecosystems. Thus, they have developed adaptive traits that are unique to landraces of that ecosystem.

Farmers’ knowledge on rice ecosystems and variety deployment
Field exercises at both the study sites have shown that farmers have intricate knowledge and understanding of rice ecosystems. They employ multiple criteria to classify and characterize the ecosystems, with the most prominent criteria being the availability of moisture in the soil and fertility status of soil. Catalan and Perez (2000) reported a similar finding while analysing the conservation and utilization of biodiversity in Mapuche communities in Chile. A high degree of similarity between farmers’ and scientists’ classification of rice ecosystems (Khush 1984) suggests that there exist commonalities in farmers’ knowledge and scientific knowledge, which gives more reason to better understand and appreciate each knowledge system for synergistic effect.

Farmers use different varieties to manage ecological diversity present on their farm. Kieft (2001) reported that rice farmers in Timor use specific varieties for specific locations, and the decision on which variety to plant is very much based on the forecasts for the next rainy season. The process of allocating different varieties to different ecosystems affects the genetic diversity of farmers’ repertoire of varieties maintained on-farm (Cleveland and Soleri 2002). Analysis of varietal allocation across different ecosystems indicates that diversity of varieties is directly associated with diversity in ecosystems because of the specificity of most varieties. The absence of a specific ecosystem in a locality results in absence of varieties associated with that ecosystem. The finding also implies that changes in specific land use systems and practices might threaten the survival of some landraces as the landrace best suited to such ecosystems can be eroded due to lack of benign environment. This has been shown statistically significant in multiple regression models, with the number of varieties maintained on-farm directly corresponding to diversity in rice ecosystems (Rana 2004).

Implications
The distribution of varieties in different ecosystems is the result of farmers’ experimentation with those varieties over the years. In other words, based on farmers’ judgement, the varieties are the ‘best fit’ under the farmers’ particular management conditions. Therefore, in order to make any intervention in the present system, researchers definitely need to know the characteristics of each ecosystem, as well as the specific traits of landraces and MVs in each ecosystem and their distribution across ecosystems.

On-farm conservation
Diversity analysis of varieties at aggregated community or landscape level ignores the influence of ecosystems in determining the position of varieties in different ecosystems. The exercise also ignores the variety × ecosystems interaction and the specificity of landraces to ecosystems. Thus, selecting landraces from an aggregated list of varieties based on weighted diversity might exclude certain strategically important landraces for conservation. Hence, there is a need for micro-level analysis of rice varietal distribution across ecosystems in order to target better conservation of specific landraces per se that might harbour genes of important traits in certain ecosystems. Another means to conserve useful genes in landraces would be through a Participatory Plant Breeding (PPB) approach, where landraces are crossed with MVs and offspring selected by farmers in the target environment (Sthapit et al. 2002). Understanding the features of ecosystems and the distribution of landraces within them will facilitate decision-making about selecting landraces for conservation. Failing to do so could result in selecting landraces with similar genetic traits for conservation (via PPB) from limited ecosystems. This would lead to the neglect of some and over-representation of others in conservation efforts.

Diversity deployment
Diversity deployment means providing farmers with options of genetic material to choose from. The introduction of new genetic material results in temporal disequilibrium because of competition between existing and new material for space in farmers’ fields, for farm labour, for capital inputs, and so forth. As time elapses, the new entrant finds its rightful place in the given environment. This is the outcome of farmers constantly trying to maintain equilibrium while meeting their multiple objectives over time. Hence, the strategy for diversity deployment ought to begin by analysing the distribution of varieties across ecosystems. Once this is done, researchers would have a clear picture of each ecosystem, along with the distribution of landraces and MVs, and the dominance of certain varieties against others and associated reasons for dominance. In the absence of this information, any diversity deployment strategy would be precarious, and new genetic material might not fit into ecosystems where it is intended to expand farmers’ choice of varieties. It could also happen that new genetic materials might compete with each other in similar ecosystems, resulting in limited impact of diversity deployment. An equally important aspect of diversity deployment is the repatriation of promising landraces, thereby increasing the chances of survival of these landraces on-farm. For repatriation of landraces to succeed, a thorough understanding needs to be established of specific ecosystems where they could fit well and compete with others.

Monitoring varietal diversity on-farm
In order to make decisions on conservation of landraces on-farm (or ex situ in genebanks) and diversity deployment at different scales, information on the varietal landscape over time is needed to observe the trends of varietal dynamics at ecosystem level. Monitoring varietal diversity on-farm is important because of its dynamic nature, with introduction of new genetic material and discarding material that no longer meets farmers’ changing needs. Therefore, monitoring of introduction and replacement of varieties at ecosystem level gives a better indication of where the introduced variety is going to fit in, and which varieties are likely to be replaced in the process.

Conclusion

Delineation of rice ecosystems can be reliably done using focus group discussion with knowledgeable farmers from the given community. The ecosystems identified and the associated varieties in each ecosystem have to be verified through transect walks. Deployment of varieties in each ecosystem is the outcome of intimate understanding of ecosystem characteristics and varietal performance, and the interaction between the two, which represent the ‘best fit’ under farmers’ given circumstances. Because of the specificity of varieties to ecosystems, aggregated information at community or landscape level may not be very useful for making decisions pertaining to landrace conservation on-farm, diversity deployment or repatriation programmes. Ecosystem characterization and varietal deployment would contribute significantly in making conservation decisions, as well as to monitor varietal dynamics over time. For wider utility, the technique needs testing in diverse settings for different crops.

Acknowledgement

Authors are highly indebted to farmers of Begnas and Kachorwa for sharing their insights on rice ecosystems and varietal distribution across ecosystems. This work is an output of a PhD Study under the IPGRI [now Bioversity International] Global Project: Strengthening Scientific Basis of In-situ Conservation of Agrobiodiversity On-farm: Nepal Country Component funded by the Netherlands Ministry of Foreign Affairs Development Cooperation DGIS (Activity number: ww104801), and IDRC and SDC.

References

Bajracharya J. 2003. Genetic diversity study in landraces of rice (Oryza sativa L.) by agro-morphological characters and microsatellite DNA markers. PhD Thesis, School of Agriculture and Forest Sciences, University of Wales, Bangor, UK.

Bellon MR, Pham JL, Jackson MT. 1997. Genetic conservation: a role for rice farmers. In: N Maxted, BV Ford-Lloyd and J Hawkes (editors). Plant conservation: the in situ approach. Chapman and Hall, London, UK.

Brush SB. 2000. The issues of in situ conservation of crop genetic resources. In: SB Brush (editor). Genes in the field: on-farm conservation of crop diversity. IPGRI, Rome, Italy, IDRC, Ottawa, Canada, and Lewis Publishers, USA. pp. 3–26.

Catalan R, Perez I. 2000. The conservation and use of biodiversity by Mapuche communities in Chile. In: C Almekinders and W de Boef (editors). Encouraging diversity: the conservation and development of plant genetic resources. Intermediate Technology Publications, UK. pp. 60–66.

Cleveland DA, Soleri D. 2002. Farmers, scientists and plant breeding: knowledge, practice and the possibilities for collaboration. In: DA Cleveland and D Soleri (editors). Farmers, scientists and plant breeding: integrating knowledge and practice. CABI Publishing, UK. pp. 1–18.

Khush GS. 1984. Terminology for rice-growing environments. IIRRI, Los Banos, the Philippines.

Kieft J. 2001. Indigenous variety development in food crops strategies on Timor: their relevance for in situ biodiversity conservation and food security. Indigenous Knowledge and Development Monitor, July 2001.

Rajasekaran B, Warren DM. 1994. IK for socio-economic development and biodiversity conservation: The Kolli Hills. Indigenous Knowledge and Development Monitor 2(2).

Rana R. 2004. Influence of socio-economic and cultural factors on agrobiodiversity conservation on-farm in Nepal. PhD Thesis, University of Reading, UK.

Sthapit B, Joshi K, Gyawali S, Subedi A, Shreatha K, Chaudhary P, Rana R, Rijal D, Upadhyaya M, Jarvis D. 2002. Participatory plant breeding: setting breeding goals and choosing parents for on-farm conservation. In: MR Bellon and J Reeves (editors). Quantitative analysis of data from participatory methods in plant breeding. CIMMYT, Mexico. pp. 104–112.

Wood D, Lenne JM. 1997. The conservation of agrobiodiversity on-farm: questioning the emerging paradigm. Biodiversity and Conservation 6:109–129.

Contact PGRN-manuscripts@cgiar.org about this page