Document(s) 20 de 38

Abstract: The paper presents results of a five-year program (1988-93) on participatory selection of beans (Phaseolus vulgaris L.) in Rwanda. It looks at the technical and social challenges of integrating farmers into on-station selection as well as issues in setting up a country-wide program on decentralized selection in community plots. Choice of farmers, trial design, and evaluation procedure can affect the technical findings but also influence the potential to institutionalize participatory selection procedures on a broad scale. Some of the trade-offs between a research-focused v. a development-focused participatory selection program are highlighted. Finally, the paper discusses participatory selection in the light of the recent and wide-scale civil disruptions in Rwanda. Farmer-centered methods are being used to evaluate possible varietal and genetic erosion, and participatory selection has been proposed as a major means for reintroducing landrace material to Rwandan communities.

Introduction: beans and bean expertise in Rwanda

The centrality of beans for nutrition is matched by their key role in agriculture. Grown by 95% of farmers, in all major regions of the country (from 1000-2200 meters), beans are sown two, sometimes three seasons a year. A third remarkable aspect lies in their genetic diversity with Rwanda providing one the most varied and vibrant bean varietal pools in the world. At least 550 local varieties are found countrywide, with important and unique types having evolved from both the MesoAmerican and Andean genepools (Scheidegger in CIAT 1993, and S. Beebe, personal communication). Households manage varietal mixtures of up to 30 components (Lamb and Hardman 1985, Voss 1992), altering blends according to different soil conditions, crop associations and seasons. Such mixture use encourages production stability as well as utilization of the country's highly diverse production niches.

Yet, while most Rwandan farmers need and directly grow beans, have been exposed to very diverse materials, and manage complex mixes, they were for many years at the fringes of the research system. The selection sequence of the Institut des Sciences Agronomiques du Rwanda (ISAR), paralleling western models, sought farmer feedback at the very last stages, in on-farm trials, if at all. Further, farmers were offered but 2-5 options — the tip of a selection funnel originally numbering some 200 entries. Follow-up surveys in 1988, showed ISAR with some laudable bean successes — but short of what could be expected in a country where a Rwanda farmer on her own may test 75-100 varieties in a lifetime (Sperling 1992).

In was in the spirit of improving performance for highly heterogeneous production environments that ISAR and International Center for Tropical Agriculture (CIAT) researchers took the first steps towards a participatory selection program in 1988. Two key questions shaped inquiry. Was there "untapped potential"? — that is, could farmers absorb and productively use a much greater range of cultivars than that currently delivered by the formal research system? Second, could breeders and farmers, working together, achieve important gains: could they target more environments, faster, and more productively?

Overview of the participatory selection program

Phase I: 1988-1990

The first phase of the participatory selection, although collaborative, remain very much research-oriented. The major conclusions are summarized in Box 1.

Box 1: Participatory selection with Rwanda bean farmers: technical results: Phase I, 1988-1990 Communities recognize differing expertise in varietal selection. These go beyond the frequently-cited divisions of gender and age. Some women are known for astutely distinguishing among varieties for particular farming contexts. Farmers' varietal criteria overlap with breeder concerns but also contain 'composite' traits. These composites, which represent combinations of features, help determine actual performance on-farm and are hard for formal breeders to anticipate. Farmers can target from station to on-farm plots, both meeting their own agronomic and socio-economic criteria as well as achieving production gains. Farmers are ready to use a wide diversity/number of cultivars.

Phase II: 1990-1993

The format: On-station, researchers wondered whether farmers could be brought a stage earlier, 5-7 seasons before normal on-farm testing. This also implied that farmers would be screening many more lines. Was there a limit on what farmers could handle?

For the three years, farmers viewed a trial normally containing about 80 lines. To minimize risk, the CIAT pathologist screened this trial earlier than usual and eliminated the most disease susceptible entries (to anthracnose, aschochyta, bean common mosaic virus and rust). So in fact, farmers screened what researchers felt was the 'largest possible reduced risk pool', some 79, 41 and 43 lines in 1990, 1991, and 1992 respectively. Bringing farmers in this early amounts to what might be termed 'prototype screening' and in any such premature collaboration, researchers should make special efforts to anticipate risks which farmers cannot.

In terms of broadening the program on-farm, the concerns of Phase II focused on how to encourage communities to select their own expert representatives and how to evolve much of the on-farm testing to where it belongs — communities themselves. The move towards 'devolution' was a healthy mixture of empowerment and economics. Communities should have the right to select their own delegates to screen on-station. Communities should also control how those 20 or 25 chosen varieties are subsequently tested in rural areas. In practical terms, such a selection program can only be widely decentralized, targeting germplasm for many different areas,if communities bear the brunt of the local-level costs (Sperling and Berkowitz 1994).

From March 1990 onwards, women experts coming to station represented the interests of three types of ad hoc local groups : farmers' research groups backed by non-governmental organizations (NGOs) for specific development projects, self-organized and independent groups of "research-oriented farmers"; and several groups of farmers united by geographic proximity in an administrative unit known as "commune".² The cultivars women selected were then managed in various types of community plots, the NGO probably serving several hundred farmers, the commune units potentially reaching up to 6,000 households. (Hence total potential population reached, 27,000 households or about 135,000 persons). Thirty to fifty farmers were normally invited to review each community plot. One or two of the selected varieties were to be given to each evaluator at harvest, eventually to be tested in their home plots the following seasons.

It is important to note concerns among scientists towards the concept of Phase II, which some saw as at the border of biological research — and moving towards extension. The participatory program came under yearly review from the Great Lakes Regional Bean Network Oversight Committee, an interdisciplinary group representing national institutes of Burundi and Zaire and Rwanda. Here, the feeling was that research itself would be needed to determine the "hows" of the program's institutionalization, not only for Rwanda, but for a range of African national partners. Partially to address issues of rigor, the program was eventually set up as an experiment in which the normal breeding sequence served as the "control" and the participatory program as the "treatment". The two schema were eventually to be compared along such parameters as number of acceptable varieties identified and adoption rates. Box 2 outlines the framework (Scheidegger, fieldnotes).

Box 2: Research experiment to compare two varietal selection frameworks
Season*	Classic procedure (ISAR)	Decentralized selection with farmer participation
1989A	Triage trial
1990A	Comparative trial
1990B	Comparative trial	Selection on-station by expert farmers — 20 varieties
1991A	Multilocation 1	Commmunity selection plot; 50 farmers per community
1991B	Multilocation 1	Farmer designed and managed trials c.150 (no researcher intervention). Independent evaluation by farmers.
1992A	Multilocation 2	Follow-up of 150 trials to identify which to test in controlled plot
1992B	Multilocation 2	WAIT
1993A	On-farm trials
1993B	On-farm trials Release of varieties
1994 A& B	Researcher managed trials to compare the varieties chosen for diffusion with the 4 most frequently multiplied by farmers. 4 replicates, 1 trial in 6 communities in Southern Rwanda
1995A	Adoption study of 6 communes
* 'A' denotes the season covering the period from September through January. For example, 1989A, extended from September 1988 through January 1989. 'B' denotes the season covering the period February through July.

Select results, Phase II

Table 1: Farmer selection of bush beans from community plots, Rwanda, 12/92-1/93*
Variety	SITES
	Sahera	Rutsatira	Gikongoro	Save	Muganza
RWR 756	X	X	X	X	X
RWR 1058	X	X		X	X
RWR 1115	X	X	X	X	X
RWR 785	X		X	X	X
RWR 779	X	X	X	X
RWR 911	X	X	X	X	X
RWR 1134	X
XAN 162	X	X	X	X	X
G484	X	X	X	X	X
RWR 719	X	X	X		X
MLB 49-89A	X			X
URUGEZI		X	X	X	X
SCAM 80 CM/5		X		X
RWR 14			X
RWR 802			X	X	X
RWR 853			X
RWR 1056			X
MLB 40 89A			X		X
RWR 1059				X
* Shading highlights varieties chosen by all farmer groups.

The participatory experiment had proposed that varieties selected by communities and which later showed wider adoption, should be brought back into the formal system and baptized as farmer-breeder varieties. Subsequent seed multiplication and distribution would have to be decentralized to meet diverse regional needs.

Perhaps the most important insights during Phase II lay with institutional concerns. Turning over both the choice of on-station representatives to communities as well as subsequent community plot testing does not always mean that community needs are served. This certainly rang true in Rwanda where relationships even at the neighborhood or "hill" level are marked by hierarchy and where women fall near the bottom of the heap no matter what the class or ethnic group. "Women have no race" goes one proverb, indicating that their power derives from their relationships to significant male others, brother, father, whatever the case may be.

In practical terms, the power structures and particularly male hierarchies, distorted the expansion of the experiment at several key points. In the selection of farmer representatives to screen on-station trials, researchers had the sense that some of the so-called community-selected experts, were neither very informed, nor very representative of community interests. For instance, one community was represented by the government agronomist's sister, and the sector head's wife. The male authorities in charge linked power with knowledge, and imputed male knowledge to their female sidekicks. If he was an important official, she must be a farmer expert.

There was also concern that key figures in charge sometimes fell short on their obligations to community participants at very last stage. The community plot was laid, evaluations completed, but seed of selected varieties was never distributed. So in theory, the data was in, but the seeds never got out to home plots. The advantages of working through administrative structures are many: these units exist countrywide, in all agro-ecological zones and potentially canvasing all farmers. They have the land and could incorporate a mandate of decentralized selection. The philosophy of such units, however, is sometimes governed by "control" rather than "service". Given their substantial strengths, researchers hoped they could be reshaped to collaborate more fully.

The experiment did thrive when women themselves had some control and when the community saw itself as a true community. The women's cooperative, supported by a Belgian NGO, was well organized and very serious about the research. Five experts were sent to station, varieties chosen were subsequently tested on designated group members' plots, and the cooperative as a whole agreed what to multiply, what to discard and what to test further. Over a ton of seed was multiplied before other communities had started to budge.

Technical and institutional concerns

While it may be a conceptual leap, cost-effective breeding hinges on identifying legitimate and representative local partners, and in some cases, expanding the local power base. In the longer-term, local partners, and particularly solidly-organized local groups, should create a demand-pull on research, reshaping the larger pool of varieties on offer and selecting from this the most promising options for localized experiments (see Ashby and Sperling 1995). Expressed in popular form, one might think of the research station as an inventory warehouse: the goods are on offer to whet clients interests/needs, with customers selecting out only what is relevant. Future stocks, even prototype models, might be developed together with clients, and certainly, with clients' needs in mind.

The shift from a focus on exploring of technical expertise to one of experimenting with institutional options was accompanied by a changing methodological emphasis.

On-station procedures

Another experiment tried to examine validity of farmer evaluations in the face of a single replicate assessment. Eight varieties of bush and climbing beans (total 16), which had been chosen by women farmers the preceding year from ISAR's trial, were grown with and without 30 t/ha of farmyard manure. While farmers normally evaluate a single replicate, during 1991, a select group scanned some six. Repeatability of farmer scores was high for clearly good or bad yielders, while scores were not fully consistent over replications for intermediate varieties. It was observed that farmers differentiate parts of higher and lower soil fertility within an experimental plot and then estimate yield for both parts separately. This way of looking at experimental plots, if less objective, may be more refined than the experimental procedure of determining total plot yield and, under highly diverse soil conditions, could result in a fairer judgement of varieties. Farmers also stated that there were no visible differences among replications, that is, there was no relation between fertilization and crop development. This qualitative assessment was in full agreement with statistical analysis of yield data (Scheidegger in CIAT 1991).

During the initial phase of the program, there was also a strong focus on direct "feedback". Scientists wanted to learn first-hand how farmers evaluate: by which criteria, the ranges of acceptability within criterion, as well as the trade-offs among varietal features. The evaluation format was comprehensive: farmers scored each variety and assessed its positive and negative traits. Interviewing was often one on one, scientist (or technician) to farmer.

As the experiment evolved, exposing farmers to a greater range of germplasm and moving towards a community (v. individual) focus, so too methods had to be retooled. From the scientist point of view, in-depth evaluation of 80 entries was no longer logistically possible, nor perhaps necessary. The evaluation format aimed for efficient procedures which encouraged sharing among farmers and gave feedback (or "feedforward") out to communities rather than channel back primarily to research offices. Farmer groups, each region sending 3-5 representatives, were given two sets of colored ribbons to indicate varieties they wanted to test in future community plots and those they felt should be eliminated. They marked as many entries as desired, negatively and positively. After the tagging, plenary field discussions focused on the varieties most often signalled, the outlyers (those with one ribbon), and any variety which particularly captured farmer interest. While one might argue that ribbons confounded results, that is, farmers visualizing others' choices might be unduly influenced, farmer representatives, eager to ferret out the most suitable varieties for their own home areas, perceived no problem. On the contrary, they enjoyed exchanging ideas and reflecting on inter-group differences. They found the final tallies particularly exciting. Ribbons allowed them not only to reflect on their own choices, but to immediately synthesize the results of five group selections. Such synthesis, usually confined to office corridors, was visually striking and illuminating. During the second phase of on-station evaluation, feedforward came at the expense of detailed feedback, with more cursory identification also the consequence of greatly enlarging the options on offer.

On-farm procedures

During the second stage, tested paradigms within communities were determined by participants themselves and in part reflected the group's orientation towards its members. The farmers' research group, technically assisted by an NGO, decentralized testing and evaluated together. A core group, designated as the research contingent, divided up the station-selected varieties and tested them on individual plots: group evaluation was then completed by means of a walking tour (PAMU 1993). The group subsequently multiplied and diffused the most promising entries. The Rwandan Program received a written report on the farmer evaluation, by which time the varieties had already been launched on their way among other community members.

The experiment within the administrative units ("communes") was conducted in a very different and more standard manner. The agronomist took control, station researchers drew up a standardized protocol (varieties sown in lines, at given densities) and some local farmers were invited to evaluate the plot and select varieties for home use. One advantage was that more farmers were exposed to a greater range of cultivars than in the previous model. Such a top-down research posture at the community level is not atypical of many local grassroots groups, who may have some trained technicians — trained under standard models. Due to their greater involvement in commune evaluations, researchers received feedback more quickly, but the progress towards adaptive testing on individual plots and further diffusion was significantly slower.

The different methods and designs used through the experiment represented trade-offs for the various actors involved. Researchers were initially disappointed by the level of farmer expertise proffered — when communities themselves controlled participant selection. Perhaps with greater experience, the power structures would have better signalled exceptional skills within the global group "women." Some scientists also lamented the decline in detailed feedback as farmers screened a larger pool of germplasm and as some subsequent community designs and evaluations ignored "researcher language" altogether, for example, no yield data. The move towards community-oriented models, however, brought important gains to local participating groups. Ribbon evaluations were more transparent; more farmers directly benefitted on farm, and, in the best of cases, farmers identified and distributed productive varieties with unusual speed. As institutionalization of the approach hinges on community participation, standard research models will have to reorient towards communities' own research and development (R&D) concerns. ³ Box 3 summaries select institutional issues of the second phase of research.

Box 3: Participatory selection with Rwanda bean farmers: institutional concerns: Phase II, 1990-1993 Differences in varietal preferences among even closely-spaced farming communities suggests that participatory selection has to be coupled early with decentralized seed multiplication programs. Scaling up of a participatory selection program, implies formal sector research must partner with organized groups of farmers, rather than individuals, to share the costs and responsibilities of widespread varietal research. Working through community institutions does not guarantee that community needs are served. Local power structures — for example, male hierarchies — can distort the fundamental premises of a "participatory" program. The challenge is to identify local organizations which represent the range of farmer interests and which can serve as research partners. Working with farmer groups demands that methods be developed which "feedforward" information to communities as well as feedback insights to the formal sector. There may be important methodological trade-offs between community R&D and formal sector R&D approaches.

Emerging research models

Institutionally, it has also become also clear that farmers can organize themselves to test quite a wide range of germplasm on-farm, although the different organizational structures and protocols used will influence which and how many households can be reached and even the number of germplasm entries potentially accommodated. Research on possible arrangements for community testing needs to be carried much further.

The results of experimental program suggest that the standard breeding models may not be using each partners', breeders and farmers, talents to best advantage, particularly in areas marked by marginal, heterogenous environments. Breeders may not be the best candidates to select for the diversity of needs/preferences nor for the difficult 'composite' traits. Breeders' unique expertise lies in their capacity to generate new genetic variability. Farmers do cross and select, but at an extremely slow rate: scientific breeding accelerates the process. Breeders might also concentrate on those constraints/opportunities which are 'invisible' to farmers: e.g. certain pathogens and diseases. In turn, the finishing of the product, targeting the variety to a particular production system, can and should be left to farmers. To pursue this goal, farmers' would need access to a wide range of germplasm (Box 4).

Box 4: Conceptualizing a new division of breeding labor
Breeders	Farmers
Create new genetic variability > > > Make accessible wide range of germplasm (local and exotic) Screen large amounts of material for "minimum" criteria Screen for key stresses "invisible" to farmers	< < < Target for agronomic conditions (performance) Target for socio-economic circumstances (preference)

Rethinking the breeding division of labor probably also demands that the scientific community rethink how they evaluate the relative success or failure of the growing number of participatory breeding trials. In Rwanda, initial stages were marked by an exclusive focus on production or impact achievements. Our conceptual framework sought to compare the standard program and experimental program in terms of "end-result" variables: for instance, number of acceptable varieties identified, number of disease resistant varieties identified, rates of adoption of the two sets of material adoption. Varietal diversity of the ISAR-released v. farmer-selected material mainly came into view as a evaluation-variable when it became clear that farmers wanted many and varied cultivars. However, aside from giving farmers access to a wider range of options on-station, the experiment was not shaped to specifically enhance genetic diversity on-farm. Much more could have been done to 'promote genetic diversity' had the participatory program been conceived with this primary goal in mind.

As the experiment evolved, 'community capacity to serve as research partners' lunged to the forefront as a process to strengthen. The technical findings alone (e.g. 'farmers can expertly target varieties) could not deliver adapted varieties to local groups. Enhancing community control and research skills therefore became a central issue in enhancing the efficiency of breeding. Within such a perspective, 'empowering communities' becomes a functional necessity for achieving cost-efficient research programs. Box 5 starts to sketch parameters along which we might start to evaluate our participatory breeding trials, according to each program's specific focus. Broadly, at least three perspectives presently guide such participatory experiments: some practitioners focus on production achievements, some on the enhancement of genetic diversity, and still others on the shifting of control (of germplasm and the breeding process itself) to communities and other grassroots organizations. Successful participatory breeding program should probably show positive indicators in all three categories. Relative emphasis will vary greatly according to the primary objective of the program.

Box 5: Participatory breeding programs: potential evaluation criteria
Functional perspectives (orientation:products)
Production/Impact Enhancement # farmer-acceptable varieties # disease-resistant varieties absolute production gains rates of adoption · ·	Genetic Diversity genetic profile of released varieties incidence of landrace parents · ·
Control/empowerment perspectives (orientation:process) degree to which: farmer skills are enhanced to more effectively cross/select themselves farmers gain fuller access to wide pool of germplasm farmers control local testing farmers are involved in decisions of varietal release · ·

After the genocide: varietal assessments and reintroductions

The aid community, particularly NGOs and various United Nations agencies, responded swiftly and on a wide scale to the agricultural crisis. During the subsequent growing season, September 1994-January 1995, large amounts of seed of key crops were distributed: 6970 MT of bean, 1707 MT of maize, and 7230 kg of vegetable seed (MINAGRI/UNREO/PNUD/FAO 1994).

The CGIAR has responded along lines of its own advantage, assessing the state of varietal and genetic erosion and developing strategies to restock germplasm in national research sites as well as on farmers' fields. A "Seeds of Hope" (SOH)⁴, initiative, is now multiplying collections of local material, breeding lines, and improved lines appreciated by farmers for possible reintroduction. For beans alone, 170 landraces have been obtained through national and international genebanks. The first nationwide surveys, conducted through a range of NGOs (CARE, World Vision, Swiss Disaster Relief, Catholic Relief Services and Medécins sans Frontières) has suggested that varietal loss has been less than anticipated: 45% of the seed sown during the first post-event season came from farmers' own stocks (Sperling 1995a). In August/September 1995, surveys will further examine this issue of varietal loss for the most vulnerable areas, those which experienced large-scale population movements.

Methodologically, in reference to farmer participatory selection, two interesting developments can be signalled.

First, SOH is looking at the complementarity and differences between farmer varietal assessments and molecular genetic assessments. Farmer assessments tend to be site specific and indicate the degree to which farmers can access desired varieties (that is, varieties which have useful traits which are available in useful combinations). Molecular assessments (using RFLP) suggest the presence/absence of genetic characters nationwide and map region variations at community (versus farm) levels. Both programs aim to determine the genetic and varietal needs of Rwandan farmers and to guide the rebuilding of genetic collections at ISAR (Sperling 1995b, and S. Beebe, personal communication).

Second, farmer participatory selection is being proffered as a major method of reintroducing germplasm at the community level, should varietal restocking be necessary (World Vision, J. Hooper, personal communication). As provenance data on the 170 landraces needs to be sharpened, the proposals suggest that entries roughly be sorted by high, medium and low-altitude adaptation and then be moved to community plots for further targeting. In the Rwandan context, farmer participatory selection thus becomes a chosen strategy for research and development initiatives but also for emergency aid and rehabilitation efforts. Let farmers help get the germplasm to where it can best be used.

References

Ashby, J. and L. Sperling, 1995. Institutionalizing participatory, client-driven research and technology development in agriculture. Development and Change, 26(4):753-770.

International Center for Tropical Agriculture (CIAT), 1993. Bean Program Annual Report. CIAT: Cali, Colombia

International Center for Tropical Agriculture (CIAT), 1991. Bean Program Annual Report. CIAT: Cali, Colombia

Iyameremye, A., n.d. Le secteur agricole durement touché. Traits d'union, Rwanda 5.

Lamb, E.M and L.L Hardman, 1985. Final report of: Survey of bean varieties grown in Rwanda. January 1984-June 1985. AID-Rwanda Local Crop Storage Cooperative Research.

Merrill-Sands, D., S. Biggs, R.J. Bingen, P. Ewell, J. McAllister, and S. Poats, 1991. Institutional Considerations in Strengthening On-Farm Client-Oriented Research in National Agricultural Research Systems: Lessons from a Nine Country Study', Experimental Agriculture, 27: 343-373.

Merrill-Sands, D., M.E.O. Arnaiz, R. J. Bingen, J. Farrington, D. Carney, A,J, Bebbington, (forthcoming), Conceptual framework for study on: the role of farmers organizations in technology change: current situation and future prospects. The Hague: ISNAR.

Ministère du Plan (MINIPLAN), 1988. Enquête nationale sur le budget et la consommation des ménages. Volume 4: consommation alimentaire en milieu rural. Kigali: République Rwandaise. Décembre 1988.

MINAGRI/UNREO/PNUD/FAO, 1994. Compte-rendu de réunion, 16 décembre 1994: Opérations d'urgence et réhabilitation agricoles.

Projet Agricole de Muganza (PAMU), 1993. Rapport Annuel 1992. Butare, Rwanda, Février 1993.

Sperling, L., 1995a. Emergency relief in Rwanda: Assessment of bean and seed use 1995A, Summary Report. SOH Assessment Document 1.

Sperling, L., 1995b. The impact of war on plant genetic resources: bean varietal assessment in Rwanda 1995A. Summary Report. SOH Assessment Document II.

Sperling, L., 1992. Farmer Participation and the Development of Bean Varieties in Rwanda. In J. Moock and R. Rhoades, eds. Diversity, Farmer Knowledge, and Sustainability. Ithaca and London: Cornell University Press.

Sperling, L. and P. Berkowitz, 1994. Partners in Selection: Bean Breeders and Women Bean Experts in Rwanda. Washington: CGIAR Gender Program.

Sperling, L., M. Loevinsohn and B. Ntabomvura, 1993. Rethinking the Farmer's Role in Plant Breeding: Local Bean Experts and On-Station Selection in Rwanda. Experimental Agriculture, 29:509-519.

Voss, J., 1992. Conserving and increasing on-farm genetic diversity:farmer management of varietal bean mixtures in Central Africa. In J.L. Moock and R.E. Rhoades, eds. Diversity, farmer knowledge and sustainability. Ithaca:Cornell University Press.

Footnotes:

1 Professor, Swiss College of Agriculture. (See 'participant list' for address of senior author.) We thank the many individuals who made the experimental program possible: in particular, Dr. Pierre Nyabyenda, Gaspard Gasana and David Cishahayo from ISAR; farmers in Save, Muganza, Sahera and Gikongoro; and NGO colleagues from the Projet Agricole de Muganza and Projet Agricole de Gikongoro. Dr. Robin Buruchara helped lead the second phase and Beatrice Ntabomvura facilitated the fieldwork throughout. (BACK)

2 In other contexts, such as Zaire, the participatory experiment was carried out with well-organized farmers' cooperatives. Unfortunately, Rwanda has a limited tradition of farmers cooperatives or with any grass-roots organizations which might lobby for farmers interests or organize collective ventures on a large-scale, for example, credit or marketing. (BACK)

3 For instance, facing similar challenges in India, the KRIBHCO project is recommending that "farmer-acceptability" data, versus the standard yield trials, be considered as sufficient evidence for varietal release (J. Witcombe, personal communication). (BACK)

4 The Seeds of Hope Initiative is a joint rehabilitation initiative of the CGIAR. Formalized in September 1995, many African NARS have contributed germplasm, field space, and advice to the initiative: those of Burundi, Ethiopia, Kenya, Malawi, Tanzania, Uganda, Zaire, Zimbabwe and, more recently, Rwanda. In addition, some seven of the International Agricultural Research Centers (IARCs) are strongly involved in the Rwandan Agricultural reconstruction:

Document(s) 20 de 38