African Crop Science Journal African Crop Science Society ISSN: 1021-9730 EISSN: 2072-6589 Vol. 5, Num. 3, 1997, pp. 285-294

Using trial follow-up surveys to assess varietal adoption: the case of beans

S. David, C. Wortmann, S. Kasozi and M. Mugisha-Mutetikka

International Centre for Tropical Agriculture (CIAT), P.O. Box 6247, Kampala, Uganda Uganda National Bean Programme, P.O. Box 7084, Kampala, Uganda

(Received 21 April, 1997; accepted 4 July, 1997)

Code Number: CS97035
Sizes of Files:
    Text:
    Graphics: Tables and line drawings (gif) - 19.9K 

Drawing on results of a survey of Ugandan farmers who previously hosted bean (Phaseolus vulgaris L.) varietal trials, this paper reports on the adoption of three new bush bean cultivars and assesses the usefulness of post-trial surveys as a methodology for obtaining reliable early feedback on varietal adoption potential. While this case study confirms the validity of post-trial surveys for verifying varietal acceptance and predicting broad trends in varietal adoption, the findings from such studies should be assessed within the context of the complexity of choices farmers make about seed use, which are not solely an expression of their varietal preference, especially among the poorest farmers. The negative implications of adverse agro-environmental conditions for the retention of seed of new varieties and hence, adoption, are clearly highlighted. The study also warns against the pitfalls of making wider inferences from the adoption behaviour of trial farmers, since the experimentation process itself may interfere, positively or negatively, with the adoption process.

Key Words: Diffusion of agricultural technology, on-farm testing, technology adoption

RESUME

A partir des resultats d'une enquete menee sur quelques fermiers ougandais qui, au prealable, avaient participeaux experimentations des varietes des haricots, le rapport fait la lumire sur l'adoption de trois nouvelles varietes des haricots sauvages et fait un compte rendu de l'utilite des enquetes des post- recensements d'essais, comme une methodologie pour obteuir une preuve de regularitedes informations precoces sur le potentiel d'adoption des varietes. Alors que ce recensement preliminaire confirme la validitedes enquetes de post-recensement d'essais pour verifier l'acceptation des varietes et prevoir les grandes tendances sur l'adoption des varietes, les decouvertes de telles etudes devraient etre etablies apres avoir tenu compte des choix divergents que les fermiers font sur l'usage des semences, qui n'est pas uniquement, une expression de leur preference varietale surtout parmi les fermiers les plus pauvres. Les implications negatives sur l'adversitedes conditions agro-environnementales pour la retention des semences de nouvelles varietes et, par la suite leur adoption, sont clairement mis au point. Le rapport attire aussi l'attention sur les dangers de faire de grandes inferences concernant le comportment des fermiers l'egard de l'adoption des varietes, ds lors que le procede d'experimentation peut avoir des effets positifs ou negatifs sur le processus d'adoption.

Mots Cles: Diffusion de technologie agricole, experience sur le champ, adoption de la technologie

INTRODUCTION

Surveys to investigate the adoption of a new agricultural technology are typically expensive, time-consuming and undertaken several years or seasons after widespread dissemination and diffusion of the technology. Cost-effective methodologies for generating empirical data on adoption rates and patterns are therefore needed to predict adoption potential for planning purposes and to satisfy donors' demands for early ex-post adoption results. Documenting adoption during the take-off phase of a new technology may also alert researchers to the need for further fine tuning in the design of the technology or improving dissemination strategies.

On-farm varietal trials (OFVTs), typically the final stage in the plant breeding cycle in commodity research programmes, generate a wealth of data from a relatively large population of farmers which are used by breeders to document the performance of promising varieties under farmers' conditions, understand farmers' va rietal preferences and support the release of new varieties. This paper explores a third possible use of data collected from trial farmers: predicting the adoption potential of new varieties through follow-up surveys, where OFVTs are considered as a first stage in technology diffusion.

Trial farmers form a logical early adoption study population for several reasons: 1. if proper records are kept, they can be easily located several seasons after trials have ended; 2. they are spread over several agro-ecological regions; and 3. since these farmers have experience with growing a number of test varieties, they provide a cost-effective opportunity to investigate the uptake of several varieties at a time. Drawing on results from a follow-up survey of Ugandan farmers involved in bean (Phaseolus vulgaris L.) varietal trials, this paper has two objectives: reporting on the adoption of three new bush bean cultivars and assessing the usefulness of post-trial surveys as a methodology for obtaining reliable early feedback on varietal adoption potential. While similar surveys have been conducted in Uganda and elsewhere (Kisakye, 1990; Janssen et al., 1991; Grisley et al., 1993), the methodology itself has not been evaluated. The use of follow-up surveys to assess the adoption potential of new cultivars is, notably, best suited to self-pollinating crops such as beans, the seed of which can be resown over several seasons. Findings from this study may therefore be relevant to research on the adoption of new varieties of groundnuts and cowpea or vegetatively propagated crops such as cassava and sweet potatoes.

Two important premises underlie the hypothesis that the adoption behaviour of trial farmers provides insight about the adoption potential of new bean varieties. The first is that farmers make a clear-cut distinction between testing a new variety and adoption and need 2-3 seasons (or years, depending on the crop) to grow and assess a variety before making the decision to adopt (Pachico and Ashby, 1983; Sperling and Loevinsohn, 1993). We also assume that two measurable behaviours: growing a variety for at least 2 seasons following trials, and seed diffusion (i.e., whether or not diffusion took place and diffusion motivated by appreciation of the new variety), are reliable indicators of farmer appreciation of a new variety and of its adoption potential.

METHODS AND MATERIALS

Trial procedures.

Common bean is the most widely grown and consumed grain legume in Uganda. In following a strategy of increasing bean productivity through the release of high yielding, stress-resistant cultivars, the Uganda National Bean Programme (UNBP) regularly carries out on-farm varietal trials in several regions of the country. Researcher-designed, farmer-managed varietal trials were supervised by the Programme's a gronomist or staff in collaborating institutions (i.e., NGOs or agricultural projects). Trials had two replicates per farm in a randomised complete block design and were laid on plots of 3 meters by 4 meters. Each season^1 farmers were provided with kits containing 250-300 seeds of 4-6 test varieties. They were instructed to follow their usual management practices and to plant a check variety of their choice. Besides help from extension in planting, the trials were completely farmer-managed. Post-harvest farmer evaluations were conducted by the UNBPgronomist on an individual and group basis. After yields were measured, farmers were at liberty to eat or resow the seed of test varieties. New seed was provided during every season of trials.

The present survey monitored three bush bean varieties developed by the UNBP: MCM 5001, CAL 96 and RWR 136 . CAL 96 and RWR 136^2 are determinate bush types (Type 1) characterised by dark red mottled, large seeds of the Calima types highly appreciated in Uganda. MCM 5001, a small seeded Carioca seed type previously unknown in Uganda, has an indeterminate (Type 2) growth habit. The on-station yield results were 919 kg ha^-1 for CAL 96 and 949 kg ha^-1 for RWR, a 27% and 30% yield advantage, respectively, over the local check^3. With a yield of 1377 kg ha^-1, MCM 5001 outyielded the local check by 80% and has resistance to bean common mosaic virus (BCMV), an important production constraint in some regions. Before the varieties were released and distributed, scientists made predictions about their adoption potential based on observations regarding regional varietal preferences and farmer evaluation data (Uganda National Bean Programme, 1991, 1992, 1993, unpubl.). CAL 96 was expected to have a high adoption rate in all regions of the country because of its seed type and colour, taste, growth habit and similarity to K20, a popular, highly marketable Calima seed type. While MCM 5001 was expected to be appreciated in the West, East and North due its yield advantage, seed size, drought and disease tolerance and good taste, low adoption of this variety in the Central region was predicted due to its small seed size and tan colour. Because of its seed colour and size, growth habit, early maturity, short cooking time and taste, RWR 136 was expected to be popular in the Central region.

Survey.

A survey of 60 farmers who had hosted varietal trials (n=48) and demonstrations (n=12) with the UNBP and its collaborators between 1991 and 1993 was conducted in April-May 1994. The survey was restricted to 7 districts in which OFVTs had been conducted during this period: Iganga, Pallisa and Kamuli in Eastern Uganda (n=17), Mubende and Mpigi in the Central region (n=22) and Bushenyi and Mbarara Districts in the West (n=21) ^4. The sample was drawn from lists of trial farmers and selected opportunistically on the basis of availability. All farmers interviewed had received the test varieties at least three seasons prior to the survey and half had received the varieties 4 or more seasons prior to being interviewed. Not all farmers surveyed had conducted trials with all three varieties: 57 farmers tested MCM 5001, 39 tested CAL 96 and 27 tested RWR 136. Most farmers participated in trials for more than 1 consecutive season: 63% of farmers who grew RWR 136, 62% of farmers who grew CAL 96 and 53% of farmers who grew MCM 5001. Seven seasons was the longest time a farmer hosted trials. With the exception of one case, demonstrations were conducted during a single season. The interview schedule included questions about varietal performance, farmers' assessment of the varieties and diffusion behavior.

Farmer characteristics.

Most farmers surveyed (77%) had been selected for trials by extension agents and therefore may not be representative of the wider population of farmers. The majority of respondents (63%) were women, the principal bean farmers in Uganda. Most lived in male-headed households (83%), but 12% were de jure female heads of households. Most respondents were experienced farmers since the vast majority (75%) were 31 years of age and above, and over a third (39%) had lived on their current farm for 21 or more years. Over half of the trial farmers surveyed (57%) had primary or a higher level of education. Several indicators suggest that trial households belong to the average and above average wealth strata. Half of all farmers hired labour on a regular basis and 39% owned a shop or business. Based on their own subjective assessment of wealth, interviewers considered only 10% of households surveyed to be of below average wealth.

RESULTS

TABLE 1

Adoption and varietal longevity.

Since the majority of farmers surveyed hosted trials for 2 or more seasons, whether or not a farmer continued to grow a test variety after trials ended can be considered as indicative of adoption, rather than testing, behaviour. More than half of trial farmers continued to sow test varieties on their own accord: 82% of farmers growing CAL 96, 67% of farmers growing MCM 5001 and 56% of farmers growing RWR 136 (Table 1). No significant difference in the adoption behaviour of farmers who hosted trials and demonstrations was observed. Regional differences in post-trial sowing of the test varieties show that relatively few farmers in all regions dropped CAL 96 after trials ended: 10% in the East, 17% in the Central region and 24% in the West. While MCM 5001 was dropped by 52% of farmers in the Central region after trials, relatively fewer farmers discontinued growing that variety in the East (18%) and West (26%). Based on the proportion of farmers across all regions who dropped RWR 136 after trials terminated, this variety was the least preferred of the three test varieties. Notably, however, the higher proportion of farmers who continued sowing both RWR 136 and CAL 96 after trials in the Central and East relative to the West, confirms the strong preference in those regions for Calima seed types. In the Central region this preference may be largely attributed to the dominating market influence of Kampala, Uganda's capital.

TABLE 2

Most studies investigating technology adoption behaviour analyse the initial decision to adopt a new practice, but not the extent, intensity, quality or completeness of the adoption process (Feder et al., 1985). Going beyond a dichotomous analysis of adoption to assess the number of seasons a test variety was planted after trials ended reveals more about adoption behaviour and patterns. As Table 2 shows, the number of farmers who sowed test varieties declined steadily over time, with the result that at the time of the survey, no farmer had sown a test variety for more than 3 seasons after trials. The number of seasons a farmer grew test varieties after trials was not associated with socio-economic characteristics such as wealth, age and household type. Farmers continued to sow CAL 96 for more seasons after trials ended (an average of 1.6 seasons vs. 1.5 for the other two varieties) compared to the other two varieties: 19% sowed that variety for three seasons compared with 8% who sowed MCM 5001 and 7% who sowed RWR 136 for the same number of seasons. However, the data show no significant difference between varieties on post-trial longevity (x^2 = 0.26; 2 d.f; P <.05), that is, no variety was more likely to be grown for more seasons than the other.

Varietal mortality.

TABLE 3

While consumption of seed was probably in some cases intentional and in others unintentional5, the higher percentage of farmers who consumed seed of RWR 136 and MCM 5001 (40%) compared to CAL 96 (17%) is likely to reflect variety-specific, voluntary rejection due to low appreciation. Compared to MCM 5001 and RWR 136, few far-mers disliked CAL 96 on any commonly mentioned characteristic. A majority of farmers were impressed by the yield advantage (95%) and taste (70%) of MCM 5001, but RWR 136 was not rated as outstanding on any characteristic by a vast majority of farmers. Farmer assessment of the yield capacity of that variety varies, with 48% regarding it to be high yielding and 30% being of the opposite view. One hypothesis requiring further investigation is that in times of food shortage, farmers will consume seed of both appreciated and non-appreciated bean varieties, with a stronger bias toward the latter.

The data suggest a reverse decision-making pattern with regard to seed sale in that farmers appear to readily sell seed of preferred newly introduced varieties. Given a strong market preference in the Central and Eastern regions of the country (represented by 65% of the sample) for Calima grain types, the higher percentage of farmers who sold most or all seed of CAL 96 (25%) and RWR 136 (13%) compared with MCM 5001 (3%) is more likely a reflection of favourable market opportunities rather than voluntary rejection, especially of CAL 96. It can therefore be hypothesised that when tempted by favourable market opportunities, most resource-poor farmers will sell all or most seed of appreciated varieties, both local and newly introduced. Given the poor storage facilities of many households, such a decision is logical with regard to local varieties, the seed of which can easily be obtained from shops, markets or other farmers. Another possible explanation for the seemingly irrational sale of seed of a well regarded variety such as CAL 96 is farmers equent inability to distinguish it from K20 and the ease with which these two Calima types could be accidentally mixed and sold. Consumption and sale of seed of test varieties may also reflect expectations on the part of some trial farmers of long-term contact with extension/researchers, assuring them of a regular supply of seed.

A number of extraneous factors, which do not reflect farmers' varietal preferences, also determined farmers' ability to retain seed of the test varieties. The most important of these was drought which seriously disrupted crop production in many parts of the country during the second season of 1993. Involuntary seed loss also resulted from situations which prohibited planting for one or more seasons, such as illness and a cultural taboo in Mbarara District against newly widowed women planting farm-saved seed. In a few cases, trial implementation caused unintentional seed loss when extension personnel collected seed from trial farmers for other uses (e.g. distribution to other farmers, to feed a training course or to sell for profit).

Diffusion.

TABLE 4

The major reason cited by respondents for diffusion was the desire to share an appreciated variety (30%). Another reason, which also indicates varietal preference, was the desire to have the seed multiplied by others to ensure its availability to the donor and the wider community (23%). Other reasons were not specifically related to farmer (i.e., donor) preferences: in response to requests (23%), as a reciprocal act of exchange (11%), instructed to do so by extension agent (7%), among other reasons (7%). Insufficient seed was the primary reason cited by non-diffusers, although reasons differed by variety. Most farmers who did not share seed of CAL 96 (43%) and MCM 5001 (37%) were reluctant to part with their seed because of the small quantities available, in contrast to nearly a third of farmers who did not distribute RWR 136 because they received no requests. The latter may suggest that trial farmers were less likely to recommend RWR 136 to other farmers, and consequently received relatively few requests for seed of that variety.

TABLE 5

TABLE 6

The data reveal different rates of post-trial sowing of the three varieties and give indications of diverse regional adoption potential, although no statistically significant trends were documented. Confirming researchers edictions, CAL 96 showed the highest adoption rate in all regions investigated. MCM 5001 showed the next highest rate of adoption, particularly in the West and East. RWR 136 was dropped after trials by the largest proportion of farmers, suggesting low adoption potential generally, although adoption was highest in the Central and Eastern Regions.

Three findings deserve more detailed discussion: 1. the lack of statistically significant variation in post-trial longevity between varieties; 2. the absence of a consistent pattern between adoption behaviour, farmer preference and seed diffusion; and 3. the effects of drought, marketing and consumption behaviour on seed retention. While the absence of differential adoption rates between varieties may have resulted from the small sample size, we conclude from the high and rapid rate of mortality for all test varieties, even appreciated ones, that adoption behaviour is influenced by many other factors besides varietal characteristics. These findings therefore suggest that, given the precarious socio-economic and agro-climatic environments under which farmers produce beans and make decisions regarding the allocation of their harvests, the adoption potential of a new bean cultivar cannot be predictably measured by a single behavioral indicator such as the number of seasons a variety was sown.

The high rate of seed loss documented by this study is evidence that the adoption of new bean cultivars is a complex and non-linear (i.e., not one time) process. While recent Ugandan data indicate a reasonably high level of success by most smallholders at retaining seed of bean landraces (David, 1996)^6, research results suggest that farmers, especially the poorest, find it more difficult to keep seed of newly introduced cultivars, even appreciated ones^7. Larger seed stocks of local varieties and the possibility of obtaining seed of landraces from various off-farm sources (shops, markets, other farmers) explains farmers eater success at retaining the former compared to new varieties, not yet fully established in local seed systems. In a given setting, the ability of individual farmers to retain seed of a preferred newly released bean variety appears to be a function of several factors, including: environmental conditions during the first few seasons after initial seed distribution, the quantity of seed received and the multiplication rate and productivity of the variety. The probability of varietal survival among individual farmers may also be boosted where strong seed exchange networks exist, since farmers often deliberately diffuse seed as a security measure (David, 1996).

The study provides little evidence to support our premise that seed diffusion is a reliable indicator of farmer appreciation of a new bean variety. As Table 4 shows, the proportion of farmers who exchanged seed of RWR 136, the least appreciated variety, was higher than the proportion who distributed the other two varieties in 1992a and 1993b. Thus, our data corroborates the observations made by Sperling and Loevinsohn (1993:451) in Rwanda that the spread of popular improved bean varieties through farmer seed networks is effective to varying degrees and the speed of diffusion varies by variety. We concur that while the diffusion behaviour of individual farmers is often an expression of varietal appreciation, it is also influenced by several varietal and nonvarietal factors, namely, the multiplication rate of a variety, seasonal yield performance, the quantity of seed initially received, social obligations governing seed exchange (which may vary by locality) and the status of individual farmers in seed exchange networks8. These factors were not considered in this paper due to the unreliability of farmers' yield estimates and the lack of data on other variables.

Finally, our findings suggest a need for reorientation of bean seed delivery strategies in Uganda. Presently, in Uganda and most countries in Eastern and Southern Africa, small quantities of seed of newly released bean cultivars are distributed free of charge to small-scale farmers by the national bean research programme and NGOs. The evidence from our survey, that even better-off trial farmers were unable to retain seed of new bean varieties over 3 seasons, suggests that the goal of promoting rapid adoption of a new bean cultivar requires a system which resupplies seed over a number of seasons until the variety becomes fully established in local seed networks and markets.

Where bean seed is to be supplied to resource-poor farmers by formal institutions outside of the seed sector (e.g. national research programmes, NGOs), the optimal number of seed deliveries should be determined by the total quantity of seed disseminated per locality, the amount of seed received by individual households, the multiplication rate of the variety and the incidence of adverse climatic conditions, among other factors. Within this context, an important role for extension should be monitoring adoption by trial farmers and encouraging the multiplication and diffusion of seed locally by these farmers, once the new variety has passed through a period of local validation. While bean seed could be informally produced by these farmers, another option is the establishment of small-scale farmer enterprises specialising in seed production (David, 1994, unpubl.; Lepiz et al., 1996).

CONCLUSION

The Ugandan experience with beans shows that follow-up surveys of trial farmers can be a cost-effective method for obtaining early, qualitative feedback on varietal acceptance and adoption potential. However, the findings suggest adoption behaviour documented by such studies should be assessed within the context of the complexity of choices farmers make about seed use which are not solely an expression of their varietal preference, especially among the poorest farmers. The negative implications of adverse agro-environ-mental conditions for seed retention and adoption were clearly highlighted, as were the pitfalls of making wider inferences from the adoption behaviour of trial farmers, since the experimentation process itself may interfere, positively or negatively, with the adoption process. In conclusion, while this paper confirms the validity of post-trial surveys for predicting broad trends in adoption and alerting researchers to the need for improving technology dissemination strategies, it warns against drawing strong conclusions about adoption in the wider population from such data.

^1 Trials were conducted during both growing seasons: March-June (season A) and August-November (season B).

^2 MCM 5001 and CAL 96 were released in 1994.

^3 Possibly due to inaccurate measurements, yields per hectare from on-farm trials or the three varieties surpassed those from on-station trials: CAL 96: 1333kg; RWR:1151kg; MCM 5001:1333 kg.

^4 While Mbarara and Bushenyi Districts are geographically located in the South-west, they represent the bean production environments of Western Uganda. Hence, these districts are referred to in this paper as Western Uganda.

^5 In some cases, farmers were forced to eat beans intended for planting during periods of food shortage. Such behaviour is construed as a survival strategy rather than as disadoption.

^6 Only 10% of the 233 farmers surveyed in two districts of Uganda were chronically seed deficient (David 1996). Other studies conducted in East and Southern Africa also report a minority of farmers in this category (Ferguson and Mkandawire, 1993;Cromwell and Zambezi, 1993; Speriling, 1994). On the other hand, over an 8 year period, 75% of Ugandan bean farmers surveyed had experienced varietal loss, either deliberately or involuntarily (David, unpubl. data).

^7 Sperling and Loevinsohn (1993)found that among Rwandan farmers who stopped growing seed of appreciated new bean varieties, the most common reasons for seed loss were agro-environmental factors (36%), followed by varietal-specific (33%) and socio-econornic reasons (30%). Among the farmers surveyed in Lushoto District, Tanzania, who no longer grew Lyamungu 85 bean variety (65% of the sample), 77% had lost the seed during a drought (Ndakidemi and Mushi, unpubl. data).

^8 In some localities in Uganda, where beans have recently become a cash crop, sale or exchange of bean seed apears to be replacing more altruistic forms of exchange (David, 1994, unpubl.). In other settings, cultural norms stipulate that diffusion of new technologies between farmers takes place only through in-kind or cash exchange.

ACKNOWLEDGEMENTS

The authors wish to acknowledge and thank all enumerators who participated in the survey which included staff from Kawanda Agricultural Research Institute, the South West Agricultural Rehabilitation Project (SWARP) and extension staff in Mpigi, Pallisa and Kamuli Districts.

REFERENCES

Cromwell, E. and Zambezi, B. 1993. The Performance of the Seed sector in Malawi: An Analysis of the Influence of the Organizational structure. ODI, London.

David, S. 1996. Local bean seed systems in Uganda: Preliminary results from surveys in two districts. In: Alternative Approaches to Bean seed production and Distribution in Eastern and Southern Africa: Proceedings of a working group meeting, 10-13, October, 1994, Kampala, Uganda. David, S. (Ed.). CIAT Africa Workshop Series, no. 32.

Feder, G., Just, R.E. and Zilberman, D. 1985. Adoption of agricultural innovation in developing countries: A survey. Economic Development and Cultural Change 33:257-97.

Ferguson, A. and Mkandawire, R. 1993. Common bean and farmer managed diversity: Regional variations in Malawi. Culture and Agriculture 45-46:14-17.

Grisley, W. 1991. Seed for bean production in Sub-Saharan Africa: Issues, problems and possible solutions In: National Research Planning for Bean Production in Uganda. Grisley, W. (Ed.). Ministry of Agriculture, Uganda and CIAT.

Janssen, W., Ruiz de Londono, N., Beltran,J. and Woolley, J. 1991. On-farm research in support of varietal diffusion: Bean production in Cajamarca, Peru. In: Planning Change in Farming Systems: Progress in On-farm Research. Tripp, R. (Ed.), pp. 191-214. John Wiley and Sons, Chichester.

Kisakye, J. 1990. Follow-up of on-farm bean cultivar trials in Uganda. In: Progress in the Improvement of Common Bean in Eastern and Southern Africa. CIAT Africa Workshop Series, no. 12.

Lepiz, R., Ashby, J. and Roa, J. 1996. Artisinal bean seed production in Latin America. In Alternative Approaches to Bean Seed Production and Distribution in Eastern and Southern Africa: Proceedings of a working group meeting, 10-13, October, 1994, Kampala, Uganda. David, S. (Ed). CIAT Africa Workshop Series, no. 32.

Pachico, D. 1992. Experiences and challenges in the assessment of international agricultural research impacts. In: Assessing the Impact of International Agricultural Research for Sustainable Development. Lee, D.R. and Kearl, S. (Eds.). Ithaca: Cornell University Press.

Pachico, D. and Ashby, J. 1983. Stages in technology diffusion among small farmers: Biological and management screening of a new rice variety in Nepal. Agricultural Administration 13:23-37.

Sperling, L. and Loevinsohn, M. 1993. The dynamics of adoption: Distribution and mortality of bean varieties among small farmers in Rwanda. Agricultural System 41: 441-453.

Sperling, L. 1994. Analysis of bean seed channels in the Great Lakes Region: South Kivu, Zaire, Southern Rwanda, and select bean-growing areas zones of Burundi. CIAT African Occasional Publications Series, no. 13, CIAT/RESAPAC, Butare, Rwanda.

Copyright 1997 The African Crop Science Society

The following images related to this document are available:

Line drawing images