This paper synthesises five years of field research on the bean seed sector within the Great Lakes region of Africa (Rwanda, Burundi and Eastern Zaire). The work, conducted jointly by farmers, three national programmes, the International Center for Tropical Agriculture (CIAT), NG0s, shopkeepers, prisons, nutritional centres, and others suggests more effective ways for tailoring seed systems towards specific agroecological and socio-economic environments. While some of the recommendations are specific to the bean crop and the region, most can be applied more broadly.

The paper is divided into five main parts: observations on the formal seed system, diagnosis of the informal channels, and reflections in three areas—seed distribution, genetic management, and seed production—where Great Lakes research has built on existing farmer knowledge and farming systems to foster an improved bean seed sector. A final section suggests a framework for choosing among the multiple strategies to strengthen seed systems for small farmers. Analyses show that the formal seed service, a government parastatal, is not necessarily meeting small farmer needs. There is a bias towards larger grain bean seed, suitable for more fertile soils (and hence better off farmers) and key, farmer-desired varieties, are dropped by the wayside. Nor are the channels through which the service diffuses operating as expected: development projects, used as intermediate producers, often show negative rates of multiplication, and farmer-to-farmer diffusion is unexpectedly slow, especially in stress environments. Despite hidden subsidies, the formal system reaches only one in 600 farmers.