Plant breeders cannot respond to every quirk of farmers' circumstances but the relevance of breeding research in poor nations can be improved. Recent innovations in germplasm screening programmes for developing nations highlight important differences in selection criteria between farmers and scientists, and among farmers themselves. Drawing on the authors' African experience with International Agricultural Research Centres working on maize and potatoes, the paper explores how breeding programmes benefit from an understanding of farmers' own detailed knowledge about the crop varieties they already cultivate. Issues considered include farmers' use of cultivar diversity, the trade-off between yield and maturity class as selection criteria, defining appropriate experimental conditions, and innovative techniques to increase farmer participation in germplasm screening.

The effect of no-tillage and mulch on soil properties and tuber yield of cassava was studied over a five year period on an infertile, acid Ultisol in south-eastern Nigeria. Mulch was applied at a rate of 12 t ha−1 as a split application once at planting and once six months after planting. No chemical fertilizers were used during the trial. Bulk density and proportion of micropores (pore radius, r, <1.4 µm) were greater, and proportion of macropores (r>14.4 µm) lower, with no-tillage. Water infiltration was increased by mulching, being greatest in tilled, mulched plots. Soil chemical properties were not significantly affected by tillage system, although levels of exchangeable calcium, magnesium and potassium were higher and acid saturation lower with mulching. Soil physical and chemical properties were best with tillage and mulching, and poorest when neither tillage nor mulch was used. Cassava tuber yield was unaffected by tillage system but increased by mulching.

Stages in oil palm inflorescence development which determine the number and weight of the fruit bunch components were identified by thinning to intervene in development and dissection of a sample of the felled palms. This showed that the ratio of female to total inflorescences and the number of flowers per spikelet were determined eight leaves prior to spikelet initiation. The number of spikelets per inflorescence was linked with the meristematic development of the primary axis. Inflorescences were prone to abortion at the onset of rapid expansion of the primary axis. The weight of the frame responded to thinning at these three developmental stages. The weight of single fruits responded shortly after the fruit set was fixed at anthesis; the response levelled off in axils of the six older leaves. At the start of an abortion phase nearly all aborted inflorescences were female, but subsequently female and male inflorescences aborted in equal proportions. Our study suggests that the number and weight of fruit bunch components are determined at a specific developmental stage but that the time to anthesis may vary considerably as a result of fluctuations in the speed of inflorescence development.

Yield components in oil palm were recorded in a spacing experiment comparing 56, 110, 148 and 186 palms ha−1. The higher densities reduced the number of female inflorescences (due to a decrease in the proportion of female to total inflorescences and an increase in the proportion of leaves with aborted inflorescences), the weight of the frame and the number of flowers per inflorescence. The results suggest that the production of seed per bunch can be increased by at least 15% by thinning around selected palms in seed gardens. The sex ratio and the components of flower number per inflorescence were more sensitive to competition for light than the weight of the frame and floral abortion. Components determining the oil and kernel extraction showed no response to density when assisted pollination was used, but with the successful introduction of insect pollinators, oil and kernel extraction increased with planting density as a result of an increase in fruit per bunch.

Twenty one strains of Brassica carinata were tested under irrigated, rainfed and saline soil conditions and five strains under late sowing conditions along with five varieties each of B. juncea, B. napus and B. campestris var. yellow sarson, brown sarson and toria. Brassica juncea had the largest seed and oil yield under irrigated conditions but B. carinata performed much better than other species under late sowing, rainfed and saline soil conditions.