Résumé
Sorghum (
Sorghum bicolor (L.) Moench) is the staple crop of Burkina Faso in West Africa where guinea race landraces are grown in low-input cropping systems. National and international breeding programs have had little success in disseminating modern varieties based on high yield potential caudatum or kafir race breeding materials, mostly introduced from other countries. It has been argued that the breeding objectives were not sufficiently oriented towards the farmers’ needs and preferences, and that they did not target the prevailing growing conditions of Burkina Faso. The objectives of the present article are (i) to identify and examine farmers’ selection criteria for sorghum varieties in the Centre-West of Burkina Faso, (ii) to compare these criteria with the breeder's agronomic observations and standard practices, and (iii) to show how the criteria of both farmers and breeders can be effectively integrated into the early stages of a pedigree breeding program. These objectives take into consideration gender differentiation, consistency of selection criteria and the interrelationship of measured quantitative traits, as well as the impact of these traits and criteria on the final selection. The present paper is based on a pedigree breeding program that was carried out over three years (2001–2003) in two villages of Burkina Faso. Options for integrating farmers’ selection expertise with that of the breeders’ were examined. Participatory selection was initiated with 53 F3/F4 progenies in field trials managed by farmers using rating and voting exercises. The breeders measured and analysed the agronomic data while the farmers evaluated a large number of progenies by means of their three most important selection criteria and a general appreciation. Farmers’ initial choices remained consistent in the selection exercises (voting) performed in subsequent years. The farmers’ methods for defining traits turned out to be more multivariate than the breeders’ formal understanding of these same traits. This was especially so for the criteria of grain quality, earliness, and productivity for which the farmers’ definition encompasses factors such as flour yield and stability across environments. However, rating results between farmer groups were variable. A disagreement between female and male ratings was especially found for the grain quality traits. The results clearly show that farmers can effectively select for traits on the basis of progeny and single plants while pursuing specific agronomic aims such as adaption. Subsequent yield improvement schemes will thus be more efficient in terms of selection intensity related to grain yield. The study is showing the way for breeders to adjust their selection criteria to suit the basic needs of small-scale farmer in semi-arid regions of sub-Saharan Africa.