Participation of farmers in plant breeding programmes has been reported to increase breeding efficiency. Farmers’ participation bridges the gap between variety development and dissemination and provides an opportunity for farmers to select varieties they prefer. The breeders on the others hand learn more about the farmers’ preferences and the environment in which the new varieties will be grown. However, the advantages of participatory breeding can best be realized when farmers’ indigenous technical knowledge (ITK) and experience complement the breeder’s scientific knowledge and skills. Cassava (Manihot esculenta Crantz) is a clonally propagated crop grown in diverse environments by small scale farmers for subsistence. Information on the roles of farmers and breeders at various stages of breeding and their ability to effectively participate in breeding programmes is limited. The objectives of this study were to determine: (1) cassava farmers’ preferences, production constraints and systems; (2) farmers’ selection criteria of cassava varieties; (3) genetic inheritance of farmer preferred traits; (4) how farmers and breeders complement each other at all stages and activities of cassava breeding. Participatory rural appraisal was conducted in three purposefully sampled districts of western Kenya based on ethnicity and agro-ecology. The results reveal that cassava is predominantly grown by small scale farmers with mean land size of 1.6 ha mainly under mixed cropping system for subsistence. The storage roots are eaten either after boiling or processing to flour. The majority of farmers (over 60%) are aware of the improved varieties but adoption rate is low (18% in some districts). The effects of pests and diseases, and the lack of high yielding varieties, capital, land, and disease free planting material are the most important constraints to cassava production. Farmers prefer tall, high yielding varieties that are resistant to diseases and pests, early maturing and long underground storability of harvestable storage roots. The districts surveyed significantly differed in popularity of utilization methods, traits preferences and relative ranking of the production constraints indicative of differences in ethnicity and agro-ecology. Three farmer groups from the three districts selected in western Kenya were used to study farmers’ variety selection criteria based on their own indigenous technical knowledge (ITK). The groups evaluated 15 (10 landraces and five improved) popular cassava varieties with concealed identities on their farms. The results revealed that farmers have effective methods of selecting varieties for most of their preferred traits. However, ITK alone cannot be used to evaluate all the important traits, such as cyanide content. The genetic inheritance of farmer preferred traits was determined through a genetic study. Six landraces and four improved varieties popular in western Kenya were crossed using the North Carolina mating design II to generate 24 full-sib families. The 24 families, represented by 40 siblings each, were evaluated at two sites, Kakamega and Alupe research station farms, in a 24 x 40 a-lattice design. General combining ability (GCA) and specific combining ability (SCA) mean squares were significant (P<0.05) for all traits evaluated except dry matter content and cyanide content. However, non-additive gene action predominated over additive gene for cassava mosaic disease (CMD) resistance, height to first branching, total number of storage roots per plant and fresh storage root yield in all environments. The best crosses were not necessarily obtained from parents with high general combining ability confirming the presence of non-additive gene action. The best performing parents per se did not necessarily have high GCA effects implying that selection based on the per se performance of parents may not always lead to development of superior hybrids. The clonal evaluation trial (CET) was established at Alupe research station and evaluated by the breeder and farmers from two districts independently. Three selection criteria were tested to determine the most appropriate approach to selection of varieties that meet both farmers’ and breeder’s preferences. The selection criteria were; farmers’ independent selection index (SI) derived from farmers’ selection criteria from each district, breeder’s negative selection and independent SI, and a participatory SI which combines farmers’ and breeder’s selection criteria. There was 14% overlap among the top 100 varieties selected by farmers from all districts and the breeder when independent SI were used. However, there was 49% overlap among the top 100 varieties selected by farmers using participatory SI and the breeder’s SI. The farmers and the breeder have a role to play in the variety development process. Varieties with traits preferred by both the farmers and the breeder are likely to enhance breeding efficiency and effectiveness. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2011.