Soybean (Glycine max (L.) Merr.) occupies an important position in the world economy of the feedstock of high quality protein and vegetable oils. However, its production is threatened by, Asian soybean rust (ASR), caused by the rust fungus Phakopsora pachyrhizi Syd. & P. Syd. This fungus is highly dependent on environmental conditions, has a wide range of hosts, and evolves rapidly into novel races, making it difficult to control. In addition, most commercial varieties are susceptible to rust, the rust has already developed resistance to triazole fungicides, and most small-scale farmers cannot afford expensive systemic fungicides to control the disease. The use of resistant varieties is the most viable, long-term option to manage ASR, especially in the small-holder soybean farming sector. This study was therefore designed to undertake the following goals: (i) to identify farmers’ preferred varieties and desired traits, their knowledge of ASR, and other key constraints affecting soybean production in Kenya; (ii) to evaluate soybean accessions for rust resistance, and to determine the correlation of rust resistance with other agronomic traits; (iii) to determine the mode of inheritance for ASR resistance and selected agronomic traits; and (iv) to determine yield stability of soybean advanced lines at multiple sites in Central and Eastern Kenya. To understand farmers’ preferred varietal characteristics, knowledge of ASR and other key constraints to soybean production, a survey was conducted using a structured questionnaire in the major soybean growing areas of Kenya. The farmers preferred local varieties because of their desirable characteristics, which included high yields, early maturity, drought tolerance and seed availability. Although the majority of the participating farmers expressed a willingness to grow improved varieties, financial limitations, seed unavailability and lack of information were the major barriers to their use of improved varieties. High yield, early maturity, adaptability and grain quality were the traits that most farmers sought in an ideal soybean variety. Knowledge of the cause of ASR was limited, and its occurrence was largely attributed to environmental factors, poor soil fertility conditions, poor agronomic practices, physiological maturity and specific species of weeds. Their investments in control methods were minimal due to a lack of technical knowledge, poor access to fungicides, and limited resources. Other constraints faced by soybean farmers included: lack of access to grain markets; lack of knowledge in processing and utilization of soybean grain; the unavailability of seeds; losses to pests and diseases; the lack of inputs such as fertilizers; frequent dry spells; and low yielding varieties. A total of 110 soybean accessions were evaluated for their rust reactions and correlations with selected agronomic traits. These included plant introductions possessing single rust resistant genes (Rpp1-4), tolerant lines, gene bank accessions, commercial varieties and advanced lines. Soybean genotypes varied significantly in their reactions to rust severity, sporulation, lesion type and area under disease progress curve (AUDPC) values. Genotypes possessing Rpp4 (G10428) and Rpp2 (G8586) resistant genes, and non-characterized genotypes MAK BLD 11.3, GC 00138-29 and Namsoy 4M, were the most resistant accessions, as indicated by low rust severity scores, low AUDPC values, red brown lesions and low sporulation scores. Other genotypes with known resistant genes including G7955 (Rpp3), G58 and Tainung 4 (Rpp1), a few tolerant lines, and one advanced line (BRS Sambaiba) were moderately resistant. All the other advanced lines, commercial varieties, gene bank accessions and collections from the farmers’ fields were highly susceptible to rust. Rust severity was positively correlated with rust sporulation, indicating that reduction of sporulation made a significant contribution towards rust resistance. An F2 population was generated from a half diallel mating design, involving 4 resistant, 2 moderately resistant and 2 susceptible genotypes selected as parents. The F2 populations along with their parents were evaluated in two environments to determine the type of gene action for rust resistance and other quantitative traits in soybeans. The results revealed that both general combining ability (GCA) and specific combining ability (SCA) were significant for most of the traits studied, indicating that both additive gene action and non-additive gene action played a major role in the inheritance of rust resistance and selected agronomic traits. The GCA/SCA ratio was close to unity for rust severity, rust sporulation, days to flowering, days to maturity and plant height. This indicated that additive gene action played a more significant role in the inheritance of these traits than non-additive gene action. Non-additive gene action was only predominant for soybean grain yield. Parental lines G10428, G8586 and Namsoy 4M were the best general combiners for improving rust resistance across the environments. The most promising parents for early flowering were G7955, G8586 and G58. Parent Maksoy 1N was the best general combiner for early maturity while parents Maksoy 1N, G58, G7955 and Nyala contributed effectively towards reduced plant height. Yield stability analysis was conducted for 30 genotypes in 6 environments, using additive main effects and multiplicative interaction (AMMI), genotype main effect and genotype x environment interaction (GGE) biplot analyses. Genotypes 916/5/19 and G7955 were identified as the high yielding and most stable across the environments. On the other hand, genotypes BRS MG46 and Sable were high yielding but unstable and specifically suitable for the environments EM2 and MW2, respectively (both environments have long rainy seasons). Environment EM2 was identified as the most discriminating and representative among the six environments. Environments IG1 and MW1 (short rainy seasons) were less informative on genotypes tested, as confirmed by short environment vectors. Environment EM1 was better for discriminating genotypes but was a poor representative of the test environments, hence it should only be utilized for developing specifically adapted genotypes. Further analysis using GGE biplot approach grouped the environments into three putative mega-environments in Central and Eastern Kenya. Overall, this study established the need to educate farmers on the cause of ASR, to develop ASR resistant varieties, and to incorporate farmers’ desired traits in the breeding programme, especially by the use of participatory breeding approaches. The resistant and moderately resistant genotypes identified in this study could be used as sources of resistant genes to develop ASR resistant varieties in Kenya. This study also established that genetic improvement for ASR resistance and selected agronomic traits in soybeans is possible based on the use of recurrent selection breeding procedures that result in the accumulation of additive gene effects. Selection of late segregating generations would be effective for soybean grain yield improvement. This study identified potential parents for ASR resistance and selected agronomic traits, but they require further breeding to improve on farmers’ desired traits. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2012.