BREEDING WITH WILD RELATIVES: SCREENING SORGHUM PRE- BREEDING MATERIAL FOR PERFORMANCE AND ADAPTATION IN WESTERN KENYA

Abstract

Sorghum (Sorghum bicolor L. Moench) is an essential cereal for food security and resilience to climate change in Kenya. Continuous utilization of traditional landraces coupled with biotic and abiotic stresses have led to low sorghum yields in western Kenya. Further, the limited amount of genetic variation within the cultivated sorghum gene pool has constrained breeding efforts to overcome major production constraints. The development of new sorghum diversity through the use of wild and weedy species to enrich cultivated sorghum germplasm offers a novel and promising opportunity. The objectives of this study were to: (i) determine agronomic performance of sorghum wild relatives (SWR) in western Kenya (ii) determine genotype by environment interaction and stability of SWR in selected sites in western Kenya. (iii) Introgress genes from sorghum wild relatives into selected cultivated Kenyan Lines and determine the amount of heterosis in the sorghum crosses for selected traits. Twenty-one (21) SWR plus 4 checks were evaluated for performance in a 5 X 5 alpha lattice replicated thrice at Kibos, Nyabisawa, Sega and Godkwer agro-ecologies. Genotype by Environment interaction (GEI) and yield stability was determined using the Additive Main Effect and Multiplicative Interaction (AMMI) model across 4 environments. A total of 9 crosses were developed and evaluated at Kibos site using RCBD replicated thrice.Results showed significant variation at p < 0.05 among sorghum tested for majority of traits studied at the four sites which indicated the existing genetic, difference among the entries tested. The mean grain yield was higher in Kibos (2.1t/ha) in comparison to the other 3 sites mainly because of higher soil fertility levels while Nyabisawa site had the least mean yield of 1.6 t/ha. Entries 466 and 588 were the most promising for grain yield. The AMMI ANOVA for grain yield showed significant effects for genotype (G), environment (E) and GEI. For grain yield, the differences among the (E) accounted for 8.7% of the total variation while the G and GEI accounted for 42.1% and 15.4% respectively of the variation. About 20% of SWR Genotypes were more stable

than the checks. A total of 9 crosses were developed and 85% of them exhibited high mid- parent heterosis on majority of the phenotypic traits tested which showed transgressive

inheritance and hence possibility for further breeding using the SWRs. The best grain yield per plant heterotic cross combination identified was (RUC26 X 586) which expressed high positive heterosis over both mid and better parent (194.27% and 137.89%, respectively).

This cross also showed significant positive average heterosis and heterobeltiosis for yield- contributing components. This study has identified and developed potentially diverse new

sorghum germplasm which can be useful in further breeding and yields improvement in western Kenya.