Use of generation mean scaling for estimation of variance components in maize breeding
Genetic studies were carried out to estimate the variance components in maize using six crosses, A x F, A X 64, A x 8, F X 64, F x 8 and 8 x 64. The four parental inbred lines (A, F, 64 and 8), the F1 , F2 and backcross generations of these crosses were grown in a randomized complete block design with three replicates at National Agricultural Research Centre, Kitale; Western Agricultural Research Centre, Kakamega; Soy and Sabwani, between April and October 1991. Diallel and generation mean analyses were done based on data from twelve randomly selected plants in each plot for plant height, ear height, grain yield, ear lengtP, 200-grain weight and kernel row number. Lodging percentage was determined for the whole plot. According to the dial leI analysis, the additive gene effects were relatively more important for all the traits except grain yield, 200-grain weight and lodging ,percentage. Non- additive gene effects played insignificant- roles in manifestation of the traits studied. The generation mean analysis revealed that both additive and dominance genetic variances were important for grain yield, ear length, 200-grain weight, plant height, ear height and kernel row number. The variability attributable to dominance genetic variance was much greater in all cases. Lodging percentage was mainly governed by additive genes. In most crosses additive x additive and dominance x dominance genetic variances were more important than additive x dominance genetic variance at all sites. In all cases the manifestation of various genetic variance components varied according to crosses and experimental sites. Among the parental cultivars, inbred lines F and 64 had high general combining ability estimates for yield and yield component traits at most sites, and hence were the most promising for yield improvement. The crosses 8 X 64, A x F, A x 8 and F x 64 also had high specific combining ability estimates for grain yield and could be used Lh further selection and hybrid development programmes. Cross 8 x 64 was the most heterotic at most sites for grain yield, plant height, ear height and ear length while crosses A x F and F x 8 were the most heterotic at most of the sites for characters, 200-grain weight and kernel row number respectively. Double cross (A x 64) x (F x 8) had the best predicted yield performance at Kitale, Kakamega and Soy.