Estimation of genetic parameters in two maize recurrent selection programs

Abstract

<p>Recurrent selection is a cyclical breeding procedure that focuses on improving the mean performance of a population by increasing the frequency of favorable alleles, while maintaining adequate genetic variability for continued selection. Iowa Stiff Stalk Synthetic (BSSS) is a maize population that has undergone continuous recurrent selection for more than 50 years as the base population for two independent selection programs (intra- and inter-population);This study was designed to estimate the mean performance and important genetic parameters in BSSS populations per se after: seven cycles of half-sib (HS) progeny selection, six cycles of S2-progeny selection, and 11 cycles of reciprocal recurrent selection (RRS). A Design II (cross-classified) mating design was constructed to give direct estimates of additive and dominance variance in the individual populations. Fourteen sets of 4 male by 4 female matings for each of the populations were evaluated in a randomized incomplete block (Reps/Sets) experiment grown in multiple environments;RRS in BSSS produced the most effective mean performance responses for grain yield in the populations per se. S2-progeny selection in BSSS did not perform up to theoretical expectations. Variance component estimates showed very little significant change for the majority of the traits with all three selection methods. In general, the largest portion of the total genetic variance for all traits consisted of additive variance. However, dominance variance for grain yield seems to be an important component in BSSS germplasm. Genetic variance by environmental interaction variance components were generally of unimportant magnitude. The importance of dominance variance in BSSS provides for more effective response from selection with testcross selection methods that can take advantage of dominance genetic effects. With adequate levels of available additive genetic variance remaining and high heritability estimates for most of the traits of interest, future response from selection should be achieved with each selection method.</p>