Bayesian estimation of heritability and genetic gain for subsets of genotypes evaluated in a larger set of genotypes in a block design

Abstract

There are situations in which a number of inbred lines are found grouped into classes, depending on their origin and phenology. Interest in such situations lies in the estimation of genotypic variations between the genotypes from individual groups, whereas all genotypes are evaluated in a single trial conducted in a randomized complete block design (RCBD) or an incomplete block design (IBD). The objective of this study was to apply a Bayesian approach to estimate genotypic variation, heritability, and genetic advances within individual groups of groups of genotypes. A set of 360 barley genotypes were evaluated in a two replicate alpha-design with blocks of 10 plots each. The standard frequentist method to estimate variance components was carried out by the restricted maximum likelihood method on the days to flower data from the IBD as well as by an RCBD by ignoring the incomplete blocks. The Bayesian approach with selection of best priors was implemented for the estimation. We noticed a substantial difference in the estimates of the various genetic parameters across the groups. The estimation of variations between the genotypes from individual groups (RCBD or IBD) is needed as the basis of many agricultural research or plant breeding/agronomy trials. The Bayesian approach uses broader inference framework to integrate the prior information on parameters with the likelihood of the current data. Therefore, the Bayesian approach presented here for estimation of heritability and genetic gain for subsets of genotypes evaluated in a larger set of genotypes in the block design is recommended for use in similar situations.