Bayesian perspective in the selection of cowpea genotypes in trials of value for cultivation and use

Abstract:

The objective of this work was to select, under the Bayesian perspective, cowpea (Vigna unguiculata) genotypes that meet high phenotypic adaptability and stability, in the state of Mato Grosso do Sul, Brazil. Data from four experiments, conducted in a randomized complete block design, were used, in which grain yield of 20 semiprostrate cowpea genotypes was evaluated. To represent non-informative prior distributions, probability distributions with high variance were used; and, to represent informative prior distributions, a metanalysis concept was adopted using information from previous studies. The comparison between the prior distributions was done using the Bayes factor. The Bayesian approach provides greater accuracy in the selection of semiprostrate cowpea genotypes, with high phenotypic adaptability and stability assessed by the Eberhart & Russell methodology. Based on the informative priors, the MNC99-507G-4, TE97-309G-24, MNC99-542F-7, and BR 17-Gurguéia genotypes are classified as with high adaptability to favorable environments. The TE96-290-12G, MNC99-510F-16, MNC99-508G-1, MNC99-541F-21, MNC99-542F-5, and MNC99-547F-2 genotypes have high adaptability to unfavorable environments.

Index terms:
Vigna unguiculata; Bayes fator; genotype x environment interaction; metanalysis; informative prior.