Implementing genomic selection in sour passion fruit population

Abstract

Sour passion fruit is an economically important tropical fruit crop with little explored genetic potential. This study aimed to provide breeders with essential estimates of genomic breeding values in economically important traits in passion fruit, using Bayesian models which may contribute to the implementation of Genomic Selection and develop new strategies for the continuity of sour passion fruit breeding programs. For this, the following Bayesian models were tested using 183 polymorphic marks: Bayesian Ridge regression, Bayes A, Bayes B, Bayes B2, Bayes Cπ and Bayesian Lasso for estimation of genomic breeding values. To achieve this, ninety-five full-sib progenies derived from the third cycle of recurrent selection of the sour passion fruit (Passiflora edulis Sims.) at Universidade Estadual do Norte Fluminense Darcy Ribeiro—UENF were used and eight fruit yield (number of fruit, total yield, mean fruit weight, fruit length, fruit width) and quality(percent pulp, skin thickness, soluble solids) traits were assessed. The Bayes Cπ (smaller deviance information criterion) yield the best genetic predictions for almost all traits. Genetic correlations in this study indicate that the number of fruit can be used as a proxy for yield. The values of genomic heritability obtained were high and ranged from 0.62 to 0.76 and predict accuracy ranged from 0.55 to 0.75, so we can to speculate that the use of two replicates in the present study was an adequate amount to obtain phenotypic mean, which was used to adjust the genomic prediction model.