Enhancing population and family selection accuracy with statistical genetics models accounting for epistatic effects for wheat breeding

Few studies have investigated the effect on the genotypic value of wheat (Triticum aestivum L.) families with the adoption of the additive and epistatic (additive × additive) relationship matrix. The objective of this study is to select F_2:3 families of wheat by means of three statistical genetics models (without pedigree information, additive, and additive plus additive × additive epistatic) and to evaluate the selection rank between the traditional model and the model with best fit of families for recombination and for deriving progenies. The experiment was composed of a total of 880 F_2:3 families of tropical wheat, from 56 populations conducted by the genealogical method, which came from a full diallel involving the cultivars BRS 254, BRS 264, and BRS 394, CD 1303, Tbio Aton, Tbio Ponteiro, Tbio Duque, and Tbio Sossego. The pedigree matrix was calculated, obtaining approximately 20 generations of ancestry of the parents. The data were analyzed in three genetic-statistical models: Model 1—without information on family relationship; Model 2—computing the additive relationship matrix; and Model 3—including the additive and epistatic (additive × additive) relationship matrix. Using the additive and epistatic (additive × additive) pedigree matrix has a significant effect on most traits. The selection revealed families of populations with potential to be used in recombinations: BRS 254/CD 1303, Tbio Ponteiro/BRS 394, and BRS 394/Tbio Ponteiro, with genetic value to derive progenies: BRS 254/Tbio Aton, Tbio Aton/Tbio Duque, and BRS 394/Tbio Aton, and with both attributes: BRS 254/CD 1303, BRS 394/Tbio Ponteiro, and Tbio Sossego/BRS 264.