Reciprocal Recurrent Genomic Selection for Total Genetic Merit in Crossbred Individuals

Abstract

Genomic Selection (GS: Meuwissen et al, 2001) aims to provide genomic estimated breeding values (gEBVs) as criteria for selection by exploiting associations between genetic markers and phenotypes. This process generally assumes additive genetic merit, implicitly regressing individual phenotype on the number of specific alleles carried, separately for each locus. gEBVs are thus based on estimated allele substitution effects at QTL or at linked marker loci. This paper aims to give proof of concept that genomic selection can be used to breed for dominance effects as well as additive effects, without having to fit both these effects in statistical models. This leads to increasing heterosis in crossbred animals though genomic selection within the contributing parental lines. Should this work under real conditions, the benefits could be substantial, particularly in the pig and poultry industries. In this paper the target is genetic merit in crossbred individuals. This has been well addressed by Ibánẽz-Escriche et al. (2009), but without consideration of non-additive QTL effects. This paper aims to exploit within-locus dominance deviations as well as additive effects while selecting within purebred parental lines. This can be done by estimating allele substitution effects on crossbred phenotypes, separately for each parental line. Consider the average effect of an allele substitution in gametes of line A, evaluated on its phenotypic impact in the AxB crossbred genotype. This depends on allele frequency in gametes contributed from line B, and is thus the same as the average effect of an allele substitution within line B, assuming no epistatic effects.