A Novel Mating Design to Optimize Genomic Selection Efficiency for Commercial Corn Breeding

Abstract

In many commercial corn (Zea. mays L.) breeding programs, lines are selected based only on general combining ability (GCA) during first-year trials. Selection for specific combining ability (SCA) is delayed until later trials, resulting in many unevaluated hybrid combinations. My objective was to determine whether a reciprocal testcross mating design enables simultaneous selection for GCA and SCA, while maintaining the same resources typical in first-year trials. Suppose B1 and B2 are Iowa Stiff Stalk Synthetic (BSSS) lines, whereas N1 and N2 are non-BSSS lines. In a reciprocal testcross design, progeny of B1 × B2 are testcrossed with N1 and N2, and progeny of N1 × N2 are testcrossed with B1 and B2. In 2019, grain yield and moisture of 1,642 hybrids from 10 BSSS and non-BSSS populations were measured at a median of three locations per hybrid across the upper Midwest. In 2020, a validation set consisting of 146 hybrids that were not tested in 2019 were evaluated at a median of five locations per hybrid. The GCA and SCA values were estimated using genomewide prediction with 11,000 SNP markers and the level of dominance was estimated using a subset of these markers. The sizes of training populations were kept constant, and the cross-year predictive ability of the reciprocal testcross design was compared with that of a standard, nonreciprocal design. Including SCA in the models marginally increased predictive abilities for reciprocal designs and the reciprocal designs produced higher predictive abilities than the nonreciprocal designs. The median level of dominance for grain yield was 1.08 indicating complete dominance. The results indicated that the reciprocal testcross mating design combined with genomic prediction could efficiently enable simultaneous selection for GCA and SCA earlier in a breeding pipeline.