Upgrading a maize breeding program via two-cycle genomewide selection: Same cost, same or less time, and larger gains

Abstract

It would be helpful to have genomewide selection schemes that increase genetic gains within the same cost and same time as currently required in breeding programs. My objective was to assess if and how two-cycle genomewide selection can increase maize (Zea mays L.) genetic gains in a cost-neutral and time-neutral manner. As per industry sources, per-unit costs were assumed equal for genotyping, doubled haploidy, testcrossing, and phenotyping at one location. The values of genomewide prediction accuracy (rMG) (0.20–0.80) and entry-mean heritability (0.60) mimicked those for maize yield. In simulations, two-cycle genomewide selection conducted as follows led to large gains: (a) genomewide selection among 200 to 240 Cycle 0 F2 plants, (b) genomewide selection among 200 to 240 Cycle 1 doubled haploids, and (c) phenotyping of 35 to 45 doubled haploids at 12 locations. The largest gains from two-cycle genomewide selection were 124 to 178% of the largest gain from phenotypic selection and 112 to 135% of the largest gains (given the same rMG) from one-cycle genomewide selection. When 8 to 10 Cycle 0 F2 plants were intermated to form Cycle 1, the gains were numerically greater with two-cycle genomewide selection than with phenotypic selection at least 78% of the time. The best resource allocations for two-cycle genomewide selection were tantamount to substituting genomewide predictions for first-year phenotyping, thereby reducing the time required to release hybrid cultivars. The results indicated no reason for commercial maize breeders to rely solely on phenotypic selection during inbred development.