Expected efficiencies of mating designs for reselection of parents

Abstract

Expected efficiencies, in terms of genetic gain from reselection of parents (backwards selection), were compared for hierarchical mating, factorial, partial factorial, modified half diallel, and partial diallel crossing designs and polycrosses; this was done in parallel with a separate study of expected efficiencies for advanced-generation (forwards) selection, assuming a fixed-resources model of 100 parents and 10 000 offspring. The present study considered a single-trait case, with variable numbers of crosses per parent, varying heritability (h2) levels (h2 = 0.1, 0.2, and 0.5), and varying ratios of specific combining ability to general combining ability variance (0, 0.5, and 1). Compared with the case of forwards selection, the relative efficiencies of the different designs were generally similar. Two notable exceptions were the comparative inefficiency of small, disconnected factorial sets for backwards selection and the generally high (but not always maximal) efficiency of polycrosses for this purpose.