Abstract
A breeding program was simulated in this study. Two alternative ways of selecting the breeding population for the following generation was compared. Phenotypic selection, which means to select just on the individual performance, and combined index selection, which means selection on predicted breeding value for each individual obtained by weighting family average and individual phenotype, were compared. The plant number (testing resource) and gene diversity (status number, Ns) were kept constant, but the breeding population size was variable and chosen for maximizing gain for the particular breeding scenario. At low and medium heritability phenotypic selection was inferior to combined index selection. Only when heritability was high phenotypic selection was as efficient (generation 1) as or more efficient (generation 5) than combined index selection. This contrasts to earlier studies done under constant breeding population size, where selection methods appeared similar. The advantage in gain of combined index selection is usually at a larger breeding population size. At limited heritability and breeding population size the difference is considerable. When breeding population size was kept rather small (<100), and the heritability limited, combined index selection can result in slightly higher gain than phenotypic selection at the same gene diversity, but this was at the cost of a much larger breeding population. Phenotypic selection and combined index selection appears as rather similar for many cases in this simple model used in this study. Considering other advantages with phenotypic selection, it may often be regarded as a competitive alternative.