Résumé
Stochastic simulations were used to compare genetic progress and inbreeding depending on whether full sib matings were allowed or not when applying individual selection. Fifteen selection cycles were considered, using a single pair mating design. Two levels of heritability (h²) were studied (0.1; 0.5) as well as four different proportions of selected animals (2, 5, 10 and 20%), four population sizes (1000, 2,000, 4,000 and 10,000 candidates) and two levels of inbreeding depression (0 and 5% decrease of phenotype for 10% increase in inbreeding). Fifty replicates were generated for each set of parameters. The simulations revealed that avoiding full sib matings became less and less relevant as population size and proportion of selected animals increased and as heritability and impact of inbreeding on phenotype decreased. In large populations (10,000 candidates), excluding full sib matings was not worthwhile at all whereas in small populations with high selection intensity, it was a first useful way to manage genetic variability. This can be illustrated in two extreme situations. With 1000 candidates, h²=0.5, 5% inbreeding depression and 2% selection pressure, the phenotypic mean after 15 generations of selection was 18.9% greater when full-sib matings were excluded rather than allowed, while inbreeding was 33% lower. Conversely, for a large population (10,000 candidates), when the proportion of selected animals was over 5%, there was no significant difference in phenotypic mean after 15 generations of selection whether excluding full-sib matings or not, while the difference in inbreeding was below 0.07% per generation.