Marker assisted selection for the improvement of two antagonistic traits under mixed inheritance

A Monte Carlo simulation was used to investigate the potential of Marker As sisted Selection (MAS) in a multiple-trait situation. Only additive effects were considered. The base population was assumed to be in linkage equilibr ium and, next, the population was managed over 15 discrete generations, 10 males and 50 females were chosen out of the 100 candidates of each sex. Per formance for two traits was simulated with an overall heritability of a giv en trait equal to 0.25 or 0.10 and the overall genetic correlation between traits was generally equal to -0.4 except in one case where it was equal to 0. The model involved one biallelic QTL, accounting for 10 or 20% of the g enetic variance of a given trait, plus polygenes. Initial allelic frequenci es at the QTL were generally equal to 0.5 but in one case were equal to 0.1 and 0.9. A marker with 120 different alleles in the 60 founder parents was simulated in the vicinity of the QTL. Two values of the recombination rate between these two loci were considered, 0.10 and 0.02. The genetic evaluat ion was based on a multiple-trait BLUP animal model, accounting (MAS) or no t (conventional BLUP) for marker information. Two sets of simulations were run: (1) a "missing data" case, with males having no record for one of the traits, and (2) a "secondary trait" case, with one trait having a weight in the aggregate genotype 4 times less than the other trait and the QTL actin g only on this secondary trait. In the first set, evaluation methods were f ound to mainly affect the accuracy of overall genetic values prediction for the trait with missing data. In comparison with BLUP, MAS led to an extra overall genetic response for the trait with missing data, which was strongl y penalised under the conventional BLUP, and to a deficit in response for t he other trait. This more balanced evolution of the two traits was obtained , however, at the expense of the long-term overall cumulated response for t he aggregate genotype, which was 1 to 2.5% lower than the one obtained unde r the conventional BLUP. In the second set of simulation, in the case of lo w initial frequency (0.1) of the QTL allele favourable to the secondary tra it, MAS was found to be substantially more efficient to avoid losing this a llele than BLUP only when the QTL had a large effect and the marker was clo se. More benefits should be expected from MAS with more specific applicatio ns, such as early selection of animals, or by applying dynamic procedures i .e. letting the respective weights to QTL and polygenic values in the selec tion criterion vary across generation.