Effects of genetic groups on estimates of variance components for a maternally influenced trait in the Spanish Limousine

Residual maximum likelihood (REML) estimates of variance components for standardized 210-day weights were obtained in the Limousine breed from Spain under different animal models. All models were applied including (Gdm) or not (G00) genetic groups for the direct and the maternal components. Genetic groups were defined according to sex of progenitor and generation in two different approaches depending on whether the full pedigree information or reduced pedigree was used. For the most complex model, which included direct, maternal and permanent environmental variances as well as the direct-maternal covariance, genetic groups were imposed either to the direct (Gd0) or to the maternal (G0m) component treating genetic groups as covariates without the QP transformation. For the Gdm case (genetic groups for both direct and maternal effects) with the full pedigree information, the estimates of the direct genetic variance and the genetic covariance between direct and maternal genetic effects showed the largest reduction, while the estimate of maternal genetic variance was only slightly reduced, with respect to G00, a model without genetic groups. This result might be due to a less important variability among group effects for the maternal component, probably due to a lower selection intensity performed for this component. However, for the Gd0 case, estimates of genetic variances and especially the direct-maternal covariance were largely overestimated with respect to the Gdm case. Inclusion of direct and maternal grouping was deemed necessary for this population. Reduction of pedigree information did not seem to have a large impact on variance component estimation. Nevertheless, deeper pedigree information together with a correct definition of the generation to assign genetic groups is needed to accommodate the effect of selection on direct and maternal components of this trait, in this population.