Integrating three genetic dimensions relating to piglet birth weight: direct and maternal effects on the mean and genetic control of residual variance

Uniformity of production traits is desired for different traits in livestock species, including the uniformity of within-litter birth weights in piglets. Birth weight (BW) in pigs is associated with increased vitality and survival until weaning. However, as uniformity of BW increases, the importance of initial weight decreases as competition between piglets decreases. The aim of this study was to estimate the direct and maternal genetic components of BW, jointly with the maternal genetic component of the residual variance for within-litter BW, and their genetic correlations. We used two distinct datasets of Swiss Large White pigs: 1) the experimental farm dataset and 2) the commercial farms dataset, comprising 43,135 and 23,313 records of individual piglet birth weight, respectively. For statistical analysis, the heteroscedastic (or canalising selection) model was used. This model assumes that both the mean BW and the residual variance are affected by systematic and random effects, with the residual variance being heterogeneous and partially under genetic control. Despite the best fitting model was the most complex one including both genetic effects for the mean trait, the results indicated that direct genetic effects, or correlations with such effects, are negligible. The genetic environmental variance for BW ranged between 0.071 and 0.131 for experimental farm and 0.037 to 0.094. The genetic correlation between the mean BW and its variability was always positive and ranged between 0.149 and 0.307 for the experimental farm and between 0.220 and 0.589 for the commercial farms. It is thus sufficient to model BW and its variability by including only the maternal genetic effect for both traits. In addition, even though moderate genetic correlations exist between the mean and the variance of BW, focusing selection on BW uniformity within litters would be preferable to creating a selection index for both traits simultaneously.