Towards assessing indirect genetic effects in dairy cattle

Abstract

Social interactions in a dairy herd may impact an individual’s production, e.g., milk yield. These interactions can have a genetic component, so-called indirect genetic effects (IGE). IGEs contribute to heritable variation in other species, but studies on IGEs in cows are limited. Knowledge is needed on appropriate methods to monitor social interactions in cows. We evaluated with simulations whether we can estimate IGEs in cows. We used milk yield as an example trait, and we assessed how herd size, direct and indirect genetic correlations, and magnitude of IGE affected the variance component estimations and breeding value accuracies. We investigated the importance of knowing the contact intensity and direction by either including or ignoring them in the estimation model. Additionally, we investigated how random noise added to the intensities would affect the estimates and breeding values. The estimated variance components were unbiased and precise for scenarios with different herd sizes of 50, 100, or 200 cows and direct and indirect genetic correlations of either − 0.6, 0, or 0.6. The IGE breeding value accuracies were 0.55–0.65 for cows when the IGE explained 30% of the phenotypic variance. When the magnitude of the IGE became smaller, the precision of the estimated variances became lower. The IGE breeding value accuracies were 0.16–0.52 for cows when the IGE explained 1.5–15% of the phenotypic variance. Using imprecise intensities or ignoring the contact direction underestimated the variance of the indirect effects, and the breeding value accuracies became lower. Ignoring the variation in intensities in the model led to unbiased variance component estimates but a larger residual variance and lower breeding value accuracies than if we used imprecise intensities. We could estimate IGE in dairy cattle with high accuracy and precision in a simulated population of 10,000 phenotyped cows distributed over 50–200 herds. A smaller IGE variance led to less precise estimates and lower breeding value accuracies. Ignoring information about the intensity of contact in the model would be worse than using imprecise intensities, and using technology that also monitors the direction of contact may be beneficial to estimate variance components of IGE.