Genetics of social interactions in livestock

Social interactions among individuals are wide-spread, both in natural and domestic populations. In livestock, the social environment provided by group or herd mates impacts performance and welfare of individuals. Because the social environment is partly the result of the genes of social partners, it may have a genetic component and can respond to selection. This expectation is confirmed by selection experiments for socially affected traits, such as mortality due to cannibalism in chicken, which have yielded responses that conflict with predictions from classical models. Here I review the quantitative genetics of socially affected traits, focussing on genetic improvement. I review the evidence for social genetic effects in livestock and fish, present methods and optimal experimental designs for the estimation of social effects, and show optimal selection strategies for the improvement of socially affected traits. Results show that social effects can contribute substantially to the heritable variation available for response to selection. In contrast to classical models, the phenotypic variance does not present an upper limit to the heritable variance that can be used for genetic improvement. Social genetic effects can be estimated with linear mixed models, using data on populations consisting of many groups. Ideally, those groups consist of a limited number of families. The optimum design for variance component estimation has two families per group. The use of groups consisting of related members greatly increases the response to selection in socially affected traits. The optimum selection index is a simple linear combination of the EBVs for direct and social effects. When such EBVs are not available, an index of individual and group performance can be used to increase response to selection. Alternatively, one may use groups composed of single families. One important category of socially affected traits is the disease status of an individual (infected / not-infected) for an infectious disease. Transmission of infection is a social process, depending for example on the contact behaviour of individuals, and on susceptibility and infectivity of individuals. Results show that breeding schemes effective for social behaviours are also effective for selection against infectious diseases. Particularly the use of groups composed of related individuals greatly accelerates response to selection. Genetic analysis of disease status data (0,1), both for GWAS and (G)BVE, can be performed using generalized linear (mixed) models with a complementary log-log link function.