Estimation of non-linear relationships between traits in livestock.

The estimation of genetic parameters such as heritabilities or genetic correlations is usually based on the assumption of linear relationships between the traits involved. However, a model with nonlinear trait interrelationships should better match the true state in biology in general and therefore also yield realistic parameter estimates, provided that an adequate estimation procedure is available. The concept for evaluating nonlinear genetic relationships was based on two different approaches, first, on the offspring-parent polynomial regression, second, on the paternal halfsib analysis theory. The usefulness of both approaches was demonstrated by means of simulated data, afterwards they were applied on the relationship between milk yield and protein percentage and milk yield and type traits in dairy cattle. In both cases, significant nonlinearity was found.