Exploring Different Model Structures for the Genetic Evaluation of Dairy Bull Fertility

The aim of study was to investigate different models for the evaluation of dairy bulls for male fertility. A dataset containing single insemination records performed on Brown Swiss cows and heifers reared in Eastern Italian Alps was used. The outcome variable (successful/ unsuccessful) was analyzed as binary trait. In the first step raw conception rate was computed for each service sire, as the mean of the outcome of all his insemination events. In the second step Bayesian threshold sire models were implemented via Gibbs sampling. Different models increasing in complexity were fitted, in order to obtain variance components and sire solution estimates. Results showed that genetic variance for direct effect(s) on conception rate is low (repeatability=0.014; heritability=0.009-0.073) and raw conception rate is poorly related to solutions from prediction models. If the service sire is a diagonal effect, rank correlations with raw conception rate are about 0.81-0.84, while those decrease to 0.74-0.78 if service sires are related by a relationship matrix, and is null (-0.01-0.06) if is the sire of the service sire to account for the direct genetic effect. Service sire fertility (sire of service sire effect) has been proven to be a different trait to embryo survival (service sire effect) giving rank correlation of 0.11-0.24. In conclusion phenotypic and genetic differences in male fertility among the service sires exist, and should be monitored thought a reliable evaluation system