Bayesian Analysis of Test Day Milk Yields in an Unbalanced Mixed Model Assuming Random Herd-Year-Month Effects


Abstract: The main environmental effects in a mixed model are comparison or contemporary group effects or more precisely herd-year-month of calving subclass effects. The controversial subject of much discussion about the choice between treating contemporary group effects as fixed or as random has not still been settled in dairy cow evaluation. However, these effects are usually treated as fixed since nonrandom associations between sires and herds may lead to biased predictions if herd-year-month effects are accounted for as random. On the other hand, treating herd-year-month effects as random would increase the effective number of daughters or the information with which an animal is being evaluated, and as a result of this, prediction error variance decreases. The main purpose of this paper is to demonstrate the implementation of the Gibbs sampler with data on test day milk yields of dairy cows in an unbalanced mixed half-sib sire model assuming random herd-year-month effects. An analysis of this kind employing the Gibbs sampler with a very large data set containing records on 23.873 cows and 689 sires is carried out for the first time. Posterior expectations of genetic and phenotypic parameters and functions of them are obtained from test day milk yields. The results of this study are then compared with those of the previous study using the same data set but assuming fixed her-year-month effects.

Keywords: test day milk yields, Gibbs sampling, genetic and phenotypic parameters

Full Text: PDF