Abstract

The aim of this thesis was to investigate different approaches for international genetic evaluations of dairy sires for udder health traits. For this purpose, deregressed national genetic evaluations were used of Holstein milk somatic cell (SC) from 12 countries, clinical mastitis (CM) from 3 of these countries as well as Ayrshire fore udder attachment from 9 weakly linked countries. Deregression and multiple-trait across-country genetic evaluations (MACE) worked satisfactorily for a lowly heritable trait (h2=0.02) such as CM. The across-country genetic correlations ranged from 0.47 to 0.97 (median = 0.88) for SC and from 0.59 to 0.89 for CM. International genetic udder health evaluations enable more efficient global selection than selection based on national evaluations. The predictive ability and reliability of international genetic evaluations were improved by including multiple traits per country (CM and SC) in international genetic evaluations. The advantage of analysing SC and CM simultaneously was most noticeable for young bulls with daughters in countries not using the within-country correlation structure in their national genetic evaluation and for prediction of CM. The average reliability of international breeding values for young bulls with most daughters in Sweden increased from 18 to 25% on Nordic CM scales. Estimates of the same genetic correlation differed considerably (up to 1.03 units) depending on estimation strategy when genetic ties between the respective pairs of countries were weak. The use of prior genetic parameters improved the predictive ability of international breeding values for weakly linked populations, especially when the uncertainty of the (co)variances was considered, i.e. in fully Bayesian MACE. For young bulls on foreign scales, the predictive ability of such analyses was 11 to 15% higher than for traditional MACE. The recommendations are that deregressed national genetic evaluations should be used as dependent variables in MACE, that multiple-trait MACE should be used for international genetic udder health evaluations, and that prior genetic correlations should be considered. The implication is that global selection against CM will be more effective, which can reduce the use of antibiotics globally.