Title: RETROSPECTIVE ANALYSIS OF THE ACCURACY OF CONVERSION EQUATIONS AND MULTIPLE-TRAIT ACROSS COUNTRY EVALUATIONS OF HOLSTEIN BULLS USED INTERNATIONALLY
Weigel, K - UNIV OF WISCONSON
Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: December 8, 1999
Publication Date: N/A
Interpretive Summary: The International Bull Evaluation Service (Interbull) provides a service that combines dairy bull evaluation data across countries using multiple- trait across country genetic evaluation (MACE) procedure in order to provide genetic estimates for all bulls on all national scales. These evaluations have been available long enough that it is now possible to compare MACE evaluations with other estimates from 4 years ago as predictors of 1999 genetic evaluations for six countries. Bulls studied had evaluations in 1999 in a particular country, but did not in 1995. Comparisons were among the MACE evaluations from 1995 and two kinds of conversion equations applied to data from other countries at that time. Correlations of the three estimates with evaluations in 1999 were .80. The MACE estimates were slightly better than estimates from conversion equations for bulls with more data in the importing country in 1999. The lack of improvement in accuracy of prediction with the MACE procedure points to the need for improvement in the procedure but may also be due to limitations in data structure.
In 1995, the multiple-trait across country genetic evaluation (MACE) procedure replaced regression-based conversion equations as the preferred method for international genetic comparisons of dairy bulls. In the present study, February 1999 estimated breeding values of 632 foreign Holstein bulls used in Canada, Germany, Italy, The Netherlands, Sweden and the US were compared with January 1995 predictions from data of the exporting country. January 1995 predicted breeding values for each importing country were from the MACE procedure, conversion equations based on the MACE evaluations, and conversion equations based on the Wilmink method. Correlations between 1999 estimated breeding values in the importing countries and 1995 predictions from foreign data were 0.799-0.804 for all methods. The MACE procedure led to selection of different bulls, because bulls were allowed to rank differently in each country, but no significant increase in accuracy of selection was observed. The lack of improvement in accuracy of prediction by MACE was most likely due to limitations in data structure. International genetic comparisons are largely driven by data from a relatively small number of proven bulls with exported semen. Data from siblings and more distant relatives provide only weak, indirect genetic links between countries, and inclusion of such data seems to provide a minimal improvement in accuracy. Limitations in data structure might be alleviated by methods that define environments by climate or management factors rather than country borders.