IMPROVING PREDICTION OF NATIONAL EVALUATIONS BY USE OF DATA FROM OTHER COUNTRIES

Authors: Powell, Rex; Norman, H; BANOS, G. - SWEDISH UNIV OF AG SCI

Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/15/1999
Publication Date: N/A
Citation: N/A

Interpretive Summary: Improvements in genetic evaluation procedures often are assessed by their ability to predict genetic estimates from added or independent data. The International Bull Evaluation Service (Interbull) has combined national bull evaluations into international evaluations since August 1994, but the benefit of using data from multiple countries is still being questioned. To determine if including multinational data through the Interbull evaluation process improved prediction of future national evaluations, national and international Holstein bull evaluations from Canada, France, Germany, Italy, The Netherlands, and the United States were compared for milk, fat, and protein yields. Later national evaluations were assumed to be improved estimates of true genetic merit. Correlations with later national evaluations generally were larger for earlier national evaluations than for Interbull evaluations, probably because of the stronger part-whole relationship between earlier and later national evaluations. However, standard deviations of difference from later national evaluation were lower for Interbull evaluations than for earlier national evaluations. For bulls with substantial increases in daughter numbers nationally and internationally, all correlations were higher and standard deviations of differences were lower for Interbull evaluations compared with earlier national evaluations. These results indicate that including multinational data improved prediction of future national evaluations and suggest that usage restrictions for Interbull evaluations can be relaxed or eliminated. Recent importation of European semen into North America will increase the importance of data from other countries in improving the accuracy of U.S. genetic evaluations.

Technical Abstract: National and international bull evaluations from Canada, France, Germany, Italy, The Netherlands, and the United States were examined to determine the usefulness of data from other countries in increasing the accuracy of prediction of national evaluations for milk, fat, and protein yields. Holstein evaluations from the International Bull Evaluation Service in February 1995 and the national evaluations that were data for the international evaluations were compared for predictive ability of national evaluations that were data for international evaluations in February 1998. The later national evaluations were assumed to be improved estimates of true genetic merit. Correlations with later national evaluations generally were larger for earlier national evaluations than for international evaluations, probably because of the larger part-whole relationship between earlier and later national evaluations. However, standard deviations of difference from later national evaluation were lower for international evaluations than for earlier national evaluations, which suggested improved prediction from inclusion of multinational data. For bulls with substantial increases in daughter numbers nationally and internationally, all correlations were higher and standard deviations of differences were lower for international evaluations compared with earlier national evaluations. Inclusion of multinational data improved the prediction of future national evaluations, especially for countries that import genetics of dairy cattle.