Author
 |
Powell, Rex |
|
Submitted to: American Dairy Science Association Proceedings
Publication Type: Abstract Only Publication Acceptance Date: 6/20/1999 Publication Date: N/A Citation: N/A Interpretive Summary: Technical Abstract: The goal of international genetic evaluation of dairy bulls is to accurately estimate bull merit across countries. Implicit in that statement is that results do not give unfair advantage to one country over another. Bias is difficult to measure, as true merit is never known. Expected genetic value of full brothers should be the same on each national scale. However, a large family or a number of families are needed for Mendelian sampling to average out. International Bull Evaluation Service (Interbull) Holstein evaluations from February 1995 were used to define full-brother families spanning countries. These two data sets used either the current procedure, including use of genetic correlations, or one with unity genetic correlations. Each set was analyzed by a model fitting evaluation on a given national scale by absorbing full-brother family and producing country solutions. Countries were included if they had at least 20 bulls in full-brother families that spanned countries. There were 1950 bulls from 7 countries in 756 full-brother families. Single-country data on an additional 2695 families with 6612 bulls were included to improve the estimate of within-family variation. Country differences were not significant with unity genetic correlation. For data with genetic correlations used, country differences were not significant for fat or protein but were significant (P<.05) for milk on US, French, and Italian scales, suggesting an artifact rather than a country bias. Significance disappeared with a requirement of 50 daughters. Thus, it appears that Interbull procedures are not advantaging particular countries but the combination of genetic correlations and limited sampling can produce results suggesting bias. This relationship needs further study. |
