Skip to main content
ARS Home » Research » Publications at this Location » Publication #119278
Title: STABILITY AND BIAS OF YIELD EVALUATION FOR HOLSTEIN BULLS IN ARTIFICIAL- INSEMINATION SERVICE

Author
item Powell, Rex
item Norman, H

Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/7/2001
Publication Date: N/A
Citation: N/A

Interpretive Summary: Concern about declines in estimated genetic merit of dairy bulls has been a continuing challenge. Accuracies of USDA genetic evaluations for bull genetic predicted transmitting abilities (PTA) for yield of milk, fat, and protein were examined. PTA from January 1995 were compared to May 2000 USDA genetic evaluations. Average PTA for the 14,012 artificial insemination (AI) bulls born in 1980 or later were stable. As expected, changes were larger for bulls having lower initial estimated accuracy and adding more new data. Emphasis was on the 540 bulls in AI service whose mean PTA changes were near zero. However, bulls in the top 10% for milk, fat, and protein declined substantially, but that bias essentially disappeared for bulls progeny tested by the major AI organizations. The standard deviation for change in PTA milk for top 10% bulls was about a third larger than expected. Again, considering only bulls progeny tested by the major studs removed the excess variation, indicating that other bulls are the source o bias and lack of stability. These bulls progeny tested outside the traditional AI organizations appeared biased in initial PTA, largely due to the inflated estimate of daughter performance. The hypothesis of inflated parent average data producing inflated early PTA was not corroborated. Stability of the base and unbiasedness of bulls progeny tested by disinterested parties were demonstrated, but concern continues regarding other bulls. To the extent that the bias is likely due to preferential management of early daughters, it will not be easy to eliminate all the decline.

Technical Abstract: Holstein bull predicted transmitting abilities (PTA) for yield of milk, fat, and protein were compared for January 1995 and May 2000, beginning and end of the 1995 genetic base definition. Changes in PTA were assessed regarding a number of factors. Overall, PTA for the 14,012 artificial insemination(AI)bulls born in 1980 or later were stable for mean but varied slightly more than expected based on increase in data. As expected, changes were larger for bulls having lower initial reliability (accuracy) and increasing more in reliability over the 5.3 yr. Emphasis was on the 540 bulls in active AI service whose mean PTA changes were near zero at -8, -1.3, and -0.5 kg for milk, fat, and protein. Bulls in the top decile for milk, fat, and protein declined 47, 2.2,and 2.2 kg, respectively, but that bias essentially disappeared for bulls progeny tested by the major AI organizations. The standard deviation for change in PTA milk for top decile bulls was 190 kg where 140 was expected. Again, considering only the major stud sampled bulls (essentially sampling code S) removed the excess variation, indicating that other bulls are the source of bias and lack of stability. The non-S bulls appeared biased in initial PTA, largely due to the inflated estimate of daughter performance. The hypothesis of inflated parent average data producing inflated early PTA was not corroborated. Stability of the base and unbiasedness of S-code bulls were demonstrated, but concern continues regarding other bulls. To the extent that the bias is likely due to preferential management of early daughters, it will not be easy to eliminate all the decline.