|Clay, John - DRMS|
Submitted to: Journal of Dairy Science
Publication Type: Abstract Only
Publication Acceptance Date: March 23, 2004
Publication Date: July 25, 2004
Citation: Kuhn, M.T., Hutchison, J.L., Clay, J.S. 2004. Prediction of service sire fertility [abstract]. Journal of Dairy Science. 87(Suppl. 1):412. Technical Abstract: Estimated relative conception rates (ERCRs) are phenotypic predictions of service sire (SSR) fertility computed by Dairy Records Management Systems. ERCRs are based on first service 70-d non-return rate using a linear model that includes the fixed effects of herd-year-month of service, parity, days in milk at breeding, birth year of SSR, and energy-corrected milk yield and the random effects of animal (cow), permanent environment (PE) of the cow and SSR by herd interaction. The ERCR, then, is simply the SSR solution. This research investigated an alternative linear model using all services. The model included the fixed effects of herd-year-season of service, state-month of breeding, parity, and stud; the linear and quadratic effects of milk yield, days-in-milk at breeding, age of cow at breeding, age of SSR at breeding and the linear effects of expected future inbreeding of the daughters (EFI) and inbreeding coefficient of the SSR and also the random effects of animal (SSR), cow and PE of the SSR. A sire's predicted conception rate (CR) was calculated as the sum of his animal, PE, and stud solutions plus the regression on his age, his EFI, and his own inbreeding coefficient. Data for prediction was the result of breeding (0,1) for 530,380 Holstein cows calving between 1995 and 1997. For unconfirmed records, the last service was considered successful unless the cow was coded as sold for reproductive reasons. To assess accuracy of the predictor, mean CR in 1998 was calculated for each SSR and correlated with predicted CR. For bulls with at least 100, 500, and 1000 matings in the predictor, the correlation between predicted and actual CR in a future year was 0.16, 0.26, and 0.36, respectively. For bulls with breedings to at least 1000 cows, in contrast to just 1000 services, the correlation between predicted and future CR was 0.52. Further research will focus on other alternative models, appropriate data edits, and methods to handle the last service where pregnancy status is not known with certainty.