Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: IMPROVING GENETIC PREDICTIONS FOR DAIRY ANIMALS USING PHENOTYPIC AND GENOMIC INFORMATION Title: Modeling Nuisance Variables for Prediction of Service Sire Fertility

Authors
item Kuhn, Melvin
item Hutchison, Jana
item Norman, H

Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: March 18, 2008
Publication Date: July 1, 2008
Repository URL: http://hdl.handle.net/10113/18873
Citation: Kuhn, M.T., Hutchison, J.L., Norman, H.D. 2008. Modeling Nuisance Variables for Prediction of Service Sire Fertility. Journal of Dairy Science. 91(7):2823-2835.

Interpretive Summary: This research analyzed what factors should be accounted for when comparing the fertility of bulls in artificial insemination using on-farm breeding records. Variables identified were herd, year, month, and lactation of breeding along with stage of lactation, level of milk yield, service number, cow age, and effect of the cow.

Technical Abstract: The purpose of this research was to determine which (available) nuisance variables should be included in a model for phenotypic evaluation of US service sire conception rate (CR), based on field data. The primary means used to compare alternative models was to split data into records for estimation and set aside records, compute predictions from the model of interest using the estimation data, and then compare predictions to bulls' average CR in the set aside data. A future yr was used as set aside data. Breedings for estimation were from January 1, 2003 to June 30, 2005 while set aside records spanned July 1, 2005 to June 30, 2006. There were 3,613,907 breedings in the estimation data and 2,025,884 in the future yr data. Bulls were required to have a minimum of 50 records for estimation and 100 breedings in a minimum of 30 herds in the future yr data to be included for comparisons, which resulted in 803 bulls for comparison. Only matings with known outcomes were included in either data set. Correlations and mean differences were the main statistics used to compare models. Nuisance variables considered were management groups based on herd-yr-season-parity-registry (HYSPR) classes, yr-state-mo, cow age, DIM, and various combinations of lactation, service number, and milk yield. Preliminary analyses led to selection of standardized lactational milk yield as the production variable for consideration. Preliminary analyses also indicated that modeling the quantitative independent variables as categorical factors provided better bull fertility evaluations than modeling the effects as linear and quadratic covariates, because the quantitative variables did not have either a linear or quadratic relationship with CR. Two general strategies for management groups were tested, one where HYSPR groups were required to have an absolute specified minimum number of records and a second where groups were combined across registry, season, and parity subclasses until a minimum group size was achieved. Minimums tested were 3, 5, 10, and 20. For the 5, 10, and 20 minimums, another alternative tested was to allow a herd-yr into the evaluation, after combining, provided it had minimum of 2, 5, or 10 records, respectively. Combining groups to a target size of 20 and allowing a herd-yr into the evaluation provided it had a minimum of 10 breedings maximized correlation with future yr CR and was chosen as the management grouping strategy for implementation. Combining groups implies that some groups have multiple seasons as well as parities which was the reason for consideration of yr-state-mo and lactation as additional factors; yr-state-mo was utilized, rather than mo alone, to allow mo effects to vary by yr as well as region of the country. The final nuisance variables selected for inclusion in the model for prediction of service sire CR were, in addition to HYSPR, yr-state-mo, lactation, service number, milk yield, cow age at breeding, and the cow effect, partitioned as permanent environment and breeding value. This model maximized correlation with future yr CR, minimized mean square error, and had a mean difference of essentially 0.

Last Modified: 10/21/2014
Footer Content Back to Top of Page