Genetic evaluation for somatic cell score with a test day model for multiple lactations.

A multiple lactation model for test day data was applied to predict genetic merit for somatic cell scores of Canadian Holsteins. The model for genetic evaluation included a fixed effect for herd test date, fixed regressions on functions of days of lactation, random effects of permanent environment within lactation, random genetic effects on animal, and residual effects. Records from the first three lactations were used and treated as different traits. Procedures for this model, developed for national genetic evaluation for somatic cell score in Canada, were found to be practical. Use of starting values from the previous genetic evaluation reduced the number of rounds necessary to reach convergence. Test day models were compared with several single-trait models based on lactation average of somatic cell score in terms of computing efficiency and ranking of animals. Differences between EBV from the test day model and EBV from a repeatability model for lactation average were small for bulls with many daughters, but differences were large with EBV from a single-trait model for first lactation average of somatic cell count. Association were desirable for EBV for somatic cell score with EBV for some udder conformation traits, but undesirable for EBV for milk and protein yield.