Associations of reproductive indices with fertility outcomes, milk yield, and survival in Holstein cows.

The study is part of a research effort investigating potential associations between genomic variation and fertility of Holstein cows. The objective was to compare the reproductive performance of Holstein cows in 3 categories of 2 reproductive indices (RI) that were developed for the allocation of cows in a ranking for potential fertility, based on the predicted probability of pregnancy. The associations between categories of the developed indices and multiple fertility variables in a large multistate population of Holstein cows were tested. In addition, we analyzed associations among the RI categories with milk yield and survival. Based on phenotypic information from individual cows, 2 reproductive indices (RI1 and RI2) were developed, representing a predicted probability that a cow will become pregnant at first artificial insemination postpartum, as a function of explanatory variables used in a logistic model. Data from a total of 11,733 cows calving in 16 farms located in 4 regions of the United States (Northeast, Midwest, Southeast, and Southwest) were available. Cows were enrolled at parturition and monitored weekly for reproductive events, health status, milk yield, and survival. To develop the indices, potential significant effects were initially tested by univariate analyses. Effects with P ≤ 0.05 were offered to the multivariate analysis, and the final models were determined through backward elimination, considering potentially significant interactions. The final model for RI1 included the random effect of farm and a complement of significant fixed effects as explanatory variables influencing a pregnancy outcome: (1) incidence of retained fetal membranes; (2) metritis; (3) clinical endometritis; (4) lameness at 35 days in milk (DIM); (5) resumption of postpartum ovulation by 50 DIM; (6) season of calving; and (7) parity number. The model for RI2 included (1) parity number; (2) body condition score at 40 DIM; (3) incidence of retained fetal membranes; (4) metritis; (5) resumption of postpartum ovulation by 50 DIM; (6) region; (7) subclinical ketosis; (8) mastitis; (9) clinical endometritis; and (10) milk yield at the first milk test after calving; as well as the interaction effects of postpartum resumption of ovulation by 50 DIM × region; mastitis × region; and milk yield at the first milk test after calving × parity number. Multivariate logistic regression, ANOVA, and survival analysis were used to test the correspondence between the resulting RI and individual fertility, milk yield, and survival from the population. To facilitate the analyses, the resulting RI values were categorized as low for cows in the lowest quartile, medium for cows within the interquartile range, or high for cows in the top quartile. We found consistent agreement between categories of the predicted RI and the measures of fertility and survival collected from individual cows. We conclude that the proposed RI represent a viable approach to refine the allocation of cows into potential low- and high-fertility populations.