Predictors of reproductive outcomes in seasonal-calving, pasture-based lactating dairy cows

Abstract

The objective of this retrospective observational study was to evaluate associations between predictor variables and reproductive performance outcomes in seasonal-calving, pasture-based dairy cows. Individual cow data were available from 5,845 lactations across 21 herds in 2022 and 2023. Cow data records included genomic proofs, milk production, calving and reproduction events, as well as health and estrus alerts from automated activity monitoring systems. The breeding start date (BSD) was defined for each herd, and reproductive outcomes included the proportion of cows submitted for artificial insemination (AI) within 21 d of BSD (SR21), first service pregnancy per AI (P/AI1), BSD to conception interval (BSD-CI), and the proportion of cows pregnant in 42 d (P42) and 84 d (P84). Several cow-level factors were evaluated in univariable and multivariable models: parity (1, 2, 3, ≥4), number of automated estrus alerts (AEA) before BSD (0, 1, ≥2), fertility subindex (FI; high [>€93.58], medium [€93.58-68.14], low [<€68.14]), severity of automated health alerts (AHA) within 50 DIM (none, mild [index >65], severe [index ≤65]), calving assistance (unassisted vs. assisted/twins), milk yield at 80 DIM, DIM at BSD, and semen type (sex-sorted vs. conventional). Several factors were associated with reproductive performance. Cows with ≥2 AEA before BSD had greater SR21 (94% vs. 85%), greater P/AI1 (52% vs. 42%), and shorter BSD-CI (23 vs. 27 d) than cows with no AEA. High- and Medium-FI cows had greater reproductive performance compared with Low-FI cows (P/AI1: 49%, 48% vs. 44%; BSD-CI: 23, 23 vs. 25 d, respectively). Cows with no or mild AHA in early lactation had greater P/AI1 (49%) and shorter BSD-CI (23 d) than those with severe AHA (44% and 25 d). Primiparous cows had greater reproductive performance (P/AI1: 50%; BSD-CI: 23 d) than multiparous cows (≥4 lactations: 45%, 24 d). The DIM at BSD was significant across all fertility outcomes, whereas milk yield was not. To identify subgroups with large differences in reproductive performance potential, variables associated with reduced reproductive performance were assigned risk score values for poor reproductive performance proportional to β-coefficients from the multivariable model. Cows were subsequently stratified by risk score into low (score ≤2, 25% of data), medium (3-8, 54% of data), and high risk (≥9; 21% of data). High-risk cows were more likely to be multiparous, have no AEA, fewer DIM at BSD, and had early-lactation health alerts. High-risk cows had lower SR21 (74%), P/AI1 (37%), P42 (51%), and P84 (71%) and reduced hazard of pregnancy (0.50, 95% CI: 0.46 to 0.55) than low-risk cows (SR21: 96%, P/AI1: 61%; P42: 80%, P84: 90%). Combining multiple predictors enabled identification of subgroups with more meaningful fertility differences than any single factor alone. These findings support the future development of targeted reproductive management strategies and decision-support tools for pasture-based, seasonal-calving systems. Further work is needed to evaluate the utility of this grouping approach for real-time on-farm decision-making.

Keywords: fertility; seasonal-calving; technology.