Association of genomic predictions for fertility traits with reproductive outcomes in dairy cows under 2 estrus detection and timed artificial insemination programs

Genomic prediction of daughter pregnancy rate (GDPR) and cow conception rate (GCCR) are fertility traits developed to help improve selection for reproductive performance. Although these traits overlap, their denominator can vary, and programs with different strategies combining estrus detection (ED) and timed AI might experience different associations with these traits (e.g., shorter interval for first service for high GDPR quartiles, but not for GCCR). The objectives were to assess days from calving to first service (TP1), AI at ED (AIE), pregnancy at the first service (P1), pregnancy loss for the first service (PL), and number of services to conception (NSFC) in 2 reproductive programs (RepP) combining variable ED length and timed AI (TAI) and their relationship with GDPR and GCCR. Holstein cows from a single farm were randomly allocated to 2 different programs, the short RepP (n = 982) or long RepP (n = 942). In the short RepP, cows were enrolled in a Presynch-Ovsynch (PGF_2α: 36 ± 3 and 50 ± 3 DIM) followed by ED and AI from 50 ± 3 to 62 ± 3 DIM. Cows not detected in estrus underwent in the Ovsynch (GnRH: 62 ± 3, PGF_2α: 69 ± 3, GnRH: 71 ± 3 and TAI :72 ± 3 DIM). In the long RepP, cows received a PGF_2α at 50 ± 3 DIM followed by AI at ED up to 81 ± 3 DIM. Cows with no ED were enrolled in an Ovsynch (GnRH:82 ± 3, PGF_2α:89 ± 3, GnRH:91 ± 3 and TAI:92 ± 3 DIM). Cows were categorized into quartiles (Q1 to Q4) considering their GDPR (qGDPR) and GCCR (qGCCR). Statistical analyses included logistic regression used for AIE, P1, and PL; Poisson regression for the NSFC; and linear regression for TP1. Models included AIE, P1, PL, NSFC, and TP1 as dependent variables, with RepP and qGDPR for GDPR and RepP and qGCCR for GCCR as independent variables. Time to pregnancy was analyzed using Cox's proportional hazard model. The short RepP had a shorter TP1 (short = 64.3 vs. long = 72.1) and fewer NSFC than the long RepP (short = 2.9 vs. long = 3.1). The long RepP had a higher AIE (short = 45.2% vs. long = 73.2%). The P1 tended to be greater in the short than the long RepP (short = 33.7% vs. long = 30.0%). Cows in Q4 (GDPR and GCCR) had lower TP1 and NSFC, greater AIE and P1 compared with the lowest quartiles (Q1). Interactions between RepP and GDPR were present for TP1, AIE, and P1, but no interactions were observed between RepP and GCCR. The short program had a shorter interval from calving to pregnancy than the long program. Cows in the highest quartiles for GDPR and GCCR had shorter intervals from calving to pregnancy and higher pregnancy hazards than the lowest quartiles. The current study revealed that cows enrolled in long RepP that relied on longer ED intervals had lower reproductive outcomes than short RepP cows, and cows ranked in the highest GDPR and GCCR quartile for most responses had better outcomes independent of the RepP used.