Associations between milk infrared-predicted plasma biomarkers of stress resilience and fertility in dairy cattle: insights for enhancing breeding programs and herd management

Fertility is a crucial aspect of dairy herd efficiency and sustainability. Among factors influencing fertility in dairy cattle, metabolic stress and systemic inflammation of animals are of main relevance, especially in the postpartum stage when ovarian activity begins and cows are inseminated. Our study aimed to infer the associations between milk infrared-predicted blood biomarkers of stress resilience and fertility traits, namely the interval from calving to first service (iCF), days open (DO), and the pregnancy rate at first service (PRF) in a multi-breed population of 89,097 dairy cows. The blood metabolites (15 blood biomarkers related to hepatic damage and function, oxidative stress, and inflammation/innate immunity) were predicted using milk Fourier-transform mid-infrared (MIR) spectroscopy. A gradient boosting machine (GBM) approach with leave-one-batch-out cross-validation (R² range from 0.45 to 0.82) was implemented to an independent calibration database of 1,367 lactating cows reared in 5 herds. Calibration equations were then applied to a population database of 1,799,186 MIR milk spectral data, that were then merged with fertility data collected by the Breeders Federation of Alto Adige (Bolzano province, Italy) generating a final database of 285,145 records. The 2 databases were merged according to the milk test day (and thus the MIR spectrum) closest to the date of insemination. The interval fertility traits were fitted as the hazard of either receiving the first service after calving at time t for iCF or becoming pregnant after calving at time t for DO in a Cox proportional-hazards model. Statistical analyses were performed including in the model the number of lactations, year of calving, and herd as fixed effects. The independent effect of the MIR-based predictions of metabolites was also included with each metabolite evaluated separately and discretized into 7 levels based on percentiles. Pregnancy rate at first service, on the other hand, was analyzed using logistic regression and the same explanatory variables. The metabolites linked to liver function and damage, such as aspartate aminotransferase, total bilirubin, and alkaline phosphatase, had a relevant influence on iCF and DO in terms of the hazard ratio (HR). Relevant results were also obtained for the biomarkers related to oxidative stress and inflammation/innate immunity. Specifically, increasing levels of ceruloplasmin, total reactive oxygen metabolites, and advanced oxidation protein products resulted in a relevant decrease in the HR of cows becoming pregnant. The logistic regression analysis did not reveal any significant effect of the aforementioned biomarkers on PRF, indicating that the effects of the stress response mainly concern the resumption of the ovarian cycle after calving. The results for the associations of the predicted biomarkers of the stress response with iCF and DO were consistent with expected physiological patterns. In conclusion, the predicted biomarkers investigated revealed to be promising novel phenotypes for assessing animal health and welfare, in the view of enhancing fertility in dairy cattle also through selective breeding, thus improving the overall efficiency of dairy herds.