Field application of milk infrared-based equations to predict blood energy metabolites and minerals: effects of cattle breed and dairy system interactions

During the postpartum period, especially over the first six weeks, dairy cows require high energy due to production demands, which is often not fully met by their diet. Assessing blood biomarker concentrations [e.g. non-esterified fatty acids (NEFA) or beta-hydroxybutyrate (BHB)] using Fourier transform mid-infrared (MIR) spectroscopy on milk has proven to be a valuable tool for monitoring cow health. Its application at the field level is particularly attractive, as MIR is already routinely used for milk quality assessment within milk recording schemes. This study evaluates the associations between predicted blood biomarkers and various sources of individual variation, with a focus on breed-specific responses in relation to days in milk (DIM), parity, and dairy system, using data collected from herds within the Parmigiano Reggiano Consortium. MIR predictions of six energy-related metabolites (glucose, cholesterol, NEFA, BHB, creatinine, and urea) and three minerals (Ca, P, and Mg) were analysed using a linear mixed model. The dataset comprised 1 343 298 records from 174 556 cows of the Holstein, crossbred, Brown Swiss, Simmental, and Reggiana breeds across 941 farms. Farms were classified into dairy systems using a non-hierarchical k-means clustering based on farming practices and herd characteristics. Four dairy systems were considered: two traditional systems (Apennines and Po Plain) and two modern systems (one with total mixed ration feeding, one without). The fitted model included the effects of breed, dairy system, parity, DIM, and the interactions of breed with DIM, parity, and dairy system, as well as month, year, animal, and herd. Results from the interaction between breed and DIM (significant for all traits, P < 0.05) revealed breed trend differences, especially between Brown Swiss and Holstein. The interaction between breed and dairy system (significant for all traits except glucose) revealed greater variation in metabolite concentrations in Reggiana breed compared to the more stable metabolic profiles observed in breeds specialised in milk production. The interaction between breed and parity (significant for all traits except Ca) indicated that multiparous cows appeared to manage energy demands more efficiently. The results highlight breed differences in metabolic blood profiles, reflecting distinct physiological strategies to meet nutritional demands. MIR-predicted energy-related metabolism mirrored expected physiological patterns, supporting the robustness of the prediction equations in this population. In addition to providing a cost-effective and routinely accessible tool for monitoring metabolic status in dairy cows, MIR predictions also represent promising novel phenotypes for selective breeding.