Blood metabolic biomarkers predicted from milk spectra are heritable in Holstein transition cows

Blood metabolic parameters provide useful information about cows’ metabolic status, especially in early lacta tion. Determining these traits is however costly and requires invasive blood samplings. We used milk mid-infrared spectroscopy to predict blood concentration of β-hydroxybutyrate (BHB), non-esterified fatty acids (NEFA), urea, cholesterol (CHO) and glucose (GLU). Blood and milk samples were collected from 680 cows between 5 and 35 days in milk in 34 multi-breed Italian farms during the morning milking. Partial least square regression was performed for prediction models development and coefficients of determination in cross-validation were 0.64 for BHB, 0.64 for NEFA, 0.85 for urea, 0.53 for CHO, and 0.33 for GLU. Subsequently, we estimated heritability (h2) with a linear animal model using blood traits obtained from a prediction set (8,277 early-lactation Italian Holstein cows, 374 herds). BHB exhibited the highest h2 (0.13±0.03), while NEFA the lowest (0.03±0.01). The h2 of the remaining traits was low too, ranging from 0.04 to 0.08. The minimum and the maximum coefficient of additive genetic variation were 3.1% (GLU) and 20.8% (BHB). Although the prediction accuracy and precision do not allow for punctual determination, predicted blood traits can be still used for selection purposes to guide genetic progress towards a reduced incidence of metabolic diseases.