Genetic aspects of immunoglobulins G concentration measured with gold standard and predicted from infrared spectra in bovine colostrum

The concentration of immunoglobulins G (IgG) is the criterion usually adopted to classify the quality of the colostrum adminis-tered to calves. Although the reference analysis of IgG (g/L) is time-consuming and expensive, no studies have evaluated the ability of infrared spectroscopy to predict colostrum IgG concentration so far. In this study, colostrum was collected on 693 Holsteins within 6 h after calving following a specific protocol. Samples were analysed through the reference analysis (radial immunodiffusion), which was carried out according to the manufacturer’s instructions (Triple J Farm, Bellingham, US) after dilution of colostrum (1:5 v/v) in pure water. Near-infrared spectra (400–2500 nm) were collected on all samples using the DS2500 (Foss, Hillerød, Denmark). After spectra quality editing, the final dataset accounted for 685 samples. The calibration set included 195 samples representative of the 9 herds and cows from parity 1 to 6, and the best prediction model for IgG (RMSE =14.2 g/L; R2 = 0.84) was reached through 20-fold cross-validation and multiplicative scatter correction and second derivative as mathematical treatment. Finally, IgG was predicted in the validation set which accounted for the remaining 490 samples (RMSE =19.4 g/L; R2 = 0.73). Variance and covariance components were estimated for both reference and predicted IgG (validation set) using a linear model which included the fixed effects of parity (5 levels: 1, 2, 3, 4, and 5 + 6), the season of calving (4 levels), year of calving (2 levels), and herd (9 levels), and the random effects of additive genetic animal (6714 individuals in the pedigree) and the residual. Means (coefficient of genetic variation) of reference and predicted IgG were 92.4 g/L (14.8%) and 90.5 g/L (15.1%), respectively. Heritabilities of IgG were 0.18 ± 0.15 (reference) and 0.21 ± 0.15 (predicted). The phenotypic correlation between the two traits was positive and strong (0.86 ± 0.01), as well as the genetic correlation (0.99 ± 0.11). Overall, results indicated that IgG can be predicted from colostrum spectra with moderate accuracy and the genetic correlation between the reference and predicted IgG is close to 1. Therefore, IgG predictions can be potentially exploited as phenotypes to improve the IgG content of bovine colostrum, leading to potential practical positive consequences on calf health.