Prediction of fatty acid composition using milk spectral data and its associations with various mid-infrared spectral regions in Michigan Holsteins

Fatty acid composition in milk is not only reflective of nutritional quality but also potentially predictive other attributes (e. g. including the cow's energy balance and its relative output methane emissions). Furthermore, a higher ratio long-chain to short-chain fatty acids or mean carbon number has been associated with negative dairy cows, whereas enhanced properties have generally levels unsaturation. We set out directly compare Bayesian regression strategies partial least squares for prediction various using Fourier-transform infrared spectrum data on 777 samples taken from 579 cows 4 Michigan herds between 5 90 d milk. identify those spectral regions that might be whether level unsaturation contribute strength these associations. These associations were based adaptively clustered windows wavenumbers mitigate distorting effects severe multicollinearity marginal involving individual wavenumbers. In general, methods, particularly variable selection method BayesB, outperformed cross-validation accuracy both groups. Strong signals wavenumber BayesB well distributed throughout mid-infrared spectrum, 910 3,998 cm−1. Carbon appeared linearly related 38 moderately highly predicted within 2,286 2,376 2,984 3,100 cm−1, nonlinear determined 1,141 1,205; 1,570 1,630; 1,727 1,768 However, no such detected Spectral where there significant relationships association may useful targets inferring proportion acids, hence balance.