Use of Raman spectroscopy as a rapid tool to discriminate milk deriving from different pasture-based diets and breeds in a seasonal, spring-calving dairy production system

This work examined the use of Raman spectroscopy to study the effects of 2 pasture-based feeding systems—perennial ryegrass pasture and multispecies swards—on bovine milk compositional parameters. The milks of both pasture-based diets were examined across 2 breeds, Holstein-Friesian and Jersey × Holstein-Friesian, throughout an entire lactation period (n = 144). The results demonstrate that Raman spectroscopy, coupled with multivariate analysis (principal component analysis [PCA] and support vector machine discriminant analysis), is a promising approach to discriminate and classify milk based on the pasture diets and breeds examined, across the 3 stages of lactation (SOL), as both separation and classification models showed excellent performance. In parallel, wet chemistry techniques were applied to the milk samples, to examine the compositional differences for parameters expected to be relevant to specific regions of the Raman spectra obtained (i.e., fatty acid profile, β-carotene, and cholesterol levels). These were analyzed by multivariate analyses (PCA and redundancy analysis), and confirmed that compositional differences in milk existed for each of the 3 factors examined, which were reflected in differences in relative intensity of spectral regions, namely the 1,660 relative. cm⁻¹ region (ascribed to C=C bonds); 1,015, 1,162, and 1,525 rel. cm ⁻¹ (ascribed to carotenoids); 2,850 to 2,950 (the most intense Raman peak, ascribed to C–H stretching region); and 3,100 to 3,400 rel. cm⁻¹ (ascribed to O–H stretching). Moreover, in this work, it was shown that unsaturation index (UI), a factor generally recognized as an indicator of nutritional and functional quality of milk, could be predicted by applying partial least squares regression to the Raman spectral data (calibration R² = 0.974 and prediction R² = 0.789), where values obtained by wet chemistry techniques were used as standard. These results suggest that the use of Raman spectroscopy and chemometrics tools could be used for discrimination, classification, and authentication of raw milk based on farm level factors, such as diet, cow breed, and SOL, and to predict milk quality factors, such as the UI.