부산대학교 도서관

This study aimed to evaluate the feasibility of mid-infrared (MIR; 5000–900 cm−1) and visible/near-infrared (Vis/NIR; 400–2500 nm) spectroscopy to discriminate organic (ORG) from conventional (CONV) bulk milk. Samples (n = 225) from 24 farms (ORG, n = 12; CONV, n = 12) located in the same area, mainly rearing Holstein-Friesian cows, under similar management conditions, except for ORG livestock spending a period of time on pasture, were collected from September 2019 to August 2020. Chemical composition of the lactation ration was similar between groups. Mid-infrared and Vis/NIR spectrum of each sample were collected. Principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA) were done using R software. For the PLS-DA, records were divided in a train set (8 farms/group) and a test set (4 farms/group), and only wavelengths with VIP >1 were retained. The PCA was not able to discriminate both groups. The PLS-DA revealed an accuracy of the model in the test set of 54.1% and 62.9%, for MIR and Vis/NIR, respectively. In conclusion, both infrared regions performed similarly, and the moderate accuracy of the PLS-DA could be related to the similarity of the selected farms between both categories. Infrared spectra from bulk milk samples of organic and conventional farms were similar. Low accuracy using partial least squares-discriminant analysis. Both MIR and Vis/NIR infrared regions performed similarly. [ABSTRACT FROM AUTHOR]