The ability of near infrared (NIR) spectroscopy to predict the milk fatty acids (FA) composition, groups of fats and nutritional-related ratios and indexes in sheep milk was evaluated using different near infrared (NIR) instruments. Detailed FA composition of 101 individual milk samples was analyzed by gas chromatography. Predictive equations were developed for oven-dried milk samples using modified partial least squares (PLS) regression method for reflectance spectra (400−2500 nm) and using PLS for transmittance (850–1050 nm) and Fourier transform near infrared (FT-NIR) reflectance (1000–2500 nm) spectra. Coefficient of determination of cross validation (R2 CV) and residual predictive deviation (RPD) were excellent for C4:0, SFA and PUFA (R2 CV above 0.9; RPD above 3). Good predictive results (R2 CV 0.80–0.90; RPD 2.9–2.0) were obtained with the reflectance spectra for C18:1n7t, C18:2n6c, C18:2 9c11t, C18:3n3 C22:5n3, MUFA, SCFA, LCFA, total C18:1, CLA, n-3, n-6, n-6/n-3 ratio, and the desirable FA, atherogenicity, thrombogenicity and peroxidability indexes. Less accurate results were obtained for C8:0, C12:0, C14:0, C18:2n6t, and hypocholesterolaemic/hypercholesterolaemic ratio. The FT-NIR instrument provided comparable results. On the other hand, only the butyric acid model can be considered satisfactory (R2 CV=0.74; RPD=1.96) with the dispersive transmittance spectra.
Estimating fatty acid content and related nutritional indexes in ewe milk using different near infrared instruments
Acuti, Gabriele
;Branciari, Raffaella;Ranucci, David;Olivieri, Oliviero;Trabalza-Marinucci, Massimo
2020
Abstract
The ability of near infrared (NIR) spectroscopy to predict the milk fatty acids (FA) composition, groups of fats and nutritional-related ratios and indexes in sheep milk was evaluated using different near infrared (NIR) instruments. Detailed FA composition of 101 individual milk samples was analyzed by gas chromatography. Predictive equations were developed for oven-dried milk samples using modified partial least squares (PLS) regression method for reflectance spectra (400−2500 nm) and using PLS for transmittance (850–1050 nm) and Fourier transform near infrared (FT-NIR) reflectance (1000–2500 nm) spectra. Coefficient of determination of cross validation (R2 CV) and residual predictive deviation (RPD) were excellent for C4:0, SFA and PUFA (R2 CV above 0.9; RPD above 3). Good predictive results (R2 CV 0.80–0.90; RPD 2.9–2.0) were obtained with the reflectance spectra for C18:1n7t, C18:2n6c, C18:2 9c11t, C18:3n3 C22:5n3, MUFA, SCFA, LCFA, total C18:1, CLA, n-3, n-6, n-6/n-3 ratio, and the desirable FA, atherogenicity, thrombogenicity and peroxidability indexes. Less accurate results were obtained for C8:0, C12:0, C14:0, C18:2n6t, and hypocholesterolaemic/hypercholesterolaemic ratio. The FT-NIR instrument provided comparable results. On the other hand, only the butyric acid model can be considered satisfactory (R2 CV=0.74; RPD=1.96) with the dispersive transmittance spectra.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
