Near-Infrared Spectroscopy to Predict Nutritional Factors of Green Forage

Green forage quality is crucial for efficient breeding. Therefore, it would be extremely important to provide a method for a rapid and cost-effective analysis of the nutritional factors of green forage. The application of Fourier-Transformed Near-Infrared (FT-NIR) spectroscopy in combination with chemometric techniques could meet the identified requirements. The objective of the present study was to develop optimized partial least squares (PLS) models for the prediction of nutritional factors of green forage, namely dry matter, crude protein, crude ash, ether extract, neutral detergent fiber, acid detergent fiber, acid detergent lignin, and crude fiber. The elected PLS models resulted from a combination of chemometrics techniques applied to FT-NIR absorbance spectra acquired on fresh and dried/ milled samples of green forage. The most accurate prediction results were obtained for dry matter in the fresh samples (coefficient of determination of cross-validation, R2CV=0.95; root-mean-square error of cross-validation, RMSECV =3.84g100g-1 fresh weight), and crude protein in the dried/milled samples (R2CV= 0.94, RMSECV = 1.99 g 100 g-1 dry weight). Future developments will be further focused on improvement of prediction accuracy by applying deep learning techniques.