Near-Infrared Spectroscopy and Multivariate Analysis for the Determination of Nutritional Value of Soybean Meal and Maize Bran

摘要

Near-infrared spectroscopy and multivariate analysis were used to determine nutritional values of soybean meal and maize bran based on crude protein, lysine, and methionine. The crude protein, lysine, and methionine mean concentrations in soybean meal were 47.0, 2.08, and 0.603 wt%, respectively, and in maize bran were 8.090, 0.250, and 0.184 wt%, respectively. Tukey tests at a significance level of 95%, partial least squares, and principal component analysis were used. Similar values for the standard errors of calibration for lysine and methionine in maize bran were obtained. These values were lower than the results obtained for crude protein. Three values were obtained for the standard errors of cross-validation. The determination coefficients were all close to 1 with the best fit for crude protein. The values of standard errors of calibration and standard errors of cross-validation for crude protein, lysine, and methionine in soybean meal were different. The determination coefficients for maize bran were lower than those for soybean meal, indicating that the latter possesses higher nutritional value. Principal component analysis indicated that some samples may be used for the production of animal feed due to a strong relationship among crude protein, digestive lysine, total lysine, digestive methionine, and total methionine. Near-infrared spectroscopy and multivariate analysis were shown to be efficient for the determination of the nutritional values of food for proteins and amino acids.