Estimating ruminal protein degradability of roasted soybeans using near infrared reflectance spectroscopy

Heat treatment by roasting is used extensively in the Midwest to increase RUP of whole soybeans (Glycine max L. Merr.). Samples of roasted soybeans (n = 266) were collected from different suppliers in Wisconsin, Minnesota, and Michigan to determine the value of near infrared reflectance spectroscopy for estimating RUP. Samples were analyzed for DM, total N, protein dispersibility index, and RUP using an inhibitor in vitro procedure. Samples were milled (1 mm), and near infrared reflectance spectra were collected. Scans from 121 samples (calibration set) were used to develop calibration equations. Standard errors and coefficients of multiple determination for calibration of near infrared reflectance spectroscopy using conventional chemical assays were, respectively, 0.26 and 0.97 (DM), 0.05 and 0.99 (total N), 1.60 and 0.71 (protein dispersibility index), and 0.98 and 0.90 (RUP). A validation set of 145 samples then was used to evaluate the accuracy of calibration equations for estimating chemical composition. Standard errors and coefficients of determination for validation were, respectively, 0.63 and 0.86 (DM), 0.12 and 0.86 (total N), 3.52 and 0.52 (protein dispersibility index), and 1.54 and 0.70 (RUP). Protein dispersibility index was poorly correlated (r(2) = 0.28) to RUP estimated by inhibitor in vitro. Results indicated that near infrared reflectance spectroscopy might be used to estimate DM, total N, and RUP in roasted soybeans.