Near-infrared evaluation of wet mink diets

Abstract

Near-infrared spectroscopy (NIRS) is a relatively new technology in the determination of chemical constituents in compound (complete) feeds. The current paper reports on the application of NIRS to the measurement of dry matter (DM), crude protein (CP) and crude fat (CF) in wet diets for mink and the development of a calibration to predict these chemical constituents. In total, 287 typical wet mink feeds were collected from commercial mink ranches located on the south shore of Nova Scotia, Canada, to provide a large representative sample population. After population structuring, using CENTER and SELECT software (ISI) during calibration development, there were 73 samples remaining in the calibration. All feeds were analyzed by standard reference methods (primary) and then scanned (secondary) with a scanning monochromator NIR, model 5000 (NIRSystems Inc.), to develop a global calibration to predict DM, CP and CF. The “best fit” treatment for the calibration was a modified partial least squares analysis with a standard normal variate (SNV) correction for scatter, a derivative math treatment of 1, 4, 4, 1 to remove particle size interference while scanning segments 1108–2492 skipping every 8 nm for prediction of DM, CP and CF. Standard error of calibration (SEC, g kg −1), proportion of explained variance (RSQ), and standard error of cross-validation (SECV, g kg −1) for DM, CP, and CF were; 11.9, 0.92 and 14.3; 11.3, 0.96 and 17.5; and 20.0, 0.92 and 22.5. Two external validation sets were used to evaluate the accuracy of the calibration with standard error of prediction (SEP), ranging from 12.7 to 19.5 g kg −1. Based on the results of the data reported the use of NIRS technology in the nutritional evaluation of complete wet diets for mink is feasible. The most promising future use for this NIRS application is as on-line quality control in a central feed kitchen.