Near infrared spectroscopy calibrations for quantifying the animal species in processed animal proteins

Near infrared (NIR) calibrations were developed for the instantaneous and simultaneous prediction of the animal species composition of processed animals proteins (PAPs). A total of 359 PAPs samples were used to built calibration and validation sets. Different derivatives and scatter correction pre-treatments were tested. Modified partial least squares regressions were developed for the prediction of the percentage of poultry by-products meat, pork meat, cattle meat, ruminant meat and non-ruminant meat in a mixture. All the equations presented R-2 and ratio performance deviation (RPD) values around 0.9 and 3, respectively. In the external validation procedure, the results obtained correspond well with the calibration statistics. This study shows the potential of NIR technology to quantify the animal species in PAPs and it clearly demonstrates the importance of the size and structure of the calibration set in the robustness of the equations.