Monitoring NIRS calibrations for use in routine meat analysis as part of Iberian pig-breeding programs

The meat industry, animal breeding programmes and consumer expectations regarding meat-product labelling require fast screening techniques for the determination of meat quality. The potential of Near Infrared Spectroscopy (NIRS) for this purpose has been clearly demonstrated. This paper reports on the development, validation and updating of NIRS models for the routine analysis of ground Iberian pork muscles (gluteus medius, masseter, longissimus dorsi and spinalis dorsi). Modified Partial Least Squared (MPLS) calibration models were obtained to predict fat, moisture and protein content. The method employed was based on the use of different Mahalanobis distances for the recalibration of models, enabling the number of recalibration samples to be reduced. The final recalibrated models displayed a Standard Error of Cross-Validation (SECV) of 0.35% and a Determination Coefficient of Cross-Validation ( R CV 2 ) of 0.99 for fat; values for moisture were SECV = 0.46% and R CV 2 = 0.97 ; while for protein were SECV = 0.52% and R CV 2 = 0.90 , indicating good spectral matching even for muscle types not included in the calibration.