Discrimination of beef meat according to the level of farming intensification through visible and near infrared spectroscopy data and chemometrics

The European livestock sector is highly diverse, comprising beef production systems which differ significantly in their levels of intensification, from grass-fed cull cows to intensive feedlot systems finishing young cattle. Different production practices are directly correlated to different quality grades of beef products, creating opportunities for differentiation in the market but also increase the risk of fraud. In this paper, we evaluated the performances of three spectral devices (two portable visible and NIR and one benchtop visible/NIR) and linear discriminant analysis in discriminating beef samples based on the level of intensification of farming systems. Discrimination performances were influenced by the characteristics of the device (such as the spectral range), as well as the sample preparation protocol and the dataset split approach chosen for the discriminant model. Overall, the best classification rates were obtained for samples belonging to the classes associated to the most distinct characteristics, including animal age and type (cull sucklers cows vs young beef) and production system (concentrates supplementation vs exclusively grass-fed systems). Misclassification was observed for classes characterized by high similarity, including animals fattened indoors in feedlots until 14-15 months of age with similar proportions of concentrates and maize silage. This study showed that visible and NIR spectroscopy combined with appropriate chemometric tools allow to detect differences in beef composition associated to different farming practices and degrees of intensification. Portable and lower-cost spectral devices achieved averagely lower classification accuracy than the benchtop instrument, providing new insights in innovation of authentication-driven technologies in the beef sector.