Enhancing transparency through the beef production supply chain using objective measurements is crucial for maintaining the premium price of Australian beef products. A four-class model has been developed using Raman spectroscopy for beef cattle sourced from four different production systems including long- and short-term grain-fed and grass- and grass-supplemented-fed cattle with an F1 score of 92%, 88%, 98% and 100%, respectively. A two-class model for grain- and grass-fed cattle showed 98.5% and 97.1% correct classification. Both models were developed using a dataset of 404 calibration samples and 101 test samples which resulted in a total of 6 and 2 misclassified samples in the four- and two-class models, respectively. Fatty acids were also utilised to separate samples by principal component analysis (PCA), and loadings revealed samples that could be separated by the concentration of omega-3 polyunsaturated fatty acids. A discriminant analysis of fatty acid data yielded a classification accuracy of between 88 and 100% of beef carcases according to production system. Thus, classification utilising Raman spectroscopy provided a reliable method to discriminate between cattle finished by different feeding systems.
- This work was financially supported by the New South Wales Department of Primary Industries (NSW DPI) and Meat & Livestock Australia (MLA). Scholarship funding was provided by the Australian Meat Processing Corporation (AMPC), the Graham Centre for Agricultural Innovation and the Australian Government Research Training Program (AGRTP).