The potential of milk MIR spectra to certify milk geographic origin

Protecting and supporting local production systems, regional authorities, as well as producers, give a very important role to milk quality. Therefore, this study was aimed to investigate the potential of mid-infrared spectroscopy (MIR) for certifying the geographic origin of milk. Because milk MIR spectral databases and extra phenotypes (breed, testday, livestock herd and origin appellation of traditional products) were available in the Belgium Walloon Region via European project OptiMIR (INTERREG IVB North West Europe Program), discrimination studies were conducted to distinguish Ardennes region (which is linked to PDO “Beurre d’Ardennes”) from the rest of Wallonia. A total of 542,733 spectral records linked to their geographic origin coming from Wallonia milk recording were used (97,369 of MIR spectra -Ardennes region and 450,326 -rest of Wallonia). A mixed model (fixed: breed, year and month of record, random: herd x year) was applied to obtain predicted MIR spectral values for all testdays and prediction errors (residuals) representing the factors not present in the model. In order to test the MIR ability to milk authentication, chemometric tools, such as partial least squat regression and linear discriminant analysis were applied to residuals for three MIR spectral regions (e.g. 930-1600 cm-1, 1710-1810 cm-1 and 2560-2990 cm-1). The classifications on not-corrected MIR spectral data were 95% and the cross-validation were 95% for Ardennes region. Results showed after correction of MIR spectra, the discriminant function constructed on the residuals spectra allowed a good discrimination. The results show that MIR spectroscopy techniques may provide useful fingerprints to detect geographic origin and could be potentially used in routine management decision and quality assurance tools.