Traceability and authentication in agri-food production: A multivariate approach to the characterization ofthe Italian food excellence elephant garlic (Allium ampeloprasum L.), a vasoactive nutraceutical

A research platform for food authentication was set up by combining stable isotope ratio analysis, metabolomics by gas and liquid mass-spectrometry and NMR investigations, chemometric analyses for food excellences. This multi-analytical approach was tested on samples of elephant garlic (Allium ampeloprasum L.), a species belonging to the same genus of common garlic (Allium ampeloprasum L.), mainly produced in southern Tuscany-(Allium ampeloprasum). The isotopic composition allowed the product to be geographically characterized. Flavonoids, like (+)-catechin, cinnamic acids, quercetin glycosides were identified. The samples showed also a significant amount of dipeptides, sulphur-containing metabolites and glutathione, the latter of which could be considered a molecular marker of the analyzed elephant garlic. For nutraceutical profiling to reach quality labels, extracts were investigated in specific biological assays, displaying interesting vasorelaxant properties in rat aorta by mediating nitric oxide release from the endothelium and exhibited positive inotropic and negative chronotropic effects in rat perfused heart.

Prediction of quality parameters in straw wine by means of FT-IR spectroscopy combined with multivariate data processing

This study represents the first attempt to combine mid infrared (MIR) spectroscopy and multivariate data processing for prediction of alcohol degree, sugars content and total acidity in straw wine. 302 Italian samples, representing different vintages, production regions and grape varieties, were analysed using FT-MIR spectroscopy and reference methods. New regression functions based on a combination of Orthogonal Signal Correction and Partial Least Squares regression are proposed for prediction of quality parameters: this approach allows overcoming the issue of matrix complexity, reducing spectral interferences and enhancing the information embodied in fingerprinting data. The models proposed are characterised by an excellent reliability, with low error in prediction (alcohol: 0.28%; sugars: 9.9 g/L; acidity: 0.29 g/L) comparable both to reference methods and table wine models. Results demonstrate that vibrational spectroscopy, combined with a proper multivariate data strategy, represents a suitable strategy for the quick and non-destructive assessment of quality parameters of straw wine.