GEOGRAPHICAL TRACEABILITY AND AUTHENTICITY OF EXTRA VIRGIN OLIVE OIL BY CHEMOMETRIC TECHNIQUES AND CHROMATOGRAPHIC FINGERPRINT

In recent years, special attention was posed on the issues related to food quality and safety. In this context, the identification and traceability of foods play a key role, as a defense, both for the producers and the consumers. Indeed, at present, only label and production records guarantee the effective traceability of foodstuff. Therefore, the necessity to develop new analytical methods that allow, a posteriori, to define the correct geographical origin, emerged. PDO Extra virgin olive oil is one of the foods made in Italy with the highest added value, because it is related to a delimited area of production. However, it is too often subjected to frauds and imitation that are difficult to detect, especially if control relies simply on reading the label. The aim of this work was therefore to develop a new analytical method that could allow the identification of PDO extra virgin olive oil, and in particular, which could allow to discriminate the Sabina PDO from other olive oil, extra virgin or not. To achieve this objective, a complex chemical and chemometric analysis were carried out. In fact, univariate analysis of some parameters, like the free acidity, peroxide number and the total content of polyphenols, is restrictive and inadequate, allowing only the distinction between different kinds of vegetable oils. On the other hand, studies in the literature suggest that the quality and quantity of polyphenols present in olive oils and multivariate analysis can be valid instruments for the geographical classification of the product. At first, through an experimental design protocol, the extraction of polyphenolic components was optimized in terms of recovery, time and cost. The identification of the compounds was performed through the use of mass spectrometry while multivariate analysis was conducted on the entire chromatogram of the phenolic fraction, recorded at 254, 280 and 340nm. The chromatographic profile of each sample was considered as a fingerprint of olive oil and with the application of chemometric methods, it was possible to extract useful chemical information for the classification and determination of the geographical origin. Before applying classification methods, it was necessary to pretreat the chromatographic data to eliminate the variability due to variations of the baseline and the shift of the retention times of the analytes. For the correction of the baseline, the algorithm "Penalyzed Asymmetric Least Squares"[1] was used. After correcting the baseline, it was necessary to pretreat further chromatographic signals to ensure that the peaks of the analytes were aligned. Operatively, the alignment of the chromatograms was performed using iCoshift algorithm, which divides the chromatograms into sever parts and for each part identifies the best alignment [2]. The chromatographic profiles of extra virgin olive oils extracts (27 Sabina PDO and 50 other origins) after being "pretreated", have been used as data for the construction of the classification model. Specifically, the method applied for discriminant classification was Partial Least Squares Discriminant Analysis (PLS-DA) [3]. The predictive capability of a multivariate classification model can be affected by the presence of a large number of variables, in our case, not all the points that constitute the chromatographic profile carry discriminant information, and a selection of portion of the chromatogram was necessary. For this purpose, the technique Backwards Interval PLS (Bi-PLS) coupled to a procedure based on Genetic Algorithms (GA) [4] was used . Once calibrated, the classification model (PLS-DA after Bi-PLS-GA) has been validated, and tested for its predictive capacity on external extra virgin olive oil samples and 90% of these were correctly classified. Of 27 samples of extra virgin olive oil (Sabina PDO and not) 24 were classified in the appropriate class of perfectly. In conclusion, the analytical-method developed, being based on the chemometric processing of the results of chemical analysis on the finished product, doesn’t rely on label and can allow detecting imitations and falsifications of Sabina PDO. Furthermore, it was demonstrated that the chromatographic fingerprint of the phenolic fraction of extra virgin olive oil may be a possible indicator of product traceability. In the future,models of traceability, similar to this, can be built for other extra virgin olive oil with the appellation of origin to revealing fraud.