Evaluation of pattern recognition techniques for the characterization of PDO olive oils by 1H-NMR fingerprinting

ALONSO-SALCES, R. M.; HOLLAND, M. V.; GUILLOU, C.; HÉBERGER, K.

Siófok (Hungría)

Conferencia; Conferentia Chemometrica 2009; 2009

Chemometric Section of the Hungarian Chemical and Hungarian Academy of Sciences

Extra-virgin olive oil is a high added value agricultural product in the European Union, in both commercial and nutritional terms. EU legislation protects it with quality labels on basis of its geographical origin such as “Protected Designation of Origin (PDO)” [Regulation (EEC) No 2081/92]. PDO olive oils are sometimes subject to adulteration, for instance, with olive oils that do not fulfil the PDO requirements or by labeling a non-PDO olive oil as a PDO one. Therefore, validated methods to guarantee the authenticity and traceability of PDO olive oils are necessary to protect both the consumer and the producer from illicit practices. The authentication of olive oils with respect to their geographical origin has been studied using various analytical approaches such as NMR (1H, 13C, 31P), NIR spectroscopy, IRMS, LC-MS and GC-MS [1-4]. NMR fingerprinting methods seem particularly useful as they can be used to generate reference fingerprints for these products, providing means to compare profiles of suspected counterfeit products with these reference data, and thus detect fraud. In the TRACE project (http://www.trace.eu.org), a statistical significant number of authentic PDO extra-virgin olive oils from EU and non EU countries (935 samples) were collected during three seasons (2005, 2006 and 2007). 1H-NMR fingerprints of these olive oils were obtained by a high throughput NMR approach and analyzed by pattern recognition techniques in order to distinguish the olive oils of a certain PDO, e.g. “Riviera Ligure” (Liguria, Italy), from the oils of other protected origins. Several supervised multivariate data analysis were performed by LDA and PLS-DA using balanced and unbalanced datasets, different cross-validation methods, different data scaling, external validation and different commercial softwares in order to evaluate the best approach for the authenticity and traceability of PDO olive oils. The procedure for data analysis and the critical steps to obtain feasible models, as well as the results provided by each software were studied.