Tunisian olive oils geographical origin discrimination using the potentiometric fingerprints recorded by an electronic tongue

The development of fast and cost-effective analytical techniques for EVOO authentication is a challenging task. Moreover, if a specific meteorological or geographical factor affects different geographical regions similarly, olive oils geographical discrimination may be a hard task using conventional analytical techniques [1]. E-noses and/or voltammetric E-tongues have already been applied to assess olive oils' geographical origin, mainly to discriminate different countries or quite different regions of the same country [2]. In this work, we used an electronic tongue (E-tongue ), with 40 lipid membrane sensors, to extract representative potentiometric fingerprints of Tunisian monovarietal olive oils that, in combination with linear discriminant analysis (LOA), could be used to classify olive oils according to the geographical origin. Aqueous ethanolic (80:20, v/v) extracts of different single-cultivar Tunisian olive oils were electrochemically analysed. According to the literature [3-6], these olive oil' extracts are rich in polar compounds that deliver different overall potentiometric responses, which can then be used to evaluate the E-tongue performance for olive oils geographical origin discrimination. The proposed E-tongue-LDA approach, based on the signal profiles of different sub-sets of sensors (seleted with the simulated annealing meta-heuristic algorithm) allowed the correct geographical origin classification of Tunisian olive oils produced from autochthonous Chemleli or Sahli cultivars (i.e., Kairouan, Sidi Bouzid and Sfax; or, Mahdia, Sousse and Kairouan; respectively). Indeed, predictive correct classifications of 92±7% and 97±8% (for repeated K-fold cross-validation) could be obtained for Chemleli or Sahli olive oils, pointing out the potential use of the E-tongue device for geographical origin identification of olive oils.This work was financially supported by POCI- 0 1-0 145-FEDER-006984-Associate Laboratory LSRE-LCM, Project UID/QUI/00616/2013--CQVR, Project UID/BI0/04469/2013--CEB and Project UID/AGR/00690/2013--CIMO all funded by FEDER, through COMPETE2020, and by national funds through. Nuno Rodrigues thanks FCT, POPH-QREN and FSE for the Ph.D. Grant (SFRHIBD/1 04038/2014 ).