Characterisation of the geographical origin of Western Greek virgin olive oils based on instrumental and multivariate statistical analysis

In this work, measurements of free acidity, peroxide content, spectrophotometric parameters, chlorophyll content, phenol, sterol, fatty acid and triacylglycerol composition, were carried out on samples of virgin olive oils (VOOs) coming from four different Greek Ionian islands, i.e. Zakynthos, Kefalonia, Lefkada and Kerkyra. An analysis of variance (ANOVA) highlighted statistically significant differences (p < 0.01) in the values of 26 analytical parameters amongst the VOOs produced in the four different geographical regions but a post-hoc test showed that no variable was able to distinguish all four origins. Analogously, a Principal Component Analysis (PCA) showed a modest grouping of VOOs according to geographical origin except for Kerkyra samples which were more distinct from others. Applying discriminant function analysis (DFA) a good separation of the four geographical groups was achieved with classification and prediction abilities equal to 97.7% and 95.3%, respectively. Moreover, the analysis of the standardised coefficients showed that the fatty acids and triacylglycerols were the most discriminant variables. This last outcome was confirmed by comparison of the prediction performances obtained applying DFA on four subdatasets containing fatty acids (69.8%), triacylglycerols (76.7%), sterols (62.8%), and remaining parameters (65.1%) together, respectively. As the results showed, the multidisciplinary approach that combines different types of analytical determinations improved the discrimination of geographical origin for Greek virgin olive oils. (C) 2012 Published by Elsevier Ltd.