Classification of PDO olive oils on the basis of their sterol composition by multivariate analysis

The sterol compositions (GLC/FID/capillary column) of monovarietal olive oils (51 samples) from the most important cultivars of northeastern Portugal (Cvs. Cobrangosa, Madural and Verdeal Transmontana) and 27 commercial samples of olive oils with protected denomination of origin (PDO) from the same region and cultivars were evaluated. beta-sitosterol, Delta(5)-avenasterol and campesterol were the most representative sterols. Cholesterol, stigmasterol, clerosterol and Delta(7)-stigmastenol were also found in all samples. All studied samples respected EC Regulation N. 2568, and in all cases total sterols were remarkably higher than the minimum limit set by legislation, ranging from 2003 to 2682 mg/kg. Results were analysed with the help of several statistical techniques, including reduction of dimensionality by principal component analysis with cross-validation of the number of components, followed by the use of canonical variate predictive biplots for model development and canonical variate interpolative biplots for approximate classification of monovarietal and PDO olive oils. These biplots proved to be a very interesting solution in the present case study, overcoming the problems of interpretation and classification that arise whenever different multivariate analyses are coupled together. (c) 2005 Elsevier B.V. All rights reserved.