Sequential (step-by-step) detection, identification and quantitation of extra virgin olive oil adulteration by chemometric treatment of chromatographic profiles

An analytical method for the sequential detection, identification and quantitation of extra virgin olive oil adulteration with four edible vegetable oils - sunflower, corn, peanut and coconut oils - is proposed. The only data required for this method are the results obtained from an analysis of the lipid fraction by gas chromatography-mass spectrometry. A total number of 566 samples (pure oils and samples of adulterated olive oil) were used to develop the chemometric models, which were designed to accomplish, step-by-step, the three aims of the method: to detect whether an olive oil sample is adulterated, to identify the type of adulterant used in the fraud, and to determine how much aldulterant is in the sample. Qualitative analysis was carried out via two chemometric approaches - soft independent modelling of class analogy (SIMCA) and K nearest neighbours (KNN) - both approaches exhibited prediction abilities that were always higher than 91% for adulterant detection and 88% for type of adulterant identification. Quantitative analysis was based on partial least squares regression (PLSR), which yielded R-2 values of > 0.90 for calibration and validation sets and thus made it possible to determine adulteration with excellent precision according to the Shenk criteria.