FTIR Spectroscopy and Multivariate Analysis Can Distinguish the Geographic Origin of Extra Virgin Olive Oils

Abstract

This work investigates whether Fourier transform infrared spectroscopy (FTIR), in combination with multivariate analysis, can distinguish extra virgin olive oils from different producing countries. Duplicate spectra were collected from 60 oils from four European countries. Two approaches to data analysis were used as follows: first, the "whole spectrum" method of partial least squares (PLS) followed by distance-based linear discriminant analysis (LDA) applied to the PLS scores, and second, a genetic algorithm (GA) for variate selection from the raw data, followed by LDA applied to the selected subset. The PLS-LDA approach produced a cross-validation success rate of 96%, whereas the GA-LDA approach achieved a 100% cross-validation success rate, from subsets comprising only eight variates. Neither the selected variate nor the whole spectrum approach was able to offer insight into the origin of the discrimination in biochemical terms. However, FTIR analysis is rapid, and this work shows that it has the required discriminatory power to potentially offer a "black box" method of screening oils to verify their country of origin.