Elemental and spectrophotometric profiling of Italian extra virgin olive oils: geographical classification, data fusion and soil-oil relationships

The authentication of extra virgin olive oil (EVOO) geographical origin requires analytical strategies capable of decoding and interpreting the multiple effects associated with cultivation areas, including the possible contribution of soil composition. This study addresses this objective through two parallel investigations. In the first, 117 commercial EVOOs from the 2024 harvest and five Italian regions were characterized by ICP-MS elemental profiling and rapid spectrophotometric determinations of total phenolic content, total flavonoid content, and antioxidant capacity. Single-block classification models were first developed for each analytical dataset, followed by a data fusion approach. Samples were grouped into three macroregions, namely Central Italy, Puglia, and Sicilia. Among the tested classifiers, PCA–LDA provides the best balance between predictive performance and interpretability, achieving an accuracy of 0.90 in external validation. A reduced discriminant model based on ten original variables was also evaluated, showing that the increased model parsimony does not result in a major performance loss. In the second investigation, orchard soils from the same five regions represented in the commercial dataset were characterized by X-ray fluorescence and compared with laboratory-produced oils obtained from olives collected at the corresponding sites. Elemental correlations were generally weak, while multivariate analyses revealed a significant but limited soil–oil association. In addition, comparison of matched commercial and laboratory-produced oils highlighted a potential effect related to production processes in elemental profiles. Overall, data fusion improved geographical classification, while the orchard-matched study provided a framework for investigating the relationships between soil composition and EVOO elemental profiles.