This study suggests original holistic approach based on comprehensive analysis of phenolic compounds (HPLC-DAD), volatile (including pentene dimers) and terpene compounds (HS-SPME-GC-MS) and sensory analysis by the Panel Test to propose practical tools for clustering and quality differentiation of monovarietal extra virgin olive oils (MEVOO). The sample-set consisted of 356 MEVOO samples from 13 major Italian cultivars analyzed across five production years. Hierarchical Cluster Analysis following genetic algorithm-linear discriminant analysis and random forest methods allowed identifying three robust cultivar clusters with similar sensory features and selecting nine significant sensory attributes. Models for samples classification in these clusters were set and externally validated. Terpenes exhibited the best relationships with sensory clustering (reaching approx. 92% accuracy), but combining terpene, volatile and phenolic compounds improved accuracy up to 98%. Finally, two simplified chemometric models based either on 8 terpenes or 12 combined terpene, volatile, and phenolic compounds were proposed for predicting MEVOOs classification in the three clusters. • Authentication and clustering of Monocultivar Extra Virgin Olive Oil by chemometrics • Simplified authentication using volatiles, terpenes and phenols data, GA-LDA and RF • Clustering was supported by sensory analysis (Panel Test), GA-LDA-HCA and RF-HCA • Understanding association of chemical compounds with sensory clusters • Comprehensive sensory/chemical characterization of 320 samples 13 Italian cultivars
Ugolini et al. (Thu,) studied this question.