Caratterizzazione di oli di oliva vergini siciliani. Nota VI: Analisi statistica su oli da cv autoctone

The application of statistical methods to the oils analyses is important to discriminate samples by their geographic characteristics and by quality properties. A study was carried out on 444 oil samples obtained by olives from 17 varieties in four crop years (1996-2000). The applied statistical methods were: descriptive analysis, Kruskal-Wallis test, discriminant analysis and cluster analysis. Descriptive analysis showed that some variables have very large interest intervals, that decreases the significativity of the data. Instead, Kruskal-Wallis test showed that Biancolilla is the most omogeneous variety in Siciliy and that it is not influenced by geographic characteristics; moreover, acidic components of oils produced in Enna are not discriminate at all, just like the varieties of this zone, Biancolilla and Moresca, have the same characteristics. Discriminant analysis showed that almost all samples can be correctly classified. Cluster analysis showed some similarity between Ogliarola Messinese and Cerasuola varieties in the common zone of growth (Palermo). In conclusion, only the statistical data processing of sterols and fatty acids compositions is necessary to characterize and discriminate oils from different olive varieties (when a high number of samples is available); if the oil samples are obtained from a restricted geographic area the complete characterization could be difficult.