ABSTRACT
A capacity to determine the provenance of high-value food products is of high scientific and economic interest. With the aim to develop a tool for geographical traceability of Croatian extra virgin olive oils (EVOO), multielement composition and 13C/12C isotope ratio in EVOO as well as the geochemistry of the associated soils were analysed in samples collected from three regions along the Croatian Adriatic coast. Soil geochemistry was shown to influence the transfer and elemental composition of EVOO. The most discriminating variables to distinguish EVOO from different regions were S, Mo, Rb, Mg, Pb, Mn, Sn, K, V and δ13C. The predictive models achieved high sensitivity and specificity, especially when carbon isotope composition was added. The results suggest that interregional geographical traceability of Croatian EVOO is possible based on matching their multielement composition with that of the soils in the provenance area.
Subject(s)
Chemometrics , Soil , Olive Oil/chemistry , Carbon Isotopes/analysis , Croatia , Plant Oils/analysisABSTRACT
The δ18O and δ13C (analyzed by isotope ratio mass spectrometry, IRMS) and concentration of 22 selected elements (analyzed by inductively coupled plasma-optical emission spectrometry, ICP-OES) in 190 Croatian microvinified and commercial wine samples from continental and coastal winegrowing areas and from three viticultural zones (B, CI, and CII) were measured to investigate whether multivariate statistical methods could provide the fingerprint for geographical origin determination. The highest power for discrimination of wines produced in Croatian winegrowing areas was achieved by general discriminant analysis (GDA) showing correct classification of 97.9% of all investigated samples, 100.0% of microvinified samples and 84.8% of commercial samples in the cross-validation matrix. The most significant markers for discrimination of coastal and continental areas found by GDA were δ18O and Co, followed by K, Rb, Sn, Li, and δ13C in descending order. GDA showed higher levels of correctly classified samples from three viticultural zones in Croatia if only microvinified samples were employed in the analysis (94.9%) than for all samples together (86.3%) or for commercial samples (66.1%) in the cross-validation matrix. The discrimination of viticultural zones B, CI, and CII in Croatia was achieved by δ18O, Co, Rb, Li, K, and Sn. The results obtained showed that the relationships between the isotopic ratios and concentrations of different considered elements combined with appropriate statistical model represent a powerful tool in discrimination of wines produced in different Croatian winegrowing areas.
