| Summary: | The aim of the present work is to develop a chemical fingerprint using multielement analysis and isotope ratios, for tracing the geographical origin of Douro and Port wines. Both wines are produced within the Douro region, located in northeastern Portugal. Although the terroir is the same the fermentation of Port Wine is interrupted with the addition of “Aguardente Vínica” (alcohol distilled from wine) which alters the sweetness of the wine according to when it is added. Thirty samples of wine were supplied by the Instituto dos Vinhos do Douro e do Porto. Samples were digested using closed vessel microwave digestion. Multielement analyses were carried out, in triplicate, in compliance with NP EN ISO/IEC 17025. Inductively Coupled Plasma Mass Spectrometry (ICP-MS) was used for the determination of 25 elements (Li, Be, B, Al, Cr, Mn, Co, Ni, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Sn, Tl, Pb, Fe, Mg, Ca, Na, P and K). The digests were further purified for strontium isolation and determination of 87Sr/86Sr ratio by Mulitcollector ICP-MS. 18Oxygen was determined by Isotope Ratio Mass Spectrometry (IRMS). Results showed that all legislated elements (B, As, Pb, Zn, Cu and Cd) complied with maximum admissible values. Cd and Tl were removed from the statistical analysis since results were below the limit of quantification in all samples. Principal Component analysis (PCA) was applied to the remaining samples. This unsupervised method returned 5 principal components (PCs) that explained 82% of total variance. Twelve parameters (Li, B, Rb, K, 87Sr/86Sr, Cr, Ni, Cu, Pb, Fe, Sr, Mn) showed influence in the PCs and therefore were used for Hieraquical Cluster Analysis (HCA). Based on the results from HCA a predictive model using Linear Discriminant Analysis (LDA) was constructed. Samples were chosen either for building the LDA model or as cross validation data, according to the results obtained in the HCA. According to the obtained results it is possible to distinguish between Port and Douro wines based on elemental inorganic analysis and Sr isotope ratios. However, the choice of samples used for the construction of the predictive model is crucial for correct identification. Data was produced in compliance with quality requirements therefore being adequate to integrate Food Composition Databases.
|
|---|