The share of adulterated wines on the global market is rising and this trend is also visible in the Czech Republic. The control authorities are confronted with an increasing number of cases of adulterated wine. A characteristic feature of grapevine cultivation in the Czech Republic is a diverse spectrum of the cultivars. The verification of wine’s varietal authenticity, next to the confirmation of geographic origin, is the toughest challenge for analytical chemists and control laboratories. The aim of this study was to assess possibilities of the discrimination and classification of Moravian varietal wines based on the elemental composition data. An important objective was to find the variables in elemental composition which are strongly associated with a particular variety. Tests were performed on three popular varieties (Rhine Riesling, Müller-Thurgau and Green Veltliner). Analysis of wine samples was carried out by the combination of ICP-MS and ICP-OES methods. Experimental data were evaluated by univariate and multivariate statistical methods, such as analysis of variance, principal component analysis and discriminant analysis. Statistically significant discriminant fuctions and predictive functions were constructed by the method of canonical discriminant analysis. These fuctions were based on elemental composition parameters: Al, Sn, Gd, Tb, Tm/Yb, Yb/Lu, Mo/Sn, Mn/Cr. The model thus created was capable of classifying known varietal wines at a succes rate of 95.83 %. The predictive capability of the model was finally tested by the cross validation method. Classification effectiveness for unknown samples was determined at 70.83 %. The results prove that the approach to varietal wine authentification presented in this paper is a promissing option in interregional varietal wine discrimination.

23.2.2018