The use of mineral and trace elements profiles for cows' and goats' cheese species prediction

摘要

Cluster, principal component, factor and canonical discriminant analysis were used for differentiation of cows' and goats' cheese species using the contents of minerals (Ca, Cu, K, Mg, Na) and risk elements (Ba, Cr, Hg, Mn, Mo, Ni, V). Slovakian cheeses' data output of cluster analysis concerning the membership of the samples to clusters resulted in 95.5% of correctly marked cheeses according to their species of origin. Recognition ability expressed as classification in canonical discriminant analysis conditions resulted in 99.3% of the total cheeses correctly classified, where Cu, Na, Ca, Hg and Mn showed the most discriminant impact on categorizing Slovakian cheeses by their affiliation to animal species. When discriminant analysis was applied to European cheeses, the classification resulted in 97.7% of cheeses correctly classified and in 97.5% of correctly classified samples after cross-validation in the prediction capability procedure. The most discriminating variables for European cheeses were Ba, Ca, Cr, Cu, Hg, K, Mg, and Na concentrations. Found results revealed that multielemental data selection and multivariate statistics are able to differentiate among animal species origin of cheeses produced in cheese-making manufactories on the territory of one or more countries.