Abstract: The contents of 31 trace elements in 42 green tea samples collected from four tea-producing counties in Tongren city, Guizhou province were determined by inductively coupled plasma mass spectrometry (ICP-MS). The correlation between these elements was analyzed. Principal components analysis (PCA), stepwise linear discriminant analysis (S-LDA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were used to develop classification models for the tea samples from different geographic origins. The results indicated that the contents of Co, Mo, Ag, Cd, Zn, Al, Y, Nd, Sc, Pr, Gd, Sm, Dy, Er, Yb, Ho, Tb, Tm and Lu in tea samples were significantly different among the different producing areas (P < 0.05). There was a synergistic correlation between some of the elements, while others showed an antagonistic correlation between each other. The OPLS-DA model displayed the best classification performance with a correct discrimination rate of 92.86%. This study revealed that the multivariate statistical classification model based on trace elements contents could effectively distinguish tea samples from neighboring producing areas, which will be useful for tea quality and safety traceability in the tea industry.

Key words: green tea; trace elements; multivariate statistics; geographical origin traceability