Abstract: The volatile constitutions of 18 green tea samples with three different types of characteristic chestnut-like aroma were characterized based on gas chromatography-mass spectrometry (GC-MS) combined with multivariate statistical data analysis including principal component analysis (PCA), hierarchical cluster analysis (HCA) and partial least squares-discriminant analysis (PLS-DA). The results showed that PCA, PLS-DA and HCA could achieve good differentiation of three chestnut flavored green teas. In the PLS-DA analysis, 18 chestnut-like green tea samples were well separated according to their aroma characteristics, and the well-explained variance (R2Y =0.843) and cross-validated predictive capability (Q2 = 0.694) indicated the model’s good feasibility. In the HCA analysis, three kinds of chestnut fragrant green tea could be clearly divided into three categories at a distance of 12, of which the chestnut-like and tender chestnut-like tea samples were closer, matching the results of sensory evaluation. In addition, 38 volatile components were identified based on variable importance in projection (VIP) score > 1, which were responsible for the discrimination of green teas with three different flavor characteristics.

Key words: gas chromatography-mass spectrometry (GC-MS), principal component analysis (PCA), hierarchical cluster analysis (HCA), partial least squares-discriminant analysis (PLS-DA)