Flavor quality evaluation and discrimination of regional tea (Camellia sinensis L.) based on metabolomics, volatileomics, and machine learning

Abstract

This study aimed to elucidate the differences in the composition of the flavor molecular basis of fresh tea leaves ( Camellia sinensis L.) from different regions and their potential for geographical discrimination. The samples from eight representative producing areas across China’s four major tea regions (Jiangbei, Jiangnan, South China, and Southwest) were analyzed using biochemical composition profiling, UPLC-Q-TOF-MS, and HS-SPME-GC-MS. Sixteen chemical indices, 171 metabolites, and 275 volatile organic compounds were identified, based on multivariate criteria. Six chemical indicators, 24 metabolites across 8 classes, and 24 volatile organic compounds across ten classes were identified as differential compounds, revealing molecular feature differences of fresh tea leaves across major Chinese tea regions and highlighting the crucial role of regional factors in the formation of the flavor basis. KEGG enrichment revealed four key metabolic pathways. Using five machine learning models in combination with the drop-column importance (DCI), three characteristic flavor markers were identified, including geraniol, 11-methoxynoryangonin, and methyl gallate. This work elucidates the molecular basis of terroir effects on tea flavor formation and provides novel omics-based insights for origin tracing and quality control.