The study focused on pre-Qingming Longjing, the most coveted harvest of the year, picked before the Qingming Festival in early spring when the buds are at their most tender. Researchers from the Hangzhou Tea Research Institute and collaborating institutions collected 55 tea samples from the 2024 spring season, spanning four historically recognized production areas within the West Lake region: Shifeng, Yunqi, Hupao, and Meijiawu. The samples covered three commercial grades, superior, first, and second, as defined by Chinese national standards. Because all leaves were processed by the same traditional Longjing method within a single season, the researchers could attribute chemical differences to geography and grade rather than to processing or vintage effects.

The analytical engine of the study was untargeted metabolomics based on ultra-high-performance liquid chromatography coupled to high-resolution mass spectrometry, a technique that catalogues thousands of small molecules in a sample without knowing in advance what is there. Tea, however, is a notoriously difficult matrix for this approach. It is loaded with structurally related polyphenols that differ only by hydroxylation, methylation, or glycosylation, and these near-twins often produce nearly identical fragmentation spectra, making database matching ambiguous. Reference standards for tea-specific compounds are scarce, compounding the problem.

The team’s key innovation addressed a subtler culprit: in-source fragment ions. During electrospray ionization, some molecules break apart before they ever reach the targeted fragmentation stage of analysis. These premature fragments register as independent signals, inflating feature tables and masquerading as metabolites that do not exist. Conventional processing platforms such as XCMS, MS-DIAL, and MZmine excel at peak detection but do not systematically flag these impostors. The researchers used a platform called AntDAS-LCHRMS, which groups ions that share nearly identical elution profiles, a bilinear signature revealing a common molecular origin, and then reconstructs a composite spectrum for each true compound.

The payoff was substantial. Applied to nearly 9,000 features that varied significantly by origin, the in-source fragment workflow collapsed the data into 5,431 compound-level identifications and produced 812 library matches, a 62.4 percent improvement over the 500 matches achieved by a conventional feature-level pipeline on the same dataset. It also exposed how misleading conventional annotation can be: in one illustrative case, fragment ions from a single flavonol glycoside were erroneously assigned to kaempferol-7-O-glucoside and even to kaempferol itself, while the joint analysis of reconstructed and experimental spectra correctly identified the compound as kaempferol-7-neohesperidoside. In total, the team confidently identified 57 compounds associated with geographical origin.

With the chemistry cleaned up, the discrimination results were striking. Principal component analysis and orthogonal partial least squares discriminant analysis separated the four production areas cleanly within every grade, and separated the three grades cleanly within every origin. Amino acids and their derivatives, including theanine, glutamine, and asparagine, the compounds behind Longjing’s prized umami sweetness, dominated the origin-discriminative markers, while flavonoid glycosides and purine alkaloids such as caffeine and theobromine, which govern bitterness and astringency, contributed in grade-dependent patterns. Three markers, theobromine, luteolin-4′-O-glucoside, and tyrosine, consistently distinguished origins across all grades, making them candidate universal fingerprints of terroir.

Grade told a different biochemical story. Within each production area, higher grades were generally associated with the enrichment of free amino acids linked to umami and freshness, consistent with the national grading standards, while flavonoids and alkaloids modulated the bitterness and astringency that define the perceived quality hierarchy. Yet the overlap of grade markers across origins was minimal, with tyrosine the only compound shared by all four areas. Quantitative measurements confirmed its dual role: tyrosine levels were markedly higher in Shifeng and Hupao than in Yunqi and Meijiawu within the same grade, and its grade-related trends ran in different directions depending on the village. Theophylline showed similarly origin-dependent grade patterns, underscoring that quality chemistry is not uniform across the West Lake region.

Crucially, the models survived the ultimate test of any authentication tool: time. Eighteen independently collected pre-Qingming samples from the 2026 season, corrected for inter-batch drift using pooled quality-control samples, were projected onto the fixed 2024 models without refitting. The models correctly assigned 88.89 percent of the 2026 samples to their geographical origin and 94.44 percent to their commercial grade, with nearly all predictions falling within the models’ confidence regions. The authors are careful to frame this as preliminary validation, since only one additional production year was tested, and tea metabolites are known to shift with annual variation in temperature, rainfall, and nitrogen supply.

The limitations are real but bounded. A portion of detected features remained unannotated because spectral libraries for complex tea matrices are still incomplete, and environmental variables such as soil properties and microclimate were not quantified, limiting mechanistic interpretation. A few misclassifications occurred in the superior-grade origin model and the Meijiawu grade model, a reminder that temporal variation cannot be entirely dismissed. The researchers suggest that future work should extend validation across multiple consecutive spring seasons and integrate volatile profiling by gas chromatography to capture aroma chemistry alongside taste chemistry.

Even with those caveats, the study delivers something the premium tea market has lacked: an objective, chemically grounded framework for verifying both where a Longjing tea grew and what grade it merits, replacing subjective expert panels with reproducible molecular evidence. As geographical indication products worldwide face growing pressure from adulteration, the message from West Lake is that the answer may lie not in stricter paperwork but in smarter data processing, teaching mass spectrometers to recognize their own fragments before asking them to catch a counterfeit.

Subject of Research: Non-volatile metabolomic authentication of geographical origin and quality grade in pre-Qingming Xihu Longjing green tea

Article Title: Non-volatile metabolomic differentiation of geographical origin and quality grade in pre-Qingming Xihu Longjing green tea using in-source fragment ion-assisted annotation

Article References: Wang, X.-C., Yang, C., Wu, H.-X., Lv, H., Zhai, M., Ma, H., Fan, Y., Yu, Y.-J., Fu, H.-Y., & She, Y. (2026). Non-volatile metabolomic differentiation of geographical origin and quality grade in pre-Qingming Xihu Longjing green tea using in-source fragment ion-assisted annotation. Food Chemistry: X, 39, Article 104494. https://doi.org/10.1016/j.fochx.2026.104494

Image Credits: AI Generated

DOI: Not provided

Keywords: Xihu Longjing, green tea, food authentication, metabolomics, UHPLC-HRMS, in-source fragment ions, geographical indication, chemometrics, theobromine, tyrosine, food fraud, mass spectrometry