The review, led by Miao Qi and Liyong Luo of Southwest University in Chongqing, assembles evidence that has long sat scattered across soil science, food microbiology, and nutrition research. Its central concept is the tea plant holobiont: the plant together with the rhizosphere, endosphere, and phyllosphere microbiota that live on and inside it. Processing microbes and the consumer’s gut bacteria are treated as downstream ecological partners in an extended field-to-gut framework. The proposition is deceptively simple. Tea quality, encompassing both flavor and health attributes, does not arise from host metabolism alone but from plant-microbe co-metabolism followed by successive microbial transformations after harvest and after swallowing.

The anomalies that break the old model are striking. A single cultivar grown under different soil microecological conditions develops measurably different quality attributes, even when the plant genetics are identical. Post-fermented dark teas such as Pu-erh and Fu brick tea lose large fractions of their catechins while gaining theabrownins and entirely new aroma compounds, transformations driven by microbial communities whose metabolic repertoire exceeds anything the plant can do. And the same tea can produce divergent physiological responses in different people, a pattern that plant chemistry alone cannot explain but that differential microbial processing in the gut plausibly can. Continuity from plantation to gut, the authors stress, does not mean a single microbiome travels intact along the chain. It may involve direct microbial transmission, ecological filtering or replacement, or simply the carryover of metabolites made by upstream microbes.

The story begins underground. Tea plantation soils host microbial communities whose assembly reflects a persistent tension between deterministic niche-filtering and stochastic dispersal, and the balance shifts with context. Across a 105-year chronosequence, fungal assembly drifted progressively from stochastic toward deterministic control, while bacterial communities remained stochastically assembled throughout, likely buffered by their faster turnover and broader metabolic versatility. Geography matters too: Zhengyan rock-tea soils, famous for oolong terroir, yielded communities assembled predominantly through stochastic processes, whereas nearby Banyan and Zhoucha soils showed deterministic dominance. Abiotic filters such as soil pH, exchangeable magnesium and calcium, clay content, and altitude constrain which microbes establish, while the host genotype selectively recruits distinct consortia at the rhizosphere interface.

Management decisions can impose selective pressures that rival natural edaphic variation, sometimes with uncomfortable consequences. Organic farming raises keystone species richness and produces more complex, stable microbial networks, and pecan-tea intercropping produces analogous structural gains. Prolonged nitrogen over-application does the opposite, eroding bacterial diversity and destabilizing community structure. Fertilization regimes optimized for short-term yield may therefore quietly strip away the microbial diversity that underpins long-term flavor complexity. Notably, soil fungal diversity indices correlate positively with multiple quality attributes in finished tea, although the authors are careful to note that this correlation does not by itself establish causation; controlled inoculation trials are needed to determine whether diverse fungi genuinely generate richer precursor pools or whether both simply respond to shared soil health.

What the rhizosphere community actually does is increasingly well characterized, and it is a story of dual-function agents. A potassium-solubilizing Burkholderia species isolated from tea roots dissolves insoluble potassium through organic acid excretion, raising available soil K and promoting accumulation in roots and leaves. Rhizosphere bacteria secrete hydrolases that accelerate organic matter turnover and release indole-3-acetic acid, coupling nutrient cycling with hormonal control of root development. Synthetic consortia built from representative strains enhanced ammonia uptake and promoted theanine biosynthesis under controlled conditions. On the defensive side, Bacillus megaterium inhibits mycelial expansion and conidial germination across a broad range of root-infecting fungi while simultaneously solubilizing phosphate and releasing siderophores. Trichoderma asperellum TC01 activates the jasmonic acid and ethylene signaling pathway, cutting anthracnose severity by roughly 58 percent. The same microbes often do both jobs at once, which means any practice that depletes dual-function populations may weaken plant defense and reduce flavor precursor supply simultaneously.

Perhaps the most provocative claim concerns theanine, the signature umami compound long assumed to be exclusively plant-made. Sun and colleagues first identified L-theanine biosynthetic capacity in endophytic bacteria of the genus Luteibacter, whose gamma-glutamyltranspeptidases synthesize the amino acid efficiently. In a controlled inoculation study, Pseudomonas knackmussii combined with ethylamine supplementation increased leaf theanine content nearly fourfold. Catechins may have microbial sources too: endophytic fungi including Pseudopestalotiopsis camelliae-sinensis and Didymella sinensis were documented as independent catechin and EGCG producers in axenic culture, though the authors caution that optimized culture conditions differ materially from the competitive interior of a living leaf, and isotope-labeling experiments will be needed to quantify the true in planta contribution. The review frames this as a microbial bypass running in parallel with plant-endogenous synthesis, and raises an uncomfortable question for breeders: programs that evaluate genotypes solely through host-genome markers may systematically underestimate the microbiome’s contribution to final metabolite levels.

Gene-level specificity makes the bypass concept more than a loose association. An endophytic strain designated V4 simultaneously downregulates catechin-associated genes CsLAR and CsANS while upregulating theanine synthase CsTS, directly adjusting the leaf’s astringent-to-umami ratio. Colonization by wild-type Herbaspirillum sp. ZXN111 reduced L-theanine accumulation, whereas a mutant lacking a single gene function increased it markedly. That flavor tuning can hinge on individual microbial gene products opens a path toward targeted genetic modification of selected endophytes as a precision tool for metabolite adjustment. Meanwhile, the phyllosphere bacterium Herbaspirillum camelliae secretes a characterized tannase that degrades EGCG and epicatechin gallate, releasing gallic acid, one of the most biochemically defined endophytic bioconversions documented in tea.

Processing then acts as a sequence of ecological filters. Fresh leaves arrive at the factory carrying a structured phyllosphere inoculum dominated by Sphingomonas and Methylobacterium, and each manufacturing step reshapes it. In Keemun black tea, rolling was the single step most predictive of the final bacterial community profile; high-temperature fixation in green tea can drive viable microbial load to negligible levels. Dark teas take the opposite trajectory, deliberately cultivating successional ecosystems in which bacteria shift from Proteobacteria toward Firmicutes and fungi rise to dominance. During the flowering stage of Fu brick tea, Aspergillus cristatus was associated with 20 to 45 percent reductions in major catechins, a near-doubling of gallic acid, and marked increases in methyl salicylate and linalool that generate the tea’s characteristic floral profile. Temperature and moisture act as habitat filters on this succession: wet-piling at around 60 percent moisture promoted Rhizomucor dominance, while dry-piling below 30 percent selectively enriched Aspergillus. Across ripened Pu-erh, Liubao, and Tibetan teas, only six dominant microorganisms and six key aroma compounds were shared, a striking testament to how sensitive flavor outcomes are to both community composition and process parameters.

The final act unfolds in the human gut, where most high-molecular-weight tea polyphenols escape systemic absorption and encounter the resident microbiota. Here the evidence turns causal. Antibiotic-induced microbiota depletion completely abrogated the anti-obesity and metabolic benefits of Fuzhuan brick tea in mice, and fecal microbiota from tea polyphenol-treated donors alleviated memory impairment and synaptic plasticity deficits when transplanted into recipient animals. Mechanistically, theabrownin from Pu-erh tea suppresses microbial bile salt hydrolase activity, altering bile acid profiles and attenuating intestinal FXR-FGF15 signaling to reduce hepatic steatosis, a mode of action in which a microbial enzyme, not a host protein, is the drug target. Because tea polyphenols have low oral bioavailability and their conversion depends on each person’s resident microbes, individual gut communities may partly explain why the same cup produces different effects in different people. The authors propose metabotype-informed nutrition as an emerging research direction, while cautioning that tea-specific metabotypes have not yet been validated in large prospective cohorts. Microbiome-guided cultivation and controlled fermentation, by contrast, offer immediate translational opportunities, from targeted biofertilizers to defined starter cultures. The cup of tea, it turns out, is never just the plant. It is a rolling ecological collaboration that begins in the soil and ends, remarkably, in us.

Subject of Research: Microbial drivers of tea flavor and bioactivity across cultivation, processing, and gut metabolism

Article Title: From rhizosphere to gut: microbial drivers of tea flavor and bioactivity

Article References: Qi, M., Xie, J., Li, S., Zeng, L., & Luo, L. (2026). From rhizosphere to gut: microbial drivers of tea flavor and bioactivity. npj Science of Food, 10(1), Article 288. https://doi.org/10.1038/s41538-026-01156-0

Image Credits: AI Generated

DOI: 10.1038/s41538-026-01156-0

Keywords: tea, Camellia sinensis, holobiont, rhizosphere microbiome, endophytes, L-theanine, EGCG, fermentation, Aspergillus cristatus, gut microbiota, polyphenols, personalized nutrition

The research team, led by Chao Sun and Mengting Zhu, worked with fresh Zijuan leaves harvested in June 2025 from a plantation in Menghai County, in China’s Yunnan Province, picking the standard one bud with three leaves. The leaves were processed into oolong tea using a carefully standardized protocol: solar and static withering, two rounds of shaking with cooling intervals, piling, fixation at 280 degrees Celsius, rolling, and a two-stage drying sequence. After a three-day rest, the finished tea was divided into four batches. One batch was left unroasted as a control, while the others were roasted for two hours at 80, 110, or 150 degrees Celsius, representing light, medium, and high roasting intensities. All compositional data were expressed on a dry-weight basis to correct for the moisture that roasting drives off, which fell from just over 4 percent in the control to about 1.1 percent in the most heavily roasted sample.

Sensory evaluation followed the Chinese national standard GB/T 23776-2018, with a panel of seven trained assessors, each with more than five years of experience, scoring the teas on a 100-point system weighted toward taste and aroma. The results told a nuanced story. The lightly roasted tea earned the highest overall score of 91.7, praised for an elegant, sweet aroma and a thick, mellow taste with only slight astringency. The unroasted control scored 89.4, with a distinctly spicy aroma but noticeable astringency. Medium roasting produced a score of 89.5 and a return of the spicy, varietal character, while the high-roasted tea scored 90.8, distinguished by a pronounced roasted aroma and the highest taste score of all, a thick and robust mouthfeel with classic oolong character.

Instrumental color analysis of the tea liquors mirrored these impressions. The control tea, rich in anthocyanins, produced a light brownish-red liquor with a visible purple hue. Light and medium roasting shifted the liquor toward a brighter brownish-yellow, a color generally considered more desirable for oolong tea. But heavy roasting reversed the trend: the liquor darkened significantly, its brightness measured by the L* parameter dropping to 16.69 compared with 20.31 in the control, and the purple tone reappeared. The researchers attribute this to two converging processes, the complexation of anthocyanins with other polyphenols under thermal stress and the accumulation of brown melanoidins generated by advanced Maillard reactions, which together darken the infusion and reduce its visual clarity.

Beneath the sensory shifts lay systematic changes in the tea’s non-volatile chemistry. Anthocyanins, the signature compounds of Zijuan, declined steadily with roasting intensity, falling from 2.62 milligrams per gram in the control to 2.11 milligrams per gram after high roasting, a loss of roughly 19.5 percent. Under intense heat, anthocyanins are prone to structural cleavage, including chalcone ring-opening, and can degrade into phenolic acids and aldehydes or participate in co-pigmentation and polymerization with oxidized polyphenols. This pigment breakdown helps explain the visual migration from a purple-tinged liquor to a brownish-red one, and may also contribute to the softening of astringency that panelists noticed in the roasted samples.

The catechins, the bitter and astringent polyphenols central to tea taste, displayed their own choreography. Epi-type catechins, which made up nearly 88 percent of the total, remained stable under light and medium roasting but dropped significantly under high roasting, with epigallocatechin falling by almost 11 percent and epicatechin by more than 12 percent. Meanwhile, the non-epi-type catechin gallocatechin surged by 51.63 percent, a hallmark of thermal epimerization in which epi-forms convert to their non-epi counterparts under heat. Gallic acid rose from 0.51 to 0.80 milligrams per gram, likely released by thermal hydrolysis of galloylated catechins such as EGCG and ECG. The net effect was a 6.33 percent decline in total catechins under high roasting, a shift that weakens the interactions between polyphenols and salivary proteins and helps explain why heavily roasted tea tastes smoother and more full-bodied despite its darker character.

Sugars and amino acids, the fuel for flavor-generating thermal reactions, followed a rise-and-fall pattern. Soluble sugars initially increased from 7.85 to 8.21 milligrams per gram under light roasting, possibly through partial hydrolysis of polysaccharides, then fell to 6.57 milligrams per gram under high roasting as they were consumed in Maillard-type pathways. L-theanine, the predominant amino acid in tea and a key contributor to umami and mellowness, declined relentlessly across the roasting gradient, from 1.19 milligrams per gram in the control to just 0.53 after high roasting. Caffeine, by contrast, stayed essentially flat across all treatments, which means the reduced bitterness of roasted tea owes more to catechin changes than to any loss of this alkaloid.

The volatile chemistry revealed the most spectacular changes of all. Using headspace solid-phase microextraction coupled with gas chromatography-mass spectrometry, the team identified 64 volatile compounds across the treatments. The unroasted control contained only 29 volatiles, dominated by floral oxygenated terpenes such as hotrienol and linalool. Light roasting actually reduced total volatile content, yet improved perceived aroma, likely because diminishing green and sulfurous notes stopped masking the remaining floral and sweet compounds. Medium roasting exploded the volatile inventory to 55 compounds and more than doubled total abundance, introducing 22 new substances including terpenes, pyrroles, and the Strecker aldehydes 3-methylbutanal and 2-methylbutanal, direct chemical evidence of Maillard and Strecker reactions consuming sugars and amino acids. High roasting pushed further, generating roast-aroma pyrazines and furfuryl pyrroles while degrading many fresh, green volatiles, restructuring the profile toward nutty and roasted notes.

Multivariate statistics sharpened the picture. A partial least squares-discriminant analysis model, validated with a 200-permutation test to rule out overfitting, cleanly separated the samples by roasting intensity and identified ten key discriminant volatiles, led by the pyranoid and furanoid oxides of linalool, followed by linalool itself, alpha-farnesene, and 2-pentylfuran. Many of these compounds are associated with floral and fruity aromas, marking them as the chemical fulcrum on which the sensory character of Zijuan oolong turns. The researchers are careful to note that these statistical discriminants are not necessarily the aroma-active compounds themselves, but they provide a powerful fingerprint of how far the roasting process has progressed.

The practical implications reach beyond one purple cultivar. The study demonstrates that roasting is not a simple dial from raw to burnt but a staged chemical program: light roasting polishes and balances, medium roasting builds complexity and varietal intensity, and high roasting delivers the roasted depth that some oolong drinkers prize. Because consumer preferences vary, the authors emphasize that no single roasting level is universally optimal; instead, producers can select intensity according to the flavor profile and market position they want. For a cultivar as chemically distinctive as Zijuan, whose anthocyanin-rich matrix responds to heat in ways conventional green-leaf teas do not, that kind of compositional roadmap could help transform an ancient craft into a precision process, one carefully controlled degree at a time.

Subject of Research: Effects of roasting intensity on the sensory quality and chemical composition of anthocyanin-rich Zijuan oolong tea

Article Title: Effect of roasting intensity on sensory quality and chemical composition of Zijuan oolong tea

Article References: Sun, C., Song, T., Hu, J., Wang, M., Chen, C., Tian, Y., Yang, Y., Luo, Q., Li, Y., Shen, S., Liu, B., & Zhu, M. (2026). Effect of roasting intensity on sensory quality and chemical composition of Zijuan oolong tea. Food Chemistry: X, 39, Article 104462. https://doi.org/10.1016/j.fochx.2026.104462

Image Credits: AI Generated

DOI: 10.1016/j.fochx.2026.104462

Keywords: Zijuan tea, oolong tea, roasting, anthocyanins, catechins, L-theanine, Maillard reaction, volatile compounds, flavor chemistry, tea sensory evaluation, purple tea, Food Chemistry