Characterizing and decoding the effects of co-culture fermentation on dark tea infusion flavor development: elucidating microbial succession and metabolic dynamics mediated by Eurotium cristatum and water kefir

Abstract

Water kefir (a probiotic-rich culture) and Eurotium cristatum (a key fungus in Fu brick tea) were co-fermented with dark tea infusion to investigate dynamic changes in volatile organic compounds (VOCs), microbial communities, and biochemical components. Co-fermentation dramatically altered microbial composition, enriching beneficial lactic acid bacteria (Liquorilactobacillus, Lacticaseibacillus) while inhibiting undesirable microbes (e.g., Saccharomyces). A total of 65 VOCs were identified, with co-fermentation promoting floral, fruity, minty, and stale aromas (e.g., linalool, methyl salicylate, acetophenone) while reducing off-flavors (e.g., 3-methylbutanoic acid). Multivariate analysis revealed core functional microbiota (Gluconobacter, Acetobacter, Eurotium, etc.) strongly associated with flavor formation. Additionally, co-fermentation modulated organic acids, free amino acids, and catechins, improving sensory attributes. Sensory evaluation confirmed superior aroma complexity (enhanced minty/fungal floral notes) and balanced taste (reduced sourness/bitter/astringency characteristics). These findings demonstrate that water kefir and E. cristatum co-fermentation synergistically improves dark tea flavor characteristics, offering a promising strategy for industrial-scale probiotic tea beverage development.

Graphical abstract

Highlights

• •Dark tea infusion was firstly co-fermented by water kefir and E. cristatum.
• •Co-fermentation drastically increased the flavors with floral, fruity, mint, and stale aromas.
• •Co-fermentation with E. cristatum increased the abundance of lactic acid bacteria.
• •Co-fermentation affected the levels of organic acids, free amino acids, and catechins.
• •Co-fermentation greatly improved the sensory characteristics of dark tea infusion.

1. Introduction

Dark tea, a post-fermented tea originating in China, is globally recognized for its health benefits and unique sensory attributes (Lin et al., 2021; Xiao et al., 2024). Traditionally produced via solid-state fermentation (SSF), dark tea faces challenges such as prolonged processing, inconsistent quality, and microbial contamination risks (Chen et al., 2021; Du et al., 2022). Liquid-state fermentation (LSF) has emerged as a promising alternative, offering advantages like enhanced quality control, scalability, and reduced contamination, thereby aligning with industrial demands for standardized tea beverages (Chen et al., 2021; Huang et al., 2023; Liang et al., 2021).

The fermentation process plays a pivotal role in defining the flavor profile and health-stimulating effects of dark tea (Guo et al., 2025; Huang et al., 2025; Xiao et al., 2024; Guo et al., 2024; Yang et al., 2025). Among the microorganisms involved, Eurotium cristatum is particularly significant in the production of Fu brick tea (FBT), a distinctive post-fermented dark tea originating from China's Shanxi and Hunan provinces. Studies have shown that E. cristatum contributes substantially to the development of FBT's characteristic aroma (Wang et al., 2024; Xiao et al., 2024). Our earlier work revealed that inoculating dark tea with E. cristatum markedly enhances its volatile organic compounds (VOCs) content, imparting a fungi flower and stale fragrance that elevates its sensory appeal (Xiao, Huang, et al., 2022). Furthermore, solid-state fermentation with E. cristatum drives the conversion of esterified catechins into degalloylated catechins, simple phenolic acids, and theabrownins through deglycosylation, glycosylation, degradation, methylation, and oxidative polymerization, ultimately refining the tea's taste (Xiao, He, et al., 2022). The unique flavor and taste of FBT are primarily attributed to extracellular enzymes secreted by E. cristatum, which catalyze the degradation and transformation of amino acids, polyphenols, and other components, thereby generating its characteristic sensory properties (Du et al., 2022; Xiao et al., 2024). Recent studies highlight the potential of E. cristatum in liquid-state fermentation (LSF) for tea-based beverages. For instance, Huang et al. (2023) employed E. cristatum in LSF to produce a fermented dark tea drink with enhanced aromatic qualities, including prominent mint, floral, and fruity notes. Similarly, instant dark tea processed via E. cristatum-mediated LSF showed elevated levels of both non-volatile compounds (e.g., non-galloylated catechins, gallic acid, alkaloids, thearubigins, and theabrownins) and VOCs (e.g., methyl salicylate and linalool), improving its overall sensory profile (An et al., 2021, An et al., 2023). Despite these advances, the potential of combining E. cristatum with probiotic starters in tea fermentation remains largely unexplored.

Probiotic-enriched foods have gained significant attention due to their positive effects on human health, with recent research largely focusing on their incorporation into dairy-based formulations (de Souza et al., 2023). Nevertheless, dairy-derived probiotic products present challenges, including allergic reactions to milk proteins, lactose malabsorption, and elevated cholesterol levels. Additionally, consumer preferences for improved taste profiles and the growing adoption of plant-based diets have driven the exploration of non-dairy substrates as viable alternatives for probiotic delivery (Corona et al., 2016; Wang, Sun, et al., 2022b). Among plant-derived fermented options, water kefir has emerged as a particularly promising non-dairy probiotic beverage, contributing to the expanding market of plant-based fermented products (Corona et al., 2016; Ma et al., 2024). Water kefir grains, often referred to as sugary kefir, have attracted considerable research interest owing to their rich and diverse probiotic microbiota (Tu et al., 2019). These grains host a complex microbial community, predominantly composed of lactic acid bacteria (LAB) (particularly Lacticaseibacillus and Liquorilactobacillus), yeasts (notably Saccharomyces), and acetic acid bacteria (AAB) (mainly Acetobacter and Gluconobacter) (Patel et al., 2022). Unlike traditional fermentation systems, water kefir microbiota demonstrates remarkable substrate versatility, efficiently fermenting various plant-based materials, including vegetables, fruits, and cereals, which enables the production of novel probiotic beverages (Corona et al., 2016; Patel et al., 2022). Although water kefir has been widely used in vegetable- and fruit-based fermentations, its application in tea fermentation remains unexplored. A key challenge is that LAB and AAB—the dominant microbial groups in water kefir—generate substantial acidity, often leading to an excessively sour taste (Patel et al., 2022; Tu et al., 2019). This high acidity may negatively impact the sensory quality of tea, likely contributing to the scarcity of studies on water kefir-fermented tea. Consequently, innovative approaches are needed to optimize both probiotic functionality and consumer acceptability in such beverages.

Co-fermentation, leveraging microbial synergies, has proven effective in enhancing flavor profiles. For instance, co-culture of Saccharomyces boulardii with Lactiplantibacillus plantarum in green tea fermentation significantly enhanced geraniol and methyl salicylate production, intensifying fruity-minty aroma profiles (Wang, Sun, et al., 2022a). Similarly, E. cristatum combined with Aspergillus niger enriched the flavor profile of instant dark teas (Chen et al., 2021). Besides, co-fermentation of red jujube juice using L. plantarum and Streptococcus thermophilus significantly enhanced the production of flavor compounds and organic acids compared to monoculture fermentation, leading to improved product quality (Li, Xu, et al., 2024). These findings underscore the potential of co-fermentation to integrate functional and sensory advantages of distinct microbial strains. We hypothesize that co-fermenting water kefir with E. cristatum could merge the probiotic richness of kefir with the aroma-enhancing properties of E. cristatum, addressing acidity issues and improving desirable sensory traits.

Herein, we propose co-fermentation of dark tea infusion with water kefir and E. cristatum to develop a novel probiotic-rich beverage with enhanced aroma and taste. This study investigates: (1) dynamic shifts in microbial communities during fermentation, (2) volatile organic compound (VOC) profiles and their correlation with microbiota, and (3) changes in key biochemical components such as organic acids, amino acids, and catechins. The findings will advance the development of novel fermented tea beverages with enhanced flavor characteristic and sensory qualities, offering practical insights for the tea industry.

2. Materials and methods

2.1. Chemicals and materials

All C6-C21 n-alkanes and ethyl decanoate (99.99 % purity) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Primary dark tea, used as the raw material for FBT production, was sourced from Anhua Tea Factory (Yiyang, Hunan, China). The fungus E. cristatum was isolated from FBT in our laboratory. Our previous studies demonstrated that E. cristatum contributes to the improvement of flavor characteristics and taste attributes of dark tea during fermentation (Xiao, He, et al., 2022, Xiao, Huang, et al., 2022). It was stored on a M40Y agar slant and regularly sub-cultured. Commercial water kefir grains were obtained from a supplier in Shenyang, Liaoning Province, China. Preliminary characterization revealed that the microbial composition was dominated by acetic acid bacteria (AAB), yeasts, and lactic acid bacteria (LAB). All other used reagents and substances were analytical grade.

2.2. Preparation of inoculum cultures

2.2.1. E. cristatum inoculum starter preparation

The E. cristatum inoculum was prepared by reviving the preserved strain through two successive cultivations on M40Y agar (containing 20 g/L maltose extract, 5 g/L yeast extract, 20 g/L agar) at 28 °C for 7 days, as previously described in our study (Xiao, Huang, et al., 2022). Subsequently, spore suspensions were obtained by washing the E. cristatum fungal spores with sterile water. The above-described microbial procedures were carried out in a sterile environment.

2.2.2. Water kefir inoculum starter preparation

In order to preserve the viability of the water kefir grains before fermentation, they were stored in a solution with brown sugar (100 g/L) as described by Tu et al. (2019). The brown sugar solution had been previously pasteurized at 121 °C for 20 min. The water kefir grains were then added to the sugar solution in a concentration of 12 % (w/v) and incubated at 28 °C for 24 h in a glass jar covered with gauze. Following incubation, a sterile sieve was used to separate the kefir grains from the liquid. The water kefir filtrate was utilized in this study for fermentation, while the kefir grains were reused for further water kefir fermentation in the sugar solution.

2.3. Preparation of fermented dark tea infusions

The dark tea infusion was prepared by powdering raw dark tea leaves through a 40-mesh sieve followed by hot water extraction (90 °C, 1:50 w/v) for 45 min. The centrifuged supernatant was sterilized (121 °C, 20 min) and cooled to ambient temperature. Fermentation was conducted in triplicate using 140 mL aliquots in 250 mL Erlenmeyer flasks. Three experimental groups were established: (1) K group: water kefir (7 mL) + sterile water (2.8 mL); (2) J group: sterile water (7 mL) + E. cristatum spores (2.8 mL, 10⁷ CFU/mL); (3) JK group: water kefir (7 mL) + E. cristatum spores (2.8 mL, 10⁷ CFU/mL). All groups were shaken (120 rpm, 28 °C) for 8 days. Samples were collected every two days from day 0 to day 8 (K0-K8, J0-J8, JK0-JK8), then stored at −20 °C for subsequent analysis. C0 represents the unfermented control (raw dark tea infusion) used for all experimental groups (K, J, and JK) in the analysis of flavor, taste, and non-volatile metabolites.

2.4. Volatile organic compounds (VOCs) analyzed by HS-SPME-GC–MS

VOCs were analyzed following our established protocol (Huang et al., 2023). Briefly, 6 mL of dark tea infusion was mixed with 0.5 g NaCl and 10 μL internal standard ethyl decanoate (10 ppm) in a headspace vial. VOCs were extracted at 80 °C for 50 min using a 50/30 μm DVB/CAR/PDMS fiber (PA, USA), then desorbed in the GC–MS inlet at 250 °C for 5 min. Chromatographic separation was performed using a 30 m × 0.25 mm i.d., 0.25 μm film thickness Agilent HP-5MS capillary column with high-purity helium as the carrier gas (flow rate: 1.0 mL/min). The temperature gradient began with an initial hold at 40 °C for 3 min, followed by programmed increases: first at speed of 2 °C/min to 90 °C, then at rate of 3 °C/min to 150 °C, subsequently to 180 °C (5 °C/min), and finally to 230 °C (15 °C/min) and maintain for 5 min. MS detection was conducted in electron impact mode (70 eV) using full-scan acquisition (mass range: m/z 35–400) with a 230 °C ion source. Compounds identification combined NIST 2017 spectral matching, retention indices (calculated via C6–C21 n-alkanes), and literature verification. The quantification of VOCs used internal standard method, while odor contributions were assessed via relative odor activity value (rOAV) derived from published odor thresholds (Huang et al., 2025; Xiao et al., 2024).

2.5. Microbial community analysis

Genomic DNA was extracted from samples using the conventional cetyl trimethylammonium bromide (CTAB) approach (Jiang et al., 2023), which is effective for low-biomass samples and diverse microbial populations. Following extraction, DNA was dissolved in 50 μL of elution buffer, while nuclease-free water served as the negative control. Fungal ITS regions were amplified using primers ITS2 (5'-TCCTCCGCTTATTGATATGC-3′) and ITS1FI2 (5'-GTGARTCATCGAATCTTTG-3′), while bacterial 16S rRNA V3-V4 regions were targeted with primers 805R (5'-GACTACHVGGGTATCTAATCC-3′) and 341F (5'-CCTACGGGNGGCWGCAG-3′), each containing unique barcodes for multiplexing. PCR amplification was conducted in 25 μL reactions including 2.5 μL of each primer (10 μM), 12.5 μL of PCR Premix, and 25 ng of template DNA, with the volume adjusted using PCR-grade water. The thermal cycling conditions consisted of initial denaturation for 30 s at 98 °C, followed by 32 cycles of denaturation (98 °C, 10 s), annealing (54 °C, 30 s), and extension (72 °C, 45 s), with a final extension for 10 min at 72 °C. The resulting amplicons were purified by gel electrophoresis (2 % agarose), cleaned using AMPure XT beads, and quantified with a Qubit fluorometer (Invitrogen, USA). Prior to sequencing, the amplicon library was evaluated for quality and fragment size distribution using the Kapa Biosciences Library Quantification Kit (Illumina) and the Agilent 2100 Bioanalyzer (Agilent Technologies, USA), respectively. High-throughput sequencing was carried out on the Illumina NovaSeq 6000 platform. Raw sequencing data (16S rRNA: PRJNA1287290; ITS: PRJNA1287301) are deposited in the NCBI Sequence Read Archive.

2.6. Determination of organic acids, free amino acids, catechins and gallic acid

Organic acids were quantified using an Agilent 1260 HPLC system with a modified method (Liu et al., 2022). Tea infusions were filtered through a 0.22 μm PVDF membrane prior to separation on an Agilent ZORBAX SB-C18 column (4.6 × 250 mm, 5 μm) kept at 40 °C, with 20 μL injection volume. The mobile phase including Eluent A (0.1 % phosphoric acid in water) and Eluent B (2 % phosphoric acid in acetonitrile), delivered at 1.0 mL/min under gradient conditions: 2.5 % B (0–10 min), 2.5–100 % B (10–15 min), 100 % B (15–20 min), and re-equilibration to 2.5 % B (20–25 min). Detection was carried out at 210 nm using a variable wavelength detector (G7114A VWD) coupled with an autosampler (G7114A) and quaternary pump (G7111A). Organic acids were quantified in mg/L of tea infusion by comparison with authentic standards. Free amino acids were analyzed by GC–MS following our recently published protocols (Huang et al., 2025; Xiao et al., 2024). Sample derivatization was performed using propyl chloroformate (Aladdin Industrial Co.) prior to chromatographic separation on an Agilent HP-5MS capillary column (30 m × 0.25 mm × 0.25 μm) with an Agilent 7000D GC–MS system. The analysis incorporated stable isotope internal standards acetic acid‑d₄ and DL-alanine-3,3,3-D3(McLean Biochemicals Ltd.; Shanghai, China). Samples (1 μL) were injected in split mode (10:1 ratio) at 260 °C, using helium carrier gas (1.0 mL/min). The temperature program began at 50 °C (5 min hold), followed by sequential ramps: 10 °C/min to 70 °C, 3 °C/min to 85 °C, 5 °C/min to 215 °C, and finally 10 °C/min to 280 °C (5 min hold). Mass spectrometry conditions included a 70 eV electron energy, ion source temperature of 230 °C, and scanning from m/z 50–800 in selected ion monitoring mode for data acquisition. The gallic acid (GA) and catechins were quantified by using an Agilent 1260 HPLC system following our previously established method (Xiao, He, et al., 2022), and the quantities were reported as mg/L of tea infusions.

2.7. Sensory evaluation

A panel of ten trained assessors (5 males and 5 females, aged 22–50 years) evaluated the aroma and taste profiles of fermented dark tea infusions. The assessment was conducted following the Chinese Tea Sensory Evaluation Standard (GB/T 23776–2018) with minor modifications (Xiao, Huang, et al., 2022). Prior to the evaluation, all panelists completed an intensive 200 h training program to enhance their ability to identify and articulate key sensory attributes. Through collaborative discussion, a standardized lexicon was established, encompassing nine aroma descriptors (fragrant, floral, sour, minty, woody, fruity, herbal, green, and fungal floral) and seven taste characteristics (mellow, refreshing, sour, sweet, bitter, astringent, and umami). Samples were presented in randomized coded order, with palate cleansing (using water) between tastings to minimize carryover effects. A 10-point hedonic scale was employed for intensity scoring, where higher values denoted more pronounced attributes. The final sensory scores represented the mean ratings from all ten panelists. Additionally, tea infusion color was quantified using a SC–80C colorimeter (Kangguang Instrument Co. Ltd., Beijing, China).

2.8. Statistical analysis

Data are recorded as mean ± standard deviation (SD) from three independent replicates. Statistically significant difference was assessed using ANOVA and further analyzed with Duncan's test (p < 0.05) through SPSS 17.0 software.

3. Results and discussion

3.1. Dynamic alteration of microbial community in the process of fermenting dark tea infusion

3.1.1. Analysis of alpha and beta diversity of microbiota

Alpha diversity analysis indicated sufficient sequencing depth (coverage >99.99 %) to capture microbial richness (Tables S1–S2). Rarefaction curves for Shannon diversity plateaued, indicating comprehensive species representation (Fig. S1). Notably, the co-fermentation group (JK) showed significantly higher Shannon and Simpson indices than the kefir-fermented group (K) in the later stages (days 6–8), indicating that E. cristatum promotes microbial diversity in kefir-fermented tea infusions. Principal component analysis (PCA) of beta diversity revealed distinct shifts in microbial communities across different fermentation periods (Fig. 1A-B). Regarding bacterial community structure, PCA demonstrated clear separation between the JK and K groups, suggesting that E. cristatum co-fermentation substantially reshaped the bacterial composition. Moreover, this divergence became more pronounced as fermentation progressed (Fig. 1A). For fungal communities, the JK and K groups exhibited similar distributions in the early fermentation stage (days 2–4). However, the fungal community structure diverged markedly between the JK and K groups in the later fermentation stage (days 6–8) (Fig. 1B). Collectively, these findings demonstrate that the incorporation of E. cristatum significantly modifies the microbial ecology of water kefir-fermented dark tea infusion, particularly in the later fermentation stages.

Fig. 1.

Principal component analysis of bacteria (A) and fungi (B) for the changes in microbial community of dark tea infusions during fermentation. Chord diagram visualizing relative abundances of the predominant five bacterial genera (C) and fungal genera (D) in the process of fermenting dark tea infusions. The relative abundance of the bacteria (top 30 genera) (E) and fungi (top 30 genera) (F) alterations in the dark tea infusions during fermentation. (G) Dynamic alteration in pH value of dark tea infusions during fermentation. (H) Network correlation analysis of the top 20 bacteria and the top 20 fungi in relative abundance. (I) Network visualized the relationship between E. cristatum and top 20 bacteria and fungi of dark tea infusions during fermentation. The line represents the Spearman correlation coefficient between the microbes, the red line denotes a positive correlation, whereas the blue line signifies a negative correlation. Note: In Fig. 1A (bacterial PCA), Fig. 1C and Fig. 1E, K0 and JK0 are merged as “K0JK0” because their bacterial communities are identical at day 0. In Fig. 1B (fungal PCA), Fig. 1D and Fig. 1F, K0 and JK0 are shown separately due to the immediate impact of E. cristatum inoculation on fungal composition. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.1.2. Change of microbial community structure during fermentation

The bacterial and fungal community composition of dark tea infusions during fermentation was analyzed at both phylum and genus levels (Fig. 1 and Fig. S2). At the phylum level, Firmicutes and Proteobacteria dominated the bacterial communities across all samples (Fig. S2A, S2C). Firmicutes was particularly abundant in non-fermented dark tea infusions (representing native water kefir microbiota), consistent with its reported prevalence in water kefir (Zannini et al., 2022). However, Proteobacteria gradually became the dominant phylum as fermentation progressed. Fungal communities were overwhelmingly dominated by Ascomycota (98.91–99.34 % relative abundance) in all samples (Fig. S2B, S2D). Basidiomycota and Zygomycota were detected but represented only minor components of the fungal microbiota. LEfSe analysis (Linear Discriminant Analysis Effect Size) revealed significant differential abundance of bacterial taxa during fermentation (Fig. S3). This method identifies taxa with statistically significant abundance differences between experimental groups (biomarkers) and provides visual representation through cladograms and LDA score histograms. The resulting cladogram (Fig. S3A—D) visually represents taxonomic hierarchies from kingdom to species (concentric circles), where node size reflects relative abundance and color indicates group-specific enrichment (yellow: non-significant; red: enriched in red-labeled group, etc.). Significant phyla are explicitly labeled, while other biomarkers are annotated with letters. Key findings from LEfSe analysis demonstrated that Firmicutes and Proteobacteria served as primary biomarkers distinguishing JK and K groups during mid-late fermentation (days 4–8) (Fig. S3A—D). Specifically, Firmicutes was more abundant in the JK group, while Proteobacteria dominated in the K group. The influence of Firmicutes as a biomarker strengthened over time, as evidenced by its high LDA scores (>4) on days 6–8 (Fig. S3c-d). At the genus level, in the early fermentation phase (days 2–4), Acetobacter and Gluconobacter served as biomarkers, with Acetobacter being more prevalent in the JK group and Gluconobacter in the K group (Fig. S3a-b). Notably, Acetobacter remained a significant biomarker until the end of fermentation (day 8). Liquorilactobacillus became a distinguishing biomarker starting from the mid-fermentation stage (day 4), showing higher abundance in the JK group (red bars in the histogram). Its discriminatory power increased progressively, with LDA scores exceeding 4 on days 6–8 (Fig. S3c-d). These results further indicate that E. cristatum co-fermentation markedly reshaped the microbial community during water kefir fermentation of dark tea infusion.

At the bacterial genus level, lactic acid bacteria (LAB) were the predominant microorganisms in the unfermented dark tea infusion sample (representing native water kefir microbiota), which were identified as Liquorilactobacillus (47.33 %), Lacticaseibacillus (26.33 %), Schleiferilactobacillus (2.29 %), and Lentilactobacillus (1.03 %). The second dominant bacteria were acetic acid bacteria (AAB) that were identified as Gluconobacter (12.60 %) and Acetobacter (9.34 %) (Fig. 1C and E). This finding aligned with Patel et al. (2022), who also reported AAB (predominantly Acetobacter and Gluconobacter) and LAB (predominantly Lacticaseibacillus and Liquorilactobacillus) were the primary bacterial genera in water kefir. As fermentation advanced, the relative abundance of Acetobacter notably increased, with the JK group showing a higher increase than the K group (Fig. 1E). It is worth noting that Liquorilactobacillus in the JK group also began to significantly increase after 6 days of fermentation, becoming one of the dominant bacterial genera. After 8 days of fermentation, its relative abundance reached to 3.66 %. Even though the abundance of Liquorilactobacillus in the K group also increased, its relative abundance (< 1 %) was much lower compared with the JK group. This result suggested that co-fermentation with E. cristatum promoted the growth of Liquorilactobacillus. The correlation analysis also found that Liquorilactobacillus and E. cristatum were highly positive correlated (r = 0.567, p < 0.05) (Fig. 1H-I). In addition, we have also found that the abundance of Lacticaseibacillus in JK2, JK4, JK6 and JK8 were 4.0-, 10.0-, 1.8-, 4.0-fold higher compared with the K groups, respectively, which implied that co-fermentation with E. cristatum also promoted the growth of Lacticaseibacillus, and the correlation analysis result also verified this statement (r = 0.650, p < 0.01) (Fig. 1H-I). These findings are consistent with Kou et al. (2024), who reported that E. cristatum stimulates LAB growth. E. cristatum secretes various hydrolytic enzymes (e.g., glycoside hydrolase, glycosyltransferase, tannase, laccase, and protease) during fermentation (Sun et al., 2023; Xiao, Huang, et al., 2022). These enzymes break down macromolecules, releasing carbon and nitrogen sources that may facilitate LAB proliferation (e.g., Liquorilactobacillus and Lacticaseibacillus). Similarly, Ponomarova et al. (2017) observed that yeast-derived nutrients (e.g., amino acids and vitamins) support LAB growth. This cross-feeding mechanism is also evident in probiotic tea fermentation, where yeast-LAB co-culture significantly enhances LAB viability (Wang, Sun, et al., 2022a).

At the fungal genus level, Saccharomyces dominated the eukaryotic genus during fermentation in both K and JK groups (Fig. 1D and F). Saccharomyces comprised 84.97–98.30 % of the total fungal relative abundance across all sequences during the fermentation process in K group. Previous studies have reported that Saccharomyces, an indispensable microbe in kefir fermentation, can hydrolyze saccharose into glucose and fructose and synthesize flavor compounds (Patel et al., 2022). It was also found that the relative abundance of Saccharomyces in the co-fermentation group was lower compared with the K group during fermentation. Eurotium was the secondary dominant fungal genus during in JK group. Notably, the relative abundance of Eurotium decreased from 31.06 % to 1.53 % after 2 days of fermentation. This finding might be ascribed to the proliferation of AAB in kefir, resulting in the generation of substantial acids that lower the pH of the dark tea infusion (Fig. 1G), thereby inhibited the growth of Eurotium. As fermentation progressed, LAB and other microbes in water kefir metabolized acids, gradually increasing the pH of the dark tea infusion. This shift enabled E. cristatum to recover and proliferate, leading to subsequent changes in the microbial community structure. This pH-dependent succession mirrors findings in kombucha fermentation studies (Cheng et al., 2024), where microbial activities are similarly modulated by acidification patterns. Notably, the abundance of Eurotium reached to 13.04 % after 8 days of fermentation. Eurotium (i.e., E. cristatum) was reported to possess glycoside hydrolases, such as galactosidases, arabinofuranosidases, rhamnosidases, glucosidases, and glucuronidases, which catalyzed the hydrolysis of glycosidic bonds in tea leaves glycosides (An et al., 2021; Sun et al., 2023), facilitating the conversion of polysaccharides into monosaccharides to accommodate the nutritional needs of other microorganisms. The correlation results revealed that Eurotium was closely correlated with Asterotremella, Alternaria, Malassezia, etc. (Fig. 1I). Thus, the introduction of E. cristatum greatly influenced the fungi community structure of water kefir during co-fermentation.

3.2. Dynamic variations in volatile organic compounds across the fermentation of dark tea infusions

A total of 65 volatile organic compounds (VOCs) were identified in dark tea infusions during fermentation (Table 1). The co-fermentation group (JK) exhibited higher total VOCs content than monoculture groups (J or K) during the later fermentation phase (6–8 days), particularly for alcohols, esters, and ketones (Fig. 2A–B). These VOCs were categorized into 15 alcohols, 8 aldehydes, 11 esters, 10 ethers, 12 ketones, 4 acids, 2 phenols, 2 hydrocarbons, and 1 other compound (Table 1, Fig. 2C). PCA and hierarchical clustering analysis (HCA) distinguished VOC profiles among groups, revealing that JK initially aligned with the kefir group (K) but shifted toward the E. cristatum group (J) by day 6–8 (Fig. 2D-E), indicating sequential microbial dominance: kefir initially shaped flavor development, whereas E. cristatum dominated thereafter. The heat-map analysis (Fig. 2F) demonstrated that the VOCs in dark tea infusions were significantly influenced by microbial cultures and fermentation duration. Furthermore, the Venn diagram (Fig. S4) illustrated both unique and shared VOCs among the experimental groups, with the co-culture group (JK) displaying overlapping yet distinct volatile profiles compared to monocultures. Notably, as shown in Fig. S4, the number of shared VOCs between co-culture and monoculture groups decreased during fermentation, while the abundance of unique compounds increased, leading to greater compositional divergence over time. For instance, while 33 common VOCs were detected between co-culture and monoculture groups after 2 days of fermentation, this number decreased to only 13 shared compounds by day 8. This substantial reduction suggests that microbial co-culture markedly reshapes metabolic pathways, further supporting the pronounced impact of co-fermentation on dark tea infusion flavor development.

Table 1.

Alteration of volatile organic compounds (VOCs) using GC–MS in the fermentation of dark tea infusions.

					Content(μg/L)
No.	Compounds	RT (min)	RIA/RIB	Odor description	0 day	2 days			4 days			6 days			8 days
					C0	K2	JK2	J2	K4	JK4	J4	K6	JK6	J6	K8	JK8	J8
	Alcohols
V1	Trans-3-hexenol	8.623	855/852	Green	n.d.	n.d.	n.d.	1.41 ± 0.18c	n.d.	n.d.	3.55 ± 0.11a	n.d.	n.d.	1.77 ± 0.28b	n.d.	n.d.	n.d.
V2	Heptanol	14.498	969/970	/	0.36 ± 0.01d	n.d.	n.d.	n.d.	n.d.	n.d.	1.35 ± 0.12c	n.d.	n.d.	13.07 ± 1.41a	n.d.	n.d.	2.67 ± 0.71b
V3	1-Octen-3-ol	14.934	977/980	Floral, Mushroom, Cucumber, Earth, Fat	1.23 ± 0.04c	n.d.	n.d.	0.57 ± 0.06c	n.d.	n.d.	0.83 ± 0.06c	n.d.	n.d.	10.14 ± 0.92b	n.d.	1.07 ± 0.10c	45.56 ± 2.25a
V4	2-Ethylhexanol	18.014	1026/1030	Green, Rose	4.54 ± 0.05a	1.22 ± 0.16e	1.28 ± 0.22e	2.72 ± 0.55c	1.16 ± 0.13e	1.11 ± 0.07e	2.11 ± 0.24d	1.11 ± 0.20e	4.63 ± 0.20a	2.09 ± 0.06d	n.d.	3.32 ± 0.04b	2.68 ± 0.29c
V5	(Z)-Linalool oxide	20.822	1067/1074	Woody, Floral	2.46 ± 0.04c	2.34 ± 0.12cd	0.87 ± 0.10f	0.61 ± 0.03f	6.43 ± 0.52b	1.65 ± 0.18de	0.72 ± 0.10f	3.04 ± 0.43c	6.70 ± 0.81b	0.97 ± 0.06ef	1.14 ± 0.06ef	12.70 ± 0.99a	0.80 ± 0.11f
V6	(E)-Linalool oxide	21.924	1083/1086	Floral	1.04 ± 0.02f	1.29 ± 0.17ef	1.34 ± 0.21ef	0.76 ± 0.17fg	4.87 ± 0.64c	1.30 ± 0.00ef	1.20 ± 0.13f	3.10 ± 0.43d	7.04 ± 1.47b	2.27 ± 0.30de	n.d.	11.63 ± 1.03a	n.d.
V7	Linalool	22.842	1096/1099	Floral, Coriander, Lavender, Lemony, Rose	18.03 ± 0.15d	16.95 ± 0.70de	19.29 ± 1.24cd	21.76 ± 1.13bc	18.20 ± 0.91d	14.32 ± 1.60ef	17.16 ± 0.61de	9.82 ± 1.42g	23.21 ± 4.21b	13.22 ± 2.63f	13.11 ± 1.06f	37.70 ± 2.04a	18.69 ± 0.46d
V8	2-Phenylethanol	23.737	1108/1116	Fruity, Honey, Lilac, Rose, Wine	0.58 ± 0.01d	28.39 ± 0.87a	1.84 ± 0.46b	1.38 ± 0.10bc	n.d.	n.d.	1.56 ± 0.13bc	n.d.	n.d.	1.22 ± 0.22c	n.d.	n.d.	n.d.
V9	(−)-trans-Chrysanthenol	26.298	1137/1142	/	0.39 ± 0.01a	n.d.	0.24 ± 0.02c	0.27 ± 0.04b	n.d.	n.d.	0.21 ± 0.04d	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V10	4-Terpineol	29.157	1170/1177	Woody, Nutmeg, Earth, Must	1.02 ± 0.02b	0.48 ± 0.02de	0.45 ± 0.06de	0.53 ± 0.03cde	0.62 ± 0.05c	n.d.	0.42 ± 0.05e	0.45 ± 0.09de	0.58 ± 0.07cd	0.59 ± 0.04cd	0.47 ± 0.06de	1.23 ± 0.21a	1.02 ± 0.06b
V11	α-Terpineol	30.373	1184/1189	Herbal, Anise, Fresh, Mint	2.18 ± 0.03ef	2.74 ± 0.21cd	2.78 ± 0.09c	2.80 ± 0.52c	2.72 ± 0.24cd	2.05 ± 0.22ef	3.07 ± 0.14c	1.80 ± 0.13fg	4.54 ± 0.42b	2.31 ± 0.25de	1.80 ± 0.12fg	7.64 ± 0.22a	1.55 ± 0.31g
V12	Nerol	33.444	1230/1228	Floral, Fruity	n.d.	1.75 ± 0.04c	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	10.75 ± 0.48a	n.d.	n.d.	2.25 ± 0.14b	n.d.
V13	Geraniol	35.082	1259/1255	Geranium, Lemony Peel, Passion Fruit, Peach, Rose	0.79 ± 0.03b	3.17 ± 0.12a	0.46 ± 0.05d	n.d.	n.d.	n.d.	n.d.	n.d.	0.45 ± 0.02d	n.d.	n.d.	0.56 ± 0.05c	n.d.
V14	Nerolidol	45.878	1563/1564	Woody, Fir, Pine	0.25 ± 0.00c	0.38 ± 0.07b	0.65 ± 0.09a	0.20 ± 0.01c	n.d.	0.28 ± 0.09c	0.23 ± 0.05c	0.27 ± 0.03c	n.d.	n.d.	0.60 ± 0.07a	n.d.	n.d.
V15	Cedrol	46.86	1602/1598	Woody, Cedarwood, Sweet	0.55 ± 0.02ab	0.38 ± 0.05cd	0.55 ± 0.02ab	0.61 ± 0.05a	0.31 ± 0.01d	0.52 ± 0.05b	0.44 ± 0.07c	0.39 ± 0.07cd	0.44 ± 0.05c	0.43 ± 0.08c	0.31 ± 0.01d	0.43 ± 0.02c	n.d.
	Aldehydes
V16	Benzaldehyde	13.658	955/962	Roasted Peppe, Bitter Almond, Burnt Sugar, Cherry, Malt, Fruity	9.58 ± 0.14a	1.94 ± 0.03cde	2.02 ± 0.04cd	1.21 ± 0.04hi	2.61 ± 0.01b	2.16 ± 0.08c	1.81 ± 0.07def	1.42 ± 0.32gh	1.56 ± 0.03fg	1.69 ± 0.41ef	1.70 ± 0.01ef	1.02 ± 0.02i	n.d.
V17	1-Ethyl-2-pyrrolecarboxaldehyde	19.257	1044/1046	Savory	0.37 ± 0.02e	1.48 ± 0.16a	1.10 ± 0.05b	0.38 ± 0.01e	n.d.	n.d.	0.49 ± 0.03d	n.d.	0.63 ± 0.11c	0.36 ± 0.09e	n.d.	n.d.	n.d.
V18	Nonanal	23.223	1102/1104	Green, Floral, Fat, Lemony	2.33 ± 0.10a	1.58 ± 0.33b	1.43 ± 0.32bcd	1.43 ± 0.35bcd	1.16 ± 0.15cd	1.34 ± 0.29bcd	1.37 ± 0.07bcd	1.10 ± 0.07d	1.16 ± 0.11cd	1.71 ± 0.29b	1.07 ± 0.23d	1.70 ± 0.06b	1.53 ± 0.11bc
V19	Safranal	31.094	1193/1201	Herbal, Fresh, Spicy, Rosemary, Tobacco	0.60 ± 0.03a	n.d.	n.d.	0.19 ± 0.05b	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V20	Decanal	31.92	1203/1206	Floral, Fried, Orange Peel, Tallow	3.20 ± 0.16a	0.97 ± 0.07d	1.00 ± 0.21d	2.20 ± 0.69b	0.64 ± 0.04d	0.87 ± 0.24d	0.76 ± 0.12d	0.57 ± 0.13d	0.95 ± 0.17d	2.09 ± 0.34b	0.58 ± 0.14d	1.54 ± 0.20c	0.86 ± 0.09d
V21	2,5-Dimethylbenzaldehyde	32.2	1208/1208	/	3.41 ± 0.31b	1.89 ± 0.32c	1.09 ± 0.07d	4.90 ± 0.55a	0.52 ± 0.12e	0.45 ± 0.05e	1.72 ± 0.29c	0.46 ± 0.04e	0.69 ± 0.09de	3.14 ± 0.21b	0.46 ± 0.14e	0.53 ± 0.02e	0.99 ± 0.04d
V22	β-Homocyclocitral	35.408	1264/1254	/	0.52 ± 0.03a	n.d.	n.d.	0.19 ± 0.00c	n.d.	n.d.	0.45 ± 0.07b	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V23	Citral	35.853	1272/1276	Lemony	1.27 ± 0.04a	0.48 ± 0.06c	n.d.	0.21 ± 0.04d	n.d.	n.d.	0.29 ± 0.03d	n.d.	0.63 ± 0.08c	0.95 ± 0.28b	n.d.	n.d.	0.60 ± 0.08c
	Esters
V24	Methyl benzoate	22.351	1089/1094	Herb, Lettuce, Prune, Violet	0.47 ± 0.02c	n.d.	n.d.	0.50 ± 0.02c	n.d.	n.d.	1.46 ± 0.08b	n.d.	n.d.	1.69 ± 0.15a	n.d.	n.d.	n.d.
V25	Ethyl benzoate	28.853	1167/1171	Fruity, Flower, Camomile, Celery, Fat	n.d.	0.83 ± 0.02cd	0.72 ± 0.00cde	0.78 ± 0.10cde	0.69 ± 0.12def	0.62 ± 0.09ef	1.63 ± 0.01b	0.37 ± 0.01g	0.63 ± 0.02ef	2.64 ± 0.28a	0.52 ± 0.02fg	0.77 ± 0.09cde	0.87 ± 0.09c
V26	Methyl salicylate	30.612	1187/1192	Wintergreen, Mint, Almond, Caramel	n.d.	0.65 ± 0.08hi	n.d.	n.d.	8.36 ± 0.70d	1.60 ± 0.14gh	0.40 ± 0.05hi	4.13 ± 0.38e	18.78 ± 1.88b	2.39 ± 0.24fg	3.66 ± 0.60ef	26.92 ± 1.27a	14.18 ± 1.67c
V27	Ethyl caprylate	31.447	1197/1196	Floral, Apricot, Brandy, Pineapple	0.65 ± 0.08c	35.69 ± 1.13a	30.97 ± 3.41b	0.43 ± 0.03c	n.d.	n.d.	0.22 ± 0.02c	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V28	Phenethyl acetate	35.132	1259/1258	Floral, Honey, Rose	n.d.	n.d.	n.d.	0.91 ± 0.09c	n.d.	n.d.	1.80 ± 0.12b	n.d.	n.d.	2.62 ± 0.60a	n.d.	n.d.	1.56 ± 0.32b
V29	Neryl formate	36.404	1282/1283	Green	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	6.09 ± 0.74a	n.d.	n.d.	0.75 ± 0.13b	n.d.
V30	Ethyl nonanoate	37.238	1296/1296	Floral	n.d.	2.08 ± 0.13b	2.66 ± 0.43a	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V31	γ-Nonanolactone	39.691	1360/1363	Coconut, Creamy, Waxy, Sweet, Buttery	n.d.	0.52 ± 0.03d	0.48 ± 0.01d	n.d.	1.70 ± 0.18a	1.31 ± 0.27c	n.d.	1.55 ± 0.21ab	n.d.	n.d.	1.41 ± 0.07bc	n.d.	n.d.
V32	Ethyl 9-decenoate	40.664	1386/1387	Fruity, Fatty	n.d.	1.11 ± 0.25b	1.57 ± 0.26a	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V33	Dihydroactinidiolide	44.996	1529/1532	Herbal, Musk, Coumarin	1.16 ± 0.078b	1.13 ± 0.09bc	1.56 ± 0.16a	1.24 ± 0.09b	0.93 ± 0.12cd	1.07 ± 0.13bc	0.62 ± 0.15e	0.95 ± 0.20cd	0.82 ± 0.02d	0.37 ± 0.11f	0.75 ± 0.06de	0.86 ± 0.07d	n.d.
V34	2,2,4-Trimethyl-1,3-pentanediol diisobutyrate	46.718	1596/1588	/	0.34 ± 0.01de	0.88 ± 0.12a	0.34 ± 0.03de	0.24 ± 0.01e	0.66 ± 0.07b	0.35 ± 0.04de	0.61 ± 0.08b	0.44 ± 0.09cd	0.56 ± 0.03bc	0.62 ± 0.06b	n.d.	0.90 ± 0.09a	0.70 ± 0.18b
	Ethers
V35	1,2-Dimethoxybenzene	26.858	1144/1148	Woody, Earth, Moss	0.43 ± 0.02c	n.d.	n.d.	0.48 ± 0.07c	n.d.	n.d.	1.15 ± 0.18b	n.d.	n.d.	1.32 ± 0.18a	n.d.	n.d.	1.35 ± 0.21a
V36	4-Methoxystyrene	27.146	1147/1156	Sweet	n.d.	n.d.	n.d.	0.95 ± 0.11	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V37	2,3-Dihydrobenzofuran	32.938	1221/1224	/	n.d.	n.d.	0.54 ± 0.05a	n.d.	0.51 ± 0.01b	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V38	2,4-Dimethoxytoluene	34.232	1244/1244	Stale	n.d.	n.d.	n.d.	0.41 ± 0.03b	n.d.	n.d.	0.69 ± 0.12b	n.d.	0.52 ± 0.16b	1.19 ± 0.55a	n.d.	n.d.	1.06 ± 0.22a
V39	Theaspirane	36.487	1283/1302	Honey	0.35 ± 0.01b	0.40 ± 0.05a	0.33 ± 0.03b	0.19 ± 0.03c	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V40	Edulane	37.795	1309/1314	/	1.21 ± 0.01b	0.85 ± 0.04c	0.98 ± 0.02c	0.82 ± 0.12c	1.63 ± 0.30a	0.48 ± 0.03d	0.52 ± 0.02d	0.54 ± 0.02d	0.46 ± 0.04d	0.24 ± 0.00e	0.49 ± 0.16d	n.d.	n.d.
V41	1,2,3-Trimethoxybenzene	37.891	1312/1313	Stale	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	0.27 ± 0.02b	n.d.	n.d.	0.34 ± 0.00a	n.d.	n.d.	n.d.
V42	4-Ethyl-1,2-dimethoxybenzene	38.298	1323/1320	Stale	n.d.	n.d.	n.d.	1.53 ± 0.10b	n.d.	1.39 ± 0.16b	1.37 ± 0.17b	n.d.	4.06 ± 0.60a	0.72 ± 0.09c	n.d.	4.09 ± 0.42a	0.52 ± 0.05c
V43	3,4-Dimethoxy styrene	39.924	1367/1369	Savory	n.d.	n.d.	n.d.	n.d.	n.d.	0.47 ± 0.03c	n.d.	n.d.	3.83 ± 0.90a	n.d.	n.d.	2.07 ± 0.12b	n.d.
V44	p-tert-Butylphenetole	41.329	1405/1418	/	0.58 ± 0.03bc	0.34 ± 0.02d	0.48 ± 0.01c	0.56 ± 0.03bc	n.d.	0.49 ± 0.02c	0.48 ± 0.08c	0.34 ± 0.03d	0.54 ± 0.11bc	0.61 ± 0.09b	0.49 ± 0.02c	0.76 ± 0.09a	n.d.
	Hydrocarbons
V45	p-Cymene	17.389	1017/1025	Citrus, Fresh, Solvent	n.d.	0.56 ± 0.01d	0.53 ± 0.06d	0.51 ± 0.04d	n.d.	n.d.	n.d.	0.49 ± 0.06d	2.76 ± 0.29a	1.09 ± 0.13b	0.75 ± 0.10c	1.16 ± 0.22b	n.d.
V46	(R)-Isocarvestrene	17.88	1024/1027	/	n.d.	0.81 ± 0.14b	0.74 ± 0.20b	0.54 ± 0.02c	1.35 ± 0.03a	n.d.	0.54 ± 0.01c	0.80 ± 0.14b	n.d.	1.48 ± 0.08a	0.78 ± 0.07b	n.d.	n.d.
	Ketones
V47	2-Heptanone	9.982	889/891	Blue Cheese, Fruity, Green, Nut, Spice	1.28 ± 0.07c	5.97 ± 0.75a	3.22 ± 0.07b	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V48	Sulcatone	15.393	985/986	Citrus, Green, Mushroom, Pepper, Strawberry	5.51 ± 0.15cd	4.50 ± 0.06e	5.36 ± 0.27cd	6.17 ± 0.74c	2.16 ± 0.20g	5.18 ± 0.90de	3.30 ± 0.12f	1.03 ± 0.13h	17.18 ± 0.41b	3.19 ± 0.40f	1.80 ± 0.28gh	26.02 ± 0.94a	n.d.
V49	Acetophenone	20.299	1059/1065	Almonds, Flower, Hawthorn, Mimosa	0.65 ± 0.01defg	0.50 ± 0.06g	0.59 ± 0.09fg	1.07 ± 0.08b	0.89 ± 0.23bc	0.76 ± 0.09cdef	0.85 ± 0.08cd	0.62 ± 0.04efg	0.83 ± 0.06cde	0.71 ± 0.08cdefg	0.54 ± 0.02fg	1.80 ± 0.26a	0.67 ± 0.08cdefg
V50	Isophorone	24.328	1114/1124	Cedarwood, Spice, Peppermint, Camphor	n.d.	n.d.	n.d.	0.44 ± 0.06a	0.37 ± 0.04b	n.d.	0.19 ± 0.01c	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V51	(E)-β-Damascenone	40.494	1382/1386	Apple, Rose, Honey, Tobacco	0.94 ± 0.01b	0.67 ± 0.04c	1.08 ± 0.03a	1.08 ± 0.15a	0.40 ± 0.06de	0.41 ± 0.04de	0.53 ± 0.02d	0.22 ± 0.03f	0.49 ± 0.11d	0.32 ± 0.15ef	n.d.	n.d.	n.d.
V52	α-Ionone	41.912	1424/1426	Woody, Violet	0.82 ± 0.05a	0.35 ± 0.02d	0.65 ± 0.02b	0.83 ± 0.04a	n.d.	0.44 ± 0.09c	0.39 ± 0.06cd	0.17 ± 0.02e	0.40 ± 0.06cd	n.d.	0.23 ± 0.05e	n.d.	n.d.
V53	3,4-Dehydro-β-ionone	42.04	1428/1423	Violet, Floral	1.76 ± 0.09b	1.45 ± 0.08cd	2.02 ± 0.04a	1.53 ± 0.07c	1.35 ± 0.05de	1.27 ± 0.18e	0.83 ± 0.07g	0.86 ± 0.08g	1.06 ± 0.12f	0.35 ± 0.16h	0.86 ± 0.12g	0.84 ± 0.04g	n.d.
V54	Dihydro-β-ionone	42.259	1435/1433	Woody, Mahogany, Orris, Dry amber	n.d.	n.d.	0.20 ± 0.02c	0.33 ± 0.01a	n.d.	n.d.	0.26 ± 0.05b	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.
V55	Geranylacetone	42.688	1449/1453	Fresh, Green, Fruity, Rose, Woody, Magnolia	0.36 ± 0.03bc	0.50 ± 0.13b	0.44 ± 0.03b	0.27 ± 0.01c	0.50 ± 0.05b	0.52 ± 0.17b	0.39 ± 0.05bc	0.39 ± 0.07bc	n.d.	0.49 ± 0.06b	0.38 ± 0.07bc	0.71 ± 0.17a	n.d.
V56	2,6-Di-tert-butyl-p-benzoquinone	43.092	1462/1471	/	0.48 ± 0.01cd	0.72 ± 0.30b	1.22 ± 0.04a	0.42 ± 0.02cd	0.69 ± 0.13b	0.46 ± 0.03cd	0.34 ± 0.03de	0.60 ± 0.11bc	0.47 ± 0.06cd	0.23 ± 0.03e	0.51 ± 0.05	n.d.	n.d.
V57	2,5-Di-tert-butyl-1,4-benzoquinone	43.257	1468/1466	/	0.56 ± 0.01cde	0.49 ± 0.01e	0.54 ± 0.02de	0.71 ± 0.06bc	n.d.	0.53 ± 0.08de	0.98 ± 0.05a	0.31 ± 0.05f	0.52 ± 0.12de	0.93 ± 0.14a	0.54 ± 0.14de	0.76 ± 0.12b	0.67 ± 0.14bcd
V58	(E)-β-Ionone	43.698	1482/1486	Violet, Raspberry, Floral, Fruity, Woody	2.63 ± 0.03a	1.38 ± 0.06d	2.27 ± 0.10b	1.61 ± 0.13c	0.98 ± 0.11e	1.46 ± 0.12cd	0.78 ± 0.06f	1.02 ± 0.11e	0.59 ± 0.09g	0.59 ± 0.07g	0.92 ± 0.15ef	0.52 ± 0.05g	n.d.
	Acids
V59	Acetic acid	3.206	612/610	Acid, Fruity, Pungent, Sour, Vinegar	n.d.	3.63 ± 0.87a	2.61 ± 0.98b	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	1.26 ± 0.61c	n.d.	n.d.
V60	2-Methylbutyric acid	9.036	866/861	Pungent, Acid, Roquefort, Cheese, Fermented	n.d.	5.61 ± 0.34a	2.90 ± 0.35b	n.d.	0.84 ± 0.04e	n.d.	n.d.	1.25 ± 0.18d	n.d.	n.d.	1.73 ± 0.13c	n.d.	n.d.
V61	3-Methylbutanoic acid	9.84	885/863	Cheese, Pungent, Acid	n.d.	30.00 ± 2.86b	33.84 ± 5.55a	n.d.	5.01 ± 0.39d	3.23 ± 0.45de	n.d.	5.67 ± 1.43d	n.d.	n.d.	17.04 ± 0.09c	n.d.	n.d.
V62	2-Amino-4-methylbenzoic acid	10.822	906/949	/	n.d.	n.d.	n.d.	n.d.	9.80 ± 0.94b	8.34 ± 0.26cd	n.d.	3.94 ± 1.36e	7.18 ± 1.14d	n.d.	8.93 ± 0.99bc	13.52 ± 1.27a	7.98 ± 1.48cd
	Phenols
V63	4-Ethyl-2-methoxyphenol	36.198	1278/1282	Clove, Phenol, Spice	0.23 ± 0.01d	0.24 ± 0.02d	0.44 ± 0.01b	n.d.	0.69 ± 0.07a	0.33 ± 0.05c	n.d.	0.39 ± 0.02b	n.d.	n.d.	0.44 ± 0.06b	n.d.	n.d.
V64	2,4-Ditert-butylphenol	44.505	1510/1519	/	2.35 ± 0.15a	n.d.	1.40 ± 0.06b	2.23 ± 0.27a	n.d.	1.07 ± 0.20c	1.44 ± 0.18b	n.d.	1.01 ± 0.26c	1.51 ± 0.15b	n.d.	0.95 ± 0.10c	0.82 ± 0.17c
	Others
V65	1,2-Benzisothiazole	32.719	1217/1221	/	2.48 ± 0.04a	1.19 ± 0.10fg	1.62 ± 0.07de	2.06 ± 0.07bc	0.86 ± 0.12g	1.32 ± 0.21ef	1.78 ± 0.05cd	0.85 ± 0.18g	1.61 ± 0.06de	2.31 ± 0.37ab	1.55 ± 0.42def	2.35 ± 0.22ab	2.38 ± 0.28ab

Data within each row marked with different letters show significant differences at p < 0.05. n.d., not detected. ^A Retention index of compounds on HP-5MS. ^B Retention index of compounds in references. Odor descriptions were from FEMA database.

Fig. 2.

The number (A) and content (B) of VOCs of dark tea infusions during fermentation. The classification of all identified VOCs in dark tea infusions (C). Principal component analysis (D), hierarchical cluster analysis (E), heat-map analysis (F) of VOCs in dark tea infusions during fermentation.

Alcohols: During late fermentation (days 6–8), alcohols became the dominant VOCs in the JK group (Fig. 2B). Notably, linalool (floral aroma) reached 37.70 μg/L in JK8—2- and 3-fold higher than in groups J8 and K8, respectively (Table 1). Similarly, (E)- and (Z)-linalool oxides (floral and woody notes) peaked at 11.63 and 12.70 μg/L in JK8, significantly surpassing monoculture groups. These compounds derive from β-primeveroside and β-glucopyranoside precursors' hydrolysis by microbial enzymes (Nie et al., 2019). E. cristatum secretes primeverosidases/glucosidases (Xiao, Huang, et al., 2022; Zheng et al., 2016), while kefir-associated lactic acid bacteria produce additional glucosidases (Hu et al., 2022), synergistically enhancing linalool and its oxides production in co-fermentation. 1-Octen-3-ol (floral, mushroom and fat aromas) accumulated markedly in E. cristatum-fermented groups (J/JK), likely via lipoxygenase/hydroperoxide lyase/alcohol dehydrogenase-mediated oxidation of linoleic acid (Chen et al., 2022; Ho et al., 2015). Conversely, kefir microbes degraded 1-octen-3-ol, explaining its absence in the K group (Table 1).

Esters: It was found that ethyl caprylate, having floral and brandy aromas, reached the highest concentration of 30.97 μg/L and 35.69 μg/L on the 2nd day of fermentation in JK group and K group, respectively (Table 1). Methyl salicylate (wintergreen/mint aroma), a key FBT volatile (Xiao, Huang, et al., 2022), surged in JK8 to 26.92 μg/L—1.9- and 7.4-fold higher than J8 and K8, respectively. This compound arises from glycoside precursors hydrolysis by β-glucosidase (Nie et al., 2019). Previous studies reported that E. cristatum, LAB and yeast could secret large amounts of extracellular enzymes including β-glucosidase during fermentation (An et al., 2023; Huang et al., 2025; Hu et al., 2022; Xiao, Huang, et al., 2022). Co-fermentation (JK) likely enhanced methyl salicylate production via synergistic β-glucosidase secretion by LAB, yeast, and E. cristatum, surpassing monoculture systems. Similarly, Wang, Sun, et al. (2022a) also reported that co-fermenting green tea with Lactiplantibacillus plantarum and Saccharomyces boulardii promoted the formation of flavors due to synergistic enzymatic activities.

Aldehydes: Low levels of aldehydes are associated with specific aroma characteristics, for instance, nonanal at low concentrations imparts green grass and fatty notes, and low concentration of benzaldehyde contributes pleasant notes of nut and almond. Aldehydes concentrations above a specific threshold cause unpleasant and irritating odors such as rancidity. Hence, the reduction of aldehydes prevents the generation of this off-odor. The levels of benzaldehyde, nonanal, safranal and decanal were highest in unfermented dark tea infusion (Table 1), but their levels in all fermentation groups (K, JK and J) dropped significantly as fermentation progressed. Sukharev et al. (2019) revealed that microorganisms could catalyze the conversion of aldehydes to acids and alcohols.

Ketones: Twelve ketones were identified in dark tea infusions, originating from glycoside precursors or carotenoid oxidation (Ho et al., 2015). Notably, sulcatone (citrus/strawberry aromas) peaked at 26.02 μg/L in JK8—13-fold higher than K8 and undetectable in J8. Similarly, acetophenone (almond/floral notes), a key FBT volatile (Xiao, Huang, et al., 2022), reached 1.80 μg/L in JK8, significantly surpassing monocultures (Table 1, Fig. 2F). Acetophenone biosynthesis occurs via the shikimate pathway: phenylalanine is transformed to trans-cinnamic acid by L-phenylalanine ammonia lyase (PAL), then sequentially hydroxylated and degraded (Zhou et al., 2018; Zubkov & Kouznetsov, 2023). Co-fermentation might be beneficial for microorganisms to secrete large amounts of PAL, synergistically promoting acetophenone accumulation.

Ethers: As for the ethers, 4-ethyl-1,2-dimethoxybenzene showed great higher in the co-fermentation group compared with J group, and this VOC was not detected in K groups during the entire process. 3,4-Dimethoxystyrene was only detected in JK groups during the fermentation period of 4–8 days (Table 1, Fig. 2F). Methoxyphenolic compounds, known for their mellow and stale aromas, were commonly found in dark tea (Cao et al., 2018). These compounds were thought to be derived from the methylation of tannins, gallic acid, and catechins by microorganisms (Wang, Li, et al., 2022). Research has shown that catechins (such as epigallocatechin gallate and epicatechin gallate) can be hydrolyzed to form gallic acid (GA) under the action of tannase and esterase (Li, Wei, et al., 2024; Wang, Sun, et al., 2022a). Then, under the action of methyltransferase produced by microorganisms, GA could transform to methoxyphenolic compounds and their derivatives (Li, Wei, et al., 2024; Wang, Li, et al., 2022).

Acids: Volatile acids (3-methylbutanoic, 2-methylbutyric, and acetic acids) contribute off-odors in fermented beverages through AAB metabolism. In our study, 3-methylbutanoic acid (pungent/rancid aroma) persisted throughout kefir (K) fermentation but was eliminated in co-fermentation (JK) by day 6 (Table 1). This acid forms via the Ehrlich pathway through microbial degradation of L-leucine, with 2-keto-4-methylvaleric acid as a key intermediate mediated by kefir-derived transaminases (Duensing et al., 2024; Kieronczyk et al., 2003). We propose E. cristatum's esterases may convert 3-methylbutanoic acid to ethyl isovalerate in JK, supported by known microbial esterase activities (Bornscheuer, 2002). Acetic and 2-methylbutyric acids (pungent/sour notes) appeared in both K2 and JK2 but at reduced levels in JK2, disappearing after day 4 (Table 1, Fig. 2F). E. cristatum likely prevented their accumulation through two mechanisms: (1) dehydrogenases converting 2-methylbutyric acid to 2-methylbutanol (Kun et al., 2022), and (2) microbial transformation of acetic acid to lactic acid (Jayabalan et al., 2007). These metabolic conversions explain the superior aroma profile of co-fermented tea, as the marked reduction of these off-odor acids significantly enhanced the beverage's sensory quality.

3.3. Key aroma-active compounds analysis during dark tea infusion fermentation

Aroma impact was assessed by rOAV, where VOCs with rOAV ≥1 are key aroma contributors and 0.1 ≤ rOAV <1 are important aroma modifiers (Huang et al., 2025). Table 2 shows that 21 VOCs with rOAV ≥ 0.1 out of the 65 VOCs identified in this study. The high rOAV of linalool (rOAV 44.618–171.373), known for its sweet, floral, fruity, and rose odors, were consistently observed in all three groups throughout the fermentation process, significantly impacting the aroma of the fermented tea infusion. (E)-β-ionone having a floral and violet scent is the carotenoid degradation derivative of β-carotene (Ho et al., 2015; Huang et al., 2023). In this study, (E)-β-ionone exhibited high rOAV (greater than 70) in all tea infusions except for the J8 group (Table 2), playing a significant role in shaping the aroma profile of fermented dark tea infusion. More importantly, as shown in Table 2, the co-fermentation group JK8 exhibited the highest rOAV for linalool, (E)-linalool oxide and (Z)-linalool oxide among all samples, which playing a role in imparting floral and sweet odors to dark tea infusion. The rOAV of sulcatone (rOAV 0.383) and methyl salicylate (0.673) were also the highest in the co-fermentation group of JK8. Sulcatone and methyl salicylate were also reported as key VOCs contributed to the formation of ‘fungal floral’ aroma in FBT. The synergistic effect by co-fermentation of kefir with E. cristatum led to the higher formation of key VOCs during fermentation (especially in the later fermentation stage), contributing to improved aroma with mint, floral notes, and fruity notes in the fermented dark tea infusion. (E)-β-damascenone (apple-like) displayed exceptional rOAVs (>100) until day 6, with its ultra-low threshold (0.002 ppb) making it greatly influence the aroma characteristics despite lower concentration. These findings demonstrate that co-fermentation improves production of critical aroma-active compounds, resulting in a more balanced and appealing aromatic profile compared to monoculture fermentations.

Table 2.

The rOAV values of VOCs during dark tea infusions fermentation.

NO.	Compounds	Threshold value (μg/kg)	Relative odor activity values (rOAV)
			0 day		2 days				4 days				6 days				8 days
			C0		K2	JK2	J2		K4	JK4	J4		K6	JK6	J6		K8	JK8	J8
V1	Trans-3-hexenol	110	n.d.		n.d.	n.d.	0.013		n.d.	n.d.	0.032		n.d.	n.d.	0.016		n.d.	n.d.	n.d.
V2	Heptanol	5.4	0.067		n.d.	n.d.	n.d.		n.d.	n.d.	0.25		n.d.	n.d.	2.42		n.d.	n.d.	0.494
V3	1-Octen-3-ol	1.5	0.822		n.d.	n.d.	0.382		n.d.	n.d.	0.555		n.d.	n.d.	6.757		n.d.	0.713	30.371
V4	2-Ethylhexanol	1280	0.004		0.001	0.001	0.002		0.001	0.001	0.002		0.001	0.004	0.002		n.d.	0.003	0.002
V5	(Z)-Linalool oxide	100	0.025		0.023	0.009	0.006		0.064	0.017	0.007		0.03	0.067	0.01		0.011	0.127	0.008
V6	(E)-Linalool oxide	60	0.017		0.021	0.022	0.013		0.081	0.022	0.02		0.052	0.117	0.038		n.d.	0.194	n.d.
V7	Linalool	0.22	81.973		77.032	87.691	98.895		82.745	65.105	77.995		44.618	105.491	60.086		59.605	171.373	84.95
V8	2-Phenylethanol	564.23	0.001		0.05	0.003	0.002		n.d.	n.d.	0.003		n.d.	n.d.	0.002		n.d.	n.d.	n.d.
V9	(−)-trans-Chrysanthenol	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V10	4-Terpineol	340	0.003		0.001	0.001	0.002		0.002	n.d.	0.001		0.001	0.002	0.002		0.001	0.004	0.003
V11	α-Terpineol	1200	0.002		0.002	0.002	0.002		0.002	0.002	0.003		0.001	0.004	0.002		0.001	0.006	0.001
V12	Nerol	680	n.d.		0.003	n.d.	n.d.		n.d.	n.d.	n.d.		n.d.	0.016	n.d.		n.d.	0.003	n.d.
V13	Geraniol	6.6	0.119		0.48	0.07	n.d.		n.d.	n.d.	n.d.		n.d.	0.068	n.d.		n.d.	0.085	n.d.
V14	Nerolidol	10	0.025		0.038	0.065	0.02		n.d.	0.028	0.023		0.027	n.d.	n.d.		0.06	n.d.	n.d.
V15	Cedrol	0.5	1.108		0.762	1.09	1.228		0.62	1.048	0.886		0.77	0.876	0.856		0.61	0.866	n.d.
V16	Benzaldehyde	750.89	0.013		0.003	0.003	0.002		0.003	0.003	0.002		0.002	0.002	0.002		0.002	0.001	n.d.
V17	1-Ethyl-2-pyrrolecarboxaldehyde	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V18	Nonanal	1.1	2.121		1.437	1.304	1.296		1.055	1.218	1.241		0.993	1.053	1.555		0.97	1.537	1.39
V19	Safranal	3	0.199		n.d.	n.d.	0.062		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.
V20	Decanal	3	1.066		0.323	0.333	0.734		0.213	0.289	0.252		0.189	0.315	0.697		0.192	0.512	0.287
V21	2,5-Dimethylbenzaldehyde	200	0.017		0.009	0.005	0.024		0.003	0.002	0.009		0.002	0.003	0.016		0.002	0.003	0.005
V22	β-Homocyclocitral	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V23	Citral	28	0.045		0.017	n.d.	0.008		n.d.	n.d.	0.01		n.d.	0.022	0.034		n.d.	n.d.	0.021
V24	Methyl benzoate	73	0.006		n.d.	n.d.	0.007		n.d.	n.d.	0.02		n.d.	n.d.	0.023		n.d.	n.d.	n.d.
V25	Ethyl benzoate	55.56	n.d.		0.015	0.013	0.014		0.012	0.011	0.029		0.007	0.011	0.048		0.009	0.014	0.016
V26	Methyl salicylate	40	n.d.		0.016	n.d.	n.d.		0.209	0.04	0.01		0.103	0.47	0.06		0.091	0.673	0.355
V27	Ethyl caprylate	19.3	0.034		1.849	1.605	0.022		n.d.	n.d.	0.012		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.
V28	Phenethyl acetate	249.59	n.d.		n.d.	n.d.	0.004		n.d.	n.d.	0.007		n.d.	n.d.	0.011		n.d.	n.d.	0.006
V29	Neryl formate	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V30	Ethyl nonanoate	377	n.d.		0.006	0.007	n.d.		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.
V31	γ-Nonanolactone	9.7	n.d.		0.054	0.049	n.d.		0.175	0.135	n.d.		0.16	n.d.	n.d.		0.145	n.d.	n.d.
V32	Ethyl 9-decenoate	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V33	Dihydroactinidiolide	500	0.002		0.002	0.003	0.002		0.002	0.002	0.001		0.002	0.002	0.001		0.001	0.002	n.d.
V34	2,2,4-Trimethyl-1,3-pentanediol diisobutyrate	14	0.025		0.063	0.024	0.017		0.047	0.025	0.044		0.032	0.04	0.044		n.d.	0.064	0.05
V35	1,2-Dimethoxybenzene	3.17	0.136		n.d.	n.d.	0.152		n.d.	n.d.	0.363		n.d.	n.d.	0.415		n.d.	n.d.	0.426
V36	4-Methoxystyrene	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V37	2,3-Dihydrobenzofuran	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V38	2,4-Dimethoxytoluene	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V39	Theaspirane	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V40	Edulane	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V41	1,2,3-Trimethoxybenzene	0.75	n.d.		n.d.	n.d.	n.d.		n.d.	n.d.	0.363		n.d.	n.d.	0.459		n.d.	n.d.	n.d.
V42	4-Ethyl-1,2-dimethoxybenzene	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V43	3,4-Dimethoxy styrene	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V44	p-tert-Butylphenetole	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V45	p-Cymene	5.01	n.d.		0.112	0.106	0.102		n.d.	n.d.	n.d.		0.098	0.551	0.217		0.15	0.231	n.d.
V46	(R)-isocarvestrene	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V47	2-Heptanone	140	0.009		0.043	0.023	n.d.		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.
V48	Sulcatone	68	0.081		0.066	0.079	0.091		0.032	0.076	0.048		0.015	0.253	0.047		0.026	0.383	n.d.
V49	Acetophenone	65	0.01		0.008	0.009	0.016		0.014	0.012	0.013		0.01	0.013	0.011		0.008	0.028	0.01
V50	Isophorone	11,000	n.d.		n.d.	n.d.	0		0	n.d.	0		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.
V51	(E)-β-Damascenone	0.002	470.333		334.5	539	540		200	203	263		108	244.5	161		n.d.	n.d.	n.d.
V52	α-Ionone	3.78	0.216		0.093	0.173	0.219		n.d.	0.116	0.103		0.045	0.105	n.d.		0.06	n.d.	n.d.
V53	3,4-Dehydro-β-ionone	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V54	Dihydro-β-ionone	1	n.d.		n.d.	0.196	0.334		n.d.	n.d.	0.255		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.
V55	Geranylacetone	60	0.006		0.008	0.007	0.004		0.008	0.009	0.006		0.007	n.d.	0.008		0.006	0.012	n.d.
V56	2,6-Di-tert-butyl-p-benzoquinone	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V57	2,5-Di-tert-butyl-1,4-benzoquinone	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V58	(E)-β-Ionone	0.007	376.143		196.429	324.714	230.286		139.429	208.429	111.143		145.143	84.857	84.143		131.857	74.857	n.d.
V59	Acetic acid	99,000	n.d.		0	n.d.	n.d.		n.d.	n.d.	n.d.		n.d.	n.d.	n.d.		0	n.d.	n.d.
V60	2-Methylbutyric acid	2200	n.d.		0.003	0.001	n.d.		0	n.d.	n.d.		0.001	n.d.	n.d.		0.001	n.d.	n.d.
V61	3-Methylbutanoic acid	490	n.d.		0.061	0.069	n.d.		0.01	0.007	n.d.		0.012	n.d.	n.d.		0.035	n.d.	n.d.
V62	2-Amino-4-methylbenzoic acid	–	–		–	–	–		–	–	–		–	–	–		–	–	–
V63	4-Ethyl-2-methoxyphenol	89.25	0.003		0.003	0.005	n.d.		0.008	0.004	n.d.		0.004	n.d.	n.d.		0.005	n.d.	n.d.
V64	2,4-Ditert-butylphenol	500	0.005		n.d.	0.003	0.004		n.d.	0.002	0.003		n.d.	0.002	0.003		n.d.	0.002	0.002
V65	1,2-Benzisothiazole	–	–		–	–	–		–	–	–		–	–	–		–	–	–

n.d., not detected.-, threshold value of the volatile is not found in the literatures.

3.4. Correlation analysis between the change of VOCs and microbial community succession

The evolution of volatile organic compounds (VOCs) during dark tea infusion fermentation exhibited a strong association with microbial community succession. To elucidate microbial contributions to VOC formation, Spearman's correlation analysis was performed between the top 20 microbial genera and all detected VOCs (Fig. 3A-B). Significant positive correlations (p < 0.05) were observed between Acetobacter and several key VOCs, including (E)-linalool oxide, (Z)-linalool oxide, methyl salicylate, acetophenone, 4-ethyl-1,2-dimethoxybenzene, and 3,4-dimethoxy styrene (Fig. 3A). This aligns with Acetobacter's known enzymatic capacity, as it expresses oxidase systems (e.g., alcohol dehydrogenase, aldehyde dehydrogenase, and catalase) that facilitate flavor precursor and volatile metabolites production (Gomes et al., 2018; Wu et al., 2018). Notably, Acetobacter may enhance methyl salicylate and linalool oxide levels via β-glucosidase-mediated hydrolysis of glycoside precursors (Tajima et al., 2001; Zhang et al., 2020; Wang, Sun, et al., 2022a). Other bacterial genera, including Gluconobacter, Liquorilactobacillus, and Lacticaseibacillus, also showed significant VOCs correlations (p < 0.05). Gluconobacter exhibited positive associations with 2,3-dihydrobenzofuran, ethyl benzoate, and ethyl nonanoate but negative correlations with 4-terpineol, 2-ethylhexanol, 1-octen-3-ol, and heptanol (Fig. 3A). This genus's membrane-bound dehydrogenases partially oxidize alcohols to ketones and acids (Kataoka, 2024), and its decline after 2 days of fermentation in the K and JK groups likely reduced alcohol oxidation. Meanwhile, lactic acid bacteria (LAB; e.g., Liquorilactobacillus and Lacticaseibacillus) contribute to VOCs formation through enzymatic activities (e.g., β-glucosidase, fatty acid dehydrogenase, and hydrolases) (Gomes et al., 2018; Parapouli et al., 2020; Stribny et al., 2016).

Fig. 3.

Correlation between microorganisms and VOCs during fermentation of dark tea infusions. Heat-map analysis of the correlation relationship between VOCs and top 20 bacteria at the genus level during fermentation of dark tea infusions (A). Heat-map analysis of the correlation relationship between VOCs and top 20 fungi at the genus level during fermentation of dark tea infusions (B). *p < 0.05, **p < 0.01. C0 is the non-fermented dark tea infusion.

At the fungal genus level, Eurotium (syn. E. cristatum) demonstrated significant positive correlations (p < 0.05) with linalool, 1-octen-3-ol, 4-ethyl-1,2-dimethoxybenzene, 3,4-dimethoxy styrene, cedrol, and 4-terpineol (Fig. 3B). This is consistent with its reported secretion of primeverosidase and glucosidase, which hydrolyze β-glucopyranoside and β-primeveroside precursors to release linalool derivatives (Hu et al., 2023; Nie et al., 2019; Zheng et al., 2016). Saccharomyces was positively associated with 2-methylbutyric acid and 3-methylbutanoic acid (Fig. 3B), likely via L-leucine degradation through the Ehrlich pathway (Babcock et al., 2017; Duensing et al., 2024). Co-fermentation with E. cristatum reduced Saccharomyces abundance, potentially suppressing these undesirable volatile acids. These findings align with study of Liang et al. (2025), who also reported that co-fermentation suppressed off-flavors production. Additionally, Melampsora correlated positively with terpenoids (e.g., linalool, α-terpineol, geraniol) and the sesquiterpene cedrol (Fig. 3B). This may be attributed to its terpene synthase (TPS) and trans-isoprenyl diphosphate synthase (IDS) enzymes, which drive terpenoid biosynthesis (Wei et al., 2020).

Collectively, Acetobacter, Gluconobacter, LAB, Eurotium, Saccharomyces and Melampsora were core functional microbes driving VOCs formation through enzymatic transformations. Co-fermentation with E. cristatum modulated microbial interactions, enhancing desirable floral/minty notes while suppressing off-flavors, underscoring its potential for improving dark tea infusion aroma characteristic.

3.5. Change of organic acids content during fermentation

Organic acids are flavor precursors and played a pivotal role in VOCs formation and sensory characteristic. Succinic, lactic, malic, and acetic acids dominated the profiles (Table 3). Co-fermentation (JK) and kefir-only (K) groups exhibited rising succinic acid levels, peaking at 1823.23 mg/L (JK) and 2834.02 mg/L (K) by day 8—3- and 5-fold higher than E. cristatum-only (J) group, respectively. Succinic acid correlated strongly with the abundance of Acetobacter and Lactococcus (Fig. S5A), and greatly positively associated with the alteration of (Z)-linalool oxide during fermentation (Fig. S5B). Acetic acid peaked transiently on day 2 in K and JK groups, likely due to early-stage AAB activity, but declined thereafter. Notably, lactic acid in the JK group peaked at 1961.48 mg/L by day 8, which is 20.5-fold higher than K group. This is due to the proliferation of LAB (Liquorilactobacillus, Lacticaseibacillus) facilitated by E. cristatum co-fermentation (Fig. S5A). Glucuronic acid, a functional metabolite linked to hepatoprotection and precursor for vitamin C and glucosamine synthesis (Nguyen et al., 2015), peaked at 150.03 mg/L in the JK group on day 6, showing positive correlations with LAB and Eurotium but negative associations with Saccharomyces. Conversely, citric and malic acid levels decreased in co-fermentation, likely due to their catabolism via the tricarboxylic acid cycle and malolactic pathways (Yin et al., 2015). These distinct organic acid profiles demonstrated how microbial community differences drove divergent metabolic pathways, ultimately shaping unique flavor profiles.

Table 3.

Changes in organic acids content during dark tea infusions fermentation.

Organic acids	Content (mg/L)
	0 day		2 days				4 days				6 days				8 days
	C0		K2	JK2	J2		K4	JK4	J4		K6	JK6	J6		K8	JK8	J8
Glucuronic acid	54.32 ± 5.60f		34.01 ± 0.19i	40.39 ± 1.33gh	70.00 ± 1.55c		36.34 ± 0.18hi	61.56 ± 0.28de	70.72 ± 2.13c		110.05 ± 3.02b	150.03 ± 2.45a	65.14 ± 1.78d		42.23 ± 0.28g	38.29 ± 0.62ghi	60.38 ± 1.00e
Oxalic acid	77.83 ± 2.07cde		78.12 ± 1.17cd	74.52 ± 0.78f	79.07 ± 2.38c		89.95 ± 0.23b	88.05 ± 0.63b	76.50 ± 0.49def		92.59 ± 0.38a	78.08 ± 0.33cd	69.98 ± 1.61g		92.58 ± 0.25a	75.58 ± 0.17ef	62.16 ± 0.21h
Tartaric acid	68.00 ± 2.64b		27.27 ± 4.51f	57.06 ± 0.36c	72.90 ± 2.74b		34.85 ± 0.80d	27.84 ± 3.43ef	84.45 ± 2.89a		33.54 ± 0.64de	37.38 ± 1.36d	86.93 ± 6.04a		32.26 ± 0.24def	37.16 ± 3.33d	86.57 ± 3.56a
Malic acid	850.67 ± 21.36ef		805.49 ± 1.33efg	865.47 ± 0.41e	1638.21 ± 14.57a		773.38 ± 2.26fg	1182.88 ± 127.97c	952.15 ± 38.3d		765.95 ± 0.53g	1282.11 ± 27.21b	737.29 ± 1.93g		759.71 ± 1.33g	392.39 ± 0.48h	141.64 ± 2.85i
Lactic acid	868.94 ± 60.38d		214.86 ± 0.84g	399.62 ± 3.34e	1237.40 ± 4.86b		282.87 ± 5.21g	520.65 ± 89.14f	918.64 ± 57.38cd		201.51 ± 7.44g	899.98 ± 33.58d	993.94 ± 32.73c		95.57 ± 1.01h	1961.48 ± 63.10a	930.66 ± 13.68cd
Acetic acid	261.62 ± 12.16d		828.77 ± 5.16b	965.66 ± 63.33a	400.03 ± 22.57c		133.46 ± 1.81f	168.98 ± 9.23ef	85.28 ± 0.70g		133.35 ± 5.69f	186.53 ± 17.06e	n.d.		129.40 ± 1.96f	143.78 ± 4.15f	n.d.
Citric acid	331.04 ± 19.12a		138.11 ± 0.77de	161.51 ± 0.48c	224.43 ± 21.18b		144.02 ± 2.35d	111.37 ± 6.41gh	125.65 ± 0.58efg		114.07 ± 1.11fg	74.98 ± 5.08i	126.48 ± 3.38efg		94.63 ± 1.42h	67.81 ± 1.42i	131.20 ± 1.90def
Succinic acid	1289.05 ± 120.84g		1134.94 ± 105.22h	1822.99 ± 78.46f	900.85 ± 5.66i		2440.53 ± 6.20d	2681.93 ± 21.3b	741.10 ± 5.25j		2555.35 ± 7.88c	2300.33 ± 17.54e	487.46 ± 12.54k		2834.02 ± 19.34a	1823.23 ± 6.68f	542.45 ± 1.29k
Fumaric acid	0.97 ± 0.02j		2.31 ± 0.04c	1.94 ± 0.06ef	2.61 ± 0.02b		2.02 ± 0.03de	1.87 ± 0.06f	2.08 ± 0.02d		1.66 ± 0.02g	2.53 ± 0.04b	1.22 ± 0.07h		1.91 ± 0.08ef	2.91 ± 0.01a	1.10 ± 0.11i

Data within each row marked with different letters show significant differences at p < 0.05.

3.6. Change of free amino acids (FAAs) content during fermentation

Amino acids serve as critical precursors for volatiles biosynthesis through microbial-mediated decarboxylation, deamination, and dehydrogenation pathways (Xiao et al., 2024; Zhang et al., 2023). Microbial fermentation significantly altered free amino acids (FAAs) profiles, with aspartic acid and glutamic acid decreasing approximately 97 % in JK and K groups after 2 days (Table 4). Tyrosine, theanine, and valine also markedly declined in JK and K groups, suggesting kefir-associated microbes preferentially utilized these nitrogen sources for their growth. Group J showed less pronounced FAAs reduction, indicating differential microbial utilization patterns. During fermentation, FAAs undergo multiple transformation pathways. They can be transformed into aldehydes via Strecker degradation or further metabolized by microbial activity into various volatile compounds, including esters, alcohols, and carbonyls (Ho et al., 2015). Of particular significance is phenylalanine (Phe), which serves as a key precursor for phenylpropanoid derivatives and volatile benzenoids. Notably, the concentration of 2-phenylethanol—a Phe-derived VOC imparting fruity, honey, rose, and wine-like aromas—showed significant elevation (p < 0.05) across all experimental groups after just 2 days of fermentation. Another important Phe-derived aromatic compound, methyl salicylate (Ho et al., 2015; Huang et al., 2025), demonstrated a similar progressive accumulation pattern throughout the fermentation process. Phenylalanine was mainly catalyzed by phenylalanine ammonia-lyase (PAL) for deamination to form methyl salicylate (Li et al., 2019). Network analysis further revealed methyl salicylate accumulation strongly correlated with phenylalanine utilization (r = 0.768, p < 0.05) (Fig. S5C). A significant positive correlation between the accumulation of methyl salicylate and the abundance of Acetobacter (r = 0.757, p < 0.05) was noted (Fig. 3A). It could be reasonably speculated that Acetobacter produced a large amount of PAL to decompose phenylalanine into derivative aromatic compounds. This indicated that Acetobacter have great potential for amino acid hydrolysis and the generation of VOCs. Thus, the catabolism of amino acids by microorganisms is a pivotal route for flavor development in fermented dark tea infusion.

Table 4.

Changes in free amino acids content during dark tea infusions fermentation.

FAAs		Content (mg/L)
		0 day		2 days				4 days				6 days				8 days
		C0		K2	JK2	J2		K4	JK4	J4		K6	JK6	J6		K8	JK8	J8
Aspartic acid	Asp	159.22 ± 3.00a		4.07 ± 0.05e	4.08 ± 0.02e	131.86 ± 5.91b		4.13 ± 0.03e	4.13 ± 0.01e	42.57 ± 1.50c		4.12 ± 0.01e	4.10 ± 0.01e	24.38 ± 1.36d		4.13 ± 0.01e	4.11 ± 0.01e	21.85 ± 1.58d
Glutamic acid	Glu	111.74 ± 3.38a		3.87 ± 0.46e	3.08 ± 0.04e	51.36 ± 6.81b		3.62 ± 0.12e	3.18 ± 0.11e	25.05 ± 2.21c		3.73 ± 0.03e	3.70 ± 0.09e	12.99 ± 1.13d		4.07 ± 0.10e	3.57 ± 0.03e	13.22 ± 1.11d
Serine	Ser	5.06 ± 0.03a		4.55 ± 0.08b	4.54 ± 0.07b	4.39 ± 0.02de		4.41 ± 0.01cd	4.30 ± 0.08ef	4.25 ± 0.03f		4.22 ± 0.02f	4.50 ± 0.09bc	4.27 ± 0.02f		4.26 ± 0.02f	4.24 ± 0.01f	4.29 ± 0.04ef
Alanine	Ala	4.34 ± 0.02a		n.d.	n.d.	1.12 ± 0.04b		n.d.	n.d.	0.99 ± 0.00c		n.d.	n.d.	0.56 ± 0.00d		n.d.	n.d.	0.24 ± 0.01e
Threonine	Thr	3.53 ± 0.01a		2.46 ± 0.00g	2.47 ± 0.02fg	3.12 ± 0.08b		2.46 ± 0.00fg	2.46 ± 0.00fg	3.06 ± 0.04c		2.46 ± 0.00fg	2.47 ± 0.01fg	2.49 ± 0.01fg		2.58 ± 0.00d	2.55 ± 0.00de	2.51 ± 0.00ef
Glycine	Gly	3.20 ± 0.02bc		3.04 ± 0.04e	3.06 ± 0.01e	3.29 ± 0.09a		3.06 ± 0.02e	3.05 ± 0.00e	3.25 ± 0.04ab		3.06 ± 0.01e	3.07 ± 0.01de	3.18 ± 0.02bc		3.06 ± 0.01e	3.06 ± 0.01e	3.14 ± 0.07cd
Proline	Pro	2.64 ± 0.03a		1.01 ± 0.00e	1.01 ± 0.00e	1.50 ± 0.04b		1.02 ± 0.00e	1.01 ± 0.00e	1.19 ± 0.00c		1.02 ± 0.00e	1.02 ± 0.00e	1.05 ± 0.00d		1.02 ± 0.00e	1.02 ± 0.00e	1.06 ± 0.00d
Tyrosine	Tyr	25.40 ± 0.24a		4.85 ± 0.07e	4.91 ± 0.01e	17.31 ± 0.40b		4.83 ± 0.02e	4.81 ± 0.01e	10.76 ± 0.23c		4.82 ± 0.01e	4.83 ± 0.01e	6.49 ± 0.07d		4.77 ± 0.02e	4.77 ± 0.00e	6.19 ± 0.12d
Lysine	Lys	8.84 ± 0.05a		8.04 ± 0.01b	8.04 ± 0.00b	8.07 ± 0.01b		8.04 ± 0.01b	8.04 ± 0.00b	8.05 ± 0.01b		8.04 ± 0.01b	8.04 ± 0.00b	8.04 ± 0.00b		8.04 ± 0.00b	8.04 ± 0.00b	8.05 ± 0.00b
Histidine	His	5.24 ± 0.07b		6.40 ± 1.00ab	7.26 ± 1.02a	3.99 ± 0.98c		6.32 ± 0.91ab	6.15 ± 0.37ab	5.23 ± 0.09b		3.63 ± 0.38c	3.91 ± 0.21c	6.84 ± 0.26a		5.35 ± 0.25b	6.35 ± 0.21ab	5.33 ± 0.25b
Phenylalanine	Phe	3.25 ± 0.05a		1.75 ± 0.02def	1.77 ± 0.01d	1.94 ± 0.01b		1.73 ± 0.00ef	1.73 ± 0.00f	1.83 ± 0.00c		1.73 ± 0.00f	1.73 ± 0.00f	1.76 ± 0.00d		1.73 ± 0.00f	1.73 ± 0.00f	1.76 ± 0.00de
Methionine	Met	3.07 ± 0.01c		3.07 ± 0.00c	3.07 ± 0.01c	3.21 ± 0.03a		3.11 ± 0.00b	3.08 ± 0.00c	3.11 ± 0.01b		3.07 ± 0.01c	3.08 ± 0.00c	3.07 ± 0.00c		3.07 ± 0.00c	3.07 ± 0.01c	3.06 ± 0.01c
Isoleucine	Ile	2.24 ± 0.03a		1.19 ± 0.02e	1.18 ± 0.00e	1.92 ± 0.06b		1.19 ± 0.01e	1.19 ± 0.01e	1.71 ± 0.01c		1.18 ± 0.00e	1.18 ± 0.00e	1.26 ± 0.02d		1.18 ± 0.00e	1.18 ± 0.00e	1.21 ± 0.01e
Leucine	Leu	1.30 ± 0.01a		0.54 ± 0.00d	0.54 ± 0.00d	0.97 ± 0.02b		0.55 ± 0.00d	0.55 ± 0.00d	0.96 ± 0.01b		0.55 ± 0.00d	0.55 ± 0.00d	0.59 ± 0.00c		0.55 ± 0.00d	0.55 ± 0.00d	0.59 ± 0.00c
Tryptophan	Typ	6.37 ± 0.01a		5.29 ± 0.01e	5.33 ± 0.02c	5.40 ± 0.01b		5.24 ± 0.00h	5.24 ± 0.00h	5.25 ± 0.00gh		5.25 ± 0.00gh	5.26 ± 0.00fg	5.31 ± 0.00cd		5.25 ± 0.00gh	5.27 ± 0.01f	5.31 ± 0.00de
γ-aminobutyric acid	GABA	1.32 ± 0.03a		0.22 ± 0.00de	0.22 ± 0.00e	0.87 ± 0.03b		0.23 ± 0.00de	0.22 ± 0.00de	0.83 ± 0.03c		0.21 ± 0.00e	0.22 ± 0.00de	0.24 ± 0.00de		0.22 ± 0.00de	0.22 ± 0.00de	0.25 ± 0.00d
Theanine	The	42.23 ± 0.39a		6.56 ± 0.07e	6.78 ± 0.15e	43.37 ± 0.92b		6.17 ± 0.00e	6.18 ± 0.01e	23.69 ± 1.18c		6.22 ± 0.02e	6.21 ± 0.01e	17.18 ± 0.22d		6.20 ± 0.01e	6.24 ± 0.03e	6.75 ± 0.03e
Cysteine	Cys	2.83 ± 0.00a		2.82 ± 0.00d	2.82 ± 0.00bc	2.83 ± 0.00b		2.82 ± 0.00d	2.82 ± 0.00d	2.82 ± 0.00cd		2.82 ± 0.00d	2.82 ± 0.00d	2.82 ± 0.00d		2.82 ± 0.00d	2.82 ± 0.00cd	2.82 ± 0.00d
Valine	Val	2.16 ± 0.02a		0.22 ± 0.00f	0.22 ± 0.00f	1.88 ± 0.10b		0.23 ± 0.00ef	0.22 ± 0.00f	1.63 ± 0.01c		0.22 ± 0.00ef	0.22 ± 0.00ef	0.30 ± 0.00d		0.22 ± 0.00ef	0.22 ± 0.00f	0.28 ± 0.00de
Total FAAs	TTA	393.97 ± 6.20a		59.96 ± 1.31f	60.36 ± 0.92f	288.40 ± 7.11b		59.16 ± 0.86f	58.35 ± 0.47f	146.24 ± 3.06c		56.37 ± 0.44f	56.89 ± 0.30f	102.83 ± 2.08d		58.51 ± 0.33f	59.01 ± 0.19f	87.89 ± 2.67e

Data within each row marked with different letters show significant differences at p < 0.05.

3.7. Alteration of catechins and gallic acid profiles during fermentation

Co-fermentation with water kefir and E. cristatum (JK group) significantly modulated catechins metabolism compared to monoculture fermentation (J group). Notably, the total esterified catechins content in group J2 decreased by 83.8 % compared to the unfermented control, whereas groups K2 and JK2 maintained substantially higher levels. In particular, JK2 showed significantly elevated concentrations of ECG, GCG, and EGCG—approximately 3.11-, 12.27-, and 5.99-fold higher than those in group J2, respectively (Fig. 4). The findings suggest that co-fermentation with kefir and E. cristatum better preserved esterified catechins compared to E. cristatum monoculture, which may be attributed to several factors. First, lactic acid bacteria (LAB) in kefir create an acidic environment, effectively preventing the chemical degradation of esterified catechins (Rodríguez et al., 2021; Tu et al., 2018). Second, certain LAB and yeast strains in water kefir exhibit antioxidant activity, potentially generating metabolites that neutralize free radicals and thus protect catechins from oxidative degradation (Chandra et al., 2020). This mechanism likely slowed the conversion of esterified catechins into gallic acid and non-ester catechins. Additionally, co-fermentation with kefir may have suppressed E. cristatum growth, consequently reducing its capacity to degrade esterified catechins (Tu et al., 2018). While degalloylated catechins and gallic acid (GA) increased in both groups, their levels were lower in JK than J, consistent with E. cristatum's role in hydrolyzing ester catechins into degalloylated catechins and GA via extracellular enzymes. Notably, co-fermentation maintained higher total catechins, as kefir microbiota appeared to modulate E. cristatum-mediated transformations of degalloylated catechins into simpler phenolic acids (e.g., phloroglucinol, salicylic acid) through demethylation, decarboxylation, and ring-fission (Sun et al., 2023; Xiao, He, et al., 2022). Since catechins contribute substantially to tea's health-promoting properties (Zhao et al., 2024; Gao et al., 2023), these findings highlight co-fermentation as a strategy to balance catechins retention and biotransformation, improving both sensory and health-promoting attributes of dark tea.

Fig. 4.

Dynamic alteration in catechins and gallic acid content of dark tea infusions during fermentation. Data marked with different letters show significant differences (p < 0.05) among K, JK and J groups at the same fermentation time.

3.8. Sensory evaluation

Based on VOCs analysis, 8-day fermented dark tea infusions exhibited superior aroma profiles compared to other fermentation periods. Thus, sensory comparisons were made between 8-day fermented and non-fermented infusions (Fig. 5). Fermented samples displayed darker coloration, with J8 (mono-fermented with E. cristatum) showing the most pronounced hue (Fig. 5A), accompanied by reduced brightness (L∗) and yellowness (b∗) (Fig. 5B). Aroma profiling revealed JK8 and J8 showed higher fungal flower and minty notes compared with K8, and JK8 showed the highest intensity of minty aroma (Fig. 5C). This is likely due to elevated levels of linalool, linalool oxides, methyl salicylate, sulcatone, and 4-ethyl-1,2-dimethoxybenzene. Taste attribute improved significantly in JK8 and J8, exhibiting stronger mellow and umami notes with reduced astringency and bitterness (Fig. 5D). By contrast, the kefir-only sample (K8) exhibited pronounced sourness, attributable to the elevated accumulation of acidic metabolites during mono-fermentation, which was perceived as unpalatable by panelists. Notably, co-fermentation with E. cristatum (JK8) mitigated this effect, yielding balanced taste characteristic. Overall, JK8 and J8 demonstrated desirable sensory acceptance, indicating that E. cristatum-mediated fermentation (mono- or co-culture) effectively enhances the sensory quality of fermented dark tea.

Fig. 5.

Sensory evaluation of the non-fermented dark tea infusion (C0) and 8th day fermented dark tea infusions (K8, JK8, J8). (A) Color of dark tea infusions, (B) color parameters, radar of sensory aroma (C) and taste attributes profile (D). Data marked with different letters show significant differences (p < 0.05).

4. Conclusion

This study demonstrates that water kefir co-fermentation with E. cristatum significantly reshapes microbial communities in dark tea infusion fermentation, increasing Liquorilactobacillus, Lacticaseibacillus, and Acetobacter while suppressing Gluconobacter and Saccharomyces. HS-SPME-GC–MS identified 65 VOCs, with co-fermentation enhancing desirable floral, minty, fruity, and woody aromas (e.g., linalool and its oxides, α-terpineol, methyl salicylate, and acetophenone) and reducing off-flavors like 3-methylbutanoic acid). Notably, 21 VOCs exhibited significant aroma contributions (rOAV ≥0.1), particularly linalool, (E)-β-ionone, and (E)-β-damascenone. Microbial-VOC correlation analysis revealed Acetobacter, Gluconobacter, Liquorilactobacillus, Lacticaseibacillus, Eurotium, Saccharomyce, and Melampsora are closely correlated with the formation of VOCs. Co-fermentation also modulated organic acids, notably increasing lactic acid (linked to LAB activity), and preserved higher levels of esterified/total catechins compared to E. cristatum monoculture. Free amino acids declined across all groups during fermentation, contributing to flavor development of fermented dark tea infusion. Sensory evaluation confirmed superior aroma complexity (enhanced minty/fungal floral notes) and balanced taste (reduced sourness/astringency) in co-fermented tea. These findings highlight the synergistic potential of water kefir and E. cristatum to produce probiotic-rich dark tea beverages with improved flavor profiles. Future research will investigate the synergistic fermentation of E. cristatum and water kefir to clarify the metabolic pathways of key VOCs responsible for improving dark tea's flavor characteristics. Additionally, pilot-scale fermentation trials will be performed to evaluate the industrial viability of this co-culture strategy in enhancing the sensory and quality characteristics of dark tea infusion.

Statement on compliance with ethical standards

All sensory assessments were performed in accordance with national research ethics guidelines following approval by Hunan Agricultural University's Institutional Review Board (Approval No. 202462).

CRediT authorship contribution statement

Yao Gao: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation. Rui Zhuo: Software, Methodology, Investigation. Hong Luo: Methodology, Investigation, Data curation. Chi-Tang Ho: Writing – review & editing. Yulian Chen: Writing – review & editing, Resources, Methodology, Investigation. Hui Zhou: Resources. Zhaoxia Qu: Resources. Huanyu Chen: Resources. Youjin Yi: Writing – review & editing, Resources. Yuanliang Wang: Writing – review & editing, Supervision, Resources. Xiaozhen Peng: Writing – review & editing, Visualization, Resources. Mingzhi Zhu: Resources, Methodology. Zhonghua Liu: Resources. Yu Xiao: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization.

Informed consent statement

All participants provided informed consent for sensory testing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.