Mixed-fermentation effects of co-fermentation with Hanseniaspora uvarum and Saccharomyces cerevisiae on the flavor quality of Ziyan tea wine Authors: Yajie Hua, Lechi Ling, Xue Xia, Yueling Zhao, Yongxian Chen, Ling Lin, Wei Xu Published: 2026 ABSTRACT AbstractZiyan tea wine is an anthocyanin-rich fermented beverage with distinctive colour and flavor. In this study, we compared single and mixed yeast fermentation. Co-fermentation with S. cerevisiae and Hanseniaspora uvarum at a ratio of 1:1 significantly increased the sensory score compared with S. cerevisiae monoculture. Enartis Ferm Aroma White and H. uvarum at a ratio of 1:2 also produced a high sensory score, but the difference from Enartis Ferm Aroma White monoculture wasn't significant. HS-SPME-GC–MS and spectrophotometric analyses showed that co-fermentation changed the volatile profile of Ziyan tea wine. Some co-fermentation treatments had higher levels of floral and fruity aroma compounds (e.g., nerol, isoamyl acetate) and lower levels of volatile acids (e.g., nonanoic acid, octanoic acid), leading to a more balanced fruity and floral aroma profile. Overall, co-fermentation changed the volatile composition and sensory properties of Ziyan tea wine and may provide a useful method for improving tea wine quality.Keywords: Ziyan tea wine, Co-fermentation, Volatile compounds, Sensory quality Introduction Tea wine is a low-alcohol beverage produced through the fermentation of tea and sugar sources, combining the chemical constituents of tea with the distinctive flavors of fermented alcoholic drinks. It possesses significant cultural and economic value and has garnered increasing attention in the field of food science in recent years. ‘Ziyan’ is a novel tea cultivar selected from high-altitude regions, where the unique ecological conditions contribute to its superior quality (Tang et al., 2026). Notably, this cultivar is rich in anthocyanins, with content as high as 2.73%. Due to the chemical property of anthocyanins to appear red in acidic conditions, tea wine fermented from ‘Ziyan’ leaves exhibits a unique purplish-red hue (Perez et al., 2022). However, studies on anthocyanin-rich tea wine produced from fresh ‘Ziyan’ tea leaves remain limited. Current fermentation of Ziyan tea wine mainly relies on traditional Saccharomyces cerevisiae (SC), often resulting in a relatively monotonous flavor profile. Yeast composition is a key factor determining the aroma and sensory quality of fermented beverages (Y. Li et al., 2025). Co-fermentation can leverage mixed-fermentation effects between yeasts to achieve a more complex flavor hierarchy and improved aroma profile. Consequently, co-fermentation involving non-Saccharomyces yeasts and Saccharomyces yeast has emerged as an effective strategy for aroma enhancement and is now widely applied. For instance, a previous study demonstrated that wines co-fermented by Torulaspora delbrueckii and S. cerevisiae contained higher concentrations of ethyl propionate, ethyl isobutyrate, ethyl dihydrocinnamate, and isoamyl acetate, endowing them with more intense fruity notes and greater aromatic complexity (Renault et al., 2015). Similarly, Zhang et al. found that co-fermentation with Pichia kluyveri HS-2-1 and S. cerevisiae NX11424 enhanced the fruity and floral aromas in ‘Cabernet Sauvignon’ wine (Zhang et al., 2020). Related studies confirm that yeast species, fermentation conditions, and microbial interactions play crucial roles in shaping the aroma, structure, and nutritional composition of beverages (Rollero et al., 2018). Although previous studies have advanced our understanding, the differential effects of single-strain versus co-fermentation in Ziyan tea wine have yet to be fully elucidated. To address this issue, the present study employed three yeast strains with representative characteristics: Enartis Ferm Aroma White (BAI), a versatile strain known for preserving varietal character while generating fermentation aromas; Hanseniaspora uvarum, an aroma-producing non-Saccharomyces yeast; and a conventional Saccharomyces cerevisiae (SC) strain, which ensures efficient alcoholic fermentation. Using ‘Ziyan’ tea, characterized by its high anthocyanin content, as the fermentation substrate, we first determined the optimal blending ratios for co-fermentation of Saccharomyces cerevisiae (SC or BAI) with H. uvarum based on sensory evaluation. The highest sensory scores were achieved at a 1:1 ratio of SC to H. uvarum (SC1;1) and at a 1:2 ratio of BAI to H. uvarum (BAI1;2). The volatile profiles of teas fermented with different yeast combinations were then comprehensively analyzed using headspace solid-phase microextraction coupled with gas chromatography–mass spectrometry (HS-SPME-GC–MS), revealing a significant enhancement in floral and fruity aromas of the co-fermentation tea wines. Furthermore, flavor metabolomics and principal component analysis were utilized to elucidate the differences in key biochemical and volatile components among the tea wines and to explore the basis for these differences. These findings demonstrate that co-fermentation with selected yeast strains effectively enhances the flavor complexity and balance of Ziyan tea wine, providing useful insights into flavor enhancement and offering a valuable reference for the development and utilization of ‘Ziyan’ and other similar specialty tea cultivars. Materials and methods Materials and reagents ‘Ziyan’ green tea was provided by Sichuan Yizhichun Tea Industry Co., Ltd. Fructose syrup was purchased from Servis Bio-Technology Co., Ltd. Two commercial Saccharomyces cerevisiae yeasts were used in this study: FermSc (Enartis, Italy) and Enartis Ferm Aroma White (Enartis, Italy). The specific strain codes were not disclosed by the manufacturers. The Hanseniaspora uvarum strain RL-B was supplied by our laboratory. A Genmini C18 HPLC column (Phenomenex, USA), an Agilent 7890B—7000D GC–MS/MS system (Agilent Technologies, USA), a DB-5MS capillary column (30 m × 0.25 mm × 0.25 μm; Agilent Technologies, USA), a 120 μm DVB/CAR/PDMS SPME fiber (Agilent Technologies, USA), and a CTC SPME autosampler system including the conditioning station and sample heating block (CTC Analytics AG, Switzerland). The hexane and reference standards possess chromatographic purity (Merck KGaA, Darmstadt, Germany). The remaining reagents, including Folin-Ciocalteu's phenol reagent, ninhydrin, and lead subacetate, were all commercially purchased and met domestic analytical grade standards. Experimental methods Preparation of tea wine ‘Ziyan’ green tea was extracted at a tea-to-water ratio of 1:50 by adding 9.6 g of tea leaves to 480 mL of purified water in a beaker. The extraction was carried out with 85 °C hot water for 10 min, assisted by ultrasonication to improve the extraction efficiency of tea components. After filtration to remove tea residues, the tea broth was collected for later use. Fructose syrup (21 mL per 100 mL of tea broth) was added as a carbon source based on the method of Xu et al. (Xu et al., 2023), and the mixture was thoroughly mixed. Saccharomyces yeast and Hanseniaspora uvarum were activated separately following the method of Cheng et al. with modifications (Cheng et al., 2020). For monoculture fermentation, Saccharomyces yeast was inoculated at 0.2 g/L, and allowed to undergo anaerobic fermentation. For co-culture fermentation, Saccharomyces yeast and H. uvarum were inoculated at cell ratios of 1:1, 1:2, and 2:1. All ratios are expressed as Saccharomyces yeast: H. uvarum. The total inoculum cell number was kept constant across all mixed fermentation treatments. The cell concentration of each yeast was adjusted according to the required inoculation ratio. All experimental containers were sterilized by autoclaving before use. The tea broth was pasteurized at 60 °C for 30 min before yeast inoculation. This treatment was used to reduce the native microbial load while limiting the thermal degradation of anthocyanins. However, microbial counts were not determined after pasteurization, and an uninoculated blank control was not included. The H. uvarum strain used in this study was selected and cultivated in our laboratory, and its cell count was confirmed to be equal to that of 0.1 g/L commercial S. cerevisiae. The total yeast inoculum was kept constant at 0.2 g/L for both single and co-fermentation treatments. Fermentation was carried out statically at 28 °C for 10d (240 h). Fermentation was carried out statically at 28 °C for 10 days (240 h) using perforated fermentation bottles. Aerobic fermentation was conducted during the first two days. The holes were then covered with tape, and fermentation continued under anaerobic conditions for the following eight days. Samples were collected on days 0, 2, 4, 6, 8, and 10. Sensory evaluation Sensory evaluation was conducted on the appearance, aroma, and taste of the Ziyan tea wine. The sensory scoring criteria were established by referencing the national standard GB/T 33405–2016 “Terminology of Baijiu Sensory Evaluation” (China National Light Industry Council, 2016), with slight modifications based on the specific characteristics of the Ziyan tea wine (Table 1). The sensory panel included 15 trained assessors, including 5 males and 10 females, from the Tea Science program at Sichuan Agricultural University. All assessors had received formal training and had experience in tea and wine sensory evaluation. Each sample was served at 20 °C, and 200 mL was placed in a white porcelain bowl. The evaluation was conducted in a dedicated sensory room under controlled lighting and temperature conditions. Each sample was coded with a random three-digit number and presented in a randomized order. Panelists evaluated appearance, aroma, and taste using a 100-point scoring system. Aroma intensity was also evaluated using a 5-point scale, where 1 indicated imperceptible and 5 indicated very strong. Panelists used water to cleanse their palates between samples. The evaluation was conducted in a separate session. Three independent fermentation samples were evaluated for each treatment. Each sample was evaluated three times by each of the 15 panelists. All panelist scores were included in the statistical analysis. The sensory data were analyzed using a randomized block ANOVA, with fermentation treatment as the main factor and panelist as the block. Tukey's test was used for multiple comparisons at P < 0.05. Table 1Sensory scoring criteria.AttributeDescriptionScoresAppearance (30)Pinkish-purple, clear and transparent, lustrous26–30Pale pink, clear, no inclusions21–25Slightly pink, slightly muddy, loss of light or oxidative discolouration0–20Fragrance (30)Harmonised wine and tea aroma with a strong smell26–30The aroma of wine is light, slightly tea-like, but no unusual aroma21–25Thin aroma of wine, no tea aroma, bad odor16–20Unscented, undesirable odours prominent0–15Taste(40)Full-bodied, mellow and harmonious, with moderate tea flavors36–40Smooth, softer, lighter tea flavors31–35Harmonised, pure and unadulterated, with a slightly bitter taste26–30Impure taste, no tea flavor, bitter, not strong enough21–25Sour, astringent, bitter, with an off-flavor0–20 Sensory scoring criteria. This sensory evaluation study was approved by the Academic Ethical and Welfare Committee of Sichuan Agricultural University (Approval No. H20240034). All participants provided written informed consent prior to the evaluation and were assured of their right to withdraw at any time. All data were anonymized to protect participant privacy. Analytical methods of non-volatile components After fermentation of the Ziyan tea wine was completed, the contents of tea polyphenols, caffeine, and free amino acids were determined according to national standards (National Technical Committee 339 on Tea of Standardization Administration of China, 2015); (All China Federation of Supply and Marketing Cooperatives, 2013);. Alcohol content was measured according to GB/T 15038–2006. A 100 mL sample at 20 °C was distilled, and the distillate was collected in a 100 mL volumetric flask. The distillate was kept at 20 °C for 30 min and adjusted to volume with distilled water. Alcohol content was measured using an alcohol meter, and the reading was corrected to 20 °C according to the standard. Results were expressed as % vol (China National Light Industry Council, 2006). The determination of anthocyanin content in tea wine followed the optimized method of Guo et al. for mulberry wine (Guo et al., 2020). Each treatment included three independent fermentation batches as biological replicates. For the biochemical analyses, each biological replicate was measured in triplicate as technical replicates. The mean of the three technical measurements was used as the value of one biological replicate. Statistical analysis was based on the three biological replicates. Analytical methods of volatile components Furthermore, the volatile aroma compounds of the Ziyan tea wine were analyzed by headspace solid-phase microextraction coupled with gas chromatography–mass spectrometry (HS-SPME-GC–MS) (Guo et al., 2020; Liu et al., 2022). Specifically, an aliquot of 1.0 mL of tea wine was placed in a 20-mL headspace vial. 10.0 μL of tert-amyl alcohol at a concentration of 8.06 g/L was added as the internal standard. The vial was equilibrated at 50 °C for 5 min. Volatile compounds were then extracted using a 120 μm DVB/CAR/PDMS SPME fiber at 50 °C for 40 min under no agitation. The fiber was desorbed in the GC injector at 250 °C for 5 min. Separation was performed using an Agilent 7890B—7000D GC–MS/MS system equipped with a DB-5MS column (30 m × 0.25 mm × 0.25 μm). Helium was used as the carrier gas at a constant flow rate of 1.2 mL/min. The oven temperature was held at 50 °C for 2 min, increased to 145 °C at 3 °C/min, increased to 230 °C at 15 °C/min, and held at 230 °C for 3 min. The mass spectrometer was operated in electron ionization mode at 70 eV. The ion source, quadrupole, and transfer line temperatures were 230, 150, and 230 °C, respectively. Data were acquired in full-scan mode over the range of m/z 35–400. No solvent delay was used. Compounds were identified by comparing their mass spectra with the NIST version library using a match score of at least 80%. Retention indices were calculated using a C7–C30 n-alkane series under the same chromatographic conditions and compared with published values. Compounds were further confirmed using authentic standards when available. Semi-quantification was performed using tert-amyl alcohol as the internal standard. The estimated concentration of each compound was calculated from the peak-area ratio between the compound and the internal standard, assuming a response factor of 1.0. Therefore, the reported concentrations are semi-quantitative values. The OAV was calculated as the estimated concentration of each compound divided by its odor threshold in water. The odor threshold and its reference for each compound are provided in Table 3. Because the concentrations were obtained by semi-quantification, the calculated OAVs should be considered estimated values. Table 3Odor value of aroma components for the four treatment samples.CASFragrance compositionChemical ClassesOdorThreshold(μg·L−1)SC1;1OAVBAIOAVBAI1;2OAVSCOAV123-92-2isoamyl acetatePentanolsSweet, fruit, banana solvent93.9774.1589243.745269.4563452.5203104-76-71-Hexanol, 2-ethylHexanolsCitrus, Floral1.281108.7122782.179869.23112501.880678-83-1isobutanolButanolethereal283001.462290.484640.83580.7266431-03-82, 3-ButanedioneKetonesSweet, creamy, with a spicy caramel flavor2.800193.67370106-25-2nerolAlcoholSweet, natural orange blossom4008.62468.540665.42033023726-93-4β-DamascenonePolycyclic CompoundsWoody, sweet, fruity and earthy with green floral notes0.12003960.298616171.92785989-33-3cis-alpha,alpha,5-trimethyl-5-vinyltetrahydrofuran-2-methanolorganic heterocyclic compoundEarthy, floral, sweet, woody0.321672.50411207.12831039.19872647.400064-19-7acetic acidAcetatesPungent, sour vinegar.160,00000.020130.0246067-64-1acetoneKetonesEther, apple, pear0.8320012815.513734998.7067100-52-7benzaldehyde C6H5CHO, the simplest aromatic aldehydeAldehydesSweet, bitter, almond0.351903.61504143.7104685.66014791.5922122-78-1phenylacetaldehyde (PAG)AldehydeFlowers, honey, roses, cherries26216.06185.46326.59206.99701730-91-2Butyric acid, 2-methyl-acidfruity-sweet3030.024834.041429.111254.2057105-54-4Butyric acid, ethyl esterEsterFruity and Juicy Pineapple Brandy81.513.66557.708812.58298.682577-53-2cedarwood alcoholAlcoholCedar, woody, sweet1.10162.574600106-22-9citronellolAlcoholLeather, Wax, Rose, Citrus5.41191.2438566.2767536.44212530.2089112-31-2formaldehydeAldehydeSweet, Waxy, Orange Peel, Citrus, Floral70.800125.944423.8338110-38-3Ethyl decanoateEsterSweet, fruity, apple, brandy11203.286062.170011.71554539.6855141-78-6ethyl acetateEsterLight and fruity sweet weed green326003.56641.64012.40911.8435106-24-1geraniolAlcoholFloral, Fruity, Rose, Citrus30000544.2969629-78-7heptadecaneAlkanesalkanes100000000.20557123-66-0Ethyl caproateEsterGreen, fruity, brandy55.3534.5092203.7295254.2638614.186378-70-6linaloolAlcoholFlower, Green23596.6851503.6242528.19561165.0873112-05-0nonanoic acidAcidWaxy, dirty and cheesy with nuances of cultured dairy products100003.45702.75920124-07-2Caprylic acidAcidGreasy, waxy, rancid, oily vegetable cheeses120406.91065425.0828265.9803974.159160-12-8phenethyl alcoholFluorinated aliphatic substancesFruit, Honey, Clove, Rose, Wine28,90053.238735.277338.182280.893834995-77-2trans-Linalool Oxide (Furan)oxolanesflower fragrance0.194355.1562362.09930098-55-5α-terpineolAlcoholPine, Iris, TillN.D.N.D.N.D.N.D.N.D.5937-11-1τ-CadinolAlcoholbalsamic earthinessN.D.N.D.N.D.N.D.N.D.84-69-5Bis(2-methylpropyl) 1, 2-benzenedicarboxylateEster–N.D.N.D.N.D.N.D.N.D.“N.D.” means that the odor threshold was not available and the OAV could not be calculated. A dash indicates that no odor descriptor was available. Odor value of aroma components for the four treatment samples. “N.D.” means that the odor threshold was not available and the OAV could not be calculated. A dash indicates that no odor descriptor was available. Data analysis Experimental data were processed using Excel 2020. One-way analysis of variance (ANOVA) and t-tests were conducted using SPSS 27.0 software. Data visualization and statistical analysis were performed using Origin 2024 and MetaboAnalyst 6.0. All experiments were analyzed with three replicates. Results and discussion Effects of different yeast fermentation strategies on biochemical indicators and sensory scores of Ziyan tea wine Based on the criteria in Table 1, the sensory quality and biochemical composition of tea wines fermented with single or mixed yeast strains were evaluated, and the results are summarized in Table 2. As shown in Table 2, co-fermentation generally resulted in lower alcohol content, higher amino acid retention, and altered contents of polyphenols and anthocyanins compared with single-strain fermentation (SC and BAI). These changes can be ascribed to the lower ethanol-producing capacity but stronger aroma-synthesis ability of non-Saccharomyces yeasts (Lin et al., 2020). Moreover, the oxidative metabolism of non-Saccharomyces yeasts can help reduce alcohol content during wine fermentation (Li et al., 2019). Based on the study by Shao et al., we speculate that the increased amino acid content was probably due to reduced consumption of amino acids via decarboxylation and deamination during alcohol production in the presence of H. uvarum, leading to higher amino acid retention than in single-culture fermentation with S. cerevisiae (Shao et al., 2024), which means its fresh and refreshing taste will be enhanced. Table 2Complete experimental results.Type of yeast additionAlcohol content(%vol)amino acids (mg·ml-1)caffeine (mg·ml-1)tea polyphenols (mg·ml-1)anthocyanin (μg·ml-1)Sensory Score (points)S. cerevisiae9.80 ± 1.08ab1.21 ± 0.01a0.51 ± 0.00a1.89 ± 0.02de3.27 ± 0.08c85.97 ± 0.54cS. cerevisiae:H. uvarum = 1:18.40 ± 0.53a1.23 ± 0.02a0.53 ± 0.00a1.94 ± 0.03 cd4.98 ± 0.39ab92.74 ± 0.72aS. cerevisiae:H. uvarum = 1:29.67 ± 1.15ab1.28 ± 0.03a0.63 ± 0.02b1.71 ± 0.01f4.30 ± 0.25a88.74 ± 0.01bS. cerevisiae:H. uvarum = 2:18.97 ± 0.85ab1.27 ± 0.13a0.58 ± 0.05a1.60 ± 0.01 g5.17 ± 0.24a79.86 ± 1.28dEnartis Ferm Aroma White11.40 ± 1.42b1.24 ± 0.03a0.62 ± 0.02b2.05 ± 0.06bc4.90 ± 0.18a89.98 ± 0.56abEnartis Ferm Aroma White:H. uvarum = 1:110.36 ± 0.64ab1.32 ± 0.09a0.53 ± 0.02a1.79 ± 0.02ef4.85. ± 0.43a74.23 ± 0.13eEnartis Ferm Aroma White:H. uvarum = 1:210.43 ± 0.61ab1.37 ± 0.15a0.52 ± 0.01a2.30 ± 0.07a4.45 ± 0.67ab90.89 ± 0.98abEnartis Ferm Aroma White:H. uvarum = 2:110.03 ± 0.80ab1.34 ± 0.14a0.54 ± 0.03a2.13 ± 0.05b3.76 ± 0.16bc81.92 ± 2.13dDifferent lowercase letters indicate significant differences among treatments with different yeast strains and inoculation ratios at P < 0.05. Complete experimental results. Different lowercase letters indicate significant differences among treatments with different yeast strains and inoculation ratios at P < 0.05. For the combination of S. cerevisiae with H. uvarum, when S. cerevisiae and H. uvarum were co-fermented at a 1:1 ratio, the tea wine exhibited the lowest alcohol content (8.4% vol), the highest tea polyphenol concentration, relatively low caffeine content, and the best overall sensory score, characterized by rich aromatic layers and a well-balanced, mellow taste with harmonious tea and wine notes. At a 1:2 ratio, the tea wine showed the highest levels of amino acids and caffeine (0.63 mg/mL), moderate alcohol content, but relatively weak aroma and less balanced taste, resulting in a medium sensory score. A 2:1 ratio led to the highest anthocyanin content; however, tea aroma and aromatic complexity were less distinct, yielding the lowest sensory score. Thus, considering alcohol content, appearance, aroma, and taste comprehensively, the optimal co-fermentation ratio of S. cerevisiae to H. uvarum is 1:1. Further optimization of the co-fermentation with Enartis Ferm Aroma White and H. uvarum revealed that at a 1:2 ratio, the tea wine achieved the highest amino acid content (1.37 mg/mL), along with suitable levels of tea polyphenols, anthocyanins, and caffeine. It presented complex aromatic layers and a full, harmonious taste, receiving the highest sensory score. At a 1:1 ratio, the aroma profile was less distinctive and the taste relatively plain, resulting in the lowest score. A 2:1 ratio offered good integration of tea and wine aromas but limited complexity, resulting in a mid-level sensory score. Therefore, taking into account alcohol content, appearance, aromatic complexity, and taste richness, the recommended co-inoculation ratio of Enartis Ferm Aroma White to H. uvarum is 1:2. Based on the above results, we conducted further analysis on the tea wines fermented with the optimal mixing ratios of Saccharomyces yeast and non-Saccharomyces strains that achieved the highest sensory scores, as well as the tea wines fermented with the two single Saccharomyces yeasts, specifically: S. cerevisiae alone, co-fermentation of S. cerevisiae and H. uvarum at a 1:1 ratio, Enartis Ferm Aroma White alone, and co-fermentation of Enartis Ferm Aroma White and H. uvarum at a 1:2 ratio. These are hereinafter referred to as SC, SC1;1, BAI, and BAI1;2, respectively, and the results are as follows. Dynamic changes in key chemical components during yeast fermentation As shown in Fig. 1, the amino acid content in tea wine fermented with any yeast strain reached its maximum on the second day, which is the final day of aerobic fermentation. This increase can be attributed to the enhanced energy supply for amino acid synthesis under aerobic conditions (Shao et al., 2024). During the subsequent anaerobic phase, amino acids in single-yeast fermentations (SC and BAI) decreased continuously, reaching the lowest level on day 10. In contrast, co-fermentations (SC1;1 and BAI1;2) exhibited a pattern of initial decrease followed by a late increase, with the lowest values observed on day 6 and day 8, respectively. This trend suggests the effects of mixed fermentation between Saccharomyces and non-Saccharomyces yeasts, which may improve the stability and accumulation of amino acids (Guo et al., 2024). Under anaerobic conditions, the decrease in amino acid content during fermentation may be related to the use of amino acids by yeasts as carbon and nitrogen sources (Mao et al., 2024). However, amino acid metabolism was not directly measured in this study. Dynamic changes in key chemical components of tea wine during 10-day fermentation with different yeasts. (A-B) Amino acid content; (C—D) caffeine content; (E-F) anthocyanin content; (G-H) tea polyphenols content. Data were analyzed by one-way analysis of variance (ANOVA) with statistical significance set at P < 0.05. Caffeine content did not change significantly during fermentation. The final caffeine content in tea wines fermented with mixed yeasts (SC1;1 and BAI1;2) was higher than that in tea wine fermented with a single yeast strain. Moreover, the rate of caffeine decrease in these co-fermentations was lower compared to single-yeast fermentation. The stability of caffeine is likely attributable to its molecular structure, as its degradation requires the involvement of specific enzymes, such as caffeine dehydrogenase. However, the yeast strains used in this study may lack the relevant metabolic capacity. Regarding anthocyanins, the main form in tea wine was delphinidin. With prolonged fermentation, acetaldehyde produced by yeast metabolism can polymerize with anthocyanins, leading to a decrease in their content (Vernhet et al., 2020). As shown by the experimental results, the final anthocyanin content of tea wine from co-fermentation of the two yeast strains was lower than that in BAI but higher than that in SC. Besides, the content of tea polyphenols also reached the maximum in the aerobic fermentation stage. The reasons may include the chemical reaction promoted by oxygen, the enzymatic action, and the enhancement of microbial metabolic activity (T. Hu et al., 2022; Kim et al., 2011). Notably, from the fourth day until the end, the tea polyphenol content of tea wine fermented by each mixed yeast was higher than that in the corresponding single-yeast wine. The higher polyphenol content in some mixed-fermentation treatments may be associated with changes in the fermentation environment (Ma et al., 2014) or enzymatic activity (Luo et al., 2024; Yao et al., 2025). However, pH, dissolved oxygen, and enzyme activities were not measured. Therefore, this explanation remains hypothetical. Collectively, these results indicate that co-fermentation helps preserve functional components in tea wine, thereby contributing to the maintenance of its health-related properties (Shang et al., 2021). However, the initial soluble solids, pH, sugar content, and final residual sugar were not measured in this study. Therefore, the effects of mixed fermentation on sugar utilization and fermentation efficiency could not be fully evaluated. These parameters should be included in future studies. Volatile composition and flavor formation as influenced by yeast combinations in Ziyan tea wine There were 61 common volatile components were detected in the four kinds of tea wine. Among them, 15 were esters, 6 were alcohols, and 6 were aldehydes (Table 3). Multivariate differentiation of volatile profiles among yeast treatments To systematically reveal the effects of different yeast fermentation strategies on volatile composition, principal component analysis (PCA) was applied (Fig. 2A). The results revealed that all samples obtained through co-fermentation formed distinct clusters according to their aroma profiles, with a clear separation among the four types of tea wines fermented with different yeasts. The differences in volatile compounds may be related to interactions between the two yeasts. However, yeast growth, substrate use, gene expression, and metabolic pathways were not directly investigated in this study. Volatile aroma profiles of tea wines fermented with four yeast treatments. (A) Principal component analysis (PCA) score plot showing the clustering of samples. (B) Circular diagram illustrating the distribution of volatile compound classes (alcohols, acids, heterocyclic compounds, terpenoids, etc.) across the four treatments: SC, SC1;1, BAI, and BAI1:2 (from inner to outer rings). The circular diagram (Fig. 2B) further illustrates the distribution of volatile compound classes across the four treatments. SC and BAI1;2 exhibited higher contents of alcohols, whereas SC1;1 and BAI showed relatively high and evenly distributed levels of acids, heterocyclic compounds, and terpenoids. Differential accumulation of shared and unique volatiles across treatments As shown in the heatmap (Fig. 3A), the relative abundance of volatile compounds clearly differentiated the four tea wine samples. In the co-fermentation samples, the levels of compounds such as nerol (with a sweet, natural orange-blossom aroma) and 2,3-butanedione (with a sweet, creamy note) were increased compared to the single-yeast fermentation samples; conversely, the concentrations of compounds like acetone (with a pungent, sweet odor) and benzaldehyde (with a bitter almond aroma) were decreased. These observations might be partially attributable to interactions via signaling molecules (e.g., specific hormones or metabolites released by yeasts) during co-fermentation, which may influence gene expression in each yeast (Zhou et al., 2019), potentially contributing to divergent aroma outcomes. A mechanism similar to that reported by Zhang et al. may also be at play in this study, in which non-Saccharomyces yeasts (e.g., Torulaspora delbrueckii) in co-fermentation reduce the formation of by-products such as excess ethanol and sulfides by competing for substrates, thereby improving the overall aroma profile (K. Hu, Jin, Xu, et al., 2018). Comparative analysis of volatile compound distribution among four tea wine fermentation treatments. (A) Heatmap displaying the relative abundance of individual volatile compounds. Colors range from pink (above average) to blue (below average). (B) Venn diagram quantifying the number of shared and unique volatile compounds. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Additionally, as can be seen from the Venn diagram (Fig. 3B), a total of 20 identical volatile compounds were detected across the four tea wines. Except for SC1;1, the other three tea wines each possessed their own unique volatile compounds that were not shared with any other tea wine, among which BAI1;2 had the highest number of unique volatile compounds. Analysis of flavor differences in Ziyan tea wine with single versus mixed yeasts As shown in Fig. 4, all four tea wines exhibited a floral-oriented flavor profile. Among them, BAI1;2 displayed relatively greater aroma complexity and higher levels of floral and fruity notes compared to the products fermented by a single yeast strain. This may be attributed to the higher β-glucosidase activity of non-Saccharomyces yeasts in the fermentation environment compared to that of Saccharomyces cerevisiae (Qin et al., 2022). This enzyme not only exhibits good ethanol tolerance but also contributes to more intense and pleasant floral and fruity aromas (Wang et al., 2024). During co-fermentation, the mixed fermentation effects of these enzymatic activities resulted in greater diversity and better harmony of aromas in the tea wine, thereby enhancing its overall flavor profile. Flavor characteristics of tea wines fermented with single and mixed yeasts. (A) Stacked line chart showing the concentration profiles of aroma compounds across different treatments; (B) Chord diagram illustrating the abundance among aroma categories; (C) Classification of aroma compounds by category, displayed as a flavor wheel. In contrast, tea wine fermented with a 1:1 ratio of S. cerevisiae to H. uvarum showed lower diversity of flavor compounds than that produced with a single yeast fermentation. However, as presented in Table 2, the sensory score of the co-fermented tea wine was superior to that of the single-yeast-fermented product. This could be attributed to its more focused and harmonious aroma profile, dominated by floral, fruity, and ethereal notes, with fewer undesirable off-flavors such as pungent acidity or solvent-like odor. It also reduced the burning sensation and astringency associated with high alcohol content (Table 2). Furthermore, this yeast combination introduced more desirable flavors (e.g., citrus, fatty notes), which enhanced the taste (K. Hu, Jin, Mei, et al., 2018) and overall flavor characteristics of the tea wine. Schematic illustration of flavor improvement by co-fermentation To further research the potential metabolic mechanisms underlying the aroma differences and to provide a rational account of key flavor compounds during co-fermentation, a schematic diagram was constructed based on the preceding results (Fig. 5). The co-fermentation with Hanseniaspora uvarum and Saccharomyces cerevisiae or Enartis Ferm Aroma White led to a lower content of caprylic acid, a compound associated with oily, waxy, rancid, and cheesy off-flavors. This reduction may be explained by competitive substrate uptake between co-cultured yeasts. This mechanism has been shown in a study by Zhang et al., could lower the metabolic rate of S. cerevisiae and weaken mitochondrial fatty acid synthesis (Zhang et al., 2018). In contrast, the levels of nerol (sweet, orange-blossom aroma) and diacetyl (creamy, popcorn-like note) were elevated in co-fermentation samples, which are crucial for enhancing fruity characteristics in wine (Yang et al., 2024). The changes in the volatile profile may be associated with fatty acid metabolism, geraniol formation, and butanoate metabolism. Previous studies have shown that mixed fermentation with different yeast species can change the production of esters and other aroma compounds. However, these metabolic pathways were not directly measured in the present study. Therefore, these explanations remain hypotheses and require further investigation. Schematic diagram of the co-fermentation strategy for improving the flavor profile of Ziyan tea wine. Upward green arrows (↑) and downward grey arrows (↓) indicate increased desirable compounds and decreased off-flavor compounds, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Previous studies have shown that co-fermentation of Torulaspora delbrueckii with S. cerevisiae significantly increased the content of medium-chain fatty acid ethyl esters and improved the fruity aroma profile (Choi et al., 2023; Hu et al., 2018b), and co-fermentation of P. fermentans with S. cerevisiae significantly enhanced the formation of aroma compounds (N. Li et al., 2020), which provides evidence supporting the hypothesis. In the co-fermentation, the content of diacetyl, imparting creamy and popcorn-like notes, was also increased. These mechanisms likely operate in the co-fermented tea wine, thereby contributing to the distinct flavor characteristics of Ziyan tea wine. Conclusion This study compared the sensory properties, major biochemical components, and volatile compounds of Ziyan tea wine produced by single and mixed yeast fermentation. Mixed fermentation changed the flavor profile and sensory properties of the tea wine. The S. cerevisiae / H. uvarum ratio of 1:1 significantly increased the sensory score compared with S. cerevisiae monoculture. The Enartis Ferm Aroma White / H. uvarum ratio of 1:2 also produced a distinctive volatile profile, although its sensory score was not significantly different from that of Enartis Ferm Aroma White monoculture. Some mixed-fermentation treatments showed higher levels of floral and fruity aroma compounds and lower levels of some volatile acids. However, the mechanisms underlying these changes were not directly investigated and require further study. Ethics and consent to participate declarations This sensory evaluation study was approved by the Academic Ethical and Welfare Committee of Sichuan Agricultural University (Approval No. H20240034). All participants provided written informed consent prior to the evaluation and were assured of their right to withdraw at any time. All data were anonymized to protect participant privacy. The study was conducted in accordance with the ethical principles of the World Medical Association's Declaration of Helsinki. CRediT authorship contribution statement Yajie Hua: Writing – original draft, Methodology, Investigation, Data curation, Conceptualization. Lechi Ling: Writing – original draft, Visualization, Resources, Investigation, Data curation. Xue Xia: Validation, Software, Resources, Investigation. Yueling Zhao: Writing – review & editing. Yongxian Chen: Writing – review & editing. Ling Lin: Writing – review & editing, Writing – original draft, Supervision, Conceptualization. Wei Xu: Writing – review & editing, Funding acquisition, Conceptualization. Ethics declaration Informed consent and patient details Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed. Studies in human This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was conducted in accordance with World Medical Association Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Participants. This study was approved by the Sichuan Agricultural University Academic Ethical and Welfare Committee. (Approval No. Approval NO. H20240034). 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. Acknowledgments This work was supported by Special Project for Guiding Local Science and Technology Development by the Central Government of Sichuan Province (2025ZYD0046). Contributor Information Ling Lin, Email: linling941106@sina.com. Wei Xu, Email: xuweianti@sicau.edu.cn. Data availability Data will be made available on request. References Associated Data Data Availability Statement Data will be made available on request.