From spontaneous to controlled: Simulating key aroma chemical characteristics of spontaneous fermentation with synthetic microbial community in Italian Riesling wine

2.1. Spontaneous fermentation and sampling

Italian Riesling grapes utilized in this experiment were sourced from Penglai, Shandong, China (37.79°N, 120.77°E). During the harvesting season (September 2022), samples were collected from five specific locations in the vineyard (One central area and four corner areas) using a five-point sampling method. Following the stipulated wine production requirements, the Italian Riesling grapes selected for this experimental study underwent manual a series of meticulous procedures. The process entailed a series of manual operations, including destemming, pressing, and crushing, culminating in a comprehensive clarification treatment. During the process of grape crushing, 40 mg/L SO₂ (industrially standard) was added to prevent interference from spoilage microorganisms during fermentation, and 20 mg/L pectinase was added to improve juice yield. The entire process was conducted under strictly controlled conditions to ensure sterility, with all equipment, operators, fermentation tanks, and the fermentation environment being subjected to rigorous sterilization procedures. Upon achieving turbidity (NTU) levels of 200, the grape juice (total sugar 165 g/L, total acid 7.62 g/L, pH 3.30) was transferred into three pristine, aseptic 20 L glass jars for the initiation of spontaneous fermentation. The fermentation temperature was meticulously regulated at a range of 14–17 °C, and the fermentation process was terminated when residual sugar levels diminished to below 4 g/L. Each fermentation tank underwent three biological replicates. Given that the total sugar content of the must was less than 200 g/L, 35 g/L of table sucrose was added to the fermenter during the peak fermentation phase to minimize the impact on the fungi. The fermentation progress was meticulously monitored by conducting specific gravity measurements on a twice-daily basis. Sampling was conducted at various stages of the fermentation process, specifically when the NTU reached 200 (F1); when a decrease in total sugar was observed (F2); when total sugar consumption reached 30 % (F3); when total sugar consumption reached 80 % (F4); and when total sugar was below 4 g/L (F5). The detailed procedure was as follows: prior to sampling, the area surrounding the sampling port of the fermentation tank was meticulously wiped and sterilized using a 75 % ethanol solution. An alcohol lamp was also ignited in the operational area to establish a relatively sterile environment for sampling. Subsequently, a sterilized pipette was employed to draw the sample, which was then transferred into a pre-sterilized centrifuge tube. It was crucial to highlight that each fermentation tank was assigned exclusive aseptic pipettes and centrifuge tubes, thereby preventing cross-contamination among different samples. All samples were stored at −80 °C until analysis. A proportion of the samples was then subjected to high-throughput sequencing, while the remainder was analyzed for its basic physicochemical properties.

2.2. Determination of physical and chemical parameters

Titratable acidity, pH, and volatile acidity were measured directly according to the Chinese national standard GB 15038-2006. The assay of yeast assimilable nitrogen (YAN) was conducted utilizing the 12,807-AMMONIA and 12,809-PAN kits (Biosystem) in conjunction with a fully automated analyzer (Biosystems, Barcelona, Spain). The samples were subjected to a centrifugal process at a speed of 12,000 rpm for five minutes. Thereafter, the upper layer of the samples was analyzed following the instructions provided in the kit. The quantification of glucose, fructose, organic acids, glycerol, and ethanol was conducted via high-performance liquid chromatography (LC1260, Agilent Technologies Co. Ltd., USA); the specific method is detailed in Huang et al. (2023).

2.3. Determination of volatile compounds

To prepare samples, 1.2 g NaCl was weighed into a 20 mL glass headspace (HS) vial pre-loaded with 5 mL of sample. A 10 μL aliquot of 4-methyl-2-pentanol (1.013 g/L) was then introduced as internal standard before the vial was hermetically sealed with a magnetic screw cap equipped with a silicone septum. The extraction of volatile compounds was strictly performed using Headspace Solid-Phase Microextraction (HS-SPME). Key operational parameters, including fiber coating type, sample incubation temperature and time, extraction time, and desorption time, were slightly modified based on our research group's established and fully validated protocol (Chen et al., 2024). After equilibrating the sample at 40 °C for 30 min, the SPME fiber was inserted and exposed for 30 min under continuous agitation at 500 rpm. Following extraction, the fiber was automatically introduced into the injector for thermal desorption over 8 min, with helium as the carrier gas at a flow rate of 1 mL/min. The injector temperature was maintained at 250 °C, and sample introduction was performed in split mode (5:1). The extracted volatiles were analyzed on a Trace 1610 GC hyphenated to an ISQ 7610 MS (both Thermo Fisher Scientific, USA). The gas chromatography temperature program was set as follows: initial temperature 50 °C held for 1 min, then ramped at 3 °C/min to 220 °C and held for 5 min. The mass spectrometer interface temperature was set to 250 °C, while the ion source and quadrupole temperatures were maintained at 250 °C and 150 °C, respectively. Electron ionization (EI) mode was used with an ionization energy of 70 eV. Compound identification was achieved by matching mass spectra to the NIST20.L library and cross-referencing retention times and diagnostic m/z fragments. Calibration standards were prepared in a synthetic wine matrix (15 % v/v ethanol, 5 g/L tartaric acid) whose pH was accurately adjusted to 3.50 with 1 M NaOH. Volatiles were quantified by the internal standard method using external calibration curves. Compounds without available standards were quantified using surrogates of similar chemical structure and carbon number. For detailed quantitative information regarding the linear fitting, the R² values, linear ranges, odor attributes, and odor categories of the volatile compounds analyzed in this study are provided in Supplementary Table 1. Volatile compounds lacking corresponding reference standards (2,4-Di-tert-butylphenol, ethyl benzoate and propanoic acid) were quantified using calibration curves of compounds with the same functional group and similar carbon chain length.

2.4. Total DNA extraction and high-throughput sequencing

The total DNA of the microbiome was extracted using the HiPure Stool DNA Kit (Magen, Guangzhou, China), in accordance with the manufacturer's instructions, with final elution in 80 μL of preheated (70 °C) Buffer AE. The concentration of DNA was determined by measuring the optical density of the samples at 260 nm using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). The integrity and quality of the samples were evaluated by subjecting them to 1 % agarose gel electrophoresis. The genomic DNA was dispatched to Genedenovo Biotech Ltd. The amplification and sequencing were conducted at the Guangzhou Research Hub (https://www.genedenovo.com/). The filtration and processing of data were conducted in accordance with the methodologies outlined in preceding studies (Zhang, Zhuang, et al., 2025; Zhang, Zhao, et al., 2025). Samples for high-throughput analysis were prepared in triplicate.

In order to amplify the fungal ITS2 (Internal Transcribed Spacer) region, the primers ITS3_KYO2F (5′-GATGAAGAACGYAGYRAA-3′) and ITS4R (5′-TCCTCCGCTTATTGATATATGC-3′) were utilized. The PCR reactions were performed in triplicate using a 50 μL mixture comprising 5 μL of 10 × KOD Buffer, 5 μL of 2.5 mM dNTPs, 1.5 μL of each primer (5 μM), 1 μL of KOD Polymerase, and 100 ng of template DNA. The amplification was initiated with an initial denaturation step at 95 °C for 2 min, followed by 27 cycles of denaturation at 98 °C for 10 s, primer binding at 62 °C for 30 s, and extension at 68 °C for 30 s.

2.5. Screening, culture, and identification of fungi

The samples were diluted using a sterile saline solution via a gradient dilution technique (F1: 10²–10⁴; F2: 10⁴–10⁶; F3: 10⁵–10⁷; F4: 10⁵–10⁷; F5: 10⁵–10⁷). The diluted solution was thoroughly mixed and 1000 μL of this was applied to WLN nutrient agar (containing 100 mg/L chloramphenicol to inhibit bacterial growth). The agar was then subjected to an incubation process at a temperature of 28 °C for a period of 2–3 days (three parallels). The most suitable dilution was selected, and the total colony count per milliliter was determined. The WLN agar plates were observed post-incubation at 28 °C for a period of 5–6 days. Thereafter, the colony color and morphology were recorded for subsequent classification and identification. The colonies, which exhibited a range of morphologies, were meticulously enumerated and documented individually. The selection of 3–5 WLN culture types for pure culture was based on each morphological stage of Italian Riesling spontaneous fermentation. Purely isolated strains were cultivated in YPD medium at 28 °C for 2 days. Glycerol (final concentration 25 %) was added, and the strains were stored at −80 °C.

The genomic DNA was extracted using the Fungi Genomic DNA Extraction Kit (Beijing Solarbio Science & Technology Co., Ltd.) in accordance with to the manufacturer's instructions. Universal primers NL1 and NL4 were utilized in the amplification and sequencing of the 26S rDNA D1/D2 region. The identification of 26S rDNA yeast strains was facilitated by the utilization of NL1 (5′-GCATATCAATAAGCGGAGGAAAAG-3′) and NL4 (5′-GGTCCGTGTTTCAAGACGG-3′). The PCR reaction was performed in a 50 μL reaction volume, comprising 25 μL 2 × Rapid Taq Master Mix, 2 μL of each 10 μM forward and reverse primer, 1 μL of DNA template (200 ng), and 20 μL of sterile deionized water. PCR program was 95 °C for 3 min (pre-denaturation), 95 °C for 15 s, 55 °C for 45 s, 72 °C for 30 s, and 72 °C for 5 min (final extension). A 4 μL of the PCR product was subjected to electrophoresis on a 1 % agarose gel for approximately 25 min. Positive products were subsequently sent to Beijing Genomics Co., Ltd. for purification and sequencing. The sequences were analyzed using the BLAST tool available at NCBI (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The range of gene sequence similarity was from 99 % to 100 %, which is consistent with the standard of ≤ 1 % difference within the same species.

Strains identified as S. cerevisiae were subsequently subjected to Interdelta fingerprinting. The primer sequence employed was delta12 (5′-TCAACAATGGAATCCCAAC-3′) and delta21 (5′-CATCTTAACACCGTATGA-3′). The PCR reaction system comprised, 12.5 μL 2 × Rapid Taq Master Mix, 1 μL upstream primer (10 μM), 1 μL downstream primer (10 μM), 0.7 μL (80 ng/μL) DNA template, and sterile deionized water was made up to 25 μL. The thermal cycling parameters for the PCR comprised pre-denaturation at 95 °C for 4 min, denaturation at 95 °C for 30 s, reversion at 46 °C for 30 s, extension at 72 °C for 90 s, and a total of 35 cycles, finally, 72 °C was used to make up for the flatness for 10 min. The amplified products were then subjected to 2 % m/v agarose gel electrophoresis for 1 h at a constant voltage of 120 V. The subjects were observed and photographed using a UV imager.

2.6. Fermentation of synthetic microbial communities

In a 250 mL flask, 150 mL of filtered (sterile 0.22 μm filter membrane) Italian Riesling grape juice (harvested in September 2024 from Penglai, Shandong) was subjected to fermentation at 18 °C (Since the total sugar content of the grape juice reached 200 g/L, no sucrose was added). The flask was inoculated with an 18-h pre-culture; specific inoculation details are provided in Table 1. Besides the aforementioned experiments, single-strain fermentations were performed using the commercial S. cerevisiae F15 and three indigenous strains (SC1, SC21, and SC23). The progression of fermentation was monitored by tracking the loss of CO₂, with data recorded at 24-h intervals. A total of three biological replicates were conducted for each fermentation vessel. The fermentation process was considered to have reached its conclusion when the rate of weight loss remained below 0.2 g/L for two consecutive days.

2.7. Statistical analysis

The results obtained are presented as mean ± standard deviation. Microbial count data were analyzed in CFU/mL units. Fungal identification results were submitted to the NCBI database (https://www.ncbi.nlm.nih.gov/) for sequence alignment and analysis. SPSS 25 (IBM® SPSS® Statistics, NY, USA) and GraphPad Prism 8.0.2 (GraphPad Software, Inc., La Jolla, CA, USA) were used to visualize fungal composition, liquid chromatography data, and volatile components, including one-way ANOVA for significant differences in raw data. Principal component analysis (PCA) was used to analyze volatile differences in Italian Riesling at different spontaneous fermentation stages, with visualization facilitated via SIMCA 14.1 (Umetrics AB, Umeå, Sweden). The O2PLS model was constructed utilizing an online platform (https://www.omicshare.com/) to explore relationships between microbes and flavor compounds. Heatmaps were visualized using TBtools (Toolbox for Biologists; version 2.042, China). A correlation analysis was performed between aroma-active compounds and yeasts, with ggplot2 software utilized in R (version 3.4.0) for graphical representation.

3.1. Dynamics of metabolites during wine fermentation

The physicochemical parameters of Italian Riesling during spontaneous fermentation at various stages are presented in Table 2. Throughout spontaneous fermentation, reducing sugars declined sharply; at the terminal stage (F5) total residual sugar fell below 4 g/L, signifying successful fermentation yet leaving a trace of residual sweetness. As sugars were depleted, ethanol and glycerol accumulated significantly, attaining final concentrations of 11.57 ± 0.01 g/L and 7.96 ± 0.23 g/L, respectively. YAN was abundant when fermentation initiated, and it declined steadily and markedly throughout fermentation. Strikingly, YAN stabilized during the terminal phase (F4-F5), ceasing its decline, an observation attributed to supplementary nitrogen released via late-stage yeast autolysis and protein hydrolysis (Christofi et al., 2022; Romano et al., 2022). Overall, total acidity and pH exhibited inverse trends. Initially, tartaric and malic acids dominated the organic-acid profile, with succinic acid presented at lower levels, which was consistent with previous reports (Ding et al., 2024). Throughout fermentation, citric acid levels remained relatively constant, while tartaric acid levels exhibited a marked increase followed by a significant decrease. Malic acid levels remained stable in the early fermentation phase (stages F1-F2), but decreased markedly from stage F3 onward, reaching relatively low levels by the end of fermentation. Succinic acid levels remained relatively constant in the first three stages, increased significantly at F4, and reached 1.04 ± 0.05 g/L at F5. Lactic acid was produced in the late fermentation stage, while acetic acid was generated during the active fermentation phase. These alterations in physicochemical properties are closely associated with microbial dynamics during fermentation. The marked shifts in the aforementioned physicochemical indices (sugars, ethanol, glycerol, YAN, organic acids, and pH) are intimately linked to the metabolic activity and population succession of yeast communities throughout fermentation.

3.2. Dynamic succession of fungal microbial during spontaneous fermentations

3.3. Composition of spontaneous wine aromas

The study identified 50 aroma compounds across 10 categories (Supplementary material Table S3). PCA demonstrated that fermentation significantly altered aroma compounds, with the first two principal components explaining 87.7 % of the variation (Fig. 3A). During the process of vinification, the transformation of grape juice into wine is accompanied by the release of various volatile compounds, which collectively contribute to the formation of the wine's aroma profile. Higher alcohols, organic acids, and esters are the key aromatic compounds in wine, significantly influencing its aromatic profile and sensory attributes (He et al., 2023). Major volatile compounds exhibited marked dynamic changes throughout alcoholic fermentation (P < 0.05). Higher alcohols, fatty acids, and esters exhibited pronounced accumulation from start to finish (Fig. 3B). Carbonyls followed a biphasic trajectory, and their concentration decreased in the early stage of fermentation, but increased significantly at the end of fermentation. Conversely, terpenes declined below their initial must concentrations at the final stage, implying partial degradation or biotransformation. Remaining compound classes accumulated steadily, reaching maximal concentrations at fermentation completion (Fig. 3B). Collectively, these dynamics underscore the decisive role of Saccharomyces metabolism in shaping the wine's aromatic profile.

As indicated by the odor activity values (OAV > 1), the following compounds were identified as key contributors to the aroma profile of Italian Riesling wine: phenylethyl alcohol, 1-propanol, isoamyl alcohol, phenol, 2-methoxy-4-ethylphenol, 2,4-di-tert-butylphenol, nonanal, geraniol, ethyl acetate, isoamyl acetate, ethyl butyrate, ethyl hexanoate, ethyl caprylate, isobutyric acid, octanoic acid, and n-decanoic acid. Fatty alcohols such as 1-propanol, isobutyric acid and isopentanol are considered to be representative of those found in wine. 1-propanol has been shown to contribute a mature fruit and alcoholic aroma to wine, while isobutyl alcohol and isoamyl alcohol may impart undesirable alcoholic and nail polish-like pungent notes (Liu et al., 2023). Phenylethyl alcohol has been shown to contribute a desirable honey and rose-like aroma to wine (Liu et al., 2023). Research suggests that higher alcohol levels below 300 mg/L promote the expression of floral and fruity aromas in wine, enhancing its aromatic complexity, whereas levels exceeding 400 mg/L may result in undesirable odors (Lai et al., 2023). Brettanomyces/Dekkera bruxellensis has been observed to produce 4-ethylphenol and 2-methoxy-4-ethylphenol during growth and metabolism, which can negatively affect wine sensory attributes (Childs et al., 2015). The study have also shown that 2-methoxy-4-ethylphenol can provide fragrance (Wang et al., 2023). This may provide a rationale for the elevated levels of volatile phenols observed in this study. Geraniol, nerol, citronellol, and α-terpineol have been identified as the most prevalent compounds in Muscat wines (Chen et al., 2021). The study identified five terpene compounds, among which d-limonene, citronellol, and geraniol contributed to the aroma of Italian Riesling wine, with geraniol being especially significant. While a small fraction of esters accumulate during grape maturation, most esters are synthesized by yeast during alcoholic fermentation. These compounds characteristically contribute fruity and floral notes to wine, thereby playing a pivotal role in shaping its aromatic profile (Dzialo et al., 2017). Fatty acid metabolism is the primary pathway for the formation of volatile acids, including isobutyric acid, octanoic acid, and propionic acid. Fatty acids function as precursors to fatty acid ethyl esters, and their metabolic evolution exerts a substantial influence on the synthesis of ethyl ester compounds in wine. These compounds elucidate the role of aroma substances in Italian Riesling wine and their influence on its sensory attributes. Elucidation of the origins and contributions of these compounds offers a scientific foundation for refining fermentation processes and improving the aromatic quality of wine.

3.4. Screening and characterization of core fungi populations

In order to delineate the core fungi community in Italian Riesling wine fermentation, O2PLS modeling and Pearman correlation analysis were employed. These tools analyzed the relationships between fungal communities, species, S. cerevisiae genotypes, and key volatile flavor compounds during spontaneous fermentation, thereby identifying key fungi communities. In the model associating fungal communities with key volatile compounds, the X matrix (key flavor compounds) explained 0.976 of the variance, while the Y matrix (fungal microbiota) explained 0.991 (R²Xcorr = 0.894, R²Ycorr = 0.991), confirming a robust model fit. The results indicated that Saccharomyces, Mycosphaerella, Penicillium, Aspergillus, and Aureobasidium exhibited correlations with key aroma compounds, excluding geraniol. In contrast, Hanseniaspora, Colletotrichum, Papiliotrema, Alternaria, and Cladosporium demonstrated correlations exclusively with geraniol (Fig. 4A).

In the model linking fungal culturomics microbiota to key volatile compounds, the X matrix (key flavor compounds) accounted for 0.946 of the variance, while the Y matrix (fungal culturomics microbiota) accounted for 0.968 (R²Xcorr = 0.946, R²Ycorr = 0.796), indicating a robust model fit. The results of the study demonstrated that, non-Saccharomyces, S. bacillaris exhibited significant correlations with key aroma compounds. S. bacillaris, a non-Saccharomyces naturally present on grape berry surfaces, has been shown to enhance terpene production (Wang et al., 2024). Furthermore, H. vineae, P. kudriavzevii, and H. uvarum showed significant correlations with geraniol, while S. cerevisiae exhibited strong correlations with key aroma compounds (such as 2-methoxy-4-ethylphenol, octanoic acid, ethyl caprylate, isobutyric acid, 2,4-Di-tert-butylphenol, n-Decanoic acid and so on, Fig. 4B). Concurrently, S. bacillaris dominated the grape must microbiota, whereas S. cerevisiae prevailed throughout alcoholic fermentation. Overall, S. cerevisiae and S. bacillaris exhibited significant correlations with the majority of core volatile compounds, with S. cerevisiae being particularly notable, a finding that aligns with previous research (Ma, Peng, et al., 2023; Ma, Yu, et al., 2023).

In further pursuit of the correlations disparate different S. cerevisiae genotypes and key aroma compounds, this study has constructed a correlation heatmap using the Pearson correlation coefficient for dominant genotypes and key aroma compounds. The results indicated that S. cerevisiae SC1 showed significant positive correlations with ethyl caprylate, isobutyric acid, n-decanoic acid, 2,4-Di-tert-butylphenol, and octanoic acid. S. cerevisiae SC20, SC21 and SC23 exhibited positive correlations with all key aroma compounds, except geraniol, exhibiting significant positive correlations, especially with ethyl acetate, phenol, 1-propanol, isoamyl acetate, isoamyl alcohol, ethyl hexanoate, phenyl ethanol, and nonanal. Furthermore, SC21 showed significant positive correlations with ethyl caprylate and 2,4-Di-tert-butylphenol (Fig. 4C). Although SC20 exhibited significant correlations with aroma compounds comparable to SC21 and SC23, its correlation strengths were lower than those observed for SC21 and SC23. Consequently, strains SC21 and SC23 were prioritized for selection. While SC1 exhibited complementary metabolic functions to SC21 and SC23, the core microbial community in Italian Riesling wine fermentation consists of one non-Saccharomyces (S. bacillaris) and three S. cerevisiae strains (SC1, SC21, SC23). The synthetic microbial community under investigation in this study enables more precise identification of different S. cerevisiae genotypes, thus distinguishing it from conventional synthetic microbial communities. Accurate identification and utilization of these core microbes can provide the wine industry with an innovative fermentation strategy, thereby enhancing product flavor complexity and market competitiveness.

3.5. Fermentation performance and aroma of synthetic microbial communities

The present study focused on constructing and analyzing a synthetic microbial community for Italian Riesling wine, with the aim of closely replicating the chemical aroma profile of traditional spontaneous fermentation by controlling the mixture of core microbial species and the intraspecific diversity of S. cerevisiae. The subsequent section of this study involved an assessment of the effects of a synthetic community composed of S. bacillaris and S. cerevisiae SC1, SC21, and SC23. As shown in Fig. 5, the fermentation performance of the synthetic microbial consortia is demonstrated as follows: Except for the SB group, all other synthetic microbial consortia groups and their corresponding single-strain controls completed the fermentation process within 10 days without encountering fermentation stagnation or delays. This indicates the well-designed synthetic microbial consortia exhibit good rationality. Analyzing glycerol yield (Fig. 5B), the commercial yeast F15 showed the lowest glycerol production, while All-△SC23 and All-△SC21 exhibited the highest glycerol yields, followed by All-△SC1 and All, which were significantly higher than those of single-strain fermentations. This suggests that the synergistic action of the synthetic microbial consortia effectively enhances glycerol generation. Regarding ethanol content (Fig. 5C), the SB group, which did not complete fermentation, showed the lowest ethanol level. The commercial yeast F15 displayed the highest ethanol content, indicating its high efficiency in ethanol conversion. The ethanol levels of SC21 and SC23 were intermediate, with All falling in between. In terms of acetic acid production (Fig. 5D), the SB group exhibited the highest yield, followed by SC1. The acetic acid production of All-△SB and All-△SC23 was slightly higher than that of the All group.

PCA was employed to analyze key aroma compounds in wines produced by All, SF, community minus one strain, and single strain of Italian Riesling grapes. The results indicated that the PC1 and the PC2 of the PCA model explained 51.09 % and 17.32 % of the variance, respectively (Fig. 6A). It is noteworthy that the distance between the SF group and the All group was relatively small, indicating minor feature differences in the original high-dimensional space, consistent with the results reported by Peng et al. (2025). Additionally, hierarchical clustering analysis of key flavor compounds demonstrated that SF and All treatment groups formed a single cluster branch, confirming the absence of significant compositional divergence between these groups (Fig. 6B). These findings suggest that precise control of core microbial species and their intraspecific diversity can effectively mimic the key aroma chemical profiles of spontaneous fermentation, thereby providing the wine industry with an innovative fermentation strategy. In single-strain fermentations, SC23 demonstrated a marked promotion ability for the synthesis of ethyl acetate, 2-methoxy-4-ethylphenol, and isobutyric acid. SC1 exhibited high efficiency in the production of 2-methoxy-4-ethylphenol, while SC21 showed exceptional performance in the synthesis of 1-propanol. SB strains, on the other hand, displayed a clear advantage in the generation of phenol, phenethyl alcohol, and ethyl hexanoate. However, the fermentation characteristics of synthetic microbial communities are not merely the simple summation of the functional contributions of their constituent strains. These microorganisms form a complex network of interactions during fermentation, encompassing various ecological relationships such as mutualism, competition, antagonism, and metabolic cross-feeding. Through carefully designed species combinations, members of synthetic communities can establish mutually beneficial symbiotic mechanisms, such as metabolic cross-feeding, enabling them to collaboratively complete complex metabolic pathways that individual strains cannot achieve independently. This results in a significant enhancement of the synthetic capacity for specific flavor compounds. This knowledge gap constitutes a core bottleneck in this research field, constraining effective quality control and the achievement of industrialized production. Importantly, the present work is confined to laboratory scale, and scaling-up its conclusions to industrial fermenters remains a pivotal hurdle. Hence, forthcoming investigations will leverage transcriptomics to unravel the molecular basis of strain–strain interactions. Deciphering these mechanisms will furnish indispensable theoretical foundations and practical directives for advancing the controllability and consistency of industrial-scale Italian Riesling vinification. In summary, this study successfully constructed a synthetic microbial community whose fermentation product closely resembled the target Italian Riesling wine in terms of key aroma compound composition, odor activity values (OAV), and multivariate statistical models. This indicates that the synthetic community possesses the chemical foundation necessary to reproduce the target aroma profile. However, this similarity requires further confirmation through future sensory analyses, such as descriptive analysis and triangle tests.

It should be noted that the conclusions of this study are primarily based on inferences from chemical data. Although OAV and multivariate statistical analyses are valuable tools for predicting sensory contributions, the absence of formal sensory evaluation—such as tests by a trained panel, recombination, or omission studies—represents a limitation of this work. Therefore, the current findings should be interpreted as achieving a chemical profile approaching that of the target aroma. Validating the actual sensory similarity remains a key objective for future research.