The Daqu bricks were fermented in the Qu-producing shop of Jing Brand Maotai Town Liquor Co., Ltd., Renhuai, Guizhou province, China. Following fermentation, Daqu was categorized by color (black, yellow, and white) and moved to the storage room for maturation with a period of 180 days. Samples were collected at intervals of 0, 30, 60, 120, 150, and 180 days from each color of Daqu, resulting in 54 samples (three replicates per color) for subsequent analysis. Once the samples were gathered, the brick from each kind of Daqu was crushed to powder for the immediate assessment of physicochemical properties, cultivable microbes, and fermenting vitalities. The powder was stored at −80°C for subsequent flavor compounds analysis and amplicon sequencing.

The analysis of moisture, acidity, and reducing sugar content in three kinds of Daqu were conducted based on our previous study with minor revisions (Dong et al., 2022). The amino acid nitrogen content of Daqu was detected following the methodology outlined in Huang’s study (Huang et al., 2021). Specifically, moisture content was measured by taking 4 g of Daqu powder in an oven for desiccation (overnight) until a constant weight, and calculating the ratio of weight loss. Before the acidity assessing, 10 g of Daqu powder was added into 200 mL of ddH₂O with 30 min of static for acids extraction. Subsequently, the supernatant was used for acid–base titration to determine the acidity. The content of reducing sugar in Daqu was evaluated via Fehling’s test in reliance on the standard reducing sugar solution. As for the amino acid nitrogen, it was evaluated based on the amphoteric behavior of amino acids. Firstly, the amino acid nitrogen was extracted form Daqu by mixing 20 g of samples into 60 mL of ddH₂O together with vibration at 120 rpm for 30 min, and the filtrate was collected and used for detecting. Subsequently, the pH value of filtrate was adjusted to 8.2 with 50 mM NaOH, together with 10 mL of methanal being added to fix the amino base. Finally, the titration with standard NaOH solution was performed to determine the end-point (pH 9.2) via a pH meter, and thus calculating the amino acid nitrogen content by the consuming volume of standard NaOH solution.

The cultivable microbes in Daqu were quantified using traditional dilution and plating method, following our previous protocols with minor adjustments (Dong et al., 2022). Specifically, Bacillus, lactic acid bacteria (LABs), yeasts, and molds were targeted for detection. Ten grams of Daqu powder were combined with 90 mL of sterile H₂O and shaken at 160 rpm for 30 min. Subsequently, a gradient dilution was performed, and 100 μL of the diluted suspension was plated on respective agar plates to cultivate cultivable microbes. LB agar, YPD agar (with ampicillin), MRS agar (with nystatin), and PDA (with ampicillin) were employed to support the growth of Bacillus, LABs, yeasts, and molds, respectively, under appropriate conditions. Following the designated incubation period, the number of visible colonies on the respective plates was enumerated.

The fermenting vitalities of Daqu was monitored by detecting neutral protease activity, α-amylase activity, glucoamylase activity, and fermentation ability. Thereinto, the neutral protease activity was determined via colorimetry at 680 nm using Folin–Ciocalteu method based on our previous study (Dong et al., 2020). The α-amylase activity was detected based on iodine-starch colorimetric assay, and the decrease in absorbance at 620 nm was indicative of α-amylase activity. Glucoamylase activity was evaluated by determining the glucose produced from starch degradation via DNS (3,5-dinitrosalicylic acid) method. Additionally, fermentation ability was assessed by quantifying the weight of CO₂ produced during sorghum juice fermentation inoculated with Daqu powder.

The flavor substances in Daqu were detected by a head-space solid-phase microextraction and gas chromatography–mass spectrometry (HS-SPMEGC–MS) according to our previous study (Dong et al., 2022). In this process, 2.0 g of Daqu powder and 3.0 g NaCl were combined in 10 mL of 10% ethanol within a headspace bottle, then internal standard substances (2-octanol and 2-ethyl hexanol) were added. After that, the flavor substances in mixture were extracted by an automatic headspace sampling system (Multipurpose Sample MPS 2XL) at 50°C for 45 min. Subsequently, the SPME fiber was introduced into the injection port, set at 250°C, for a duration of 5 min. Compounds separation was achieved using an Agilent HP-5 column (30 m × 0.25 mm; 0.25 μm film thickness) and a DB-FFAP column (60 m × 0.25 mm; 0.25 μm film thickness), followed by analysis via GC–MS analysis employing the Agilent 7890B GC system and the 5977C mass selective detector. Compounds identification was carried out by comparing mass spectral profiles with a match quality of ≥80 in the NIST14 database.

The third-generation amplicon sequencing was employed to analyze the composition, structure, and succession of microbial community in three types of Daqu during maturation, specifically on day 0, 30, 90, and 180. Initially, total DNA was extracted from 0.2 grams of Daqu powder using the E.Z.N.A.^® Soil DNA Kit (Omega Bio-Tek, United States) as per the provided instructions. After the concentration and quality assessment of total DNA were conducted, it served as the template for library construction. For the full-length 16S rDNA library of bacteria, amplification was performed using Q5 high-fidelity DNA polymerase (NEB, United States) with primers 27F (AGRGTTTGATYNTGGCTCAG) and 1492R (TASGGHTACCTTGTTASGACTT) (Callahan et al., 2019). The amplification parameters were as follows: 98°C for 5 min, 98°C for 30 s, 55°C for 30 s, and 72°C for 45 s for 30 cycles, and final at 72°C for 5 min. Simultaneously, the full-length ITS library of fungi was constructed by Q5 high-fidelity DNA polymerase using primers ITS1-F (CTTGGTCATTTAGAGGAAGTAA) and ITS4 (TCCTCCGCTTATTGATATGC) (Kuang et al., 2021). The amplification parameters for this library were: 98°C for 5 min, 98°C for 30 s, 53°C for 30 s, and 72°C for 20 s for 30 cycles, with a final step at 72°C for 5 min. Subsequently, the purified PCR products of these two libraries were subjected to high-throughput sequencing (HTS) using the PacBio Sequel II platform. The post-sequencing data were analyzed bioinformatically using QIIME2, with the main analytical workflow referenced from Yang’s study (Yang et al., 2024).

The raw data generated from HTS has been securely archived in NCBI under the BioProject accession number PRJNA 1034761 and PRJNA 1034762. Statistical analyses in this study involved data processing with Excel (version 2019), and the determination of significance was conducted using Origin (version 9.0), with a significance threshold set at 0.05.

The physicochemical properties of the three types of Daqu during the maturation process were shown in Figure 2, emphasizing key parameters such as moisture content, acidity, reducing sugar, and amino acid nitrogen levels. A gradual reduction in moisture content was observed among the three types of Daqu, with levels decreasing from 8.09 to 7.08% for black Daqu, 8.43 to 7.20% for yellow Daqu, and 7.68 to 6.89% for white Daqu (Figure 2A). Notably, all three types reached a final moisture content below 9%, adhering to the standard for qualified Daqu, preserving enzymatic and microbial activities (Paredes-López et al., 1988; Xia et al., 2022). While the dropping trend of acidity during maturation was similar across the three colored Daqu, black Daqu exhibited the highest acidity, white Daqu the lowest, and yellow Daqu intermediate, consistent with findings in Zhang’s study (Zhang et al., 2022). At the end of the maturation period, the acidity values for these three types of Daqu were 1.03, 0.56, and 0.35 mmol/10 g, respectively (Figure 2B), which were lower than those in previous studies (Deng et al., 2020; Zhang et al., 2022). In terms of reducing sugar, initial content varied significantly among the Daqu types but showcased an overall dropping trend, stabilizing at a low level of approximately 1.40% by the end of maturation (Figure 2C). This low reducing sugar content is conducive to microbial stabilization but insufficient for microbial growth in Daqu. Amino acid nitrogen content showed an initial increase followed by a decrease, with black Daqu having the lowest final content (4.11 g/kg), white Daqu next (4.28 g/kg), and yellow Daqu having the highest content (4.78 g/kg) (Figure 2D).

The counts of cultivable microbes in the three-color Daqu manifested an overall raising trend followed by dwindling during the maturation process (Table 1), aligning with the previously noted decrease in moisture content (Figure 2A). In cases of Bacillus, the number ranged from 6.33 × 10⁵ to 9.03 × 10⁸ cfu/g, with white Daqu exhibiting the highest count, followed by yellow Daqu, and black Daqu (Table 1). This Bacillus count falls within a reasonable range reported in previous studies, where the maximal order of magnitudes for Bacillus is around 10⁸ in Daqu (Sakandar et al., 2020). The prevalence of Bacillus in cultivable microbes may be attributed to its ability to survive in harsh conditions (Liu et al., 2019; Sakandar et al., 2020). The number of LABs were lower (6.30 × 10⁵ to 6.21 × 10⁷ cfu/g) than those of Bacillus in the three Daqu and exhibited an opposite trend, with black Daqu having the highest LAB count, followed by yellow Daqu and white Daqu (Table 1). This finding aligns with previous research indicating that LABs contribute to substrate acidity, inhibiting the growth of Bacillus (Zou et al., 2018). The quantity of cultivable fungi was significantly lower than that of bacteria, with yeast counts ranging from 3.62 × 10² to 3.32 × 10⁵ cfu/g and mold counts varying from 7.66 × 10² to 3.53 × 10⁶ cfu/g (Table 1). Among these, white Daqu had the highest counts of yeast and mold, while black Daqu had the lowest counts, suggesting a synergistic effect of yeast and mold during the maturation process.

The fermenting vitalities of Daqu play a crucial role in the production of sauce-flavor baijiu, influencing the initiation of baijiu fermentation (Wang B. et al., 2018; Li H. et al., 2023). In this study, we conducted a comprehensive analysis of neutral protease, α-amylase, glucoamylase, and fermentation ability in the three types of Daqu to evaluate their fermenting vitalities. Neutral protease activity displayed an initial increase followed by a decrease in all three types of Daqu (Figure 3A). White Daqu exhibited the final activity at 80.61 U/g, followed by yellow Daqu at 48.51 U/g, and black Daqu at 9.98 U/g on day 180 of maturation (Figure 3A). This result is consistent with the changes in the counts of Bacillus mentioned earlier (Table 1), as Bacillus is recognized for its superior protease production (Contesini et al., 2017). The activities of α-amylase, glucoamylase, and fermentation vitalities all exhibited similar trends, with initial increases followed by a decline and subsequent stabilization (Figures 3B–D). Overall, white Daqu demonstrated the highest activity, followed by yellow Daqu, and black Daqu had the lowest (Figures 3B–D). These results align with the counts of yeast and mold in cultivable microbes (Table 1), indicating that a higher count of mold and yeast promotes the production of α-amylase and glucoamylase (Wang X. D. et al., 2018), ultimately resulting in stronger fermentation activity (Yang et al., 2022). In summary, the analysis of the fermenting vitalities of the three types of Daqu suggests that white Daqu has the highest quality and provides stronger fermentation capabilities, contributing to the initiation of sauce-flavor baijiu brewing.

The flavor substances in Daqu serve as the precursors to determine the final flavor and characteristics of baijiu (Li H. et al., 2023). A total of 160 different chemicals were detected, and the fluctuations in flavor substances during the maturation of the three types of Daqu are intricate. Here, we focused on analyzing alcohols, organic acids, esters, pyrazines, aldehydes and ketones (Figure 4). The content of alcohols in the colored Daqu all exhibited an overall decreasing trend throughout the maturation process (Figure 4A). Yellow Daqu initially had the highest alcohols content (79.79%) on day 0, dropping to 30.36% on day 180, while black Daqu started with 44.82%, decreasing to 24.33% at maturity (Figure 4A). In contrast, white Daqu showed the lowest initial alcohols content of 1.97%, declining to 0.68% on day 180 (Figure 4A). Organic acids content also decreased during maturation in these three types of Daqu, albeit with a milder decline than the variation observed in alcohols, ranging from 1.60 to 3.75% (Figure 4B). Ultimately, black Daqu had the highest organic acids content (1.85%), followed by yellow Daqu (0.73%), and white Daqu (0.54%) at the end of maturation (Figure 4B), in line with changes in acidity mentioned earlier (Figure 2B). Esters content varied significantly among the three types of Daqu, with white Daqu having the highest overall content, fluctuating throughout maturation, while black Daqu and yellow Daqu showed a similar upward trend (Figure 4C). Thus, the decline in alcohols and organic acids observed here can be attributed to their substantial consumption in the formation of esters. Pyrazine substances, crucial flavor compounds in sauce-flavor baijiu, contribute to pleasant aromas of roasted nuts and cocoa (Mortzfeld et al., 2020; Shi X. et al., 2022). They were considered one type of the characteristic flavor components that differentiate sauce-flavor baijiu from other types of baijiu (Zhang et al., 2013; Niu et al., 2022). The content of pyrazines in the three-color Daqu showed big differences (Figure 4D). White Daqu and black Daqu maintained high levels of pyrazines (>10%), with white Daqu reaching 19.18% and black Daqu at 12.37% on day 180 (Figure 4D). In contrast, the content of pyrazines in yellow Daqu remained at low level, reaching 4.53% on day 180 (Figure 4D). Aldehydes and ketones exhibited an increasing trend in both black Daqu and yellow Daqu, reaching 9.17 and 10.96% at the end of maturation, respectively, whereas white Daqu experienced a sharp increase to 13.04% during day 60–120, followed by a rapid decrease to 4.45% in the end (Figure 4E). Other flavor substances exhibited varying changes among the three types of Daqu (Figure 4F). In summary, the diverse changes in flavor substances during the maturation process of the three types of Daqu indicate each type’s unique characteristics, suggesting potential combinations based on differences in flavor substances for sauce-flavor baijiu brewing.

Through third-generation HTS, we analyzed the microbial community of the three-color Daqu during maturation. The results indicated that the sequencing coverage exceeded 0.991 (Table 2), demonstrating the depth and reliability of the amplicon sequencing. Regarding bacterial α diversity, the Chao1 and Shannon indices for the three types of Daqu generally exhibited a trend of initial decrease followed by an increase as they mature (Table 2). Overall, the Chao1 richness was highest in black Daqu, followed by white Daqu, and lowest in yellow Daqu. The Shannon index was higher in black and white Daqu compared to yellow Daqu. For fungal α diversity, the Chao1 richness decreased for all three types of Daqu, indicating a decline in fungal richness during Daqu maturation. This decrease may be associated with the decline in moisture content during Daqu maturation since the lower moisture levels, below 9%, can reduce free water in microbial cells, hampering their growth and reproduction. In contrast, the fungal Shannon index remained relatively stable, with white Daqu exhibiting the highest value, suggesting a more even distribution of fungi during maturation. In summary, the primary diversity differences among the three types of Daqu are reflected in the Chao1 richness, with less variation in the Shannon index (Table 2).

To further assess the structure of microbial community during Daqu maturation, a Principal Component Analysis (PCA) based on Bray-Curtis distances was conducted. The β diversity results revealed that the structural differences in bacterial community among the three types of Daqu were more pronounced than for fungal community. In the bacterial β diversity plot, samples representing the three types of Daqu were more scattered, indicating greater differences (Supplementary Figure S1A). In contrast, in the fungal β diversity plot, samples from all three Daqu types clustered closely and were not easily distinguishable (Supplementary Figure S1B). This implies that the differences in microbial diversity mainly exist in the bacterial community rather than the fungal community.

To investigate the composition of microbial community in Daqu, differences in taxa at the phylum and genus levels were analyzed during the maturation of the three-color Daqu. For bacterial composition, Firmicutes, Actinobacteriota, Bacteroidota, Proteobacteria, Fusobacteriota, and Verrucomicrobiota constituted the predominant phyla (Figure 5A). Among them, Firmicutes emerged as the most abundant phylum, which was congruent with Gan’s study (Gan et al., 2019). However, the abundance of Firmicutes varied among three kinds of Daqu. In black Daqu, there was an initial increase followed by decrease (49.59 to 88.80 to 32.17%) (Figure 5A). Yellow Daqu exhibited a minor decline from 97.61 to 91.07%, while Firmicutes remained relatively stable in white Daqu, with abundances consistently above 80.52% (Figure 5A). As for fungal composition, the predominant phyla included Ascomycota, Mucoromycota, Basidiomycota, Mortierellomycota, Chytridiomycota, and Rozellomycota (Figure 5B). Ascomycota was the most dominant phylum, collaborating with previous studies (Deng et al., 2020; Zhang et al., 2022). The abundance of Ascomycota in black Daqu maintained above 94.29%, yellow Daqu exhibited fluctuations ranging from 80.51 to 99.77%, and white Daqu displayed an initial increase followed by a decrease (80.03 to 97.43 to 74.23%) (Figure 5B).

At the genus level, the composition of microbial community in the three-color Daqu showed even greater differences (Figures 5C,D). For bacterial, Virgibacillus, Kroppenstedtia, Bacillus, Scopulibacillus, Candidatus Sulcia, Oceanobacillus, Enterobacter, Saccharopolyspora, Pediococcus, and Staphylococcus were the top 10 abundant genera (Figure 5C). However, their relative abundances in the microbial community changed significantly during Daqu maturation. In black Daqu, Kroppenstedtia (31.58%), Candidatus Sulcia (17.49%), and Virgibacillus (10.27%) constituted the top 3 dominant bacterial genera (Figure 5C). Kroppenstedtia, in particular, was the most dominated genera in black Daqu, and its relative abundance rapidly increased from 6.48% (day 0) to 68.20% (day 30), followed by a continue decrease to 3.11% on day 180 (Figure 5C). Previous studies have unveiled that Kroppenstedtia is the core microbe in connection with the biosynthesis of organic acids (lactic acid and short chain fatty acids) (Zhu C. et al., 2023; Zhang et al., 2024), explaining the highest acidity observed in black Daqu during maturation. In yellow Daqu, Virgibacillus (54.66%), Kroppenstedtia (21.47%), and Oceanobacillus (8.57%) made up the top 3 dominant bacterial genera (Figure 5C). Virgibacillus was the main dominant genus in yellow Daqu, with its relative abundance quickly rosing from 54.83% (day 0) to 71.45% (day 30) and then gradually dropping to 33.67% (day 180) (Figure 5C). This result differs from previous studies where Virgibacillus was a dominant bacterium but not the most dominant in Daqu (Zhu C. et al., 2023; Zhu Q. et al., 2023). The roles and functions of Virgibacillus in Daqu production and baijiu brewing are not clear and needs further study. In white Daqu, Bacillus (35.53%), Virgibacillus (22.13%), and Scopulibacillus (10.01%) were the top three dominant bacterial genera (Figure 5C). Bacillus was the most dominant genus, and its relative abundance gradually decreased during maturation (from 49.44 to 25.86%) (Figure 5C). Despite the decreasing trend, the abundance of Bacillus remained above 25%, suggesting that it could still function effectively. Bacillus is known for secreting amylase and proteases and playing essential roles in liquefaction and saccharification (Ding et al., 2013; Wang et al., 2014, 2020), contributing to the higher fermenting vitalities observed in white Daqu (Figure 3). In addition, the average abundance of Kroppenstedtia in three kinds of Daqu followed the order: black Daqu, yellow Daqu, and white Daqu. Kroppenstedtia was found to be positively correlated with the production of various amino acids in Daqu (Zhu C. et al., 2023), which might promote the Maillard reaction with reducing sugar, leading to the formation of the yellow-brown color and contributed to the color difference in these three-color Daqu. Therefore, Kroppenstedtia, Virgibacillus, and Bacillus were identified as the three core bacteria in these colored Daqu, with Kroppenstedtia being the potential key microbe related to color formation.

For fungal genera, differences were smaller than those of bacteria among the three types of Daqu. The top 10 abundant fungal genera included Thermoascus, Thermomyces, Aspergillus, Rasamsonia, Rhizomucor, Pichia, Candida, Mortierella, Gelasinospora, and Monascus (Figure 5D). Among them, Thermoascus was the most dominant genera shared in all three types of Daqu, which was similar to Zhu’s study (Zhu C. et al., 2022). The relative abundance of Thermoascus changed in the same manner, initially decreasing, then increasing, and finally decreasing again (Figure 5D). In black Daqu, Thermoascus (57.38%), Aspergillus (17.18%), and Thermomyces (12.48%) constituted the top 3 dominant fungal genera, and the relative abundance of Thermoascus ranged from 35.92 to 92.58% (Figure 5D). In yellow Daqu, the top 3 dominant fungal genera belonged to Thermoascus (59.84%), Rasamsonia (6.85%), and Thermomyces (6.08%), whereas the abundance of Thermoascus varied from 36.51 to 73.91% (Figure 5D). In white Daqu, Thermoascus (39.35%), Thermomyces (25.01%), and Rhizomucor (7.50%) comprised the top 3 fungal genera, while the abundance of Thermoascus fluctuated from 6.44 to 65.88% (Figure 5D). Here, Thermoascus is known to produce various enzymes such as catalase, endoglucanase, glucosidase, keratinase, and chitinase, which are able to degrade starch, cellulose or protein from raw materials, providing basic nutrition for other microbes’ metabolism that contributed to flavor formation (Jain et al., 2014; Cai et al., 2021). Thus, Thermoascus was identified as the most important fungal genera during Daqu maturation.

Physicochemical factors play a crucial role in shaping and influencing the composition and succession of microbial community during spontaneously fermentation processes (Guan et al., 2020). Therefore, canonical correlation analysis (CCA) was employed to examine the impact of physicochemical factors on the microbial community of the three types of Daqu. The results, as shown in Figure 6, indicated that acidity had the most substantial impact on microbial community, as the arrow length representing acidity was the longest among the four physicochemical factors. Regarding bacterial communities, the influence of the four physicochemical factors was as follows: acidity > reducing sugar > moisture > amino acid nitrogen (Figure 6A). Specifically, the dominant bacterium Kroppenstedia in black Daqu showed a positive correlation with acidity, reducing sugar, and moisture, while it had a negative correlation with amino acid nitrogen (Figure 6A). Kroppenstedtia is known to correlate with the biosynthesis of organic acids, evidencing its positive relationship with acidity (Zhang et al., 2024). Its negative correlation with amino acid nitrogen contradicted a previous study that found Kroppenstedtia to be positively connected to the production of various amino acids in Daqu (Zhu C. et al., 2023). This discrepancy might be caused by the color formation in Daqu, where amino acids were consumed via the Maillard reaction to form the dark color. In yellow Daqu, the dominant bacterium Virgibacillus displayed a positive correlation with reducing sugar, moisture, and amino acid nitrogen, but a negative correlation with acidity (Figure 6A). In white Daqu, the dominant bacterium Bacillus exhibited a positive correlation with amino acid nitrogen but a negative correlation with the other three physicochemical properties (Figure 6A). Here, Bacillus is the typical microbe for producing protease (Wang et al., 2014; Contesini et al., 2017), thus hydrolyzing protein to amino acids or small peptides, interpretating this positive correlation. When considering fungal communities, the influence of these four physicochemical factors was as follows: acidity > amino acid nitrogen > moisture > reducing sugar (Figure 6B). The dominant fungus Thermoascus in all three types of Daqu shared a positive correlation with all four physicochemical factors (Figure 6B), indicating Thermoascus’ ability to adapt to the maturation process of Daqu. Therefore, acidity was the most critical factor affecting the microbial community of these three kinds of Daqu during maturation.

During the maturation of Daqu, microbes utilize substrates such as starch, protein, and others, producing a plethora of flavor compounds and various enzymes. These, in turn, have a profound impact on the subsequent production of baijiu, making microbes pivotal in the maturation of Daqu and significantly influencing the quality of the final product. To explore these connections, we employed Spearman correlation analysis to unravel the relationships between dominant microbes and flavor compounds, as well as fermenting vitalities. The results, presented in Figure 7, delineated the influence of dominant bacteria and fungi on flavor compounds and fermenting vitalities. In black Daqu, the dominant bacterium Kroppenstedia exhibited a substantial positive correlation with acids and pyrazines, along with a significant negative correlation with amylase, fermenting vitality, and esters (Figure 7A). These correlations may be attributed to the unique characteristics of Kroppenstedia, as reported previously. Kroppenstedtia was identified as a core microbe related to the production of organic acids (lactic acid and short chain fatty acids) (Zhu C. et al., 2023; Zhang et al., 2024), and positively correlated with the formation of certain pyrazines such as tetramethylpyrazine and 2,3,5-trimethyl pyrazine (Zhang et al., 2021). Within yellow Daqu, the dominant bacterium Virgibacillus demonstrated a positive correlation with neutral protease and alcohols, while displaying a negative correlation with pyrazines (Figure 7A). However, the functions of Virgibacillus in fermented food is not clear and warrant further investigation. In white Daqu, the dominant bacterium Bacillus displayed a positive correlation with all enzymes, fermenting vitality, and pyrazines, while holding a negative correlation with alcohols (Figure 7A). Bacillus is known to produce various hydrolytic enzymes, including amylase, protease, and lipase, to degrade macromolecules and hence promoting the formation of flavor compounds during baijiu brewing (Jiang et al., 2021; Li Z. et al., 2023). Especially, the pyrazine compounds, providing sauce-flavor baijiu with its unique flavor characteristic, were closely linked to Bacillus (He et al., 2019). Considering fungi, the dominant fungus Thermoascus across all three types of Daqu showed a positive correlation with amylase, glucoamylase, alcohols, and esters (Figure 7B), hinting that Thermoascus could promote the formation of flavor by producing various enzymes and collaborating with others functional microbes (Jain et al., 2014; Cai et al., 2021). Notably, the formation of pyrazine substances, characteristic of sauce-flavor baijiu, exhibited a significant positive correlation with Aspergillus, Rhizomucor and Rhizopus (Figure 7B), implying that these fungi also played a crucial role in the formation of pyrazines. This conclusion complements the prevailing belief that Bacillus is the core microbe responsible for pyrazine formation, indicating the substantial contributions of molds to this process.