Microbial Diversity and Volatile Flavor Compounds in Tibetan Flavor Daqu

In this study, the microbial diversity in Tibetan flavor Daqu was analyzed based on single molecule real-time sequencing (SMRT). The volatile flavor compounds in Daqu were detected using the headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC-MS). In addition, the correlation between various microbes and volatile flavor compounds was explored. Our studies indicated that the dominant bacterial genera in Tibetan flavor Daqu were Oceanobacillus, Kroppenstedtia, Virgibacillus, Enterococcus, Pediococcus, Streptomyces, Saccharopolyspora, Leuconostoc, uncultured_bacterium_f_Lachnospiraceae and Lactobacillus. The dominant fungal genera were Wickerhamomyces, Monascus, Aspergillus and Rhizomucor. 101 volatile compounds were detected in the Daqu samples, including alcohols, acids, esters, aldehydes, hydrocarbons, ketones, ethers, aromatics and pyrazines, and 10 key flavor compounds were identified using the relative odor activity value (ROAV). The results of our correlation analysis showed that Enterococcus was mainly associated with the synthesis of aldehydes such as trans-2-octenal, and uncultured_bacterium_f_lachnospiraceae was associated with the synthesis of most aldehydes. This paper has systematically investigated the physicochemical indices, microbial community structure and flavor compounds of Tibetan flavor Daqu, which is helpful in gaining a deeper understanding of the characteristics of Tibetan flavor Daqu.


Introduction
Liquor is a distilled spirit unique to China. There are many types of liquor, which include three main categories: Maotaiflavor liquor, Luzhou flavor liquor and Light flavor liquor [1]. In addition, there are some less well-known liquors that are also of excellent quality, such as Tibetan flavor liquor (TF liquor). TF liquor is a distilled liquor produced via the saccharification and fermentation of Daqu under certain conditions, and it is favored by Tibetans. TF liquor is pure with a natural and coordinated compound aroma dominated by ethyl caproate and supplemented by ethyl acetate. The taste of TF liquor is sweet and refreshing, and it is mellow and has a long remaining taste [2].
During the fermentation process of liquor, Jiuqu, also known as Daqu, plays a decisive role [1]. Daqu is the saccharifying and fermenting agent of liquor, which directly affects the flavor, quality and taste of liquor. Daqu is divided into many types according to different flavors, such as Maotai flavor Daqu, Luzhou flavor Daqu and Light flavor Daqu [1,2]. Tibetan flavor liquor is fermented with Tibetan flavor daqu (TF Daqu). TF Daqu is a medium-temperature Daqu; the process diagram of its production is shown in Figure 1. TF Daqu is made of high-quality barley, wheat, highland barley, peas and maize with crushed Tibetan medicinal materials. It is made via the traditional TF Daqu adobe house cultivation process and the top culture temperature is 35-48 • C. Its stubble is clear, hard and tidy with an intense fragrance of Daqu and Tibetan medicinal materials, and the storage period is over 4 months [2]. Because Daqu is naturally inoculated and produced via an open fermentation process, the microorganisms in Daqu are very complex [1,2]. To date, there have been many research studies on the functions and types of microorganisms in Daqu. In general, the microorganisms in Daqu can be mainly divided into four categories: bacteria, molds, yeasts and actinomycetes [3]. Bacillus, one of the most representative bacteria, can secrete a variety of degrading enzymes, such as protease, amylase and glycosylase, which will decompose proteins, starch and other macromolecular substances during the fermentation of liquor to produce a variety of flavor substances [3]. Lactic acid bacteria can not only synthesize extracellular polysaccharides and oligosaccharides, but also produce the aroma substances of liquor via the Maillard reaction and produce a large amount of lactic acid, which can enhance the sweetness and mellowness of the wine at the same time [4]. Aspergillus in Daqu can produce enzymes related to starch saccharification and protein hydrolysis, thus affecting the flavor of liquor, while playing a vital role in the saccharification and esterification abilities of Daqu during its fermentation [5]. Rhizopus can produce organic acids, including succinic acid, lactic acid and fumaric acid, which are crucial in the formation of the characteristic taste of liquor [6]. Yeast in Daqu is the primary microbe in the brewing process. During the manufacturing of liquor, Saccharomyces cerevisiae transforms glucose into alcohol and produces esters, higher alcohols and organic acids [6]. In addition, Daqu is also known as the aroma-producing agent of liquor. During the Daqu manufacturing process, the metabolites of the microorganisms and decomposition products of the raw materials directly or indirectly contribute to the flavor substances in liquor, giving it a variety of unique flavors [7]. At present, many studies have introduced common Daqu in detail, including Maotai flavor Daqu, Luzhou flavor Daqu and Light flavor Daqu [6][7][8], but there are few reports on TF Daqu. Therefore, exploring the microbial diversity and volatile flavor compounds of TF Daqu can help to enhance the quality of TF Daqu, which is of far-reaching significance in the brewing of liquor. Because Daqu is naturally inoculated and produced via an open fermentation process, the microorganisms in Daqu are very complex [1,2]. To date, there have been many research studies on the functions and types of microorganisms in Daqu. In general, the microorganisms in Daqu can be mainly divided into four categories: bacteria, molds, yeasts and actinomycetes [3]. Bacillus, one of the most representative bacteria, can secrete a variety of degrading enzymes, such as protease, amylase and glycosylase, which will decompose proteins, starch and other macromolecular substances during the fermentation of liquor to produce a variety of flavor substances [3]. Lactic acid bacteria can not only synthesize extracellular polysaccharides and oligosaccharides, but also produce the aroma substances of liquor via the Maillard reaction and produce a large amount of lactic acid, which can enhance the sweetness and mellowness of the wine at the same time [4]. Aspergillus in Daqu can produce enzymes related to starch saccharification and protein hydrolysis, thus affecting the flavor of liquor, while playing a vital role in the saccharification and esterification abilities of Daqu during its fermentation [5]. Rhizopus can produce organic acids, including succinic acid, lactic acid and fumaric acid, which are crucial in the formation of the characteristic taste of liquor [6]. Yeast in Daqu is the primary microbe in the brewing process. During the manufacturing of liquor, Saccharomyces cerevisiae transforms glucose into alcohol and produces esters, higher alcohols and organic acids [6]. In addition, Daqu is also known as the aroma-producing agent of liquor. During the Daqu manufacturing process, the metabolites of the microorganisms and decomposition products of the raw materials directly or indirectly contribute to the flavor substances in liquor, giving it a variety of unique flavors [7]. At present, many studies have introduced common Daqu in detail, including Maotai flavor Daqu, Luzhou flavor Daqu and Light flavor Daqu [6][7][8], but there are few reports on TF Daqu. Therefore, exploring the microbial diversity and volatile flavor compounds of TF Daqu can help to enhance the quality of TF Daqu, which is of far-reaching significance in the brewing of liquor.
In this study, SMRT (single molecule real-time) and HS-SPME-GC-MS (headspace solid-phase microextraction-gas chromatography-mass spectrometry) were used to detect the microbial diversity and volatile flavor compounds in TF Daqu, respectively. The correlation between the microbial diversity and volatile flavor compounds was also analyzed. The results of the present study supply some theoretical foundation for the quality im- provement of TF Daqu and provides theoretical support for further optimization of the microorganisms in liquor and product quality control.

Sample Collection and Pretreatment
The TF Daqu samples were collected from the Tibetan flavor liquor distillery in Tianzhu County, Gansu Province. TF Daqu blocks were randomly selected from five different batches. Samples from the same batch were crushed and mixed to form a mixed sample. Thus, a total of 5 mixed samples were formed and numbered ZQ1, ZQ2, ZQ3, ZQ4 and ZQ5. The samples were preserved at 4 • C to detect the flavor compounds, and stored at −80 • C for analysis of the microbial diversity.

DNA Extraction, PCR Amplification and High-Throughput Sequencing
A Power Soil DNA isolation kit was used to extract DNA from Daqu. The extracted DNA was then determined for its purity, concentration and integrity. To analyze the bacteria in the samples, the full-length 16S rRNA gene was amplified with 27F(5 -AGRGTTTGATYNTGGCTCAG-3 )/1492R(5 -TASGGHTACCTTGTTASGACTT-3 )primers. The amplification procedure was as follows: pre-denaturation at 95 • C for 2 min; denaturation at 98 • C for 10 s, annealing at 55 • C for 30 s, extension at 72 • C for 1.5 min, a total of 25 cycles; and extension at 72 • C for 2 min. ITS1F(5 -CTTGGTCATTTAGAGGAAGTAA-3 )/ITS4R(5 -TCCTCCGCTTATTGATATGC-3 )primers amplified the full-length ITS sequences of Daqu for fungi. The amplification procedure was as follows: pre-denaturation at 95 • C for 5 min, 8 cycles (denaturation at 95 • C for 30 s, annealing at 55 • C for 30 s, extension at 72 • C for 45 s), 24 cycles (denaturation at 95 • C for 30 s, annealing at 60 • C for 30 s, extension at 72 • C for 45 s), and extension at 72 • C for 5 min. The products were purified, quantified and homogenized, and the resulting library was established and quality checked. Finally, the qualified libraries were tested on the PacBio sequencing platform.

Analysis of the Microbial Diversity
The original sequences were derived and the effective sequences were obtained after identifying, filtering and removing the chimeras. The effective sequences were clustered according to their similarity and the OTU (operational taxonomic units) were divided. According to the results of our OTU analysis, the samples were analyzed taxonomically at various taxonomic levels. Alpha diversity analysis, which includes the Simpson, Shannon and Chao1 indices of the samples, was used to study the species diversity and richness. The microbial community structure was analyzed using the relative abundance of species at the phylum and genus level.

Analysis of the Volatile Flavor Compounds in Daqu
Using HS-SPME-GC-MS SPME fiber (50/30 µmDVB/CAR/PDMS; Supelco, Bellefonte, PA, USA) was used to sample volatile flavor compounds. A total of 3.00 g of the crushed Daqu sample was placed in a 15 mL headspace bottle and subsequently sealed. The bottle was then placed in a thermostat water bath heated at 60 • C for 15 min and the SPME fiber was inserted at 60 • C for 50 min to carry out the extraction. The extraction head was inserted into the GC-MS inlet and desorbed at 250 • C for 3 min for GC-MS analysis.
The volatile flavor compounds were detected using GC-MS (Trace GC 1310-ISQ mass spectrometer; Thermo Scientific, Austin, TX, USA). The GC-MS conditions used were in accordance with the method reported in a previous study with some slight modification [8]. A DB-5MS flexible quartz capillary column (30 m × 0.25 mm × 0.25 µm) was applied. The heating procedure was carried out as follows: The temperature was initially set at 40 • C, then increased to 150 • C at a rate of 5 • C/min for 3 min, and then heated to 250 • C at a rate of 5 • C/min for 2 min. High-purity helium at a flow rate of 1.0 mL/min was used as the carrier gas. The MS conditions were as follows: An EI source served as the ion source and its temperature was 230 • C, the temperature of the connecting port was 280 • C and the electron energy was 70 eV. The quality scanning range was 40-450 amu.
The obtained spectra were searched and analyzed using the Wiley spectral library provided with the instrument and the peaks with similarity of <80% and siloxane-type impurities were screened and removed, and the volatile compounds in the TF Daqu samples identified. The peak-area normalization method was used to determine the relative content of the compounds. The different flavor compounds were analyzed and compared using software such as Excel.

Identification of the Key Flavor Compounds in Daqu Using ROAV (Relative Odor Activity Value)
Referring to the method described by Cai et al. [9], ROAV was applied to determine the key flavor compounds in the Tibetan Daqu samples. The ROAV max was defined as 100 for the volatile compound with the largest contribution to the aroma of the samples and the ROAVs of the other volatile compounds (A) were determined using the following formula: where C%A and C%max denote the relative content of each aromatic component and the aromatic component with the largest contribution to the aroma, respectively. C%A and C%max were calculated by GC-MS. TA and Tmax denote the odor threshold of each aromatic component and the aromatic component with the largest contribution to the aroma, respectively. TA and Tmax were obtained by querying the odor threshold table.

Correlation Analysis of the Microbial Community Diversity and Flavor Compounds in TF Daqu
SPSS software was used to calculate correlation coefficients and p-values. The Spearman correlation coefficients and p-values were used to study the correlation between the dominant flora and key flavor compounds in Daqu. A significant correlation was defined as a correlation coefficient >0.6 and p < 0.05.

Sequencing Results and Alpha Diversity
The sample sequences were processed using statistical methods to obtain high-quality sequences for subsequent analysis. As shown in Table 1, the number of effective bacterial sequences in the five TF Daqu samples ranged from 5150 to 6321 and the number of effective sequences of fungi ranged from 6069 to 7530. In addition, the proportion of bacterial effective sequences exceeded 80% and the proportion of fungal effective sequences exceeded 90% in all of the Daqu samples. The alpha diversity reflects the species diversity and richness of single samples. Among them, the Chao1 index is applied as a measure of species richness and the Simp- son and Shannon index reflects the species diversity [10]. In addition, the coverage was calculated; the larger the coverage value, the higher the probability that the species will be detected in the samples. The alpha diversity indexes are shown in Table 1.
Among the five TF Daqu samples studied, the Chao1 index of the bacteria ranged from 37.429 to 95.375, the Simpson index was between 0.739 and 0.943, and the Shannon index was between 2.726 and 5.090. The Chao1 index, Simpson index and Shannon index of ZQ1 and ZQ5 were relatively high, indicating that the bacterial community richness and diversity were higher than those of the other samples studied. For fungi, the Chao1 index ranged from 14.000 to 48.091, the Simpson index was between 0.060 and 0.214, and the Shannon index was between 0.277 and 0.821. The Chao1 index, Simpson and Shannon index of ZQ3 were relatively high, indicating that the diversity and richness of fungal communities were higher than those of the other samples studied. The coverage of the sequencing for the five Daqu samples were all over 99%, meaning that the amount of data for this sequencing was sufficient.

Microbial Community Structure of TF Daqu
The sequencing results show that there were 16 phyla, 158 genera and 171 species of bacteria in the five TF Daqu samples. In addition, there were 5 phyla, 35 genera and 39 species of fungi in the samples (Table S2).
The relative abundance of the bacterial communities was analyzed at the phylum level ( Figure 2A). Firmicutes was predominant in the five Daqu samples, accounting for >50% of the total, followed by Proteobacteria and Actinomycetes. This is consistent with the findings reported in previous studies. Xie et al. [11] studied the bacteria community and dynamic succession of sesame flavor liquor Daqu and their findings indicated that the most predominant flora was Firmicutes. In addition, Tian et al. [12] reported that Firmicutes was the most important phylum using a metagenomics-based study on the diversity of bacterial communities in Shilixiang liquor Daqu. Among the five TF Daqu samples studied, the Chao1 index of the bacteria ranged from 37.429 to 95.375, the Simpson index was between 0.739 and 0.943, and the Shannon index was between 2.726 and 5.090. The Chao1 index, Simpson index and Shannon index of ZQ1 and ZQ5 were relatively high, indicating that the bacterial community richness and diversity were higher than those of the other samples studied. For fungi, the Chao1 index ranged from 14.000 to 48.091, the Simpson index was between 0.060 and 0.214, and the Shannon index was between 0.277 and 0.821. The Chao1 index, Simpson and Shannon index of ZQ3 were relatively high, indicating that the diversity and richness of fungal communities were higher than those of the other samples studied. The coverage of the sequencing for the five Daqu samples were all over 99%, meaning that the amount of data for this sequencing was sufficient.

Microbial Community Structure of TF Daqu
The sequencing results show that there were 16 phyla, 158 genera and 171 species of bacteria in the five TF Daqu samples. In addition, there were 5 phyla, 35 genera and 39 species of fungi in the samples (Table S2).
The relative abundance of the bacterial communities was analyzed at the phylum level ( Figure 2A). Firmicutes was predominant in the five Daqu samples, accounting for >50% of the total, followed by Proteobacteria and Actinomycetes. This is consistent with the findings reported in previous studies. Xie et al. [11] studied the bacteria community and dynamic succession of sesame flavor liquor Daqu and their findings indicated that the most predominant flora was Firmicutes. In addition, Tian et al. [12] reported that Firmicutes was the most important phylum using a metagenomics-based study on the diversity of bacterial communities in Shilixiang liquor Daqu. The relative abundance of the fungal communities was analyzed at the phylum level ( Figure 2B). Ascomycota was the main phylum among the five TF Daqu samples, its relative abundance ranging from 97.07 to 99.35%. This is consistent with the findings of Jiang et al. [13] during their study on the microbial flora and dynamic succession during the manufacture of Northern Jiang flavored Daqu liquor, which reported that the main fungus was also Ascomycota. The relative abundance of the bacterial communities was analyzed at the genus level ( Figure 3A). It is considered that the dominant genus is when the relative abundance is greater than 1%. In sample ZQ1, the dominant bacteria were Oceanobacillus, Kroppenstedtia, Virgibacillus, Enterococcus and uncultured_bacterium_f_Lachnospiraceae. In the samples ZQ2, ZQ3 and ZQ4, the dominant bacteria were Oceanobacillus, Kroppenstedtia, Virgibacillus, Enterococcus, Pediococcus, Streptomyces, Saccharopolyspora, Leuconostoc and Lactobacillus. In sample ZQ5, the dominant bacteria were Oceanobacillus, Kroppenstedtia, Enterococcus and uncul-tured_bacterium_f_Lachnospiraceae. Among them, the relative abundance of Oceanobacillus in ZQ4 was 45.71%, while it was only 15.60% in ZQ5. The relative abundance of Virgibacillus in ZQ3 was 13.92%, while it was only 0.60% in ZQ5. It can be seen that although the TF Daqu samples contain a large number of bacterial communities, the distribution of the dominant bacteria in each sample was not uniform. Chen et al. [14] found that the dominant bacterial groups in the middle of special flavor liquor Daqu were Oceanobacillus, Kroppenstedtia, Lactobacillus and Bacillus. During analysis of the microbial diversity in various types of high-temperature Daqu, Wang et al. [4] found that the dominant bacteria were composed of Bacillus, Brevibacterium, Kroppenstedtia, Lentibacillus, Staphylococcus, Saccharopolyspora, Streptomyces and Thermoactinomycetes. Cai et al. [15] found that the core bacterial flora of low-temperature Daqu was dominated by Lactobacillus, together with Saccharomyces, Bacillus and Streptomyces.
The relative abundance of the fungal communities was analyzed at the phylum level ( Figure 2B). Ascomycota was the main phylum among the five TF Daqu samples, its relative abundance ranging from 97.07 to 99.35%. This is consistent with the findings of Jiang et al. [13] during their study on the microbial flora and dynamic succession during the manufacture of Northern Jiang flavored Daqu liquor, which reported that the main fungus was also Ascomycota.
The relative abundance of the bacterial communities was analyzed at the genus level ( Figure 3A). It is considered that the dominant genus is when the relative abundance is greater than 1%. In sample ZQ1, the dominant bacteria were Oceanobacillus, Kroppenstedtia, Virgibacillus, Enterococcus and uncultured_bacterium_f_Lachnospiraceae. In the samples ZQ2, ZQ3 and ZQ4, the dominant bacteria were Oceanobacillus, Kroppenstedtia, Virgibacillus, Enterococcus, Pediococcus, Streptomyces, Saccharopolyspora, Leuconostoc and Lactobacillus. In sample ZQ5, the dominant bacteria were Oceanobacillus, Kroppenstedtia, Enterococcus and uncultured_bacterium_f_Lachnospiraceae. Among them, the relative abundance of Oceanobacillus in ZQ4 was 45.71%, while it was only 15.60% in ZQ5. The relative abundance of Virgibacillus in ZQ3 was 13.92%, while it was only 0.60% in ZQ5. It can be seen that although the TF Daqu samples contain a large number of bacterial communities, the distribution of the dominant bacteria in each sample was not uniform. Chen et al. [14] found that the dominant bacterial groups in the middle of special flavor liquor Daqu were Oceanobacillus, Kroppenstedtia, Lactobacillus and Bacillus. During analysis of the microbial diversity in various types of high-temperature Daqu, Wang et al. [4] found that the dominant bacteria were composed of Bacillus, Brevibacterium, Kroppenstedtia, Lentibacillus, Staphylococcus, Saccharopolyspora, Streptomyces and Thermoactinomycetes. Cai et al. [15] found that the core bacterial flora of low-temperature Daqu was dominated by Lactobacillus, together with Saccharomyces, Bacillus and Streptomyces. The relative abundance of fungal communities was analyzed at the genus level (Figure 3B). The unclassified fungi were not discussed. In sample ZQ1, ZQ3 and ZQ4, the dominant fungi were Wickerhamomyces, Monascus, Aspergillus and Rhizomucor, and in sample ZQ5, the dominant fungi was Wickerhamomyces. Hui et al. [16] used SMRT to identify The relative abundance of fungal communities was analyzed at the genus level ( Figure 3B). The unclassified fungi were not discussed. In sample ZQ1, ZQ3 and ZQ4, the dominant fungi were Wickerhamomyces, Monascus, Aspergillus and Rhizomucor, and in sample ZQ5, the dominant fungi was Wickerhamomyces. Hui et al. [16] used SMRT to identify the microbial characteristics of koji and reported that the main fungal genus was Wickerhamomyces. Aspergillus and Rhizoctonia were found to be the dominant genera in Maotai flavor Daqu [17].
Since TF Daqu is a medium-temperature Daqu, its fermentation temperature is~30-50 • C. This environment is suitable for most microorganisms to survive, so the microorganism communities are more abundant. The findings of this study suggest that the number of bacteria in Daqu was more abundant than that of fungi, which was consistent with the results of previous studies [18]. Although the microorganisms in TF Daqu have been reported in previous Daqu liquors, there are some differences between the microbial composition of each Daqu sample when compared with previous studies, which may be caused by the differences in the raw materials, techniques and environment used for the production of Daqu (Table S2) [18].

Volatile Flavor Compounds in TF Daqu
The volatile flavor components in TF Daqu were determined using HS-SPME-GC-MS. There were 101 kinds of volatile compounds detected in the TF Daqu samples studied, including alcohols, aldehydes, esters, acids, hydrocarbons, ketones, ethers, aromatics and pyrazines ( Table 2). Note: "-" not perceived. Table 2 shows that the flavor compounds in TF Daqu mainly consisted of esters, hydrocarbons and alcohols, and their relative contents account for 37.14, 30.17 and 7.64% of the total flavor compounds contents, respectively, while the relative contents of the other kinds of flavor compounds were <7%.
Esters determine the flavor type of liquor. In this study, esters were the compounds with the highest content and relatively more types than the other flavor compounds found in the TF Daqu samples. Among all of the flavor compounds, the relative content of methyl hexadecanoate was the highest (9.87%), followed by methyl linoleate (9.49%). Fan et al. [19] reported that the volatile compounds were mainly esters and alcohols in Fen flavor Daqu. Le et al. [20] also reached a similar conclusion during their analysis of the flavor substances in Fen Daqu. The hydrocarbon content in TF Daqu was second only to that of esters, and the relative content of tetradecane and dodecane was higher, accounting for 8.99 and 5.65%, respectively. Alcohols are the primary compound in liquor. In addition, the types and content of alcohols in TF Daqu were also high. Among them, the relative content of phenethyl alcohol was the highest, accounting for 4.31%.
Sun et al. [21] found that alcohols, esters and pyrazines were the main fragrance substances of sauce flavor Daqu. Meanwhile He et al. [22] detected 60 volatile substances in strong flavor Daqu, of which there were 42 esters, so esters were the main aroma components in strong flavor Daqu. The difference in the composition of the flavor substances in Daqu makes each Daqu liquor have a different flavor; the reason for the difference is that the microorganisms and enzymes in Daqu have different compositions and ratios [23].

Identification of the Key Flavor Compounds in TF Daqu
By analyzing the odor threshold of flavor compounds (Table 3), it was found that the relative content of trans-2-nonenal in all of the flavor compounds accounted for 0.37% and that the odor threshold was 0.09 µg/m 3 , which contributes the most to the integral flavor of TF Daqu. There were 10 compounds with an ROAV ≥ 1 in the TF Daqu samples studied. These were 1-octen-3-ol, phenethyl alcohol, phenylacetaldehyde, (E)-2-octenal, nonanal, (E)-2-nonenal, decanal, methyl laurate, isovaleric acid and eugenol; they are considered the main flavor compounds in TF Daqu. There were 12 compounds with 0.1 ≤ ROAV < 1, namely 1-octanol, hexanal, benzaldehyde, methyl nonanoate, 2-methylbutyric acid, octanoic acid, D-Limonene, dodecane, octadecane, 2-nonanone, 2,3,5-trimethylpyrazine and 2-pentylfuran; these volatile compounds play a modifying role in the overall flavor of TF Daqu, among which the ROAV of 2-pentylfuran was 0.97. Therefore, it can be considered to have an essential modifying effect on the integral flavor of TF Daqu. According to the ROAVs of the volatile flavor compounds, the key flavor compounds in TF Daqu were mainly composed of aldehydes and alcohols. These flavor compounds are usually produced by microorganisms in Daqu during the fermentation process or produced by microorganisms degrading starch compounds in the raw materials during glycolysis [24]. 1-Octen-3-ol has a fruity, grass and dusty flavor, and phenethyl alcohol and phenylacetaldehyde have a rosy aroma [25]. Trans-2-nonenal has an oily and grassy flavor [26]. Nonanal has the fragrance of grass and orange [27]. Trans-2-octenal has an oily odor [28]. These flavor compounds contribute to the unique flavor of TF Daqu.

Correlation Analysis of the Microbial Community Diversity and Flavor Compounds in TF Daqu
The correlation between the dominant flora and key flavor compounds of Daqu was analyzed using the Spearman correlation coefficients and p-values. The correlation heat map is shown in Figure 4. According to the ROAVs of the volatile flavor compounds, the key flavor compounds in TF Daqu were mainly composed of aldehydes and alcohols. These flavor compounds are usually produced by microorganisms in Daqu during the fermentation process or produced by microorganisms degrading starch compounds in the raw materials during glycolysis [24]. 1-Octen-3-ol has a fruity, grass and dusty flavor, and phenethyl alcohol and phenylacetaldehyde have a rosy aroma [25]. Trans-2-nonenal has an oily and grassy flavor [26]. Nonanal has the fragrance of grass and orange [27]. Trans-2-octenal has an oily odor [28]. These flavor compounds contribute to the unique flavor of TF Daqu.

Correlation Analysis of the Microbial Community Diversity and Flavor Compounds in TF Daqu
The correlation between the dominant flora and key flavor compounds of Daqu was analyzed using the Spearman correlation coefficients and p-values. The correlation heat map is shown in Figure 4. Among the dominant bacterial genera, a negative correlation was observed between Oceanobacillus and trans-2-nonenal (p < 0.05). A positive correlation was observed between Virgibacillus and eugenol, and a negative correlation was found between Virgibacillus and decanal (p < 0.05). Enterococcus had a significant positive correlation with trans-2-octenal (p < 0.01), a positive correlation with trans-2-nonenal and decanal, and a negative correlation with methyl laurate (p < 0.05). Pediococcus exhibited a significant negative correlation with trans-2-octenal (p < 0.01), a negative correlation with trans-2-nonenal and decanal, Among the dominant bacterial genera, a negative correlation was observed between Oceanobacillus and trans-2-nonenal (p < 0.05). A positive correlation was observed between Virgibacillus and eugenol, and a negative correlation was found between Virgibacillus and decanal (p < 0.05). Enterococcus had a significant positive correlation with trans-2-octenal (p < 0.01), a positive correlation with trans-2-nonenal and decanal, and a negative correlation with methyl laurate (p < 0.05). Pediococcus exhibited a significant negative correlation with trans-2-octenal (p < 0.01), a negative correlation with trans-2-nonenal and decanal, and a positive correlation with methyl laurate (p < 0.05). Streptomyces had a positive correlation with eugenol and a negative correlation with decanal (p < 0.05). A negative correlation was observed between Saccharopolyspora and decanal (p < 0.05). Leuconostoc exhibited a considerable negative correlation with trans-2-nonenal (p < 0.01) and a negative correlation with trans-2-octenal (p < 0.05). Uncultured_bacterium_f_Lachnospiraceae was positively correlated with trans-2-octenal, trans-2-nonenal and decanal (p < 0.05). A positive correlation was observed between Lactobacillus and methyl laurate and a negative correlation with trans-2-octenal p < 0.05). Among the dominant fungi, a positive correlation was observed between Wickerhamomyces and 1-octene-3-ol (p < 0.05). A positive correlation was observed between Fusarium and nonanal (p < 0.05).
For these ten key flavor compounds, Enterococcus and uncultured_bacterium_f_Lachnospiraceae contributed the most to the flavor compounds. Figure 4 shows Enterococcus was associated with the synthesis of aldehydes such as trans-2-octenal. Uncultured_bacterium_f_Lachnospiraceae was associated with the synthesis of most of the aldehydes detected.
Enterococcus has good biosafety and probiotic properties, is often used to accelerate the fermentation process in fermentation production, and can hydrolyze proteins and esters to endow foods with good flavor and metabolize bacteriocins [29]. Lachnospiraceae may be a potential probiotic. All members of the Lachnospiraceae family have fermentation properties, of which some have strong hydrolyzing activities, such as pectate lyase, pectin methylesterase, α-amylase, xylanase and α-L-arabinofuranosidase [30]. Oceanobacillus are present in Maotai flavor Daqu, Luzhou flavor Daqu and Fen flavor Daqu. During the fermentation of liquor, Oceanobacillus are capable of producing various enzymes such as protease, amylase, cellulase and esterase. In addition, Oceanobacillus promote the esterification and saccharification ability of Daqu to some extent [31]. Kroppenstedtia is the dominant group of high-temperature sesame-flavored Daqu and its ability to secrete cellulase is strong, which contributes to the liquefaction ability of Daqu and the alcohol production in this liquor [32]. Virgibacillus are widely found in the natural environment and can utilize most carbohydrates as carbon and energy sources. In addition, Virgibacillus is able to produce exoenzymes, such as amylase, protease, inulinase and gelatinase [33]. Pediococcus is often considered to be spoilage bacteria in wine and has the ability to synthesize extracellular polysaccharides, which give the wine a sticky and thick texture. In addition, Pediococcus can develop a variety of enzymes that enable the production of desirable fragrant substances in wine [34]. Streptomyces is an important type of microorganism that is widespread in all kinds of ecosystems. Streptomyces has the capacity to secrete alkaline phosphatase, esterase and phosphate hydrolase, which are likely to have a significant role in the composition of the flavor components or precursors in Maotai flavor liquor [4]. Saccharopolyspora is mainly found in the natural surroundings and can produce important bioactive substances [35]. Leuconostoc is present in many different environments and is critical for the preparation and fermentation of many dairy products, vegetables and grains. In addition, it generally forms the buttery taste of dairy products [36]. Lactobacillus is considered a probiotic, which is beneficial to human health. In addition, it is present in some fermented food products where it is helpful in preservation, aroma and nutrition [36]. Monascus is widely present in Daqu, fermented grains, brewing mash, etc. Monascus has been widely used for the production of Monascus pigments used to color traditional foods [37]. In addition, Monascus can produce a variety of beneficial metabolites as a fermenting bacterium for traditional Chinese foods. Aspergillus is widely distributed in the environment and not only can use monosaccharides, but also produces numerous enzymes to decompose proteins, polysaccharides and other organic macromolecules. Aspergillus is also commonly used in food fermentation, for example, koji fermentations used in the production of soy sauce and miso [38]. The role that these species play in TF Daqu needs further study.

Conclusions
The results of our study have shown that microorganisms are highly abundant in TF Daqu and the bacterial diversity was higher than the fungal diversity. A variety of volatile flavor compounds were detected in TF Daqu, which mainly consist of esters, hydrocarbons and alcohols. The results of our correlation analysis showed that Enterococcus and uncultured_bacterium_f_Lachnospiraceae contribute more to the flavor compounds. Through the characterization of the microbial communities and flavor compounds of TF Daqu, our understanding of TF Daqu has deepened, providing a direction for further research on the formation of flavor compounds in liquor and the enhancement of Daqu quality. This study has a guiding significance for producing liquor and enhancing the quality of the final product.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods12020324/s1, Table S1: Species statistics by class of samples. Table S2: Differences in the dominant microorganisms of different types of Daqu.
Author Contributions: Conceptualization, W.Z. and P.W.; Methodology, P.W.; Formal analysis, R.Z.; Data curation, R.Z.; Writing-original draft, Y.L. and H.Q. Y.L. and H.Q. are contributed equally to this paper; Funding acquisition, W.Z. and H.Q. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by the National Natural Science Foundation of China (31960486, 21864002) for financial support.

Data Availability Statement:
The data of microbial diversity has been submitted to the SRA of NCBI, and the bioproject number is PRJNA898104. Other data are available from the authors.