Structural Characteristics and Formation Mechanism of Microbiota Related to Fermentation Ability and Alcohol Production Ability in Nongxiang Daqu

Fermentation ability and alcohol production ability are important quality indicators of Chinese liquor Daqu, reflecting microbial growth and metabolic capacity and ethanol production capacity of Daqu microbiota, respectively. However, information on the microbial community related to the fermentation ability and alcohol production ability is unclear. In this study, fermentation functional microbiota (FFM) and alcohol functional microbiota (AFM) were obtained by correlating fermentation ability and alcohol production ability with Daqu microbiota. FFM and AFM consisted of 50 and 49 genera, respectively, which were basically the same at the phylum level but differed at the genus level. Correlation analysis showed that FFM and AFM were mainly affected by moisture, acidity, and humidity in the early stage of Daqu fermentation, and oxygen content was a critical factor for microbial succession in the middle stage of fermentation. FFM and AFM had commensal or synergistic interactions with multiple microbes. Function predictions indicated that fermentation functional bacterial microbiota was active in product synthesis and transport-related metabolic functions, and alcohol functional bacterial microbiota was very active in raw material utilization and its own metabolic synthesis. This study reveals the structural characteristics and formation mechanism of FFM and AFM, which is important for control of Daqu quality.


Introduction
Chinese Baijiu is one of the six major distilled spirits in the world and is widely loved by consumers for its unique aroma and taste. Daqu, the saccharifying fermenting agent in the solid-state fermentation process of Chinese Baijiu, provides raw materials, microbes, and enzymes vital to Baijiu fermentation. While brewing Baijiu, Daqu is generally added after Jiupei (a mixture of fermented grains) has cooled, mixed well, and then fermented in the pit. During fermentation, Daqu is rich in hydrolytic enzymes and microflora, thus allowing the simultaneous saccharification and fermentation of grains. During Baijiu production, starch is hydrolyzed to glucose, which is further converted to ethanol. This process is continuous, progressive, and involves various microbes and enzymes. Liquefaction activity, saccharification activity, fermentation ability, and alcohol production ability are biochemical indicators reflecting Daqu quality. While liquefaction activity and saccharification activity are related to enzymes, fermentation ability and alcohol production ability are more concerned with microbes in Daqu.
Fermentation ability is expressed in terms of carbon dioxide produced by Daqu using fermentable sugars. Multiple metabolic pathways in microbial metabolism can produce carbon dioxide, accompanied by the production of a large number of intermediate metabolites that contribute to the formation of flavor compounds. Daqu with high fermentation

Sampling
Daqu bricks were collected from a Nongxiang Baijiu enterprise in Yibin, China. The bricks were collected at different time points (days 1, 2, 5, 6, 12, and 13) during Daqu fermentation. A Daqu brick was taken from four locations of a Qu-room each time ( Figure S1). Each Daqu brick was crushed and stored in a sterile plastic bag. A total of 24 samples were collected, and a part of them was analyzed for physicochemical properties in real time, while the other part was stored at −80 • C for total DNA extraction.

Physicochemical Properties
Total acidity was assessed using the acid base titration method. The moisture of Daqu was determined by the direct drying method at 105 • C. The temperature, humidity, and oxygen content at each sampling point of the Daqu bricks were monitored in real time using the sensors of an online environmental detection system. Fermentation ability and alcohol production ability were determined as follows: 1 g of Daqu powder was added to 150 mL of sterilized saccharification solution, which was placed in an incubator at 30 • C, and the fermentation was stopped after 72 h. Fermentation ability was evaluated by measuring the weight change of the fermented system before and after fermentation, i.e., the amount of carbon dioxide produced ((unit: g/(1 g-72 h); symbol: U). Alcohol production ability was expressed as the ethanol content in the fermentation broth at the end of fermentation (unit is mL/100 g). Because of the low ethanol content in some samples, a modified method of residual alcohol determination in Jiupei was used to determine the ethanol content in the samples [12]. Potassium dichromate oxidizes ethanol to acetic acid, and hexavalent chromium is reduced to trivalent chromium; thus, the ethanol content of the fermentation broth was determined by the colorimetric method. The method was as follows: 100 mL of fermentation solution was taken for distillation, an appropriate amount of distillate was taken in a colorimetric tube, 1 mL of 2% potassium dichromate and 5 mL of concentrated sulfuric acid were added. The solution was shaken well, heated together with the standard tube for 10 min, and cooled. Finally, the optical density was measured at 600 nm with a spectrophotometer.

DNA Extraction and Processing
To obtain the genomic information of the microbial community, the total DNA in the samples was extracted using a modified cetyltrimethylammonium bromide extraction method [13]. The V3-V4 region of the bacterial 16S rRNA gene was amplified using the universal primer pair 338F/806R, and the internal transcribed spacer (ITS) region of the fungal gene was amplified using the universal primer pair ITS1F/2043R. PCR amplification was performed as follows: 95 • C for 3 min; 95 • C for 30 s, 55 • C for 30 s, 72 • C for 45 s, 27 cycles; 72 • C for 10 min, ending at 4 • C. PCR products were separated on a 2% agarose gel, purified using the AxyPrep DNA gel extraction kit (Axygen Biosciences, Union City, CA, USA) and quantified using the Quantus™ fluorometer (Promega, Madison, WI, USA). Amplicons were sequenced using the Illumina MiSeq platform at Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China). The raw sequence reads generated in this study were all deposited in the NCBI database (accession number: PRJNA869206).

Statistical Analysis
Spearman correlations between the two indicators (i.e., fermentation ability and alcohol production ability) and Daqu microbiota were analyzed using the pheatmap package in R v3.3.1. Redundancy analysis (RDA) or canonical correspondence analysis (CCA) was performed to reveal the correlations between associated microbial communities and environmental factors using the vegan package in R. To reveal the effects of microbial interactions on the associated communities, Procrustes analysis and co-occurrence network analysis were performed using the vegan and igraph packages in R, respectively. Phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt2) analysis based on 16S rRNA and ITS sequence data was performed to gain additional insights into the metabolic potential of the related communities. The gene family counts for each predicted sample were derived from the MetaCyc metabolic pathway database (https://metacyc.org/, accessed on 5 June 2022). The differential functions of the two microbial communities were analyzed using linear discriminant analysis effect size (LEfSe; http://huttenhower.sph.harvard.edu/galaxy/, accessed on 20 June 2022).

Variations in the Fermentation Ability and Alcohol Production Ability
Variations in the fermentation ability and alcohol production ability are shown in Figure S2. At the beginning of fermentation, fermentation ability was high, but it gradually decreased with an increase in temperature, eventually dropped to 0.22 (U). In the first two days of fermentation, alcohol production ability was low, and but it gradually increased to 3.03 (mL/100 g), the highest level, which lasted for 4 days, before rapidly decreasing and finally returning to the initial level. This might be due to microbial succession during fermentation, which leads to changes in fermentation ability and alcohol production ability.

Structural Characteristics of Functional Microbiota
High-throughput sequencing technology was used to obtain information on Daqu microbiota, in which there were 211 genera in 12 bacterial phyla and 63 genera in 4 fungal phyla. The community barplots in Figure 1 show the composition and dynamics of Daqu microbiota. The relative abundance (RA) of each microbial species in the community barplot was greater than 0.01%. The richness of bacterial genera increased with the increase in days of fermentation, but the community changes usually tended to be stable ( Figure 1a). The abundance of fungal genera decreased with the increase in days of fermentation, and the rate of fungal succession was rapid ( Figure 1b).
To obtain the associated microbial communities, we performed correlation analysis of Daqu microbiota with fermentation ability and alcohol production ability. The microbial communities that showed significant correlation (|R| > 0.4, p < 0.05) with the two indicators were termed fermentation functional microbiota (FFM) and alcohol functional microbiota (AFM). The fermentation functional bacterial microbiota (FFBM) comprised 30 genera in 5 phyla-Cyanobacteria, Bacteroides, Proteobacteria, Firmicutes, and Actinobacteria. A total of 7 genera were positively correlated and 23 genera were negatively correlated with fermentation ability (Figure 2a). The alcohol functional bacterial microbiota (AFBM) consisted of 33 genera in 5 phyla-Bacteroides, unclassified_k__norank_d__Bacteria, Proteobacteria, Firmicutes, and Actinobacteria. Thirty genera were positively correlated and three genera were negatively correlated with alcohol production ability. At the phylum level, both had the same four phyla, but at the genus level, 24 genera were specific to FFBM, 27 genera were specific to AFBM, and only 6 genera were shared by FFBM and AFBM. In FFBM, the dominant genera were norank_f__norank_o__Chloroplast, Saccharopolyspora, Thermoactinomyces, and Pseudomonas. In AFBM, the dominant genera were Saccharopolyspora, Thermoactinomyces, and Pediococcus. In AFBM, Pediococcus (with RA reaching 3.5%) was the dominant genus on the first day and the absolute dominant genus on the second day of fermentation. In FFBM, norank_f__norank_o__Chloroplast (with RA reaching 52%) was the dominant genus on the first day of fermentation. RA of both subsequently decreased with fermentation time. Pseudomonas in FFBM is a dominant genus in the fermentation pre-period and significantly regulates bacterial diversity and community composition [19]. In addition, it regulates the spinability of dimethylbutyrate, which is an important flavor compound [20]. Saccharopolyspora and Thermoactinomyces were the dominant genera shared by the two communities. Saccharopolyspora is a key genus in the middle stage of Daqu fermentation [9] and may be involved in amino acid metabolism, sugar metabolism, and ester synthesis in Daqu fermentation [21]. Thermoactinomyces is a thermotolerant bacterial genus that is dominant in both Daqu and heap fermentation [22] with complete metabolic pathways for acetaldehyde and acetic, propionic, and butyric acids [23], which are crucial to the formation of volatile compounds [24]. To obtain the associated microbial communities, we performed cor of Daqu microbiota with fermentation ability and alcohol production ab bial communities that showed significant correlation (|R| > 0.4, p < 0.0 nus in the middle stage of Daqu fermentation [9] and may be involved in amino acid metabolism, sugar metabolism, and ester synthesis in Daqu fermentation [21]. Thermoactinomyces is a thermotolerant bacterial genus that is dominant in both Daqu and heap fermentation [22] with complete metabolic pathways for acetaldehyde and acetic, propionic, and butyric acids [23], which are crucial to the formation of volatile compounds [24]. Fermentation functional fungal microbiota (FFFM) comprised 20 genera in 4 phyla-Mucoromycota, unclassified_k__Fungi, Ascomycota, and Basidiomycota ( Figure 2b). Eighteen genera were positively correlated with fermentation ability, of which yeasts accounted for one-half-Cutaneotrichosporon, Kazachstania, Trichosporon, Candida, Wickerhamomyces, Issatchenkia, Apiotrichum, Hyphopichia, and unclassified_o__Saccharomycetales, indicating that carbon dioxide production was closely related with yeasts, which is consistent with previous investigations [3]. Alcohol functional fungal microbiota (AFFM) included 16 genera in the same 4 phyla (i.e., Mucoromycota, unclassified_k__Fungi, Ascomycota, and Basidiomycota). A total of 4 genera were positively correlated and 12 genera (mainly molds and yeasts) were negatively correlated with alcohol production ability. Both FFFM and AFFM were identical at the phylum level, but at the genus level, 12 genera were exclusive to FFFM, 8 genera were exclusive to AFFM, and 8 genera were common to FFFM and AFFM. Dominant genera in FFFM were Thermoascus, Thermomyces, Issatchenkia, unclassified_k__Fungi, Aspergillus, Kazachstania, Trichosporon, unclassified_p__Ascomycota, and Candida. Dominant genera in AFFM were Thermoascus, Thermomyces, Issatchenkia, Fermentation functional fungal microbiota (FFFM) comprised 20 genera in 4 phyla-Mucoromycota, unclassified_k__Fungi, Ascomycota, and Basidiomycota ( Figure 2b). Eighteen genera were positively correlated with fermentation ability, of which yeasts accounted for one-half-Cutaneotrichosporon, Kazachstania, Trichosporon, Candida, Wickerhamomyces, Issatchenkia, Apiotrichum, Hyphopichia, and unclassified_o__Saccharomycetales, indicating that carbon dioxide production was closely related with yeasts, which is consistent with previous investigations [3]. Alcohol functional fungal microbiota (AFFM) included 16 genera in the same 4 phyla (i.e., Mucoromycota, unclassified_k__Fungi, Ascomycota, and Basidiomycota). A total of 4 genera were positively correlated and 12 genera (mainly molds and yeasts) were negatively correlated with alcohol production ability. Both FFFM and AFFM were identical at the phylum level, but at the genus level, 12 genera were exclusive to FFFM, 8 genera were exclusive to AFFM, and 8 genera were common to FFFM and AFFM. Dominant genera in FFFM were Thermoascus, Thermomyces, Issatchenkia, unclassified_k__Fungi, Aspergillus, Kazachstania, Trichosporon, unclassified_p__Ascomycota, and Candida. Dominant genera in AFFM were Thermoascus, Thermomyces, Issatchenkia, unclassified_k__Fungi, Dipodascus, Rhizomucor, Kazachstania, unclassified_p__Ascomycota, and unclassified_f__ Lichtheimiaceae. Six genera were common to both communities, three genera were unique to FFFM, and three genera were unique to AFFM. Thermoascus, Thermomyces, Issatchenkia, and unclassi-fied_k__Fungi were the dominant genera shared by the two communities, and the sum of RA of these four genera was greater than 74% on each day of fermentation. Thermoascus and Thermomyces can be used as biomarkers for soy sauce flavor Daqu for black and white types, respectively [25], and are the dominant genera in different rounds of heap fermentation [26] as well as the dominant genera in the early stage of pit fermentation [27], which can produce various thermophilic glycoside hydrolases [28] to accelerate the decomposition and utilization of starchy raw materials, promote microbial growth, and produce Baijiu flavor [29]. Aspergillus is also a dominant genus in Daqu at the fermentation stage, heap stage, and early stage of pit fermentation [26,27,30], and it synthesizes various carbohydrate hydrolases that directly participate in the starch, pyruvate, and tricarboxylic acid cycle (TCA) pathways [31], laying the material foundation for producing liquor and aroma by other strains. Kazachstania is the dominant genus in the early stage of Daqu fermentation, and it can replace the dominant genera Aspergillus, Thermoascus, and Thermomyces as the absolute dominant genus in the middle and late stages of liquor fermentation [27]. It has some acid tolerance and can thus participate in the biosynthesis of acetic acid, lactic acid, ethyl acetate, and ethyl lactate [32,33], contributing greatly to the late stage of Baijiu aroma development. Issatchenkia is the absolute dominant genus in the first two days, with RA reaching 85% on the second day of fermentation. It synthesizes pectinases and lipases [34] that are associated with the production of higher alcohols and volatile acids in grain fermentation [22,35]. Trichosporon is a key genus in Daqu fermentation [9], and it is associated with phenylethanol synthesis in sauce-flavored Daqu [29], which has a rosy aroma and is an important Baijiu flavor compound. Candida is the dominant genus in different stages [26,27], and it produces various hydrolytic enzymes [36], which have the ability to synthesize acetic acid and many alcohols [33,37]. We can conclude that Aspergillus, Thermoascus, and Thermomyces in the fungal community were important fermentation starters in liquor fermentation, and these fungal genera could break down starch raw materials in the early stage of fermentation to form various small molecules that provide the material basis for the metabolic synthesis of the other strains. The four non-Saccharomyces yeast strains Kazachstania, Issatchenkia, Trichosporon, and Candida play an articulate role and can synthesize various flavor chemicals in the middle and late stages of Baijiu fermentation, contributing to the aroma of the spirit.
In terms of total genera, FFBM and AFBM accounted for nearly one-seventh and onesixth of the total bacterial genera, respectively, and FFFM and AFFM accounted for nearly one-third and one-fourth of the total fungal genera, respectively. In terms of dominant genera, FFBM and AFBM contained 4 and 3 of the 17 dominant genera in the bacterial community, respectively, and FFFM and AFFM each contained 9 of the 15 dominant genera in the fungal community. Thus, FFM and AFM play pivotal roles in the whole Daqu community.

Effects of Biotic and Abiotic Factors on Functional Microbial Communities
To investigate the effects of environmental factors on the correlated communities, RDA or CCA was performed. As shown in Figure 3a-d, the two axes explained 44.15%, 55.96%, 41.92%, and 66.10% of the total variance of AFBM, AFFM, FFBM, and FFFM differentiation, respectively, indicating that the physicochemical properties could explain the variation in microbial communities to some extent. AFBM and FFBM were affected by moisture, acidity, and oxygen content to a significant level (Table S1) (r2 > 0.4, p < 0.01). Moisture, acidity, humidity, oxygen content, and temperature were significantly correlated with AFFM and FFFM distribution (r2 > 0.4, p < 0.01). Therefore, the effects of environmental factors on FFM and AFM succession were approximately the same, i.e., FFM and AFM were mainly affected by moisture, acidity, and humidity in the early stage of Daqu fermentation, and oxygen content became a critical factor for microbial succession in the middle stage of fermentation. Further, change in moisture was the main factor affecting succession in FFBM and AFBM (r2 ≥ 0.60, p < 0.01), which is in line with the fact that the main driving force of the Daqu bacterial community is moisture [11,38], and all factors significantly contributed to succession in FFFM and AFFM, with humidity having the most significant effect (r2 ≥ 0.84, p < 0.01).
Procrustes analysis was performed to study microbial interactions. As shown in Figure 4a-d, the correlation between bacterial and fungal communities and AFM and FFM reached a significant level (p < 0.05), indicating that both bacterial and fungal communities could drive the succession of AFM and FFM. By co-occurrence network analysis in Figure 4e, (|R| > 0.4, p < 0.05), we found that AFM succession was affected by three bacterial genera and multiple yeast genera, and FFM succession was associated with one bacterial genus and three fungal genera. AFBM had higher co-occurrence with Candida, Trichosporon, and Cutaneotrichosporon, whereas AFFM succession was more significantly influenced by Trichosporon, Cutaneotrichosporon, and Aspergillus. Nesterenkonia had some influence on FFBM succession. Kodamaea, Fusarium, and Clavispora had an effect on FFFM succession. Therefore, microbial interactions could influence succession of FFM and AFM. The significant positive correlation between both communities and multiple microbes indicated that several microbes had a commensalism or synergism effect with both communities.

Potential Functions of the Two Microbial Communities
To understand the functions of the associated microbial communities, functions were predicted using PICRUSt2. The MetaCyc pathway abundance data showed that 71 metabolic pathways were common to both FFFM and AFFM, except for one metabolic pathway that was unique to AFFM. Because of the low abundance of some metabolic pathways, only the main metabolic pathways were analyzed, i.e., the top 54 in abundance. The analysis of FFFM and AFFM functions (Figure 5a) revealed 6 metabolic pathways for carbohydrate biosynthesis/degradation; 7 metabolic pathways for precursor metabolites and energy production; 15 metabolic pathways for nucleotide metabolism (which might be involved in DNA/RNA biosynthesis); 7 amino acid metabolic pathways; 6 cofactor, carrier, and vitamin biosynthesis metabolic pathways; 7 fatty acid and lipid biosynthesis/degradation metabolic pathways; and 2 secondary metabolite biosynthesis metabolic pathways, indicating that the community was more active in all aspects of metabolic functions, especially in the utilization of raw materials for decomposition and autogenous reproduction. Enzymes encoding saccharification and ethanol fermentation-related enzymes might be associated with Thermoascus and Thermomyces [10], which were the absolute dominant fungi shared by FFFM and AFFM, which is consistent with our previous discussion that both fungi are important fermentation starter strains in Baijiu production.   The functional abundance of FFFM and AFFM (Figure 5a) was relatively consistent day to day. The metabolic pathways that were not significantly different between the two communities (LDA ≥ 3.0, p ≤ 0.05) were found using LEfSe on community functions, which showed that FFFM and AFFM functions were consistent. On further analysis, 13 metabolic pathways directly produced carbon dioxide and one metabolic pathway produced ethanol-superpathway II for adenosine nucleotide ab initio biosynthesis, heme biosynthesis I (aerobic), L-proline biosynthesis II (from arginine), TCA cycle II (fungi), mevalonate pathway I, L-valine biosynthesis, L-tyrosine degradation I, glycine superpathway for heme synthesis, stearic acid biosynthesis III (fungi), L-leucine degradation I, tetrapyrrole biosynthesis II (from glycine), octanoyl-[acyl carrier protein] biosynthesis (mitochondria), and chitin degradation to ethanol. Thus, carbon dioxide production pathways were more abundant than ethanol production pathways.
The clusters of orthologous groups (COG) functional abundance data indicated that the same 23 broad categories of functions were present in both FFBM and AFBM (Figure 5b,c). Analysis of FFBM and AFBM functions revealed that these functions were more abundant. It was more active in energy production and conversion, amino acid transport and metabolism, carbohydrate transport and metabolism, translation ribosome structure and biogenesis, transcription, and inorganic ion transport and metabolism. This indicated that the community was in a high rate of reproduction and had a high level of metabolic processes, which might help produce more intermediate metabolites that contribute to flavor formation. This also has some similarity to the function of hightemperature Daqu bacterial communities [39]. Further, LEfSe of community functions was performed to select differential metabolic pathways (LDA ≥ 3.0, p ≤ 0.05). The functions that showed significant differences in FFBM were coenzyme transport and metabolism, post-translational modification of protein turnover chaperones, cell motility, inorganic ion transport and metabolism, intracellular transport secretion and vesicular transport, and signal transduction mechanism ( Figure S3), indicating that FFBM was active in product synthesis and transport-related metabolic functions. In AFBM, the functions that showed significant differences were carbohydrate transport and metabolism, acid transport and metabolism, transcription, recombination, and repair, indicating that AFBM was very active in raw material utilization and its own metabolic synthesis.
We analyzed the specific pathways to obtain detailed information. MetaCyc pathway abundance data showed 360 and 329 metabolic pathways in FFBM and AFBM, respectively, of which 326 metabolic pathways were common to both, 34 pathways were unique to FFBM, and only 3 pathways were unique to AFBM. LEfSe of metabolic pathway abundance differences showed that 62 metabolic pathways were significantly different (LDA ≥ 3.0, p ≤ 0.01), including 44 in FFBM and 18 in AFBM, a nearly three-fold difference (Figure 5d). FFBM had 10 amino acid biosynthesis pathways; 12 cofactor, carrier, and vitamin biosynthesis pathways; 12 carbohydrate metabolism pathways; 2 nucleic acid processing pathways; 2 precursor metabolism and energy production pathways; 3 secondary metabolite biosynthesis pathways; and 1 sulfur compound metabolism pathway. In addition, five of these metabolic pathways produce carbon dioxide-the superpathway of heme biosynthesis from glutamate, L-tryptophan biosynthesis, the superpathway of L-phenylalanine biosynthesis, heme biosynthesis II (anaerobic), and pyridoxal 5 -phosphate biosynthesis I. Thus, the community was mainly active at higher levels of primary and secondary metabolism, which was consistent with the differential analysis of COG functions. In AFBM, there were two amino acid biosynthesis pathways; two cofactor, carrier, and vitamin biosynthesis pathways; three carbohydrate metabolism pathways; two precursor metabolism pathways; two energy production pathways (both for lactic acid fermentation, of which iso-lactic acid fermentation produces ethanol); two secondary metabolite biosynthesis pathways; five nucleotide metabolism pathways; and two cell wall biosynthesis pathways. This coincided with the differential analysis of COG functions, where the nucleotide metabolism pathways confirmed that AFBM was active in its own transcription and recombination, and that lactate was produced in three carbohydrate metabolism pathways and two precursor metabolism and energy production pathways, indicating a high level of phosphorylation at the substrate level at this time.
In Figure 5d, some differential metabolic pathways of FFBM were connected before and after: (i) pentose phosphate pathway → branching acid biosynthesis → superpathway of aromatic amino acid biosynthesis (L-tryptophan biosynthesis, L-phenylalanine biosynthesis superpathway). The pentose phosphate pathway can produce D-erythrose-4phosphate, which can be used as a starting substrate in branching acid biosynthesis, an important intermediate that can participate in the biosynthesis of many essential metabolites.
(ii) L-glutamate and L-glutamine biosynthesis/L-serine and glycine biosynthesis superpathway I → tetrapyrrole biosynthesis I (from glutamate)/tetrapyrrole biosynthesis II (from glycine) → heme biosynthesis from glutamate/heme biosynthesis II (anaerobic). Heme is an iron-containing cofactor found in many essential proteins; in addition to its role in oxidative metabolism, it plays a molecular regulatory role in transcription, translation, protein targeting, protein stability, and cell differentiation. Metabolite delivery is one of the ways in which microbial interactions work [40], suggesting a strong mutual cooperation among microbes within FFBM.
To understand the interactions within the functional microbial communities, genuslevel co-occurrence network analysis was performed for FFM and AFM. As shown in Figure 6a,b, all nodes of the FFM and AFM co-occurrence networks were clustered into four modules. FFM mainly consisted of module 1 (40.92%) and module 2 (40.92%), and AFM mainly consisted of module 1 (53.06%) and module 2 (36.73%). However, the FFM network structure was more complex than the AFM network structure, and its modules 1 and 2 were strongly linked, indicating stronger FFM interactions and more active metabolic levels compared with the AFM network structure.
Based on the functions analysis, FFFM and AFFM functions converged in terms of kind and abundance. However, carbon dioxide production pathways were more than ethanol production pathways, indicating that the fermentation ability index was more widely representative of community activity. The number of differential functions was higher in FFBM than in AFBM, and the differential functions of FFBM indicated that the community was more advantageous for both substrate utilization and product synthesis. In summary, FFM was more widely functional and had a more active community than AFM.  Based on the functions analysis, FFFM and AFFM functions converged in terms of kind and abundance. However, carbon dioxide production pathways were more than ethanol production pathways, indicating that the fermentation ability index was more widely representative of community activity. The number of differential functions was higher in FFBM than in AFBM, and the differential functions of FFBM indicated that the community was more advantageous for both substrate utilization and product synthesis. In summary, FFM was more widely functional and had a more active community than AFM.

Conclusions
In the present study, by analyzing the structure of microbial communities related to fermentation ability and alcohol production ability, we found that FFM and AFM were approximately the same at the phylum level but differed greatly at the genus level, and both occupied an important position in Daqu microbiota. Environmental factors significantly affected the structure the correlated communities. FFM and AFM were mainly affected by moisture, acidity, and humidity in the early stage of Daqu fermentation, and oxygen content was a critical factor for microbial succession in the middle stage of fermentation. Microbial interactions could drive the succession of both FFM and AFM, and multiple microbes had commensalism or synergism interactions in both communities. Function predictions revealed that FFFM and AFFM functions were relatively consistent, which was particularly active in the utilization of raw materials for decomposition and autogenous reproduction. FFBM was active in product synthesis and transport-related metabolic functions, while AFBM was very active in the utilization of raw materials and its own metabolic synthesis.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods11172602/s1, Figure S1: Sampling points of Daqu in the Qu-room; Figure S2: Variations in the fermentation ability and alcohol production ability. Red represents fermentation ability, blue represents alcohol production ability; Figure S3: Functional differences between FFBM and AFBM. Red represents the differential metabolic pathways in AFBM, green represents the differential metabolic pathways in FFBM. (Function prediction based on COG); Table S1: Functional community and environmental factor permutation test.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.