Fungal Biomarkers in Traditional Starter Determine the Chemical Characteristics of Turbid Rice Wine from the Rim of the Sichuan Basin, China

The fungal community in Qu plays a key role in the formation of turbid rice wine (TRW) style. The Sichuan Basin and its surrounding areas have become one of the main TRW production regions in China; however, the fungal community in Qu and how they affect the characteristics of TRW remain unknown. Therefore, this study provided insight into the fungal biomarkers in Qu from Guang’an (GQ), Dazhou (DQ), Aba (AQ), and Liangshan (LQ), as well as their relationships with compounds in TRW. The main biomarkers in GQ were Rhizopus arrhizus, Candida glabrata, Rhizomucor pusillus, Thermomyces lanuginosus and Wallemia sebi. However, they changed to Saccharomycopsis fibuligera and Mucor indicus in DQ, Lichtheimia ramose in AQ, and Rhizopus microsporus and Saccharomyces cerevisiae in LQ. As a response to fungal biomarkers, the reducing sugar, ethanol, organic acids, and volatile compounds were also changed markedly in TRWs. Among important volatile compounds (VIP > 1.00), phenethyl alcohol (14.1–29.4%) was dominant in TRWs. Meanwhile, 3-methyl-1-butanol (20.6–56.5%) was dominant in all TRWs except that fermented by GQ (GW). Acetic acid (29.4%) and ethyl palmitate (10.1%) were dominant in GW and LW, respectively. Moreover, GQ biomarkers were positively correlated with acetic acid and all unique important volatile compounds in GW. DQ biomarkers had positive correlations with unique compounds of acetoin and ethyl 5-chloro-1,3,4-thiadiazole-2-carboxylate in DW. Meanwhile, the AQ biomarkers were positively correlated with all AW unique, important, and volatile compounds. Although there were not any unique volatile compounds in LW, 16 important volatile compounds in LW were positively related to LQ biomarkers. Obviously, biomarkers in different geographic Qu played vital roles in the formation of important volatile compounds, which could contribute specific flavor to TRWs. This study provided a scientific understanding for future efforts to promote the excellent characteristics of TRW by regulating beneficial fungal communities.


Introduction
Turbid rice wine (TRW) is one of the most popular alcoholic beverages in east Asia [1]. In China, it is consumed mainly by Han and ethnic minorities in the southern rice-growing regions because of its rich nutrition, such as amino acids, polypeptides, vitamins, and other bioactive components [2,3]. TRW is generally made by semisolid-state fermentation of glutinous rice, millet or other cereals [4] with TRW starter (Qu) in a relatively closed pottery jar [5]. In this case, the stand or fall of Qu directly affects the quality of TRW products.
The effective data were denoised using the Divisive Amplicon Denoising Algorithm 2 (DADA2), and the final amplicon sequence variants (ASVs) were then generated by removing the redundant and low occurrence (n < 5 within all samples). QIIME2′s classsklearn algorithm was used for species annotation for each ASV using a pre-trained Naive Bayes classifier.

Rice Wine Fermentation on a Lab-Scale
The main brewing process of turbid rice wine is depicted in Figure 2. At room temperature, 250 g glutinous rice was soaked in distilled water for 4 h, then drained and steamed for 35 min at 100 °C. After the glutinous rice cooled down to room temperature, it was mixed with 1 g Qu and transferred into a pottery jar. Then, the mixture was supplemented with 200 mL distilled water to ferment at 25 °C for 3 days. A 50 mL fermentation broth of each TRW sample was stored at 4 °C prior to testing. TRW fermented by GQ, DQ, AQ, and LQ was named GW, DW, AW, and LW, respectively. All experiments were repeated in triplicate.
The effective data were denoised using the Divisive Amplicon Denoising Algorithm 2 (DADA2), and the final amplicon sequence variants (ASVs) were then generated by removing the redundant and low occurrence (n < 5 within all samples). QIIME2's classsklearn algorithm was used for species annotation for each ASV using a pre-trained Naive Bayes classifier.

Rice Wine Fermentation on a Lab-Scale
The main brewing process of turbid rice wine is depicted in Figure 2. At room temperature, 250 g glutinous rice was soaked in distilled water for 4 h, then drained and steamed for 35 min at 100 • C. After the glutinous rice cooled down to room temperature, it was mixed with 1 g Qu and transferred into a pottery jar. Then, the mixture was supplemented with 200 mL distilled water to ferment at 25 • C for 3 days. A 50 mL fermentation broth of each TRW sample was stored at 4 • C prior to testing. TRW fermented by GQ, DQ, AQ, and LQ was named GW, DW, AW, and LW, respectively. All experiments were repeated in triplicate.

Reducing Sugar and Ethanol Analysis
The reducing sugar was determined by the 3,5-dinitrosalicylic acid (DNS) assay [5]. The ethanol content was determined using a gas chromatograph (GC-2020 Plus, Shimadzu, Japan) equipped with a capillary column of WONDA CAP WAX (30 m × 0.25 mm × 0.25 µm) and detector of DET1 [9].

Non-Volatile Organic Acids Analysis
Organic acid analysis was carried out via HPLC (Waters 2695, Waters Corporation, Milford, MA, USA) equipped with an Aminex HPX-87H ion exchange column (300 × 7.80 mm, 9 µm film thickness, Bio-Rad Laboratories, Inc., Hercules, CA, USA) at a temperature of 60 °C [10]. The injection volume was 20 µL, and the mobile phase was 7 mmol/L H2SO4 solution (pH 2.2) at the flow rate of 0.60 mL/min. The ultraviolet detector was set at 210 nm.

Volatile Compounds Analysis
The volatile compounds were extracted via headspace solid phase microextraction (SPME) equipped with a 75 µm Carboxen/PDMS StableFlex fiber (Supelco, Bellefonte, PA, USA) for 30 min at 80 °C and transferred to a gas chromatography inlet to desorb at 250 °C for 10 min. Then, they were analyzed by an Agilent 6890N GC coupled with an Agilent 5973i quadrupole mass detector (Agilent Technologies, Inc, Palo Alto, CA, USA) [10]. The separations were carried out by a HP-5MS capillary column (30 m × 0.25 mm, 0.25 µm film thickness, Agilent Technologies, Inc, Palo Alto, CA, USA).

Reducing Sugar and Ethanol Analysis
The reducing sugar was determined by the 3,5-dinitrosalicylic acid (DNS) assay [5]. The ethanol content was determined using a gas chromatograph (GC-2020 Plus, Shimadzu, Japan) equipped with a capillary column of WONDA CAP WAX (30 m × 0.25 mm × 0.25 µm) and detector of DET1 [9].

Non-Volatile Organic Acids Analysis
Organic acid analysis was carried out via HPLC (Waters 2695, Waters Corporation, Milford, MA, USA) equipped with an Aminex HPX-87H ion exchange column (300 × 7.80 mm, 9 µm film thickness, Bio-Rad Laboratories, Inc., Hercules, CA, USA) at a temperature of 60 • C [10]. The injection volume was 20 µL, and the mobile phase was 7 mmol/L H 2 SO 4 solution (pH 2.2) at the flow rate of 0.60 mL/min. The ultraviolet detector was set at 210 nm.

Volatile Compounds Analysis
The volatile compounds were extracted via headspace solid phase microextraction (SPME) equipped with a 75 µm Carboxen/PDMS StableFlex fiber (Supelco, Bellefonte, PA, USA) for 30 min at 80 • C and transferred to a gas chromatography inlet to desorb at 250 • C for 10 min. Then, they were analyzed by an Agilent 6890N GC coupled with an Agilent 5973i quadrupole mass detector (Agilent Technologies, Inc, Palo Alto, CA, USA) [10]. The separations were carried out by a HP-5MS capillary column (30 m × 0.25 mm, 0.25 µm film thickness, Agilent Technologies, Inc, Palo Alto, CA, USA).

Statistical Analyses
The differential fungi and volatile compounds in samples were discovered via partial least squares discriminant analysis (PLS-DA) with SIMCA software (version 14.1) (Umetrics, MKS Umetrics AB, Umea, Sweden), then the important volatile compounds with variable importance for the projection (VIP) > 1.00 were visualized by circus (http://mkweb.bcgsc. ca/tableviewer/visualize/ accessed on 2 August 2022). The fungal biomarkers in Qu were analyzed by linear discriminant analysis effect size (LEfSe) with the Kruskal-Wallis test (p < 0.05), Wilcoxon test (p < 0.05), and LDA threshold score > 2.00 on the Galaxy website (http://huttenhower.sph.harvard.edu/galaxy/, accessed on 2 August 2022). The chemical data were analyzed using IBM SPSS software (version 21) by one-way analysis of variance (ANOVA) with least significant difference (LSD) test at p = 0.05. Correlation analysis was performed using Spearman's correlation between fungal communities and volatile compounds on the online platform of OmicShare tools (https://www.omicshare.com/tools/Home/Soft/ ica2, accesses on 3 September 2022) and visualized via the Cytoscape software (version 3.9.1).

Fungal Profiles of Different Qu
A total of 25 fungi genera and 28 fungal species were found in different Qu from the rim of the Sichuan Basin ( Figures S1 and 3A). The dominant genera were Rhizopus, Lichtheimia, Saccharomycopsis, Saccharomyces, and Candida, which accounted for more than 96% of the total abundance of fungi in each Qu. At species level, Rhizopus arrhizus (R. arrhizus) was predominant in Qu, accounting for 75.1%, 65.0%, 49.8%, and 53.8% in GQ, DQ, AQ, and LQ, respectively. However, the sub-dominant fungi in Qu were obviously different, which were Candida glabrata (C. glabrata, 10.4%) and Saccharomycopsis fibuligera (S. fibuligera, 7.47%) in GQ, S. fibuligera (27.9%) and Rhizopus microsporus (R. microspores, 3.84%) in DQ, Lichtheimia ramose (L. ramose, 41.8%) and Saccharomyces cerevisiae (S. cerevisiae, 6.24%) in AQ, and R. microsporus (21.6%) and S. cerevisiae (20.0%) in LQ ( Figure 3A). These results indicated that the fungal structure in Qu was significantly affected by the different geographical environments around the Sichuan Basin. The PLS-DA further showed that fungal communities were obviously divided into four groups based on all fungi in Qu, where R 2 X and R 2 Y were 0.664 and 0.966, respectively, and Q 2 of the model was 0.911 ( Figure 3B). Among these species, 12 fungi were verified as biomarkers ( Figure  3C) and most dominant species belonged to biomarkers. R. arrhizus (75.1%), C. glabrata

The Chemical Indexes in TRW
Chemical indexes of rice wine often reflect the quality or type of TRW. Reducing sugars in GW and DW from the parallel fold ridge valley belt area (181.9-248.9 g/L) were obviously higher than AW and LW from the plateau area (121.6-128.1 g/L) ( Figure 4A). By contrast, the ethanol contents in TRW fermented by Qu from plateau (11.4-12.4%) were higher than that from the parallel fold ridge valley belt area (4.56-5.85%) ( Figure 4B). The composition and amounts of organic acids are also important contributors to the sour taste of rice wine [4,11]. A total of 7 organic acids were identified and quantified, including L-lactic acid, succinic acid, quinic acid, pyruvic acid, citric acid, tartaric acid and shikimic acid ( Figure 4C). GW had the highest content of total organic acids (1.14 g/L), then followed by AW (0.921 g/L) and by DW (0.669 g/L) and then LW (0.553 g/L). L-lactic acid was absolutely predominant in GW (85.8%), DW (69.3%), AW (67.5%), and LW (41.2%). Except for GW (1.65%), succinic acid was also dominant in DW (19.0%), AW (22.8%), and LW (40.5%). Additionally, tartaric acid was only detected in GW and DW, but shikimic acid in GW.

The Chemical Indexes in TRW
Chemical indexes of rice wine often reflect the quality or type of TRW. Reducing sugars in GW and DW from the parallel fold ridge valley belt area (181.9-248.9 g/L) were obviously higher than AW and LW from the plateau area (121.6-128.1 g/L) ( Figure 4A). By contrast, the ethanol contents in TRW fermented by Qu from plateau (11.4-12.4%) were higher than that from the parallel fold ridge valley belt area (4.56-5.85%) ( Figure 4B). The composition and amounts of organic acids are also important contributors to the sour taste of rice wine [4,11]. A total of 7 organic acids were identified and quantified, including Llactic acid, succinic acid, quinic acid, pyruvic acid, citric acid, tartaric acid and shikimic acid ( Figure 4C). GW had the highest content of total organic acids (1.14 g/L), then followed by AW (0.921 g/L) and by DW (0.669 g/L) and then LW (0.553 g/L). L-lactic acid was absolutely predominant in GW (85.8%), DW (69.3%), AW (67.5%), and LW (41.2%). Except for GW (1.65%), succinic acid was also dominant in DW (19.0%), AW (22.8%), and LW (40.5%). Additionally, tartaric acid was only detected in GW and DW, but shikimic acid in GW.

Volatile Compound Profiles in TRW
Volatile compound, the main source of aroma, is one of the main factors affecting the quality characteristics of alcoholic products [12]. A total of 162 volatile compounds were identified in TRWs, including 27 alcohols, 48 esters, 10 aldehydes, 21 ketones, 17 acids, 17 alkanes, and 22 other compounds ( Figure S2). Of these, 108 were found in GW, 92 in DW, 87 in AW, and 52 in LW ( Figure S3). Except for 31 volatile compounds which were commonly shared in all TRWs, 34, 21, and 20 unique volatile compounds were discovered in GW, DW, and AW, respectively ( Figure S3B). Interestingly, there were not any unique compounds in LW. The AW had the highest content of total volatile compounds with 51.7 mg/L, followed by DW with 47.9 mg/L, LW with 37.2 mg/L, and GW with 20.5 mg/L ( Figure 4D). Among them, alcohols contributed the greatest to DW (57.8%), AW (77.3%), and LW (58.2%), while there was only 25.2% in GW, in which acids were the most abundant, at 35.6%. In DW, AW, and LW, acids made up 2.45-6.68%. Esters were the second abundant volatile compounds, accounting for 19.1%, 13.2%, 10.3%, and 23.3% in GW, DW, AW, and LW, respectively. The ketones were higher in GW (10.6%) and DW (14.8%) than in AW (3.34%) and LW (6.43%). The aldehydes, alkanes, and other compounds made up 0.597-6.53%. Obviously, the different Qu led to different profiles of volatile compounds in TRWs.

Correlations between Fungi and Volatile Compounds
Fungi in Qu are closely linked to the formation of volatile compounds in TRWs [7]. A total of 22 fungal species were significantly correlated with 158 volatile compounds (|r| ≥ 0.7, p < 0.05) ( Figure 7A). Most biomarkers in Qu had positive correlation with almost all unique compounds in its corresponding TRW, which confirmed the contribution of biomarkers to the formation of specific flavor compounds.

Correlations between Fungi and Volatile Compounds
Fungi in Qu are closely linked to the formation of volatile compounds in TRWs [7]. A total of 22 fungal species were significantly correlated with 158 volatile compounds (|r| ≥ 0.7, p < 0.05) ( Figure 7A). Most biomarkers in Qu had positive correlation with almost all unique compounds in its corresponding TRW, which confirmed the contribution of biomarkers to the formation of specific flavor compounds.

Discussion
Fungi in Qu are considered a crucial influence on the quality of TRW [5,6,13]. As one of the main TRW production regions in China, the Sichuan Basin and its surrounding areas has formed its own characteristics of TRWs; unfortunately, the fungi in Qu and how they affect the specific compound formation in TRWs had never been elucidated. Therefore, analysis of fungi in Qu and further exploration of their correlations with chemical characteristics of TRW are essential to deeply understand the cause of the regional TRW formation around the Sichuan Basin.
The fungal community of Qu was distinctly different in various geographical locations [2,6,[14][15][16]. Thus, fungal biomarkers could distinguish different Qu [6,17]. In this study, fungal biomarkers in Qu were obviously different ( Figure 3C). GQ had the most biomarkers, which might be the reason for the most unique volatile compounds in GW ( Figure S3B). Additionally, C. glabrata and S. fibuligera, as the biomarkers of Qu from hill and mountain areas, were also widely found in Hong Qu from hill areas in Fujian province [18][19][20]. L. ramose, S. cerevisiae, and R. microspores, as biomarkers of Qu from plateaus, were commonly found in Da Qu around China [21]. In general, the fungal communities in Qu from the rim of the Sichuan Basin varied between geographic locations.
Responding to the differences in fungal communities, the chemical characteristics of TRWs had distinct difference. R. arrhizus can produce L-lactic acid [22]. It was the most abundant fungus in GQ, which might result in the large amount of L-lactic acid in GW ( Figure 4C). S. fibuligera can secrete amounts of extracellular hydrolases to hydrolyze polysaccharides into sugars [23,24], and then sugars are converted into ethanol by S. cerevisiae [25]. In GQ and DQ, S. fibuligera was in higher relative abundance, while S. cerevisiae was in low abundance ( Figure 3A), resulting in higher reducing sugars and lower ethanol content in GW and DW ( Figure 4A,B). C. glabrata could utilize both glucose and xylose to produce alcohols [2,26]. It had higher abundance than S. cerevisiae in GQ, which might be the main producer of alcohols in GW. L. ramose and R. microsporus can produce hydrolases to hydrolyze polysaccharides into fermentable sugars [27][28][29], which would then be efficiently converted to ethanol by S. cerevisiae. The high abundance L. ramose and R. microsporus in AQ and in LQ, respectively, together with the higher abundance S. cerevisiae, resulted in higher ethanol content and lower reducing sugars in AW and LW ( Figures 3A and 4A,B). Obviously, those predominant fungi in Qu led to different chemical characteristics of TRWs.
Volatile compounds were one of the main influences on the sensory characteristic of TRW [1]. Of the important volatile compounds (VIP > 1.00), phenethyl alcohol, was dominant in all TRWs (14.1-29.3%) ( Figure 6), which might play a vital role in the flavor of TRW, since it had quietly elegant rosy and honey aromas and was used as an important characteristic component of rice wine [5]. 3-Methyl-1-butanol, one of the major aliphatic alcohols [13] with a banana flavor [30], had been characterized as an important aroma compound in Qingke liquors from the Qinghai-Tibetan plateau [31], which also showed high proportions in TRWs except GW. It accounted for the highest proportion in AW (56.5%) and LW (36.0%) (Figure 6), suggesting that it might be the most important compound affecting the specific flavor of AW and LW from plateaus. However, in GW, the acetic acid with vinegar aroma was the highest, which might lead to the different flavor of GW from other TRWs with lower acetic acid content ( Figure 6). Additionally, unique important compounds in TRWs might be one of the reasons for the characteristic flavor of TRWs. For instance, formic acid in GW could endow it a sour, strong irritation flavor [32], and acetoin in DW could give it a pleasant buttery odor [33]. In LW, although there were no unique compounds, the esters such as ethyl palmitate, ethyl oleate, and ethyl linoleate with higher content might give a unique flavor to LW ( Figure 6). These esters have their own aromas, for example, ethyl palmitate has fruity, candy, and perfume-like aromas [8]; ethyl oleate has a faint, floral note; ethyl linoleate has a waxy, creamy, fatty, coconut odor [34]. Therefore, the dominant or unique important volatile compounds, which made important contributions to the distinctive flavor of TRWs, were a significant difference in TRWs fermented by different regional Qu.
The metabolism of fungi in Qu can produce complex compounds during TRW brewing, and a compound is usually the result of combined actions of fungi [8]. Of GQ biomarkers and important volatile component variable (VIP > 1.00), the acetic acid with the highest content in GW (6.04 mg/L) ( Figure 6) was positively correlated with R. arrhizus, R. pusillus, T. lanuginosus, and P. kudriavzevii ( Figure 7B, Table S1). Additionally, the unique important compounds of GW, namely formic acid, methyl 2-ethylacetoacetate, methyl pyruvate, and 1-methoxy-2-propanol, all had highly positive correlations with the GQ biomarkers R. arrhizus, C. glabrata, R. pusillus, T. lanuginosus, and W. sebi ( Figure 7B, Table S3), which directly certified the functions of these fungi on the formation of unique flavor in GW. Among those biomarkers, R. arrhizus, as the most abundant fungus in GQ, could hydrolyze racemic acetates to produce (R)-(+)-alcohols and acetic acid [35], which might be the main cause of the massive amount of acetic acid in GW ( Figures 3A, 6 and S2). C. glabrata, the second dominant fungus in GQ, could produce more esters and important terpene substances with the high activity of β-glucosidase [36], which might increase the floral and fruity aroma in GW. In DW, the unique important compounds acetoin and ethyl 5chloro-1,3,4-thiadiazole-2-carboxylate were positively associated with the DQ biomarkers S. fibuligera and M. indicus ( Figure 7B, Table S3). Acetoin, in particular, with a pleasant buttery odor and as a key aroma contributor in wines [33,37] showed the high ratio of 7.70% in DW ( Figure 6), which implied its key contribution to the special flavor of DW. The content of acetoin in Zaopei increased when S. fibuligera was inoculated in Sichuan-style Xiaoqu [23], suggesting that S. fibuligera with high abundance in DQ might also produce acetoin in DW by itself or by influencing other microorganisms. In AW, all unique important volatile compounds were positively correlated with L. ramose ( Figure 7B, Table S3). L. ramose, as one of the main functional microbes involved in the main flavor compounds' development in Daqu [38], was also predominant in AQ and was determined as a biomarker (Figure 3), which implied it was irreplaceable in generating the unique flavor of AW. In LW, the ethyl oleate, (S)-(+)-citramalic acid, and furaneol in higher proportions had positive correlations with S. cerevisiae; methyl isobutyrate and methyl acrylate in higher proportions were positively correlated with R. microspores; and ethyl palmitate (10.1%) and ethyl linoleate (3.35%) had positive correlations with S. cerevisiae and R. microspores (Figures 6 and 7B, Tables S1 and S2). S. cerevisiae, as an effective ethanol producer, is widely used in making wine, bread, and beer [17] and could produce secondary metabolites, such as amino acids, organic acids, and volatile flavor substances [39]. R. microsporus could secrete amylases to hydrolyze starch into sugar for further microbial utilization [29] and was positively correlated with volatile alcohols, acids and esters [18]. Moreover, it could produce lipase to synthesis ethyl oleate under solid-state fermentation [40], which might be the reason for the higher content of ethyl oleate in LW (Figures S2 and 6). Consequently, the LW biomarkers S. cerevisiae and R. microspores were the main contributory factors to the particular flavor production in LW. In short, most biomarkers in Qu showed strongly positive correlation to the important volatile compounds in corresponding TRW, thus proving their key role in the formation of the unique style of TRW.
Notably, the correlation analysis was based on statistical methods but did not imply that these compounds were directly produced by these fungal biomarkers, which might account for the positive association of some non-biomarkers with important volatile compounds (Tables S1-S3). Metagenomics, transcriptomics, metabonomics, and community reconstruction can be used to further investigate the functions of these fungi.

Conclusions
In the current study, R. arrhizus was found absolutely predominant in Qu from different regions in the rim of the Sichuan Basin; however, Qu from different areas had different fungal biomarkers. The biomarkers in GQ mainly were R. arrhizus, C. glabrata, R. pusillus, T. lanuginosus, and W. sebi, while S. fibuligera and M. indicus were the biomarkers in DQ, L. ramose in AQ, and R. microsporus and S. cerevisiae in LQ. Responding to fungal biomarkers, the chemical characteristics were also markedly changed in TRWs. GW and DW had higher reducing sugars and lower ethanol contents, which were opposite to AW and LW. Among important volatile compounds (VIP > 1.00), acetic acid (29.4%) and phenethyl alcohol (19.6%) were dominant in GW. However, they shifted to phenethyl alcohol (29.3%), 3-methyl-1-butanol (20.6%), acetoin (7.70%), and (R, R)-2, 3-butanediol (6.20%) in DW, 3-methyl-1-butanol (56.5%) and phenethyl alcohol (14.1%) in AW, and 3-methyl-1-butanol (34.0%), phenethyl alcohol (17.0%), and ethyl palmitate (10.1%) in LW. Moreover, biomarkers in Qu showed strongly positive correlation to the important volatile compounds in corresponding TRW. Among these, GQ and AQ biomarkers were positively correlated with all unique important volatile compounds in GW and AW, respectively. Meanwhile, GQ biomarkers had positive correlation with acetic acid, DQ biomarkers with acetoin and ethyl 5-chloro-1,3,4-thiadiazole-2-carboxylate, and LQ biomarkers with 16 important volatile compounds in LW. This study provides a more comprehensive and in-depth insight into the regional Qu from the rim of the Sichuan Basin and may improve the quality and flavor of TRW by regulating the key fungi in Qu.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods12030585/s1, Figure S1: Fungal communities in 4 regional Qu varieties at genus level. Results are shown as the mean from three biological replicates; Figure S2: Heatmap of volatile compound contents in rice wine fermented using different Qu. The contents of volatile compound were processed logarithmically; Figure S3: Types of volatile compound in rice wine fermented using different Qu: A. Stack graph of types of volatile compound classified into esters, alcohols, ketones, acids, aldehydes, alkanes, and other compounds; B. Venn diagram of types of volatile compound in different rice wine samples. Table S1: The correlation coefficient (|r| ≥ 0.7, p < 0.05) between fungi and important volatile compounds with ≥ 1 mg/L of total volatile compounds through all TRWs; Table S2: The correlation coefficient (|r| ≥ 0.7, p < 0.05) between fungi and important volatile compounds with < 1 mg/L of total volatile compounds through all TRWs; Table S3: The correlation coefficient (|r| ≥ 0.7, p < 0.05) between fungi and the unique important volatile compounds in respective TRW.
Funding: This work was funded by National Natural Science Foundation of China (31571935), Chengdu Science and Technology Project (2020-YF09-00007-SN), and Talent Introduction Grogram of Xihua University, China (Z211010).

Data Availability Statement:
Raw data obtained in this study are available from the corresponding author upon reasonable request.