Characterizing Relationship of Microbial Community in Xiaoqu and Volatiles of Light-aroma-type Xiaoqu Baijiu

Abstract

As an essential fermentative starter of light-aroma-type Xiaoqu Baijiu, Xiaoqu plays a crucial role in the formation of flavor characteristics. Herein, we explored the correlations between the microbiota and volatiles via polyphasic detection and analysis methods. Significant differences in microbial community structure were observed among three typical Xiaoqu from different regions of China. Moreover, the diversity of bacterial community was much higher than that of fungal community in Xiaoqu samples. In total, fourteen dominant bacterial and five dominant fungal genera were detected in Xiaoqu samples, of which Herbaspirillum, Weissella, and Rhizopus were the common dominant genera. The correlation network analysis showed that 74 different volatiles were correlated with 19 dominant genera. Importantly, the results suggested that the difference of locations and the diversity and relationship of microbiota in Xiaoqu truly affected the abundance and composition of volatiles in Xiaoqu Baijiu. Overall, our study provides scientific information on identifying the key microbiota contributing to the formation of Baijiu flavor.

Introduction

As one of the six well-known distillates (alongside whiskey, brandy, vodka, gin, and rum) in the world, Chinese Baijiu occupies a special position in traditional Chinese culture. And it was often regarded as an important present by Chinese people in the special days. The production of Chinese Baijiu is a typical spontaneous solid-state fermentation process (Xu et al., 2010), using sorghum, maize, wheat, rice and glutinous rice as the main raw materials. The technical operations process includes raw materials treatment, grains steeping, grains steaming, starter preparation, cooling and grains blending, fermentation, and distillation etc. As the fermentation catalyzer, starter is termed as Jiuqu or Qu which is usually called koji in Japan. In China, Qu is usually divided into Daqu (big starter), Xiaoqu (small starter) and Fuqu (Table 1). Qu is the main source of microorganisms and enzymes in the fermentation, and produces a variety of metabolites that contributed to the abundance and diversity of volatiles in Chinese Baijiu (Zheng et al., 2011; Zhu and Tramper, 2013; Zhu et al., 2015). These volatile compounds, including esters, alcohols, acids, aldehydes, as the flavor components affect the flavor and taste of Chinese Baijiu.

Table 1. Comparison of Daqu, Xiaoqu and Fuqu.

Qu	Raw materials	Dominant microorganisms	Main aroma of Baijiu	Characteristics
Daqu	barley, wheat, and/or peas	filamentous fungi, yeasts, bacteria, and actinomycetes	sauce aroma, strong aroma, and light aroma	big size, long production cycle (40–60 days), large amount of usage (< 20%), more flavour compounds and low rate of alcohol production
Xiaoqu	rice, rice bran or clay	yeasts, Rhizopus, Mucor and lactic acid bacteria	light aroma, rice aroma	small size, short production cycle (7–14 days), excellent saccharification and fermentation capacity, small amount of usage (0.5–1%), high rate of alcohol production
Fuqu	bran	Rhizopus or Aspergillus	light aroma, sauce aroma, strong aroma, and sesame aroma	short production cycle (5–10 days), high utilization, excellent saccharification and fermentation capacity, high rate of alcohol production

Baijiu can be classified as twelve types of aroma depending on the differences of Qu and liquor flavor in China. The most popular categories are sauce-aroma-type, strong-aroma-type, and light-aroma-type. As one of the most famous Chinese Baijiu, light-aroma-type Xiaoqu Baijiu uses Xiaoqu as the saccharifying and fermenting agent. Xiaoqu-making is the process of microbial fermentation on the solid raw materials, such as grains, wheat bran, rice flour or chaff. The production process of Xiaoqu can enrich various microorganisms from the surrounding environment (Zheng and Han, 2016). According to our study, the common types of microbes in Xiaoqu are yeasts, Rhizopus, Mucor, lactic acid bacteria and other unknown species. Among them, Saccharomyces cerevisiae is the most important alcohol-producing yeast and Wickerhamomyces anomalus performs well at ester-producing. It is also reported that the microbiota of Xiaoqu is also closely related to the formation of flavor characteristics of light-aroma-type Xiaoqu Baijiu (Jin et al., 2017; Tang et al., 2019). Thus, it is necessary to study the effect of microorganisms on volatile compounds.

Traditionally, microbes were isolated and identified mainly by culture-dependent and -independent techniques (Chen et al., 2014; Ricciardi et al., 2015). Some species have been studied extensively through pure or mixed cultures in order to highlight the effect on flavor components (Howell et al., 2006; Singhania et al., 2009; De Filippis et al., 2016). However, how microbiota affects the quality of fermented foods still remained unknown owing to technical limitations and the complexity of microbial influences on flavor components (Wu et al., 2013). In recent years, the research strategy on Baijiu microorganisms has gradually shifted from traditional culture to molecular biology and information biology. Revealing the microbial diversity in traditional fermentation has become a research hotspot by high-throughput sequencing (HTS) method (Ercolini, 2013). HTS combined with metatranscriptomics sequencing were used to identify the core microbiota included 10 fungal and 11 bacterial genera in Chinese sauce-aroma-type Baijiu production (Zhewei et al., 2017). It was investigated that the microbial community dynamics during the light-aroma-type Xiaoqu Baijiu fermentation process by culture-dependent and culture-independent method, and suggested Bacillus, Lactobacillius (L.pontis, L. acetotolerans) and Lactococcus (L. piscium) were the dominant bacteria (Dong et al., 2020). The previous studies showed that the community structure of bacteria was generally more complex than that of fungi in Xiaoqu. For instance, the previous study showed that the microbial communities were characterized and 17 bacteria and seven fungi were detected by HTS in Xiaoqu from three different regions in China, which illustrated that different flavours of Xiaoqu Baijiu were mainly caused by the microbial diversity in Xiaoqu from different regions (Wu et al., 2017). Although the structure of the microbial community in traditional fermented foods were largely illustrated, the correlation between the microbiota and flavor components in light-aroma-type Xiaoqu Baijiu remains unclear (Papagianni, 2014). Recently, correlation analysis was previously adopted to establish relationships between microorganisms and flavor components (Tang et al., 2019). Tang et al. combined polyphasic detection and analysis methods, such as HTS to explore the relationships between microbial diversity of Xiaoqu from four different regions in China and volatiles of Xiaoqu. It was reported that there were significant differences in microbial community diversity among Xiaoqu samples from different regions, and 24 dominant bacteria and seven dominant fungi in Xiaoqu were correlated with 20 different volatiles in Xiaoqu. This is a new and scientific perspective to better understand the relationship of microbiota and volatiles. However, the functional correlation between the dominant microbiota and important flavor components in light-aroma-type Xiaoqu Baijiu is rarely studied and needs to be further established.

In this study, we employed HTS method to examine the structure and function of microbes in Xiaoqu from three major light-aroma-type Baijiu production regions in China, including Sichuan Xiaoqu, Yunnan Xiaoqu and Hubei Xiaoqu. Sequencing results showed that Herbaspirillum, Weissella, and Rhizopus were the common dominant genera in three Xiaoqu samples. In addition, we combined Head Space solid phase microextraction (HS-SPME) and gas chromatography-mass spectrometer (GC-MS) to investigate the major volatile components in light-aroma-type Xiaoqu Baijiu produced by the three typical Xiaoqu, respectively. On this basis, we further explored the functional correlations between the dominant microbiota in Xiaoqu and important metabolites in the Baijiu through correlation analysis. This results provided us with novel insight into the microbial community responsible for light-aroma-type Xiaoqu Baijiu production, which will improve the understanding of relationship between microbes and flavor compounds.

Material and Methods

Sample collection    Three typical Xiaoqu made of clay and bran have been used to produce light-aroma-type Xiaoqu Baijiu by solid-state fermentation for decades. Xiaoqu samples were collected in biological duplicates from three typical light-aroma-type Xiaoqu Baijiu enterprises located in Qionglai city (Sichuan province, China), Yuxi city (Yunnan province, China), and Huangshi city (Hubei province, China), and labeled as SC, YN, and HB, respectively. All samples were transferred into sterile bags and stored at -20 °C.

DNA extraction, PCR amplification and amplicon sequencing    DNA extraction, PCR amplification and amplicon sequencing were conducted by Personsal Biotechnology Co., Ltd (Wuhan, China). Total Genomic DNA was extracted from 5 g of sample and purified using the E.Z.N.A.^®Soil DNA Kit (omega bio TEK, Norcross, GA, USA), following the manufacturer's instructions. The bacterial V3-V4 hypervariable regions of the 16S rDNA genes and the fungal ITS (internal transcribed spacer) were amplified by standard bacterial V3-V4 (F: 5′-ACTCCTACGGGAGGCAGCA-3′, R: 5′-GGACTA CHVGGGTWTCTAAT-3′), and standard fungal ITS1 (F: 5′-GGAAGTAAAAGTCGTAACAAGG-3′, R: 5′-GCTGCGT TCTTCATCGATGC-3′), respectively. PCR was carried out in 25 µL and the program as follows: Initial denaturation for 2 min at 98 °C, followed by 25–30 cycles of 98 °C for 15 s (denaturation), 55 °C for 30 s (annealing), 72 °C for 30 s (extension), and 72 °C for 5 min (final extension). A final hold step of 10 °C was performed. The number of cycles would be regulated according to the sample to ensure that qualified target bands were amplified with the least number of cycles. Sample-specific 7-bp barcodes (index sequence) were incorporated into the primers for multiplex sequencing. PCR amplicons were quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA), pooled in equal amounts, and then sequenced using an Illumina MiSeq platform (Illumina, San Diego, CA, United States) at Personsal Biotechnology Co., Ltd (Wuhan, China). The information of PCR amplification of samples including sequencing platform and index sequence was shown in Table S1.

Table S1. Information of PCR amplification of samples

Group ID	Sample ID	Primer name (Amplification area)	Sequencing platform	Barcode (Index sequence)
SC	SCB1	standard bacterial V3-V4	MiSeqPE250	ACACGTA
	SCB2	standard bacterial V3-V4	MiSeqPE250	ACAGATC
	SCB3	standard bacterial V3-V4	MiSeqPE250	ACAGCAT
YN	YNB1	standard bacterial V3-V4	MiSeqPE250	ACTACGT
	YNB2	standard bacterial V3-V4	MiSeqPE250	ACTAGAC
	YNB3	standard bacterial V3-V4	MiSeqPE250	ACTTGCA
HB	HBB1	standard bacterial V3-V4	MiSeqPE250	ACTCAAG
	HBB2	standard bacterial V3-V4	MiSeqPE250	ACTGTTG
	HBB3	standard bacterial V3-V4	MiSeqPE250	ACGTTCT
SC	SCF1	standard fungal ITS1	MiSeqPE250	TCAGAGA
	SCF2	standard fungal ITS1	MiSeqPE250	TGACTGA
	SCF3	standard fungal ITS1	MiSeqPE250	TGAGACT
YN	YNF1	standard fungal ITS1	MiSeqPE250	TGTCAGT
	YNF2	standard fungal ITS1	MiSeqPE250	TGTGTCA
	YNF3	standard fungal ITS1	MiSeqPE250	CAACTGT
HB	HBF1	standard fungal ITS1	MiSeqPE250	CATCAGA
	HBF2	standard fungal ITS1	MiSeqPE250	CATGTCT
	HBF3	standard fungal ITS1	MiSeqPE250	CACAAGT

GC-MS analysis of the volatiles in light-aroma-type Xiaoqu Baijiu    Baijiu produced by SC, YN, and HB using the light-aroma-type Xiaoqu Baijiu production process were abbreviated as SCB, YNB, and HBB, respectively. Volatile compounds produced by the different Xiaoqu using light-aroma-type Xiaoqu Baijiu process were determined by HS-SPME-GC-MS conducted on an Agilent 7890B GC system coupled with an Agilent 5977B MSD.

The sampling to HS-SPME-GC were treated as follows: 8 mL of sample and 10 µL of 2-ethyl hexanol (internal standard) were placed into a 20-mL headspace vial with 5 g of NaCl. The sample was then incubated at 75 °C for 5 min and extracted for 30 min at the same temperature. The MS operation conditions were as follows: The electron impact (EI) energy was set at 70 eV. Ion source temperature, MS Quadropole temperature, and GC–MS interface temperature were set at 230 °C, 150 °C, and 250 °C, respectively. SCAN mode was from 25 to 550 amu. The GC operation conditions were as follows: The samples were analyzed on an Agilent DB-FFAP column (length of 60 m, thickness of 0.25 µm, and diameter of 0.25 mm). The GC injection port temperature was set at 250 °C. The initial oven temperature was maintained at 50 °C for 2 min, and increased to 230 °C at 4 °C/min and maintained for 30 min. The carrier gas was high purity helium with flow rate of 1.42 mL/min. The split ratio was 10:1.

The unknown compounds were identified by matching with the NIST14.L spectrum database and their retention indices (RI) were calculated using C8–C30 n-alkanes. Relative quantities of each compound were determined by comparison of its peak area to the integrated of peaks of the total.

HTS and sequence processing    HTS were conducted by Illumina Miseq platform. All sequencing data have been submitted to the Sequence Read Archive (SRA) of the National Center for Biotechnology Information (NCBI) with accession number PRJNA608055. A series of wrong or questionable sequences would be existed in the raw HTS data. In order to ensure the reliability and accuracy of the analysis results, it is necessary to further evaluate the quality of the effective sequence extracted above, so as to obtain the sequences that can be used for subsequent analysis. Illumina-sequencing errors were minimized by removing reads with low quality scores (Q < 20). Forward and reverse sequencing reads obtained were then fused with each other by the program FLASH v1.2.7 (Magoc and Salzberg, 2011). Potentially suspicious reads and Chimeric reads were eliminated with QIIME v1.8.0 (Caporaso et al., 2010).

Bioinformatics and statistical analysis    Remaining reads were clustered into operational taxonomic units (OTUs) at 97% sequence similarity with VSEARCH in QIIME (Blaxter et al., 2005; Angly et al., 2016), and then were subjected to molecular taxonomic identification with the Greengenes database (Release 13.8) and the UNITE database (Release 7.0) to obtain microbial taxonomy information (DeSantis et al., 2006; Kõljalg et al., 2013). The OTUs whose sequencing reads were less than 0.001% in all the samples were also removed (Bokulich et al., 2013). Rarefaction curves were used to evaluate the sufficiency of the sequencing and alpha diversity indices were used to reflect the richness and the diversity of the microbial community was mainly calculated using the OTU table in QIIME. Beta diversity analysis was performed to investigate the structural variation of microbial communities across samples using non-metric multidimensional scaling (NMDS) by R packages (Ramette, 2007). Microbial community structure and flavor fingerprint were plotted using Origin (v8.5 or OriginPro 2019b). In order to explore the correlations between the dominant microbiota in Xiaoqu and important volatiles in Baijiu, Spearman's correlation coefficient was calculated via IBM SPSS Statistics (version 19.0), and visualized as a correlation network with |RHO| > 0.6 and P < 0.01 in Cytoscape (v3.6.1) (Shannon et al., 2003).

Results

The diversity of microbial community in Xiaoqu samples    According to statistics, a total of 362071 effective sequences of 16S rDNA gene and 596554 effective sequences of ITS were obtained (Table 2). All the effective sequences were clustered into operational taxonomic units (OTUs) at 97% sequence similarity, and there were 2154 OTUs of bacteria and 79 OTUs of fungi (Table 2). The rarefaction curves based on the OTUs suggested that the sequencing depth was enough to represent the microbial diversity of samples (Figure 1).

Table 2. Reads, OTUs and Alpha diversity indices.

Sample ID		Reads	OTUs	Alpha diversity indices
				Simpson	Shannon	Chao1	ACE
16S	SC	362071	2154	0.5193 ± 0.1471	2.240 ± 0.617	407 ± 99	409 ± 93
	YN			0.6813 ± 0.0338	4.020 ± 0.319	754 ± 115	766 ± 93
	HB			0.9379 ± 0.0327	6.130 ± 0.481	771 ± 98	787 ± 100
ITS	SC	596554	79	0. 0476 ± 0.0068	0.210 ± 0.026	21 ± 2	21 ± 2
	YN			0.0099 ± 0.0021	0.057 ± 0.012	21 ± 6	21 ± 6
	HB			0.5047 ± 0.3160	1.500 ± 0.848	43 ± 9	43 ± 9

Fig. 1.

Rarefaction analysis of the different Xiaoqu samples. (A) Bacteria; (B) Fungi.

The alpha diversity indices can reflect community diversity. Chao1 and ACE indices are focused on reflecting the richness of community, while Shannon and Simpson indices can reflect the richness and evenness. The Shannon index, Chao1 index, ACE index, and Simpson index were analyzed to detect the alpha diversity of bacterial and fungal community (Table 2). All alpha diversity indices in sample HB were the highest. It showed that the abundance and diversity of bacterial and fungal community in sample HB were the highest. The bacterial Chao1 in samples SC, YN and HB were 407 ± 99, 754 ± 115, and 771 ± 98, respectively. The bacterial ACE in samples SC, YN and HB were 409 ± 93, 766 ± 93, and 787 ± 100, respectively. However, the fungal Chao1 and ACE in samples SC, YN and HB were 21 ± 2, 21 ± 6, and 43 ± 9, respectively. It suggested that the abundance and diversity for bacteria in samples YN and HB were more similar and for fungi in samples SC and YN were more similar. On the whole, the diversity of bacterial community was much higher than that of fungal community. Moreover, NMDS analysis revealed that Xiaoqu samples were divided into three distinct groups consistent with identified taxa from three major light-aroma-type Baijiu production regions in China (Figure 2). It indicated that the microbial composition of samples was obviously different, especially between sample HB and others.

Fig. 2.

Results of NMDS sequencing of Xiaoqu samples SC (red and circle), YN (olive and square), and HB (blue and triangle). (A) Bacteria; (B) Fungi.

Phylogenetic structure of the microbial community in Xiaoqu samples    Figure 3 displays the taxonomic composition of microbial communities in genus levels. In total, fourteen dominant bacterial genera (Herbaspirillum, Staphylococcus, Lactobacillus, Weissella, Acetobacter, Pseudomonas, Bacillus, Sphingobacterium, Klebsiella, Stenotrophomonas, Pediococcus, Geobacillus, Enterobacter and Pelomonas) and five dominant fungal genera (Rhizopus, Apiotrichum, Candida, Aspergillus, and Trichosporon) were detected across all samples at average relative abundance of 1% or more, and a large number of groups were rare and had < 1% of all sequences (Figure 3). In addition, all samples were dominated by Herbaspirillum, Weissella, and Rhizopus (Figure 3). Obviously, the genus Rhizopus (accounting for 98.05%, 99.51%, and 60.97% in SC, YN, HB, respectively) was the most abundant fungal genus (Figure 3B). Except for Rhizopus, only the genera Herbaspirillum (accounting for 25.62%, 64.12%, and 4.65% in SC, YN, and HB, respectively) and Weissella (accounting for 3.14%, 3.22%, and 18.82% in SC, YN, and HB, respectively) were the common predominant genus (Figure 3A).

Fig. 3.

Taxonomic classifications of microbial communities at the genus level in Xiaoqu samples from three major light-aroma-type Baijiu production regions in China. The class “unidentified” refers to the group that could not be precisely assigned to any known taxonomic group at the genus level. (A) Bacteria; (B) Fungi.

Analysis of volatiles in light-aroma-type Xiaoqu Baijiu    A total of 87 major volatile components in Xiaoqu Baijiu were identified by GC-MS, and 59, 58, and 67 major components in SCB, YNB, and HBB, respectively. All volatiles were divided into six categories, including esters (28), alcohols (21), acids (7), aldehydes (8), ketones (5), and others (18). Heatmap analysis of flavor components in Xiaoqu Baijiu suggested that HBB was classified into an alone cluster, while the remaining YNB and SCB were divided into another cluster (Figure 4). As shown in figure 4, 3-methylbutanol was the most abundant volatile component in Xiaoqu Baijiu (accounting for 25.04%, 21.69%, and 21.12% in SCB, YNB, and HBB, respectively). Moreover, the most abundant esters in SCB, YNB, and HBB were ethyl caprylate (12.24%), ethyl caprylate (14.66%), and ethyl acetate (16.94%), respectively.

Fig. 4.

Heatmap of flavor components in SCB, YNB, and HBB with cluster analysis.

In addition, esters and alcohols were the main flavor components, accounting for a similar proportion in SCB, YNB, and HBB (Figure 5). Esters accounted for 59.25%, 61.68% and 57.39% in SCB, YNB, and HBB, respectively; Alcohols accounted for 37.84%, 35.20% and 35.77% in SCB, YNB, and HBB, respectively. In contrast, the contents of acids, aldehydes, ketones, as well as other flavors in HBB were higher than that in SCB and YNB. Such as, acids accounted for 0.59%, 0.66% and 2.37% in SCB, YNB, and HBB, respectively.

Fig. 5.

Proportions of volatile compounds (esters, alcohols, acids, aldehydes, ketones, and others) in SCB, YNB, and HBB, respectively.

Correlation analysis of dominant microbiota and between dominant microbiota and volatiles    The correlation network analysis showed that 74 different volatile components in light-aroma-type Xiaoqu Baijiu were correlated with fourteen dominant bacterial genera and five dominant fungal genera (Figure 6). According to the interactions between the dominant genera of bacteria and fungi and the volatile components of Xiaoqu Baijiu, all dominant genera in Xiaoqu samples could be divided into four categories. Firstly, the genus Pediococcus showed positive correlations with 21 volatile compounds which were some acids and other components with low relative content, and Pediococcus showed no significant correlations with other dominant genera; Secondly, six flavor components were associated with Klebsiella and Trichosporon, and the genus Klebsiella displayed negative correlation with Trichosporon. Klebsiella and Trichosporon mainly related to esters, such as ethyl caprylate, 2-phenethyl acetate, and ethyl caproate; Thirdly, eight genera including six bacterial genera (Herbaspirillum, Bacillus, Stenotrophomonas, Geobacillus, Pelomonas, and Pseudomonas) and two fungal genera (Rhizopus and Candida), displayed correlations with 20 volatile compounds which were main esters and alcohols, such as ethyl acetate, ethyl caprate, isopentyl acetate, 1-nonanol, and 1-octanol; Finally, eight genera including six bacterial genus (Staphylococcus, Weissella, Sphingobacterium, Enterobacter, Acetobacter and Lactobacillus) and two fungal genera (Apiotrichum and Aspergillus) showed correlations with 28 volatile compounds belonged to all categories of volatile components, especially alcohols, acids and esters, such as 3-methylbutanol, isobutyl alcohol, phenylethyl alcohol, acetic acid, methyl isocyanate, and 2,4-di-tert-butylphenol. Besides, the correlations of dominant genera Staphylococcus and Candida with other dominant genera were negative.

Fig. 6.

Correlation network diagram among dominant genera and between dominant genera and volatiles. Microbes and volatiles are respectively represented by square and circle modules; Positive and negative correlations among dominant genera are respectively represented by red and blue edges; Positive and negative correlations between dominant genera and volatiles represented by green and yellow edges with arrow.

Discussion

In the present study, we characterized and compared the microbial community in three typical light-aroma-type Xiaoqu, checking the effect of geography on microbial community in Xiaoqu, and revealed the effect of dominant genera of bacteria and fungi on volatile compounds in light-aroma-type Xiaoqu Baijiu. In total, fourteen dominant genera of bacteria (four, eight, and nine in SC, YN, and HB, respectively) and five dominant genera of fungi (two, one, and four in SC, YN, and HB, respectively) were detected in three typical Xiaoqu samples (Figure 3). In addition, all Xiaoqu samples were dominated by Herbaspirillum, Weissella, and Rhizopus (Figure 3). At the genus level, the diversity of bacterial community was relatively more complex than that of fungal community (Figures 2, 3), which was consistent with the previous studies (Gou et al., 2015; Wu et al., 2017; Tang et al., 2019). However, compared with the proportions of the dominant genera in different Xiaoqu samples, there were great differences in microbial community diversity among these samples. For example, the average relative abundance of Staphylococcus (accounting for 64.63%, 0.26%, and 0.19% in SC, YN, and HB, respectively) and Candida (accounting for 0.42%, 0.05%, and 13.5% in SC, YN, and HB, respectively) varied greatly (Figure 3). The bacterial and fungal community compositions in samples SC and YN were more similar, while visible differences were observed by comparison to sample HB. These differences may be caused by various factors such as raw materials, process parameters and microorganisms in external environment (Lu et al., 2016; Wang et al., 2016; Wang et al., 2018).

Prior research showed that the diversity and interaction of microbial community in starter could contribute to the abundance and diversity of volatiles (Zheng et al., 2011; Zhu and Tramper, 2013; Zhu et al., 2015). The flavor characteristic of light-aroma-type Xiaoqu Baijiu was mainly attributable to ethyl acetate, ethyl lactate, ethyl caprylate, and ethyl hexanoate (Guo et al., 2102). In our study, these esters in Baijiu also accounted for high relative abundance except ethyl lactate. The most abundant esters were ethyl caprylate, ethyl caprylate, and ethyl acetate in SCB, YNB, and HBB, respectively (Figure 4). The network analysis revealed that the genus Candida positively correlated with ethyl acetate (Figure 6); The genus Klebsiella showed positive correlations with ethyl caprylate and ethyl caproate (Figure 6). As shown in figure 6, the genera of Weissella, Staphylococcus, Lactobacillus, Acetobacter, Sphingobacterium, Klebsiella, Enterobacter, Apiotrichum, and Aspergillus positively correlated with acids, and the relative abundance of these genera in sample HB was also obviously higher than that in samples SC and YN. Moreover, the relative abundance of acids in HBB (2.37%) was obviously higher than that in SCB (0.59%) and YNB (0.66%) (Figure 5). For instance, the relative abundance of Lactobacillus was 0.0016%, 0.31%, and 39.94% in samples SC, YN, and HB, respectively. Studies suggested that Lactobacillus could produce lactic acid and degrade 2-nonenal (Zaunmüller et al., 2006; Vermeulen et al., 2007). Our studies also illustrated these genera contributed to the formation of acids (Figures 3–6). In other words, it may explain why the content of acids in HBB was clearly higher than that in SCB and YNB. Therefore, the diversity of microbial community in Xiaoqu from different regions truly resulted in the significant difference in abundance and composition of volatiles in Xiaoqu Baijiu. That is to say, we can construct the association network between the microorganisms and the flavor fingerprint by the methods of this study in order to improve the quality of Xiaoqu and finally improve the quality of Xiaoqu Baijiu.

As we know, the major origin of these microorganism in the stater is from the nature, which determines the microbial community for Baijiu fermentation. Different regions confer to the microbal discrepancy in starter. And the differences in microbiota induce the varation in the production of the key flavor commpontents. So our results indicated that geography could affect the factor characteristics of light-aroma-type Xiaoqu Baijiu through the microbial construction in Xiaoqu.

Till now, there are few studies on the microorganisms of Xiaoqu and flavor in Baijiu. Wu et al. (2017) surveyed the microbial communities in Xiaoqu from three different regions in China by culture-dependent and culture-independent methods. A total of 17 bacteria and seven fungi were detected and the dominant species in three Xiaoqu were completely different. They speculated the difference in flavours of Baijiu may be resulted mainly from the bacterial and fungal diversity in Xiaoqu from different regions. Tang et al. (2019) established relationships between microorganisms and flavor components in Xiaoqu from four different locations through correlation analysis on the basis of HTS and GC-MS. Staphylococcus, Weissella, Rhizopus, and Candida were the dominant genera. The results indicated there were significant differences in microbial community diversity among Xiaoqu samples from different regions, which resulted in a discrepancy of volatile profiles and interaction relationship among the genus in Xiaoqu. In our study, fourteen dominant genera of bacteria (four, eight, and nine in SC, YN, and HB, respectively) and five dominant genera of fungi (two, one, and four in SC, YN, and HB, respectively) were detected in three typical Xiaoqu samples. And all Xiaoqu samples were dominated by Herbaspirillum, Weissella, and Rhizopus. In addition to the common dominant genera reported, we also found other rare dominant genera that were significantly associated with flavor components, such as Pediococcus and Herbaspirillum. Moreover, we reported 87 flavor compounds in three different Xiaoqu Baijiu. To our knowledge, this is the first time to explore the functional relationships between microbes in Xiaoqu from three typical Chinese regions and flavor components of light-aroma-type Xiaoqu Baijiu by HTS, SPME-GC-MS, and correlation analysis.

In conclusion, the microbiota in the starter, Qu, play a crucial role in the production of violate components and the formation of flavor of light-aroma-type Xiaoqu Baijiu. The results suggested the diversity and coordination of microbial community in Xiaoqu starter truly affected the abundance and composition of volatiles of Xiaoqu Baijiu, and the regional factors influenced the microbial community diversity, which resulted in different flavor profiles. Furthermore, our studies provided scientific information on identifying the key microbiota contributing to the formation of Baijiu flavor, conducting the Xiaoqu manufacturing process as well as improving the quality and flavor of Xiaoqu Baijiu.

Acknowledgements    We gratefully acknowledge financial support from Jing Brand Research Institute, Jing Brand Co., Ltd.

References

•  Angly,  F. E.,  Heath,  C.,  Morgan,  T. C.,  Tonin,  H.,  Rich,  V.,  Schaffelke,  B.,  Bourne,  D. G., and  Tyson,  G. W. (2016). Marine microbial communities of the Great Barrier Reef lagoon are influenced by riverine floodwaters and seasonal weather events. Peer J., 4, e1511.
•  Blaxter,  M.,  Mann,  J.,  Chapman,  T.,  Thomas,  F.,  Whitton,  C.,  Floyd,  R., and  Abebe,  E. (2005). Defining operational taxonomic units using DNA barcode data. Philos. Trans. R. Soc. B-Biol. Sci., 360, 1935-1943.
•  Bokulich,  N. A.,  Subramanian,  S.,  Faith,  J. J.,  Gevers,  D.,  Gordon,  J. I.,  Knight,  R.,  Mills,  D. A., and  Caporaso,  J. G. (2013). Quality-filtering vastly improves diversity estimates from Illumina amplicon sequencing. Nat. Methods, 10, 57-59.
•  Caporaso,  J. G.,  Kuczynski,  J.,  Stombaugh,  J.,  Bittinger,  K.,  Bushman,  F. D.,  Costello,  E. K.,  Fierer,  N.,  Pena,  A. G.,  Goodrich,  J. K.,  Gordon,  J. I.,  Huttley  G. A.,  Kelley  S. T.,  Knights  D.,  Koenig  J. E.,  Ley  R. E.,  Lozupone  C. A.,  McDonald  D.,  Muegge  B. D.,  Pirrung  M.,  Reeder  J.,  Sevinsky  J. R.,  Turnbaugh  P. J.,  Walters  W. A.,  Widmann  J.,  Yatsunenko  T.,  Zaneveld  J., and  Knight  R. (2010). QIIME allows analysis of high-throughput community sequencing data. Nat. Methods, 7, 335-336.
•  Chen,  B.,  Wu,  Q., and  Xu,  Y. (2014). Filamentous fungal diversity and community structure associated with the solid state fermentation of Chinese Maotai-flavor liquor. Int. J. Food Microbiol., 179, 80-84.
•  De Filippis,  F.,  Genovese,  A.,  Ferranti,  P.,  Gilbert,  J. A., and  Ercolini,  D. (2016). Metatranscriptomics reveals temperature-driven functional changes in microbiome impacting cheese maturation rate. Sci. Rep., 6, 21871.
•  DeSantis,  T. Z.,  Hugenholtz,  P.,  Larsen,  N.,  Rojas,  M.,  Brodie,  E. L.,  Keller,  K.,  Huber,  T.,  Dalevi,  D.,  Hu,  P., and  Andersen,  G. L. (2006). Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl. Environ. Microbiol., 72, 5069-5072.
•  Dong,  W.W.,  Yang,  Q.,  Liao,  Y.X.,  Liu,  Y.C.,  Hu,  Y.L.,  Peng,  N.,  Liang,  Y.X., and  Zhao,  S.M. (2020). Characterisation and comparison of the microflora of traditional and pure culture xiaoqu during the baijiu liquor brewing process. J. Inst. Brew..
•  Ercolini,  D. (2013). High-throughput sequencing and metagenomics: moving forward in the culture-independent analysis of food microbial ecology. Appl. Environ. Microbiol., 79, 3148-3155.
•  Gou,  M.;  Wang,  H.,  Yuan,  H.,  Zhang,  W.,  Tang,  Y., and  Kida,  K. (2015). Characterization of the microbial community in three types of fermentation starters used for chinese liquor production. J. Inst. Brew., 121, 620-627.
•  Guo,  J.H.,  Yan,  X., and  Fan,  W.L. (2012). Aroma compounds of fresh distillates from different runs of fermented grains in Chinese light aroma type liquor. Sci. Technol. Food Ind., 33, 52-1317.
•  Howell,  K. S.,  Cozzolino,  D.,  Bartowsky,  E. J.,  Fleet,  G. H., and  Henschke,  P. A. (2006). Metabolic profiling as a tool for revealing Saccharomyces interactions during wine fermentation. FEMS Yeast Res., 6, 91-101.
•  Jin,  G.,  Zhu,  Y., and  Xu,  Y. (2017). Mystery behind Chinese liquor fermentation. Trends Food Sci. Technol., 63, 18-28.
•  Kõljalg,  U.,  Nilsson,  R. H.,  Abarenkov,  K.,  Tedersoo,  L.,  Taylor,  A. FS,  Bahram,  M.,  Bates,  S. T.,  Bruns,  T. D.,  Bengtsson-Palme,  J.,  Callaghan,  T. M.,  Douglas,  B.,  Drenkhan,  T.,  Eberhardt,  U.,  Dueñas,  M.,  Grebenc,  T.,  Griffith,  G. W.,  Hartmann,  M.,  Kirk,  P. M.,  Kohout,  P.,  Larsson,  E.,  Lindahl,  B. D.,  Lücking,  R.,  Martín,  M. P.,  Matheny,  P. B.,  Nguyen,  N. H.,  Niskanen,  T.,  Oja1,  J.,  Peay,  K. G.,  Peintner,  U.,  Peterson,  M.,  Põldmaa,  K.,  Saag,  L.,  Saar,  I.,  Schüßler,  A.,  Scott,  J. A.,  Senés,  C.,  Smith,  M. E.,  Suija,  A.,  Taylor,  D. L.,  Telleria,  M. T.,  Weiß,  M., and  Larsson,  K. (2013). Towards a unified paradigm for sequence-based identification of fungi. Mol. Ecol., 22, 5271-5277.
•  Lu,  Z.M.,  Liu,  N.,  Wang,  L.J.,  Wu,  L.H.,  Gong,  J.S.,  Yu,  Y.J.,  Li,  G.Q.,  Shi,  J.S., and  Xu,  Z.H. (2016). Elucidating and regulating the acetoin production role of microbial functional groups in multispecies acetic acid fermentation. Appl. Environ. Microbiol., 82, 5860-5868.
•  Magoc,  T. and  Salzberg,  S. L. (2011). FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics, 27, 2957-2963.
•  Papagianni,  M. (2014). Curr. Biochem. Eng., 1, 2-8.
•  Ramette,  A. (2007). Multivariate analyses in microbial ecology. FEMS Microbiol. Ecol., 62, 142-160.
•  Ricciardi,  A.,  Guidone,  A.,  Ianniello,  R. G.,  Cioffi,  S.,  Aponte,  M.,  Pavlidis,  D.,  Tsakalidou,  E.,  Zotta,  T., and  Parente,  E. (2015). A survey of non-starter lactic acid bacteria in traditional cheeses: culture dependent identification and survival to simulated gastrointestinal transit. Int. Dairy J., 43, 42-50.
•  Shannon,  P.,  Markiel,  A.,  Ozier,  O.,  Baliga,  N. S.,  Wang,  J. T.,  Ramage,  D.,  Amin,  N.,  Schwikowski,  B., and  Ideker,  T. (2003). Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res., 13, 2498-2504.
•  Singhania,  R. R.,  Patel,  A. K.,  Soccol,  C. R., and  Pandey,  A. (2009). Recent advances in solid-state fermentation. Biochem. Eng. J., 44, 13-18.
•  Tang,  Q.,  He,  G.,  Huang,  J.,  Wu,  C.,  Jin,  Y., and  Zhou,  R. (2019). Characterizing relationship of microbial diversity and metabolite in sichuan Xiaoqu. Front. Microbiol..
•  Vermeulen,  N.,  Czerny,  M.,  Gänzle,  M. G.,  Schieberle,  P., and  Vogel,  R. F. (2007). Reduction of (E)-2-nonenal and (E, E)-2, 4-decadienal during sourdough fermentation. J. Cer. Sci., 45, 78-87.
•  Wang,  J.,  Zhong,  Q.,  Yang,  Y.,  Li,  H.,  Wang,  L.,  Tong,  Y.,  Fang,  X., and  Liao,  Z. (2018). Comparison of bacterial diversity between two traditional starters and the round-koji-maker starter for traditional cantonese chi-flavor liquor brewing. Front. Microbiol., 9, 1053.
•  Wang,  Z.M.,  Lu,  Z.M.,  Shi,  J.S., and  Xu,  Z.H. (2016). Exploring flavour-producing core microbiota in multispecies solid-state fermentation of traditional Chinese vinegar. Sci. Rep., 6, 26818.
•  Wu,  H.,  Zhang,  S.,  Ma,  Y.,  Zhou,  J.,  Luo,  H., and  Yang,  J. (2017). Comparison of microbial communities in the fermentation starter used to brew Xiaoqu liquor. J. Inst. Brew., 123, 113-120.
•  Wu,  Q.,  Chen,  L., and  Xu,  Y. (2013). Yeast community associated with the solid state fermentation of traditional Chinese Maotai-flavor liquor. Int. J. Food Microbiol., 166, 323-330.
•  Wu,  Q.,  Chen,  L., and  Xu,  Y. (2013). Yeast community associated with the solid state fermentation of traditional Chinese Maotai-flavor liquor. Int. J. Food Microbiol., 166, 323-330.
•  Zaunmüller,  T.,  Eichert,  M.,  Richter,  H., and  Unden,  G. (2006). Variations in the energy metabolism of biotechnologically relevant heterofermentative lactic acid bacteria during growth on sugars and organic acids. Appl. Microbiol. Biotechnol., 72, 421-429.
•  Zheng,  X.W.,  Tabrizi,  M. R.,  Nout,  M. R., and  Han,  B.Z. (2011). Daqu-a traditional Chinese liquor fermentation starter. J. Inst. Brew., 117, 82-90.
•  Zheng,  X.W. and  Han,  B.Z. (2016). Baijiu, Chinese liquor: history, classification and manufacture. J. Ethnic Foods, 3, 19-25.
•  Zhewei,  S.,  Hai,  D.,  Yan,  Z., and  Yan,  X. (2017). Unraveling core functional microbiota in traditional solid-state fermentation by high-throughput amplicons and metatranscriptomics sequencing. Front. Microbiol., 8, 1294.
•  Zhu,  D.W.,  Wu,  Z.Y.,  Luo,  A.M., and  Gao,  H. (2015). Characterization and detection of toxoflavin-producing Burkholderia in rice straws and Daqu for Chinese Maotai-flavour liquor brewing. J. Inst. Brew., 121, 290-294.
•  Zhu,  Y. and  Tramper,  J. (2013). Koji-where East meets West in fermentation. Biotechnol. Adv., 31, 1448-1457.