After quality control filtration, 103, 078 ~ 140, 822 bacterial sequences were obtained with an average length of 162–376 bp. In total, 3,204 OTUs were generated using cluster analysis. In addition, 93, 715 ~ 121, 901 fungal sequences were obtained, with an average length of 105–396 bp. A total of 1,634 OTUs were generated using cluster analysis. The coverage of samples in each group was greater than 0.999, and p-values of the Chao1 index were 0.93 and 0.32 (Figure 1). Abundance and diversity of microbial communities can be reflected by alpha diversity (Schloss et al., 2011). The dilution curve, Shannon curve, and alpha diversity indices (Supplementary Figure S3 and Supplementary Tables S1, S2) indicated that the metagenomic sequencing data obtained were representative of the samples.

The Chao1 index of the bacteria in Group F was the highest among the groups (Figure 1A). The Chao1 index of fungi, indicating fungal richness, in Group E was the highest (Figure 1B). The number, evenness, and diversity of species in the samples are indicated by the Shannon and Simpson indices (Wang et al., 2018). The Shannon and Simpson indices for bacteria in Group F were the highest (Figure 1A), while these indices for fungi were both the highest in Group M (Figure 1B). In conclusion, the bacterial diversity of FG in the early stage of ultra-long fermentation was higher than that in the middle and later stages. The fungal diversity in the middle stage was higher than that in the early and late stages.

There were 110 bacterial OTUs in the different groups, accounting for 5.32% of total bacterial OTUs. The maximum number of bacterial OTUs (958) was highest in Group F, accounting for 46.30% of the total bacterial OTUs in the groups, indicating that Group F had the highest bacterial diversity (Figure 1C). There were 36 fungal OTUs in the different groups, accounting for 3.64% of the total fungal OTUs (Figure 1D). The maximum number of fungal OTUs in Group E was 724, accounting for 73.28% of the total fungal OTUs, indicating that Group E had the highest fungal diversity. The fungal OTU number in Group M was higher than that in Groups F and E. This indicated that the diversity of fungi in the FG at the middle stage was the highest among the different stages of the brewing process, which was consistent with the results of the Shannon and Simpson indices.

As shown in Figure 2A, the bacteria in FG could be classified into 13 phyla. The dominant phylum in FG was Firmicutes, and its relative abundance in different samples was more than 30.00%. In sample AE180, the relative abundances of Firmicutes and Proteobacteria were 39.82 and 59.73%. As shown in Figure 2D, the fungi in FG could be classified into six phyla. Ascomycota in FG was the dominant phylum, and its relative abundance in different samples was higher than 80.00%. At the end of ultra-long fermentation, the relative abundance of Ascomycota was 99.93%. Ascomycota is a key phylum that plays an important role in brewing different Baijiu styles, such as soy sauce (Song et al., 2017) and Luzhou-flavored Baijiu (Zhang et al., 2021). Comparing with the normal fermentation (60 days), the dominant phylum of bacteria in FG was changed to Firmicutes and Proteobacteria instand of Firmicutes, the dominant phylum of fungi changed to Ascomycota instand of Ascomycota and Basidiomycota during the ultra-long fermentation process (180 days). It reported that Ascomycota is the main fungus in FG of sauce-flavored, strong-flavored, and light-flavored baijiu, indicating this phylum is one of the key fungal microflora in the brewing of baijiu (Luo et al., 2023).

As shown in Figure 2B, the bacteria in FG can be classified into four main genera: Lactobacillus, Mitochondria, Bacillus, and Comamonas. The relative abundance of Lactobacillus in AF10, AF20, AM30, and AM45 was higher than 35.00%, and Lactobacillus was the dominant bacteria in FG. Lactic acid is an important flavor substance, which is helpful in increasing the thickness and reducing the irritability of Baijiu, and Lactobacillus is an important lactic acid-producing bacterium in FG (Wang et al., 2017). It is likely that the lactic acid in FG is mainly produced during the middle and late stages of the ultra-long fermentation period. The relative abundance of Bacillus was higher than 2.50% during the late stage of ultra-long fermentation. Bacillus has high temperature resistance, enzyme production, and fragrance production, and its metabolites include pyrazines, acids, methyl esters, and other flavor substances (Li et al., 2014). Comparing with the normal fermentation (60 days), the dominant genera of bacteria in FG was changed to Mitochondria and Lactobacillus instand of Comamonas and Lactobacillus. In addition, the dominant genera of fungi changed to Monascus and Penicillium instand of Malbranchea, Thermomyces, and Monascus during the ultra-long fermentation process (180 days).

As shown in Figure 2E, the dominant genera in the early and middle stages were Pichia (AF1, AF5, AF10, AF20, and AM30), and their relative abundances were higher than 58.00%. The relative abundance of Pichia was low during the middle (AM45, AM60, and AM90) and late stages of fermentation. Wickerhamomyces and Pichia play major roles in aroma production and metabolism during Baijiu fermentation (Wang et al., 2019). The relative abundance of Monascu was 72.04% in the later stages of ultra-long fermentation, indicating that Monascu was the dominant fungal genus. Monascus is a unique functional fungus that appears in FG and plays an important role in promoting the formation of esters during the fermentation of Baijiu (Xu et al., 2021, 2022).

Linear discriminant analysis effect size (LEfSe) is helpful in understanding the differences in microflora and microbial species between groups. As shown in Figures 2C,F, 29 bacterial genera were upregulated in the early stages, including Kocuria, Pantoea, Saccharopolyspora_virgula. Five bacterial and four fungal genera were upregulated during the middle stage. At the later stage, Bacteroides, Flavobacteriacea, and Mollicutes_RF39 were among the 16 upregulated bacterial genera. Among the seven fungal genera, Sporidiobolaceae, Rhodotorula, and Sporidiobolales were upregulated.

Tang et al. (2022) reported that the acidity and alcohol content increased followed the extension of fermentation time, which led to the death of aerobic microbiota, such as molds and most types of yeasts, but survival of anaerobes and facultative anaerobic microorganisms. In our study, amplicon sequencing was used and mainly analyzed the relative abundance of microbial community in different samples of FG at the genus level. Further, the strength of the results should be validated by the culture-dependent analysis, and we are planning to tackle these questions in future work.

As shown in Figure 3A, there were eight, nine, two, and two species of bacteria with an abundance greater than 1.00% in the early stages of the ultra-long fermentation. There was no common dominance of bacterial genera, and there were at least two bacterial genera with high abundance (more than 1.00%) that are important bacterial genera in FG, such as Lactobacillus and Bacillus. The relative abundances of Lactobacillus, Mitochondria, and Burkholderia were 34.02, 48.95, and 8.04% at the end of the ultra-long fermentation. These were the dominant bacterial genera in FG. In addition, Lactobacillus were identified as the most abundant bacteria during the ultra-long fermentation of CFB, which was consistent with the results revealed in previous studies (Pang et al., 2018). However, the significant difference between this work and previous studies was the prokaryotic microbial community structure in FG at different stages during the ultra-long fermentation process.

According to Figure 3C, there were two, two, one, and one species of fungi with relative abundances greater than 1.00% during the early stages of the ultra-long fermentation. There was no common dominance of fungal genera in FG, and at least two of the fungal genera with high abundance (more than 1.00%) are considered important in FG, such as Pichia and Monascus. In our research, Pichia, Monascus, and Aspergillus were considered to be the main fungal genera during different stages of the ultra-long fermentation. During the early stage, the relative abundances of Pichia reached 99.54%. However, at the late stage, the relative abundances of Pichia reduced to 0.12%, which indicated that Pichia played vital roles in the early stage of fermentation. Pichia can play a key role in aroma production and metabolism. Monascus plays an important role in promoting the generation of esters, whereas Aspergillus can produce a variety of enzymes, organic acids, and fatty acids to facilitate the generation of aromatic esters (Wang et al., 2019; Xu et al., 2021, 2022; Cheng et al., 2023). The relative abundances of Pichia were 58.33, 97.07, 99.54, and 96.65% in the early stages of the ultra-long fermentation, which may have been influenced by the stacking fermentation process (Wang et al., 2019; Cheng et al., 2022). The relative abundances of Monascus, Penicillium, and Talaromyces at the end of the ultra-long fermentation (AE180) were 72.04, 16.91, and 9.99%, respectively. This indicated microbial communities were dominated by esterification microorganisms at the late stage of the ultra-long fermentation.

Microbial association networks are predominantly used to clarify the assembly differences of community species caused by environmental differences or experimental treatments, with the aim of finding key bacterial groups or species that can leverage changes in community composition (Tang et al., 2023). The composition and function of the core microbial flora determine the style and quality of Baijiu (Wang et al., 2020; Shen et al., 2021). As shown in Figure 3B, Bacillus, Lactobacillus, and Mitochondria were positively correlated with other bacterial genera. Bacillus and Lactobacillus had pronounced negative correlations. Bacillus positively correlated with Comamonas and Mitochondria. As shown in Figure 3D, Pichia, Aspergillus, and Thermoascus were positively correlated with other fungal genera. There was a pronounced negative correlation between Pichia, Aspergillue, and Thermoascus. Monascus was positively correlated with Filobasidium and Aspergillus. Amplicon analysis revealed fundamental information about microbial succession and the positive correlations between the main microbiota structure and major endogenous factors. Furthermore, existence of correlations between the main microflora with important flavor metabolites (Song et al., 2017; Wang et al., 2019). In addition, these six genera are play important roles in the process of the major flavor metabolites and formation in fermented grains, indicating that these six genera were the main microflora in FG during the ultra-long fermentation period.

S. cerevisiae plays a dominant role in the succession of fungal communities (Liu et al., 2021; Shen et al., 2021). Microorganisms are constantly domesticated during ultra-long fermentation, such as Lactobacillus and Monascus, among which Monascus plays an important role in promoting the formation of esters such as ethyl caproate and ethyl acetate (Xu et al., 2021, 2022). The influences of environmental factors and the interaction among microorganisms led to differences in fungal species and abundance in FG, including Monascus, Penicillium, and Talaromyces. These factors also influenced bacteria in FG, such as Lactobacillus, Mitochondria, and Burkholderia.

Extraction and concentration processes were performed simultaneously using headspace solid-phase microextraction, which is a simple, rapid, and inexpensive technique. The required sample volumes are also relatively small (Wang et al., 2020). A total of 88 volatiles, including six alcohols, 43 esters, eight aldehydes and ketones, 13 acids, and 18 other compounds, were detected in FG using HS-SPME-GC–MS (Figure 4F and Supplementary Table S3). These volatile components were accurately characterized, and so many volatile components were detected may related to the time of ultra-long fermentation. Hu et al. (2021) reported that a total of 71 volatiles including 33 esters, 14 alcohols, 9 fatty acids, 5 phenols, and 10 other compounds were detected by HS-SPME-GC–MS in FG of strong-flavor Baijiu. As we know, the RI comparison is a method to identify unknown compounds by comparing the RI of an unknown compound and known compound. If their RI are similar, the unknown compound may be considered as the known compound (Chen et al., 2022). In our research, RI combination with MS as an auxiliary qualitative method (Supplementary Table S3), the results showed that most of the important aroma substances were formed in the early stages, and some, including ethyl caproate, acetic acid, and creosote, increased during the later stages of the ultra-long fermentation process (Figure 4G). The unique volatile content in FG increased sharply from days 1 to 30, and these compounds were mainly esters (Figures 4B,F).

The number of volatile compounds in FG in the later stage was lower than that in the early stage. The content of esters, alcohols, and others in the middle stage was the highest among the three different stages and was highest on the 30th day (Figure 4). Tang et al. (2022) identified 64 major volatile in fermented grains of light-flavor Xiaoqu baijiu by HS-SPME-GC–MS, which fermented for 98 days, and indicated that the contents of esters and alcohols increased, while a noteworthy decrease of the acids, aldehydes and ketones, and others contents was observed along with the fermentation time. In this study, the content of many volatiles assigned to esters, alcohols, and others showed significant differences during the ultra-long fermentation process for 180 days. Comparing with the normal fermentation (60 days), the content and proportion of esters and acids are increased during the ultra-long fermentation process (180 days). To our knowledge, the quality of baijiu is determined by the content and proportion of flavor compounds that distilled from FG, which are related to the variety and content of flavour compounds in FG. Herein, the appropriate fermentation time of FG is important to improve the quality of baijiu.

For bacteria (Figure 5A), Bacillus, Pseudomonas, Weissella, and Mitochondria were positively correlated with reducing sugar and starch, and negatively correlated with acidity and moisture. Lactobacillus was positively correlated with acidity and moisture, and negatively correlated with starch and reducing sugar content. Acidity had little effect on the composition and function of the bacterial flora. For fungi (Figure 5B), Pichia was positively correlated with reducing sugars and starch and negatively correlated with acidity and moisture. Aspergillus was positively correlated with reducing sugar and starch and negatively associated with acidity and moisture.

Reducing sugars, acidity, and temperature determine and maintain bacterial community changes during the fermentation process, and S. cerevisiae plays a dominant role in the succession of the fungal community (Liu et al., 2021). The p-values of the permutation test indicated that the effect of environmental factors on the composition and function of the fungal flora (pa = 0.152) was higher than that on the bacterial flora (pb = 0.003). It is well known that physicochemical properties (moisture content, acidity, residual starch, and sugar contents) are responsible for the microbial community changes that occur in microbial ecosystems (Jung et al., 2014). Luo et al. (2023) reported that total acidity and reducing sugar of FG played important roles in promoting the formation of core microbiota and succession of dominant taxa, which are similar to our research results. In addition, the more thorough fermented grains caused by the extended fermentation time resulted in the changes of these physicochemical indicators.

The physicochemical parameters of FG, such as starch and reducing sugars, are important factors affecting the composition and function of the bacterial and fungal flora. However, acidity had little effect on the composition and function of the bacterial flora. Previous studies demonstrated high concentrations of water, ethanol as well as high acidity could facilitate the growth of Lactobacillus in the brewing process (Liu and Miao, 2020). The high acid environment in the later stages of ultra-long fermentation formed a micro-environment that was conducive to the growth of Lactobacillus, which promoting the increase of acid in FG (Supplementary Figure S2A).

A total of 37.14 and 52.84% of the variation in metabolites was explained in groups by the top two RDA axes, indicating that the correlations between microbial communities and volatile components were strong (Figures 5C,D). This indicated that most of the volatile components, including alcohols, acids, and esters, were positively correlated with the dominant genera of FG, including Pseudomonas, Lactobacillus, Novibacillus, Pichia, and Thermoascus. Lactobacillus is regarded as the core functional microorganism responsible for increasing acidity, and can produce lactic acid, ethanol, and acetic acid through heterolactic fermentation (Song et al., 2017; Tang et al., 2019). S. cerevisiae can metabolize ethanol and provide acid resistance during the Baijiu brewing process. This plays an important role in the diversity of tastes and flavors with established qualities (Wang et al., 2019, 2020; Xu et al., 2022). Pichia is the main non-alcoholic yeast during the Baijiu brewing process and is predominantly used to form volatile compounds in Baijiu (Wang et al., 2021).

The production of volatile compounds, such as alcohols and esters, is related to the degradation of different sugars and amino acids by microorganisms, which can also lead to the generation of other volatile compounds such as furanones and pyrazines during fermentation (Lee et al., 2019). However, some microbes had the activity to coordinate with flavor-producing microbes to improve flavor compounds, but they are not flavor compound producers (Wang et al., 2019). The essence of baijiu-brewing is the process of microbial growth and accumulation of metabolites, and the synergistic effect among populations in FG is pivotal related to the flavor compounds and quality of baijiu. In addition, it reported that the environmental microbiota can drive microbial succession during fermentation, and flavor metabolism of microbial could subtly shaping the quality of baijiu (Luo et al., 2023). Meanwhile, our research results indicating that microbial succession influence by physicochemical parameters of FG, also. To our knowledge, CFB fermentation is a multifarious microecological fermentation system and open fermentation environment. Therefore, the changes in microbial succession and flavor metabolism during this brewing process require further study.

A total of 65.69 and 5.43% of the variation in metabolites was explained in groups by the top two RDA axes, indicating that the correlations between physicochemical parameters and volatile components were strong (Figure 5E). This indicated that most of the volatile components, including alcohols, acids, and esters, were positively correlated with physicochemical properties of FG. In our research, we found that the reducing sugar and starch of FG are positively correlated with alcohols, but negatively correlated with esters. In addition, the acidity of FG are positively correlated with esters. Base on the above dates, indicating that the changes of physicochemical parameters will influence the kinds or contents of volatile components in FG. Further, to improve the quality and aroma of baijiu by the control of physicochemical parameters of FG during fermentation process, is possible.

The microbial composition and environmental factors can take a great influence on community succession during fermentation of Baijiu (Tang et al., 2022), indicating that those two factors are the main driver that induced the succession of fermentation from the front to the end stage. To our knowledge, microbial community and physicochemical parameters of FG is interactional, which lead to the differences of microbial metabolisms including flavor compounds.

Network analysis was conducted to evaluate the effects of microorganisms on flavor substances. Figure 6A shows that Lactobacillus was the bacterial genus with the largest number of connections and that it was positively correlated with the anabolism of 48 volatile components, including ethyl caproate, butyric acid, and trimethyl pyrazine. Bacillus was negatively correlated with the anabolism of six volatile components. Weissella was negatively correlated with the anabolism of eight volatile components, including ethyl caproate, ethyl heptanate, and ethyl niacin. Bacillus subtilis and Bacillus licheniensis can metabolize and produce many flavor compounds that are essential to the quality of Maotai-flavor Baijiu (Li et al., 2014; Xu et al., 2022).

Pichia was the fungal genus with the largest number of connections, and it was positively correlated with the anabolism of 34 volatile components, including ethyl caproate, caproic acid, and 3-furfural. Moreover, it was negatively correlated with the anabolism of six volatile components, including isoamyl lactate, (Z)-ethyl valerate, and N-capric acid (Figure 6B). Pichia is an important aroma-producing fungus in the brewing of Baijiu, which can increase the enrichment of volatile flavors, such as acids and esters (Wang et al., 2021; Zhang et al., 2021; Huang et al., 2023).

In conclusion, Lactobacillus, Bacillus, Weissella, and Pichia were the core microbial genera involved in metabolizing the volatile components of FG. This provided abundant volatile components and precursors for CFB brewing and contributed to the style and aroma formation of CFB. Functional correlations between the core microbiota and important metabolites, such as volatile components, remain to be established and researched in the CFB brewing process (Song et al., 2017; Wang et al., 2019). In addition, considerable research is required to identify the correlations between microbial and volatile components in FG, especially to identify the functional microorganisms at species level, and we are planning to tackle these questions in future work to reveal the functional microorganisms in FG. In addition, we are playing to study on spatial heterogeneity of active microbial community in FG based on meta transcriptome and culture-dependent, and those works will be benefit for the explain and clear of the mechanism related to the changes of physicochemical parameters, microbial community and volatile components during ultra-long fermentation process of CFB.