Bacterial community and functional prediction differences between white and red sour soup Authors: Runcheng Zhou, Wenjing Zhang, Kecan Chen, Zhiteng Lin, Xumei Yi, Jingwei Pan, Xin Chen, Anping Yang Published: 2026 ABSTRACT Sour soup is a traditional plant-based fermented food from Qiandongnan Prefecture, Guizhou Province, China. It is divided into two types: white sour soup and red sour soup. White sour soup is made by semi-anaerobic fermentation of oryza sativa var. glutinosa washing water in earthenware jars for 7–10 days. Red sour soup is prepared by crushing and mixing solanum lycopersicum, red capsicum annuum, and zingiber officinale with oryza sativa var. glutinosa washing water at a ratio of 3:1:0.5:5, followed by semi-anaerobic fermentation for 10–15 days. To date, no study has systematically compared the bacterial community and functional differences between the two types of sour soup. In this study, six samples of each type were analyzed by 16 S rRNA gene sequencing and PICRUSt2 functional prediction. The results showed that their community structures were significantly separated (p < 0.05). White sour soup was dominated by Proteobacteria (93.66%) and enriched with Lactobacillus (LDA = 3.63), Acetobacter (LDA = 5.67), and Clostridiales (LDA = 4.21). Red sour soup exhibited higher diversity, with Proteobacteria decreasing to 39.09% and co-existing with multiple other phyla including Chloroflexota (19.72%), and specifically enriched with Akkermansia (LDA = 4.26). Functionally, white sour soup showed higher predicted abundances in host circulatory system (FC = 3.26, p < 0.001) and host immune system (FC = 1.65, p < 0.01) pathways, whereas red sour soup showed higher abundance in the host digestive system pathway (FC = 0.46, p < 0.01). This study systematically reveals, for the first time, raw material-driven bacterial differentiation and provides a theoretical foundation for the functional development of traditional fermented foods.Graphical abstract Introduction Sour soup is a traditional plant-based fermented food originating from the Qiandongnan Miao and Dong Autonomous Prefecture in Guizhou Province, China, with a history spanning several millennia (Li et al. 2025; Liu et al. 2024). As an important component of the Miao and Dong culinary cultural heritage, this naturally fermented product provides a valuable model system for studying microbial ecology in traditional plant-based fermentation and has attracted increasing attention from the global scientific community. Sour soup is primarily divided into two distinct types, white sour soup and red sour soup, which fundamentally differ in raw material composition, processing techniques, and microbial fermentation trajectories (Li et al. 2024; Zhang et al. 2025a, b, c). White sour soup uses the rice washing water of glutinous rice (Oryza sativa var. glutinosa) or regular rice as the sole substrate. The rice washing water is filtered through double-layered gauze to remove rice bran and impurities and is then immediately transferred to traditional earthenware jars to undergo spontaneous anaerobic fermentation relying on naturally occurring lactic acid bacteria from the environment and raw materials (Wang et al. 2023a, b). Approximately 2 cm of clean water is added to the groove at the jar mouth to maintain an semi-anaerobic environment. During this process, organic acids and amino acids accumulate as the primary metabolites, imparting the characteristic sour and umami flavors to the product (Wang et al. 2020). A published study have indicated that regular consumption of white sour soup contributes to digestive health and the maintenance of acid–base balance (Zhang et al. 2025a, b, c). In contrast, the raw material substrate of red sour soup is more complex, consisting of fresh tomato (Solanum lycopersicum), red chili pepper (Capsicum annuum), and ginger (Zingiber officinale), which are mixed with glutinous rice washing water at a traditional volume ratio of approximately 3:1:0.5:5 (Zhou et al. 2022). The mixture is chopped into a homogenate and then co-fermented in earthenware jars under the same semi-anaerobic conditions for approximately 10 to 15 days. This extended fermentation period and diverse substrate composition enrich the product with organic acids, lycopene, minerals, and phenolic compounds (Lin et al. 2020). Previous literature evidence has indicated that red sour soup possesses probiotic-like functions, including promoting digestion, regulating gut microbiota, antioxidant activity, and immune-enhancing effects (Liu et al. 2024). Bacterial community structure serves as the core biotechnological driver of food fermentation systems, and its community assembly and functional metabolism directly determine the flavor quality, nutritional properties, and technological characteristics of fermented foods (Zhang et al. 2024; Stewart J et al., 2025). Therefore, understanding the microbiological characteristics of Sour soup is essential for the quality control and functional development of these traditional products. In recent years, 16S rRNA gene high-throughput amplicon sequencing has become the primary methodological approach for characterizing the microbiome of fermented foods. This technique enables a comprehensive, culture-independent profiling of bacterial community composition and diversity with relatively high resolution and throughput. Wang et al. (2020) characterized the bacterial community of Guizhou sour soup using Illumina MiSeq sequencing technology and elucidated the correlations between the identified microorganisms and volatile flavor compounds. Lin et al. (2020) investigated the role of bacterial communities in generating the characteristic aroma of fermented sour soup using similar molecular methods. Furthermore, Liu et al. (2024) documented the quality characteristics and microbial communities of individual components of separately fermented red sour soup. Zhang et al. (2025a, b, c) further examined the effects of different raw materials on bacterial community assembly and flavor compound profiles in fermented red sour soup. However, to date, there remains a lack of systematic comparative studies on the bacterial community structure and predicted functional potential between white sour soup and red sour soup. Based on the above background, samples of white sour soup and red sour soup were collected from Shihua Village, Bibo Town, Kaili City, Qiandongnan Prefecture, Guizhou Province (latitude 26.46°N, longitude 107.76°E). All samples exhibited excellent flavor and texture. We employed 16S rRNA gene sequencing technology to comprehensively characterize bacterial diversity, utilized linear discriminant analysis effect size (LEfSe) and partial least squares discriminant analysis (PLS-DA) to identify key differential bacteria, and applied PICRUSt2 to predict functional potential. These findings will help elucidate the microbial assembly patterns driven by different raw material substrates and fermentation environments, thereby providing a theoretical foundation for subsequent process optimization and functional mechanism studies of traditional plant-based fermented foods. Materials and methods Sample processing and collection Both the white sour soup and red sour soup were collected in Shihua Village, Bibo Town, Kaili City, Qiandongnan Miao and Dong Autonomous Prefecture, Guizhou Province, China (Latitude: 26.46° N, Longitude: 107.76° E). The raw material for white sour soup is rice-washing water from glutinous rice, with the water temperature at ambient conditions (approximately 15–20 °C). The rice-washing water is filtered through double-layer gauze to remove rice bran and impurities, then immediately transferred into a traditional earthenware jar for fermentation. The fermentation period lasts approximately 7–10 days (Wang et al. 2023a, b). As for red sour soup, the raw materials consist of local tomato, red chili pepper, and ginger. These three ingredients are mixed with glutinous rice-washing water at a traditional volume ratio of approximately 3:1:0.5:5, respectively. The mixture is chopped into a homogenate and then transferred into an earthenware jar for fermentation, with a fermentation period of about 10–15 days (Zhou et al. 2022). Both types of sour soup utilize traditional earthenware jars as fermentation vessels. After filling, the groove at the rim of the jar is filled with clean water to a depth of approximately 2 cm, thereby maintaining an semi-anaerobic environment within the jar (Wang et al. 2023a, b). The ambient fermentation temperature ranges from 18 to 25 °C, with a relative humidity of 65%–80%. During fermentation, the jars are placed in a dark location, protected from direct sunlight. After fermentation was completed, samples were collected from the middle part of each jar using a long-handled stainless steel sampling spoon that had been sterilized by high-temperature high-pressure autoclaving, and were immediately flash-frozen in a portable liquid nitrogen container (− 196 °C). Subsequently, the samples were transported to the laboratory and stored in an ultra-low temperature freezer at − 80 °C until genomic DNA extraction and 16S rRNA gene sequencing were performed. 16S rRNA gene library construction, quality control, and sequencing Library construction, quality control, and sequencing of the 16S rRNA gene were performed according to standard experimental protocols. Genomic DNA was extracted using either the cetyltrimethylammonium bromide (CTAB) method or the sodium dodecyl sulfate (SDS) method (Wang et al. 2024). The CTAB method was applied to polysaccharide-rich fermentation samples, whereas the SDS method was used for routine cell lysis. The integrity of the extracted DNA was assessed by 1.2% agarose gel electrophoresis. Using quality-verified DNA as template, the V3–V4 hypervariable region of the bacterial 16S rRNA gene was amplified with the forward primer (5′-ACTCCTACGGGAGGCAGCA-3′) and reverse primer (5′-GGACTACHVGGGTWTCTAAT-3′). The Polymerase chain reaction (PCR) reaction mixture (25 μL) consisted of 12.5 μL of 2 × Taq PCR Master Mix, 1 μL of forward primer (10 μM), 1 μL of reverse primer (10 μM), and 1 μL of template DNA (10–50 ng), with sterile double-distilled water added to reach the final volume. Three technical replicates were performed for each sample, and sterile double-distilled water was used as a negative control. The PCR cycling parameters were as follows: initial denaturation at 95 °C for 3 min, then 25 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 45 s, followed by a final extension at 72 °C for 10 min. The PCR products were verified by 2% agarose gel electrophoresis and purified using AMPure XP beads to remove primer dimers and non-specific amplification products (Xu et al. 2026). After purification, the PCR products were quantified by Qubit fluorometry, pooled in equimolar ratios, and used to construct the sequencing library with the Illumina TruSeq Nano DNA LT kit. Following fragment selection on 2% agarose gel electrophoresis, the library was quality-checked on an Agilent Bioanalyzer 2100 system. Qualified libraries were subjected to PE300 paired-end sequencing on the MiSeq platform. Bioinformatics analysis The 16S rRNA gene sequencing data were processed and analyzed using the QIIME 2 platform (version 2024.2). Raw paired-end reads were trimmed of primer sequences using Cutadapt (version 4.6), followed by sequence merging and quality control (q-score ≥ 20) using VSEARCH (version 2.22.1). After uniform length truncation, denoising was performed using the Deblur algorithm to obtain amplicon sequence variants (ASVs). Based on the ASV abundance table and representative sequences, taxonomic assignment was conducted using a Naive Bayes classifier against the Silva 138 database with a confidence threshold of 0.7 (Kan et al., 2026). Alpha diversity analysis was performed using a rarefied ASV table to calculate the Chao1 index, Shannon index, Simpson index, Faith’s phylogenetic diversity (Faith_pd) index, Pielou’s evenness (Pielou_e) index, and Good’s coverage index. Beta diversity analysis was conducted based on the Bray–Curtis distance matrix, followed by principal coordinate analysis (PCoA), with non‑metric multidimensional scaling (NMDS) applied for validation. Intergroup differences were evaluated using permutational multivariate analysis of variance (ADONIS), analysis of similarities (ANOSIM), and the multi‑response permutation procedure (MRPP). Differential species analysis was carried out using LEfSe (version 1.0), with a linear discriminant analysis (LDA) score threshold of 2.0 and 1,000 permutation tests, to identify biomarkers significantly enriched between groups. Functional potential prediction was performed using PICRUSt2 (version 2.6.3), which infers metagenomic functions based on 16S rRNA gene sequences (Qin et al. 2026). The predicted functions were mapped to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database to obtain a functional pathway abundance table at Levels 1–3. Functional differences were assessed using Welch’s t‑test or Student’s t‑test, followed by false discovery rate (FDR) correction using the Benjamini–Hochberg method, with a significance threshold of adjusted p < 0.05. Statistical analysis This study employed a completely randomized design. Six independent fermentation jar samples were collected for each of the white sour soup and red sour soup, resulting in a total of 12 biological replicates (n = 6 per group). All quantitative data are presented as mean ± standard deviation. Normality was assessed using the Shapiro–Wilk test, and homogeneity of variances was evaluated using Levene’s test. Comparisons of alpha diversity indices and functional pathway abundances between the two groups were performed as follows: the independent samples Student’s t‑test was applied for data that were normally distributed with equal variances; Welch’s t‑test was used for normally distributed data with unequal variances; and the Mann–Whitney U test was employed for non‑normally distributed data. Categorical variables were compared using Fisher’s exact test. Correlation analysis was conducted using Spearman’s rank correlation (for non‑normally distributed data) or Pearson correlation (for normally distributed data). Multiple comparisons were corrected using the Bonferroni method. Results Bacterial diversity analysis The Chao1 index, Shannon index, Simpson index, Pielou_e index, Observed_species index, Faith_pd index, and Goods_coverage index all reflect the alpha diversity within samples. Higher values of these indices indicate greater community richness in the samples and Fig. 1 illustrates the differences in these indices between the white and red sour soup. In the analysis of microbial community alpha diversity, the Chao1 and Observed_species indices were used to estimate community richness, while the Shannon and Simpson indices were employed to assess community diversity. The Faith_pd index was utilized to measure phylogenetic diversity, and the Pielou_e index was applied to evaluate species evenness. Additionally, Good’s coverage index was used to represent sequencing coverage. Significant differences were observed between the white and red sour soup samples in both bacterial richness and coverage (p < 0.05).Fig. 1Chao1 index, Shannon index, Simpson index, Pielou_e index, Observed_species index, Faith_pd index, and Goods_coverage index of white sour soup and red sour soup, respectively. “W” represents white sour sSoup, and “R” represents red sour soup. The same applies below. A p-value of less than 0.05 was considered to indicate a statistically significant difference Chao1 index, Shannon index, Simpson index, Pielou_e index, Observed_species index, Faith_pd index, and Goods_coverage index of white sour soup and red sour soup, respectively. “W” represents white sour sSoup, and “R” represents red sour soup. The same applies below. A p-value of less than 0.05 was considered to indicate a statistically significant difference Figure 2A shows that as the number of samples increased, the species accumulation curve for bacteria continued to rise, but the rate of increase gradually diminished. The increase in species number was nearly zero upon the addition of the 12th sample. This indicates that the samples included in this study covered almost all bacterial species present in the sour soup, and no significant increase in species number would be expected with further sample addition. Figure 2B illustrates the beta diversity of bacteria between the white and red sour soup. The stress value was 0.077, which is below the threshold of 0.2, indicating that the analysis is reliable. Under this stress level, the overlap between white sour soup and red sour soup was minimal, suggesting substantial differences in community composition between the two groups. Furthermore, ADONIS yielded a P-value of 0.003, indicating that the bacterial community composition differed significantly between white and red sour soup. ANOSIM produced a P-value of 0.005, demonstrating that the differences between groups were greater than those within groups, reflecting a high degree of community structure separation. MRPP also gave a P-value of 0.003, confirming that samples within the same group were more similar to each other.Fig. 2Species accumulation curve (A) and beta diversity analysis (B) of white sour soup and red sour soup, respectively Species accumulation curve (A) and beta diversity analysis (B) of white sour soup and red sour soup, respectively Bacterial composition and difference analysis Based on the bacterial community analysis results at the phylum, genus, and species levels, white sour soup and red sour soup exhibited distinctly different community structure characteristics (Fig. 3). At the phylum level, white sour soup exhibited Proteobacteria as the absolute dominant bacterial group, with a relative abundance of 93.66%, followed by Actinobacteriota (3.98%) and Firmicutes_D (1.15%), indicating a highly concentrated community structure (Fig. 3A). In contrast, red sour soup displayed significantly higher bacterial diversity. The proportion of Proteobacteria decreased to 36.10%, while other phyla such as Chloroflexota (19.72%), Nitrospirota_A (10.15%), Actinobacteriota (9.15%), and Acidobacteriota (5.41%) were present at relatively high abundances, presenting a more balanced pattern of multi-phylum coexistence.Fig. 3Top 10 bacterial phylum (A), genus (B) and species C levels of white sour soup and red sour soup, respectively Top 10 bacterial phylum (A), genus (B) and species C levels of white sour soup and red sour soup, respectively At the genus level, the dominant genus in white sour soup was Acetobacter (92.76%), followed by Actinophytocola (1.89%) and Lacticaseibacillus (0.91%), indicating a highly simple community structure with extremely high dominance (Fig. 3B). In contrast, red sour soup exhibited a completely different genus-level composition, with the main taxa including an unclassified genus of the family Dongiaceae (16.49%), CF-167 (15.30%), Nitrospira_C (10.15%), Pseudomonas_E (9.27%), and Mediterraneibacter_A (7.26%), with the “others” category accounting for 25.27%, further confirming the complexity and heterogeneity of its community composition. At the species level, Unidentified_Acetobacter occupied an absolutely dominant position in white sour soup (92.11%), followed by Actinophytocola oryzae (1.81%) and Unclassified_Lacticaseibacillus (0.77%), indicating an extremely simple community structure (Fig. 3C).. In red sour soup, the dominant species included an unclassified species of the family Dongiaceae (16.49%), CF-167 sp005879775 (15.30%), Unidentified_Nitrospira_C (10.15%), Unclassified_Pseudomonas_E (9.27%), and SIO2C1 sp010672925 (6.48%), with the “others” category accounting for 25.27%, indicating the presence of numerous low-abundance but highly diverse bacterial taxa in red sour soup. The Venn diagram analysis results showed that white sour soup and red sour soup shared 17 bacterial phyla, with white sour soup containing 4 unique phyla and red sour soup containing 2 unique phyla (Fig. 4A). At the genus level, the two types of sour soup shared 64 bacterial genera, with white sour soup containing 140 unique genera and red sour soup containing 47 unique genera (Fig. 4B). At the species level, white sour soup and red sour soup shared 61 bacterial species, with white sour soup containing 197 unique species and red sour soup containing 64 unique species (Fig. 4C).Fig. 4Venn diagram analysis of bacterial communities in white sour soup and red sour soup at the phylum (A), genus (B), and species C levels Venn diagram analysis of bacterial communities in white sour soup and red sour soup at the phylum (A), genus (B), and species C levels Differences in bacterial characteristics between the white and red sour soup samples were determined using the Kruskal–Wallis test. LEfSe analysis identified 13 significantly enriched biomarkers in the white sour soup samples (Fig. 5A). The core differential bacterial taxa were primarily fermentative bacteria, including the phylum Firmicutes, specifically the genus Lactobacillus (LDA = 3.63), class Bacilli (LDA = 3.67), and order Clostridiales (LDA = 4.21), as well as the genus Acetobacter (LDA = 5.67) from the phylum Proteobacteria and its higher taxonomic ranks. In contrast, 4 significantly enriched biomarkers were identified in the red sour soup samples (Fig. 5B), with the core differential bacterial groups being the order Azospirillales (LDA = 4.89) from the phylum Proteobacteria, and the genus Akkermansia (LDA = 4.26) from the phylum Verrucomicrobiota, along with their related taxonomic units.Fig. 5LDA scores of bacterial groups (A) and LEfSe cladogram showing taxonomic distribution (B) LDA scores of bacterial groups (A) and LEfSe cladogram showing taxonomic distribution (B) PICRUSt2-based functional potential prediction PCoA analysis of PICRUSt2-predicted functional genes revealed a clear separation between white sour soup and red sour soup in terms of predicted functional composition (Fig. 6A). Along the PCoA1 axis, red sour soup samples were mainly distributed in the negative region (ranging from -0.54 to 0.02), whereas white sour soup samples were concentrated in the positive region (ranging from 0.07 to 0.21), with almost no overlap between the two groups on PCoA1. Along the PCoA2 axis, red sour soup samples showed a relatively scattered distribution (ranging from -0.31 to 0.29), while white sour soup samples were relatively clustered (ranging from 0.09 to 0.12).Fig. 6PCoA based on the functional potential prediction (A) and KEGG functional category (B). Differences were considered statistically significant at p < 0.05 (n = 6) PCoA based on the functional potential prediction (A) and KEGG functional category (B). Differences were considered statistically significant at p < 0.05 (n = 6) In addition, a KEGG functional classification analysis was performed to predict the potential direct effects that the microorganisms present in the two types of sour soup may have on a human host's various biological pathways following the consumption of sour soup. Analysis based on KEGG functional classification revealed significant differences between the two types of sour soup in six predicted functional pathways related to host–microbe interactions (p < 0.05), among which four pathways showed higher predicted abundance in white sour soup and two in red sour soup (Fig. 6B). The differences were primarily concentrated in the categories of organismal systems and genetic information processing. Specifically, the predicted abundance of circulatory system-related pathways was significantly higher in white sour soup than in red sour soup (FC = 3.26, p < 0.001), and immune system-related pathways were also significantly higher (FC = 1.65, p < 0.01). Additionally, pathways associated with the nervous system (FC = 1.20, p < 0.05) and replication and repair (FC = 1.24, p < 0.05) exhibited higher predicted abundance in white sour soup. In contrast, red sour soup showed significantly higher predicted abundance of digestive system pathways (FC = 0.46, p < 0.01) as well as higher abundance of developmental and regenerative pathways (FC = 0.09, p < 0.05). Notably, although not reaching statistical significance, the cellular community–prokaryotes pathway showed higher predicted abundance in red sour soup (FC = 0.65). No significant differences were observed between the two types of sour soup in basal metabolic pathways such as amino acid metabolism, carbohydrate metabolism, energy metabolism, and lipid metabolism (p > 0.05). Discussion In this study, 16S rRNA gene sequencing combined with PICRUSt2 functional prediction was used to systematically compare the differences in bacterial community structure and potential functional characteristics between white sour soup and red sour soup from the Qiandongnan region of Guizhou Province. The results showed significant differentiation between the two types of sour soup in terms of bacterial diversity, community composition, and predicted functions. In terms of bacterial community structure, white sour soup exhibited a highly concentrated community, with Proteobacteria being absolutely dominant, and significantly enriched with Lactobacillus, Acetobacter, and Clostridiales. Lactobacillus, the core functional genus in traditional rice-based sour soup, lowers the pH to 3.0–4.0 through homo- and heterolactic fermentation, thereby inhibiting spoilage microorganisms, and synthesizes lactic acid, acetic acid, acetoin, and amino acid derivatives, imparting the characteristic sour and umami flavor to the soup (Megan R al. 2022; Garcia ALY et al. 2024). Acetobacter, a strictly aerobic acetic acid bacterium, oxidizes ethanol to acetic acid, and its high abundance suggests that the fermentation process may enter an aerobic oxidation stage in the later phase (Ali Z et al. 2025). This genus is generally recognized as safe (GRAS) microorganism (Reinoso ÁM et al. 2026; Liu et al. 2025). Notably, the presence of Clostridiales was detected in white sour soup. This order includes both toxigenic anaerobic spore-forming bacteria such as Clostridium botulinum and Clostridium perfringens, as well as non-pathogenic Clostridium butyricum. Although 16S rRNA gene sequencing cannot precisely distinguish pathogenic from probiotic species at the species level, the occurrence of Clostridiales suggests that the anaerobic environment of the fermentation system may favor sporulation (Yue et al. 2026). Previous studies (Sivamaruthi S B et al. 2018; Ruiz B A et al. 2024) have detected Clostridium species and their toxin genes in traditional fermented foods. Therefore, the safety of white sour soup requires further validation in the future through specific detection of botulinum toxin and enterotoxin-related genes. Unlike white sour soup, red sour soup exhibited higher bacterial diversity and a more complex community structure. Proteobacteria coexisted with multiple other phyla including Chloroflexota, Nitrospirota_A, Actinobacteriota, and Acidobacteriota, with significant enrichment of Azospirillales and the genus Akkermansia of the phylum Verrucomicrobiota. Akkermansia muciniphila is a recently highlighted bacterial species in gut microbiome research, possessing potential probiotic functions such as maintaining the intestinal barrier, regulating immunity, and improving metabolism (Xu et al. 2025). The probiotic effects of food-derived Akkermansia require the fulfillment of three prerequisites: accurate strain-level identification, validation of the viability of live bacteria in the gastrointestinal tract, and experimental confirmation of functional mechanisms through in vitro or in vivo assays (Ma et al. 2024). Therefore, the findings of this study only suggest that red sour soup is a potential source of this genus and should not be directly equated with health benefits. Compared with other fermented foods (Fan et al. 2025; Xu et al. 2024), the bacterial diversity level of red sour soup is higher than that of kimchi and Sichuan pickles, which are dominated by lactic acid bacteria, and is more similar to that of naturally fermented vegetable mixture systems. This high diversity may confer a broader metabolic potential, but it also increases the uncertain risk of the coexistence of opportunistic pathogens (Li et al. 2026). Functional prediction results based on PICRUSt2 showed significant separation between the two types of sour soup in terms of predicted functions. White sour soup exhibited higher predicted abundances in pathways related to the circulatory system, immune system, nervous system, and replication and repair, whereas red sour soup showed higher predicted abundances in pathways related to the digestive system and development and regeneration. The high abundance of Clostridiales and Bacilli in white sour soup coincides with the enrichment of the replication and repair pathway, as spore formation requires a large number of genes involved in DNA replication and repair. Meanwhile, certain clostridia (e.g., Clostridium butyricum) can produce butyrate, which has immunomodulatory and neuroprotective effects (Choi IS et al. 2024). The association of Akkermansia with the digestive system pathway in red sour soup is consistent with known functions of this genus in degrading mucin and promoting short-chain fatty acid production (Jian et al. 2023), although validation through metabolomics and culture-based isolation is still needed. Furthermore, no significant differences were observed between the two types of sour soup in fundamental metabolic pathways (amino acid, carbohydrate, energy, and lipid metabolism), indicating that regardless of the raw materials, the microbial communities of the sour soup systems can maintain the stability of core metabolic functions. This finding is consistent with reports on other traditional fermented foods (Wang et al. 2022; Lee et al. 2025). It should be noted that PICRUSt2 is an indirect functional prediction based on 16S rRNA gene sequences. Its results represent the metabolic potential of the microbial community rather than the actual level of gene expression, and are subject to limitations including database coverage, 16S copy number variation, and the presence of uncultured microorganisms. In summary, the novelty of this study lies in the first systematic comparison of bacterial community differences between white sour soup and red sour soup, revealing raw material-driven community differentiation, while identifying potential spore-forming related taxa in white sour soup and the enrichment of the genus Akkermansia in red sour soup. However, this study has several limitations. Firstly, the pathogenic and probiotic types of bacteria in white sour soup and red sour soup have not yet been distinguished. Secondly, the functional differences predicted by PICRUSt2 have not been validated by in vitro experiments. Thirdly, parallel detection of physicochemical indicators and toxin genes was not performed. Future research should employ multi-dimensional approaches including metagenomic sequencing, selective culture and isolation of key genera, metabolomics, and in vitro models to scientifically evaluate the safety and health value of traditional sour soups. Conclusion This study, using 16S rRNA gene sequencing combined with LEfSe differential analysis and PICRUSt2 functional prediction, systematically elucidated the significant differences in bacterial community structure and predicted functional characteristics between white sour soup and red sour soup from the Qiandongnan region of Guizhou Province. White sour soup exhibited Proteobacteria as the absolutely dominant phylum and was significantly enriched with fermentation-related taxa such as Lactobacillus, Acetobacter, and Clostridiales, presenting a highly concentrated community structure. In contrast, red sour soup displayed higher bacterial diversity, with the proportion of Proteobacteria decreasing to 36.10% and coexisting with multiple other phyla including Chloroflexota, Nitrospirota_A, and Actinobacteriota, and was specifically enriched with the genus Akkermansia of the phylum Verrucomicrobiota. PICRUSt2 functional prediction revealed that white sour soup had higher predicted abundances in pathways related to the circulatory system, immune system, nervous system, and replication and repair, whereas red sour soup showed higher predicted abundances in pathways related to the digestive system, development, and regeneration. However, no significant differences were observed between the two types of sour soup in fundamental metabolic pathways (amino acid, carbohydrate, energy, and lipid metabolism), indicating that their microbial communities can maintain the stability of core metabolic functions. These findings reveal the shaping effects of raw material substrates and fermentation environments on the microbial community assembly and functional potential of traditional plant-based fermented foods. White sour soup is dominated by lactic acid bacteria and acetic acid bacteria with a clear flavor formation mechanism, but the detection of Clostridiales suggests the possible presence of spore-forming related taxa in its anaerobic fermentation system, necessitating future safety assessment through toxin gene detection. Red sour soup exhibits higher bacterial diversity and is enriched with the genus Akkermansia, which has attracted considerable attention in gut health research. However, the probiotic functions of its food-derived strains still require confirmation through species-level identification, viability validation of live bacteria, and in vitro and in vivo functional assays. This study provides preliminary microbiome-based clues and a theoretical foundation for the safety evaluation, process optimization, and functional mechanism studies of traditional sour soups.