The following samples of buffalo milk and dangke were collected from three villages in Enrekang, South Sulawesi, Indonesia: P1 – Rogo Village, P2 – Rogo2 Village, and P3 – Sumbang Village. Raw buffalo milk was obtained from a fourth site (P4) in the same region. Micro-, small-, and medium-sized enterprises (UMKM) processed approximately 4 liters of fresh buffalo milk, yielding a total of 12 liters used in this study. Dangke preparation followed the traditional method used by the Enrekang community. Fresh buffalo milk was heated to 85 °C, and 12 mL of diluted papaya (Carica papaya) latex was added as a natural proteolytic enzyme to induce milk coagulation. Salt was added, and the mixture was gently stirred until visible curds were formed. The curds were then separated from the whey and molded using coconut shell-shaped containers until they solidified.

Both buffalo milk and dangke samples were collected directly from local producers who rear buffaloes and prepare dangke onsite. All samples were transported under cooled conditions (4 °C) to the laboratory for further DNA extraction and metagenomic analysis. This traditional method is known to yield white-colored dangke with predominant microbial activity from LAB, reflecting the rich microbial diversity typically found in fermented Indonesian food.

DNA was extracted using a modification of the method reported by (Purwantiningsih et al., 2025). The DNA concentration was quantified using dual measurement methods: spectrophotometric analysis using a NanoDrop instrument and fluorometric assessment using a Qubit system (Thermo Fisher Scientific, Waltham, MA, USA). Quality assessment included electrophoretic separation on an agarose gel with subsequent visualization and documentation using a Gel-Doc EZ imaging system (Bio-Rad Laboratories, Hercules, CA, USA). Samples (10 mL) were centrifuged at 2,500 rpm for 10 min, and the obtained pellet was washed with saline and rinsed with sterile distilled water. Genomic DNA was extracted from the pellet using a ZymoBIOMICS DNA Miniprep Kit D4300 (Zymo Research, Cambridge, UK). DNA quality was assessed using agarose gel electrophoresis and subsequent visualization using Gel-Doc EZ imaging (Bio-Rad Laboratories).

The 16S rRNA gene for identifying bacterial species was amplified using the 27F and 1492R primers following the method described by Kim et al. (2011). The 27F primer (5’- AGAGTTTGATCMTGGCTCAG-3’) and 1492R primer (5’- GGTTACCTTGTTACGACTT-3’) produced an amplification of approximately 1,500 bp. DNA amplification was conducted in a total volume of 25 μL consisting of 0.3 μL of bacterial DNA samples, 11.9 μL of NFW, 0.15 μL of forward primer, 0.15 μL of reverse primer, and 12.5 μL of KOD Multi & Epi (KME-101). PCR analysis was set as follows: initial denaturation at 95 °C for 1 min, followed by 35 cycles of denaturation at 95 °C for 15 s, annealing at 60 °C for 10 s, and extension at 72 °C for 10 s. Nanopore sequencing was performed using MinKNOW software version 23.04.5. Base calling was performed using Guppy version 6.5.7, with a high-accuracy model (Wick et al., 2019). The quality of the FASTQ files was visualized using NanoPlot and the quality of filtering was performed using NanoFilt (Coster et al., 2018).

The filtered reads were classified using a previously described centrifuge classifier (Kim et al., 2016). Bacterial and archaeal indices were constructed using the NCBI 16SRefSeq database (https://ftp.ncbi.nlm.nih.gov/refseq/TargetedLoci/). Downstream analyses and visualizations were performed using Pavian (https://github.com/fbreitwieser/pavian), Krona Tools (https://github.com/marbl/Krona), and R Studio using R version 4.2.3 (https://www.R-project.org/).

The electrophoresis results (Figure 1) demonstrated successful amplification of the 16S rRNA gene across all tested samples (P1–P3 and buffalo milk), as indicated by clear and distinct bands at approximately 1,500 bp. The DNA molecular marker in the leftmost lane served as a size reference, confirming that the amplified fragments consistently corresponded to the correct gene region across all samples.

Although minor DNA degradation was observed in some dangke samples, the amplicon integrity was sufficient for sequencing. All PCR products met the quality control standards required for high-throughput sequencing, confirming their suitability for downstream metagenomic analyses using the Oxford Nanopore system. The uniform band size across all samples indicated specific and consistent amplification of the 16S rRNA region.

Figure 2 shows that milk processing significantly influenced microbial diversity. P1 exhibited a less complex microbial ecosystem because of standardized or controlled fermentation using pasteurized milk or starter cultures. In contrast, P2 showed slightly higher microbial diversity, suggesting exposure to ambient microbiota and the use of raw milk. P3 demonstrated the highest microbial richness, including Clostridia, Actinomycetia, and Chitinophagia, indicating spontaneous fermentation in traditional dangke. The presence of Clostridia microorganisms plays a role in proteolysis and aroma formation, while the potential bioactivity of products and Chitinophaga (a class of the phylum Bacteroidetes) microorganisms plays a less-controlled sanitation, raw materials, or equipment surfaces. Buffalo milk (P4) exhibited the broadest taxonomic diversity, harboring classes such as Flavobacteria, Sphingobacteria, and Bacteroidia.

Amplicon sequence variant (ASV) analysis revealed the following key insights into microbial overlap. Three hundred and thirty-eight ASVs were shared among all four samples, indicating a core microbiome (Figure 3A). P3 had the highest number of unique ASVs (913), reflecting its microbial complexity. Five hundred and forty-seven ASVs were common to P1, P2, and P3 (Figure 3B); however, only 67 ASVs were shared exclusively between P1 and P2, suggesting a weak microbial overlap in the absence of P3. Three hundred and fifty-seven ASVs were shared between buffalo milk (P4), P1, and P2, with P4 containing 909 total ASVs, supporting its role as a microbial source (Figure 3C). Figure 3D–F presents the shared and unique ASV counts between groups. For instance, P2 had 803 unique ASVs compared to 491 for P1, indicating a greater environmental or compositional influence by P2.

The overlap between P3 and P4 was particularly high (e.g., 344 and 333 ASVs were shared), suggesting microbial transfer from raw milk into spontaneously fermented milk (Figure 3G–J). Figures 3H and 3I show the highest ASV richness (869 and 788 ASVs, respectively), confirming P3 as the most microbiologically rich sample (Figure 3H–I).

The implications of high-throughput sequencing and the high resolution generated by next-generation sequencing allow the identification of novel microbial taxa and environmental adaptation patterns. This highlights the microbial responses to different processing environments and reveals the role of raw milk microbiota in shaping fermented dairy ecosystems. These findings are vital for future culture development, probiotic strain isolation, and the optimization of traditional fermentation practices for commercial applications.

Taxonomic profiling revealed that LAB were the predominant microbial group in all analyzed samples. The phylum Bacillota (formerly Firmicutes) was identified as the most abundant taxonomic group, with Enterococcus faecium (45%) and Enterococcus durans (19%) showing particular dominance in the samples presented in Figure 4B and Figure 4D. Although Bacillota was also dominant (Figure 4C), its relative abundance was significantly lower (Figure 4A), suggesting sample-specific variation in microbial composition.

Notably, samples with the highest viable cell counts, particularly those shown in Figures 4B and 4D, contained heat-resistant LAB strains, such as E. faecium and E. durans. These bacteria are known to withstand thermal treatment and are well-adapted to dairy fermentation environments. Their presence is crucial not only for flavor and texture development but also for microbial safety and preservation.

Understanding the factors that influence shifts in the microbiome is essential because certain microbes contribute positively to sensory attributes, whereas others may generate spoilage metabolites. Given that microbial activity drives the biochemical transformation of fermented dairy products, knowledge of the underlying microbiota is critical for product quality and safety, especially for products intended for prolonged storage.

Principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity revealed notable differences in microbial community composition among the samples (Figure 6). In the left panel, the first two principal coordinates (PCoA1 and PCoA2) accounted for 78.37%, 56.12 %, and 22.25% of the total variation, respectively. In the right panel, PCoA1 and PCoA3 explained 70.49% of the total variation.

The PCoA plot exhibited distinct clustering patterns, where the spatial distances between points reflected the degree of microbial dissimilarity. Notably, the buffalo milk sample (P4) formed a separate cluster from all dangke samples (P1–P3), indicating a significantly different microbial profile in raw milk than in the fermented products. In contrast, the dangke samples clustered tightly together, suggesting relatively similar microbial community structures regardless of the production site. This similarity likely reflects shared fermentation processes or dominant microbial taxa.

The gray vectors shown in the plot represent the contribution of specific microbial features (e.g., bacterial genera or species) to the differentiation among samples. The length of each vector is proportional to its influence on community separation, with longer vectors indicating higher discriminatory power. These findings demonstrate that while fermentation tends to standardize microbial communities, the microbiota of raw buffalo milk remains distinct and more variable.

Principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity revealed notable differences in microbial community composition among the samples (Figure 6). As shown in the left panel, the first two principal coordinates (PCoA1 and PCoA2) accounted for 78.37%, 56.12%, and 22.25% of the total variation, respectively. As shown in the right panel, PCoA1 and PCoA3 explained 70.49% of the total variation.

The PCoA plot exhibits distinct clustering patterns, in which the spatial distances between points reflect the degree of microbial dissimilarity. Notably, buffalo milk sample P4 formed a separate cluster from all dangke samples (P1–P3), indicating a significantly different microbial profile in raw milk than in the fermented products. In contrast, the dangke samples clustered tightly together, suggesting relatively similar microbial community structures regardless of the production site. This similarity likely reflected shared fermentation processes or dominant microbial taxa.

The gray vectors shown in the plot represent the contribution of specific microbial features (e.g., bacterial genera or species) to differentiation among samples. The length of each vector was proportional to its influence on community separation, with longer vectors indicating higher discriminatory power. These findings demonstrate that, while fermentation tends to standardize microbial communities, the microbiota of raw buffalo milk remains distinct and more variable.

All three samples exhibited similar electrophoretic band characteristics, that is thick and well-defined bands. As previously reported by (Haris et al., 2003), DNA concentration significantly affects fragment visualization. Excessive concentrations produce thick bands, whereas insufficient quantities yield faint or undetectable bands (Table 1). Electrophoresis separates DNA fragments by molecular weight and charge, with migration rates dependent on the DNA size and the applied electric current (Novitasari et al., 2014).

Total DNA samples are composites of various bacterial genomes. Although universal 16S rRNA primers can “attach” to different bacterial DNA templates, they preferentially amplify dominant bacterial populations. This banding pattern reflects the microbial diversity in the dangke. This is in agreement with the findings of (Randazzo et al., 2010) that traditional cheeses have greater microbial diversity than industrial varieties. This difference stems from the use of raw milk and cross-contamination with potential pathogens (Delcenserie et al., 2014).

Gel electrophoresis is a fundamental technique in molecular biology that separates charged biomolecules, including nucleic acids and proteins, using a stationary colloidal matrix under an electric field (Rudge & Monnig, 2000). Figure 1 shows the presence of distinct bands on the agarose gel, confirming successful DNA amplification. The leftmost lane displays the DNA molecular weight marker (1 kb ladder) as a size reference. All sample lanes consistently exhibited amplified products of approximately 1,500 bp, corresponding to the expected size of the 16S rRNA amplicons.

Milk processing methods (such as pasteurization or raw, unpasteurized milk) significantly influence microbial richness and diversity, affect fermentation outcomes, and result in more diverse microbial taxa (Figure 2). P1 exhibited a less diverse or simpler microbial ecosystem, likely because of standardized or controlled fermentation using a starter culture or pasteurized milk. The taxonomic abundance was quantified by comparing the number of sequence reads classified into specific taxa. Data were visualized at the genus level to highlight the predominant microbial groups in each sample. Notably, high-depth metagenomic sequencing enables precise identification of novel species in traditional fermented dairy products. This capability provides critical data for developing culture biology, strain screening for optimized fermentation, and the industrial-scale production of fermented dairy products. Microbiome sequencing, using the 16S rRNA gene, is a powerful tool for assessing and quantifying the structure and diversity of microbial communities in ecosystems (Choi et al., 2020). 16S rRNA gene sequencing is widely used to characterize microbial taxonomic composition and phylogenetic diversity (Aßhauer et al., 2015).

The results for P2 suggest a slightly more diverse fermentation environment than for P1. The factors influencing diversity included the use of raw milk and exposure to ambient microbiota. The difference in bacterial diversity between dangke products may result from the use of different raw buffalo milk depending on where the dangke is produced. The raw materials differ markedly because their quality depends on the milking process and cross-contamination during production (McHugh et al., 2021). In addition, as previously explained by (Kamimura et al., 2019), variations in microbial profiles depend on processing technologies, storage conditions, seasonal fluctuations, and climatic factors.

P3 showed greater microbial diversity, including a significant presence of Clostridia, Actinomycetia, and Chitinophagia. This indicated that natural or fermentation processes and factors influence diversity, including the use of unpasteurized. The deacidification of cheese is supported by the genus Clostridia (Clostridium tyrobutyricum) during ripening determines the intensification of the metabolic activity of C. sporogenes and C. beijerinckii and, consequently, an increase in butyric acid fermentation (Le Bourhis et al., 2005).

The presence of Chitinophagia also suggests anaerobic or less stringent sanitation conditions during production. Specific bacterial populations remain viable throughout the fresh milk transportation and milking processes (Vithanage et al., 2017). These findings highlight the urgent need for improved hygiene practices, proper handling, and better storage conditions in traditional dangke production to reduce contamination and ensure consumer safety (Syah et al., 2025) Thermophilic bacteria are resilient and produce heat-stable enzymes that maintain their activity under high-temperature conditions. Several microbial species generate extracellular peptidases capable of withstanding pasteurization temperatures (Glück et al., 2016).

Buffalo milk (P4) exhibited the most diverse microbial profile, including Flavobacteria, Sphingobacteria, and Bacteroides. Raw milk naturally harbors a wide range of environmental- and host-associated microbes. Dairy-associated microbiota frequently exhibit proteolytic and lipolytic activities that contribute to the characteristics of the final product. According to a previous study by (Rizvi et al., 2021), psychrophilic microorganisms exhibit distinct thermal adaptation profiles, characterized by a minimum growth temperature of 0 °C, an optimal growth temperature of ≤15 °C, and a maximum threshold around 20 °C. These cold-adapted microbes produce enzymes with flexible molecular architectures and reduced structural rigidity(Jin et al., 2022). Certain psychrotrophic strains exhibit heat tolerance, and psychrotrophic microorganisms are important in the low-temperature preservation of foods (Erkmen & Bozoglu, 2016).

Venn diagram analysis (Figures 3A–J) revealed important patterns in microbial diversity and similarity among samples (P1, P2, P3, and buffalo milk [P4]). A range of 338–547 ASVs were shared across multiple samples, suggesting that common microbial communities are likely involved in basic functions, such as metabolism or fermentation. P3 and P3 consistently showed the highest microbial richness (913 ASV), with many similarities among the samples (547 ASV). This is because of the BAL content in dangke. LAB plays a pivotal role in product maturation through substrate metabolism, using residual lactose and other carbohydrates, citrate, peptides, and amino acids, ultimately generating volatile aromatic compounds. In addition, bacteria have bioprotective effects because of the production of bacteriocins and other antimicrobial metabolites (Ristagno et al., 2012). A previous study corroborated these findings (Marino et al., 2019), reporting a high prevalence of LAB species (particularly Lactococcus spp. and Weissella) in the traditionally produced buffalo milk product, Mozzarella, from local markets. The LAB Lactobacillus plantarum IIA-2C12 and Lactobacillus acidophilus IIA-2B4 show extraordinary proteolytic activity against casein and sarcoplasmic proteins (Afiyah et al., 2015). Microorganisms present in cheese influence its flavor profile by producing volatile compounds (Percival & Percival, 2017). The bacterial community structure in samples is influenced by several factors, such as raw milk quality, fermentation conditions, and the addition of inoculation cultures (Yang & Yu, 2019).

Microbiota diversity in dairy products originates from multiple sources, including teat surfaces, milking equipment, barn hygiene, and manure contamination. However, despite high-temperature treatment, a small population of microorganisms may remain in the final product (Quigley et al., 2013). The microbiome of cheese plays a vital role in determining its organoleptic and physicochemical properties, which affect its quality and safety (Yeluri Jonnala et al., 2018). Therefore, it is essential to implement additional control measures to reduce the transmission of pathogenic bacteria in dairy products. Metagenomic analysis was used to investigate diversity (Qu et al., 2024).

LAB constitutes a natural microbial community in milk, with significant populations of Lactobacillus, Streptococcus, and Enterococcus consistently reported in bovine milk (Yuan et al., 2022). The present study identified heat-resistant LAB strains, including E. faecium, which exemplifies a broad spectrum of thermotolerant LAB. These findings suggest that pasteurized milk is a valuable source for isolating heat-resistant LAB strains, indicating that in the presence of LAB, antimicrobial activity is also increased by antimicrobial peptides produced during fermentation (Hanifah et al., 2016). Proteomic analysis of two bacteriocin-producing E. faecium strains isolated from milk revealed significant antibacterial activity against Staphylococcus aureus and Listeria monocytogenes(Aspri et al., 2017). The bacteria were reported to exert bacteriocinogenic properties (Bagde & Vigneshwaran, 2019), making their antimicrobial metabolites valuable for the natural preservation of livestock-derived products (e.g., milk and meat). LAB, widely used to increase the functionality of various foods, is limited by its short lifespan. In addition, LAB are categorized as probiotics and can extend the shelf life and maintain the quality of products (Sulaiman et al., 2016). LAB isolated from Indonesian beef has been recommended as a preservative agent (Sihombing et al., 2015). There is a relatively high abundance of Lactobacillaceae in dry areas of cheddar cheese (Choi et al., 2020). The production process of traditional fermented milk products is relatively open and poorly controlled; microorganisms present in the production environment can easily enter containers and participate in the fermentation process (Melo Pereira et al., 2022). LAB isolates from dangke produced pH values, titratable acidity, and viscosity that were relatively good compared to commercial cultures and even tended to be better, specifically regarding the total number of LAB from fermented milk (Syah et al., 2024).

The microbiome of P1 reflected a controlled fermentation process using pasteurized milk, characterized by low microbial diversity, but a strong dominance of LAB. These conditions favor the growth of beneficial fermentative bacteria, such as Enterococcus faecium and Staphylococcus condimenti, while minimizing the presence of environmental or spoilage microbes, indicating a technological intervention to ensure product safety and consistency. The resulting phylogenetic tree exhibited limited stability, as indicated by low bootstrap values at specific nodes. Bootstrap values are displayed numerically on the phylogenetic branches to represent the confidence level. Higher bootstrap values indicate greater consistency in the clade arrangement (Simbolon & Aji, 2021). However, as (Chatrou et al., 2012) explained, despite their low bootstrap scores, phylogenetic trees provide more robust results than those of morphology-based clustering. 

Phylogenetic trees fundamentally depict ancestral relationships among species, genes, and organisms (Baum, 2008), and serve as critical tools for reconstructing evolutionary events, refining taxonomic classifications, and enhancing crucial insights into biodiversity. Branch length correlates with evolutionary divergence, with longer branches signifying greater genetic changes and more distant relationships between clades. Environmental factors affect the growth of microorganisms and, therefore, change the composition of the microbiota. Correlative analysis of the relationship between the metagenomic composition and environmental gradients can help elucidate the main ecological factors and principles of microbial community formation (Satoh et al., 2023).

The microbial profile of P2 revealed controlled fermentation using raw or minimally processed milk. LAB and environmental microbes reflect a less selective fermentation environment, possibly because of traditional processing methods, limited hygiene practices, or fermentation. Compared to P1, P2 exhibited greater microbial complexity. A phylogenetic tree of the microbiota was constructed using evolutionary-distance-based methods, specifically the neighbor-joining algorithm (Saitou & Nei, 1987). This method optimizes branch length estimation by selecting sequence joins that best reflect the actual evolutionary distances between sequences (Dharmayanti, 2011). Bacillus is a genus of gram-positive, rod-shaped bacteria belonging to the phylum Firmicutes (Setiaji et al., 2023). In P2, Bacillota clustered with Lactobacillaceae and Weissella and the facultative anaerobic Weissella confusa, which belongs to the family Leuconostocaceae(Salazar et al., 2018).

P3 is a traditionally fermented dangke product made from raw milk with minimal processing. It exhibited higher microbial diversity than P1 and P2 and was characterized by beneficial LAB and various environmental microorganisms. The detection of genera such as Shinella, Acinetobacter, Brevundimonas, and Mixta, many of which are environmentally derived, suggests limited hygiene control or exposure to environmental microbiota during processing (e.g., unclean equipment, ambient air, or manual handling). These findings highlight how the microbial community structure varies between samples, reflecting phylogenetic constraints and functional adaptations during dairy product processing. Non-random microbial assembly patterns that result in parallelism between host phylogeny and microbial similarity have been described as phylosymbiosis (Tang et al., 2021).

Sample P4 represented a microbiome shaped by raw buffalo milk. Its microbial diversity reflected this condition, with the dominance of Pseudomonadota and Bacteroidetes and the presence of families such as Caulobacteraceae, Xanthomonadaceae, Sphingobacteriaceae, Flavobacteriaceae, and Weeksellaceae, indicating a highly diverse and environmentally influenced community. Genera such as Shinella, Brevundimonas, Sphingomonas, and Pseudoxanthomonas, which are typically found in soil, water, and plant environments, suggest extensive environmental exposure during handling or storage. P4 (buffalo milk) was dominated by Pseudomonadota (formerly Proteobacteria), with Caulobacteraceae showing a strong phylogenetic affinity. These gram-negative bacteria display diverse morphologies (rods, spirals, and cocci) and metabolic strategies (phototrophic, heterotrophic, and chemolithotrophic) with varying oxygen requirements (Wangka et al., 2020).

The principal axis (PCoA1) explained 56.12% of the variation, while PCoA2 and PCoA3 explained 23.59% and 14.72% of the total variation, respectively. Samples P1 and P2 tended to cluster, indicating relatively similar microbiota compositions, while P3, P4, and buffalo milk separated, indicating differences in their microbial communities. Dominant ASV vectors (e.g. ASVs 199592, 53345, and 1352) were strongly associated with sample divergence along the PCoA axes, indicating that specific taxa significantly differentiated the microbiota profiles between samples. These results confirmed that the source and fermentation substrate (buffalo milk) significantly influenced the final product microbiome (p<0.05).

Based on Quantitative Insights into Microbial Ecology, clusters P1–P2 were associated with ASV vectors 199592 and ASV1296, which are annotated as Lactococcus (e.g. L. lactis), as well as ASVs 53345/ASV1352 (Streptococcus/Enterococcus), reflecting the dominance of homolactic LAB in the final product. Sample P3 extended along PCoA2 toward ASV 1281 (Weissella), consistent with the contribution of heterofermentative LAB to the formation of volatile compounds (e.g. acetoin/diacetyl). In contrast, P4 and buffalo milk samples were driven by ASVs 65913/33946/2756 (Staphylococcus/Bacillus/Pseudomonas/Acinetobacter), indicating the influence of environmental flora and aerobic microbes from raw materials/equipment. Overall, LAB taxa, particularly Lactococcus and Streptococcus/Enterococcus, were the primary drivers of microbiota composition convergence in dangke, while environmental taxa separated buffalo milk samples and their products. LAB dominance benefits human health by serving as a probiotic that supports digestive health. The antimicrobial properties of LAB in cheese whey are beneficial (Mutmainna et al., 2021).

In addition, kefir contains lactic acid bacteria (LAB) made from milk, colostrum, and a milk-colostrum mixture that has antimicrobial properties against S. aureus, S. typhimurium, E. coli, and P. aeruginosa (Setyawardani et al., 2020). These lactic acid bacteria are expected to inhibit the growth of pathogenic microbes and maintain the stability of the intestinal microflora of Indonesians (Astawan et al., 2012). These findings underscore how processing variables, particularly milk treatment, sanitation, and fermentation management, play critical roles in shaping the microbial community structure during dangke production. This highlights the potential of microbiome profiling as a tool for authenticating traditional food processes and optimising microbial composition to improve product quality and safety. Temperature, humidity, storage time, and contamination encourage bacterial growth (Mutmainna et al., 2023). These findings demonstrate that buffalo milk maintains a substantially distinct microbiota compared to dangke fermented dairy products. However, further research is required to elucidate the specific mechanisms governing bacterial activity.