Metagenomic Analysis of Thoracic Empyema Etiology Through Next-Generation Sequencing Enhances Conventional Culture Techniques

Abstract

Background

This study aimed to analyze the microbiome of thoracic empyema using metagenomic methods and compare the results with conventional culture methods to increase diagnostic accuracy and enhance antibiotic therapy.

Materials and Methods

This study involved 30 patients with thoracic empyema from hospitals in Riau Province, Indonesia. Pleural fluid samples were collected for culture analysis and identification using the Vitek 2 compact system and metagenomic analysis. Patient clinical data were also collected.

Results

Culture methods showed a 40.0% positive rate, with Gram-negative bacteria (Klebsiella pneumoniae and Pseudomonas aeruginosa) predominating. Metagenomics showed a 56.7% positive rate, identifying a more diverse microbiome, including fungi (29.4% abundance), other Gram-negative bacteria (26.5%), and anaerobic bacteria (22.5%). Comparison of the two methods showed 36.7% complete agreement and 23.3% partial agreement, with 40% disagreement, with a Kappa coefficient of 0.416 and P-value of 0.016 (P<0.050).

Conclusion

Metagenomic NGS offers significant advantages in detecting the microbiome of thoracic empyema, particularly fungi and anaerobic bacteria, which are often missed by conventional culture methods. This has the potential to improve diagnostic accuracy and optimize antibiotic therapy. Further research with larger sample sizes is needed.

Graphical Abstract

INTRODUCTION

Thoracic empyema is a global health issue that is accompanied by high mortality and morbidity rates, as it is defined by the accumulation of fluids in the pleural cavity [1,2,3]. Empirical treatment with broad-spectrum antibiotics is still extensively used due to the difficulty of providing appropriate antibiotics caused by the limitations of diagnostic methods in identifying the cause of thoracic empyema [4]. The sensitivity of the culture method is approximately 40–60%, despite being used as the standard for identifying the bacteria that induce thoracic empyema [5,6,7].

Metagenomic methods can identify pathogens within a few hours to accelerate diagnosis and treatment, thus decreasing mortality [3,8,9]. The entire genome of microorganisms, such as bacteria, viruses, and fungi, can be analyzed simultaneously by metagenomic methods using Next Generation Sequencing (NGS). This methodology provides comprehensive genetic information, including identifying specific strains and antibiotic resistance genes [10,11].

In Indonesia, identifying pathogens that cause empyema is still predominantly limited to culture methods, especially aerobic cultures. The applicability of molecular methods with NGS is still limited due to the influence of geographical factors and variations in antimicrobial sensitivity with microorganism patterns. Based on these challenges, the authors are interested in conducting research that analyzes microorganisms that cause thoracic empyema using the NGS method in Indonesia. The aim was to analyze the microorganisms causing thoracic empyema using conventional culture methods, metagenomics, and abundance concentration analysis of microorganisms.

MATERIALS AND METHODS

1. Study design and participants

The study was a cross-sectional study conducted in five hospitals in Riau Province, Indonesia. Participants who were acceptable for inclusion were those over 18 who met the diagnostic criteria for thoracic empyema and agreed to participate in the study. This study used a total sampling method for patients treated between September 1, 2023 and November 30, 2023, so all participants met the specified inclusion and exclusion criteria. The study did not include samples with a volume less than 3.5 mL and a DNA concentration less than 2 ng/µL post-extraction. Data on patient characteristics, clinical manifestations, and laboratory examination results were collected.

2. Ethics statement

This study received ethical approval from the Faculty of Medicine Ethics Committee, Universitas Riau, with letter number B/094/UN19.5.1.1.8/UEPKK/2023.

3. Conventional culture analysis

Culture analysis involved culturing empyema fluid on aerobic and anaerobic blood culture media, with bacterial identification conducted using the Vitek 2 (bioMérieux, Marcy-l'Étoile, Rhône-Alpes, France) compact system.

4. Next generation sequencing (NGS)

DNA extraction was conducted employing the QIAamp DNA Mini and Blood Mini kits (Qiagen, Hilden, Nordhein-Westfalen, Germany) following the product manual. DNA samples with a concentration of 50 ng in 10 µL were quantified using a Qubit 4 Fluorometer (Thermo Fisher Scientific, MA, USA). The samples were also pooled to achieve a total concentration of 50-100 fmoles in 10 mM Tris-HCl, pH 8.0 with 50 mM NaCl. Sequencing was conducted for ten hours using a MinION R9.4.1 flow cell within the GridION Mk1 system (Oxford Nanopore Technologies, Oxford, UK).

5. Metagenomic analysis

Metagenomic analysis included the filtration of all raw sequence reads and their demultiplexing via Minknow software to ensure they were sorted by sample barcode. Each sample had an assessment for total read count, average read length, and N50 before further analysis. Following evaluation, all reads for each sample received processing via the wf-metagenomic v2.2.0-g356d8a9 workflow for taxonomic labelling, using the Kraken2 NCBI database for 16S, 18S, 28S, and ITS as the default parameters. This workflow generated alpha diversity indices for each sample. Beta diversity metrics were generated using the R package Phyloseq. The results, which include diversity figures such as heatmaps, stacked bar plots, and PCA plots, were presented using ggplot2 and Matplotlib.

Samples with a total read count less than 30 and reads that did not correspond to species in the NCBI database were excluded from the analysis. After this, the relative abundance for each read in the appropriate samples was calculated, with reads exhibiting an abundance of less than 1% were excluded [10]. The formula used to calculate the percentage of abundance of microorganism X in sample Y is as follows [10]:

$$\frac{\text{reads of microorganism }\mathrm{X}\text{ in sample }\mathrm{Y}}{\text{total reads in sample }\mathrm{Y}} \times 100$$

The total abundance of bacteria in a sample equals 100. Similarly, to calculate the percentage of abundance of microorganism class X in group Y, the following formula is used:

$$\frac{\text{total reads of microorganism class }\mathrm{X}\text{ in group }\mathrm{Y}}{\text{total reads in group }\mathrm{Y}} \times 100$$

The total abundance of each bacterial class within each group equals 100.

The species identified by NGS were grouped into nine categories: anaerobes, Enterobacteriaceae, other Gram-negative bacteria, Staphylococcus aureus, Streptococcus pneumoniae, Streptococcus anginosus group, other Gram-positive bacteria, Mycobacterium, and fungi [10].

6. Statistical analysis

The etiological appropriateness of thoracic empyema was analyzed based on conventional culture and metagenomic methods, followed by statistical testing using the Kappa test. The criteria used to determine the level of agreement between culture methods and metagenomic methods are as follows [12]:

• 1. Complete agreement

  • 1) Both methods yield negative results.

  • 2) Both methods identify the same microorganisms, at least at the genus level.

  • 3) Cases where metagenomic sequencing detected mixed microbial sequences and conventional culture demonstrated polymicrobial growth with complete matching of all microorganisms at least at the genus level.

• 2. Partial agreement

  • 1) Only one microorganism is isolated in culture, while mixed sequences are obtained from the metagenomic method.

  • 2) The metagenomic method detects only one microorganism in a polymicrobial culture.

  • 3) When both methods detect polymicrobial infection with at least one overlapping genus but not complete concordance.

• 3. No agreement

  • 1) Both methods identify different genera.

  • 2) The culture method yields a negative result while the metagenomic method is positive.

  • 3) The metagenomic method yields a negative result while the culture method is positive.

RESULTS

1. Characteristics and clinical manifestations of thoracic empyema patients

This study analyzed 36 samples, excluding 6 owing to insufficient concentration. The predominant demographic of patients was male (76.7%) and aged 18-59 years (80.0%), with most cases identified as community-acquired empyema (93.3%). The causes comprised pneumonia (30.0%), pulmonary tuberculosis (16.7%), malignancy (16.7%), and unknown causes (26.7%). Risk factors included smoking (46.7%) and diabetes mellitus (10.0%). The predominant symptoms included dyspnea (63.3%) and cough (53.3%). Laboratory tests showed leukocytosis and elevated neutrophils, with most patients presenting turbid pleural fluid, normal protein levels, glucose levels were not decreased, and increased Lactate Dehydrogenase (LDH). Only 3.3% required intensive care unit (ICU) admission, while 46.7% underwent water-sealed drainage (WSD) without surgery. The mean length of stay was 11.6 days, with 80% recovering (Table 1).

Table 1. Characteristics and clinical manifestations of thoracic empyema patients (N=30).

Variable		N (%)
Age group, years
	18–59	24 (80.0)
	60–69	4 (13.3)
	≥70	2 (6.7)
Sex
	Male	23 (76.7)
	Female	7 (23.3)
Type of empyema
	Community-acquired	28 (93.3)
	Hospital-acquired	2 (6.7)
Etiology
	Pneumonia	9 (30.0)
	Pulmonary tuberculosis	5 (16.7)
	Malignancy	5 (16.7)
	Post-surgery	0
	Spontaneous esophageal rupture	0
	Unknown	8 (26.7)
Comorbidities/risk factors
	Smoking	14 (46.7)
	Poor oral hygiene	4 (13.3)
	Diabetes mellitus	3 (10.0)
	Hypertension	2 (6.7)
History of antimicrobial administration
	Antibiotics	2 (6.7)
	Anti-tuberculosis drugs	6 (20.0)
	No history	22 (73.3)
Clinical manifestations
	Fever	7 (23.3)
	Chest pain	5 (16.7)
	Shortness of breath	19 (63.3)
	Cough	16 (53.3)
	Productive cough	5 (16.7)
Turbid pleural fluid
	Yes	28 (93.3)
	No	2 (6.7)
Leukocyte count (/μL), median (IQR)		11,193 (8,230–13,750)
Neutrophils (%), mean±SD		76.6±11.2
Pleural fluid protein (g/dL), mean±SD		3.6±1.1
Pleural fluid glucose (mg/dL), median (IQR)		85.6 (58.0–93.5)
Pleural fluid LDH (U/L), median (IQR)		1,695 (575–3,054)
ICU admission
	Yes	1 (3.3)
	No	29 (96.7)
Water-sealed drainage treatment
	Yes	14 (46.7)
	No	16 (53.3)
Surgery
	Yes	0
	No	30 (100)
Length of hospital stay (days), mean±SD		11.6±7.2
Outcome
	Recovered	24 (80.0)
	Died	2 (6.7)
	Discharged against medical advice	4 (13.3)

Values are presented as n (%), unless otherwise indicated.

ICU, intensive care unit; IQR, interquartile range; LDH, lactate dehydrogenase; SD, standard deviation.

2. Microorganism pattern of thoracic empyema by culture method

The culture method identified bacteria in 12 of 30 samples, resulting in a positive rate of 40.0%. Among these, 33.3% were polymicrobial. 16 isolates were identified, including 68.8% Gram-negative and 31.2% Gram-positive bacteria. The most prevalent Gram-negative bacteria were Klebsiella pneumoniae (18.8%) and Pseudomonas aeruginosa (12.5%). Only one pathogenic bacterium, S. pneumoniae (6.3%), was identified, and no anaerobic bacteria or fungi were isolated (Table 2).

Table 2. Microorganism pattern of thoracic empyema by culture method.

Microorganism		Isolates, n (%)
Gram-negative bacteria		11 (68.8)
	Klebsiella pneumoniae	3 (18.8)
	Pseudomonas aeruginosa	2 (12.5)
	P. putida	1 (6.3)
	Burkholderia cepacia	1 (6.3)
	Acinetobacter baumannii	1 (6.3)
	Providencia rettgeri	1 (6.3)
	Pasteurella multocida	1 (6.3)
	Vibrio mimicus	1 (6.3)
Gram-positive bacteria		5 (31.2)
	Streptococcus pneumoniae	1 (6.3)
	S. alactolyticus	1 (6.3)
	Staphylococcus aureus	1 (6.3)
	S. epidermidis	1 (6.3)
	Enterococcus faecalis	1 (6.3)
Total		16 (100)

3. Microorganism pattern and abundance causing thoracic empyema based on metagenomics

The metagenomic method identified bacteria in 17 of 30 samples, resulting in a positivity rate of 56.7%. All samples showed mixed sequencing among several microorganism groups. Figure 1 shows the composition of these groups, while Table 3 describes their patterns and cumulative abundance. The fungal group had the highest abundance at 29.4%, as verified by ITS PCR in 6 samples, with DNA fragments of 700-1.300 bp. The Gram-negative bacteria group included 61 species identified in 9 samples, dominating 5 samples with an abundance of 26.5%. Anaerobic bacteria included 43 species identified in 9 samples, dominating 4 with an abundance of 22.5%. S. pneumoniae was identified within 3 samples, dominant in 2 with an abundance of 9.8%. Other groups included some species with lower abundances (Table 3). Figure 2 presents a heatmap showing the distribution of microbial intensity in thoracic empyema samples.

Figure 1. Microbial composition and relative abundance detected by metagenomic sequencing in each thoracic empyema sample.

Table 3. Microorganisms profile and abundance in thoracic empyema detected by metagenomic sequencing.

Microorganism groups	Number of species	Positive samples, n	Relative abundance (%)
Fungi	7	12	29.4
Other Gram-negative bacteria	61	9	26.5
Anaerobes	43	9	22.5
Streptococcus pneumoniae	1	3	9.8
Other Gram-positive bacteria	28	7	8.2
S. anginosus group	2	3	1.2
Enterobacteriaceae	8	3	1.0
Staphylococcus aureus	1	1	0.7
Mycobacterium	2	3	0.7

Figure 2. Heatmap depicting the distribution and relative abundance of microcroorganisms in thoracic empyema samples detected by metagenomic analysis.

4. Analysis of microorganism differences based on culture and metagenomic methods

The metagenomic method showed a higher positive rate for identifying causes of thoracic empyema (56.7% vs. to 40.0% for culture). Table 4 establishes the concordance among the methodologies: 36.7% overall concordance, 23.3% partial concordance, and 40.0% discordance. Complete concordance included both negative results and mixed sequences found by utilizing metagenomics. Differences occurred when metagenomics yielded positive outcomes while culture did not, and conversely. A Kappa coefficient 0.416 was observed, accompanied by a P-value of 0.016 (P<0.050). This indicates a moderate and statistically significant agreement between the NGS and culture results.

Table 4. Concordance of results of the culture method and the metagenomics method.

Criteria		n (%)
Complete concordance
	Both methods yielded negative results	11 (36.7)
Partial concordance
	One microorganism isolated by culture with mixed sequences detected by metagenomics	6 (20.0)
	Both methods detected polymicrobial infection with at least one overlapping genus	1 (3.3)
No concordance
	Different genera identified by both methods	3 (10.0)
	Culture-negative but metagenomic-positive results	7 (23.3)
	Metagenomic-negative but culture-positive results	2 (6.7)
Total		30 (100)

DISCUSSION

This study advances NGS metagenomic analysis to improve the diagnosis and treatment of thoracic empyema. This study described the demographic and clinical characteristics of patients and compared the diagnostic yield of metagenomic sequencing with conventional culture. The largest sample group included males aged 18–59, consistent with previous studies indicating that men are at higher risk of empyema [4,12,13,14,15]. A study by Yamazaki A, et al. showed that community empyema was the most common type reported (93.3%), with a higher incidence than hospital empyema [16]. Most patients (73.3%) had not received antibiotics before in this study and by study of Corcoran JP, et al. [3]. Smoking habits (46.7%) and diabetes mellitus (10.0%) are the most common risk factors in empyema cases [3,27]. Shortness of breath (63.3%) and cough (53.3%) are the most common symptoms [16].

The analysis of pleural fluid in this study showed 93.3% of samples were cloudy, with normal protein levels (median 3.6 g/dL), median glucose 85.6 mg/dL, not decreased, and increased LDH levels (median 1,695 U/L). The results correlate with the study by Barnes et al., which identified purulent and cloudy pleural fluid in 86.7% of positive cultures. Other empyema cases showed glucose levels below 40 mg/dL and LDH over 1,000 U/L [9,17]. The main management of empyema is the administration of antibiotics and evacuation of pleural fluid through chest tube drainage or WSD [9]. 3.3% of patients required ICU treatment and 46.7% underwent WSD in this study and in line with the study of Yamazaki A, et al. who reported most empyema patients received WSD (63.9%) and surgery (11.0%) [16].

The mean length of hospital stay was 11.6 days (range 1–35 days). Some other studies show the average length of stay ranges from 9 to 23 days [3,4,16]. The outcome in this study was that 80% of patients recovered, 13.3% were discharged against medical advice and 6.7% died. Efforts to reduce the incidence of death in empyema, the study of Brims JFH, et al. recommends surgery for cases that are unresponsive to antibiotics or haver delayed WSD insertion or persistent infection characterized by fever, leukocytosis, and increased inflammatory markers [18].

The positivity level of culture examination in this study was 40.0%, higher than Chen et al. at 26.7% and lower than Barnes et al. at 49.0% [5,19]. The low positivity may result from low microorganism concentration, prior antibiotic use, and difficulties in detecting fastidious microorganisms, including anaerobic bacteria, which require specific growth conditions [5,6,7,9,11,18,19]. Most patients (73.3%) had not received antibiotics prior to sample collection. Thus, antibiotic exposure was not dominant and is unlikely to have substantially influenced the overall diagnostic yield. Additionally, infections with multiple microorganisms can hinder identification, as faster-growing species may inhibit others [9]. Errors in specimen collection or transportation and inadequate agar plates can also affect results. Studies indicate that using blood culture tubes for pleural fluid can improve microorganism identification rates from 38.0% to 59.0% [3,9,20]. This study's culture method involved culturing samples within an hour of collection and using appropriate aerobic and anaerobic blood culture media.

In this study, 66.7% of samples were monobacterial, while 33.3% were polymicrobial, aligning with findings from Chan K et al. [16,21]. Of the isolates, 68.8% were Gram-negative and 31.2% Gram-positive, with K. pneumoniae (18.8%) and P. aeruginosa (12.5%) being the most common Gram-negative bacteria, consistent with Corcoran JP et al. and Marks et al. [3,4]. The study identified one fastidious bacterium, S. pneumoniae (6.3%), mirroring Jain et al., who reported 5% [22]. The distribution of Gram-negative and Gram-positive bacteria may vary by empyema classification and regional profiles [19]. The TORPIDS study indicated anaerobic bacteria were the primary contributors to empyema, but this study did not identify anaerobic isolates despite using appropriate media [10]. Fungal isolates were also not found due to the absence of fungal-specific media, although some yeast fungi were grown on aerobic blood culture bottles. Contrasting results were observed in other studies reporting S. pneumoniae and S. anginosus as common causes of community empyema, suggesting variations in bacterial patterns influenced by geography and antibiotic usage [15].

NGS metagenomic examination is a powerful method for identifying pathogenic microbes through their genetic sequences [19,23]. This technique offers faster turnaround times and higher accuracy than traditional culture methods [10]. In this study, the metagenomic positivity level was 56.7%, lower than Rosario I et al. at 81.7% and Nikolaos IK et al. at 79% [10,12]. Differences may arise from sample size, study duration, and extraction methods [24].

The study identified fungi as the most abundant group at 29.4%, with dominant species including Spiromyces aspiralis and Neoolivea tectonae from the Zygomycota division, which can be opportunistic pathogens [23]. These findings align with the study by Henan Li et al., which identified polymicrobial fungi at 45% in Rhizopus and Mucor, which belong to the Zygomycota division [23]. Although fungi were detected at a relative abundance of 29.4%, this finding should be interpreted with caution. Fungal empyema is clinically rare, and the detection of fungal DNA may reflect environmental contamination, colonization, or opportunistic organisms rather than true pathogenic infection. Moreover, relative abundance values are subject to genome size bias, and the absence of Reads Per Kilobase normalization in our analysis may have inflated fungal representation. Importantly, none of the patients received antifungal therapy, yet clinical improvement was achieved with antibiotics and drainage alone, suggesting that the fungi detected were unlikely to play a primary pathogenic role. Future studies with larger cohorts, strict contamination control, and quantitative normalization methods are needed to clarify the pathogenic relevance of fungi in thoracic empyema. A complete list of all organisms identified, including fungi, anaerobes, and Gram-negative/positive bacteria, is provided in the Supplementary material.

Gram-negative bacteria were the second most abundant, comprising 26.5% of the samples and anaerobic bacteria were also present at 22.5%, highlighting challenges in culturing these organisms. These are consistent with Nikolaos IK et al. findings [10,25]. Anaerobic bacteria were also present at 22.5%, highlighting challenges in culturing these organisms. S. pneumoniae was detected in three samples (9.8%), with S. aureus and Mycobacterium at lower abundances (0.7% each), aligning with Nikolaos IK et al. [10].

The positivity rate of microbiome detection analyzed by the metagenomic method in this study, reached 56.7%, higher than the culture method's 40.0%. The results coincide with the study by Rosario I, et al., which showed 54.9% identification with culture and 81.7% with metagenomics [23]. Miranda et al. also reported 35.0% identification with culture and 70.0% with metagenomics, followed by the study by Henan Li et al. which showed that the culture method identified 54.0% and mNGS identified 95.0% with high significance (P<0.001) [21,23]. Diagnostic accuracy was also rated higher in metagenomics (74.5%) compared to culture (64.7%) [21]. This study shows differences in bacterial patterns for identifying fungi, anaerobic bacteria, and fastidious bacteria (S. pneumoniae) compared to culture methods. Qian Yi-Yi et al. successfully identified 16 bacterial or fungal pathogens with metagenomic methods, including difficult pathogens such as M. tuberculosis and Nocardia spp., which involve prolonged incubation times, and Mycoplasma pneumoniae, which cannot be cultured under standard conditions. The metagenomic method provides a short turnaround time, enabling the rapid identification of pathogens [13].

The analysis in this study also examined the level of agreement between culture and metagenomic examinations, categorized into three groups: complete concordance (36.7%), partial concordance (23.3%), and no concordance (40.0%). These results are in concordance with the study by Rosario I, et al., which recorded a complete concordance of 48.8%, consisting of 13.4% negative results in both methods, 29.3% identification of the same isolates, and 6.1% mixed sequences detected by metagenomics. Partial concordance was found in 14.6% of cases, with 2.4% of culture-positive results in mixed sequences and only one microorganism identified by culture, while 12.2% of mixed sequences were identified by metagenomics. No concordance was recorded at 36.6%, with 3.7% of samples identifying different genera, 28% identified by metagenomics with negative culture results, and 4.9% identified by culture with negative metagenomics [12].

There are 2 samples (7.0%) with positive cultures but negative metagenomics in this study, possibly due to the suboptimal specimen volume for extraction. Many studies explain that the NGS metagenomic method performs poorly detecting Candida spp. at 21.7% (5/23) and Klebsiella spp. at 21.7% (5/23) from positive culture results. This may be caused by the sequencing depth, which affects the sensitivity value of mNGS [13,26]. Some literature also mentions that increasing the number of sequencing reads can enhance sensitivity. Another limitation of mNGS is its inability to distinguish the pathogenicity status of the detected pathobionts [13,21,26].

Tan J, et al. also explained that mNGS testing can simultaneously identify pathogens and target genes in clinical samples without requiring pre-sequence amplification. Metagenomics can identify 90.1% of pathogens detected through culture and other types that are not detected [27]. mNGS testing has been shown to improve pathogen detection in critically ill patients with suspected infections and unclear diagnosis, and several studies recommend its use earlier than conventional culture in this setting (He et al., 2023; Guo et al., 2024). Early detection of microorganisms causing severe infections can provide timely clinical interventions and targeted antibiotic treatment [28].

Metagenomics has several limitations. Although it is pricier than 16S sequencing, especially for species identification, the main concerns revolve around cost and complexity. Metagenomic methods may obscure diseases and complicate microbial detection because human DNA comprises 75% of the sequencing results [29]. The risk of misidentifying or misclassifying rare pathogens highlights the importance of having high-quality and comprehensive databases. Furthermore, metagenomic techniques can identify contaminants from reagents, equipment, or the environment, which may result in false positives. Additional drawbacks of metagenomics include difficulties in identifying antibiotic-resistant genes and inconsistencies in bioinformatics analysis [30].

This study found that the metagenomic approach has a higher positivity rate compared to conventional culture (56.7% vs. 40.0%) and achieved a complete concordance of 36.7%. Importantly, our study demonstrates that metagenomic sequencing can optimize antibiotic therapy by identifying anaerobes and fastidious bacteria that are frequently missed by culture, thereby supporting more rational and targeted antibacterial use. Although fungi were detected, patients improved without antifungal therapy, suggesting that their clinical significance remains uncertain. These findings highlight the potential of metagenomics to enhance diagnostic accuracy and antimicrobial stewardship, while underscoring the need for further studies to clarify the role of fungi in thoracic empyema. The use of metagenomics can assist in selecting antibiotics, minimizing the use of broad-spectrum antibiotics, and reducing antibiotic resistance. These results should inform Indonesia's policies on the prevention and management of thoracic empyema, particularly regarding infection control and the rational use of antibiotics. However, the study had a small sample size; thus, future research should include a larger number of samples and broader geographical coverage to enhance the findings.

SUPPLEMENTARY MATERIAL

Supplementary material

Bacterial abundance profiles per sample.