Long-read whole-genome sequencing dataset of microbial communities from industrially and municipally impacted freshwater wetlands in South Africa

Abstract

This article describes a long-read whole-genome shotgun sequencing dataset generated from microbial communities inhabiting industrially and municipally impacted freshwater wetlands in South Africa. Surface water samples were collected from five strategically selected sites exposed to distinct anthropogenic pressures, including industrial effluent discharge, sewage overflow, greywater inputs, informal settlement runoff, and landfill leachate to generate a unique microbial genomic data. Environmental DNA was extracted and sequenced using the PacBio Sequel IIe platform, producing high-fidelity long reads suitable for improved assembly contiguity and functional reconstruction. Post-quality control processing yielded 4.9 × 10⁴ to 1.6 × 10⁵ HiFi reads per sample, corresponding to 0.34–1.02 Gb of high-accuracy sequence data per site. Long-read assemblies generated between 16,080 and 54,670 predicted protein-coding genes per sample. Taxonomic classification using Kaiju assigned 94.1–99.8% of assembled sequences to reference taxa. Domain-level profiles were exclusively bacterial dominated, with few rare or undetected (0.000–0.001%) archaeal, eukaryotic, or viral representation. Phylum-level composition was strongly dominated by Pseudomonadota (83–95%), followed by Bacillota (3–10%) and Bacteroidota (1–14%), with Actinomycetota consistently below 1%. Functional annotation using the DRAM pipeline identified 9390–31,251 KEGG orthologs, 969–3039 MEROPS peptidases, 13,454–45,103 Pfam domains, and 202–776 carbohydrate-active enzyme (CAZy) genes across assemblies. Distilled metabolic modules indicated the presence of near‑complete electron transport chain complexes (I–V), denitrification-associated pathways, sulfur oxidation and dissimilatory reduction genes, and diverse carbohydrate degradation functions; methanogenesis‑associated modules were not detected among the annotated metabolic pathways recovered in this dataset. The dataset provides genomic coverage of urban wetland microbiomes shaped by mixed industrial and municipal stressors and represents one of the few long-read metagenomic resources available for southern African freshwater wetlands. The availability of assembled contigs, gene annotations, metabolic reconstructions, enables reuse for comparative environmental genomics, biogeochemical modelling, bioremediation gene discovery, resistome screening, and microbial ecology investigations. This high-fidelity long-read sequencing resource expands opportunities for structural and functional analyses of anthropogenically influenced wetland ecosystems and supports future research in environmental biotechnology, bioinformatics-driven ecosystem monitoring, and microbial adaptation to urban pollution gradients.

Specifications Table

Subject	Biology, Earth and Environmental Science, Microbiology, Metagenomics profiling
Specific subject area	Anthropogenically impacted wetland ecosystems; comparative environmental genomics; long-read sequencing; Metagenomics-driven monitoring
Type of data	Tabular (CSV), Figures, FASTQ/FASTA files, Analyzed, Filtered, Processed etc.
Data collection	Surface water was collected from five industrially and municipally impacted freshwater wetland sites influenced by effluent discharge, sewage overflow, greywater, informal settlement runoff and landfill leachate. Environmental DNA was extracted and sequenced on a PacBio Sequel II platform to generate high-fidelity long reads sequencing. Assemblies were produced using Flye and taxonomic and functional profiles were generated using Kaiju and DRAM. Low-quality reads were excluded during preprocessing to ensure high-confidence genomic data.
Data source location	Environmental samples were collected from five freshwater wetland sites located in the Tshirela and Sharpeville regions of Gauteng Province, South Africa. Sampling points are situated at approximate coordinates: SP1 at 26.6663°S, 27.8407°E; SP2 at 26.6698°S, 27.8492°E; SP3 at 26.6748°S, 27.8596°E; SP4 at 26.6727°S, 27.8601°E; and SP5 at 26.6722°S, 27.8604°E. These sites form part of a connected floodplain wetland system characterized by perennial flow and progressive downstream transition from urban settlement and horticultural interfaces (SP3) to greywater and sewage-influenced drainage zones (SP4), and to landfill-affected downstream filtration zones (SP5). SP1 and SP2 represent lacustrine and riverine segments influenced by industrial effluent discharge, sewage overflow, and urban runoff. All samples were collected within wetlands located in close proximity (100–500 m) to residential
	settlements, road infrastructure, and anthropogenic disturbance sources. The generated long-read whole-genome sequencing data are archived in the NCBI Sequence Read Archive (SRA) under the associated BioProject.
Data accessibility	Repository name: NCBI Sequence Read Archive (SRA) under the associated BioProject Data identification number: PRJNA1425911. Direct URL to data: https://www.ncbi.nlm.nih.gov/sra/PRJNA1425911. Data identification number: PRJNA1425911. Direct URL to data: https://www.ncbi.nlm.nih.gov/sra/PRJNA1425911. Instructions for accessing these data: Direct Access. Repository name: www.Figshare.com DOI - Is a supplementary Data: https://doi.org/10.6084/m9.figshare.32064504.
Related research article	N/A

1. Value of the Data

• •This dataset provides one of the first publicly accessible long-read whole-genome metagenomic resources from industrially and municipally impacted freshwater wetlands in the Tshirela and Sharpeville regions of Gauteng Province, South Africa. The use of PacBio HiFi sequencing improves taxonomic resolution and gene recovery relative to short-read approaches, enabling robust characterisation of dominant microbial taxa and their functional potential. However, although the sequencing platform is capable of generating approximately 2–4 million reads per run, the number of reads obtained and deposited in this study is lower than this theoretical output. Despite this limitation, the dataset remains valuable for comparative analyses, methodological benchmarking, and future meta-analyses, and it establishes a foundational long-read reference for freshwater wetland metagenomics in Southern Africa.
• •The dataset allows researchers to directly associate microbial genomic structure with environmental gradients, supporting cross-disciplinary applications in wetland ecology, hydrochemistry, and systems-level environmental modelling.
• •Taxonomic profiles provide a detailed gene inventory for Sulphur cycling, nitrogen transformations, carbohydrate degradation, and aerobic respiration. These annotations can be reused for comparative environmental genomics and metabolic reconstruction studies.
• •Raw data can be reused for novel assembly, binning, and functional inference analyses.
• •The dataset provides a reference framework for assessing microbial responses to mixed anthropogenic pressures, serving as a baseline for longitudinal monitoring, pollution impact assessment, and restoration-oriented microbial surveillance programs.
• •The dataset thus supports applications in environmental biotechnology, urban wetlands ecosystem management, and microbial resource discovery.
• •All processed data products are hosted on Figshare.

2. Background

Freshwater wetlands provide essential ecosystem services, including water purification, nutrient retention, flood regulation, and biodiversity support, yet they are increasingly impacted by urbanization, wastewater discharge, and industrial activities [1,2]. Wetlands within the Vaal River catchment are exposed to combined pressures from mixed anthropogenic pollution [2]. Existing monitoring frameworks rely on physicochemical measurements and vegetation indicators [3], while microbial genomic characterization remains limited, particularly profiling using long-read sequencing technologies [4].

Long-read sequencing (PacBio HiFi Sequel IIe platform) has improved metagenomic assembly accuracy and gene recovery compared to short-read platforms [5,6]. However, available long-read datasets from urban wetlands in sub-Saharan Africa remain scarce. This dataset generated a high-resolution genomic resource representing microbial communities from five hydrologically connected wetland sites influenced by sewage infrastructure failures, industrial effluent discharge, landfill leachate, and informal settlement runoff in the Tshirela and Sharpeville. Surface water samples were sequenced using the PacBio Sequel IIe platform, and assemblies were generated for downstream taxonomic and functional annotation.

The dataset integrates long-read sequences to facilitate comparative environmental genomics, bioinformatic benchmarking, and ecosystem monitoring applications.

3. Data Description

The dataset comprises high-fidelity long-read metagenomic data generated from five freshwater wetland sampling sites (SP1–SP5) within the Tshirela and Sharpeville wetland system, South Africa, and is publicly accessible under the NCBI BioProject accession PRJNA1425911, with corresponding SRA accessions (SP1–SP5) and repository https://www.ncbi.nlm.nih.gov/sra/PRJNA1425911 to ensure reproducibility and permanent access. The archive is organized into five primary directories. The raw HiFi reads folder contains five compressed FASTQ files (SP1–SP5_HiFi.fastq.gz), each including post–quality-controlled PacBio HiFi reads composed of read identifiers, nucleotide sequences, and Phred+33 quality scores. Sequencing statistics per sample include 49,093–160,103 reads, 341,410,795–1035,094,499 total bases, mean read lengths, read-length standard deviations, duplicate read percentages, quality score distributions, and GC content. The Flye assembly’s directory provides assembled contig files (SP1–SP5_flye_contigs. fasta), with assembly metrics including total assembly length (16,789,676–57,869,027 bp), 272–1056 contigs, largest contig size, N50 values (83,196–617,890 bp), GC percentage, and mean coverage, with no scaffolds reported. Taxonomic classifications are stored in Kaiju output as SP1–SP5. kaijuReport files containing tab-delimited fields for percent classified, read counts, taxon ID, rank, and taxon name, visualized in Fig. 1 as radial Krona plots. Functional annotations are contained in DRAM annotations, including annotations.tsv, which provides gene-level KEGG ortholog, Pfam domain, MEROPS peptidase, and CAZy assignments, and metabolism_summary.xlsx, summarizing pathway completeness and metabolic module matrices; annotated features per assembly range from 16,080–54,670 predicted genes, 9390–31,251 KEGG orthologs, 969–3039 MEROPS entries, 13,454–45,103 Pfam domains, and 202–776 CAZy genes. The environmental metadata directory includes physicochemical and hydrochemical datasets (Physicochemical_parameters.xlsx, Cations_Anions.xlsx, BOD_COD.xlsx) and site descriptors (Site_Metadata.docx), documenting pH, conductivity, temperature, salinity, redox potential, total dissolved solids, biological and chemical oxygen demand, major cations and anions, and geographic coordinates.

Fig. 1.

Taxonomic composition of wetland water metagenomes across pollution-impacted sites based on Kaiju annotation. Radial hierarchical (Krona) plots showing the taxonomic profiles of metagenomic reads from five sampling sites (sp1–sp5) representing different levels of pollution impact.

Tables and Figures accompanying the dataset include Table 1, Long-Read PacBio HiFi Metagenomic Profiling of Anthropogenically Polluted Wetland Waters in Tshirela and Sharpeville Townships, South Africa; Table 2, Sampling points physicochemical parameters measured across the impacted Wetland Waters in Tshirela and Sharpeville Townships, South Africa, Fig. 1 (Kaiju-based taxonomic profiles), and Fig. 2 (DRAM-derived functional module heatmaps).

Table 1.

Long-Read PacBio HiFi Metagenomic Profiling of Anthropogenically Polluted Wetland Waters in Tshirela and Sharpeville Townships, South Africa^¥.

Samples	SP1	SP2	SP3	SP4	SP5
Sequence statistics
Number of Reads	152,210	149,306	49,093	160,103	112,938
Total Number of Bases	1035,094,499	893,561,450	341,410,795	1018,532,486	846,747,271
Mean Read Length	6800.44	5984.77	6954.37	6361.73	7497,45
Read Length Std Dev	3468.6	2717.61	3359.55	3295.0	4016,8
Number of Duplicate Reads (%)	442 (0.29%)	316 (0.21%)	34 (0.07%)	203 (0.13%)	276 (0.24%)
Phred Type	33	33	33	33	33
Quality Score Mean (Std Dev)	38.82 (4.71)	38.9 (4.57)	38.79 (4.75)	38.89 (4.56)	38,74 (4,83)
Quality Score (Min/Max)	3.0/40.0	3.0/40.0	3.0/40.0	3.0/40.0	3.0/40.0
GC Percentage	52.96%	52.27%	57.16%	52.1%	54.98%
Assembly (Flye) statistics
Total length (bp)	35,613,637	39,422,501	16,789,676	57,869,027	36,208,916
Number of contigs	688	1056	272	672	807
Largest contig (bp)	3009,442	3903,794	4033,946	4108,331	4235,679
Scaffolds	0	0	0	0	0
N50 (bp)	138,661	90,837	617,890	582,055	83,196
GC (%)	48.91	51.57	58.97	53.94	55.25
Mean coverage	26	19	17	15	21

^¥All values represent ranges across samples of each site. Statistics are based on post-QC PacBio HiFi reads generated using CCS (minimum passes ≥3, Q ≥ 30). Assemblies were produced using Flye with default metagenomic parameters.

Table 2.

Sampling points physicochemical parameters measured across the impacted Wetland Waters in Tshirela and Sharpeville Townships, South Africa^†.

Sampling Points	SP1 (26.6663°S, 27.8407°E;)	SP2 (26.6698°S, 27.8492°E)	SP3 (26.6748°S, 27.8596°E)	SP4 (26.6727°S, 27.8601°E)	SP5 (26.6722°S, 27.8604°E)
Multi-parameter Results
DO (mg/L)	3.50 ± 0.35	2.15 ± 0.07	2.675 ± 0.30	3.77 ± 0.95	1.515 ± 0.49
Conductivity (µS/cm)	795.00 ± 35.63	655.00 ± 261.63	440.00 ± 28.28	575.50 ± 260.92	548.00 ± 124.45
pH	8.04 ± 0.28	7.45 ± 0.35	7.43 ± 0.39	7.55 ± 0.07	7.40 ± 0.01
Temperature ( °C)	19.70 ± 0.38	22.05 ± 1.91	22.55 ± 0.64	23.40 ± 1.56	22.95 ± 2.76
Resistance (kΩ)	1.25 ± 0.78	1.66 ± 0.64	2.29 ± 0.15	1.93 ± 0.88	1.87 ± 0.43
ORP (RmV)	198.00 ± 45.95	491.90 ± 54.59	759.55 ± 228.47	843.10 ± 227.56	1032.95 ± 286.99
PD (mV)	220.00 ± 63.16	442.25 ± 16.62	767.30 ± 230.36	764.85 ± 352.35	1021.25 ± 270.47
Salinity (psu)	0.43 ± 0.10	0.34 ± 0.16	0.23 ± 0.04	0.22 ± 0.04	0.36 ± 0.07
TDS (ppm)	398.00 ± 27.95	323.50 ± 127.99	219.00 ± 14.14	212.70 ± 24.47	349.00 ± 42.43
BOD (mg/L)	0.63 ± 0.12	2.03 ± 0.72	1.40 ± 0.30	1.17 ± 0.38	1.33 ± 0.06
COD (mg/L)	442.67 ± 107.27	537.33 ± 577.39	326.67 ± 166.64	478.00 ± 393.75	738.67 ± 186.57
Cations (mg/L)
Sodium	45.03 ± 2.23	56.21 ± 1.47	28.20 ± 1.24	29.07 ± 0.80	58.37 ± 1.08
Potassium	13.31 ± 0.99	11.22 ± 1.68	6.64 ± 1.56	8.47 ± 0.60	12.78 ± 1.85
Magnesium	46.18 ± 1.17	35.04 ± 2.42	21.23 ± 0.47	23.72 ± 1.67	48.14 ± 2.48
Calcium	75.36 ± 1.99	67.16 ± 2.33	43.21 ± 1.55	45.49 ± 1.51	85.83 ± 2.02
Ammonium	2.85 ± 1.56	13.18 ± 1.40	23.12 ± 2.60	33.16 ± 3.80	30.12 ± 1.80
Anions (mg/L)
Fluoride	0.332 ± 0.127	0.196±0.112	0.240±0.128	0.149 ± 0.529	0.17 ± 0.10
Bromide	0.460 ± 0.610	0.12 ± 0.93	0.716 ± 0.013	0.021 ± 1.06	0.57 ± 0.04
Chloride	34.97 ± 8.05	36.83 ± 7.06	30.58 ± 3.87	26.39 ± 2.16	34.67 ± 9.28
Nitrite	12.10 ± 1.05	11.13 ± 1.07	19.63 ± 3.19	16.14 ± 1.02	13.18 ± 1.35
Nitrate	2.29 ± 0.02	8.30 ± 8.42	31.18 ± 5.82	25.63 ± 8.85	23.30 ± 18.99
Sulfate	121.71 ± 150.71	153.64 ± 101.84	80.38 ± 25.77	72.29 ± 29.60	130.10 ± 44.96
Phosphate	0.077 ± 0.013	0.080 ± 0.025	0.081 ± 0.004	0.090 ± 0.001	0.61 ± 0.06

^†Values represent mean ± standard deviation of physicochemical parameters measured across the impacted Wetland Waters in Tshirela and Sharpeville Townships, South Africa^†.

Fig. 2.

Functional potential of wetland water metagenomes under varying pollution impact based on DRAM annotation. a) Heatmap showing the completeness of core metabolic modules and electron transport chain (ETC) complexes across sampled sites (sp1–sp5). b) Presence/absence heatmap of key functional genes and pathways inferred from DRAM annotations.

Finally, Table 2 includes physicochemical and hydrochemical datasets, documenting pH, conductivity, temperature, salinity, redox potential, total dissolved solids, biological and chemical oxygen demand, major cations and anions, and geographic coordinates.

4. Experimental Design, Materials and Methods

4.1. Study sites

Sampling was conducted within the Tshirela and Sharpeville wetland system situated between Vanderbijlpark and Vereeniging in southern Gauteng Province, South Africa and represents a floodplain-associated perennial freshwater system influenced by urban, industrial, and landfill-derived inputs. Five sampling points (SP1–SP5) were selected to represent distinct local exposure settings along the wetland/stream corridor. Geographic coordinates for each sampling point are reported in the accompanying repository metadata (approximate coordinates: SP1 at 26.6663°S, 27.8407°E; SP2 at 26.6698°S, 27.8492°E; SP3 at 26.6748°S, 27.8596°E; SP4 at 26.6727°S, 27.8601°E; and SP5 at 26.6722°S, 27.8604°E). SP1 represents a lacustrine floodplain wetland receiving precipitation, runoff, sewage discharge, and industrial effluent. SP2 is a riverine segment impacted by sewer overflows and informal waste deposition. SP3–SP5 form a downstream gradient along a stream corridor influenced by horticultural runoff, domestic greywater discharge, and landfill leachate inputs. Hydric soils, perennial flow regimes, and dominant hydrophytes (Phragmites australis, Typha latifolia, Eichhornia crassipes) characterize all sites.

4.2. Field measurements

Dissolved oxygen (DO mg/L) and temperature ( °C) were measured in situ using a YSI 55–12FT (Yellow Springs Instruments, USA), according to the manufacturer’s instructions. In addition, onsite measurements were captured using a handheld multiparameter water-quality instrument (field logs), used to record physicochemical indicators such as pH, Electrical conductivity (µS/cm), total dissolved solids (mg/L), resistivity (kΩ), Salinity (psu), PD (mV), TDS (ppm), and oxidation–reduction potential (RmV) after routine calibration and stabilization at each sampling location. Electrodes were rinsed with distilled water between measurements. Triplicate readings were recorded at each site. Biological Laboratory physicochemical parameters, such as COD (mg/L), were quantified (COD analyzer following APHA Standard Methods and ISO guidelines and BOD (mg/L) was quantified using the OxiTop™ respirometric bottle system under standard incubation conditions (20 °C for 5 days); all accompanying datasets are included in Table 2.

4.3. Sampling design and DNA extraction, library preparation, and long-read sequencing

Five-litre surface water grab samples were collected from each wetland sampling point (SP1–SP5) using sterile polypropylene containers, immediately placed on ice in the field, transported to the laboratory under chilled conditions, and processed for downstream analyses within 12 hours of collection to minimise post-sampling microbial community shifts and DNA degradation. To concentrate microbial biomass from the low cell density aquatic matrix characteristic of the wetlands impacted by sewage inputs, industrial discharge, and landfill seepage, the full 5 L of water per sample was filtered through 0.22 µm polyethersulfone membrane filters, after which total community DNA was extracted from the folded, biomass-loaded filters using the DNeasy PowerSoil Kit (Qiagen), selected over conventional water extraction kits because the filter trapped biomass from such contaminant-rich systems behaves chemically and physically like sediment, being embedded in organic flocs, enriched in humic and fulvic acids, heavy metals, and sulfides, while the inhibitor removal chemistry and bead beating lysis of the PowerSoil workflow reliably produced high molecular weight DNA suitable for long-read sequencing. DNA concentrations were quantified fluorometrically using the Qubit system (Thermo Fisher Scientific) and assessed purity and integrity with a NanoDrop 8000 spectrophotometer prior to library preparation. Absorbance ratios at 260/280 nm and 260/230 nm were examined to evaluate potential protein and organic contaminants, respectively; all extracts satisfied established quality thresholds (A260/280 ≥ 1.7; A260/230 ≥ 1.8) and therefore required no additional purification. To preserve molecular integrity, DNA was maintained at 4 °C during short-term handling and aliquoted (1–5 µL) into sterile 1.8 mL cryovials for storage at −80 °C, thereby minimizing freeze–thaw stress. Frozen aliquots were thawed gradually at 4 °C (12–16 h) before downstream processing. Fragment size distribution and integrity were verified using the Agilent Femto Pulse capillary electrophoresis platform, which provides high-resolution sizing from 75 bp to 165 kb. Samples with modal fragment sizes ≤10 kb proceeded directly to library construction, whereas preparations enriched in fragments >15 kb were gently sheared using Covaris g-TUBE devices to obtain an average insert size of 10 kb. SMRTbell libraries were prepared using the SMRTbell Express Template Prep Kit 2.0 (Pacific Biosciences) according to the 10 kb metagenomic shotgun workflow, incorporating DNA damage repair, end-repair, adaptor ligation, and exonuclease cleanup steps. Approximately 1.5 µg of high-quality genomic DNA was used per library. Sequencing was conducted by Inqaba Biotech Ltd. (Pretoria, South Africa) on the PacBio Sequel IIe platform using Circular Consensus Sequencing (CCS) chemistry to generate high-fidelity (HiFi) reads. The resulting HiFi reads were exported in FASTQ format for subsequent assembly, taxonomic classification, and functional annotation analyses.

4.4. Quality control, assembly, and annotation of metagenomes

Raw long-read sequencing data were processed using the PacBio SMRT Link software suite (Pacific Biosciences, Menlo Park, CA, USA) to generate high-fidelity (HiFi) reads via the Circular Consensus Sequencing (CCS) algorithm under default metagenomic filtering parameters, a workflow recognized for its accuracy in long-read data generation [5,7,8]. Each 8 M SMRT Cell produced approximately 2–4 million HiFi reads, corresponding to 20–40 Gb of sequence data per run. Read-length distribution analysis confirmed median fragment sizes between 10 and 15 kb, consistent with the prepared insert libraries, and quality score evaluation demonstrated that the majority of reads achieved mean Phred scores ≥ Q30, ensuring high base-calling confidence. HiFi reads were exported in FASTQ format and converted using SAMtools to enable compatibility with downstream metagenomic pipelines [8]. Additional quality control measures retained reads within the 5–30 kb length range and mean quality ≥ Q30, while screening for residual adapter sequences and low-complexity regions. All raw and quality-filtered datasets were deposited in the NCBI Sequence Read Archive under the designated BioProject accession PRJNA1425911, ensuring transparency and reproducibility.

Quality-filtered HiFi reads were subsequently analyzed within the Department of Energy Systems Biology Knowledgebase (KBase) Narrative environment, an open and reproducible computational platform for integrative systems biology [9]. Individual read inputs (not a ReadsSet) were processed using a documented workflow comprising file unpacking, long-read assembly, taxonomic classification, batch assembly creation, and metabolic annotation. De novo assembly was conducted using Flye v2.9.4 in metagenome mode [10], generating per-sample contig FASTA outputs. Assembly performance was evaluated using total assembly length, contig counts, N50 values, GC content, and mean coverage to assess structural integrity and completeness across samples.

Taxonomic classification of both unassembled reads and assembled contigs was performed using Kaiju v1.9.0 in greedy-5 mode, which permits up to five mismatches during extension of maximum exact matches [[6], [7], [8], [9], [10], [11], [12]]. Classification was conducted against the NCBI BLAST non-redundant (nr) protein database supplemented with eukaryotic sequences, producing tab-delimited kaijuReport outputs (SP1.kaijuReport–SP5.kaijuReport). These outputs were visualized using hierarchical Krona plots to represent the taxonomic structure and relative abundance distributions across sampling sites.

Functional annotation of assembled metagenomes was performed using the DRAM pipeline v1.4.6 [11,12], which integrates gene prediction via Prodigal v2.6.3 and multi-database functional assignment across KEGG orthology, UniProt, Pfam, MEROPS, and dbCAN/CAZy repositories. DRAM generated gene-level annotation tables (annotations.tsv) and summary metabolic matrices (metabolism_summary.xlsx), enabling comparative evaluation of metabolic module completeness and pathway presence/absence across wetland sites. This integrative annotation framework provided a structured representation of microbial functional capacity while maintaining standardized database interoperability.

Throughout sequencing and bioinformatic processing, rigorous quality assurance procedures were implemented, including scheduled calibration of field and laboratory instruments, strict cold-chain sample preservation, incorporation of blank controls, and triplicate laboratory measurements to detect analytical variability. Data curation, validation, and statistical handling were performed using established analytical tools to ensure consistency, reproducibility, and methodological transparency across all generated metagenomic datasets.

Limitations

This long-read metagenomic dataset derived from industrially and municipally impacted freshwater wetlands in the Tshirela and Sharpeville regions of Gauteng Province, South Africa, is subject to several methodological and logistical limitations inherent to environmental sampling and high molecular weight DNA sequencing. Microbial biomass was collected from discrete locations during a single sampling campaign, and therefore, the dataset does not capture fine-scale temporal variability, seasonal dynamics, or episodic pollution events that are characteristic of wetland ecosystems. DNA extraction from water-borne microbial fractions may further underrepresent organisms present at very low abundance or those strongly associated with sediments and biofilms, potentially biasing community composition toward planktonic taxa. Although PacBio HiFi sequencing provides high per-base accuracy and long read continuity, variability in DNA yield and sequencing depth across samples, coupled with mechanical shearing and the application of stringent read length and quality filters (5–30 kb; ≥Q30), may have limited assembly contiguity and gene recovery, particularly for rare taxa. In addition, the number of reads generated and deposited is substantially lower than the theoretical capacity of the platform, constraining overall genome coverage and precluding exhaustive genome reconstruction. Finally, taxonomic and functional annotations rely on existing reference databases, which may incompletely represent novel or region-specific wetland microorganisms. These limitations should be carefully considered when interpreting the dataset, which is best suited for exploratory, comparative, and methodological applications rather than comprehensive community resolution.

CRediT Author Statement

Onyedikachi Ubani: Conceptualization, study design, field sampling, laboratory processing, DNA extraction and library preparation, bioinformatics workflow development, data curation, formal analysis, visualization, and writing – original draft preparation. Veronica M. Ngole-Jeme: Methodology refinement, supervision, project administration, resource provision, critical review and editing of the manuscript, and overall research oversight.

Ethics Statement

This study was conducted exclusively on environmental samples collected from freshwater wetland ecosystems and did not involve human participants, vertebrate animals, or the use of personal, clinical, or social media-derived data. All procedures were limited to environmental sampling of water and associated microbial biomass for genomic analysis. Consequently, the research did not require institutional human or animal ethics approval under applicable national and institutional guidelines. Field sampling was performed in accordance with relevant environmental regulations, and no protected species or sensitive personal data were involved at any stage of the study.

Data Availability

• NCBI"Long-Read PacBio HiFi Metagenomic Profiling of Anthropogenically Polluted Wetland Waters in Tshirela and Sharpeville Townships, South Africa (Original data)