Physico-chemical and microbiological analysis of the Loopspruit River, North West Province: A metagenomic approach

South Africa is a water-scarce country with a semi-arid climate in most of its northern and central regions. Furthermore, a large portion of the available surface water contributes to the mining and agricultural sectors, the primary economic drivers of the country (DWAF 2009; Von Bormann & Gulati 2016). Nonetheless, the latter activities are amongst the most significant causes of pollution, resulting in water quality degradation (Von Bormann & Gulati 2016). Furthermore, recent studies have illustrated that the various surface water systems are also impacted by various other non-point and point source pollutants such as heavy metals, pesticides, faecal matter, acid mine drainage and bacteria (Jordaan & Bezuidenhout 2016). This comes as no surprise as the open nature of surface water systems makes them prone to pollution emanating from anthropogenic activities and natural processes such as climate change and water-rock interactions (Akhtar et al. 2021).

The Loopspruit River, a sub-catchment of the Mooi River catchment, is located within the Upper Vaal catchment (Van der Walt et al. 2002). Various studies have identified the Vaal catchment as one of the most exploited and polluted river basins in South Africa; with reports of microplastics, bacterial contamination (Acrobacter, Aeromonas and Sphingomonas), persistent organic pollutants such as organochlorine insecticides and polychlorinated biphenyls, perfluorinated alkyl compounds, alkylphenol ethoxylates, brominated flame retardants, metal pollution, and pharmaceuticals present in the catchment (Wepener et al. 2011; Jordaan & Bezuidenhout 2016; Bezuidenhout et al. 2017; Groffen et al. 2018; Pheiffer et al. 2018; Chokwe & Okonkwo 2019; Weideman et al. 2020). Consequently, the Vaal River catchment's water quality compliance is often poor, owing to mining, agricultural, urbanization, and industrial activity (Iloms et al. 2020). Furthermore, the Loopspruit River which supports crop farming and grazing is mainly impacted by dry land agricultural activities, watershed from the surrounding mines, and informal dwellings (Van der Walt et al. 2002; Jordaan & Bezuidenhout 2016). Nonetheless, other studies conducted in this region, the Mooi River catchment, have reported the presence of potential indicator organisms, some harbouring antibiotic resistance efflux pump and virulence genes (Jordaan & Bezuidenhout 2016; Molale & Bezuidenhout 2016). The latter studies attributed the microbial pollution to poorly treated wastewater effluent and urban as well as informal settlements located in the vicinity of this catchment.

Water contaminated with faecal matter appears to be the major source of microbial pollution in the environment across the globe (Hamiwe et al. 2019; Ball et al. 2021). While Yang et al. (2020) note that microbial pollution is a major global problem as it is a human health concern and a significant health hazard to the ecosystem. Furthermore, various studies have reported the presence and/or likelihood of pathogenic traits in faecal-indicator bacteria (Burnet et al. 2021). Thus, monitoring the presence of faecal indicator bacteria, such as E. coli, Enterococcus sp. and Clostridium sp., can be useful in determining the microbial quality of a water source and provide data on faecal contamination in the environment. However, water quality is not only limited to the biological properties of water but also includes the physical, chemical and aesthetic properties (Bui et al. 2019). Thus, water quality can be affected by nutrients and minerals such as nitrogen, phosphates, calcium, and sodium, temperature, pH, and dissolved solids.

The aim of this study was to evaluate the water quality of the Loopspruit River from a physico-chemical and microbiological perspective. The microbial diversity and physico-chemical data can be used to account for historical changes in land use activities and predict future changes.

The Loopspruit River covers an area with an annual rainfall of between 604 and 620 mm (DWAF 2009). Moreover, populations are expected to increase annually in the surrounding Loopspruit River areas, which may lead to an increase in possible point source pollution. The land use surrounding the Loopspruit River in terms of anthropogenic activities is primarily agricultural (crop farming and grazing) together with gold mining activities (Van der Walt et al. 2002). The major land activities in this sub-catchment surrounding the Loopspruit River are natural (65%), agricultural (30%), urban and mining (3%), and wetlands (1%) (Bezuidenhout et al. 2017). Non-point pollution issues may arise as a result of urban runoff. The wastewater treatment plant (WWTP) in the sub-catchment may be a source of point pollution.

The Loopspruit River area has a warm and humid climate, with typical summer rainfall and dry winters. The warm to hot (15–36 °C) summer months account for up to 90% of the annual rainfall between November and March. Winters are mostly dry and cold (−2 to 26 °C), with ground frost. The annual evaporation rate in the region ranges between 1,500 and 1,700 mm (Barnard 2000). The Loopspruit's hydrological flow is estimated to be 5,463 m³/day (Rafundisani & Dhaver 2015).

From 1997 to 2008, several water monitoring stations along the Loopspruit River collected physicochemical data. Since then, no recorded data or microbiological data have been recorded, necessitating the need to update the data in this current study.

The map (Figure 1) was based on the data obtained from the North-West University's (NWU) database containing data of the NASA: ASTER 90 m Digital Elevation Dataset; DEA: 2013–2014 SA National Land Cover Dataset; DWAF: Hydrology Dams and SANBI: NFEPA River Network. The land use data were obtained from the WRC Report No K5/2347//3 (Bezuidenhout et al. 2017).

Sample sites along the Loopspruit River were chosen with anthropogenic activity and potential anthropogenic influences in mind. Feedlots, mining operations, agricultural operations, and wastewater treatment plants are all included. Sample locations ranged from Fochville to Klipdrift Dam and finally Potchefstroom (Table 1).

For the duration of the study (2018–2019), samples were collected in triplicate at each sample site (totalling 24 for each sampling season). In the first year, sampling took place during the wet season (April–May 2018) and then again during the dry season (July 2018). In the study's second year (2019), sampling occurred during the wet season of April–May 2019 and the dry season of June–July 2019. The different sampling times for the dry season were due to some logistical issues. Water was sampled in one-litre, clear, sterilized glass Schott-Duran^® bottles, in triplicate. The samples were stored on ice for 18–24 h. One additional sample was taken in 2018 and 2019's dry season for the microbial diversity determinations. Samples were filtered with 250 to 1,000 mL through a sterile 0.2 μm nitrocellulose membrane filters (Whatman GE Healthcare Life Sciences, Buckinghamshire, UK), and the filters were stored at −80 °C until needed for eDNA extraction or faecal pollution indicator screening.

A calibrated multimeter probe (Oakton PCStestr™ 35 Thermo Fisher Scientific, USA) was used to determine specific physical parameters such as temperature (°C), electrical conductivity (mS/m), pH and total dissolved solids (TDS) (mg/L) on-site. Nutrient analysis was conducted within 8 h of the unfiltered sample collection. The chemical constituents such as (⁠⁠): 8153, nitrates (⁠⁠): 8039, sulphates (⁠⁠): 8051, phosphate (⁠⁠): 8048 and the chemical oxygen demand (COD): 8000, was determined using a HACH^® spectrophotometer (HACH DR 2800)^TM (HACH, USA) and reagents according to the manufacturer's instructions. In short, pre-programmed methods (stored programs, eg. when testing for the program code number, 8051, was entered) were used to measure the sample nutrients. This consisted of HACH reagents (sachets or liquid drippers, depending on the test) to be mixed with a specific volume of the sample in a test tube and allowing their respective chemical reactions to occur. After the reaction duration was completed, the contents of the test tube were aliquoted into a 10 mL cuvette and inserted into the HACH^® spectrophotometer, pre-set to the desired programme and analysed the sample. Each test for ⁠, ⁠, ⁠, and COD had their own respective method and durations and were followed. Values measured were compared to the South African target water quality ranges (TWQR) for environmental water (DWAF 2009).

Faecal streptococci (Enterococcus sp.) and E. coli were enumerated via the membrane filtration method. Triplicate samples (100 mL) were filtered through 0.45 μm pore size membrane filters with a diameter of 47 mm [(PALL Life Sciences, Mexico) (CAT No: GN-6 Metricel Membrane 66191)]. These filters were placed on MLG (Oxoid, UK), and KF-Streptococcus selective agar media. KF-Streptococcus selective agar was supplemented with 1 mL per 100 mL media (Sigma and Aldrich, Germany) 2,3,5-triphenyltetrazolium chloride (TTC) at 60 °C.

The filters on the MLG agar plates were incubated at 37 °C for 18–24 h, and the KF-Streptococcus agar plates were incubated at 37 °C for 48 h. Green colonies on MLG agar were counted as putative E. coli, and light pink or flat dark red colonies on the KF-Streptococcus agar were counted as putative enterococci. Colony numbers were recorded, and colony-forming units (CFU) per 100 mL were calculated.

Genomic DNA was extracted from presumptive E. coli and Enterococus sp. isolates using a bacteria Chemagic DNA kit (PerkinElmer, Germany) following the manufacturer's instructions.

eDNA was exctracted with the NucleoSpin Soil Kit (Macherey-Nagel) according to the manufacturer's instructions. E. coli genomic rDNA was used as a template for PCR amplification of the rDNA gene using the primer 27F 5′ AGAGTTTGATCMTGGCTCAG-3′ and 1492R 5′-CGGTTACCTTGTTACGACTT-3′ (Jordaan & Bezuidenhout 2016) resulting in a final volume of 25 μL. Enterococus genomic DNA was used and amplified by using the primer set 341F 5′-CCTACGGGAGGCAGCAG-3′ and 907R 5′-CCGTCAATTCCTTTGAGTTT-3′ (Muyzer et al. 1993), in a final volume of 25 μL. All PCR's were done with the ICycler Thermocycler (Bio-Rad, USA), with an initial denaturation at 95 °C for 300 seconds followed by 35 cycles of denaturation, annealing and extension for 30 seconds at 95 °C, 30 seconds at 52 °C and 60 seconds at 72 °C, respectively. The final extension was done at 72 °C for 180 seconds. The DNA quality and quantity were determined with NanoDropTM 1000 Spectrophotometer (Thermo Fischer Scientific, USA) and agarose gel electrophoresis to confirm the presence of the PCR product.

Microbial genomic DNA (eDNA) was obtained from membrane filters and normalised to a concentration of ∼10 ng/μL. The sequence library preparation guide was followed as per the manufacturer's instructions (Illumina Inc.). Microbial diversity was determined using the 16S rRNA gene (V3–V4) region (≈460 bp) analysis with locus-specific primers 341F 5′ CCTACGGGNGGCWGCAG–3′ and 805R 5′–GACTACHVGGGTATCTAATCC–3′ (Herlemann et al. 2011). Illumina forward and reverse overhang adapters (Illumina Inc., CA, USA) were attached to the 5′–end of forward and reverse primers, respectively. The Illumina MiSeq sequencing run, de-multiplexing and secondary analyses of the reads were done with the MiSeq reporter software (Illumina Inc., California, USA).

The QIIME2 pipeline was used to analyse the Next Generation Sequencing (NGS) data by eliminating the effect of random sequencing errors, deleting unreliable data from the libraries (q-value <25) and removing reads shorter than 200 bp to classify the sequences to operational taxonomic units (OTUs) with 97% similarity. Taxonomic information of sequences by the Ribosomal Database Project (RDP) classifier for the 16S rRNA gene was assigned at a confidence cut-off of 0.5.

The QIIME2 output tables were further processed through the R programming language to create the final OTU table to use for further analysis. Metadata was added to the R table and an OTU file was created and analysed using the web-based tool, MicrobiomeAnalyst. Alpha diversity was determined using the Chao1, and Shannon diversity indices and beta diversities were determined with the Bray–Curtis dissimilarity distance distribution. The OTU data were also used for the metagenome analysis using PICRUSt.

This BioProject accession number is PRJNA727989.

R was used for all statistical analyses (RStudio, version 2022.02.3 + 492 ‘Prairie Trillium’ Release.). Analysis of variance (ANOVA) was used to analyse the randomized complete block design, using each independent experiment as a block, with repeated measures where possible. A one-way ANOVA was used for normally distributed data and Kruskal Wallis for non-normally distributed data to determine the difference in means between sites/seasons. Molecular variance analysis (AMOVA) was applied in R with the vegan library's bioenv package to determine the best set of environmental variables with maximum (rank) correlation. The non-metric multidimensional scaling (NMDS) plot was created as vectors along with the best subset of taxas to determine if the sampling locations plotted on the NMDS were statistically significant. A Tukey's HSD test was used to determine any variance difference in the upstream vs. downstream groups and/or sites. Furthermore, the vegan package was also used to determine the range of diversity indices. Observed OTUs (species richness) were determined with the Shannon index (for OTU richness and evenness) and Chao1 index for a qualitative measurement of alpha-diversity. Chao1 was used to account for the skewed data, which is usually the case in microbial data. The beta-diversity analyses utilised clustering of samples using the Bray-Curtis distance metric (for the samples to be weighted by their abundance) and visualized using the ggplot2 package in R to create the NMDS visuals. Only statistically significant variables (p < 0.05) were selected for further analysis.

The water quality parameters observed during the 2018–2019 sampling sessions of the two wet and two dry seasons are shown in Tables 2 and 3. In 2018, there was a statistical significance (p < 0.05) between the wet and dry seasons (Table 2) indicated as p = 0.02, whereas in 2019, there were no statistically significant differences (p > 0.05) between the wet and dry seasons with p = 0.63. Most of the physico-chemical parameters in Table 2 were within the target water quality range (TWQR) for South Africa. The TWQR were set by the Directorate National Water Resource Planning of the Department of Water Affairs and Forestry (DWAF). The TDS and COD, exceeded the TWQR at most of the sites in 2018 (DWAF 2009). Total dissolved solids ranged from 588–806.3 mg/L (TWQR for TDS is <450 mg/L) and the COD ranged from 99–131.7 mg/L (TWQR for COD is <100 mg/L). Temperature ranged from 10.2–14.5 °C during the cold dry season. Phosphates at sites KW03 and KA05 exceeded the TWQR with 1.1 and 1.3 mg/L (TWQR for is <0.4 mg/L), respectively.

Table 2

Physico-chemical data for the dry and wet seasons of 2018

View Large

Table 3

Physico-chemical data for the dry and wet seasons of 2019

View Large

During the wet season, the pH, ⁠, and COD were all below the TWQR. Total dissolved solids remained above the TWQR during the wet season ranging from 540.3–783.5 mg/L. Phosphate levels were significantly above the TWQR at sites KW03, TS04 and KA05 with concentrations of 4, 2.9 and 2.7 mg/L, respectively. Nitrates exceeded the TWQR at sites KW03 to GP08 ranging from 6–11 mg/L. In 2018, water quality at all sampling sites were closer to the TWQR during the dry season as compared to the wet season.

In 2019, exceeded the TWQR (0.4 mg/L) at all sampling sites with values ranging from 0.4 to 5.25 mg/L (Table 3). Total dissolved solids were higher than the TWQR at most sampling sites in 2019, except for sites TS04 and KA05 which had values of 413.67 and 363.33 mg/L during the wet season. Nitrite exceeded the TWQR at most sites, ranging from 6 to 39 mg/L. Nitrate exceeded the TWQR at four of the eight sites, with MU01, TS04, KA05 and VA07 at concentrations of 25.97, 8.03 11.67 and 6.17 mg/L, respectively. The remaining parameters were all within the TWQR apart from COD at KW03 (114.67 mg/L). As expected, water quality along the Loopspruit River was better during the dry season as compared to the standards set by the TWQR.

The World Health Organization (WHO) international standards had noticeable differences in TDS, nitrates, and COD standards. However, nitrites stand out, despite the fact that the WHO and TWQR standards for nitrites do not differ significantly. In Table 2 of 2018, nitrites and COD levels exceeded WHO standards in both seasons. The same is true in Table 3 for both the wet and dry seasons in 2019. The remaining WHO and TWQR parameters do not differ significantly.

Figure 5(b) shows that the Shannon-Wiener species diversity at site MD02 was the highest with a median of ∼3.68, with the whiskers ranging from a minimum of ∼2.62 to a maximum of ∼4.4. Site MU01 had the second-highest bacterial diversity, with a median of 3.6. Site VA07 had the lowest bacterial diversity which was significantly different (p < 0.05) from the other sites except for MU01 and MD02.

Geographically close sites formed associations as illustrated in Figure 6. Sites TS04 and KA05 are grouped with a range of ∼0.15 to ∼0.28. Both sites MU01 and KW03 display similarities close to zero (0) on both the y-axis and the x-axis. Sites VA07 and GP08 are also similar in terms of bacterial abundance. Sites KD06 and MD02 are on the outermost ranges, where KD02 has a distance distribution of −4.5 on the x-axis, and KD06 lies on a distance distribution value of >0.3.

The R script was modified to include genera from Escherichia (E. coli), Enterococcus and Clostridium spp. Clostridium spp ranged from ∼8% to 50% throughout all the sites. GP08 showed the highest Clostridium levels followed by KD06, VA07 and MU01. E. coli and Enterococcus were not detected in the Bray-Curtis dissimilarity dendrogram.

This study investigated the impacts of physico-chemical and microbiological parameters of the Loopspruit River. The nutrients that were investigated were ⁠, ⁠, and and COD as well as the following physical parameters: pH, temperature, and TDS. The study pointed out high nutrient deposits into the Loopspruit where agricultural practices are dominant and possible faecal pollution in the urban settings resulting in increased E. coli and Enterococcus levels. Additionally, the bacterial communities of the sample sites were generated to indicate the bacterial community composition at sites throughout the Loopspruit River.

pH remained relatively constant during the wet and dry seasons of 2018 and 2019, ranging from 7.40–9.40 (Tables 2 and 3). Various anthropogenic activities such as runoff from agricultural fields, mining activities and infiltration of treated or untreated wastewater can influence pH levels (Dallas & Day 2004). These are activities that are present in the Loopspruit River. In 2018, in the dry season, the pH at the agricultural sites was higher possibly due to the increase in total organic content (Xu & Zhang 2021) while grazing animals would use the Loopspruit River as a water source. However, during the wet season of 2018 and both wet and dry seasons of 2019, the pH at KD06 exceeded the TWQR (pH 8.84–9.40) also with the suspected increase of organic content, seeing it is a site for effluent from a WWTP.

A study by Jordaan et al. (2019) on the response of bacterial communities to anthropogenic contaminants in the Wonderfonteinspruit River showed similar results. The authors found that pH in the Wonderfonteinspruit River ranged from 7.17 to 8.29 and that this could be a result of dolomitic dissolution at the headwaters of the Wonderfonteinspruit River system. The Wonderfonteinspruit River system is located just north of the Loopspruit River.

Temperature is one of the main driving forces that affects almost all physico-chemical equilibria and biological processes (Delpla et al. 2009). The temperature in the Loopspruit River ranged from 8.4 °C at site MD02 in the cold dry season of 2019 to 25.5 °C during the warm wet season of 2019. Phosphate, nitrite and nitrate levels as well as TDS, increased as temperature increased (Tables 2 and 3). A downward trend in TDS, phosphates, nitrites and nitrates can be seen in 2018 when the temperature decreased.

The high TDS levels throughout the Loopspruit River decreased during the wet seasons which can be explained by a dilution effect. TDS remained above the TWQR downstream of urban land use areas. A study conducted in India, by Kora et al. (2017), screened physico-chemical and bacterial parameters and found that during their dry seasons the TDS levels ranged from 768.4–814.6 mg/L in August 2014 and 715.2–793.0 mg/L in September 2014. The authors also indicated that the TDS levels also exceeded the Indian water standards of 500 mg/L, but remained within the international WHO standards. This trend is similar to the trend observed in Loopspruit River during cold and dry seasons when the TDS levels are also elevated.

Van der Walt et al. (2002) found that TDS levels are usually elevated in dolomitic headwaters within the Mooi River catchment. This supports the findings in this study that the hard waters (dolomitic) of the Loopspruit River contributed to elevated levels of TDS. Acidic waters generated by mining activities result in the dissolution of dolomites (CaMg(CO₃)₂). This may contribute to the rise in macronutrients such as magnesium (Mg) and calcium (Ca) in the Mooi River catchments systems, including the Loopspruit River (Van der Walt et al. 2002), which increases the TDS levels. During the warm wet season, the TDS levels were slightly lower, potentially due to a dilution effect. Changes in TDS could also be due to changes in land-use activities.

Sulphate levels at sites MU01, MD02 and KW03 were high because of mining activities around these sample sites. During mining activities, processed debris is deposited in mine tailings dams where excess sulphates are carried through surface runoff to the Loopspruit River. Sulphate levels were lower at sample sites downstream of the WWTP (KD03) likely due to a dilution effect. High sulphate levels downstream of rural farming areas were likely the result of fertiliser use and subsequent surface runoff and sewage effluent. Similarly, the Toplluha River in Kosovo in 2019 had sulphate levels ranging from 28.9–189.9 mg/L which is below the WHO international standards (Shehu 2019).

The changes in COD correspond with the change in temperature where the COD levels increased during the cold dry seasons due to the decrease in oxidizable organic matter and lack of dilution within the Loopspruit River. Higher COD at lower temperatures might be a result of decreased microbial activity, lowering the breakdown of organic material (Iloms et al. 2020). Since there are several anthropogenic activities along the Loopspruit River, high COD levels observed in the present study are likely a result of these activities. An example of this is in 2019's wet season, where the COD in KW03 exceeded the TWQR (114.67 mg/L). This can possibly be due to that some organic substances that are not biodegradable, such as detergents, cause a high COD in wastewater (Majumdar & Sinha 2021).

Nitrite levels during 2018 remained relatively constant through the wet and dry seasons at site MU01 but increased from 4 to 34.3 mg/L in the wet season of 2019. In 2019 at site KA05, the increased from 11 to 39 mg/L. The increase seems like outliers and may be the result of a particular event, such as a large rainfall event during the week of sampling. Nitrate in the wet season of 2019 at sites MU01 and KA05 could be a result of fertiliser leachate that was collected and deposited in the Loopspruit River through surface runoff. There are agricultural activities upstream of both MU01 and KA05. The highest and levels were detected downstream of the agricultural areas of MU01, TS04 and KA05. Compared to international standards, the Loopspruit River has higher levels of nitrites such as in the study of Shehu (2019).

Phosphates remained relatively constant during both wet and dry seasons in 2018, except for increased levels at site KW03, with 1.1 mg/L during the dry season and 4 mg/L in the wet season. In 2019, levels at KW03 ranged from 0.40–5.25 mg/L exceeding the TWQR of 0.4 mg/L. Possible land use near KW03 is a military base where minimally treated sewage is deposited into the Loopspruit River. The phosphate level at KA05 was 1.25 mg/L, and this may largely be due to agriculture.

All physico-chemical parameters were acceptable according to the TWQR except for phosphates and TDS levels. Both were higher than the TWQR, and both are indicative of faecal pollution and agriculture in the Loopspruit River. These results correlate with the notion that increases with runoff from agricultural fields and sewage from urban settings (Jarvie et al. 2006).

A carbon, nitrogen and phosphorus ratio (C:N:P) of 60:7:1 is generally considered to be ideal for microbial growth to help the soil nutrients for agricultural practices (Frossard et al. 2016). When the microorganisms are in a poor phosphorus environment, they undergo ‘maintenance mode’ to remove the nutrient limitation resulting in eutrophication. This is indicative of pollution from agricultural activities depositing these nutrients in, for example the Loopspruit River through surface runoff. These high carbon and nitrogen levels promote cyanobacterial growth resulting in a eutrophic river system such as the high nitrate and nitrite levels at MU01 and KA05 which are downstream of agricultural practices.

The faecal contamination visual representation presented in Figure 3 indicates possible sources of contamination of the Loopspruit River. Surface runoff that carries water over land, picks up and washes faecal material and other chemical constituents into the Loopspruit River. The results depicted in Figure 3 show that the WWTP contributed to both E. coli and enterococci during the wet season. During the dry season, urban land use is a major contributing factor to both E. coli and enterococci levels because of increased anthropogenic activities, compared to other land uses with less anthropogenic activities. Agricultural activities contributed a moderate amount of E. coli and enterococci during the wet seasons and only enterococci during the dry seasons.

An increase in coliform concentration found in warm wet seasons can be seen in the current study where the urban land use (MU01, MD02 and GP08) generated significant faecal bacteria in the dry season. Increased rainfall with more runoff events and higher temperatures both contribute to higher faecal coliform counts during the wet season (Kim et al. 2017).

Bacterial alpha-diversity is one of the main driving forces behind nutrient cycling, litter decomposition, degradation of toxins, gas emissions and plant productivity (Delgado-Baquerizo & Eldridge 2019). The alpha-diversity is maintained by environmental factors such as pH, temperature, nutrient content, rainfall and climatic condition (Yashiro et al. 2016).

Alpha-diversity indices reflect the consistency and abundance of microbial communities. In the present study, significant bacterial diversity was found at sites MU01, MD02 and KW03 (Figure 5). Although these sites have the highest diversity, with MU01 and KW03 showing similar diversity, MD02 had a different species abundance. Dominating at all the sites were Proteobacteria, followed by Bacteroidetes, Cyanobacteria, Actinobacteria and Verrucomicrobia. This agrees with the findings that recorded the same dominating bacterial diversity by Jordaan et al. (2019) within the Wonderfonteinspruit catchment area, a neighbouring catchment. The physico-chemical and microbiological network analysis (Figure 9) shows that Proteobacteria favours conditions with a high EC, whereas the Bacteroidetes favour areas/periods with higher temperatures. Previous studies done on the effect of EC on bacterial diversity suggested that Bacteroidetes were positively correlated with the EC values (Kim et al. 2016). This is in contrast with the results of the present study. However, Proteobacteria and Verrucomicrobia results obtained here are similar to those found in a survey done by Kim et al. (2016) with a positive correlation to EC.

Beta-diversity differentiates between various microbial community compositions at all the sample sites. A previous study showed that the land use of an area actively shapes the bacterial community composition (Cai et al. 2018). This phenomenon can be seen in Figure 6 where agricultural land use (TS04 and KA05) grouped together and so did urban land-use activities (MU01, KW03, VA07 and GP08). The different land-use practices also showed unique plots where sample site MD02 is located downstream of mining activities, and KD06 downstream of the Klipdrift Dam. A study by Keshri et al. (2015) showed that the bacterial communities isolated near mining activities in other parts of South Africa are similarly dominated by Proteobacteria, Firmicutes, Bacteroidetes and Actinobacteria. Abundant genera such as Flavobacterium and Arcicella from the Bacteroidetes group were found at the first sample site of the Loopspruit River.

Within Actinobacteria there are potentially obligate and/or opportunistic pathogens including (but not limited to) Leifsonia (Evtushenko et al. 2000). Sites KD06, VA07 and GP08 along the Loopspruit River are downstream of a military base, urban and industrial land use activities, respectively. On the bacterial genera level, Flavobacterium makes up to 30% of the bacterial community at all sites as the most abundant bacteria. This was followed by Hydrogenophaga, previously classified under Pseudomonas, Acidovorax, Pseudomonas, Verrucomicrobia and Massilia and include potential pathogens, similar to the findings of Leclerc et al. (2002). Moreover, Clostridium spp. ranged from ∼8% to 50%, this can be indicative of possible faecal and organic pollution in the Loopspruit River. The diversity of the abundant genera in this study showed that this may be indicative of domestic sewage pollution in the Loopspruit River.

The water quality of any body of water is determined by the physico-chemical and microbial parameters of the water body. Correlating the physico-chemical data with the microbiological data provides an overall picture of how these bacteria make use of specific nutrients provided by the physico-chemical environment. One drawback related to using a Constrained Correspondence Analysis (CCA) plot is that it shows only the statistically significant (p < 0.05) physico-chemical and microbiological data, whereas a predicted metagenome analysis provides a better holistic perspective. The bacterial community in each environment is dependent on the relations of various factors such as pH, temperature, and the range of nutrients (Zhang et al. 2010). The negative correlation between temperature and Firmicutes, Nanoarchaeaeota, and Patescibacteria is the result of these organisms being thermophilic with an optimum growth rate at 50 °C to 90 °C (Bell et al. 2018; John et al. 2019; Toshchakov et al. 2021). The reason for the negative correlation is the low temperatures these organisms were found at, i.e. in 8–25 °C temperature range.

The results of this study provided an overview of the water quality of the Loopspruit River and indicate that it is generally at an acceptable water quality standard based on the physical, chemical, biological and aesthetic properties of water (Bui et al. 2019). Based on the research done, there is evidence that faecal pollution from urban, industrial, and agricultural practices in the Loopspruit River catchment is potentially deteriorating the water quality in this river. The nitrogen in ⁠, ⁠, and contribute to eutrophication in their dissolved inorganic forms, whereas contributes to algal blooms. Total coliforms in the environment are generally harmless, but pathogenic E. coli and Clostridium perfringens cause symptoms such as nausea, vomiting, and diarrhoea.

The monitoring plan for both physico-chemical and microbiological parameters needs to be put in place to maintain environmental sustainability for all the sectors reliant on the Loopspruit River. Integrated water quality management procedures should be put in place in the Loopspruit River to ensure that this resource remains available for agriculture and urban development. To safeguard and ensure desired water quality standards in aquatic ecosystems, researchers, management practice implementations and governmental bodies should work together to develop and continuously implement better water quality monitoring programs. Farmers and relevant stakeholders in collaboration with the Department of Human Settlements Water and Sanitation (DHSWS), which is the custodian of all water monitoring in South Africa, should drive processes to ensure the water in this catchment is protected

Funding from the National Research Foundation (NRF) through a grant holder bursary for LB (TTK170518231376) from Dr Molale-Tom, and grant UID 113824 (CCB) is hereby acknowledged. The opinions expressed and conclusions arrived at are those of the authors and are not necessarily to be attributed to the funders. The authors would also like to thank Dr Charlotte Mienie for the Sanger and MiSeq sequencing and Mr Schalk Combrinck for assisting with the sampling.

North-West University, Faculty of Natural and Agricultural Sciences Ethics number NWU-01376-20-A9

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.