The study aimed at assessing the microbiological quality and suitability of roof-harvested rainwater (RHRW) for crop irrigation and domestic use. In total, 80 rainwater tanks (246 samples) across three rural villages (Ga-Molepane, Jericho and Luthngele) were visited. Culture-based techniques were used to isolate bacterial microbes and identities were confirmed using matrix-assisted laser desorption/ionization time of flight (MALDI-TOF-MS). Uncultured fungal populations were also identified using pyrosequencing. Salmonella spp. (3%), Listeria monocytogenes (22%), total coliforms (57.7%), Escherichia coli (30.5%), Enterococcus spp. (48.8%), Pseudomonas spp. (21.5%) were detected in RHRW samples after rainfall. Fungal sequences belonging to species known to cause fever, coughing and shortness of breath in humans (Cryptococcus spp.) were identified. This study indicates that RHRW quality can be affected by external factors such as faecal material and debris on rooftops. The use of untreated RHRW could pose a potential health risk if used for irrigation of crops or domestic use, especially in the case of a relative high population of immunocompromised individuals. This study does not dispute the fact that RHRW is an alternative irrigation water source but it recommends treatment before use for domestic purposes or for watering crops.

INTRODUCTION

Approximately 11% of the world's population is without access to quality clean water. Despite all the development and infrastructure, an estimated 1,800 million people will still lack access to safe water by 2025 (UN 2012). Due to current water shortages, the use of potable municipal drinking water for irrigation in South Africa is not encouraged (Gemmell & Schmidt 2012). As a result, microbial quality of alternative water sources has received increased attention not only in South Africa but worldwide. Harvested rainwater is a realistic alternative source that can be used for irrigation and drinking purposes. Roof-harvested rainwater (RHRW) has been used for potable and non-potable purposes in many countries (Ahmed et al. 2014). Despite public health perception, rainwater poses a low risk to human health (Ahmed et al. 2014; Dobrowsky et al. 2014; Chidamba & Korsten 2015a). Various microorganisms, including human pathogens, could be present in dust, faeces of birds, insects and sometimes small mammals present on rooftops (Ahmed et al. 2010). During rainfall, faeces and other debris harbouring pathogens could be washed into rainwater harvesting tanks thus degrading its quality (Pachepsky et al. 2011). Pathogens have been reported in RHRW (Lye 2002; Lee et al. 2012; Ahmed et al. 2014) including bacteria, viruses and parasites, many of which could lead to waterborne illnesses (Hageskal et al. 2009).

The most frequently studied group of microorganisms with respect to microbial quality of water is bacteria. The presence of potentially pathogenic microorganisms, such as Listeria monocytogenes, Campylobacter spp., Salmonella spp., Escherichia coli (E. coli), Cryptosporidium spp. and Aeromonas spp. in RHRW has been reported (Ahmed et al. 2012; De Kwaadsteniet et al. 2013). Fungi are usually ignored with respect to microbial quality of water, mainly because many disease outbreaks related to use of contaminated water are caused by pathogenic bacteria (Hageskal et al. 2009). However, many fungal species have been detected in water including Fusarium oxysporum Schltdl., Fusarium moniliforme Sheld., Aspergillus flavus Link, Penicillium chrysogenum Thom., Penicillium corylophilum Dierckx, Acremonium strictum Gams, Phialophora bubakii Schol. and Phoma exigua Sacc. (Monchy et al. 2011); some of these are pathogenic to humans or are known plant pathogens. Hageskal et al. (2009) stated that Aspergillus fumigatus Fresen is one of the most important fungal pathogens detected in water that could cause infections in immunocompromised individuals. The presence of pathogenic fungi in water has been considered as a chronic and possibly underestimated problem (Hageskal et al. 2009). Plant pathogens play an important role in infecting irrigated crops during the seedling stage (Bell et al. 2006), growing conditions or post-harvest (Prusky 1996). No study has reported on the link between RHRW and crop disease or decay potential due to the presence of plant pathogens.

Reports on the role of irrigation water in the contamination of produce, subsequently leading to foodborne illness outbreaks are increasing (Allende & Monaghan 2015). The transfer of any pathogen in water used for irrigation onto crops could be risky if produce is consumed fresh. Studies have linked the presence of pathogens on fresh produce to contaminated water (Pachepsky et al. 2011; Du Plessis et al. 2015). The presence and subsequent transfer of fungal pathogens from irrigation water onto fresh produce has not been fully investigated.

Traditional culturing assays underestimate microbial diversity due to methodological constraints and limitations including the ones used for fungal populations. Previous studies on fungal analysis of water used only culturing methods. Disadvantages of culturing fungi include (1) inability to grow non-sporulating fungi and (2) diverse nutrient requirements (Hageskal et al. 2009). As a result, reliance on molecular methods has received global recognition. The 454 pyrosequencing platform is a recent type of next generation sequencing technology. This technique is a good diagnostic tool for detection and identification of microbial pathogens. Despite being an expensive technique, the 454 pyrosequencing technique is rapid and flexible with 99% accuracy (Liu et al. 2015). Therefore, this study aims to assess the microbiological quality of RHRW for irrigation and domestic purposes using culturable and non-culturable approaches for bacteria and fungi, respectively. In addition, the presence of potential human pathogens (bacterial and fungal) was also assessed as part of the microbial population.

MATERIALS AND METHODS

Study sites and sample collection

Sampling was conducted at three rural villages in South Africa: Ga-Molepane (Limpopo Province), Jericho (North West Province) and Luthengele (Eastern Cape Province). All villages experience serious water shortages, hence the communities rely on RHRW. Ga-Molepane and Luthegele villages have no boreholes and households depend entirely on RHRW during the wet season supplemented with stream water during dryer months. Domesticated animals use the stream as a water source. Jericho village has boreholes which often are not adequate to supply the community with enough water for domestic use. At Luthengele village, leafy green vegetables (mainly cabbage) are grown but RHRW is not used for watering. In total, 246 RHRW samples were collected from water tanks during the 2013–2014 rainy seasons (September to May). The strategy was to sample before and after the onset of the main rainy season. Randomly selected households (80) were used for this study. Each site was visited twice in 2013 and repeated in 2014 following the same strategy, before and after onset of the rainy season and sampling from the same households. Sanitary factors, i.e., overhanging trees, animal faecal material and other debris were identified on the rooftops. Samples were collected from the outlet taps located close to the base of the water tanks, in sterilized 2 litre (L) bottle containers. Prior to collecting water samples, the tap was sterilized with 70% ethanol, and water was allowed to run for 30 to 60 s to flush out stagnant water from the tap (Ahmed et al. 2010). After collection, the samples were stored on ice in cooler boxes during transportation. All samples were brought back to Pretoria University Plant Pathology Laboratories, and processed within 24 hr.

Enumeration and isolation of bacteria

Microbes in water were concentrated by filtering 750 mL of water through 0.45-μm pore-sized (47 mm diameter) nitrocellulose membranes (Sartorious Stedim Biotech, Gottingen, Germany) by vacuum filtration (Shi et al. 2013), and membranes aseptically transferred into 9 mL of sterile tryptic soy broth (TSB) (Merck, Johannesburg, South Africa). The tubes were vortexed vigorously for 2 to 5 minutes (min) to detach the bacteria from the membranes. After serial dilutions, 100 microlitre (μL) aliquots were plated out in triplicates onto Reasoner's 2A agar (R2A) (Merck) for heterotrophic plate counts. Membrane filters in 9 mL TSB were incubated at 37 °C for 24 hr with agitation (150 rpm). Afterwards, 0.1 mL of the pre-enriched broth was inoculated into 10 mL Rappaport-Vasidallis and Listeria enrichment broth for selective enrichment of Salmonella (Gomba et al. 2016) and L. monocytogenes, respectively. Rappaport-Vasidallis was incubated at 42 °C for 24 hr while Listeria enrichment broth was incubated at 37 °C for 18–24 hr. A loopful of each selective enrichment culture with growth was streaked onto xylose lysine deoxycholate agar (for Salmonella spp.) and Oxford-Listeria selective agar (for L. monocytogenes) plates.

All media were incubated at 37 °C for 24–48 hr except R2A plates (25 °C for 24–72 hr) and Salmonella enrichment broth. All media used were obtained from Merck. After incubation, colonies recovered using R2A were counted, recorded and expressed as colony-forming units per mL.

Indicator system

Densities of total coliforms, E. coli, Pseudomonas and Enterococci were determined with Colilert-18, Pseudolert-18 and Enterolert-18 chromogenic substrate tests kits and Quantitray 2000 trays (Idexx, Westbrook, Maine, USA), as per the manufacturer's instructions. For highly contaminated samples, 1 ml was homogenized in 99 mL sterile water to satisfy the 100 mL requirement for quantification using Quantitray 2000 trays. The manufacturer's most probable number (MPN) data table was used to generate microbial density estimates per 100 mL.

After incubation (37 °C for 18–24 hr), the backing material of each Quantitray was disinfected with 70% ethanol and a sterile razor blade used to pierce the backing material of three fluorescence positive wells per tray (Chidamba & Korsten 2015a). A loop full of well content was streaked onto E. coli, Enterococcus spp. and Pseudomonas spp. selective agar (Merck) as described before (Chidamba & Korsten 2015a). All presumptive isolates were purified on nutrient agar (NA) (Merck) and confirmed with matrix-assisted laser desorption ionization (MALDI-TOF-MS) analysis (MALDI Biotyper® system, Bruker Daltonics, Billerica, MA, USA).

Matrix-assisted laser desorption ionization identification

Presumptive isolates were purified on NA prior to MALDI-TOF-MS analysis. The automated process from MALDI-TOF-MS measurement to species identification occurs without user intervention (Pinto et al. 2011; Chidamba & Korsten 2015b). Using default settings, the manufacturer's MALDI Biotyper 3.0 software (Bruker Daltonics, Germany) analyzed spectra of the isolates and compared it with reference spectra in the database. A logarithmic score is generated corresponding to similarity of spectral patterns and interpreted as per manufacturer's instructions. Results are conveyed as log (score) values ranging from 0 to three levels. Scores designate high confidence level species identification (≥2,300), species identification (≥2,000), genus identification (1,700–1,999) and no identification (<1,700).

DNA extraction

Genomic DNA (gDNA) was isolated directly from filter paper, with a ZM fungal/bacterial DNA miniprep™ kit (Zymo Research Corporation, Inqaba Biotech catalogue # D6005) as per the manufacturer's specifications. The DNA concentration was determined with the Qubit 2.0 Fluorometer (Lifescience Technology, Johannesburg, South Africa). Isolated gDNA was stored at −20 °C and kept on ice during further experimental procedures.

Pyrosequencing analysis

DNA samples were sent to Inqaba Labs (Pretoria) for tag encoded pyrosequencing of a portion of the fungal internal transcribed spacer (ITS) region using the Roche GS FLX + 454 pyrosequencer (GATC Biotech, Konstanz, Germany). The fungal ITS region was amplified using primer set ITS1F (CTTGGTCATTTAGAGGAAGTAA) and ITS4R (TCCTCCGCTTATTGATATGC) (McHugh et al. 2014). Bacterial pyrosequencing is not presented in this paper but has been presented in another paper (Chidamba & Korsten 2015b). The matrix-assisted laser desorption/ionization time of flight (MALDI-TOF-MS) database restricted fungi identification, hence we relied on 454 fungal pyrosequencing to identify isolates.

Pyrosequencing data processing and analysis

Pyrosequencing data analysis for selected samples (Tanks 1–5) was performed using QIIME as established in previous studies (McHugh et al. 2014), with modifications. Selection of samples for analysis was based on population diversity. Sequences with less than 200 bp and quality score less than 25 were removed from the pyrosequencing-derived data sets using the UCLUST pyrosequencing pipeline initial processing (Cleary et al. 2012). The web-server SnoWMAn 1.15 (http://snowman.genome.tugraz.at) (Stocker et al. 2011) was used with the following settings for the taxonomic-based analysis: analysis type: UCLUST pipeline; chimera filtering reference database: Chimerarefdbs_09_Feb_2012_aligned; Mothur alignment reference database: fungalITS_24-Sep-2013-current_prokMSA_aligned; Mothur chimera checking reference database: chimerarefdbs_39-Feb-2012_16S_aligned and taxonomic reference database: greengenes_24-Mar-2010; rarefaction method: RDP; taxonomy: RDP; confidence threshold: 80%; include taxa covering more than 1%.

Operational taxonomic units for rarefaction analysis were clustered at 97% (species level), 95% (genus level) and 80% (phylum level) similarity cut-offs (Yang et al. 2014). The calculation of ACE (abundance-based coverage estimator), Evenness, Shannon, Chao1 and Species Richness diversity indices and the generation of principal coordinate analysis plots were generated using the open source software package QIIME (http://qiime.sourceforge.net), designed for combining heterogeneous experimental data sets, analysis of high throughput community sequencing data and obtaining new insights about various microbial communities (Caporaso et al. 2010).

Statistical analysis

Statistical analyses were conducted with SAS statistical software (version 9.3). Analysis of variance tests were performed and the least significance difference test used to derive statistical differences (P> 0.05) of microbial loads among all sampling areas.

RESULTS

Microbial quality of RHRW

Out of 246 samples, average total coliforms (57.7%), E. coli (30.5%), Enterococcus spp. (48.8%), Pseudomonas spp. (21.5%) and total aerobic bacteria (100%) were detected after rainfall. Prevalence of total coliforms, Enterococcus and Pseudomonas from all sites was not significantly different before and after rainfall. Before the rainy season, total coliforms were more prevalent in water from Ga-Molepane (43%), followed by Luthengele (35%) and Jericho (31%). Total coliforms were more prevalent in water samples from Ga-Molepane (73.7%) while the least prevalence was detected in samples from Luthengele after rainfall (Table 1).

Table 1

Average microbial loads in RHRW before and after raina

Season Site Sample number APC (cfu/mL) Total coliformsb E. coli Enterococci Pseudomonas 
 Ga-Molepane 38 5.32 ± 0.68 (100%) 1.2 ± 1.2 (43%) 0.52 ± 0.87 (23.5%) 1.45 ± 1.53 (27.1%) 0.45 ± 1.05 (31.3%) 
1 (Before rainfall) Jericho 89 5.91 ± 1.13 (100%) 1.22 ± 1.1 (31% 0.02 ± 0.1 (5%) 2.03 ± 0.98 (13.6%) 0.48 ± 1.11 (14%) 
 Luthengele 119 5.61 ± 0.71 (100%) 1.11 ± 1.34 (35%) 0.35 ± 0.67 (18.7%) 1.52 ± 1.47 (45.8%) 0.19 ± 0.62 (17%) 
 Ga-Molepane 38 5.26 ± 0.69 (100%) 1.62 ± 1.25 (73.7%) 0.82 ± 1.02 (31.6%) 0.44 ± 0.7 (42.1%) 0.98 ± 1.1 (36.8%) 
2 (After rainfall) Jericho 89 5.67 ± 1.03 (100%) 1.47 ± 1.14 (74%) 0.66 ± 0.85 (29.2%) 0.13 ± 0.35 (48.3%) 0.8 ± 1.27 (21.3%) 
 Luthengele 119 5.36 ± 0.73 (100%) 2.19 ± 1.13 (40.3%) 0.26 ± 0.49 (31.1%) 0.62 ± 0.72 (51.2%) 0.52 ± 1.08 (16.8%) 
Season Site Sample number APC (cfu/mL) Total coliformsb E. coli Enterococci Pseudomonas 
 Ga-Molepane 38 5.32 ± 0.68 (100%) 1.2 ± 1.2 (43%) 0.52 ± 0.87 (23.5%) 1.45 ± 1.53 (27.1%) 0.45 ± 1.05 (31.3%) 
1 (Before rainfall) Jericho 89 5.91 ± 1.13 (100%) 1.22 ± 1.1 (31% 0.02 ± 0.1 (5%) 2.03 ± 0.98 (13.6%) 0.48 ± 1.11 (14%) 
 Luthengele 119 5.61 ± 0.71 (100%) 1.11 ± 1.34 (35%) 0.35 ± 0.67 (18.7%) 1.52 ± 1.47 (45.8%) 0.19 ± 0.62 (17%) 
 Ga-Molepane 38 5.26 ± 0.69 (100%) 1.62 ± 1.25 (73.7%) 0.82 ± 1.02 (31.6%) 0.44 ± 0.7 (42.1%) 0.98 ± 1.1 (36.8%) 
2 (After rainfall) Jericho 89 5.67 ± 1.03 (100%) 1.47 ± 1.14 (74%) 0.66 ± 0.85 (29.2%) 0.13 ± 0.35 (48.3%) 0.8 ± 1.27 (21.3%) 
 Luthengele 119 5.36 ± 0.73 (100%) 2.19 ± 1.13 (40.3%) 0.26 ± 0.49 (31.1%) 0.62 ± 0.72 (51.2%) 0.52 ± 1.08 (16.8%) 

APC, aerobic plate counts; (), prevalence percentages.

aValues are log mean ± standard deviation.

bValues are MPN estimates unless otherwise stated.

For season 1, concentrations of E. coli isolated from Ga-Molepane were significantly different from E. coli isolated from Jericho, but not significantly different from Luthengele E. coli isolates. Higher levels of E. coli were detected in water samples from Ga-Molepane (31.6%), followed by Luthengele (31.1%) and Jericho (29.2%) after rainfall. At least 45 and 51% of water samples from Luthengele represented Enterococcus spp. before and after rainfall, respectively, while 21.3% accounted for Pseudomonas spp. in RHRW from Jericho after rainfall. In general, water samples from Ga-Molepane had the highest concentrations of faecal coliforms, E. coli, Enteroccoccus spp. and Pseudomonas spp. (Table 1). Total aerobic bacteria ranged from a geometric mean of 5.26 (Ga-Molepane) to 5.91 (Jericho) log CFU/mL (Table 1). Means for aerobic plate counts from Ga-Molepane, Jericho and Luthengele were significantly different from one another, with samples from Jericho recording the highest and samples from Ga-Molepane recording the lowest.

Salmonella spp. and L. monocytogenes were present in RHRW from Ga-Molepane, Jericho and Luthengele villages. In total, 2 and 29 (before rainfall), 5 and 43 (after rainfall) RHRW samples tested positive for Salmonella spp. and L. monocytogenes, respectively (Table 2). Generally, the prevalence of L. monocytogenes and Salmonella spp. increased after rainfall at all sampling sites (Table 2). The prevalence of L. monocytogenes was 18.2 (before rainfall) and 31.1% (after rainfall) in RHRW samples from Jericho. Salmonella spp. was not detected before rainfall in water from Ga-Molepane and Jericho but was present after rainfall. The number of Salmonella positive samples increased (from 0 (before rainfall) to 2 (after rainfall)) in RHRW samples from Ga-Molepane.

Table 2

Prevalence of human pathogenic bacteria in RHRW

  Number of positive samples/total number of samples (%)
 
Before rain
 
After rain
 
Site Salmonella L. monocytogenes Salmonella L. monocytogenes 
Ga-Molepane 0 (0) 2 (10.5) 2 (10.5) 4 (21.1) 
Jericho 0 (0) 8 (18.2) 1 (2.2) 14 (31.1) 
Luthengele 2 (3.4) 19 (32.2) 2 (3.4) 25 (41.6) 
  Number of positive samples/total number of samples (%)
 
Before rain
 
After rain
 
Site Salmonella L. monocytogenes Salmonella L. monocytogenes 
Ga-Molepane 0 (0) 2 (10.5) 2 (10.5) 4 (21.1) 
Jericho 0 (0) 8 (18.2) 1 (2.2) 14 (31.1) 
Luthengele 2 (3.4) 19 (32.2) 2 (3.4) 25 (41.6) 

Matrix-assisted laser desorption time of flight

Bacterial isolates from all sampling sites, as identified by MALDI-TOF-MS, were tabulated (Table 3). Of the total, opportunistic pathogens (Staphylococcus spp. (Niazi et al. 2010) (13%), Enterococci spp. (Shioya et al. 2011) (7%) and Pseudomonas spp. (de Bentzmann & Plésiat 2011) (3%)) and known potential human pathogens (L. monocytogenes (6%) and Salmonella spp. (4%)) were identified.

Table 3

Frequency of selected significant dominant bacterial isolates from RHRW as identified by matrix-assisted laser desorption time of flight mass spectrometrya

Species name Ga-Molepane Jericho Luthengele 
Aeromonas caviae 
Aeromonas jandaei 
Aeromonas veroni 
Bacillus cereus 
Bacillus pumilus 
Enterobacter amnigenus 
Enterobacter asburiae 
Enterobacter cancerogenus 
Enterococcus faecalis 
Enterococcus faecium 15 
Enterococcus durans 12 
Enterococcus mundtii 
Escherichia vulneris 11 
E. coli 15 11 
Klebsiella pneumonia 
Listeria monocytogenes 
Morganellamorganii 
Pantoea agglomerans 
Pseudomonas otidis 
Pseudomonas taetrolens 
Salmonella spp. 
Staphylococcus auricularis 12 
Staphylococcus capitis 
Species name Ga-Molepane Jericho Luthengele 
Aeromonas caviae 
Aeromonas jandaei 
Aeromonas veroni 
Bacillus cereus 
Bacillus pumilus 
Enterobacter amnigenus 
Enterobacter asburiae 
Enterobacter cancerogenus 
Enterococcus faecalis 
Enterococcus faecium 15 
Enterococcus durans 12 
Enterococcus mundtii 
Escherichia vulneris 11 
E. coli 15 11 
Klebsiella pneumonia 
Listeria monocytogenes 
Morganellamorganii 
Pantoea agglomerans 
Pseudomonas otidis 
Pseudomonas taetrolens 
Salmonella spp. 
Staphylococcus auricularis 12 
Staphylococcus capitis 

aValues are in percentages calculated from the total isolates identified.

Of the total bacteria isolated from Ga-Molepane, E. coli prevalence was 15%, the highest compared to other sites. Various species belonging to Enterococci were identified. These included E. faecalis (only Ga-Molapane village) and E. faecium (all sites) (Table 3). Non-human pathogenic environmental bacterial isolates, among them, Acinetobacter spp. and Anaerococcus spp., were also detected (results not shown).

Fungal pyrosequencing

At phylum level classification, fungal sequences were classified into five phyla (Figure 1). These include Ascomycota, Basidiomycota, Chytridiomycota, Glomeromycota and Zygomycota. Ascomycota dominated the data set (32.1%), while 45.7% of the sequences were not classified to any known fungal phyla.
Figure 1

Phylum level pyrosequencing classification of selected RHRW samples.

Figure 1

Phylum level pyrosequencing classification of selected RHRW samples.

Of the identified sequences 17, 4.8 and 0.09% were classified to Basidiomycota, Chytridiomycota and Glomeromycota, respectively (Figure 1). There were notable differences in fungal diversity in the selected RHRW samples. Fungal sequences (95%) from Tank 1 were not identified while all sequences in Tank 4 were identified. Ascomycota was identified in four out of the five tanks (Tanks 2, 3, 4 and 5). The highest concentration of Chytridiomycota (70%) was identified in Tank 2. Only two fungal phyla (Basidiomycota and Zygomycota) were identified in Tank 1 RHRW (Figure 2). Tanks 2 and 3 had more diverse fungal populations with five identified sequences represented in both RHRW tanks.
Figure 2

Species level relative abundance profiles using 18S rRNA sequence classifications of a water tank with RHRW.

Figure 2

Species level relative abundance profiles using 18S rRNA sequence classifications of a water tank with RHRW.

Classification at species level revealed a diverse fungal population. Sordariomycetes species and Davidiella tassiana were more predominant in Tank 3 water samples (29.3% and 18.1%, respectively) than other water samples (Figure 2). Sequences of species in Tank 3 (13.9%) remained unclassified. Other species also identified included Cryptococcus spp., Dothideomycetes spp., Tremellales spp. and Knufia perforans Sterflinger.

DISCUSSION

In this study, microbial quality of RHRW was tested for the presence of bacterial indicators (total coliforms, E. coli and Enterococci), opportunistic (Pseudomonas spp.) and potential pathogenic microorganisms (L. monocytogenes and Salmonella spp.), and fungal organisms. To assess the suitability of RHRW for irrigation and domestic use conventional culture-based methods were used. There were significant variations of E. coli and aerobic plate counts between the three rural villages sampled. A previous study reported similar variations in the microbiological quality of RHRW (Evans et al. 2006a). The study further reported extensive disparities in the level of water pollution and bacterial composition between sites, and over time. Results of our study indicate the presence of potential human pathogenic bacteria in RHRW from all the sites tested.

Salmonella spp. (3%) and L. monocytogenes (22%) were detected in RHRW samples from all three sampling areas. Ahmed et al. (2008) detected Salmonella spp. but not L. monocytogenes in RHRW. A previous study indicated that drinking RHRW containing Salmonella spp. caused diarrhoea and abdominal pains (De Kwaadsteniet et al. 2013). Based on observations during this study, RHRW was being contaminated with rooftop faecal material and dust. The source of this contamination was not specifically studied in this paper but it has been observed and previously described in other studies (Evans et al. 2006b; Ahmed et al. 2012).

The rainfall patterns of Jericho and Ga-Molepane are similar and the rainfall events occur seasonally. Hence, E. coli concentration in RHRW is significantly higher immediately after rainfall events. Luthengele area, on the other hand, experiences rain events more frequently. As a result, there is little time for accumulation of faecal material and other debris on rooftops. This could partially explain E. coli concentrations not being significantly different before and after rain events. The low coliforms and E. coli counts detected in RHRW before the start of the rainy season may suggest that the rooftops are more consistently washed down by pouring rainwater and are therefore not consistently highly contaminated. However, the presence of Enterococci spp., L. monocytogenes and Salmonella spp. suggests otherwise. The mere presence of L. monocytogenes in water does not mean infection is possible upon consumption. However, the pathogen can proliferate in storage tanks leading to high quantities suitable for the onset of listeriosis (Morobe et al. 2009; Golberg 2015).

In a recent study, low counts of E. coli were observed but detection of faecal indicator bacteria (Enterococci spp.) suggested the water may be contaminated (Chidamba & Korsten 2015a). Although the concentration of E. coli most likely decreased with time, Enterococci and other pathogens persisted owing to their resilience (Ahmed et al. 2012). This study therefore finds RHRW a potential hazard if used for domestic purposes or to irrigate crops due to the presence of well-known human pathogens. However, it is worth noting that using polluted irrigation water does not always result in contaminated crops.

Basic systems of roof-water harvesting are used in the sampling sites visited. The system is deprived of first flush diverters and filtration apparatus (Chidamba & Korsten 2015a). Due to this deprivation, microbiological quality of the RHRW is compromised and treatment prior to use is thus necessary. The absence of these first flush diverters contribute to increased chances of contamination of RHRW, hence faecal material and other debris on the roof and/or gutter washes into rainwater tanks. Chidamba & Korsten (2015a) established a link between indicator bacteria and pathogens isolated from rooftop surfaces and RHRW. The presence of such pathogens could be transmitted to leafy green vegetables when RHRW is used for irrigation purposes. Several studies have linked irrigation water with pathogens isolated from fresh produce (Allende & Monaghan 2015; Gelting et al. 2015), therefore, substandard irrigation water is a contamination risk factor of fresh green leafy vegetables. The presence of human pathogens on irrigated crops due to the use of contaminated irrigation water is well described (Pachepsky et al. 2011).

In this study, heterotrophs were isolated in RHRW at all sampling sites in concentrations higher than recommended for domestic and irrigation purposes according to the Department of Water Affairs Standard, South Africa (DWAF 1996). Concentrations higher than 1,000 CFU/mL in water used for drinking indicate contamination or after-growth in the water and pose an increased risk of infectious disease transmission if consumed (DWAF 1996). Even though heterotrophic bacterial counts indicate the general microbial quality of water, they do not indicate possible faecal pollution (DWAF 1996). However, DWAF (1996) recommends that water with high concentrations of heterotrophs requires treatment prior to use. The prevalence of Enterococci was higher than E. coli in this study. This is consistent with other reports (Spinks et al. 2006; Ahmed et al. 2011). These results suggest that RHRW from these sampling sites is not fit-for-irrigation of fresh products consumed raw unless treated prior to use according to the South African irrigation water standards (DWAF 1996).

The reliance of total coliforms and indicator bacteria for water quality assessment has come under scrutiny in the recent past (Chidamba & Korsten 2015a), as findings emerged that pathogenic E. coli could not be isolated from water samples positive for indicator bacteria of faecal origin. However, the use of total coliforms and indicator bacteria to evaluate water quality is by far the most widely used method in laboratories. The simultaneous use of E. coli and Enterococci for potential faecal contamination assessment is recommended (Savichtcheva & Okabe 2007; Ahmed et al. 2011; Chidamba & Korsten 2015a), and our study also supports this approach.

Although the presence of human bacterial pathogens has been widely reported in RHRW (Ahmed et al. 2011), the same cannot be said about human fungal pathogens. Gikas & Tsihrintzis (2012) mentioned that water harvesting systems could transmit microorganisms able to affect human health. Conventional culture techniques underestimate microbial populations (Ahmed et al. 2011), hence utilization of molecular methods is increasingly used to provide a better understanding of spp. present. Pyrosequencing of the ITS 1 and ITS 2 hypervariable regions of the 18S rRNA was used to investigate the fungal communities in RHRW. Results of this study indicated a diverse fungal community in RHRW.

Pencillium and Cryptococcus spp. were mostly detected in RHRW. Some Penicillium spp. are pathogenic to plants and others can affect human health, such as P. digitatum (Pers.) Sacc. (Hageskal et al. 2009; Oshikata et al. 2013). Further, mycotoxins and other metabolites that are produced by these organisms have been detected in storage water tanks (Paterson et al. 1997). Prolonged water storage may therefore lead to increased mycotoxin concentration, and could pose a potential risk to consumers with persistent consumption (Hageskal et al. 2009). In such a case, immunocompromised individuals are at a higher risk due to their diminished immune system. HIV prevalence in rural communities in South Africa has been reported (Shisana et al. 2014) and is considered high (39.4%) by world standards.

Human pathogenic yeasts, Cryptococcus spp., belonging to a diverse filamentous fungal division Basidiomycota, were also detected in RHRW. Cryptococcus spp. is an opportunistic human pathogen of global importance (Loftus et al. 2005), which has been isolated in bird faeces (Soltani et al. 2013). Effects of continuous exposure to Cryptococcus spp. include fever, coughing and shortness of breath.

Monchy et al. (2011) highlighted the significance of further characterization of human pathogenic microbes and understanding their effects on human health due to persistent exposure. The risk posed by the presence of sequences belonging to phylotypes with a known pathogenic history cannot be underestimated. The presence of unidentified sequences signify their absence in cluster analysis databases, hence the need for research to identify more environmental fungal populations and collection of curated sequences for fungal identification (Liu et al. 2015). The use of RHRW with human pathogenic isolates for domestic use and irrigation purposes without any prior treatment could therefore be a cause of concern if pathogens persist up to the point of consumption. Future studies should focus on the link between fungal spp. in RHRW and crop disease or decay.

CONCLUSION

This study concluded that RHRW collected in three rural villages was of poor microbiological quality, therefore not fit for domestic use or irrigation purposes unless treated prior to use as per national guidelines (DWAF 1996). The presence of known human pathogenic microorganisms and faecal indicators in RHRW indicate a potential risk if the crops are also irrigated with this water and the product is consumed raw. Better maintenance of RHRW systems are required to reduce contamination levels. Development of multiple detection systems using faecal indicators, E. coli, Enterococci and next generation sequencing techniques could holistically improve water quality assessment.

ACKNOWLEDGEMENTS

This study was part of a solicited research project (K5/1875/4: Evaluation of the risks associated with the use of rain-water harvested from rooftops, for domestic use and homestead food gardens; and groundwater for domestic use and livestock watering) funded by the Water Research Commission and co-funded by the Department of Agriculture, Forestry, and Fisheries, Republic of South Africa. We acknowledge the editorial input by Mrs. Annancietar Gomba and Dr Erika du Plessiss. We are grateful for the assistance of Dr Lizyben Chidamba, Sphumelele Mkhize and Dennis Makuwa.

REFERENCES

REFERENCES
Caporaso
J. G.
Kuczynski
J.
Stombaugh
J.
Bittinger
K.
Bushman
F. D.
Costello
E. K.
Fierer
N.
Peña
A. G.
Goodrich
J. K.
Gordon
J. I.
Huttley
G. A.
Kelley
S. T.
Knights
D.
Koenig
J. E.
Ley
R. E.
Lozupone
C. A.
McDonald
D.
Muegge
B. D.
Pirrung
M.
Reeder
J.
Sevinsky
J. R.
Turnbaugh
P. J.
Walters
W. A.
Widmann
J.
Yatsunenko
T.
Zaneveld
J.
Knight
R.
2010
QIIME allows analysis of high-throughput community sequencing data
.
Nature Methods
7
(
5
),
335
336
.
de Bentzmann
S.
Plésiat
P.
2011
The Pseudomonas aeruginosa opportunistic pathogen and human infections
.
Meeting report
.
Environ. Microbiol.
13
(
7
),
1655
1665
.
De Kwaadsteniet
M.
Dobrowsky
P. H.
van Deventer
A.
Khan
W.
Cloete
T. E.
2013
Domestic rainwater harvesting: microbial and chemical water quality and point-of-use treatment systems
.
Water Air Soil Pollut.
224
,
16
29
.
Department of Water Affairs
1996
South African Water Quality Guidelines
.
2nd edn
.
Vol. 4
.
Department of Water Affairs
,
Pretoria
,
South Africa
.
Dobrowsky
P. H.
De Kwaadsteniet
M.
Cloete
T. E.
Khan
W.
2014
Distribution of indigenous bacterial pathogens and potential pathogens associated with roof-harvested rainwater
.
Appl. Environ Microbiol.
80
(
7
),
2307
2316
.
Evans
C. A.
Coombes
P. J.
Dustan
R. H.
Harrison
T.
Martin
A.
Harris
J. N.
2006a
Coliforms, biofilms, microbial diversity and the quality of roof-harvested rainwater
.
Water Res.
40
(
1
),
34
44
.
Gelting
R. J.
Baloch
M. A.
Zarate-Bermudez
M.
Hajmeer
M. N.
Yee
J. C.
Brown
T.
Yee
B. J.
2015
A systems analysis of irrigation water quality in an environmental assessment of an E. coli O157:H7 outbreak in the United States linked to iceberg lettuce
.
Agric. Water Manag.
150
,
111
118
.
Gomba
A.
Chidamba
L.
Korsten
L.
2016
Prevalence and serovar diversity of Salmonella spp. in primary horticultural fruit production environments
.
Food Control
69
,
13
16
.
Hageskal
G.
Lima
N.
Skaar
I.
2009
The study of fungi in drinking water – the review
.
Mycol. Res.
113
,
165
172
.
Liu
P.
Wang
X.
Li
J.
Qin
W.
Xiao
W.
Zhao
X.
Jiang
H.
Sui
J.
Sa
R.
Wang
W.
Liu
X.
2015
Pyrosequencing reveals fungal communities in the rhizosphere of Xinjiang Jujube
.
BioMed. Res. Int.
1
8
,
doi:10.1155/2015/972481
.
Loftus
B. J.
Fung
E.
Roncaglia
P.
Rowley
D.
Amedeo
P.
Bruno
D.
Vamathevan
J.
Miranda
M.
Anderson
I. J.
Fraser
J. A.
Allen
J. E.
Bosdet
I. E.
Brent
M. R.
Chiu
R.
Doering
T. L.
Donlin
M. J.
D'Souza
C. A.
Fox
D. S.
Grinberg
V.
Fu
J.
Fukushima
M.
Haas
B. J.
Huang
J. C.
Janbon
G.
Jones
S. J.
Koo
H. L.
Krzywinski
M. I.
Kwon-Chung
J. K.
Lengeler
K. B.
Maiti
R.
Marra
M. A.
Marra
R. E.
Mathewson
C. A.
Mitchell
T. G.
Pertea
M.
Riggs
F. R.
Salzberg
S. L.
Schein
J. E.
Shvartsbeyn
A.
Shin
H.
Shumway
M.
Specht
C. A.
Suh
B. B.
Tenney
A.
Utterback
T. R.
Wickes
B. L.
Wortman
J. R.
Wye
N. H.
Kronstad
J. W.
Lodge
J. K.
Heitman
J.
Davis
R. W.
Fraser
C. M.
Hyman
R. W.
2005
The genome of the Basidiomycetous yeast and human pathogen Cryptococcus neoformans
.
Science
307
(
5713
),
1321
1324
.
Monchy
S.
Sanciu
G.
Jobard
M.
Rasconi
S.
Gerphagnon
M.
Chabé
M.
Cian
A.
Meloni
D.
Niquil
N.
Christaki
U.
Viscogliosi
E.
Sime-Ngando
T.
2011
Exploring and quantifying fungal diversity in freshwater lake ecosystems using rDNA cloning/sequencing and SSU tag pyrosequencing
.
Environ. Microbiol.
13
(
6
),
1433
1453
.
Morobe
I. C.
Obi
C. L.
Nyila
M. A.
Gashe
B. A.
Matsheka
M. I.
2009
Prevalence, antimicrobial resistance profiles of Listeria monocytognes from various foods in Gaborone, Botswana
.
Afr. J. Biotech.
8
(
22
),
6383
6387
.
Oshikata
C.
Tsurikisawa
N.
Saito
A.
Watanabe
M.
Kamata
Y.
Tanaka
M.
Tsuburai
T.
Mitomi
H.
Takatori
K.
Yasueda
H.
Akiyama
K.
2013
Fatal pneumonia caused by Penicillium digitatum: a case report
.
BMC Pulm. Med.
13
,
16
.
doi:10.1186/1471-2466-13-16.
Pachepsky
Y.
Shelton
D. R.
McLain
J. E. T.
Patel
J.
Mandrell
R. E.
2011
Irrigation waters as a source of pathogenic microorganisms in produce: a review
.
Adv. Agronomy
113
,
74
105
.
Paterson
R. R. M.
Kelly
J.
Gallagher
M.
1997
Natural occurrence of aflatoxins and Aspergillus flavus in water
.
Lett. Appl. Microbiol.
25
,
435
436
.
Pinto
L.
Poeta
P.
Radhouani
H.
Coelho
C.
Carvalho
C.
Rodrigues
J.
Torres
C.
Vitorino
R.
Domingues
P.
Igrejas
G.
2011
Proteomic study in an Escherichia coli strain from seagulls of the Berlengas Natural Reserve of Portugal
.
Genetics
1
(
1
),
36
41
.
Prusky
D.
1996
Pathogen quiescence in postharvest diseases
.
Annu. Rev. Phytopath.
34
,
413
434
.
Shioya
K.
Michaux
C.
Kuenne
C.
Hain
T.
Verneuil
N.
Budin-Verneuil
A.
Hartsch
T.
Hartke
A.
Giard
J.
2011
Genome-wide identification of small RNAs in the opportunistic pathogen Enterococcus faecalis V583
.
PLoS One
6
(
9
),
1
11
.
Shisana
O.
Rehle
T.
Simbayi
L. C.
Zuma
K.
Jooste
S.
Zungu
N.
Labadarios
D.
Onoya
D.
2014
South African National HIV Prevalence, Incidence and Behaviour Survey, 2012
.
HSRC Press
,
Cape Town
,
South Africa
.
Soltani
M.
Bayat
M.
Hashemi
S. J.
Zia
M.
Pestechian
N.
2013
Isolation of Cryptococcus neoformans and other opportunistic fungi from pigeon droppings
.
J. Res.Med. Sci.
18
(
1
),
56
60
.
Spinks
J.
Phillips
S.
Robinson
P.
Buynder
P.
2006
Bushfires and tank rainwater quality: a cause for concern?
J. Water Health
4
,
21
28
.
Stocker
S.
Snajder
R.
Rainer
J.
Trajanoski
S.
Gorkiewicz
G.
Trajanoski
Z.
Thallinger
G. G.
2011
SnoWMAn: high-throughput phylotyping, analysis and comparison of microbial communities
. In
Proceedings of the 110th ASM General Meeting
,
San Diego, CA
,
USA
,
23–27 May
,
23, 5
.
United Nations (UN)
2012
The Millennium Development Goals Report
.
ISBN 978-92-1-101258-3
.