Rural communities within low-income countries frequently rely on a range of drinking-water sources, and each water source varies in its potential for biological contamination. The extent and source of biological contamination in primary drinking sources within Kien Svay, Kandal, Cambodia, were determined by fecal indicator bacteria (FIB) measurements, 16S rDNA genetic markers for human and bovine fecal Bacteroides, presence of the bloom-forming Microcystis species, and the microcystin toxin mcyD gene marker. Thirteen wells, 11 rain barrels, 10 surface-water sites, and five sediment samples were examined during the dry and wet seasons. Surface water was commonly contaminated with FIB, with up to 1.02 × 105Enterococcus sp., 6.13 × 104E. coli, and 2.91 × 104 total coliforms per 100 mL of water. Human and bovine Bacteroides were detected in 100 and 90% of the surface water samples, respectively. Concentrations of FIB in rain-barrels varied by site, however 91% contained human Bacteroides. Microcystis cells were found in 90% of surface water sites, with many also containing microcystin gene mcyD, representing the first report of microcystin-producing cyanobacteria in surface waters of Cambodia. The study results show that many potential drinking-water sources in Cambodia contain harmful bacterial and algal contaminants, and care should be taken when selecting and monitoring water options.
Maintaining and monitoring proper biological (Knappett et al. 2011; Bain et al. 2014) and chemical (Smith et al. 2000; Polizzotto et al. 2008) standards of water quality is a challenge for many low- and middle-income countries. Many of the poorest rural communities in these countries rely on multiple sources of water. These water sources are frequently contaminated with microbial pathogens of fecal origin; pathogens can be bacterial, viral, or protozoan and can cause illnesses such as Legionnaire's Disease, polio, typhoid fever, cholera, gastroenteritis, and meningitis. In addition to pathogens of fecal origin, other water-based microbial threats to human health may include toxin-producing cyanobateria, such as Microcystis, which, during algal bloom formation, produce elevated concentrations of microcystin (Carmichael 2001) that can lead to severe liver damage or death if consumed (Falconer et al. 1983; Nishiwaki-Matsushima et al. 1992; Jochimsen et al. 1998). Microcystin is a cyclic peptide that acts as a phosphatase inhibitor and occurs as over 90 different variants with small sequence changes at L-amino acid positions 2 and 4 (Shimizu et al. 2014).
Given the range of potential biological contaminants of drinking-water quality, as well as the variety of drinking water sources that may be utilized within low-income countries, potential threats to human health from drinking water are often neglected based on typical monitoring strategies. Water sources may be broadly classified as improved or unimproved according to the mode in which water is accessed and the likelihood of pathogenic contamination (Joint Monitoring Programme for Water Supply and Sanitation; www.wssinfo.org/). If water quality is measured, abundant fecal indicator bacteria (FIB) are typically used as a measure of fecal contamination in water sources (WHO 2008). In low-income countries, FIB are typically enumerated by culture methods on artificial media, a process which takes 24–48 hours of incubation and often underestimates true abundances of viable but nonculturable bacteria (Colwell et al. 1985; Pommepuy et al. 1996; Oliver 2005). In contrast, molecular-based quantitative PCR (qPCR) and traditional PCR methods are much faster methods for enumerating FIB, giving results in just hours. These molecular methods allow quantification of organisms in low abundances in the environment, but also require instrumentation, personnel, and supplies that may be unavailable within low-income communities. Currently, strategies for effectively making informed water-use decisions in rural areas of low-income countries are lacking, and enhanced monitoring strategies that integrate diagnostic measures of water quality are needed (Bain et al. 2014).
This work compares and assesses multiple types of microbial contamination for the most commonly used water sources in rural parts of Cambodia, where the main drinking water sources include tubewells, surface water, and rainwater (National Institute of Statistics 2011); current public sanitation is poor; and no centralized water treatment exists. This was accomplished by: (a) examining the water quality of possible drinking-water sources in the rural Kien Svay District; (b) determining the abundances of FIB in surface-water sediments; (c) detecting the nonpoint source of the contamination in surface waters using PCR primers specific for human and bovine anaerobic Bacteroides spp.; and (d) qualitatively testing surface water for the presence of Microcystis as well as microcystin-producing cyanobacteria.
Study site and water sampling
Surface water chemistry and chlorophyll
Samples from each site were collected in triple-rinsed 1L HDPE bottles and passed through a 0.22 μM Nucleopore polycarbonate filter (Whatman) before acidification, if needed, and kept at 4 °C until tested. Cations were analyzed using a Varian 820 inductively coupled plasma-mass spectrometer (Bruker Daltonics, Billerica, MA). Samples were examined using a collision-reaction interface with hydrogen gas at 75 mL/min to correct ArCl (mass 75) interference on As(75). Anion analysis was conducted using ion chromatography on a Dionex DX-500 (ThermoScientific, Waltham, MA) fitted with an AS-22 anion analytical column while nitrate, ammonium and phosphate were measured using a Lachat 8000 flow injection analyzer (Hach Co., Loveland, CO). Total organic carbon (TOC) samples were measured using a high temperature catalytic oxidation procedure with a Shimadzu TOC-5050 total organic carbon analyzer (Shimadzu Sc. Inst., Columbia, MD). Temperature, pH, and dissolved oxygen were measured using a calibrated Hanna HI 9828 multiparameter meter (Hanna Instruments, Woonsocket, RI). Bulk surface water chemical characteristics are described in Supplementary Data Section S2 and Supplementary Table S1 (available in the online version of this paper). Additionally, total chlorophyll a (chl a) concentrations were established for surface water samples, and the methods are described in the Supplementary Data Section S3.
Ten surface-water, 13 well, 11 rain-barrel, and five sediment samples were examined in May and October 2012. Surface-water samples were tested in both seasons from ∼5–15 cm depth. Water was collected using the grab sample technique in sterile Whirl-Paks or VWR sampling bags for the surface-water and rain-barrel samples. Well water was purged of ∼10 L of water before sampling to clear stagnant well water in the screened PVC tube. Sediment samples were collected from surface water sites SS1–5 with sterile sampling bags from the sediment surface to ∼20 cm depth. All samples were kept at 4 °C for less than 8 h before membrane filtration. Four grams of wet sediment samples were diluted into 40 mL of sterile phosphate buffered saline solution and placed on a tube rotator at ∼20 rpm for 30 min in 50 mL conical tubes (VWR). After rotating, tubes were allowed to settle before processing.
A presumptive test for the presence of coliforms in water was performed using Difco Presence-Absence broth by adding 100 mL of sample water to 50 mL of sterile media into glass bottles and incubating at 37 °C for at least 24 hrs. A color change from purple to yellow with or without gas production (i.e. bubble formation) indicated samples were positive for coliforms. Enterococcus, E. coli, and total coliform densities were estimated using plate counts as described in Section 4 of the Supplementary Data.
Bacteroides and Microcystis spp.
All water types were collected for DNA analysis and filtered through 47 mm diameter 0.2 μm filters. Sediments were also collected in sterile bags and were used for extraction. DNA was extracted from filters or sediments using the Mobio Powersoil DNA Isolation Kit. To determine human and bovine fecal contamination, the 16S rDNA gene was used with BoBac and HuBac PCR primers for Bacteroides (Layton et al. 2006). These 25 μL PCR reactions consisted of 12.5 μL GoTaq 2 master mix (Promega), 1X BSA, 1 μM forward and reverse primers, sterile water, and up to 5 μL DNA. Positive controls contained linear plasmid DNA from PCR amplicons (kindly provided by Alice Layton) and negative controls contained no DNA with sterile water. Reaction protocols for all PCR reactions can be found in Table S2 of the Supplementary Data. The presence of Microsystis spp. was analyzed in surface water and sediments using primer sets MICR (Neilan et al. 1997) for all Microcystis and the mcyD gene (Kaebernick et al. 2000) for those spp. containing the microcystin toxin gene cassette. Positive controls were conducted using extracted DNA from the toxic spp. Microcystis LE-3 (kindly provided by Steven Wilhelm) from Lake Erie. PCR products were visualized with ethidium-bromide stained 1–2% agarose gels. Reaction conditions were as described above.
RESULTS AND DISCUSSION
FIB in drinking water sources
The main drinking-water sources in the sampling area are well, rain, and surface water. Surface freshwater, such as pond or river water, is characterized as an unimproved water resource due to the likely presence of pathogenic bacteria or harmful chemicals (Joint Monitoring Programme for Water Supply and Sanitation; www.wssinfo.org/). Despite this, approximately one-quarter of the rural population in Cambodia still use surface water for drinking during the dry season (National Institute of Statistics 2011). Use of well water is discouraged in some sampling sites, due to high levels of arsenic (As) (Polizzotto et al. 2008) or manganese (Buschmann et al. 2007), making rainwater and surface water the main drinking-water sources when chemical contamination in groundwater is problematic.
All samples were quickly tested for total coliforms using 100 mL of sample water and PA broth. All surface water samples from May and October had positive reactions in the PA test as exhibited by a color change from purple to yellow (Supplementary Table S3, available in the online version of this paper), indicating the presence of coliforms. Only two wells and one rain barrel were negative for coliforms, suggesting that these improved water sources are also contaminated with coliform bacteria, almost as often as surface water. Interestingly, all three sites negative for coliforms were positive for human Bacteroides based on PCR analysis, suggesting that those sites also were fecally contaminated. This observation may be due to the increased sensitivity for PCR reactions or the existence of viable but nonculturable coliforms (Bernhard & Field 2000). Only around 0.01–1% of bacteria in environmental samples are able to be cultured on artificial media (Oliver 2005), indicating all plate counts are likely underestimates of true values.
Rain barrels in Cambodia are often large ceramic basins filled with water collected from rooftops during the wet monsoon season. Using rain water for drinking, cleaning, or bathing is popular in rural parts of Cambodia, with approximately 7–40.8% of the population using rain water as a primary water source, depending on the season (National Institute of Statistics 2011). Some barrels are covered when full or not in use so that dust and other environmental contaminants are not introduced to the water, but this practice is highly unpredictable. In this study, rain-barrel water was the second most contaminated water source (Figure SB), with Enterococcus spp. ranging from 10–6.6 × 103, E. coli ranging from 10–1.85 × 103, and total coliforms ranging from 90–1.24 × 104 CFU/100 mL. It is possible that some dust, particulates, fecal droppings, and leaves landed in open vessels (WHO 2008). More often, however, the water likely gets contaminated during storage and at point-of-use from poor hygiene, as identified with PCR analysis and described below. No rain barrels tested in Cambodia had valves or taps to release water, which likely enabled microbial growth, as potentially contaminated scoops and hands were frequently in contact with the stored water.
Quantities of bacteria found in surface waters varied temporally and spatially, but were higher in surface water than in any other water source tested. Samples were collected from nine sites in May and 10 sites in October. Total counts of Enterococcus, E. coli, and general coliforms fluctuated across sites in May, whereas total abundances were relatively stable in October (Figure 2(c) and 2(d)). Enterococcus and E. coli are better indicators of fecal contamination than total coliforms (WHO 2008), and up to 1.02 × 105Enterococcus and 6.13 × 104E. coli were found during the dry months. Other work has shown that in the Mekong Delta in Vietnam, E. coli was present more often during the wet season compared to the dry season, a finding that was attributed to bacteria in water and sediment runoff (Isobe et al. 2004). During our sampling, October was not the month with the most precipitation in 2012. The largest weather events occurred in September, and rains were relatively lighter in October than in previous years, with fewer flooding events and lower water levels (MRC Regional Flood Management and Mitigation Centre 2012). The moderately dry wet season in 2012 could be a potential reason for no marked increase in FIB.
Sediments from surface water sites also harbored FIB. Wet sediment samples were collected in October of 2012, and samples from all five sites were contaminated with E. coli and general coliforms, whereas only three sites were contaminated with Enterococcus (Supplementary Figure S1).
Bacteroides and Microcystis
Culturing organisms to identify fecal contamination is a traditional method used to determine FIB in developing countries that can take days for results and likely underestimates true contaminant burdens. A faster and more accurate, albeit more expensive, method that can determine nonpoint sources of contamination is through PCR analysis. Therefore, surface (May and October), well, and rain-barrel water was tested for the presence of anaerobic Bacteroides spp. using host-specific gene primers for human and bovine fecal contamination. Surface water and sediments were also tested for the bloom-forming cyanobacterium Microcystis. Since some Microcystis spp. have the ability to produce the hepatotoxin microcystin, the mcyD gene was tested for the presence of toxic cyanobacteria.
Human Bacteroides was the most prevalent fecal contaminant in all water types and sediments (Tables 1 and S3). All surface sites (in both May and October) and sediments contained human Bacteroides spp. (Table 1). It would be expected that, as an improved water source, rain-barrels would have lower amounts of FIB, although 72.7% of the barrels also harbored these bacteria. Well water was the least contaminated, with only one site (or 7.7%) polluted with human Bacteroides spp. (Table S3). Soil has the ability to affect bacteria leaching into groundwater or well water by its physical properties (e.g. pH, clay content, particle size, etc.) which may be the reason for the decrease in the microbial burden in these well waters.
|Sample type .||Human Bacteroides .||Bovine Bacteroides .||Total Microcystis .||mcyD .|
|Sample type .||Human Bacteroides .||Bovine Bacteroides .||Total Microcystis .||mcyD .|
Surface waters and sediments were tested for bovine Bacteroides. Eighty percent of surface-water samples from May and 90% from October were contaminated with bovine feces (Table 1). These sites are situated in rural settings in Cambodia, where there is no enforced federal or local regulation of farm animal waste. Cows, pigs, chickens, and dogs often roam free in these lands, so it is not unexpected that ponds and rivers would contain animal feces. Furthermore, no sediment collected from surface water sites contained bovine Bacteroides. Since Bacteroides spp. do not survive as long as other FIB in the environment, it has been suggested that Bacteroides be used for more recent contamination (Ballesté & Blanch 2010). Given this fact, it is not surprising that the DNA marker for bovines was detected in surface water but not in sediments.
The population of Cambodia (and all of Southeast Asia) is increasing (Trinh Thi 2010), and anthropogenic and agricultural inputs of nutrients to freshwater systems are also growing, likely leading to continued eutrophication and increasing algal populations. Microcystis spp. and other toxic algae have already been detected through microscopic analysis in Lake Tonle Sap (Mizuno & Mori 1970; Ohtaka et al. 2010) and Phnom Penh raw water (Ministry of Rural Development Department of Rural Water Supply 2002). Here, other surface water sites, including rivers, wetlands, and ponds, were analyzed for total Microcystis and toxic cyanobacterial populations to assess the potential for future toxic algal bloom formation. It was found that 90% of surface waters (May and October) and 60% of sediments contained Microcystis. Of these sites, 30% of May and 50% of October samples contained toxic Microcystis spp., based on amplification of the mcyD gene (Table 1). Sediments also retained toxic Microcystis in 40% of the samples.
High chl a concentrations of up to 39 μg/L were found in surface waters (Supplementary Figure S2), which suggests there is a healthy phytoplankton community. Size-fractionated chl a analysis was not performed, so it is not known whether most of the phytoplankton quantified were in the pico-, nano-or micro-size fraction. Low oxygen, high temperatures, and adequate nutrients would also suggest that these samples are eutrophic, as described in the Supplementary Data (available in the online version of this paper). Elevated temperatures have been shown to increase the occurrence of blooms (Paerl & Otten 2013). Cambodia is in the tropical region, expanding the length of the year for bloom formation and the potential for health hazards. This is the first report of Microcystis spp. able to produce microcystin in Cambodian waters.
Microcystis are able to survive in sediments due to their buoyancy, and this may instigate bloom formation when conditions become favorable (Preston et al. 1980; Rinta-Kanto et al. 2009). Microcystis cells were present in 60% of sediments in Cambodian freshwater ecosystems, with some of these being toxic. PCR does not establish viable cells vs. dead cells, so it is not known whether the Microcystis present were living. Interestingly, SS4 surface waters and sediments did not contain Microcystis, based on the lack of a PCR amplicon using the MICR Microcystis-specific 16S primers, but they did contain the microcystin-producing gene mcyD (Table 1). Other cyanobacteria, such as Anabaena, Planktothrix, and Nostoc spp., are able to produce microcystin (Codd et al. 2005). These other genera are often found in eutrophic freshwater systems, so it is likely other cyanobacteria caused the positive reactions. To fully assess whether or not these potentially toxic microcystin-producing cyanobacteria are a threat to the ecosystem, humans, or animals, the concentration of microcystin present and abundances of Microcystis cells in these waters should be addressed in the future. The WHO guideline for microcystin is 1 μg/L (WHO 2011), and it is important to know whether cell-bound and/or extracellular microcystin concentrations meet or exceed this limit in Cambodian drinking-water sources.
Implications for drinking-water use
When assessing the suitability of water for drinking, the most common drinking water types should be analyzed for the potential health risks from FIB contamination. As a developing country, much of rural Cambodia does not have the convenience of clean, piped water, and people rely on ground, rain, or surface water for consumption. All of these water sources pose at least some risk of diarrheal disease from inadequate public sanitation and human pathogenic bacteria. This study aimed to address the extent of contamination in all of these sources and found that, in Cambodia, all sources were not suitable for immediate drinking. With approximately 40% of rural populations in Cambodia lacking access to improved drinking-water sources, which sources pose the least risk of illness needs to be determined.
Within our study site, even improved drinking water sources – well and rain-barrel water – were at risk of biological contamination, implying that development efforts should focus on resource protection, point-of-use treatment, and hygiene education in order to minimize health hazards associated with water sources that are generally deemed safe. Improved hygiene is especially necessary for rain water collected in barrels, as fecal contamination was exhibited in almost all barrels tested. Periodic cleaning of barrels (e.g. when they are not in use during the dry season), as well as the use of an appropriate lid, would provide better conditions for maintaining the safety of drinking water.
This work also highlights the broader challenges in choosing, managing, and monitoring drinking-water sources in rural areas of low-income countries. As an unimproved water source, surface water should be avoided as much as possible. High amounts of FIB suggest that these waters harbor microbial contaminants, but such sources are often used when designated improved water sources (e.g. well water) contain chemical contaminants (e.g. arsenic) that are difficult to remove. When the safety of all potential water sources is questionable due to the likely presence of biological and/or chemical contaminants, decisions about water options must consider the costs and reliability of household-level water treatment (e.g. through boiling, filtering, or chemical treatment) that is essential even for many ‘improved’ drinking-water sources. Finally, to truly ensure the safety of water sources, more comprehensive water-quality monitoring needs to be established, in order to recognize the ranges of biological and chemical contaminants that are commonly found in rural water sources and to accommodate the reliabilities of different monitoring techniques (e.g. culture-based vs. molecular methods) that have different sensitivities for contaminants.
The water quality of several drinking-water sources commonly used in Cambodia was found to have exceeded the Cambodian national standard of 0 fecal coliforms or E. coli/100 mL, which are based on the WHO guidelines used for risk assessment. No source was consistently safe to drink as determined by culture-based or molecular methods for FIB. Human waste was the most prevalent source of contamination, but animal waste was also a problem in surface water.
With the increased urbanization of Cambodia, nutrient inputs are likely to surge. Surface waters tested were already eutrophic, and with more nutrient loadings, algal blooms may also become problematic in the future. We found that most surface waters already contained the bloom-forming cyanobacterium, Microcystis. This is the first report of potential cyanotoxin-producing cyanobacteria in Cambodia. In order to reduce the likelihood of harmful algal blooms, anthropogenic sources of nutrients to water sources need to be limited.
This work was supported by North Carolina State University. The authors would like to thank Sarah Seehaver, Elizabeth Gillispie, Steven Wilhelm, Alice Layton, Kim Hutchison, Guillermo Ramirez, Andrew Shantz, Nuon Phen, and members of Resource Development International-Cambodia for their assistance with analyzing samples and DNA preparation.