South African rivers generally receive waste from inadequate wastewater infrastructure, mines, and farming activities, among others. The uMsunduzi River in KwaZulu-Natal, South Africa, is among these recipients with recorded poor to very poor water quality. To identify parts of the uMsunduzi River that are polluted by Cryptosporidium and Escherichia coli (E. coli), this study mapped out pollutants emanating from point and non-point sources using the Soil and Water Assessment Tool (SWAT). Streamflow calibration in the upper and lower reaches of the catchment showed good performance with R2 of 0.64 and 0.58, respectively. SWAT water quality output data were combined with a Quantitative Microbial Risk Assessment (QMRA) to understand the microbial health implications for people using river water for drinking, recreational swimming, and non-competitive canoeing. QMRA results for Cryptosporidium and pathogenic E. coli showed that the probability of infection for most users exceeds the acceptable level for drinking and recreation as outlined in the South African water quality guidelines, and by the World Health Organization (WHO). The results of this study can be used as a baseline to assess the economic and health implications of different management plans, resulting in better-informed, cost-effective, and impactful decision-making.

  • SWAT model identified areas of the uMsunduzi River that were polluted by Cryptosporidium and E. coli.

  • SWAT water quality output data were used in QMRA.

  • QMRA investigated the impacts of river water on canoeists, recreational swimmers, and those who drink the water.

  • uMsunduzi River water is not suitable for drinking and recreation according to QMRA results.

  • These results can inform policies and decision-making within the catchment.

Graphical Abstract

Graphical Abstract
Graphical Abstract

Surface water pollution is strongly associated with land and water uses within a catchment. These include impact from residences with insufficient sanitation infrastructure in the vicinity of streams, poorly operated and maintained sanitation infrastructure, dense settlements in areas prone to surface runoff, and intensive livestock farming (Luyt et al. 2012). Low-income countries continue to have a high incidence of diarrhoea among young children under 5 years of age, which is one of the leading causes of death, ill-health, and disability (Naghavi et al. 2015). In such communities with poor service and infrastructure, the spread of diarrhoeal diseases is exacerbated by inadequate and unsafe water supplies and inadequate sanitation and hygiene (Araya et al. 2016). Shirley et al. (2012) found that most attributable cases of moderate-to-severe diarrhoea in the youngest children were due to four pathogens: rotavirus, Cryptosporidium, enterotoxigenic E. coli producing heat-stable toxin (ST-ETEC; with or without co-expression of heat-labile enterotoxin), and Shigella.

Cryptosporidium has been identified in South Africa as a causative agent of diarrhoeal diseases in immunocompromised, HIV-positive children and adults (Ojuromi & Ashafa 2018). In the study performed in KwaZulu-Natal by Jarmey-Swan et al. (2001), Cryptosporidium was found to be most prevalent (39.3%) (1,100/2,800) in the <1-year age group. The quality of South Africa's water resources is linked to the inadequate wastewater infrastructure and the subsequent contribution of untreated or poorly treated wastewater discharges to the already deteriorating water sources (Mema 2010; Osuolale & Okoh 2017; Ojuromi & Ashafa 2018; Sekwadi et al. 2019). The lack of infrastructure in South Africa forces rural and peri-urban households to supplement their drinking water intake with water of poor microbial quality (Luyt et al. 2012).

In response to water-related challenges, hydrological modelling has been developed to analyse, understand, and explore solutions for sustainable water management, to the decision makers and operational water managers (Baffaut & Benson 2009; Mander et al. 2017; Abdullah et al. 2019). Using models to understand and manage water quality can make it easier to promote better management. By generating more data over time and space, they can supplement limited observational data and allow for scenarios to explore the impact of different management measures on water quality in a watershed. Catchment-scale water quality models have become a key tool for understanding, evaluating, and predicting the adverse effects of contamination on river water quality. Among others, models that provide refined simulation for water quality include Agricultural Nonpoint Source pollution model (AGNPS/AnnAGNPS), Soil and Water Assessment Tool (SWAT), Stormwater Management Model (SWMM), and Hydrologic Simulation Program Fortran (HSPF). Of these, HSPF and SWAT are highly recommended to users (Yuan et al. 2020).

In SWAT, there is a microbial sub-model that deals with the fate and transport of both persistent and less persistent microorganisms (Abdullah et al. 2019). SWAT can, therefore, simulate microbial transport in catchments as well as identify high-risk areas and estimate peak loads of microorganisms such as E. coli and Cryptosporidium (Bergion et al. 2017; Zhang et al. 2019). Output data from a SWAT model was successfully used as input data to conduct quantitative microbial risk assessment (QMRA) by Bergion et al. (2020). QMRA assesses the probability of infection due to exposure to microorganisms, based on the dose a person ingests and dose-response models that provide the human response to specific pathogens (Haas et al. 2014). The risk of being infected by microbial pathogens is dependent on the level of pollution of the water and the amount of polluted water consumed (Müller et al. 2001).

The aim of this study was to utilise SWAT as a water quality modelling tool to map out point and non-point source pollution in the uMsunduzi catchment in KwaZulu-Natal, South Africa, to identify high-risk areas. Concentrations of Cryptosporidium and E. coli were studied. The concentration outputs from SWAT were used to perform a QMRA to better understand the health implications on communities that use polluted waters for drinking and non-competitive recreation purposes.

The 875 km2 uMsunduzi catchment is in the KwaZulu-Natal province, South Africa. The uMsunduzi River, a 115 km watercourse in length, passes through Pietermaritzburg (29°37′S 30°23′E) and is a major tributary of the uMngeni River. It flows through rural and urban dwellings as seen in Figure 1.

Figure 1

River flow monitoring gauge positions, water quality monitoring stations, broken sewers, weather station within the uMsunduzi catchment, as well as subbasins as delineated by SWAT. Please follow the link to view in full layout: https://www.google.com/maps/d/viewer?mid=1NQXdFcqDjShMQxRgn4p8ILzpj42NpgdG&ll=-29.63672034343441%2C30.475072992298692&z=11.

Figure 1

River flow monitoring gauge positions, water quality monitoring stations, broken sewers, weather station within the uMsunduzi catchment, as well as subbasins as delineated by SWAT. Please follow the link to view in full layout: https://www.google.com/maps/d/viewer?mid=1NQXdFcqDjShMQxRgn4p8ILzpj42NpgdG&ll=-29.63672034343441%2C30.475072992298692&z=11.

Close modal

Due to the variation in settlement types, the catchment is divided into three units: the upper catchment rural area, urban area in the middle, and lower catchment rural area. The rural areas are characterised by subsistence farming (stock and crops) and rural developments. The urban area is dominated by the urban stretch of Pietermaritzburg that is a mix of formal city, residential and industrial suburbs, and informal settlements. The uMsunduzi River enters a canalised reach at Camps Drift, then passes through the city in an incised narrow channel and is joined by small tributaries before picking up treated effluent from Darvill Wastewater Treatment Plant (WWTP). The lower catchment rural area is dominated by the rural Valley of a Thousand Hills, where the uMsunduzi River confluences with the uMngeni River.

The city of Pietermaritzburg has a dry-winter humid subtropical climate (Köppen climate classification: Cwa), with distinct dry and wet seasons. The average temperatures are ∼21 °C in summer (December–February) and ∼12 °C in winter (June–August). The monthly average precipitation during dry months (June–August) is 23 and 140 mm during the rainy season (December–February). The dominant soil types in the study area, according to the Food and Agricultural Organisation (FAO 2015) are Hutton, Griffin, Mispah, and Glenrosa. The main land use types are grasslands. It is important to note that these are slowly changing into informal dwellings due to urbanisation.

Water quality model

QSWAT3 v1.1 for QGIS3 was used to setup the hydrological model for the period 1999–2018, for this study. The years 1999–2003 were reserved for the warm-up period prior to the defined calibration and validation stages to allow the stable performance of the model. Calibration and validation of the model using river flow data were performed for the periods 2004–2011 and 2012–2018, respectively. The input data included topographical, geographical, and geological information as well as meteorological data as indicated in Table 1. The uMsunduzi catchment was divided into 35 subbasins. The water quality output of this model was used in the QMRA as discussed in the subsequent sections.

Table 1

Data for the hydrological and water quality model of the uMsunduzi River

DataFile typeResolutionSource
Digital Elevation Model Raster 20×20 m National Geo-spatial Information (NGI) 
Landuse/landcover 2018 Raster 1,700×1,700 m South African National Biodiversity Institute (SANBI) 
Soil type Raster 250×250 m Food and Agriculture Organization (FAO) 
Meteorological data Text One station, Dailya South African Weather Services (SAWS) 
River flow data Text Two stations, Dailya Department of Water Affairs (DWA) 
Water quality monitoring data Text Four Points, Weeklya Umgeni Water 
DataFile typeResolutionSource
Digital Elevation Model Raster 20×20 m National Geo-spatial Information (NGI) 
Landuse/landcover 2018 Raster 1,700×1,700 m South African National Biodiversity Institute (SANBI) 
Soil type Raster 250×250 m Food and Agriculture Organization (FAO) 
Meteorological data Text One station, Dailya South African Weather Services (SAWS) 
River flow data Text Two stations, Dailya Department of Water Affairs (DWA) 
Water quality monitoring data Text Four Points, Weeklya Umgeni Water 

aRefer to Figure 1 for location within the catchment.

The hydrological parameter optimisation was performed manually. The Nash–Sutcliffe efficiency coefficient (NSE), the coefficient of determination (R2), percent bias (PBIAS), and the ratio of the root mean squared error to the standard deviation of measured data (RSR) were used for the evaluation of model performance. These are represented by Equations (1)–(3), respectively.
formula
(1)
formula
(2)
formula
(3)
where Q is discharge, the is the mean discharge, and the indices m and s stand for measured and simulated, respectively and i/n are the ith and nth measured or simulated data.

Sadeghi & Arnold (2002) developed a SWAT bacterial module for predicting microorganism loads and concentrations in rivers. The SWAT microbial sub-model considers the fate and net transport of microorganisms derived from defined pollution sources. The two microorganisms considered in this study were E. coli and Cryptosporidium, chosen to represent bacteria and protozoan parasites, respectively. Bacteria may be less persistent in the environment due to inactivation by environmental factors compared to the protozoan parasites, whose (oo)cysts are more resistant to environmental conditions (Sadeghi & Arnold 2002). The input parameter values for both microorganisms are shown in Table 2.

Table 2

SWAT parameter values for E. coli and Cryptosporidium

MicroorganismSWAT abbreviationUnitValueReference
Both BACTFDDB (Partition coefficient) Fraction 0≤1 0.9 Coffey et al. (2013)  
Both BACTKDQ (Bacteria soil partitioning coefficient) Constant 175 Arnold et al. (2012)  
Both FRT_SURFACE (Fraction of fertiliser applied to 10 mm of soils) Fraction 0≤1 0.5 Tang et al. (2011)  
E. coli WDLPQ (Die-off factor for less persistent bacteria in soil solution at 20 °C) 1/day 0.092 Westrell (2004)  
E. coli WDLPRCH (Die-off factor for less persistent bacteria in streams at 20 °C) 1/day 0.18 Westrell (2004)  
E. coli WDLPS (Die-off factor for less persistent bacteria adsorbed to soil particles at 20 °C) 1/day 0.023 Bougeard et al. (2011)  
E. coli WDLPF (Die-off factor for less persistent bacteria on foliage at 20 °C) 1/day 0.016 Bougeard et al. (2011)  
E. coli WOF_LP (Fraction of less persistent organisms washed off in rainfall events) Fraction 0≤1 0.5 Bougeard et al. (2011)  
Cryptosporidium WDPQ (Die-off factor for persistent bacteria in soil solution at 20 °C) 1/day 0.005 Westrell (2004)  
Cryptosporidium WDPRCH (Die-off factor for persistent bacteria in streams at 20 °C) 1/day 0.032 Westrell (2004)  
Cryptosporidium WDPS (Die-off factor for persistent bacteria adsorbed to soil particles at 20 °C) 1/day 0.003 Coffey et al. (2013)  
Cryptosporidium WDPF (Die-off factor for persistent bacteria on foliage at 20 °C) 1/day 0.03 Tang et al. (2011)  
Cryptosporidium WOF_P (Fraction of persistent organisms washed off in rainfall events) Fraction 0≤1 0.8 Tang et al. (2011)  
MicroorganismSWAT abbreviationUnitValueReference
Both BACTFDDB (Partition coefficient) Fraction 0≤1 0.9 Coffey et al. (2013)  
Both BACTKDQ (Bacteria soil partitioning coefficient) Constant 175 Arnold et al. (2012)  
Both FRT_SURFACE (Fraction of fertiliser applied to 10 mm of soils) Fraction 0≤1 0.5 Tang et al. (2011)  
E. coli WDLPQ (Die-off factor for less persistent bacteria in soil solution at 20 °C) 1/day 0.092 Westrell (2004)  
E. coli WDLPRCH (Die-off factor for less persistent bacteria in streams at 20 °C) 1/day 0.18 Westrell (2004)  
E. coli WDLPS (Die-off factor for less persistent bacteria adsorbed to soil particles at 20 °C) 1/day 0.023 Bougeard et al. (2011)  
E. coli WDLPF (Die-off factor for less persistent bacteria on foliage at 20 °C) 1/day 0.016 Bougeard et al. (2011)  
E. coli WOF_LP (Fraction of less persistent organisms washed off in rainfall events) Fraction 0≤1 0.5 Bougeard et al. (2011)  
Cryptosporidium WDPQ (Die-off factor for persistent bacteria in soil solution at 20 °C) 1/day 0.005 Westrell (2004)  
Cryptosporidium WDPRCH (Die-off factor for persistent bacteria in streams at 20 °C) 1/day 0.032 Westrell (2004)  
Cryptosporidium WDPS (Die-off factor for persistent bacteria adsorbed to soil particles at 20 °C) 1/day 0.003 Coffey et al. (2013)  
Cryptosporidium WDPF (Die-off factor for persistent bacteria on foliage at 20 °C) 1/day 0.03 Tang et al. (2011)  
Cryptosporidium WOF_P (Fraction of persistent organisms washed off in rainfall events) Fraction 0≤1 0.8 Tang et al. (2011)  

The uMsunduzi River is impacted by point and non-point source pollution emanating from rural settlements, informal settlements, townships, formal urban and industrial areas, and the Darvill WWTP (Dabrowski et al. 2013). The main faecal sources in the uMsunduzi catchment can be summarised as: Darvill WWTP, broken sewers in the urban area, dry toilets, and faecal droppings from grazing livestock.

To compare the model output with the measured concentrations, four subbasin output points that closely coincide with water quality monitoring points by the Msunduzi Local Municipality and Umgeni Water along the river were considered in the model. These were: uMsunduzi downstream Henley Dam – rural area (subbasin 10); urban areas (subbasin 8); major stream, city, informal settlements, industrial areas, WWTP, and waste dump site (subbasin 4); and rural areas and confluence with the uMngeni River (subbasin 1) as shown in Figure 1.

The Darvill WWTP serves 51.6% of the 163,993 uMsunduzi Local Municipality households (STATSSA 2011). The Darvill WWTP is currently being upgraded from 65,000 to 100,000 m3/day to meet current and future demands (Ramnath et al. 2019), which means that it often operated above its capacity and the system was overflowing, releasing untreated wastewater directly to the uMsunduzi River. The wastewater from the Darvill WWTP was put into the SWAT model as a point source with an average discharge of 75,000 m3/day. According to Ramnath et al. (2019), the inflows to the WWTP were, for a number of months in 2018, below the 12-month moving average (∼75,000 m3/day), with the average inflow in October 2018 only ∼59,000 m3/day. This is an unseasonal low, and Umgeni Water is concerned that not all the wastewater is reaching the Darvill WWTP. This implies that ∼16,000 m3/day could be attributed to losses due to broken or leaking sewers. This volume of wastewater was distributed over the subbasins within the urban section of the catchment in the SWAT model, where, as shown in Figure 1, there are broken sewers. This data was supplied by Mr Vilakazi of the Duzi-Umgeni Conservation Trust (DUCT). These leaking sewers were included as a point source contribution in each of these subbasins. For this study, the concentration of Cryptosporidium used as input in SWAT was 5 (0–17) oocysts/10 L based on 13 samples collected from the Darvill WWTP effluent during 2016–2017 (de Jong 2020, pers. comm.). E. coli concentration in wastewater was estimated to be 50,000 counts/100 ml, a median of 3,615 samples from monitoring data supplied by Umgeni Water taken between January 2004 and December 2018.

The remaining 48.4% of the households within the municipality use dry toilets or are without any toilet at home, a small percentage of which may have septic tanks (STATSSA 2011). Data showing the distribution of these dry toilets were not available; however, these toilets are used mainly in informal and rural settlements. Defaecating in open spaces could also be a human source of microorganisms in this catchment. It was, however, not considered in this study due to difficulties in quantifying its contribution in the runoff.

The SWAT grazing function was used to represent livestock manure production during grazing. Only cattle and sheep/goats were considered for this study as the numbers of other animals were low, and relevant prevalence and infectivity data in Africa are scarce. Due to the warm climate and general subsistence farming practices in Pietermaritzburg, these animals are grazing all year round. No data could be found that account for the distribution of the livestock herds over the catchment, thus, the estimated produced faecal matter was distributed evenly over the modelled grassland landuse area of 15,943 ha. To account for different faecal production depending on age, the fraction of animals that are young was determined using a guide by Vermeulen et al. (2017). The average Cryptosporidium concentrations in livestock faeces were calculated from the prevalence of infection and the concentration in faeces of infected animals. The assumed input data for E. coli and Cryptosporidium are presented in Table 3. The E. coli concentrations shown in Table 3 do not differentiate between pathogenic and non-pathogenic E. coli.

Table 3

Cryptosporidium and E. coli concentration estimates for SWAT input

HostLivestock in catchment (#)aAge differentiation (#)bAnimal density in grazing area (#/ha)Manure production (kg/d/1,000 kg)bAnimal mass (kg)bManure production (kg/ha/day)bCryptosporidium prevalencebCryptosporidium concentration infected livestock (oocysts/g)bCryptosporidium concentration total livestock (oocysts/g)cE. coli concentration (#/g)d
Cattle 35,345 32,517 2.04 86 250 32.07 0.17 1.0×10+02 1.7×10+01 2.0×10+05 
Calves 2,828 0.18 125* 1.39 0.29 7.9×10+04 2.3×10+04 4.2×10+05 
Goats 29,560 25,126 1.58 41 30 1.42 0.16 2.0×10+02 3.2×10+01 6.6×10+04 
Kids 4,434 0.28 15* 0.13 0.11 2.5×10+04 2.7×10+03 6.6×10+04 
Sheep 9,920 8,432 0.53 40 28 0.43 0.25 1.6×10+02 4.0×10+01 6.6×10+04 
Lambs 1,488 0.09 14* 0.04 0.13 2.0×10+05 2.6×10+04 6.6×10+04 
HostLivestock in catchment (#)aAge differentiation (#)bAnimal density in grazing area (#/ha)Manure production (kg/d/1,000 kg)bAnimal mass (kg)bManure production (kg/ha/day)bCryptosporidium prevalencebCryptosporidium concentration infected livestock (oocysts/g)bCryptosporidium concentration total livestock (oocysts/g)cE. coli concentration (#/g)d
Cattle 35,345 32,517 2.04 86 250 32.07 0.17 1.0×10+02 1.7×10+01 2.0×10+05 
Calves 2,828 0.18 125* 1.39 0.29 7.9×10+04 2.3×10+04 4.2×10+05 
Goats 29,560 25,126 1.58 41 30 1.42 0.16 2.0×10+02 3.2×10+01 6.6×10+04 
Kids 4,434 0.28 15* 0.13 0.11 2.5×10+04 2.7×10+03 6.6×10+04 
Sheep 9,920 8,432 0.53 40 28 0.43 0.25 1.6×10+02 4.0×10+01 6.6×10+04 
Lambs 1,488 0.09 14* 0.04 0.13 2.0×10+05 2.6×10+04 6.6×10+04 

aData source from STATSSA (2011).

bCalculations and values after Vermeulen et al. (2017).

cCalculated using concentration in infected livestock and prevalence.

*0.5 of the adult animal was used to calculate body mass of the respective young as no data could be found.

Quantitative microbial risk assessment (QMRA)

The QMRA tool version 2018-06-07 was used for microbial risk assessment to consumers accessed via https://www.analyticacloud.com/view?invite=24&code=56994&subName=Chalmers. This tool was developed by Abrahamsson Ansker & Heinicke (2009) and is managed by the Centre for Drinking Water Research ‘DRICKS’ on behalf of the Swedish drinking water industry. For the purpose of this study, water treatment steps were omitted in the QMRA tool, to represent untreated water ingested during recreational activities and direct domestic use.

Source water characterisation and exposure

The outputs from the SWAT model in terms of concentration of the selected microorganisms along the river were used as input values for the QMRA. Considering that not all E. coli are pathogenic, 8% of E. coli was assumed as pathogenic following various studies such as Haas et al. (2000) and Mbanga et al. (2020). The exposure routes investigated were direct ingestion of the uMsunduzi River water during recreational swimming, canoeing training, and drinking. These exposure scenarios are detailed below.

Recreational swimming

This exposure scenario was considered for subbasins 1, 4, 8, and 10. This is because it was observed during the study that the population along this stretch of the catchment habitually swim in the river during warm periods. The exposed population was categorised into children and adults. The volume ingested during swimming was estimated based on the ranges of 37–47 mL for children and 16–24 mL for adults per 45 minute event as reported in Dufour et al. (2006) as shown in Table 4.

Table 4

Water volumes ingested during swimming, canoeing, and drinking

CategoryMinMax
Ingested during swimming (L/d)a 
 Children 0.0493 0.0627 
 Adults 0.0213 0.0320 
Ingested during canoeing (L/d)a 
 Juvenal 0.0039 0.0077 
 Junior 0.0077 0.0116 
 Senior 0.0116 0.0154 
Ingested during drinking (L/d)a 
 Juvenal 0.0013 0.63 
 Junior 0.63 0.773 
 Senior 0.773 0.952 
CategoryMinMax
Ingested during swimming (L/d)a 
 Children 0.0493 0.0627 
 Adults 0.0213 0.0320 
Ingested during canoeing (L/d)a 
 Juvenal 0.0039 0.0077 
 Junior 0.0077 0.0116 
 Senior 0.0116 0.0154 
Ingested during drinking (L/d)a 
 Juvenal 0.0013 0.63 
 Junior 0.63 0.773 
 Senior 0.773 0.952 

aIn the QMRA dose-response modelling, the volume was assigned a uniform distribution with minima (Min) and maxima (Max).

Canoeing

This exposure scenario was considered for subbasin 8 as training takes place in this stretch of the river. Training information was obtained from Mr Z. L. Mthalane (pers. comm., 2019), a coach at two canoe clubs and a seasoned Dusi Canoe Marathon and Dusi Non-stop participant. The exposed population was categorised into three: juvenal (10–14 years old), junior (15–18 years old), and senior (>18 years old) based on the competition categories and the required hours for training. These weekly training hours are 3–5 (juvenal), 6–10 (junior), and 11–14 (senior). Based on these schedules, 0.5–1 h (juvenal), 1–1.5 h (junior), and 1.5–2 h (senior) per day were assumed. These hours, together with ingestion volume of 5.8 mL per 45 minute event, were used to estimate the ingested volumes during canoeing events as reported in Dorevitch et al. (2011). The subsequently calculated daily minima and maxima are presented in Table 4.

Drinking Water

This exposure scenario was considered for subbasins 10 and 1, which are in the upper and lower rural parts of the catchment, respectively. The exposed population was categorised into three: juvenal, junior, and senior. Table 4 shows daily ingestion volumes for South Africans adapted from Steyn et al. (2001).

Dose-response modelling

The dose-response functions provide the important link between exposure dose and the likelihood of occurrence of a negative consequence like infection, illness, severe illness, or death (Smeets et al. 2010; Haas et al. 2014). The ingested dose (μ) was estimated using Equation (4).
formula
(4)
where μ is the ingested dose of pathogens, v is the ingested volume, and c is the concentration of the targeted pathogens. Ingested volume was based on the values in Table 4, using a uniform distribution with minima and maxima, according to the values reported in the table. Concentration was defined as the fitted lognormal distribution based on the simulated daily pathogen concentrations output from the SWAT modelling for each subbasin. For each subbasin, fitting was conducted using the @RISK 8 software, and the lognormal distribution was one of the top three ranking distributions according to the Akaike Information Criterion (AIC).
In the QMRA tool, the Exponential dose-response model was used to calculate the probability of infection by pathogenic E. coli (O157:H7) and Cryptosporidium (Equation (5)).
formula
(5)
where P(inf) is the probability of infection, μ is the pathogen dose, and r is a Beta function. In the QMRA tool, applying Monte Carlo simulations and defining a Beta function for r in the Exponential function approximates an exact Beta-Poisson model. The Beta function (α, β) was used for Cryptosporidium (0.115, 0.176) (Teunis et al. 2002) and pathogenic E. coli (0.37, 37.65) (Teunis et al. 2004).

Water quality model

The SWAT model's predictions for streamflow were in good agreement with monitored data as shown in Table 5. The good agreement is corroborated by the visual comparison between simulated and observed flow for calibration and validation periods, using monthly data (Figure 2). All evaluation metrics shown in Table 5, i.e., NSE>0.5, RSR≤0.7, and PBIAS≤±0.25, except for the PBIAS for validation for subbasin 10, are referred to as satisfactory, with the PBIAS only slightly out of range (Moriasi et al. 2007). Parameters optimised during the calibration process are shown in Table 6.

Table 5

SWAT calibration and validation results

Calibration
Validation
IndexSubbasin 1Subbasin 10Subbasin 1Subbasin 10
R2 0.58 0.69 0.64 0.73 
NSE 0.57 0.61 0.58 0.59 
RSR 0.31 0.29 0.25 0.28 
PBIAS −0.05 0.24 0.14 0.27 
Calibration
Validation
IndexSubbasin 1Subbasin 10Subbasin 1Subbasin 10
R2 0.58 0.69 0.64 0.73 
NSE 0.57 0.61 0.58 0.59 
RSR 0.31 0.29 0.25 0.28 
PBIAS −0.05 0.24 0.14 0.27 
Table 6

Parameters optimised during SWAT calibration

VariableDescriptionDefault Value(s)Calibrated ValueInput file
Flow CN2 Initial SCS runoff curve number for moisture condition 45–85 49 .mgt 
SOL_K Saturated hydraulic conductivity 2.6–52 mm/h 7.54 mm/h .sol 
SOL_BD Soil moist bulk density 0.9–2.5 g/cm3 1.36 g/cm3 .sol 
ESCO Soil evaporation compensation factor 0.95 0.9018 .sol 
SOL_AWC Available water capacity of the soil 0.12–52 mm H2O/mm Soil 0.127 mm H2O/mm Soil .sol 
MSK.CO2 Coefficient for impact of low flow 3.5 3.65 .bsn 
CANMX Maximum water storage in canopy 0 mm 0.3 mm .hru 
Base flow ALPHA-BF Groundwater recession factor 0.084 d 0.27 d .gw 
GW-DELAY Groundwater delay 31 d 28 d .gw 
VariableDescriptionDefault Value(s)Calibrated ValueInput file
Flow CN2 Initial SCS runoff curve number for moisture condition 45–85 49 .mgt 
SOL_K Saturated hydraulic conductivity 2.6–52 mm/h 7.54 mm/h .sol 
SOL_BD Soil moist bulk density 0.9–2.5 g/cm3 1.36 g/cm3 .sol 
ESCO Soil evaporation compensation factor 0.95 0.9018 .sol 
SOL_AWC Available water capacity of the soil 0.12–52 mm H2O/mm Soil 0.127 mm H2O/mm Soil .sol 
MSK.CO2 Coefficient for impact of low flow 3.5 3.65 .bsn 
CANMX Maximum water storage in canopy 0 mm 0.3 mm .hru 
Base flow ALPHA-BF Groundwater recession factor 0.084 d 0.27 d .gw 
GW-DELAY Groundwater delay 31 d 28 d .gw 
Figure 2

Graphical comparison of simulated streamflow against observed flow for subbasins 1 and 10 calibration (January 2004–December 2011) and validation (January 2012–December 2018) periods. Subbasin 1 is depicted in grey and subbasin 10 in orange. Continuous and dashed lines represent simulated and observed data, respectively. The blue line in the secondary vertical axis shows the daily precipitation. Please refer to the online version of this paper to see this figure in colour: https://dx.doi.org/10.2166/wh.2022.266.

Figure 2

Graphical comparison of simulated streamflow against observed flow for subbasins 1 and 10 calibration (January 2004–December 2011) and validation (January 2012–December 2018) periods. Subbasin 1 is depicted in grey and subbasin 10 in orange. Continuous and dashed lines represent simulated and observed data, respectively. The blue line in the secondary vertical axis shows the daily precipitation. Please refer to the online version of this paper to see this figure in colour: https://dx.doi.org/10.2166/wh.2022.266.

Close modal

Figure 2 and 3 show the daily simulated concentrations of Cryptosporidium and E. coli in subbasins 10, 8, 4, and 1. In Figure 4, these are plotted against weekly monitored E. coli concentrations between January 2004 and December 2018. The output from the SWAT model shows slightly low variability, a standard artefact observed in microbiological modelling (Iqbal & Hofstra 2019). In general, the trend in the SWAT model simulations follows the observed data patterns in most subbasins. The model may not have captured some of the extremities observed in the monitoring data; however, the simulated peaks are one order of magnitude less than those of the monitoring data. Comparing the monitored and simulated mean of the E. coli concentration shows that for subbasins 4 and 1 the SWAT model overestimates, while for subbasin 10 and 8, it slightly underestimates the concentrations.

Figure 3

Simulated Cryptosporidium concentrations in basins 1, 4, 8, and 10. The black line represents subbasin 10, the grey line represents subbasin 4, the orange line represents subbasin 8, and the blue line represents subbasin 1. Please refer to the online version of this paper to see this figure in colour: https://dx.doi.org/10.2166/wh.2022.266.

Figure 3

Simulated Cryptosporidium concentrations in basins 1, 4, 8, and 10. The black line represents subbasin 10, the grey line represents subbasin 4, the orange line represents subbasin 8, and the blue line represents subbasin 1. Please refer to the online version of this paper to see this figure in colour: https://dx.doi.org/10.2166/wh.2022.266.

Close modal
Figure 4

Simulated and monitored (#cfu/100 ml) E. coli concentrations in basins 10, 8, 4, and 1 (respectively from top to bottom). The black line represents combination of sources, the grey line represents human sources, the orange line represents animal sources, and the blue circles represent monitored E. coli. Please refer to the online version of this paper to see this figure in colour: https://dx.doi.org/10.2166/wh.2022.266.

Figure 4

Simulated and monitored (#cfu/100 ml) E. coli concentrations in basins 10, 8, 4, and 1 (respectively from top to bottom). The black line represents combination of sources, the grey line represents human sources, the orange line represents animal sources, and the blue circles represent monitored E. coli. Please refer to the online version of this paper to see this figure in colour: https://dx.doi.org/10.2166/wh.2022.266.

Close modal

Figure 3 shows that subbasin 4 has the highest Cryptosporidium concentration, followed by subbasin 1, then subbasins 8 and 10, respectively. The trends are generally such that concentrations are higher downstream the WWTP than upstream. Cryptosporidium concentrations analysed between September 2015 and March 2016 by Adeyemo (2019) in effluents from the wastewater treatment plants in Durban show a range of 50–70 oocysts/10 L (six samples). In the study by Dungeni & Momba (2010) in the Gauteng province (South Africa), the concentrations in effluents were <0.1–4 oocysts/10 L, in 14 samples taken between January and April 2008, weekly. In a study performed between March 2016 and March 2017 with fortnightly sampling, Razzolini et al. (2020) detected Cryptosporidium concentrations of <0.003–2.6 oocysts/10 L in wastewater effluent in Sao Paulo, Brazil. In Shanghai China, Ma et al. (2016) detected 0–0.1 oocyst/10 L Cryptosporidium concentrations in wastewater effluent from three WWTPs. In this study, values provided by de Jong (2020, pers. comm.), 5 oocysts/10 L, were used as these were from the Darvill WWTP specifically. Due to a short time-series, there is a possibility of underestimation. More continuous monitoring can help reduce this uncertainty. Robertson et al. (2020) suggest that the likelihood of Cryptosporidium contamination of drinking water in Africa is probably greater than in many other countries, due to water supplies being limited and a general lack of catchment control for securing the supply of water, as surface waters are both used as drinking water and for livestock drinking.

In reference to Cryptosporidium, DWAF (1996a) for domestic use, states that 1 oocyst/10 L (0.1 oocyst/L) is enough to pose the risk of infection. No guidelines are given for protozoa in recreational use.

In the DWAF (1996b) guidelines for recreational use, the effects of E. coli on human health based on their concentration are outlined as follows: The concentrations of 0–130 counts per 100 mL may pose a low risk of gastrointestinal illness for full-contact recreational water use; negligible effects are expected if these levels occur in isolated instances only. The concentrations of 200–400 counts per 100 mL may pose some risk of gastrointestinal effects, particularly if this occurs frequently. Above 400 counts per 100 mL will pose increasing risk of gastrointestinal effects. The criteria proposed for full-contact recreation (swimming) are also recommended for intermediate contact (canoeing) (DWAF 1996b).

In the DWAF (1996a) guidelines for domestic use, the effects of faecal coliforms on human health based on their concentration are as follows: 0 will pose negligible risk of microbial infection; 0–10 counts per 100 mL will pose a slight risk of microbial infection with continuous exposure and negligible effects with occasional or short-term exposure; 10–20 counts per 100 mL will pose risk of infectious disease transmission with continuous exposure and slight risk with occasional exposure; and >20 counts per 100 mL will pose significant and increasing risk of infectious disease transmission.

In subbasins 10 and 8, animal sources contribute more than human sources to the combined E. coli concentrations, on the other hand, subbasins 4 and 1 show high human source impact. The model has successfully captured the trends of different sources in different settlements within the catchment. For subbasin 10 (upper rural), human contributions are lower than animal contributions, highlighting the intensive and uncontrolled livestock farming. As subbasin 8 is a transition between rural and urban areas, human contributions are still lower than animal contributions in these parts. Broken sewers do contribute in these parts, albeit low. In subbasin 4, there is a significant contribution from human sources (Figure 3) from the WWTP and broken sewers. Gemmell & Schmidt (2013) pointed out that the water quality of the uMsunduzi River is poor, mainly due to faecal pollution, and deteriorates as the river passes through the city.

The data in Figures 2 and 3 show that E. coli concentrations in the uMsunduzi River render the water unsuitable for drinking, swimming, and canoeing due to the high possibility of contracting gastrointestinal diseases, according to DWAF (1996a, 1996b).

There are several limitations to this study in terms of the microbial estimates in the SWAT model. Cryptosporidium loads in the effluent from the WWTP were represented as a point value rather than using a time series. The quantification of the contribution from broken sewers was an estimate since the reporting of this did not include any information on discharge. There are also limitations in available data on pathogen parameters, for example, the proportion of pathogenic E. coli. In the townships and rural areas, pit latrines are used as sanitation infrastructure, but in the study area, their contribution to the surface waters could not be quantified. Also, microbial contributions from direct stream deposits from livestock and wildlife were not estimated. Defaecating in open spaces could also be a human source of microorganisms in this catchment. It was, however, not considered in this study due to difficulties in quantifying its contribution in the runoff.

Quantitative microbial risk assessment results

Table 7 shows the probability of infection (Pinf), 50th and 95th percentile, from different exposure scenarios to Cryptosporidium and pathogenic E. coli, respectively. Table 7 also shows Pinf calculated using monitored E. coli. The Pinf calculated using simulated data and using the monitoring data generally fall into the same order of magnitude. To put the results of this study into the context of the existing guidelines, the concentrations stated by DWAF were recalculated into the probability of infection. In the DWAF (1996a) guidelines for domestic use, the effect of faecal coliforms on human health based on their concentration is 0 counts/100 mL for negligible risk of microbial infection. In terms of QMRA inputs, uncertainty lies with the ingested volumes for exposure routes.

Table 7

Probability of infection, Pinf, by Cryptosporidium and pathogenic E. coli across subbasins

Exposure Scenario (per day)Age groupLocationPinf 50th (95th) Percentile CryptosporidiumPinf 50th (95th) Percentile Modelled E. coliPinf 50th (95th) Percentile Monitored E. coli
Swimming Adults Sub 10 1.78×10−04 (2.59×10−033.23×10−01 (9.99×10−016.34×10−02 (9.83×10−01
Sub 8 2.4×10−03 (1.91×10−022.51×10−01 (9.94×10−013.02×10−01 (1.00×10+00
Sub 4 1.96×10−03 (2.37×10−024.58×10−01 (1.00×10+002.15×10−01 (1.00×10+00
Sub 1 1.01×10−03 (1.41×10−022.99×10−01 (9.98×10−012.12×10−02 (5.10×10−01
Children Sub 10 3.17×10−04 (5.35×10−035.79×10−01 (1.00×10+001.18×10−01 (1.00×10+00
Sub 8 5.08×10−03 (3.87×10−024.33×10−01 (1.00×10+005.38×10−01 (1.00×10+00
Sub 4 3.90×10−03 (4.99×10−027.55×10−01 (1.00×10+004.04×10−01 (1.00×10+00
Sub 1 2.44×10−03 (2.90×10−025.23×10−01 (1.00×10+004.41×10−02 (7.99×10−01
Canoeing Juvenal Sub 8 3.14×10−05 (6.52×10−045.99×10−02 (6.7×10−017.18×10−02 (8.99×10−01
Junior Sub 8 6.34×10−05 (1.13×10−031.00×10−01 (8.54×10−011.24×10−01 (9.74×10−01
Senior Sub 8 9.25×10−05 (1.58×10−031.38×10−01 (9.23×10−011.69×10−01 (9.94×10−01
Drinking Juvenal Sub 10 1.34×10−03 (3.49×10−029.65×10−01 (1.00×10+007.78×10−01 (1.00×10+00
Sub 1 8.15×10−03 (1.78×10−019.54×10−01 (1.00×10+001.71×10−01 (1.00×10+00
Junior Sub 10 4.00×10−03 (6.47×10−021.00×10+00 (1.00×10+009.90×10−01 (1.00×10+00
Sub 1 2.81×10−02 (3.10×10−011.00×10+00 (1.00×10+004.22×10−01 (1.00×10+00
Senior Sub 10 5.51×10−03 (7.87×10−021.00×10+00 (1.00×10+001.00×10+00 (1.00×10+00
Sub 1 3.28×10−02 (3.63×10−011.00×10+00 (1.00×10+005.06×10−01 (1.00×10+00
Exposure Scenario (per day)Age groupLocationPinf 50th (95th) Percentile CryptosporidiumPinf 50th (95th) Percentile Modelled E. coliPinf 50th (95th) Percentile Monitored E. coli
Swimming Adults Sub 10 1.78×10−04 (2.59×10−033.23×10−01 (9.99×10−016.34×10−02 (9.83×10−01
Sub 8 2.4×10−03 (1.91×10−022.51×10−01 (9.94×10−013.02×10−01 (1.00×10+00
Sub 4 1.96×10−03 (2.37×10−024.58×10−01 (1.00×10+002.15×10−01 (1.00×10+00
Sub 1 1.01×10−03 (1.41×10−022.99×10−01 (9.98×10−012.12×10−02 (5.10×10−01
Children Sub 10 3.17×10−04 (5.35×10−035.79×10−01 (1.00×10+001.18×10−01 (1.00×10+00
Sub 8 5.08×10−03 (3.87×10−024.33×10−01 (1.00×10+005.38×10−01 (1.00×10+00
Sub 4 3.90×10−03 (4.99×10−027.55×10−01 (1.00×10+004.04×10−01 (1.00×10+00
Sub 1 2.44×10−03 (2.90×10−025.23×10−01 (1.00×10+004.41×10−02 (7.99×10−01
Canoeing Juvenal Sub 8 3.14×10−05 (6.52×10−045.99×10−02 (6.7×10−017.18×10−02 (8.99×10−01
Junior Sub 8 6.34×10−05 (1.13×10−031.00×10−01 (8.54×10−011.24×10−01 (9.74×10−01
Senior Sub 8 9.25×10−05 (1.58×10−031.38×10−01 (9.23×10−011.69×10−01 (9.94×10−01
Drinking Juvenal Sub 10 1.34×10−03 (3.49×10−029.65×10−01 (1.00×10+007.78×10−01 (1.00×10+00
Sub 1 8.15×10−03 (1.78×10−019.54×10−01 (1.00×10+001.71×10−01 (1.00×10+00
Junior Sub 10 4.00×10−03 (6.47×10−021.00×10+00 (1.00×10+009.90×10−01 (1.00×10+00
Sub 1 2.81×10−02 (3.10×10−011.00×10+00 (1.00×10+004.22×10−01 (1.00×10+00
Senior Sub 10 5.51×10−03 (7.87×10−021.00×10+00 (1.00×10+001.00×10+00 (1.00×10+00
Sub 1 3.28×10−02 (3.63×10−011.00×10+00 (1.00×10+005.06×10−01 (1.00×10+00

A value of 0.8 counts per L (based on 1 count/100 mL), accounting for pathogenic portion, was used together with different ingestion scenarios in Table 4, to calculate the associated probability of infection. For domestic use, the recalculated 50th percentile risk values were 6/10,000 (juvenile), 2/1,000 (junior), and 3/1,000 (senior). For swimmers and canoeists, according to DWAF (1996b), low risk of gastrointestinal illness is considered as below 8/1,000.

For Cryptosporidium, for domestic use, the guidelines by DWAF (1996a) state that 1 oocyst/10 L is enough to pose the risk of infection; and no guidelines are given for protozoa in recreational use. Therefore, a concentration of 1 oocyst/10 L as a point value was used together with different ingestion scenarios in Table 4 to calculate the probability of infection. The 50th percentile by DWAF (1996b) for swimming is 9/10,000 children and adults 4/10,000 adults with respect to Cryptosporidium. For canoeing: 8/100,000 Juvenile, 2/10,000 Junior, and 2/10,000 Seniors were calculated based on the DWAF (1996b).

The findings of this study are summarised below in terms of the 50th percentile of probability of infection calculated using the SWAT output data.

Recreational swimming

In reference to Cryptosporidium, subbasins 1, 4, and 8 show generally higher probability of infection than the calculated 50th percentile by DWAF (1996b). In reference to pathogenic E. coli, the probability of infection is above the DWAF ‘low risk category’, with the highest risks in subbasin 4.

Canoeing

In subbasin 8, the probability of infection with Cryptosporidium is lower than the recalculated values from DWAF (1996b) and higher for Junior and Senior canoeists. The probability of infection with pathogenic E. coli during canoeing in subbasin 8 was found to be higher than the risk suggested by DWAF (1996b) of 8/1,000 canoeists across all scenarios.

Drinking water

QMRA results for Cryptosporidium indicate that the probability of infection, to the population that drink untreated river water in subbasin 1 is higher than DWAF (1996a) recommendations for protozoa in drinking water. This risk may be reduced by boiling the water before drinking. On the contrary, in subbasin 10, for all groups, the probability of infection is lower than the thresholds in DWAF (1996a). Based on pathogenic E. coli, QMRA results show that in subbasin 1 and 10 the probability of infection is higher than the criteria set in DWAF (1996a). Smeets et al. (2010) used 2.74×10−07 per person per day as the drinking risk benchmark, based on the work by Signor & Ashbolt (2009). Both subbasins 1 and 10, across all drinking scenarios, are above the 2.74×10−07 per person per year.

Mbanga et al. (2020) quantified E. coli contamination and the potential health hazards that workers at the Darvill WWTP and nearby informal communities may face after exposure to waterborne pathogenic E. coli in the WWTP and the uMsunduzi River. The authors found that the daily probability of infection with pathogenic E. coli following intentional uptake of 100 mL of the river water upstream and downstream from Darvill is 97.6 and 90.8%, respectively.

Several local studies (WRC 2002; Gemmell & Schmidt 2013; Mbanga et al. 2020) highlight that the water quality of the uMsunduzi River is poor to very poor, with apparent faecal pollution which impacts on Inanda Dam which serves the eThekwini Municipality with potable water. The findings of this study agree with these studies where they concluded that the uMsunduzi River water did not meet the requirements of DWAF for drinking and recreational uses, in terms of E. coli.

Risk of infection during swimming is highest in the urban parts of the catchment for both children and adults, considering both pathogens. This part receives contribution from sewers, WWTP, as well as animals. Canoeing was considered only for training events in this study. Canoeing competitions do take place in the uMsunduzi catchment. One of these is the historic Dusi Canoe Marathon taking place in February each year, along the uMsunduzi River to the uMngeni River all the way to Durban. It attracts 900–1,600 paddlers each year. Hay (2017) reported that a survey done in 2016, immediately after the race, revealed that 40% of the paddlers contracted mild to severe gastrointestinal infections. This illness has been termed the ‘Dusi guts’ by the paddlers. It is also reported that seasoned paddlers take antibiotics before the race, as a preventative measure.

Those who drink the water from the river are at the highest risk of infection by both studied pathogens. The rural communities and those in peri-urban areas are affected the most. These communities supplement the lack/shortage of treated water with the water from the polluted uMsunduzi. These communities are not always aware of the risks and how they can protect themselves.

This study focused on exposure per event for risk calculations and did not account or relate to the frequency of these events. To extend this approach to annual risk, the frequency of drinking river water may be considered to occur on a daily basis. However, canoeists may not train every day of the year, the training may intensify closer to competitions compared to outside competitions. Additionally, recreational swimming may be more intense during hotter days of the year (September to March) as opposed to the colder days. These aspects are important to consider if results are reviewed and compared to an annual risk level. Nonetheless, the effect of risk mitigation can be evaluated on a per event or a daily risk level as well (Signor & Ashbolt 2009).

The aim of this study was to utilise SWAT and QMRA to highlight areas of high pollution and the risk that the water may pose to the health of consumers. Based on the water quality modelling results of this study, it can be concluded that the uMsunduzi River is highly polluted with pathogens, and the use of untreated water from the river may result in a high risk of infection.

In the urban section, human sources contribute more than animal sources, indicating to the contributions from broken sewers. In terms of the QMRA, people who use the river for domestic and recreational purposes are at risk of infection by Cryptosporidium and pathogenic E. coli.

Investing in water treatment facilities, regulation of livestock practices, and safe sanitation systems for communities in need are likely to provide long-term and reliable improvement in the uMsunduzi River water quality. Indirect reuse through the river is not often considered when talking about wastewater reuse. The lack of treated water in rural areas and informal settlements usually forces people to reuse water without giving it much thought. Supporting rural communities without reticulated water is only possible if the water is not directly drawn from the river. Controlling how children access the river and educating the people about boiling their drinking water will help reduce the exposure to pathogens.

The authors would like to thank Sanele Vilakazi (Duzi-Umgeni Conservation Trust (DUCT) River Care Team, Pietermaritzburg, South Africa); Anton de Jong (Uppsala University, Sweden); Zethule Mthalane (Canoeist, Pietermaritzburg, South Africa); Frida Fischer, Yazan Hamdan, Amanda Hansson, Elin Josefsson, and Lisa Sundström (Chalmers University, Sweden); and Debbie Trollip (Umgeni Water, Pietermaritzburg, South Africa) for providing valuable input and data.

The authors are grateful to the following funders: ‘ClimAQua – Modelling climate change impacts on microbial risks for a safe and sustainable drinking water system’ grant number 2017-01413 funded by Formas – the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning. ‘Cryptosporidium and Giardia: water quality modelling, microbial source tracking, and risk assessment’ grant number SA2015-6274 funded by STINT – the Swedish Foundation for International Cooperation in Research and Higher Education.

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

Abdullah
J.
,
Muhammad
N. S.
,
Asiah
S.
,
Farahain
N.
,
Amin
M.
,
Tahir
W.
&
Township
S.
2019
Modelling trends in hydrological modelling
.
Jurnal Teknologi (Sciences and Engineering)
4
(
1
),
1
11
.
Abrahamsson
J. L.
,
Ansker
J.
&
Heinicke
G.
2009
MRA – A Model for Swedish Waterworks
.
Stockholm
,
Sweden
.
Adeyemo
F. E.
2019
Occurrence and Variations of Cryptosporidium and Giardia in Wastewater Treatment and Receiving River Basins
.
Durban University of Technology
,
Durban
,
South Africa
.
Araya
P.
,
Hug
J.
,
Joy
G.
,
Oschmann
F.
&
Rubistein
S.
2016
The Impact of Water and Sanitation on Diarrhoeal Disease Burden and Over-Consumption of Antibiotics
.
London
.
Arnold
J. G.
,
Moriasi
D.
,
Gassman
P.
,
Abbaspour
K. C.
,
Srinivasan
R.
,
Santhi
C.
,
Harmel
R. D.
&
Griensven
A. V.
2012
SWAT: model use, calibration, and validation
.
2012 American Society of Agricultural and Biological Engineers
55
(
4
),
1491
1508
.
ISSN 2151-0032 1491
Baffaut
C.
&
Benson
V. W.
2009
Modeling flow and pollutant transport in a Karst watershed with SWAT
.
Transactions of the ASABE
52
(
2
),
469
479
.
Bergion
V.
,
Sokolova
E.
,
Åström
J.
,
Lindhe
A.
,
Sörén
K.
&
Rosén
L.
2017
Hydrological modelling in a drinking water catchment area as a means of evaluating pathogen risk reduction
.
Journal of Hydrology
544
,
74
85
.
doi:10.1016/j.jhydrol.2016.11.011
.
Bergion
V.
,
Lindhe
A.
,
Sokolova
E.
&
Rosén
L.
2020
Accounting for unexpected risk events in drinking water systems
.
Exposure and Health
(0123456789)
.
doi:10.1007/s12403-020-00359-4
.
Bougeard
M.
,
Saux
J. L.
,
Teillon
A.
,
Belloir
J.
,
Mennec
C. L.
,
Thome
S.
,
Durand
G.
&
Pommepuy
M.
2011
Combining modeling and monitoring to study fecal contamination in a small rural catchment
.
Journal of Water and Health
,
467
482
.
doi:10.2166/wh.2011.189
.
Coffey
R.
,
Dorai-Raj
S.
,
O'Flaherty
V.
,
Cormican
M.
&
Cummins
E.
2013
Modeling of pathogen indicator organisms in a small-scale agricultural catchment using SWAT
.
Human and Ecological Risk Assessment
19
(
1
),
232
253
.
doi:10.1080/10807039.2012.701983
.
Dabrowski
J.
,
Bruton
S.
,
Graham
M.
,
Hill
T.
,
Murray
K.
,
Rivers-moore
N.
&
Deventer
H. V.
2013
Linking Land Use to Water Quality for Effective Water Resource and Ecosystem Management
.
Pretoria
,
South Africa
.
Dorevitch
S.
,
Panthi
S.
,
Huang
Y.
,
Li
H.
,
Michalek
A. M.
,
Pratap
P.
,
Wroblewski
M.
,
Liu
L.
,
Scheff
P. A.
&
Li
A.
2011
Water ingestion during water recreation 5
.
Water Research
5
(
45
),
2020
2028
.
doi:10.1016/j.watres.2010.12.006
.
Dufour
A. P.
,
Evans
O.
,
Behymer
T. D.
&
Cantu
R.
2006
Water ingestion during swimming activities in a pool: a pilot study
.
Journal of Water and Health
04
(
4
).
doi:10.2166/wh.2006.017
.
Dungeni
M.
&
Momba
M. N. B.
2010
The Abundance of Cryptosporidium and Giardia spp. in Treated Effluents Produced by Four Wastewater Treatment Plants in the Gauteng Province of South Africa
.
Pretoria
,
South Africa
.
DWAF
1996a
South African Water Quality Guidelines: Volume 1 Domestic Use
, 2nd edn.
Department of Water Affairs and Forestry
,
Pretoria
,
South Africa
.
DWAF
1996b
Water Quality Guidelines: Volume 2 Recreational Use
, 2nd edn.
Department of Water Affairs and Forestry
,
Pretoria
,
South Africa
.
FAO
2015
World Reference Base for Soil Resources
.
Nature Nanotechnology. Viale delle Terme di Caracalla
.
doi:10.1038/nnano.2009.216
.
Gemmell
M. E.
&
Schmidt
S.
2013
Is the microbiological quality of the Msunduzi River (KwaZulu-Natal, South Africa) suitable for domestic, recreational, and agricultural purposes?
Environmental Science and Pollution Research
20
(
9
),
6551
6562
.
doi:10.1007/s11356-013-1710-1
.
Haas
C. N.
,
Thayyar-madabusi
A.
,
Rose
J. B.
&
Gerba
C. P.
2000
Development of a dose-response relationship for Escherichia coli O157:H7
.
International Journal of Food Microbiology
1748
,
153
159
.
Haas
C. N.
,
Rose
J. B.
&
Gerba
C. P.
2014
Quantitative Microbial Risk Assessment
, 2nd edn.
John Wiley & Sons, Inc
,
New Jersey
.
Hay
D.
2017
Our Water Our Future: Securing the Water Resources of the uMngeni River Basin
.
Pietermaritzburg
,
South Africa
.
Iqbal
M. S.
&
Hofstra
N.
2019
Human and ecological risk assessment: an international modeling Escherichia coli fate and transport in the Kabul River basin using SWAT
.
Human and Ecological Risk Assessment: An International Journal
25
(
5
),
1279
1297
.
doi:10.1080/10807039.2018.1487276
.
Jarmey-Swan
C.
,
Bailey
I. W.
&
Howgrave-Graham
A. R.
2001
Ubiquity of the water-borne pathogens, Cryptosporidium and Giardia, in KwaZulu-Natal populations
.
Water SA
27
(
1
),
57
64
.
doi:10.4314/wsa.v27i1.5011
.
Luyt
C. D.
,
Tandlich
R.
,
Muller
W. J.
&
Wilhelmi
B. S.
2012
Microbial monitoring of surface water in South Africa: an overview
.
International Journal of Environmental Research and Public Health
9
(
8
),
2669
2693
.
doi:10.3390/ijerph9082669
.
Ma
J.
,
Feng
Y.
,
Hu
Y.
,
Villegas
E.
&
Xiao
L.
2016
Human infective potential of Cryptosporidium spp., Giardia duodenalis and Enterocytozoon bieneusi in urban wastewater treatment plant effluents
.
Journal of Water Health
176
(
10
),
139
148
.
doi:10.2166/wh.2016.192.Human
.
Mander
M.
,
Jewitt
G.
,
Dini
J.
,
Glenday
J.
,
Blignaut
J.
,
Hughes
C.
,
Marais
C.
,
Maze
K.
,
Waal
B. V. D.
&
Mills
A.
2017
Modelling potential hydrological returns from investing in ecological infrastructure: case studies from the Baviaanskloof-Tsitsikamma and uMngeni catchments, South Africa
.
Ecosystem Services
.
doi:10.1016/j.ecoser.2017.03.003
.
Mbanga
J.
,
Abia
A. L. K.
,
Amoako
D. G.
&
Essack
S. Y.
2020
Quantitative microbial risk assessment for waterborne pathogens in a wastewater treatment plant and its receiving surface water body
.
BMC Microbiology
20
(
1
),
1
12
.
doi:10.1186/s12866-020-02036-7
.
Mema
V.
2010
Impact of Poorly Maintained Wastewater and Sewage Treatment Plants: Lessons from South Africa
.
Council for Scientific and Industrial Research (CSIR)
,
Pretoria
.
Moriasi
D. N.
,
Arnold
J. G.
,
Liew
M. W. V.
,
Bingner
R. L.
,
Harmel
R. D.
&
Veith
T. L.
2007
Model evaluation guidelines for systematic quantification of accuracy in watershed simulations
.
Transactions of the ASABE 2007 American Society of Agricultural and Biological Engineers
50
(
3
),
885
900
.
ISSN 0001-2351
Müller
E. E.
,
Ehlers
M. M.
&
Grabow
W. O. K.
2001
The occurrence of E. coli O157:H7 in South African water sources intended for direct and indirect human consumption
.
Water Research
35
(
13
),
3085
3088
.
doi:10.1016/S0043-1354(00)00597-2
.
Naghavi
M.
,
Wang
H.
,
Lozano
R.
,
Davis
A.
,
Liang
X.
&
Zhou
M. GBD
2013
Mortality and Causes of Death Collaborators 2015
Global, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990 -2013: a systematic analysis for the Global Burden of Disease Study 2013
.
The Lancet
385
(
9963
),
117
171
.
doi:10.1016/S0140-6736(14)61682-2
.
Ojuromi
O.
&
Ashafa
A.
2018
Cryptosporidiosis in Southern Africa: review of prevalence and molecular epidemiology of a neglected disease
.
Annals of Tropical Medicine and Public Health
11
(
6
),
108
118
.
doi:10.4103/ATMPH.ATMPH
.
Osuolale
O.
&
Okoh
A.
2017
Human enteric bacteria and viruses in five wastewater treatment plants in the Eastern Cape, South Africa
.
Journal of Infection and Public Health
10
(
5
),
541
547
.
doi:10.1016/j.jiph.2016.11.012
.
Ramnath
A.
,
Meier
K.
&
Mjwara
S.
2019
Infrastructure Master Plan 2019 Volume 7: Wastewater
.
Pietermaritzburg
,
South Africa
.
Available from: www.umgeni.co.za.
Razzolini
M. T. P.
,
Breternitz
B. S.
,
Kuchkarian
B.
&
Bastos
V. K.
2020
Cryptosporidium and Giardia in urban wastewater: a challenge to overcome
.
Environmental Pollution
257
,
113545
.
doi:10.1016/j.envpol.2019.113545
.
Robertson
L. J.
,
Johansen
Ø. H.
&
Kifleyohannes
T.
2020
Cryptosporidium infections in Africa – How important is zoonotic transmission? A review of the evidence
.
Frontiers in Veterinary Science
7
,
1
25
.
doi:10.3389/fvets.2020.575881
.
Sadeghi
A. M.
&
Arnold
J. G.
2002
A SWAT/microbial sub-model for predicting pathogen loadings in surface and groundwater at watershed and basin scales
. In:
Total Maximum Daily Load (TMDL) Environmental Regulations: Proceedings of the March 11–13, 2002 Conference
(
Saleh
A.
, ed.).
ASAE Publication Number 701P0102
,
Fort Worth, Texas
,
USA
, pp.
56
63
.
Sekwadi
P. G.
,
Ravhuhali
K. G.
,
Mosam
A.
,
Essel
V.
,
Ntshoe
G. M.
&
Govender
N.
2019
Waterborne outbreak of gastroenteritis on the KwaZulu-Natal Coast, South Africa, December 2016/January 2017
.
Epi
146
,
1318
1325
.
Shirley
D. A. T.
,
Moonah
S. N.
&
Kotloff
K. L.
2012
Burden of disease from cryptosporidiosis
.
Current Opinion in Infectious Diseases
,
555
563
.
doi:10.1097/QCO.0b013e328357e569
.
Smeets
P. W. M. H.
,
Rietveld
L. C.
,
Dijk
J. C. V.
&
Medema
G. J.
2010
Practical applications of quantitative microbial risk assessment (QMRA) for water safety plans
.
Water Science & Technology
61
(
6
),
1561
1568
.
doi:10.2166/wst.2010.839
.
STATSSA
2011
Census 2011 Summary, SA Library Cataloguing
.
doi:10.1136/bmj.286.6372.1195
.
Tang
J.
,
Mcdonald
S.
,
Peng
X.
&
Samadder
S. R.
2011
Modelling Cryptosporidium oocysts transport in small ungauged agricultural catchments
.
Water Research
45
(
12
),
3665
3680
.
doi:10.1016/j.watres.2011.04.013
.
Teunis
P. F. M.
,
Chappell
C. L.
&
Okhuysen
P. C.
2002
Cryptosporidium dose response studies: variation between isolates
.
Risk Analysis
22
(
1
),
175
185
.
doi:10.1111/0272-4332.00014
.
Teunis
P.
,
Takumi
K.
&
Shinagawa
K.
2004
Dose response for infection by Escherichia coli O157:H7 from outbreak data
.
Risk Analysis
24
(
2
),
401
407
.
doi:10.1111/j.0272-4332.2004.00441.x
.
Vermeulen
L. C.
,
Benders
J.
,
Medema
G.
&
Hofstra
N.
2017
Global Cryptosporidium loads from livestock manure
.
Environmental Science and Technology
51
(
15
),
8663
8671
.
doi:10.1021/acs.est.7b00452
.
Westrell
T.
2004
Microbial Risk Assessment and its Implications for Risk Management in Urban Water Systems
.
Linköping University
,
Linköping, Sweden
.
WRC
2002
State of the River Report – uMngeni River and Neighbouring Rivers and Streams
.
Pretoria
,
South Africa
.
Yuan
L.
,
Sinshaw
T.
&
Forshay
K. J.
2020
Review of watershed-scale water quality and nonpoint source pollution models
.
Geosciences (Switzerland)
10
(
1
),
1
33
.
doi:10.3390/geosciences10010025
.
Zhang
Q.
,
Gallard
J.
,
Wu
B.
,
Harwood
V. J.
,
Sadowsky
M. J.
,
Hamilton
K. A.
&
Ahmed
W.
2019
Synergy between quantitative microbial source tracking (qMST) and quantitative microbial risk assessment (QMRA): a review and prospectus
.
Environment International
130
,
104703
.
doi:10.1016/j.envint.2019.03.051
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).