key: cord-0921848-4g99vzpp
authors: Petterson, Susan; Li, Qiaozhi; Ashbolt, Nicholas
title: Screening Level Risk Assessment (SLRA) of human health risks from faecal pathogens associated with a Natural Swimming Pond (NSP)
date: 2020-10-05
journal: Water Res
DOI: 10.1016/j.watres.2020.116501
sha: 4a48b64a87ac1e581cf2ac73dbecb18271c2d5a5
doc_id: 921848
cord_uid: 4g99vzpp

Natural swimming ponds (NSPs) are artificially created bodies of water intended for human recreation, characterised by the substitution of chemical disinfection with natural biological processes for water purification. NSPs are growing in popularity, however little is known regarding the public health risks. A screening level risk assessment was undertaken as an initial step in assessing the first Canadian public NSP located in Edmonton, Alberta. Risk of enteric pathogens originating from pool bathers was assessed under normal conditions and following accidental faecal release events. The performance of the natural treatment train for health protection was quantified with and without the addition of UV disinfection of naturally-treated water, and compared to the US EPA benchmark to provide a reference point to consider acceptability. Estimated levels of pathogen contamination of the pond was dependent upon the discrete number of shedders present, which in turn depended upon the prevalence of infection in the population. Overall performance of the natural disinfection system was dependent upon the filtration rate of the natural treatment system or turnover time. Addition of UV disinfection reduced the uncertainty around the removal efficacy, and mitigated the impact of larger shedding events, however the impact of UV disinfection on the natural treatment biome is unknown. Further information is needed on the performance of natural barriers for pathogen removal, and therefore challenge studies are recommended. Given the identified risks, the pool is posted that there is risk from accidental faecal releases, as in any natural water body with swimmers. Screening level risk assessment was a valuable first step in understanding the processes driving the system and in identifying important data gaps.

Natural swimming ponds (NSPs) are artificially created bodies of water intended for human recreation, characterised by the substitution of chemical disinfection with natural biological processes for water purification. The first NSP in recent times was built in Austria in the early 1980s, and by 2010 more than 20,000 NSPs had been constructed of which a hundred were open to the public (Littlewood 2005) cited in (Casanovas-Massana and Blanch 2013) .

Despite the growing popularity of these natural systems, little is known regarding the adequacy of the natural processes for enteric pathogen removal to provide adequate water quality for swimmers. Also, it is well known by health authorities that most recreational disease is not identified in normal health surveillance programs (Fewtrell and Kay 2015) , hence the majority go unreported even though there may be significant impacts on society including lost work days (Dwight et al. 2005) . In general, the chlorine-resistant parasitic protozoan, Cryptosporidium hominis is the leading cause of gastroenteritis in swimming pools, and results from faecal accidents/releases (Suppes et al. 2016 ) referred to in this paper as bather shedding. Human faeces may contain pathogens and is a known pathway of infection in recreational water environments (Chalmers 2012 , Dale et al. 2010 , Graciaa et al. 2018 , Pond 2005 . Hence, in a natural pool, without chemical disinfectant residual, expected bather shedding of pathogens will go untreated until the pool water is passed through sufficient 'natural' barriers, with human enteric viruses the most numerous and infectious of these enteric pathogens (Ashbolt 2015) .

While the social and health benefits associated with recreational water environments are well recognized, various outbreaks have been associated with microbiological contamination of recreational waters in natural ponds and lakes (Blostein 1991 , Paunio et al. 1999 , Sinclair et al. 2009 ) and inadequately disinfected swimming pools (Barna and Kádár 2012, Sinclair et al. 2009 ). Hence, the efficacy and reliability of disinfection treatments (natural or artificial) is central to maintaining the microbial safety of recreational waters.

The design and operation of the first Canadian NSP in Edmonton was the focus of this microbial risk assessment. The NSP treatment system designed by Polyplan Kreikenbaum Group GMBH (www.polyplan-umwelt.de) was part of a municipal pool upgrade. As part of the planning process, potential faecal pathogen risks were evaluated, using a Quantitative Microbial Risk Assessment (QMRA) framework. It is recommended to undertake QMRA at increasing levels of detail referred to as a tiered approach (WHO 2016) , beginning simply and only increasing complexity as needed. This paper documents the first Screening Level Risk Assessment of pathogen risk undertaken of a NSP. The objective of the screening level assessment, was to characterize the system in a very simplistic way, based on average flow rates (within a simple box flow model) and using available literature data assess factors driving illness risks from accidental ingestion of enteric viruses, bacteria and parasitic protozoa, and to identify future data collection needs for the purpose of quantify risks adequately to support health protection.

The NSP is part of a redevelopment of a public pool at Borden Park, Edmonton, Canada (https://www.edmonton.ca/activities_parks_recreation/borden-park-outdoor-pool.aspx).

The design consists of a large rectangular main pool, a shallow children's pool (referred to as kiddie pool), and an area of floor nozzles designed for children's water play. The water flow is connected between all pools (see Figure 1 ) via the pump well (B1) and is directed to external filtration units (Neptune filter, Hydrobotanic and submerse filters) for purification.

The public pool is fenced from wildlife, and fed by potable water from the mains drinking water supply, therefore the significant only source of faecal contamination considered in the assessment was from bathers. The first step in any QMRA is to undertake a problem formulation, defining clearly the purpose and scope of the assessment (WHO 2016). To address each class of microbial pathogen (viral, bacterial, parasitic protozoan) for the first tier in the QMRA process, the following reference pathogens were selected to represent each microbial group: Norovirus, Campylobacter jejuni and Cryptosporidium hominis; with Norovirus and C. jejuni representing some of the most prevalent pathogenic enteric viruses and bacteria reported in Edmonton sewage (Banting et al. 2016 , Qiu et al. 2015 and known to infect recreational swimmers (Guy et al. 2018) ; C. hominis was selected as oocysts are known to be far more resistant to environmental decay processes than Giardia cysts (Hamilton et al. 2018 ). Risk to adults and children were considered separately and compared with the gastrointestinal benchmark of the U.S. EPA for freshwaters of 35 cases of gastroenteritis per 1000 swimming events (EPA. 2012).

Exposure scenarios included in the risk assessment were for unintentional shedding during normal operation (nominal load = 252 adults and 17 children;

high bather load estimated as 1.5 times the nominal value =378 adults and 26 children) and

larger faecal release events.

First the magnitude of pathogen contamination due to unintentional bather shedding was estimated; and the level of treatment required in order to achieve the benchmark risk was quantified. The capacity of the designed filtration system for treating the pathogen load at the benchmark risk level was then assessed, including the system response to larger pathogen release incidents. The overall objective of the assessment was to assess the adequacy of the proposed treatment system for managing pathogen risks to bathers, and to identify future research needs to better characterize safety.

Enteric pathogens are transmitted by the faecal-oral route, and therefore the presence of pathogens in the water column is caused by shedding of faecal material by swimmers. In this model, the pathogen loading under typical operating conditions was estimated as

where N is the discrete number of people visiting the facility per day infected with a reference pathogen, f s is the amount of faecal material (grams) shed per person per day, and c rp is the concentration of reference pathogens in the faeces of infected individuals (organisms.g -1 ). The number of visitors shedding (shedders) reference pathogens on any given day will be a discrete number. Given the number of visitors (n), and the point prevalence (pp) of shedding in the general population the discrete number of shedders on any given day was modelled using a binomial distribution with parameters n and pp. A sample of the number of shedders was generated using Monte Carlo simulation (10, 000 iterations).

Faecal shedding: Unintentional shedding of faecal material by bathers is widely documented (Elmir et al. 2009 , Elmir et al. 2007 , Gerba 2000 . The amount of faeces shed per person is highly uncertain and was estimated relying on measurements from the literature of faecal indicator concentrations in bathing waters. Details of this analysis is included in supplementary material. Given the uncertainty in shedding mass, a reference distribution to describe the variability in faecal excretion was implemented in the model. A reference distribution is recommended to explicitly represent a plausible range, when there is a lack of data to fully describe a distribution (WHO 2016). We chose a triangular distribution (a distribution often used as a rough modelling tool where the range and most likely value within that range can be estimated (Vose, 2008) ) for faecal excretion with a mode equal to 0.6 and lower and upper bounds of 0.06 and 6 grams.

Pathogen shedding density: Reported concentrations of pathogens in the faeces (shedding density) is variable. The shedding density appears to vary between individuals, and over the course of an infection. Bambic et al. (2011) reviewed the published data and reported the following relevant to our chosen reference pathogens: a median Campylobacter concentration of 10 7 CFU.g -1 , ranging 10 1 -10 8 ; for Norovirus the median was 10 8 copies.g -1 , and a range of 10 4 -10 10 ; and for Cryptosporidium the median was 10 5 oocysts.g -1 , ranging 10 3 -10 6 . The range limits and medians were used to define triangular distributions for the shedding density of each reference pathogen.

Using a model constructed in Mathematica® (Wolfram International, version 11.1) a Monte Caro simulation (10, 000 random samples) was undertaken to obtain a random sample of the concentration of each reference pathogen in the main pool and the kiddie pool, assuming complete mixing. The sensitivity of the estimated concentration to each of the model input variables (number of shedders, pathogen density in faeces and magnitude of faecal shedding by bathers) was evaluating using the Spearman Rank Correlation Coefficient to compare each of the input random samples with the generated concentration sample.

Quantitative microbial risk assessment (QMRA) was used to assess the amount of treatment required in order to achieve the U.S. EPA benchmark risk. Using the approach illustrated in Figure 2 and the sample of pathogen concentrations from unintentional bather shedding, the required Log 10 reduction was estimated. A random sample representing the variability in the treatment requirements was generated using the pathogen concentration samples described in the previous section. A summary of the model input assumptions is included in Table 2 .

A simple box model was applied, assuming complete mixing within each component of the model over each time step, to evaluate the overall pathogen reduction capacity of the treatment system. Given the flow paths represented in Figure 1 and the standard and maximum flow rates included in Table 3 , together with the assumed Log 10 removal capacity of each of the barriers, the change in pathogen concentration over time was estimated.

Five removal barriers were considered in the scoping of the QMRA: a wetland system consisting of zooplankton filtering, a hydro-botanic filter and submerse sand/root filter; a commercial designed Neptune™ surface spray gravel media filter; and UV disinfection ( Figure 1 ). However, there is very limited published data on the performance of these NSP barriers (Bruns and Pepper, 2019) , so in combination with related literature estimates we provided reasonable point estimate and plausible ranges in Log 10 removals for each barrier to be applied within the Screening level QMRA model (Table 4 ). As such, triangular distributions (defined by mode [min and max]) were selected to describe the variability and uncertainty associated with the Log 10 removals.

Zooplankton filtering: Zooplankton grazing is proposed to provide important in-situ disinfection in NSPs (Bruns and Peppler, 2019 ), yet requires careful consideration as to how well it may be expected to perform as a barrier for protection of human health. Studies have shown that free-living environmental protozoa can ingest human enteric bacteria and protozoan pathogens including Cryptosporidium oocysts (Agasild and Nõges 2005 , Connelly et al. 2007 , Stott et al. 2003 , Trout et al. 2002 ). Yet very little is known regarding the rate that zooplankton can clear (oo)cysts from the surrounding water, however rates of 22-24 mL·grazer -1 ·day -1 and 15-19 mL·grazer -1 ·day -1 for Cryptosporidium oocysts and Giardia cysts respectively have been reported (Connelly et al. 2007 ). The filtration capacity of zooplankton reported by (Eydeler and Spieker 2010) cited in (Bruns and Peppler 2019) ranged from 8.5 to 64.8 mL·grazer -1 ·day -1 for Rotatoria, Copepoda and Cladocera protozoa.

The more pressing issue relates to the poorly documented fate of ingested pathogens within zooplankton and their faecal pellets. Rather than inactivate and digest all ingested pathogens, there is growing evidence that pathogens may be concentrated within zooplankton, where they may be protected from external environmental stressors ultimately favoring their survival (Bichai et al. 2010 , Bichai et al. 2014 , Bichai et al. 2008 , Folkins et al. 2020 , Tang et al. 2011 . So, while faecal pellets may accumulate within sediments and on plant surfaces, and unknown fraction may stay suspended or be resuspended. For these reasons, zooplankton filtering was not included as a barrier for pathogen removal within this screening-level assessment of the NSP system.

Two parallel filters were proposed in the design of the NSP including: 1. Hydro-botanic filter with a bed thickness of 2.00 m and a design retention time of 6-8 hours; and 2. Submerse filter with a bed thickness of 2.00 m; and a design retention time of 4-5 hours.

While these two filters operate in parallel, regarding their likely removal efficiency they are considered here together due to the similarity of mechanisms. Mechanisms of pathogen removal within these filters include adsorption to soil and filter substratum, predation (noting issues identified for zooplankton filtering above), sedimentation, physical filtering of larger organisms and inactivation due to environmental exposure (pH, temperature, sunlight

[only for the hydro-botanic filter]). Both filters have been estimated to achieve 10%

reduction of E. coli (Bruns and Peppler 2019) .

Limited data is available in the literature specifically for hydro-botanic and submerse filters however a range of removal performances for microorganisms have been reported for wastewater systems using subsurface wetlands, rock filters and reed beds (reviewed by

Verbyla (2015)). Bacterial reductions are in the range of 1.25 -2.5 Log 10 ; viruses 0.5 to 2

Log 10 and parasitic protozoa between 0.4 and 3 Log 10 (Adhikari et al. 2013 , Bastos et al. 2010 , Garcia et al. 2010 , Gerba et al. 1999 , Jackson and Jackson 2008 , Karim et al. 2004 , Reinoso et al. 2008 , Stevik et al. 2004 , Vidales-Contreras et al. 2012 , Vidales et al. 2003 , Vymazal 2005 ; however noting that a key driver of removal is often the retention time within the filter, and retention times in these studies were typically in the order of days rather than hours proposed for the NSP. While the design only assumes 10% removal for each of these filters, it seems reasonable to assume that at least 1 Log 10 would be achieved for bacteria and protozoa, with a lower value of 0.5 Log 10 for viruses.

Neptune filter: The Neptune filter system has a gravel medium depth of 2.00 to 2.20 m; and a retention time of 20 min was assumed by designers to have a higher E. coli removal efficacy of 90% in comparison to the hydro-botanic and submerse filters. Challenge studies with coliphage (assumed to be a reasonable human enteric virus surrogate to mimic their) on similar filters has shown between 95 and 99% reduction (Bruns and Peppler 2019) .

UV disinfection: UV disinfection is considered as an option in the design of the treatment system. The most extensive meta-analysis review of data relating to pathogen sensitivity to UV has is by Hijnen et al. (Hijnen et al. 2006 Hijnen et al. (2006) , it is noted that if human adenovirus results were used, the Log 10 would be 0.5, rather than 2.6.

Given a starting level of contamination of 500 Campylobacter per L; 10 000 Norovirus per L and 10 Cryptosporidium oocysts per L; the flow path model was run to estimate the decrease in concentration over time, and the subsequent Log 10 reduction in the main pool and the kiddie pool, at 1 minute time steps for 24 hours. The model was run with and without the proposed UV disinfection units.

Given the importance of system water turnover time through treatment on the overall estimated pathogen reduction, estimated Log 10 reductions were evaluated for turnover periods of 1, 2, 4, 6 and 8 hours. Noting that current design turnover of the main pool was specified at 11 and 7.4 hours for standard and maximum flow rates respectively; and in the kiddie pool, 4 and 2.7 hours respectively.

In addition to regular contamination, the response of the system to a larger scale faecal release event was investigated. The flow model was applied to assess the response to an accidental release by an ill swimmer. The fate of a large pathogen release (faeces or vomitus) 10 10 Campylobacter; 10 12 Norovirus and 10 8 Cryptosporidium oocysts occurring in either the main pool, kiddie pool or at the floor nozzles was modelled in all three locations over 24 hours.

The Monte Carlo sample of the number of shedders visiting the pool per day is illustrated for the main pool and the kiddie pool in Figures 3 & 4 , respectively. The number of shedders was driven by the estimated point prevalence of each reference pathogen in the population (Table 1) Figure 7 . When prevalence was low, the estimated concentration was primarily driven by the number of shedders present in the pool. As the prevalence of infection increased (for example from Cryptosporidium to Norovirus), and hence the likelihood of one or more shedders increased, the modelled concentration was then driven by the pathogen density in faeces.

The estimated decrease in average pathogen concentration over time, and subsequent Log 10 reduction in the main swimming pool is illustrated for Norovirus in Figure 9 (Campylobacter and Cryptosporidium are illustrated in Figures S.2 and S.3) . The lower 2.5% and upper 97.5% quantiles of the Monte Carlo simulation are illustrated with dashed lines around the solid median line. In all cases, the maximum Log 10 reduction that could be achieved over a 24 h period in the main pool, both with and without the application of UV disinfection, was around 1 Log 10 . In each case the achievable Log 10 reduction was limited by the flow rate through the external filtration/treatment system. The dashed lines (95% quantile interval of the Monte Carlo sample) represent the influence that the uncertainty in assumed pathogen removal performance by treatment had on the overall estimated reduction in pathogen concentration, and hence the achieved Log 10 reduction. This uncertainty was highest for viruses, since a low reduction was assumed across treatment barriers, and the plausible range included in the stochastic analysis was broad. However, the addition of UV disinfection eliminated the impact of this uncertainty on estimated pathogen reduction in all cases, clearly demonstrating the limiting factor of system return flow rate and pool water dilution. For the Kiddie pool, a higher overall Log 10 reduction could be achieved within the kiddie pool, in comparison to the main pool, due to the decrease in turnover time (4 h in comparison to 11 h for the main pool) for the system flow rates (illustrated in Figures S.4 ,S.5 and S.6) The maximum achievable reduction in concentration with UV disinfection was more than 3.5 Log 10 for all pathogens (with the maximum flow rate) over 24 hours.

The relationship between turnover time and achievable Log 10 reduction for each reference pathogen is illustrated in Figure 9 . Reducing the turnover time increases the achievable Log 10 reduction. Nevertheless, achieving greater than 3 Log 10 reduction in concentration over 8 h, regardless of treatment barriers, would require an hourly turnover time.

Aside from the unintentional faecal wash-off assumed by normal bathing, accidental faecal releases (AFR) or vomitus releases can and do occur in public swimming pools. This situation was modelled as an event that could occur in one of three locations: the main pool, the kiddie pool or the floor nozzles ( Figure 2 ). The full results for event analysis are included in the Supplementary Material. Highest risks and hence greatest removal performance was required for Norovirus, followed by Campylobacter. Even under the modelled event scenarios, the risks from Cryptosporidium were calculated to be low. When the release occurred within the kiddie pool or the main pool, the recovery of the system was slow within that pool. UV disinfection did not improve this recovery period. Nevertheless, UV disinfection reduced or prevented the risks being carried across the entire system. For example for Norovirus risks following an event in the kiddie pool ( Figure S.8) , UV disinfection reduced the peak reduction requirements in the main pool from over 4 Log 10 to below 2 Log 10 , and the risk at the floor nozzles was reduced from requiring around 5 Log 10 to just over 2 Log 10 . For Campylobacter, for a pathogen release in the main pool (Event 3), the inclusion of UV disinfection totally mitigated the risk in the kiddie pool or at the floor nozzle ( Figure S.13 ).

It is well established that the prime source of pathogen risks in artificial swimming pools are the swimmers, who unintentionally or accidentally (due to illness) shed pathogens into the water column. Hence, we estimated the likely level of pathogen contamination of the Edmonton natural swimming pool under normal operation and during potential event situations to explore how events and which pathogen groups may drive risk. Therefore, risks were summed on a Log scale and the combined risk of illness was driven by the highest risk group (enteric viruses).

The approach applied for normal operation was based on the number of visitors to the pool and relied on assumptions regarding the incidence of infection in Alberta, and estimates of faecal and pathogen shedding from the literature. The estimated concentration was variable and uncertain, given the extreme ranges of values reported in the literature (for example shedding of noroviruses was estimated to range from 10 1 to 10 7 copies.g -1 ). Furthermore, our prevalence calculations were likely to be an overestimate given that we assumed that all people would be equally likely to attend the pool, regardless of their health status or whether they were recovering from an infection. It is unlikely that an individual who was actively ill, would desire to attend the NSP, yet asymptomatic carriers would. In the absence of quantitative estimates on this, and recognizing that asymptomatic and extended shedding of pathogens (over weeks post illness) has been widely reported in the literature, it seemed wise to be cautious with the prevalence estimates and include all cases. It is important to note that the variability in the number of shedders is a critical driver of the pathogen risk, and additional attention may be needed during periods of known higher community infection rates (e.g. during an outbreak). This points to the importance of public education programs that encourage those who are or who have been unwell, to avoid swimming (as signposted at the Edmonton NSP). A message perhaps made easier given the COVID-19 pandemic.

When shedders were present and engaged in water activities the pathogen concentration and hence treatment requirements to maintain safety of other swimmers was estimated.

Risk from Cryptosporidium were consistently low in the calculations, however a reduction of up to 2 Log 10 and 4 Log 10 were estimated to be required within the main pool for The Alberta Health swimming pool standards (Alberta Health 2014) stipulate that a traditional chlorinated swimming pool constructed after November 2006 must have a turnover time of four hours. While the turnover time for a recirculating water spray park is lower at two hours, the standards also state that if 'a wading pool or recirculating water spray park is connected to a swimming pool, the turnover time for the swimming pool shall apply to the wading pool or water spray park'. Under the assumptions of the presented model, a four hour turnover time would achieve less than 1 Log 10 reduction for all reference pathogens. Recently the provincial agency undertaking recreational water quality testing (ProvLab) moved to qPCR testing for Enterococcus spp. to indicate potential faecal contamination along with molecular testing for a human faecal marker (HF183) to detect faecal loadings, which are used on a daily basis at the Edmonton NSP to inform management.

Given that the full-scale efficiency of the external filtration system is limited by turnover time, lack of knowledge regarding the impact of zooplankton grazing on viable pathogen numbers in the water column is a critical data gap in assessing the safety of the system.

Several studies have highlighted that pathogens are internalized by zooplankton in freshwater systems (Burnet et al. 2017 , Connelly et al. 2007 , Hahn and Höfle 2001 . The question however remains as to whether internalization leads to permanent removal/inactivation of the pathogen, or whether pathogens are actually protected and their survival enhanced by internalization (Bichai et al. 2014 , Bichai et al. 2008 , Neogi et al. 2014 ). Therefore, not only is an approach for quantifying the internalization rate of pathogens by zooplankton under full scale conditions needed, but also an understanding of the ultimate fate of internalized pathogens and what health risk they may pose to bathers.

In addition, other internal inactivation processes including predation (considering now the entire microbiome/biofilm rather than zooplankton alone) and sunlight disinfection may be important health protection mechanisms. Unfortunately, the role of the natural microbiome in inactivation of pathogens is poorly understood, and at this stage not quantifiable. Investigation of the microbial inactivation within the site-specific water matrix of the NSP would be of great value.

While data does exist on sunlight inactivation of pathogens in fresh water (Bolton et al. 2010 , Dahl et al. 2017 , Davies-Colley et al. 1999 , Fujioka and Yoneyama 2002 , Mendez-Hermida et al. 2005 , Silverman et al. 2015 , Sinton et al. 2007 , Sinton et al. 2002 , the relevance of these data to the NSP situation is uncertain. Sunlight is expected to be effective at the very surface, however the extent that solar radiation can penetrate the water column is variable and likely to be affected by site specific water quality and resuspension of particulates due to bather activities. In-situ measurement of solar radiation during pool operation, at appropriate depths, would facilitate incorporation of sunlight disinfection kinetics from literature values.

To improve the safety of the system, UV disinfection was a design suggestion for the pool water return lines. However, the inclusion of UV had limited benefit on overall performance during normal operation, but could reduce the spread of infectious pathogens between pools following a faecal accident. The UV disinfection units are considered to be particularly valuable on the outflow of the floor nozzles, given the higher likelihood of faecal contamination by children in this zone, and the need to protect the larger pools from this contamination. The Log 10 reduction value assumed for viruses was based on calicivirus results reviewed by Hijnen et al. (2006) representative for human noroviruses (2.6 Log 10 ), but not as conservative as expected from some human adenovirus (0.5 Log 10 ). Hence the protection provided by UV disinfection may be less for some human enteric viruses than shown by these preliminary modelling results for Norovirus.

The predicted performance of the UV units was based on a review of published laboratory studies for a range of different pathogens (Hijnen et al. 2006 ).There are two important limitations linked with using this laboratory data. Firstly, the performance of UV disinfection in natural waters will be suppressed by the variable organic content of the water. Validation of the UV units would be required to ensure that the required effective UV dose is achieved using an approach similar to that recommended for drinking water treatment (USEPA 2006b) Secondly, the impact of UV disinfection on the natural biome of the water column and hence the natural elimination efficiency of the overall system is poorly understood.

Disrupting the natural biome through UV disinfection must be undertaken with caution.

Overall, in situ challenge testing trials are recommended once the system is operational to investigate the impact of the UV disinfection on microbial survival.

A Screening Level Risk Assessment is a valuable starting point for assessing waterborne risks from enteric pathogens, and to identify risk drivers and research needs. Even with the simplistic box flow model applied to the NSP described, the following important lessons regarding the likely performance of the system were identified:

 Enteric pathogen risks associated with natural swimming pools depend upon how many swimmers are infected. Modelling pathogen concentration needs to account for the likelihood of one or more shedders being present;

 The overall performance of the filtration system was driven by the system water turnover time, and using the 11 and 4 h designed turnover time would achieve less than 1 Log 10 reduction for all reference pathogens evaluated;

 Natural disinfection mechanisms for NSP are poorly understood, making reliance upon them for health protection in public pools challenging. Specific data is needed to better describe the fate of pathogens in natural waters; and  Performance of natural filtration barriers in the NSP environment are poorly understood and challenge studies of in situ systems are recommended to reduce uncertainties in the current study. Under reporting factor 27.2 (Thomas, Murray et al. 2013) 288 (Tam, Rodrigues et al. 2012) 48.5 (Thomas, Murray et al. 2013) Mean duration of 21 (Havelaar, van 28.5 (Tu, McCaig et al. 

Volume of water (unintentionally) ingested 10 mL (Dufour, Behymer et al. 2017) 38 mL (Dufour, Behymer et al. 2017 )

Norovirus Cryptosporidium Dose-response model: Exact Beta-Poisson Parameters α=0.024; β=0.011 (Teunis, van den Brandhof et al. 2005) α=0.0044; β=0.0022 (Messner, Berger et al. 2014) α=0.115; β=0.176 (Teunis, Chappell et al. 2002) Probability of illness given infection P ill 0.2 (Black, Levine et al. 1988 

Modeling seasonal variation in bacteriophage removal in constructed wetlands using convection-dispersion equation

Cladoceran and rotifer grazing on bacteria and phytoplankton in two shallow eutrophic lakes: in situ measurement with fluorescent microspheres

Pool Standards, Health System Accountability and Performance

Microbial contamination of drinking water and human health from community water systems

Quantification of pathogens and sources of microbial indicators for QMRA in recreational waters

Evaluation of various Campylobacter-specific quantitative PCR (qPCR) assays for detection and enumeration of Campylobacteraceae in irrigation water and wastewater via a miniaturized most-probable-number-qPCR assay

The risk of contracting infectious diseases in public swimming pools: a review

Posttreatment of UASB reactor effluent in waste stabilization ponds and in horizontal flow constructed wetlands: a comparative study in pilot scale in Southeast Brazil

Role of predation by zooplankton in transport and fate of protozoan (oo) cysts in granular activated carbon filtration

Predation and transport of persistent pathogens in GAC and slow sand filters: A threat to drinking water safety?

Protection of waterborne pathogens by higher organisms in drinking water: a review

Experimental Campylobacter jejuni infection in humans

Shigellosis from swimming in a park pond in Michigan

A review of the factors affecting sunlight inactivation of micro-organisms in waste stabilisation ponds: preliminary results for enterococci

Hygienic quality of public natural swimming pools (NSP)

How does the cladoceran Daphnia pulex affect the fate of Escherichia coli in water

Characterization of microbial populations associated with natural swimming pools

Waterborne outbreaks of cryptosporidiosis

Impact of zooplankton grazing on the excystation, viability, and infectivity of the protozoan pathogens Cryptosporidium parvum and Giardia lamblia

A practical model for sunlight disinfection of a subtropical maturation pond

Reported waterborne outbreaks of gastrointestinal disease in Australia are predominantly associated with recreational exposure

Inactivation of faecal indicator micro-organisms in waste stabilisation ponds: Interactions of environmental factors with sunlight

Ingestion of swimming pool water by recreational swimmers

Estimating the economic burden from illnesses associated with recreational coastal water pollution-a case study in Orange County

Quantitative evaluation of enterococci and Bacteroidales released by adults and toddlers in marine water

Quantitative evaluation of bacteria released by bathers in a marine water

Recreational Water Quality Criteria

Recreational water and infection: a review of recent findings

Interactions Between Human Reovirus and Free-Living Amoebae: Implications for enteric virus disinfection and aquatic persistence

Sunlight inactivation of human enteric viruses and fecal bacteria

Contaminant removal processes in subsurface-flow constructed wetlands: a review

Assessment of enteric pathogen shedding by bathers during recreational activity and its impact on water quality

Optimization of artificial wetland design for removal of indicator microorganisms and pathogenic protozoa

Outbreaks associated with untreated recreational water-United States

Campylobacter in recreational lake water in southern Quebec, Canada: presence, concentration, and association with precipitation and ruminant farm proximity

Grazing of protozoa and its effect on populations of aquatic bacteria

Cryptosporidium and Giardia in wastewater and surface water environments

Immunity to Campylobacter: its role in risk assessment and epidemiology

Inactivation credit of UV radiation for viruses, bacteria and protozoan (oo)cysts in water: a review

Viruses in wetland ecosystems

Landscaping and Landscape Development Research Society (2011) Recommendations for planning, construction, servicing and operating of outdoor swimming pools with biological water purification (swimming and bathing ponds)

Natural swimming pools: an inspirational guide for construction and maintenance

Effect of batch-process solar disinfection on survival of Cryptosporidium parvum oocysts in drinking water

Fractional Poisson-A Simple Dose-Response Model for Human Norovirus

The role of wetland microinvertebrates in spreading human diseases

Swimming-associated outbreak of Escherichia coli O157 [ratio] H7

Water recreation and disease: plausibility of associated infections: acute effects, sequelae, and mortality, World Health Organization

Assessment of human virus removal during municipal wastewater treatment in

Efficiency of natural systems for removal of bacteria and pathogenic parasites from wastewater

Sunlight inactivation of viruses in open-water unit process treatment wetlands: modeling endogenous and exogenous inactivation rates

Viruses in recreational water-borne disease outbreaks: A review

Sunlight inactivation of Campylobacter jejuni and Salmonella enterica, compared with Escherichia coli, in seawater and river water

Sunlight inactivation of fecal indicator bacteria and bacteriophages from waste stabilization pond effluent in fresh and saline waters

Shedding of oocysts in immunocompetent individuals infected with Cryptosporidium

Retention and removal of pathogenic bacteria in wastewater percolating through porous media: a review

Predation of Cryptosporidium oocysts by protozoa and rotifers: implications for water quality and public health

Cryptosporidium risk from swimming pool exposures

Longitudinal study of infectious intestinal disease in the UK (IID2 study): incidence in the community and presenting to general practice

Zooplankton and aggregates as refuge for aquatic bacteria: protection from UV, heat and ozone stresses used for water treatment

A reconsideration of Campylobacter dose-response relation

Cryptosporidium dose response studies: variation between isolates

Norwalk Virus: How Infectious is It?

Estimates of the burden of foodborne illness in Canada for 30 specified pathogens and unspecified agents, circa

Rotifers ingest Giardia cysts

Norovirus excretion in an aged-care setting

Economic analysis for the final long term 2 enhanced surface water treatment rule

Ultraviolet disinfection guidance manual for the final long term 2 enhanced surface water treatment rule, United States Environmental Protection Agency

Pathogen Removal in Natural Wastewater Treatment and Resource Recovery Systems: Solutions for Small Cities in an Urbanizing World

Transport and removal of coliphage PRD1 in constructed wetlands

Virus removal from wastewater in a multispecies subsurface-flow constructed wetland

Risk analysis: a quantitative guide

Removal of enteric bacteria in constructed treatment wetlands with emergent macrophytes: a review

Quantitative Microbial Risk Assessment for Water Safety Management, World Health Organization

Reactivity of enveloped virus genome, proteins, and lipids with free chlorine and UV254

Best estimate of elimination capacity (log 10 reduction) (