This study assesses the infection risks associated with non-swimming recreational activities in Marina Reservoir, Singapore. Apart from serving as a reservoir for potable water use, Marina Reservoir is also a place where various water recreations are carried out. Quantitative microbial risk assessment (QMRA) was applied to Marina Reservoir and its four main feeders. QMRA was performed based on the occurrence data of the potential pathogen, Salmonella spp. and bacteria indicator, Enterococcus spp. through Monte Carlo simulation techniques. The results suggest that microbial risks from Salmonella and Enterococcus at the reservoir were below the United States Environmental Protection Agency (USEPA) acceptable freshwater recreational illness rate of 0.8%. All upstream catchments showed acceptable microbial risks from Salmonella. However, the probability of gastrointestinal illness risks calculated for Enterococcus exceeded the USEPA guideline values in one of the upstream catchments, but far enough upstream from recreational areas that are used frequently.

INTRODUCTION

Singapore is an island country located in Southeast Asia. In recent years, Singapore has rapidly expanded the use of water catchments and reservoirs for recreational purposes. Marina Reservoir, which is located in the Central Business District, has become one of the main recreational reservoirs in Singapore. It is used for activities like canoeing, kayaking and boating. As such, the microbial water quality is important to protect users' health. Routine water quality monitoring was conducted in the reservoir and its feeders from December 2011 to July 2013. The current recreational water quality standard in Singapore for Enterococcus is 200 counts/100 mL (National Environmental Agency, Singapore) (NEA, 2008). However, the question as to how significant the health risk to humans is always the main concern to recreational water users. Quantitative microbial risk assessment (QMRA) is a theoretical approach that can be used to estimate the risk to human health, by predicting infection or illness rates for given densities of particular pathogens, assumed rates of ingestion and appropriate dose–response models for the exposed population (Haas et al. 1999). In this study, QMRA was conducted based on the knowledge of the presence of the potential pathogen, Salmonella and the bacteria indicator, Enterococcus in the reservoir.

METHODS

Sample collection and bacteria detection

Water samples were collected from Marina Reservoir (A) and four upstream catchments (B, C, D and E) (Figure 1). Monthly sampling was carried out at all five sampling sites between December 2011 and July 2013.

Figure 1

Marina Reservoir catchment plan (Daryl, 2008) (a) and sampling sites (b).

Figure 1

Marina Reservoir catchment plan (Daryl, 2008) (a) and sampling sites (b).

Salmonella were isolated and enumerated using a modified version of Rajabi's method (Rajabi et al. 2011). The samples (200, 50, 5 and 0.5 mL) were used to determine the most probable number (MPN) per 100 mL of Salmonella present by addition to equal volumes of sterile 1% buffered peptone water in triplicate. Broth cultures were incubated at 37 °C overnight with shaking. These cultures (1 mL of each) were subsequently transferred into 9 mL of Salmonella enrichment medium for selective enrichment at 37 °C overnight with shaking. The broth cultures were then streaked for isolation onto Salmonella Shigella Agar (Becton Dickinson & Co, New Jersey U.S.) plates and incubated at 37 °C overnight. Salmonella-positive samples were further confirmed by Salmonella genus-specific polymerase chain reaction.

Enterococcus were detected and enumerated using Enterolert™ (IDEXX Laboratories, Westbrook, Maine) and reported as MPN/100 mL.

QMRA

The probability distribution function curve of Salmonella and Enterococcus at each station was generated on the basis of the monthly sampling data. As the data obtained in this study gave a poor fit to the known distribution type (i.e. normal, lognormal or uniform), distribution function curves were generated based on empirical data distribution. Random samplings under the distribution function curve were conducted using MATLAB with the emprand function (Shrestha 2007). To account for ingestion of water through splashing, the ingestion rate used was proportional to the ingestion rate for 10 minutes' swimming duration obtained from Dufour et al.'s study (2006). The ingestion volume was randomly sampled between the maximum and minimum ingested, for both adults and children. The risk of infection from Salmonella was estimated by the β-Poisson model (Haas et al. 1999), and the exponential dose–response model developed by Haas et al. (1999) and Stone et al. (2008) was used to estimate the risk of gastrointestinal illness arising from Enterococcus. Monte Carlo simulation was performed using MATLAB (The MathWorks, Inc.).

RESULTS AND DISCUSSION

The cell concentration of Salmonella and Enterococcus in the reservoir and its feeders are summarized in Table 1. Station B has the highest Salmonella and Enterococcus cell counts, with geometric means of 28.64 MPN/100 mL and 1253.46 MPN/100 mL, respectively. Meanwhile, Station A has the lowest counts of both Salmonella and Enterococcus (5.95 MPN/100 mL and 5.42 MPN/100 mL, respectively). Based on the current recreational water guidelines in Singapore, the geometric mean Enterococcus cell concentrations at all stations were below the guideline value of 200 counts/100 mL, except Station B.

Table 1

Summary of Salmonella and Enterococcus cell concentrations at the sampling locations

  Salmonella cell concentration (MPN/100 mL)
 
Enterococcus cell concentration (MPN/100 mL)
 
Station Min. Max. Geometric mean Min. Max. Geometric mean 
31.2 5.95 36.9 5.42 
10 91.1 28.64 31.3 26234 1253.46 
1.7 90.1 11.11 16 980 72.86 
1.7 90.1 8.85 365.4 34.47 
5.7 91.1 27.65 12.2 2419.6 101.48 
  Salmonella cell concentration (MPN/100 mL)
 
Enterococcus cell concentration (MPN/100 mL)
 
Station Min. Max. Geometric mean Min. Max. Geometric mean 
31.2 5.95 36.9 5.42 
10 91.1 28.64 31.3 26234 1253.46 
1.7 90.1 11.11 16 980 72.86 
1.7 90.1 8.85 365.4 34.47 
5.7 91.1 27.65 12.2 2419.6 101.48 

The Monte Carlo simulation outputs of cumulative probability risk of illness per event for adults and children from Salmonella and Enterococcus are shown in Figures 2 and 3. Marina Reservoir (A) showed the lowest microbial risk among the sampling stations. As there is no epidemiological study carried out in Singapore's recreational waters up to date, the acceptable illness risk to recreational water users proposed by USEPA was adopted in this study. After 10,000 iterations, the QMRA suggested that the microbial risk from Salmonella at all stations was well below the USEPA acceptable freshwater recreational illness rate of 0.8% (Figure 2). The probability of illness was found to be more diverse at different stations when examined with Enterococcus rather than Salmonella (Figure 3). Station B was found to exceed the acceptable level for both adults and children based on the Enterococcus model.

Figure 2

Cumulative probability risk of illness per event for adults (a) and children (b) based on the Salmonella model.

Figure 2

Cumulative probability risk of illness per event for adults (a) and children (b) based on the Salmonella model.

Figure 3

Cumulative probability risk of illness per event for adults (a) and children (b) based on the Enterococcus model.

Figure 3

Cumulative probability risk of illness per event for adults (a) and children (b) based on the Enterococcus model.

CONCLUSIONS

The QMRA results suggest that the illness risks at Marina Reservoir were below the USEPA acceptable level. However, the upstream catchment, Station B, showed illness risks that exceeded the guideline value based on the Enterococcus model. The location of the sampling point in the catchment is sufficiently far upstream – i.e., away from areas where recreational activities frequently take place – for this not to be a problem.

The estimated risks based on the Salmonella model were significantly lower than those derived from the Enterococcus model. This suggests that Salmonella may not be a cause of concern for the Marina Reservoir. The large discrepancy between the Enterococcus and Salmonella QMRA models shows the need to choose a better indicator or other potential pathogens (i.e., enteric viruses) to evaluate the health risks in the reservoir. This work is in progress.

ACKNOWLEDGEMENT

This research was funded by Singapore National Research Foundation under its Environmental & Water Technologies Strategic Research Programme and administered by the Environment & Water Industry Programme Office (EWI) of the Public Utilities Board (PUB; Ref: 1002-IRIS-32 [IDD 90301/1/24]). We would like to thank National University of Singapore, National Environmental Research Institute (NERI) and PUB for supporting this research.

REFERENCES

REFERENCES
Daryl
L. R.
2008
Sediment and Organic Carbon Variability in a Tropical Urban River System. M.A. Thesis
.
National University of Singapore
,
Singapore
.
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
4
(
4
),
425
430
.
Haas
C. N.
Rose
J. B.
Gerba
C. P.
1999
Quantitative Microbial Risk Assessment
.
John Wiley & Sons, Inc.
,
New York
,
United States
.
NEA
2008
Recreational water quality.
Retrieved from National Environment Agency
,
Singapore
. .
Rajabi
M.
Jones
M.
Hubbard
M.
Rodrick
G.
Wright
A. C.
2011
Distribution and genetic diversity of Salmonella enterica in the Upper Suwannee River
.
International Journal of Microbiology
201
,
1
9
.
Shrestha
D. L.
2007
Random number from empirical distribution
.
Stone
D.
Harding
A.
Hope
B.
Slaughter-Mason
S.
2008
Exposure assessment and risk of gastrointestinal illness among surfers
.
Journal of Toxicology and Environmental Health Part A
71
(
24
),
1603
1615
.