Article pubs.acs.org/est
Decay of Coliphages in Sewage-Contaminated Freshwater: Uncertainty and Seasonal Effects Jianyong Wu,*,† Yiping Cao,‡ Brianna Young,† Yvonne Yuen,† Sharon Jiang,† Daira Melendez,† John F. Griffith,‡ and Jill R. Stewart† †
Department of Environmental Sciences and Engineering, Gillings School of Global Public Health, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599, United States ‡ Southern California Coastal Water Research Project Authority, Costa Mesa, California 92626, United States S Supporting Information *
ABSTRACT: Understanding the fate of enteric viruses in water is vital for protection of water quality. However, the decay of enteric viruses is not well characterized, and its uncertainty has not been examined yet. In this study, the decay of coliphages, an indicator for enteric viruses, was investigated in situ under both sunlit and shaded conditions as well as in summer and winter. The decay rates of coliphages and their uncertainties were analyzed using a Bayesian approach. The results from the summer experiments revealed that the decay rates of somatic coliphages were significantly higher in sunlight (1.29 ± 0.06 day−1) than in shade (0.96 ± 0.04 day−1), but the decay rates of male-specific (F+) coliphages were not significantly different between sunlight (1.09 ± 0.09 day−1) and shaded treatments (1.11 ± 0.08 day−1). The decay rates of both F+ coliphages (0.25 ± 0.02 day−1) and somatic coliphages (0.12 ± 0.01 day−1) in winter were considerably lower than those in summer. Temperature and chlorophyll a (chla) concentration varied significantly (p < 0.001) between the two seasons, suggesting that these parameters might be important contributors to the seasonal variation of coliphage decay. Additionally, the Bayesian approach provided full distributions of decay rates and reduced the uncertainty, offering useful information for comparing decay rates under different conditions.
■
INTRODUCTION Fecal indicator bacteria (FIB) such as Escherichia coli and fecal coliforms are routinely monitored in freshwater as surrogates for human pathogens and are the basis for public health protection. However, many studies have shown that FIB and human enteric viruses have different fates and transport in surface water. FIB are also more susceptible to waste treatment processes than human enteric viruses, rendering FIB inadequate as indicators of viruses.1−5 Coliphages have been proposed as an alternative indicator because they are morphologically and functionally more similar to human enteric viruses than FIB.6 Coliphages and other viruses have similar persistence in the environment, and their presence has been correlated with viral pathogens in water.3,7−10 Furthermore, several epidemiological studies have shown significant correlations between coliphage concentrations and risk of swimming related illness, making coliphages a promising indicator for water quality monitoring.11−13 Currently, somatic and male-specific (F+) coliphages are two common groups of bacteriophages suggested as indicators for fecal contamination. Somatic coliphages are DNA bacteriophages that infect coliform bacteria through their outer membrane, while F+ coliphages are a group of bacteriophages that infect Gram-negative bacteria (including E. coli) via the pilus.6 © XXXX American Chemical Society
To estimate risks of viruses to human health and make informed management decisions, it is crucial to understand the fate and transport of viruses in ambient water. One of the most important factors that affect the fate of viruses in water is sunlight.6,14−16 Studies have revealed that sunlight can significantly increase the inactivation rates of FIB, bacteriophages and human viruses in freshwater and seawater,14,15,17−22 demonstrating that solar inactivation is an important factor for the natural reduction of fecal contaminants in water. Temperature is another important factor that influences the decay of both enteric viruses and coliphages, as it was reported that these viruses decay faster at higher water temperatures.23,24 A meta-analysis showed that the inactivation of bacteriophages and enteric viruses in water is more rapid at temperatures above 50 °C, and that the inactivation of these viruses is more sensitive to temperature in simple matrices than in complex matrices.25 Because temperature may be correlated with solar radiation,26 sunlight and temperature may have synergistic effects on the decay of coliphages. For example, temperature Received: August 4, 2016 Revised: October 3, 2016 Accepted: October 6, 2016
A
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
attenuation in SJM before returning to the Creek to continue discharge into the Upper Newport Bay and the ocean. In Situ Microcosms. The decay experiments were conducted in duplicate in two seasons, during summer (from August 18 to 28, 2014) and winter (from January 9 to 19, 2015). Dialysis bags (6−8 kDa, Spectra/Por 4, SpectrumLabs, Rancho Dominguez, CA) containing sewage (5% v/v) seeded into unaltered ambient water from the field site were suspended from PVC frames placed in ambient water so that the bags floated at approximately 15 cm below the surface of water. Each set of seeded bags was also accompanied by two control bags containing only ambient water. All bags contained 1 L of sample or control water. Sewage used for seeding was primary influent from the Orange County Sanitation District (Costa Mesa, CA) collected the night before the field experiment and stored in 4 °C before use. Ambient water was collected from SJM and large debris such as leaves were removed. In the summer, half of the PVC frames were covered with shade cloth so that one set of bags was exposed to direct natural sunlight while the other set was in the shade. In the winter, no shade cloth was used and all bags were exposed to natural sunlight. Duplicate bags from each treatment were retrieved daily and processed for E. coli, total coliforms and F+ and somatic coliphages. On day 7 and day 10, one control bag was retrieved for each treatment and processed the same as the sample bags. Sample Processing. Dialysis bags were retrieved and placed in a capped bucket filled with ambient water for transport from the site to the laboratory. Sample water from inside the bags was shaken gently before pouring the sample into an acid-cleaned bottle (regular wash followed by 10% HCl for at least 15 min, then rinsed three times with DNA-free water). Duplicate samples were collected each day around 8 am over 11 days in both summer (August 18- 28, 2014) and winter (January 9−19, 2015). In whole, 42 samples in summer and 22 samples in winter were collected for testing coliphages (Figure S1). For testing physicochemical water quality variables and indicator bacteria (total coliforms and E. coli), water samples were processed within 6 h of sample collection. For coliphage analysis, water samples were shipped overnight on ice on weekdays or stored at 4 °C over weekends until shipment on ice on the following Monday to the University of North Carolina at Chapel Hill. The samples were normally received around 10:30 am and analyzed for coliphages at noon (12:00 pm in the Eastern Standard Time, or 9:00 am in the Pacific Standard Time). Total holding time was therefore about 25 h. For the sample collected on Friday, coliphages were detected at the noon of the following Monday, with the total holding time of about 73 h. Water Quality and Meteorological Data Collection. Water quality parameters including temperature, specific conductivity, salinity, chlorophyll a (chla) and dissolved oxygen (DO) were measured with a YSI 6600 sonde (YSI Inc., Yellow Springs, OH) in the field. Turbidity was measured in the lab using a DRT-15CE turbidimeter (HF Scientific Inc., Fort Myers, FL). Chla was measured because it is an indicator of algae biomass. Nonpurgeable organic carbon (NPOC) data were obtained from literature.38 Water pH varied little for both periods, ranging around from 7.2 to 7.6. Weather conditions in the local area were also recorded in both experimental periods, including air temperature, rainfall (precipitation), relative humidity and visibility. The measurement of solar radiation and the estimation of light intensity in water column were
was found to play an important role in sunlight-mediated inactivation of F+ coliphage MS2.20 Currently, little is known about the decay of enteric viruses in sewage-contaminated water because there are many different types of enteric viruses and most are difficult to culture.27,28 A few studies have used coliphages as a surrogate to understand decay rates of enteric viruses under exposure to sunlight. For example, Sinton et al, reported sunlight inactivation rates ranging from 0.084 to 0.246 per hour for sewage-associated somatic coliphages in freshwater, simulated estuarine water and seawater. The inactivation rates for F+RNA coliphages in these types of samples ranged from 0.065 to 0.278 per hour.18 To date, seasonal effects on the decay rates of coliphages in sewage-contaminated freshwater have not been well characterized. Since temperature and other factors vary at different seasons, it is expected that the decay rates of coliphages may also vary seasonally. Overall, the decay of coliphages is still poorly characterized and subject to uncertainty. The uncertainty may come from two primary sources: one is the natural variability resulting from the effects of environmental factors, such as sunlight and temperature, and the other is the measurement uncertainty during laboratory determination of coliphages in water samples. A conventional method to calculate decay rates of microorganisms is based on the Chick-Watson model (as shown in eq 1 below), which is also used to model first-order reactions in chemical kinetics.29−31 In this model, the first-order decay rate is commonly assumed to be a constant. The Bayesian approach, a method of statistical inference based on Bayes’ rule,32−34 however, has become increasingly popular for incorporating uncertainty in modeling because this approach takes advantage of prior information and measured data to make statistical inferences, which can improve the estimation.33,34 For microbial water quality studies, the Bayesian approach is particularly useful because the distribution of microbes is not uniform in water and because microbes may be present at concentrations below the detection limit for a given detection method, which may lead to measurement uncertainty when estimating microbial concentrations.35−37 Due to this uncertainty, a Bayesian approach has been proposed to estimate the first-order decay rate of fecal indicator bacteria and to compare the effects of parameter estimation on microbial water quality models.35 However, this approach has not been previously applied to address the uncertainty of decay rates of coliphages, nor has it been used to evaluate the effects of sunlight and seasonal variation on coliphage decay. The main objectives of this study are to (1) estimate the decay rate of coliphages with Bayesian statistics considering the uncertainty attributed to measurement errors; (2) examine the variability of coliphage decay caused by sunlight and seasonal effects; and (3) compare the decay rates of coliphages with that of indicator bacteria (E. coli and total coliforms).
■
MATERIALS AND METHODS Field Site. The field experiments were conducted inside the San Joaquin Marsh (SJM, 33° 39′ 57.9″ N, 117° 50′ 46.8″ W), which is one of Irvine Ranch Water District’s natural treatment systems. This site consists of a series of ponds connected by channels and receives water directly from the adjacent San Diego Creek. The San Diego Creek is a typical freshwater creek of southern California. The creek flows through the landscape of the city of Irvine and mostly receives urban runoff from the surrounding land. The creek water is treated only by natural B
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology described in detail in a previous study.38 Briefly, the Simple Model of the Atmospheric Radiative Transfer of Sunshine (SMARTS) was used to estimate the intensity of incident UVB on the water surface in 30 min intervals for the middle day of the experimental period. The intensity of the transmitted UVB was calculated using the membrane transmittance, site-specific water absorbance, and the solar zenith angle according to Fresnel’s equations.38 Enumeration of Indicator Bacteria and Coliphages. Water samples collected in summer were analyzed for total coliforms and E. coli by Colilert-18 (IDEXX Laboratories, Inc., Westbrook, Maine, USA) at appropriate dilutions. The standard single agar layer plaque assay method was used to quantify F+ and somatic coliphages following EPA Method 1602.39 Laboratory prototype bacteriophage strains MS2 (ATCC 15597-B1) and PhiX174 were spiked in sterile deionized water as positive controls for F+ and somatic coliphages, respectively. Sterile deionized water was used as the negative control. Decay Rate Calculation with the Convention Method. The decay rates of F+ and somatic coliphages, total coliforms and E. coli were first calculated using the conventional firstorder decay equation:30,40,41 C = C0e−kt or ln C = ln C0 − kt
(1)
C = C0e−k(t − t0)or ln C = ln C0 − k(t − t0)
(2)
(3)
λi = CiVi
(4)
(5)
ln Ci = ln C0 − k(ti − t0) + ϵi
(6)
ϵi ∼ Norm(0, σ )
(7)
k ∼ Unif(a , b)
(8)
Where yi is the count of coliphages in a sample, C is the concentration of coliphages in the sample, V is the volume of the sample, i is the index of time (in days), and ϵ is the residual error term of ln C (the concentration of coliphages in logtransformed format). Pois, Norm, and Unif denote Poisson, normal and uniform distributions, respectively. σ is the standard deviation, a parameter of ϵ following a normal distribution, and a (the minimum value) and b (the maximum value) are the parameters of a uniform distribution of k. The distributions and values of these parameters were assigned based on the literature35 and assessed by sensitivity analysis described below. For the decay without a lag period, eq 5 was used, while eq 6 was chosen if the decay had a lag period. Similar models were constructed to estimate the decay rates of total coliforms and E. coli. A free Bayesian inference package, OpenBUGS version 3.2.3, was used to encode the model. Gibbs sampling, a Markov Chain Monte Carlo (MCMC) algorithm, was implemented to sample from the posterior distributions of the decay rates. A single MCMC chain was simulated for each parameter for 5000 iterations, of which the first 2000 iterations were abandoned as a “burn-in”. The remaining iterations were thinned in a ratio of one to three. As a result, a total of 1000 samples were left to represent the posterior distributions of the decay rates. The convergence of the MCMC iteration was diagnosed by monitoring the trace plots and autocorrelation plots of these parameters. The OpenBUGS code along with the explanation and the data, are included in the Supporting Information. Uncertainty and Sensitivity Analysis. In the Bayesian models, an error term ϵ was included in the models to address the uncertainty in estimation of coliphage concentrations. However, the values of three prior parameters: σ, a, b, were unknown, therefore, assigned arbitrarily. To evaluate how the choices of prior parameters influenced model outcomes, uncertainty and sensitivity analyses were conducted following the literature.36 Briefly, each parameter was given three different values from small to large and these values were inputted into the model individually, changing only one parameter for each simulation. The decay rate of somatic coliphages under sunlight treatment was chosen as the model outcome to evaluate the influence of these parameters. The sensitivity of these parameters to the model outcome was illustrated in a series of plots with the decay rate charted against different parameter values, and the uncertainty of the model outcome was described by its mean, standard deviation and 95% credible intervals (CIs). Analysis of Seasonal Effects on Coliphage Decay. A variance decomposition method was used to determine influential factors for the seasonal variation of coliphage decay and apportion the seasonal effects.37,42 For a model y = f(x1, x2, ... xn), the total variance of the dependent variable y can be decomposed into the partial variance attributable to each factor (x1, x2,...,xn) and the interactions of these factors. The factor that contributes to a larger partial variance has a larger influence on the dependent variable.
Where, C is the concentration of coliphages or bacteria at time t, C0 is the initial concentration of coliphages or bacteria at time 0 (day 0), t is the time in days, k is the first-order decay rate constant, and e is a mathematical constant, approximately equal to 2.71828. For our calculations, t0 was the time when decay started, after any measured lag period. In general, eq 1 was used to calculate decay rates. If the decay of coliphages or bacteria had a lag period, a modified formula (eq 2) was used. The lag period is the period when the population of bacteria or coliphages remains relatively stable before the beginning of logarithmic decay,30,40 which was determined based on the plots of the concentrations of coliphages and bacteria against time. According to our experimental data, the decay of somatic coliphages had no lag period. The decay of F+ coliphages and total coliforms lagged 1 and 2 days, respectively. For E. coli, the decay lagged 4 days under shade and lagged 2 days under sunlight. Therefore, the decay rate for somatic coliphages was calculated using eq 1, while the decay rates for other measured microbes were calculated using eq 2. Decay Rate Calculation with Bayesian Statistics. The above eqs (eqs 1 and 2), representing typical decay analyses, do not account for the uncertainty in decay rate calculations. To address this issue, a Bayesian model similar to that described previously35 was constructed to estimate the decay rates for both F+ and somatic coliphages. In this model the count number of coliphages in a sample was assumed to follow a Poisson distribution, the concentration of coliphages was assumed to follow a log-normal distribution, and the decay rate, k, was assumed to have a uniform distribution. The likelihood functions and the prior distributions for key prior parameters in the Bayesian model were specified as below:
yi |λi ∼ Pois(λi)
ln Ci = ln C0 − kti + ϵi
C
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. Decay curves of somatic and F+ coliphages with the shaded treatment and the sunlit treatment.
Table 1. First-Order Decay Rates of Coliphages and Indicator Bacteria in Water Calculated Using Bayesian Statistics and the Conventional Method (Unit: Day−1)a Bayesian method season summer
treatment shade
sunlight
winter a
NA
conventional method
indicators
mean
SD
2.5% quantile
97.5% quantile
Mean
SD
F+ coliphages somatic coliphages E. coli total coliforms F+ coliphages somatic coliphages E. coli total coliforms F+ coliphages somatic coliphage
1.11 0.96 0.55 0.48 1.09 1.29 1.35 0.93 0.25 0.12
0.08 0.04 0.05 0.03 0.09 0.06 0.08 0.08 0.02 0.01
0.97 0.89 0.44 0.42 0.92 1.18 1.21 0.78 0.22 0.09
1.28 1.03 0.65 0.54 1.29 1.43 1.50 1.09 0.29 0.14
1.09 1.04 0.62 0.60 1.16 1.34 1.71 1.35 0.35 0.17
0.30 0.22 0.37 0.47 0.45 0.26 0.78 0.53 0.19 0.10
SD: standard deviation; NA: not available.
■
To determine the contributions of influential factors to the seasonal variation of the decay rates of coliphages, first, we apportioned the seasonal effect to the effect attributable to solar radiation and the remaining seasonal effect attributable to the seasonal factor. Here, the seasonal effects were the combined effects due to the changes in meteorological factors or water quality variables between summer and winter, including solar radiation, temperature, precipitation, turbidity, NPOC, and chla. The seasonal factor was the combination of meteorological factors or water quality variables excluding solar radiation because it was taken as a standalone factor. In the second step, we analyzed which specific factors composing the seasonal factor played an important role in the remaining seasonal effect. Based on literature,6 we narrowed down the factors that might affect coliphage decay. Then we conducted one-way ANOVA to analyze the variation of each of those factors between two seasons, including temperature, turbidity, NPOC, chla, relative humidity and precipitation. A factor without a significant difference between two seasons may have a minor contribution to the remaining seasonal effect (see the Supporting Information for detailed analysis).
RESULTS
Characteristics of Field Water Quality and Weather Conditions. Measures of water temperature, chla concentration, DO concentration and turbidity are presented in the Supporting Information and were higher in the summer than the winter, as expected (Figure S2, Table S2). The initial (day 0) concentrations of F+ and somatic coliphages in the summer were 943 PFU/100 mL and 5.53 × 104 PFU/100 mL, respectively. In the winter decay experiment, the initial (day 0) concentrations of F+ and somatic coliphages were 70 PFU/100 mL and 1.81 × 104 PFU/100 mL, respectively. Weather conditions in both experimental periods are shown in Figure S3 and S4. On average, air temperature, relative humidity and visibility in summer were 22.4 °C, 66% and 10 mi during sampling, respectively, which were higher than that in winter. Precipitation was not observed during the summer experiment but was observed at Day 3 during the winter experiment, totaling 0.11 in. With regard to solar radiation, light intensity based on UVB (280 nm-315 nm) in water at the depth of 15 cm under sunlight treatment (summer), shaded D
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 2. Posterior distributions of the decay rate (k) of coliphages and indicator bacteria in summer.
Figure 3. Posterior distributions of the decay rate (k) of coliphages in winter.
treatment (summer) and in winter were 1.162 × 10−3 W/m2, 2.904 × 10−4 W/m2, and 8.275 × 10−5 W/m2, respectively. One-way ANOVA showed that temperature (both water and air), chla and DO concentrations were significantly different between two seasons (p < 0.001). Turbidity and NPOC concentration were also significantly different between two seasons (p = 0.037 and 0.031, respectively) (Table S2). Two
meteorological variables, relative humidity and precipitation, were not significantly different between two seasons (p > 0.05). Effects of Sunlight on Coliphage Decay Rates during Summer. Raw decay profiles indicated a larger sunlight effect on somatic coliphages but minimal sunlight effect on F+ coliphages (Figure 1). In the shaded treatment, the concentration of F+ coliphages gradually decreased after day 1 and was below the detection limit (1 PFU/100 mL) after day E
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
(1.11 and 1.09 day−1, respectively). The mean decay rates of somatic coliphages obtained with conventional and Bayesian methods were also similar. In the winter experiment, the mean decay rates of F+ and somatic coliphages calculated with the conventional method were much higher than those calculated with Bayesian approach, though it is unknown whether the differences were significant. However, the Bayesian approach yielded smaller standard deviations of decay rates. With the Bayesian method all standard deviations were below 0.1 while standard deviations of the conventional method ranged from 0.10 to 0.78, nearly 4−10 times higher than those obtained by the Bayesian approach. In addition, Bayesian statistics generated the distribution of decay rates as shown in Figure 2, which could not be obtained with the conventional method. Sensitivity Analysis. The effects of the choice of three prior parameters, σ, a, and b, on the estimation of decay rate of somatic coliphages were examined. When σ increased from 1 to 8, the mean decay rate changed little, ranging from 0.96 to 0.98 day−1 (Table 2). Similarly, when b increased from 1 to 8, the
7. The decay curve of F+ coliphages in the sunlight treatment almost overlapped with the curve in the shaded treatment. For somatic coliphages, the concentration decreased from day 1 to day 10 and remained just above the detection limit at day 10. However, in the sunlight treatment, the curve became slightly steeper in contrast to the curve in the shaded treatment and the concentration was below the detection limit after day 7 (Figure 1). This observation of different sunlight effects on F+ and somatic coliphages was confirmed by decay modeling using the Bayesian approach (Table 1, Figure 2). For somatic coliphages, the decay rate increased from 0.96 day−1 in the shade (standard deviation (SD): 0.04 day−1; 95% CI: 0.89−1.03 day−1) to 1.29 day−1 under sunlight (SD: 0.06 day−1; 95% CI: 1.18−1.43 day−1). This increase is significant because the 95% CIs of the decay rate for somatic coliphages in sunlight and shade did not overlap. For F+ coliphages, the mean decay rate in the shade was 1.11 day−1 (SD: 0.08 day−1; 95% CI: 0.97−1.28 day−1), which was not statistically different from the mean decay rate of 1.09 day−1 under sunlight (SD 0.09 day−1; 95% CI: 0.92−1.29 day−1). Comparison of Coliphage Decay Rates between Summer and Winter. The decay rates of coliphages in winter were significantly different from those in summer. According to the posterior distributions obtained by the Bayesian approach, the decay rates of F+ and somatic coliphages in winter were clearly lower than those in summer, either under shaded treatment or sunlight treatment (Figure 3). In winter, the mean decay rate of F+ coliphages was 0.25 day−1 (SD: 0.02 day−1, 95% CIs: 0.22 day−1 to 0.29 day−1), which is less than one-fourth of that obtained in summer. The decay rate of somatic coliphages was much lower in winter, with the mean of 0.12 day−1 (SD: 0.01 day−1, 95%CIs: 0.09−0.14 day−1) (Table 1, Figure 3). The decay rates of F+ and somatic coliphages were not significantly different in summer, whether the decay rates were compared under shaded treatment or sunlight treatment (Table 1, Figure 2). However, in winter, the decay rate of F+ coliphages was significantly higher, nearly two times as high as that of somatic coliphages (Table 1, Figure 3). Comparison of Decay Rates between Coliphages and Indicator Bacteria. Using a Bayesian approach, the decay rates of total coliforms and E. coli in summer were calculated (Table 1). In the summer shaded treatment, the mean decay rate of E. coli was 0.55 day−1 (SD: 0.05 day−1, 95% CI: 0.44− 0.65 day−1). The decay rate of total coliforms was slightly lower than that of E. coli, with a mean of 0.48 day−1 (SD: 0.03 day−1, 95% CI: 0.42−0.54 day−1). Interestingly, E. coli and total coliforms had significantly smaller decay rates than either F+ or somatic coliphages in shaded conditions (Figure 2). Under summer sunlight exposure, the decay rates of both E. coli and total coliforms increased nearly two times compared to the shaded treatment, with mean values of 1.35 and 0.93 day−1, respectively. Among all of the measured indicators, the decay rate of E. coli was the highest, while total coliforms had significantly lower decay rates compared to the other indicators. Comparison of Conventional and Bayesian Approaches for Decay Rate Calculation. The decay rates of coliphages calculated with both methods are presented in Table 1. In the summer experiment, the mean decay rates of F+ coliphages in shaded and sunlit conditions calculated with the conventional method were 1.09 and 1.16 day−1, respectively, which were close to the means obtained with Bayesian statistics
Table 2. Sensitivity of the Decay Rate of Somatic Coliphages to the Model Prior Parameters, Sigma (σ), a and b decay rate (unit: day−1) prior parameter
parameter value
mean
SD
2.5% quantile
97.5% quantile
σ
1 2 4 8 1 2 4 8 −0.3 0 0.3 0.6
0.96 0.97 0.98 0.97 0.95 0.96 0.96 0.96 0.96 0.96 0.96 0.96
0.04 0.07 0.15 0.29 0.03 0.04 0.04 0.04 0.04 0.04 0.04 0.04
0.89 0.83 0.70 0.43 0.88 0.89 0.89 0.88 0.88 0.89 0.88 0.88
1.04 1.13 1.27 1.54 1.00 1.04 1.04 1.04 1.03 1.04 1.04 1.04
b
a
mean decay rate only changed from 0.95 to 0.96 day−1. The mean decay rate was constant at 0.96 day−1 when a changed from −0.3 to 0.6 (Table 2). The sensitivity plots also demonstrated that the change in the values of these parameters did not have much influence on the decay rates (Figure S5). However, an increase in the σ value increased the standard deviation and 95% CI of the decay rate, thus increasing the uncertainty in estimating the decay rates. In contrast, changes in parameters a and b did not change the standard deviation and 95% CI of the decay rates (Figure S5). Therefore, these two parameters had no influence on the decay rate calculation. Analysis of Seasonal Effects on Coliphage Decay. For somatic coliphages, when the decay rate under sunlight treatment in summer was compared with that in winter, 12.1% of the variation was attributable to solar radiation, 49.7% of the variation was attributable to the seasonal factor and 38.2% of the variation was attributable to their interaction (Table S1). When the decay rate under shaded treatment in summer was compared with that in winter, 96.2% of the variation was attributable to the seasonal factor and 2.9% of the variation was attributable to their interaction. Only 0.9% of the variation was attributable to solar radiation (Table S1). The analysis suggests that temperature might have the largest contribution to the seasonal variation of the decay rate, F
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
decay rate of coliphages as a study showed that viruses had higher decay rates in eutrophic water than in oligotrophic water.48 However, the effect of chla and temperature might be interacted because the concentration of chla is temperaturedependent.49 The strong influence of temperature on virus decay has been reported in many studies.23−25 Besides affecting chla, temperature has direct effects on coliphage decay (e.g., affect the attachment and multiplication of viruses).6 Overall, temperature is speculated to be the primary factor that is responsible for the variation of the decay rates of coliphages between summer and winter. Since solar radiation will be attenuated with depth through the water column, the decay rates of coliphages may be affected by water depth of sampling. A recent study in the same location showed that solar radiation intensity was attenuated rapidly from the surface to ∼10 cm deep and varied very little below 10 cm depth in an unmixed system.38 In this study, water depth is not an influential factor of the decay of coliphages because the samples were collected near 15 cm below the surface of the water. The decay of coliphages may be also influenced by water flow. In a mixed system of which water is flowing, water flow may affect the time and intensity of solar radiation that viruses are exposed along the transport pathway.38 It should be noted that the decay rate measured in an unmixed system in this study may be different from that measured in a mixed system even if all other conditions are same. The results of the summer decay study showed that the decay rates of E. coli and total coliforms were much lower than those of coliphages in shaded conditions, which is partly consistent with previous studies. A study in California beach sand found that the decay rate of E. coli in the dark was lower than that of F+ coliphages but higher than that of somatic coliphages,50 while another study observed a higher decay rate of E. coli in the dark in waste stabilization pond effluent compared to coliphages.18 By comparing sunlight inactivation rates for different indicators from waste stabilization pond effluent, Sinton et al.18 found that E. coli are more susceptible to sunlight than somatic coliphages and F+RNA phages. These results indicated that coliphages were more resistant to sunlight than E. coli and total coliforms, suggesting coliphages might be more stable indicators for public health risks than E. coli and total coliforms in sewage-contaminated water. Both the conventional first-order decay kinetics model and the Bayesian approach were used for calculating decay rates. The results showed that the Bayesian approach provided the distribution of decay rates and reduced the uncertainty of decay rate estimates, which is important for comparing decay rates under different conditions. In the decay rate calculations, there are several steps that can lead to uncertainty. First, the concentrations of microbial indicators are typically measured from a limited number of samples and the distributions of these microorganisms are not uniform. Second, variability and experimental errors associated with the detection methods are often inevitable, which also increase uncertainty. Moreover, the decay rate is assumed to be constant in the conventional method, which is not true and can increase uncertainty; whereas in Bayesian statistics, the decay rate is treated as a random variable. Lower uncertainty indicates that estimates are more reliable, which is important for decision making. For example, in this study, it is difficult to tell whether decay rates of somatic coliphages significantly differed between the shaded and the sunlight treatments based on the conventional method because of the large standard deviations associated with rate
followed by chla. Turbidity and NPOC might have minor effects on the seasonal variation of the decay rate. Since the decay rate of F+ coliphage did not significantly vary under sunlight or shaded treatment, the seasonal variation of the decay rate was mostly attributable to the seasonal factor, including temperature, chla, turbidity and NPOC.
■
DISCUSSION We have demonstrated the use of a Bayesian approach to address the uncertainty of the decay rates of coliphages in sewage-contaminated freshwater under different conditions. We found that the decay of coliphages was significantly affected by sunlight and varied dramatically between summer and winter. We also revealed that the Bayesian approach offered several advantages over the conventional method, as it provided more reliable estimates of the distributions of the decay rates by taking into consideration the uncertainty in the decay rate estimation. To our knowledge, this is the first study that used Bayesian statistics in estimating the decay of viruses in water. The results are expected to better help understand the decay of viral pathogens in water and to introduce an approach to address uncertainty in decay estimations. Our results showed that sunlight significantly affects the decay of somatic coliphages but not F+ coliphages, suggesting that F+ coliphages might be more resistant to sunlight than somatic coliphages. A previous study, Sinton et al.17 found that F+ coliphages might be slightly more resistant to sunlight than somatic coliphages in freshwater, but less resistant in seawater. The same study17 also showed that somatic coliphages were highly susceptible to the UVB spectrum (near 338 nm) of sunlight. Our results also showed that sunlight significantly increased the decay rates of somatic coliphages, which is not surprising. A few studies have shown that indicator microorganisms are sensitive to sunlight in some matrices, including seawater, fresh, and saline waters.14,17,18 However, no significant effects of sunlight on F+ coliphage decay were observed in this study, which is different from the previous report on F+ coliphages decay in freshwater and seawater.14 This discrepancy may be related to the interaction of environmental factors with sunlight,43 which should be further investigated. The results also showed that the decay rates of both types of coliphages were significantly different between summer and winter, suggesting seasonal variation influences the decay of enteric viruses in water. Based on variance decomposition, oneway ANOVA and published literature, we identified four important factors, including turbidity, NPOC, chla, and temperature that contributed to the seasonal effect besides solar radiation. Turbidity may affect the decay of microorganisms through two ways: (1) by reducing the UV light transmittance in water; and/or (2) by shielding microorganisms from exposure to UV light.44 According to a study showing similar inactivation rates of coliphage MS2 caused by UV light in water with differences in turbidity,45 turbidity was expected to have a small contribution to the seasonal effect. NPOC measures organic matter in water, which can both attenuate light and produce reactive oxygen species.21,46 Therefore, NPOC may decease or increase the decay rates of coliphages and its total effects may be offset. Predation by algae or other microorganisms is another mechanism of virus decay. Chla is commonly used as an indicator for algae biomass, and a higher concentration of chla indicates a higher algae biomass.47 A higher concentration of chla might be associated with a higher G
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
■
ACKNOWLEDGMENTS We thank Emily Bailey from UNC-Chapel Hill for helping to facilitate this research.
estimates. However, it became evident that the decay rate of somatic coliphages exposed to sunlight was higher than the shaded treatment based on the Bayesian approach. In Bayesian statistics, the choice of prior parameter values is often subjective,34 which may influence model results. The results of the sensitivity analysis showed that the change in parameter values had little influence on the decay rate estimates. Only the increase in prior parameter σ led to an increase in the standard deviation of the decay rate estimates. This is expected because σ represents the standard deviation of the concentration of a specific indicator. When the concentration of a specific indicator has a large standard deviation, its decay rate will also have a large standard deviation. This study has several limitations. First, the decay rates of coliphages were measured in one place in California. Further studies need to investigate the decay of coliphages in other locations whose environmental conditions may differ. Second, there was a relatively long holding time (∼25−73 h) between sample collection and coliphage analysis, during which time the coliphages may have undergone decay. However, the decay is expected to be very slow at low temperatures (≤4 ◦C).51,52 Moreover, we are only able to separate the seasonal effects on coliphage decay attributable to solar radiation from those attributable to other factors, but cannot quantify the effects attributable to each specific factor without a more complicated experimental design.43 Finally, the effects of environmental factors on coliphage decay vary depending on specific genogroups of coliphages.6,24,51 Further studies to identify the genogroups of coliphages would provide additional information to understand the decay of specific coliphages under different environmental conditions. Even with these limitations, this work provides major contributions toward understanding the decay of coliphages in water. The findings are useful for further validating coliphages as an indicator of fecal contamination, modeling the fate and transport of viruses as well controlling viral pathogens in water. This study is also one of the first to introduce Bayesian modeling in decay rate calculations, an approach that can be used more broadly in water quality modeling, which allows us to understand effects of uncertainty and parameter estimation when modeling microbial contaminants, thereby improving the reliability of water quality models.
■
■
REFERENCES
(1) Havelaar, A. H.; van Olphen, M.; Drost, Y. C. F-Specific RNA bacteriophages are adequate model organisms for enteric viruses in fresh water. Appl. Environ. Microbiol. 1993, 59 (9), 2956−2962. (2) Sedmak, G.; Bina, D.; MacDonald, J.; Couillard, L. Nine-year study of the occurrence of culturable viruses in source water for two drinking water treatment plants and the influent and effluent of a wastewater treatment plant in Milwaukee, Wisconsin (August 1994 through July 2003). Appl. Environ. Microbiol. 2005, 71 (2), 1042− 1050. (3) Skraber, S.; Gassilloud, B.; Gantzer, C. Comparison of coliforms and coliphages as tools for assessment of viral contamination in river water. Appl. Environ. Microbiol. 2004, 70 (6), 3644−3649. (4) Tree, J. A.; Adams, M. R.; Lees, D. N. Chlorination of indicator bacteria and viruses in primary sewage effluent. Appl. Environ. Microbiol. 2003, 69 (4), 2038−2043. (5) Wu, J.; Long, S.; Das, D.; Dorner, S. Are microbial indicators and pathogens correlated? A statistical analysis of 40 years of research. J. Water Health 2011, 9 (2), 265−278. (6) U.S.EPA. Review of Coliphages As Possible Indicators of Fecal Contamination for Ambient Water Quality, 820-R-15-098; Washington, D.C, 2015. (7) Bonilla, T. D.; Nowosielski, K.; Cuvelier, M.; Hartz, A.; Green, M.; Esiobu, N.; McCorquodale, D. S.; Fleisher, J. M.; Rogerson, A. Prevalence and distribution of fecal indicator organisms in South Florida beach sand and preliminary assessment of health effects associated with beach sand exposure. Mar. Pollut. Bull. 2007, 54 (9), 1472−1482. (8) Borrego, J. J.; Córnax, R.; Morinigo, M. A.; Martínez-Manzanares, E.; Romero, P. Coliphages as an indicator of faecal pollution in water. Their survival and productive infectivity in natural aquatic environments. Water Res. 1990, 24 (1), 111−116. (9) Havelaar, A. Bacteriophages as model organisms in water treatment. Microbiol. Sci. 1987, 4 (12), 362−364. (10) Payment, P.; Franco, E. Clostridium perfringens and somatic coliphages as indicators of the efficiency of drinking water treatment for viruses and protozoan cysts. Appl. Environ. Microbiol. 1993, 59 (8), 2418−2424. (11) Colford, J. M.; Wade, T. J.; Schiff, K. C.; Wright, C. C.; Griffith, J. F.; Sandhu, S. K.; Burns, S.; Sobsey, M.; Lovelace, G.; Weisberg, S. B. Water quality indicators and the risk of illness at beaches with nonpoint sources of fecal contamination. Epidemiology 2007, 18 (1), 27−35. (12) Abdelzaher, A. M.; Wright, M. E.; Ortega, C.; Hasan, A. R.; Shibata, T.; Solo-Gabriele, H. M.; Kish, J.; Withum, K.; He, G.; Elmir, S. M. Daily measures of microbes and human health at a non-point source marine beach. J. Water Health 2011, 9 (3), 443−457. (13) Griffith, J. F.; Weisberg, S. B.; Arnold, B. F.; Cao, Y.; Schiff, K. C.; Colford, J. M. Epidemiologic evaluation of multiple alternate microbial water quality monitoring indicators at three California beaches. Water Res. 2016, 94, 371−381. (14) Noble, R. T.; Lee, I. M.; Schiff, K. C. Inactivation of indicator micro-organisms from various sources of faecal contamination in seawater and freshwater. J. Appl. Microbiol. 2004, 96 (3), 464−472. (15) Whitman, R. L.; Nevers, M. B.; Korinek, G. C.; Byappanahalli, M. N. Solar and temporal effects on Escherichia coli concentration at a lake Michigan swimming beach. Appl. Environ. Microbiol. 2004, 70 (7), 4276−4285. (16) Ferguson, C.; Husman, A. M. D.; Altavilla, N.; Deere, D.; Ashbolt, N. Fate and transport of surface water pathogens in watersheds. Crit. Rev. Environ. Sci. Technol. 2003, 33 (3), 299−361. (17) Sinton, L. W.; Finlay, R. K.; Lynch, P. A. Sunlight inactivation of fecal bacteriophages and bacteria in sewage-polluted seawater. Appl. Environ. Microbiol. 1999, 65 (8), 3605−3613.
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.6b03916. Water quality and weather conditions of the decay studies, the sensitivity analysis of parameters for Bayesian model and the OpenBUGS code for estimating the decay rates of F+ and somatic colipahges with Bayesian Statistics (PDF)
■
Article
AUTHOR INFORMATION
Corresponding Author
*Phone: +1 919-208-1666; fax: 919-966-7911; e-mail: jianyong.
[email protected]. Notes
The authors declare no competing financial interest. H
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Waters: Effects of Water Absorbance and Depth. Environ. Sci. Technol. 2016, 50 (10), 5068−5076. (39) U.S. EPA, Method 1602. Male-Specific (F+) and Somatic Coliphage in Water by Single Agar Layer (SAL) Procedure, EPA-821R-01-029; Washington, DC, 2001. (40) Eichmiller, J. J.; Borchert, A. J.; Sadowsky, M. J.; Hicks, R. E. Decay of genetic markers for fecal bacterial indicators and pathogens in sand from Lake Superior. Water Res. 2014, 59, 99−111. (41) Rogers, S. W.; Donnelly, M.; Peed, L.; Kelty, C. A.; Mondal, S.; Zhong, Z. R.; Shanks, O. C. Decay of bacterial pathogens, fecal indicators, and real-time quantitative PCR genetic markers in manureamended soils. Appl. Environ. Microbiol. 2011, 77 (14), 4839−4848. (42) Saltelli, A.; Tarantola, S.; Chan, K.-S. A quantitative modelindependent method for global sensitivity analysis of model output. Technometrics 1999, 41 (1), 39−56. (43) Davies-Colley, R. J.; Donnison, A. M.; Speed, D. J.; Ross, C. M.; Nagels, J. W. Inactivation of faecal indicator microorganisms in waste stabilisation ponds: Interactions of environmental factors with sunlight. Water Res. 1999, 33 (5), 1220−1230. (44) Passantino, L. Effect of low turbidity and algae on UV disinfection performance. J. Am. Water Works Assoc. 2004, 96 (6), 128−137. (45) Liu, W. J.; Zhang, Y. J. Effects of UV intensity and water turbidity on microbial indicator. J. Environ. Sci. 2006, 18 (4), 650−653. (46) Chung, H.; Sobsey, M. D. Comparative Survival of Indicator Viruses and Enteric Viruses in Seawater and Sediment. Water Sci. Technol. 1993, 27 (3−4), 425−428. (47) Boyer, J. N.; Kelble, C. R.; Ortner, P. B.; Rudnick, D. T. Phytoplankton bloom status: Chlorophyll a biomass as an indicator of water quality condition in the southern estuaries of Florida, USA. Ecol. Indic. 2009, 9 (6), S56−S67. (48) Bongiorni, L.; Magagnini, M.; Armeni, M.; Noble, R.; Danovaro, R. Viral production, decay rates, and life strategies along a trophic gradient in the North Adriatic Sea. Appl. Environ. Microb 2005, 71 (11), 6644−6650. (49) Murata, N.; Fork, D. C. Temperature dependence of chlorophyll a fluorescence in relation to the physical phase of membrane lipids algae and higher plants. Plant Physiol. 1975, 56 (6), 791−796. (50) Yamahara, K. M.; Sassoubre, L. M.; Goodwin, K. D.; Boehm, A. B. Occurrence and persistence of bacterial pathogens and indicator organisms in beach sand along the California coast. Appl. Environ. Microbiol. 2012, 78 (6), 1733−1745. (51) Lee, H. S.; Sobsey, M. D. Survival of prototype strains of somatic coliphage families in environmental waters and when exposed to UV low-pressure monochromatic radiation or heat. Water Res. 2011, 45 (12), 3723−3734. (52) Mendez, J.; Jofre, J.; Lucena, F.; Contreras, N.; Mooijman, K.; Araujo, R. Conservation of phage reference materials and water samples containing bacteriophages of enteric bacteria. J. Virol. Methods 2002, 106 (2), 215−224.
(18) Sinton, L. W.; Hall, C. H.; Lynch, P. A.; Davies-Colley, R. J. Sunlight inactivation of fecal indicator bacteria and bacteriophages from waste stabilization pond effluent in fresh and saline waters. Appl. Environ. Microbiol. 2002, 68 (3), 1122−1131. (19) Kohn, T.; Nelson, K. L. Sunlight-mediated inactivation of MS2 coliphage via exogenous singlet oxygen produced by sensitizers in natural waters. Environ. Sci. Technol. 2007, 41 (1), 192−197. (20) Romero, O. C.; Straub, A. P.; Kohn, T.; Nguyen, T. H. Role of temperature and Suwannee River natural organic matter on inactivation kinetics of rotavirus and bacteriophage MS2 by solar irradiation. Environ. Sci. Technol. 2011, 45 (24), 10385−10393. (21) Silverman, A. I.; Peterson, B. M.; Boehm, A. B.; McNeill, K.; Nelson, K. L. Sunlight Inactivation of Human Viruses and Bacteriophages in Coastal Waters Containing Natural Photosensitizers. Environ. Sci. Technol. 2013, 47 (4), 1870−1878. (22) Wommack, K. E.; Hill, R. T.; Muller, T. A.; Colwell, R. R. Effects of sunlight on bacteriophage viability and structure. Appl. Environ. Microb. 1996, 62 (4), 1336−1341. (23) Fong, T. T.; Lipp, E. K. Enteric viruses of humans and animals in aquatic environments: Health risks, detection, and potential water quality assessment tools. Microbiol. Mol. Biol. Rev. 2005, 69 (2), 357− 371. (24) Long, S. C.; Sobsey, M. D. A comparison of the survival of F +RNA and F+DNA coliphages in lake water microcosms. J. Water Health 2004, 2 (1), 15−22. (25) Bertrand, I.; Schijven, J. F.; Sanchez, G.; Wyn-Jones, P.; Ottoson, J.; Morin, T.; Muscillo, M.; Verani, M.; Nasser, A.; Husman, A. M. D.; Myrmel, M.; Sellwood, J.; Cook, N.; Gantzer, C. The impact of temperature on the inactivation of enteric viruses in food and water: a review. J. Appl. Microbiol. 2012, 112 (6), 1059−1074. (26) Bristow, K. L.; Campbell, G. S. On the relationship between incoming solar radiation and daily maximum and minimum temperature. Agricult. Forest Meterol. 1984, 31 (2), 159−166. (27) Cantalupo, P. G.; Calgua, B.; Zhao, G.; Hundesa, A.; Wier, A. D.; Katz, J. P.; Grabe, M.; Hendrix, R. W.; Girones, R.; Wang, D. Raw sewage harbors diverse viral populations. mBio 2011, 2 (5), e00180− 11. (28) Metcalf, T.; Melnick, J.; Estes, M. Environmental virology: from detection of virus in sewage and water by isolation to identification by molecular biology-a trip of over 50 years. Annu. Rev. Microbiol. 1995, 49 (1), 461−487. (29) Chick, H. An investigation of the laws of disinfection. J. Hyg. 1908, 8 (01), 92−158. (30) Crane, S. R.; Moore, J. A. Modeling Enteric Bacterial Die-Off - a Review. Water, Air, Soil Pollut. 1986, 27 (3−4), 411−439. (31) Watson, H. E. A note on the variation of the rate of disinfection with change in the concentration of the disinfectant. J. Hyg. 1908, 8 (04), 536−542. (32) Gelman, A.; Carlin, J. B.; Stern, H. S.; Rubin, D. B. Bayesian Data Analysis; Taylor & Francis, 2014; Vol. 2. (33) Gilks, W. R.; Richardson, S.; Spiegelhalter, D. J. Introducing Markov Chain Monte Carlo; Chapman & Hall/CRC, 1996; Vol. 1, p 19. (34) Hoff, P. D. A First Course in Bayesian Statistical Methods; Springer Science & Business Media, 2009. (35) Gronewold, A. D.; Qian, S. S.; Wolpert, R. L.; Reckhow, K. H. Calibrating and validating bacterial water quality models: A Bayesian approach. Water Res. 2009, 43 (10), 2688−2698. (36) Wu, J. Y.; Gronewold, A. D.; Rodriguez, R. A.; Stewart, J. R.; Sobsey, M. D. Integrating quantitative PCR and Bayesian statistics in quantifying human adenoviruses in small volumes of source water. Sci. Total Environ. 2014, 470, 255−262. (37) Wu, J. Y.; Simmons, O. D.; Sobsey, M. D. Uncertainty analysis of the recovery of hollow-fiber ultrafiltration for multiple microbe classes from water: A Bayesian approach. J. Microbiol. Methods 2013, 93 (3), 161−167. (38) Maraccini, P. A.; Mattioli, M. C. M.; Sassoubre, L. M.; Cao, Y. P.; Griffith, J. F.; Ervin, J. S.; Van De Werfhorst, L. C.; Boehm, A. B. Solar Inactivation of Enterococci and Escherichia coli in Natural I
DOI: 10.1021/acs.est.6b03916 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
