Persistence of fecal indicator bacteria and associated genetic markers from wastewater treatment plant effluents in freshwater microcosms

The U.S. Environmental Protection Agency (EPA) has accepted the use of a quantitative PCR (qPCR) method as a rapid alternative for monitoring recreational waters for fecal contamination (U.S. EPA 2012a). Fecal contaminations such as combined sewer overflows, leaky septic systems and runoff of land-deposited municipal, livestock and other fecal wastes are possible sources of microbial pathogens with human health risks in recreational waters (Gerba & Smith 2005; Nevers et al. 2014). However, treated wastewaters discharged by wastewater treatment plants (WWTPs) are considered to be a main source of pathogens in many of these waters (Payment et al. 2001; Koivunen et al. 2003; Ottoson et al. 2006a; Castro-Hermida et al. 2008; Carducci et al. 2009; Okoh et al. 2010; Hendricks & Pool 2012). The fate of microbial pathogens that may survive current wastewater treatment processes, therefore, remains an important question.

qPCR analysis methods have been used extensively to study the decay of fecal indicator bacterial (FIB) genetic markers as potential surrogates for microbial pathogens. A number of studies have investigated the persistence of such markers from various FIB groups in marine and freshwater microcosms (Seurinck et al. 2005; Wery et al. 2006; Okabe & Shimazu 2007; Bae & Wuertz 2009; Bell et al. 2009; Walters & Field 2009; Walters et al. 2009; Jeanneau et al. 2012; Korajkic et al. 2014; Bae & Wuertz 2015; Cloutier & McLellan 2017). Most of these persistent studies have used raw sewage or fecal material as the seed material in microcosms. Findings from these studies may not accurately reflect the fate of genetic markers present in normal WWTP effluents. Under normal operating conditions, WWTP effluents have been treated by various primary and secondary processes and then typically disinfected by either chlorine or ultraviolet (UV) radiation. These treatment processes are known to reduce the concentrations and viability of the fecal microbial populations in the wastewater (George et al. 2002, Villanova et al. 2004; Ottoson et al. 2006b; Wéry et al. 2008; Varma et al. 2009; Wen et al. 2009). Less is known about the effects of these treatments on the persistence of the nucleic acid components of these organisms after they are released with the treated wastewater effluents into ambient receiving waters. Chern et al. (2014) reported no immediate significant reduction of FIB genetic markers caused by secondary wastewater disinfection processes. Other studies have reported differences between chlorination and UV irradiation with regard to reduction and viability of culturable bacteria (Blatchley et al. 2007; Crockett 2007), but similar studies have not examined the influence of these treatments on the persistence of FIB genetic markers.

Knowledge about the persistence of FIB genetic markers originating from treated wastewater could be useful in understanding how well these analytes perform as surrogates for similarly treated microbial pathogens and thus in their ability to predict potential health risks from these pathogens. In this study, we examined the relative persistence of genetic markers from a range of different FIB groups in freshwater microcosms seeded with non-disinfected or disinfected secondary-treated wastewater effluents based on determinations of their decay rate constants. The FIB groups included enterococci and Escherichia coli, both facultative anaerobic bacteria groups that are recommended by the U.S. EPA for water quality testing (U.S. EPA 1986, 2012a), and the strict anaerobe groups, Clostridium and Bacteroides, among which the former are known to be environmentally persistent due to their production of spores and the latter are commonly used as indicators in microbial source tracking. The decay rate constants of culturable organisms in each of these groups were also determined the influence of different disinfection methods (UV and chlorination). Experiments were performed in the winter, fall, spring and summer seasons to determine any temperature effects, and data from the fall and spring season experiments were pooled (fall/spring) due to their similarity in water temperatures. Two different sources of simulated receiving waters for the microcosms, such as Lake Michigan and the Ohio River, were selected for the study to represent different water body types that can be impacted by WWTP effluents. This study provides a direct comparison of the environmental persistence of a variety of different FIB groups and their associated genetic markers coming from treated wastewater including the effects of disinfection.

Surface water samples (20 L) were collected in 2008 from Lake Michigan near Porter, IN and in 2009 from the Ohio River near California, OH. Samples were collected during the summer, spring/fall, and winter season in the mornings at approximately 9 AM. Both secondary non-disinfected and disinfected wastewater samples (1 L) were collected from four different WWTPs across southern Ohio. All plants used the activated sludge treatment process for secondary wastewater treatment. Two plants disinfected by UV radiation (daily average discharge of 5–16 million gallons daily (MGD)) and the other two by chlorination (daily average discharge of 27–144 MGD). All samples were held on ice during transport to the laboratory for immediate processing.

A total of nine microcosms were prepared for each of the six experiments (two receiving water sources, each collected in three different seasons). Each microcosm contained a total volume of 1.5 L in 2 L flasks with tops covered by aluminum foil. One of the nine microcosms was used as a control containing only Lake Michigan or Ohio River water. The remaining eight microcosms were a mixture of freshwater and 20% of the secondary-treated or disinfected, secondary-treated wastewater effluent collected from one of the four different treatment plants. Microcosms were stored in the dark at temperatures similar to the ambient Lake or River water temperature at the time of collection (4, 20 and 25 °C) with constant agitation on a New Brunswick shaker (Innova Model 2150, New Brunswick Scientific, Enfield, CT) at a speed of approximately 65 rpm throughout the study period. In this study, seasons correspond to the different water temperatures during sample collection. Duplicate samples were collected for analysis on the initial day (D₀) followed by sampling on days 1, 2, 4 and 6.

Volumes of 1, 5 and 10 mL water samples from each microcosm were filtered in duplicate through a 0.45 μm pore size (47 mm in diameter) cellulose nitrate membrane filter (Sartorius, Bohemia, NY). Filters were placed onto mEI (U.S. EPA 2002) agar and followed by incubation overnight at 41 °C for the enumeration of enterococci. For E. coli enumeration, modified mTEC (U.S. EPA 2014) agar was used and followed by incubation at 44.5 °C (with a 2-h initial incubation at 35 °C). Bacteroides Bile Esculin (BBE) (Livingston et al. 1978) agar was used and followed by incubation at 36 °C for 48 h for culturing of the Bacteroides fragilis group. Filters placed onto mCP (U.S. EPA 1996) agar were incubated overnight at 44.5 °C for the enumeration of Clostridium perfringens. Anaerobic conditions using the GasPak™ EZ container system (BD Diagnostics, Franklin Lakes, NJ) were used for incubation of the BBE and mCP plates. Presumptive C. perfringens colonies were confirmed as described in the Membrane Filter Method for C. perfringens in the Information Collection Rule Microbial Lab Manual (U.S. EPA 1996).

Duplicate aliquots of 50 mL from each microcosm were filtered through a 0.4 μm pore size (47 mm in diameter) polycarbonate membrane filter (GE Osmonics, Minnetonka, MA). Filters were placed into 2 mL screw cap extraction tubes prefilled with 0.3 g of glass beads. Cells were lysed after the addition of 600 μL of AE Buffer (Qiagen, Valencia, CA) containing 0.2 μg mL⁻¹ of salmon DNA (Sigma, St. Louis, MO) into the extraction tubes followed by bead milling and then centrifugation as described previously by Haugland et al. (2005). DNA extracts were stored at 4 °C no longer than 24 h prior to qPCR analysis.

qPCR was used to estimate the concentrations of genetic markers of enterococci, E. coli, Clostridium spp. and the B. fragilis cluster in the microcosms. Primers and TaqMan^® hydrolysis probe sequences used to amplify the genetic markers are shown in Supplementary Material, Table S1. For the purpose of consistency, this paper will subsequently refer to the group of Bacteroides species detected by the qPCR (Okabe et al. 2007) and the species detected by BBE culture medium as B. fragilis group. Similarly, the Clostridium species detected by the qPCR (Rinttilä et al. 2004; Supplementary Material, Table S1) and by mCP culture medium will both be referred to as Clostridium unless otherwise specified. Total reaction volumes of 25 μL containing 12.5 μL of TaqMan^® Universal PCR Master Mix or Gene Expression Master Mix for the E. coli assay (Applied Biosystems, Foster City, CA), 2.5 μL of 2 mg L⁻¹ bovine serum albumin, 1 μmol L⁻¹ of each primer, 80 nmol L⁻¹ of probe and 5 μL of 5-fold diluted extracts were used. Amplification conditions consisted of an initial incubation at 50 °C for 2 min then 95 °C for 10 min followed by 40 PCR cycles of 95 °C for 15 s and 60 °C for 1 min. All reactions were carried out in an ABI Model 7900 DNA thermal cycler (Applied Biosystems, Foster City, CA).

Calibrator cells of E. coli (ATCC #25922), Enterococcus faecalis (ATCC# 29212), C. perfringens (ATCC# 13124) and Bacteroides thetaiotaomicron (ATCC # 29741), a member of the B. fragilis group (Okabe et al. 2007), were prepared as previously described by Chern et al. (2009). Approximately 1 × 10⁶E. coli, 4.6 × 10⁴C. perfringens, 3 × 10⁵E. faecalis and 3.0 × 10⁵B. thetaiotaomicron cells were present in each calibrator sample. Serial dilutions of calibrator sample extracts were prepared to determine amplification efficiency and detection limits for each of the assays. qPCR analyses for salmon DNA that was added to all samples at identical concentrations prior to DNA extraction were used as a control to determine total DNA recovery in the extracts and to signal potential PCR interference (Haugland et al. 2005; U.S. EPA 2012b). All samples were analyzed in duplicate. Filter blanks and no template controls were also analyzed as negative controls along with the calibrator samples and unknown test samples.

Concentrations of each genetic marker were estimated as calibrator cell equivalents (CCE) by the delta–delta C_t calculation method as previously described (Chern et al. 2011, 2014) using the mean amplification efficiency value generated from multiple analyses of serially diluted calibrator sample extracts. CCE estimates below the mean detection limits were adjusted to one-half of the mean detection limit values. Geometric mean concentration estimates from duplicate filters of each sample were calculated and extrapolated to 100 mL sample volumes. Concentrations by culture were expressed as log₁₀ colony-forming units (CFUs). Rates of decay for each FIB group, both genetic markers and culture-based, in the microcosm were then estimated using the first-order Chick–Watson kinetic model (Chick 1908; Watson 1908): ln (C/C₀) = −kt, where C₀ is the initial concentration (CFU or CCE per 100 mL), k is the rate constant (day⁻¹) and C is the concentration (CFU or CCE per 100 mL) at time t, in days. Average decay rate constants over the four different FIBs are sometimes referred to the convenience of discussion, particularly when their trends are similar. This is not meant to imply that these decay rate constants are independent of FIB. Intra-FIB differences are discerned later in the discussion.

Analyses of variance (ANOVAs) were performed separately for criterion variables (CCE, CFU) as well as their differences (CCE − CFU). The model utilized the following as independent, explanatory, variables:

• type of organism (four indicator classes),

• season (winter vs. spring/summer/fall),

• source of spiked freshwater (river, lake) and

• effluent disinfection status (non-disinfected, chlorine-disinfected, UV-disinfected).

All potential interactions among the above were included, yielding a total of 10 potential explanatory variables. The four WWTP sources of wastewater used for spiking were modeled as a random factor, representative of WWTPs in general. Based on the estimated mean decay rate constants as modeled via SAS version 9, p-values (significance level = 0.05) were used in the evaluation of the difference between seasons, disinfected and non-disinfected secondary wastewater effluent and disinfection methods with respect to stepwise log₁₀ reductions for the individual CCE, CFU and CCE–CFU. ANOVA results for significant model variables are tabulated separately to facilitate discussion in the results section.

Amplification efficiency estimates of each qPCR assay were generated from the slope of the serial dilution analyses and ranged from 0.92 to 1.04. Results indicated that the mean calibrator cell detection limits, which were estimated as described in Chern et al. (2011), were approximately 86, 315, 75 and 134 CCE per 50 mL filtered water sample volume based on analyses of 5-fold extract dilutions for the enterococci, E. coli, B. fragilis group and Clostridium assays, respectively. No inhibition or contamination was detected from analysis of the salmon DNA sample processing controls and from analysis of the filter blanks and no template controls. Analyses of positive and negative control test organisms along with sterility controls were performed with each culture method and exhibited acceptable results.

After the addition of wastewater into the lake- or river-receiving water microcosms, genetic marker concentrations increased, on average, by approximately 0.6 logs for Clostridium up to 1.3 logs for enterococci (Table 1), and the concentrations of the different markers increased substantially in at least 84% (37/44) of the microcosms, indicating that most of the markers initially present in the majority of these microcosms originated from the added wastewater. A similar or greater increase in concentration was seen for culturable organisms in the seeded microcosms; however, levels of culturable organisms in the microcosms seeded with disinfected wastewater were not sufficiently high to accurately determine decay rate constants. Consequently, decay rate constants for culturable organisms were based on the non-disinfected wastewater-seeded microcosms only in this study. Overall, decay rate constants of the cultured FIB or genetic FIB marker did not significantly differ in the different receiving water microcosms, i.e. lake vs. river (Table 2), except for the genetic marker decay rates in winter where slower decay occurred in river than lake water (p < 0.0001) (Table 2). No significant differences were observed in the averaged decay rate constants of the genetic markers from the four FIB groups in freshwater microcosms seeded with the disinfected and non-disinfected secondary-treated wastewater effluent (Table 3). In addition, the type of disinfection (chlorination vs. UV) had no significant effect on the rates of genetic marker decay (Table 3). Decay rate constants of the averaged FIB culture data and of the averaged genetic marker data for the four FIB groups were not significantly different between the summer and spring/fall seasons, so decay rate constants from these seasons were combined in further analyses (Tables 4 and 5).

However, significant differences in decay rate constants of both culturable organisms and genetic markers were detected between winter and summer/spring/fall seasons and among the different FIB groups in the seeded microcosms (p < 0.05) (Tables 4 and 5). Differences between the decay rate constants of culturable organisms and their genetic markers also differed significantly among the different FIB groups in the summer/spring/fall season but not in the winter season (Table 5). Summer/spring/fall conditions (warmer water) had the greatest effect on accelerating B. fragilis genetic marker decay, while Clostridium genetic marker decay showed a relatively little seasonal variation (Table 5). Cultured enterococci, E. coli and B. fragilis group shared similar decay rate constants that were faster than cultured Clostridium in the summer/spring/fall season, while in the winter cultured enterococci and E. coli both were observed to have slightly faster decay rate constants than the cultured B. fragilis group and Clostridium markers. This contrasted with the similar decay rate constants of Clostridium, E. coli and enterococci genetic markers that decayed faster than B. fragilis group genetic markers in the winter (Table 5). Except for Clostridium, summer/spring/fall conditions accelerated both culturable organism and genetic marker decay rate constants than winter conditions. These results are also illustrated in Figure 1 where the persistence of both cultured FIB and their associated genetic markers was generally greater in the winter season than in the summer/spring/fall seasons.

Results of this study could provide important new insights into the persistence of genetic markers from FIB target organisms occurring in secondary-treated wastewater. Season or temperature had the strongest influence on the decay of these markers among the environmental parameters tested but only in terms of winter vs. summer, spring and fall conditions, even though the magnitude of this effect differed among the various FIBs. These results are consistent with previous results, indicating that water temperature was one of the single most important environmental factors in determining FIB genetic marker decay (Balleste & Blanch 2010; Bae & Wurtz 2015). However, wastewater disinfection and type of wastewater disinfection did not have a significant effect on the persistence of these markers in this study.

The persistence of culturable FIB has been reported to be influenced by numerous biotic and abiotic factors in the environment (Scheuerman et al. 1988; Okabe & Shimazu 2007; Jackson et al. 2008; Bell et al. 2009; Balleste & Blanch 2010; Korajkic et al. 2013). FIB genetic markers have been described as more environmentally persistent than corresponding culturable organisms (Walters et al. 2009; Korajkic et al. 2014). This conclusion was supported by results from this study for the traditional FIB groups, E. coli and enterococci. However, the larger decay rate constants seen for the B. fragilis group and Clostridium genetic markers compared to the culturable organisms in summer/spring/fall in this study appeared to differ from this more widely seen trend and may warrant further investigation. It should also be noted that there was no functional culture medium available to enumerate the same group of Clostridium species that was detected by the Clostridium genetic marker assay (see Supplementary Material, Table S1) or vice versa. Therefore, comparisons of decay rate constants between C. perfringens culture and the genetic markers detected by the qPCR assay in this study may not be appropriate. In contrast, it is believed that a highly similar group of species is detected by the selective culture medium and the genetic marker assay used in this study for the B. fragilis group. Similar trends in differences between culture and qPCR decay rate constants have been reported in another study using two specific Bacteroides strains from this group as seeding material (Balleste & Blanch 2010). At least in this study, using treated secondary effluents as the seed material for the microcosms, a reasonable hypothesis for the faster decay rate of B. fragilis group genetic markers than culturable cells could be that most of the qPCR target sequences are being contributed by cells that have already died and are deteriorating even before disinfection. This hypothesis is supported by the relatively high concentrations of B. fragilis CCE compared with CFU estimates in the D₀ microcosms containing non-disinfected as well as disinfected wastewater as shown in Table 1, although care should be taken when comparing these estimates (Haugland et al. 2014).

One common observation reported in several studies investigating culturable FIB persistence is that Clostridium is more persistent while Bacteroides is less persistent than other FIB indicators even when using different sources of these microbes to seed microcosms (Fiksdal et al. 1985; Lisle et al. 2004; Hijnen et al. 2009). These relationships could be extended to the genetic markers examined in this study where the Clostridium marker was also found to be the most persistent while the B. fragilis group marker was the least persistent, at least in summer, spring and fall conditions. The differences seen between the decay rate constants of genetic markers from the different FIB groups hold open the possibility that the fate of one of these groups may more closely mirror the fate of viral or protozoan pathogen groups in at least some environmental conditions.

For the microcosms to achieve winter temperature conditions and to minimize the number of different variables, all microcosms in this study were kept in controlled temperature rooms in the dark. The reported decay rate constants should be conservative since sunlight is known to accelerate the decay of culturable cells (Davies-Colley et al. 1999; Sinton et al. 2002). Results by Dong et al. (2014), using artificially constructed microcosms with pure cultures of Bacteroides ovatus, also indicated that DNA decayed faster in daylight conditions under full spectrum irradiation than in the dark; however, the effect of sunlight on accelerating the decay of E. coli, enterococci and Bacteroides genetic markers from sewage under more natural conditions was less evident (Wanjugi et al. 2016). The relative and/or absolute decay rate constants of E. coli, enterococci and Bacteroides genetic markers in this study appeared to be consistent with those reported in other studies for these organisms from raw sewage while under dark conditions (Wanjugi et al. 2016; Liang et al. 2017).

As with most small-scale controlled microcosm studies, there are limitations in the experimental approach of this study and the conclusions that can be drawn. For example, the volumes used for each analysis substantially decreased the total volumes of the microcosms over the period of the experiments which may have affected other parameters. If any such effects occurred, they would presumably have affected the decay rate constants equally for all potential factors that were investigated (organism, disinfection type, temperature, etc.). The primary focus of this study was not to exhaustively investigate the many environmental factors that could contribute to the decay of FIB and their genetic markers which, as discussed above, have been investigated in numerous other studies, but rather to focus on the differences in decay rate constants between different groups of potentially relevant FIB and investigate just a few parameters that could specifically affect the decay of these organisms and their genetic markers in treated wastewater effluents. Another potential limitation of this study was suggested by the relatively low concentrations of some of the different FIB groups and their genetic markers in some of the wastewater-seeded microcosms compared to corresponding receiving water microcosms. This observation would suggest that decay rate constant estimates for the FIB groups and their genetic markers in these microcosms could be strongly influenced by the decay of the indigenous organisms in the receiving waters. These microcosms, however, provided only a small percentage of the data used to estimate the overall FIB decay rate constants in this study and are thus believed not to strongly influence these estimates.

This study presents results from four WWTPs. We treated these as a random selection of WWTPs, adding the variance among plants to the more stable within plant variability. This allowed projection of the studied WWTP results to WWTPs in general. Adding WWTPs to this base would enable increased power in evaluating effects and in representativeness of the results. The same holds true for incorporating additional sources of receiving waters, which, in our case, were regarded as fixed effects out of necessity, given that only two sources were used. Differing receiving water characteristics could potentially yield valuable information on microbial decay relative to source water properties.

This study presents genetic marker decay rate constants from a range of FIB target groups originating from treated wastewaters. Knowledge of such decay rate constants should aid in determining if genetic markers may be suitable to serve as surrogates for pathogens in these effluents. The relationship between the persistence of FIB markers and enteric viruses and protozoa is of particular interest because both have been shown to be less effectively attenuated than culturable bacteria by the disinfection regimes commonly used at WWTPs (Chang et al. 1985; Tree et al. 1997). In addition, monitoring for genetic markers may be a better option for predicting the fate of pathogens in disinfected wastewater since, as reflected by this study, the lower concentrations of culturable bacterial indicators from these sources can make them more difficult to accurately quantify. Despite this, a limitation to the accurate determination of decay rate constants of genetic markers originating from treated wastewaters in receiving waters can be the relative concentration of these markers in the wastewaters compared to the receiving waters in microcosm studies of this nature. Studies on other WWTPs and source waters are needed to confirm the extent that results from the present study apply to a diverse range of conditions. In addition, further studies are needed to compare the persistence of FIB and their genetic markers with pathogens. Such studies could support the development of generalized models for predicting the health risks posed by WWTP effluents in ground water (Shende & Chau 2019) or at nearby recreational beaches (Boehm et al. 2018). Genetic markers from enterococci have already been shown to be predictive of associated health risks at WWTP-impacted beaches (U.S. EPA 2012a). Monitoring genetic markers could further aid in the routine assessment of water quality and in the protection of public health at other recreational beaches.

We thank several WWTPs in Ohio, Dawn Shively of USGS, and Tetra-Tech-Clancy Environmental (Saint Albans, VT) for their help in providing and collecting wastewater, Lake Michigan, and Ohio River water samples, respectively. We also thank Al Dufour and Jerry Stelma for their review of this manuscript.

Information has been subjected to U.S. EPA peer and administrative review and has been approved for external publication. Any opinions expressed in this paper are those of the authors and do not necessarily reflect the official positions and policies of the U.S. EPA. Any mention of trade names or commercial products does not constitute endorsement or recommendation for use.

None declared.

All relevant data are available from an online repository or repositories (https://doi.org/10.23719/1519085).