Differential decomposition of bacterial and viral fecal indicators in common human pollution types

Abstract

Understanding the decomposition of microorganisms associated with different human fecal pollution types is necessary for proper implementation of many water quality management practices, as well as predicting associated public health risks. Here, the decomposition of select cultivated and molecular indicators of fecal pollution originating from fresh human feces, septage, and primary effluent sewage in a subtropical marine environment was assessed over a six day period with an emphasis on the influence of ambient sunlight and indigenous microbiota. Ambient water mixed with each fecal pollution type was placed in dialysis bags and incubated in situ in a submersible aquatic mesocosm. Genetic and cultivated fecal indicators including fecal indicator bacteria (enterococci, E. coli, and Bacteroidales), coliphage (somatic and F+), Bacteroides fragilis phage (GB-124), and human-associated genetic indicators (HF183/BacR287 and HumM2) were measured in each sample. Simple linear regression assessing treatment trends in each pollution type over time showed significant decay (p ≤ 0.05) in most treatments for feces and sewage (27/28 and 32/40, respectively), compared to septage (6/26). A two-way analysis of variance of log10 reduction values for sewage and feces experiments indicated that treatments differentially impact survival of cultivated bacteria, cultivated phage, and genetic indicators. Findings suggest that sunlight is critical for phage decay, and indigenous microbiota play a lesser role. For bacterial cultivated and genetic indicators, the influence of indigenous microbiota varied by pollution type. This study offers new insights on the decomposition of common human fecal pollution types in a subtropical marine environment with important implications for water quality management applications.

Graphical Abstract

1. Introduction

The average human excretes 100 g or more feces per day (Rose et al., 2015) suggesting that a minimum of 3.2 × 10⁷ kg of feces are produced each day in the United States alone (USCB, 2015). As a result, proper waste management practices are paramount to protect public and ecological health. In the United States, there are more than 14,700 wastewater treatment facilities designed to collect, treat, and dispose of human waste (USEPA, 2008) These facilities collect waste through a system of more than 1.2 million miles of sewer pipes (USEPA, 2002a). In addition, approximately 20% of United States households depend on onsite septic tank systems (USEPA, 2008). Despite all these sanitation efforts, human fecal waste continues to impair environmental waters due to numerous situations such as: untreated sewage resulting from combined sewer overflow events after heavy rains and leaky sewer lines due to aging infrastructure, septage originating from faulty septic systems often due to lack of routine maintenance, and raw feces from illicit waste disposal, stormwater run-off, or recreational bather accidents.

Bacterial (E. coli and enterococci) general fecal indicators are routinely used to monitor for fecal pollution in environmental waters. However, these general fecal indicators are also common in the feces of other animal groups (Sadowsky and Whitman, 2011, Soller et al., 2010a), making it difficult to distinguish human fecal pollution from waste introduced by agricultural practices and local wildlife. A quantitative microbial risk assessment study suggests that fecal contamination originating from human sources presents a greater public health risk compared to fecal contamination originating from many other animal groups such as gull, chicken, and swine (Soller et al., 2010b). Because many impaired environmental waters likely contain a mixture of fecal waste from human and other animal sources with variable predicted public health risks, many water quality managers are turning to fecal source identification technologies to compliment general fecal indicator approaches. These host-associated methods are specifically designed to detect and, in some instances, quantify levels of fecal pollution in water samples (reviewed in (Boehm et al., 2013, Harwood et al., 2013)). Two human-associated methods in particular, HF183 (Bernhard and Field, 2000, Green et al., 2014) and HumM2 qPCR (Shanks et al., 2010), are consistently top performers in validation studies (Boehm et al., 2013, Layton et al., 2013), are being implemented in laboratories across the State of California (Griffith et al., 2013), and are currently under consideration by the United States Environmental Protection Agency for the development of national standardized protocols (Shanks et al., 2016). Even though there is a growing precedence for the use of these methods, it is important to note that they cannot differentiate between sewage, septage and raw human feces.

Although all three pollution types contain human feces, each source has a distinct route to environmental waters. Different physical and chemical stressors encountered by each pollution type may lead to changes in microbial composition that influence decay in an ambient water setting. For example, it is estimated that around 12–15% of the bacterial community in untreated sewage is derived from raw feces (Newton et al., 2015, Shanks et al., 2013). This shift in microbiota composition is likely due to a combination of mixing, residence time, and other selective conditions found in the sewer line environment. The resulting secondary human fecal pollution source has been shown to respond differently to the detrimental effects of ambient sunlight and indigenous microbiota in fresh and marine water decomposition experiments (Korajkic et al., 2013a, Korajkic et al., 2014, Sassoubre et al., 2015). It is likely that considerable changes in human fecal community structure occur during septic system processing that could also influence decomposition behavior. Furthermore, indicator methods relying on cultivation of bacteria or viruses compared to molecular technologies may exhibit variable decomposition patterns under different conditions in environmental waters polluted by any combination of sewage, septage, or raw fecal waste. Potential differences could have serious ramifications for some water quality applications, as well as provide new insights to improve indicator selection strategies for water quality management.

In this study, the decomposition of select cultivated and molecular indicators originating from fresh fecal matter, septage, and primary effluent sewage is reported with a focus on the influence of solar irradiation and microbiota indigenous to marine water. Ambient marine water mixed with each fecal pollution type was placed in dialysis bags and incubated in situ in a submersible aquatic mesocosm over a six day period. Experimental treatments included: A) exposure to both sunlight and marine microbiota, B) exposure to sunlight with reduced marine microbiota, C) exposure to marine microbiota and reduced sunlight (shading), and D) reduced marine microbiota and shading. Genetic and cultivated fecal indicators including fecal indicator bacteria (enterococci, E. coli, and Bacteroidales), coliphage (somatic and F+), Bacteroides fragilis phage (GB-124), and human-associated genetic indicators (HF183/BacR287 and HumM2) were measured in each sample. Findings suggest that indicators exhibit different decomposition trends across human fecal pollution types with important implications for water quality management applications.

2. Materials and methods

2.1. Site description and ambient water properties

The experiment was conducted in the Gulf of Mexico at Fort De Soto Park (Tierra Verde, FL; GPS coordinates: 27°38′17″N and 82°43′07″W) during April 2013. Over the study period, light intensity and temperature conditions were measured on an hourly basis using Onset Temperature and Light HOBO® UA 002–08 data loggers (Onset Computer Corporation Bourne, MA) placed at 5 cm (light treatment) and 30 cm (shaded treatment) below the water surface. Solar insolation incident on a horizontal surface and daylight cloud coverage information for the Fort De Soto Beach was retrieved from the National Aeronautics and Space Administration Atmospheric Science Data Center (http://eosweb.larc.nasa.gov/). Forty liters of ambient marine water was collected 18 h prior to the start of the experiment and held at 4 °C to minimize changes in indigenous microbial populations. For reduced marine microbiota treatments (B and D), ambient marine water was filtered through a 0.45 μm, 0.22 μm nitrocellulose membranes (Fisher Scientific, Pittsburg, PA) followed by NanoCeram cartridge filtration (Argonide, FL), as previously described (Korajkic et al., 2013a, Korajkic et al., 2014). Culturable enterococci, E. coli, and aerobic/facultative anaerobic heterotrophs in filtered marine water were enumerated on mEI, modified mTEC and trypic soy agar respectively as previously described (APHA, 2004, USEPA, 2002a, USEPA, 2002b); and found to be negligible [< 5 colony forming units (CFU) per 1 L of filtered marine water].

2.2. Human fecal pollution collection and preparation

Three human fecal pollution types were tested including primary effluent sewage, septage, and fresh fecal material. The sewage sample was collected from the Falkenburg Road Advanced Wastewater Treatment Plant (Tampa, FL). Septage represented a mixture of residential homes and was collected from a local service septic pumping truck (Tampa, FL). Ten fecal samples were collected from healthy volunteers. All fecal sources were collected in sterile containers and prepared on the same day as mesocosm deployment (day⁰) and immediately stored on ice (4 °C) with a holding time of less than 2 h. Sewage and septage dialysis bags consisted of 20 mL of primary sewage effluent or septage and 180 mL of ambient marine water (1:10 dilution). A composite of human material was prepared by mixing 75 g human feces from each individual with 1500 mL of sterile phosphate buffered water (0.0425 g/L KH₂PO₄ and 0.4055 g/L MgCl₂; pH 7.2) (Eaton et al., 2005). The fecal composite mixture was hand-shaken vigorously for 2 min until no visible solids were present, allowed to settle for 5 min followed by mixing 20 mL of the mixture with 180 mL of ambient marine water (final dilution: 1:200).

2.3. Submersible aquatic mesocosm

A submersible aquatic mesocosm was employed as previously described (Korajkic et al., 2013a, Korajkic et al., 2014). Briefly, the device consisted of a polyvinyl chloride pipe frame that encases 200 mL dialysis tubing diffusion bags situated at two depths including 1) a top level at 5–10 cm below the water surface (sunlight; Treatments A and B), and 2) a bottom level roofed with a heavy-duty, black plastic sheet at 25–30 cm below the water surface (shaded; Treatments C and D). Mixtures of each human fecal pollution type were placed in dialysis bags consisting of 75 mm flat width, 13–14 kDA pore size regenerated cellulose dialysis tubing (Spectrum Laboratories, Rancho Dominguez, CA), which are reported to have minimal sunlight attenuation [< 10%; (Korajkic et al., 2014)]. Duplicate dialysis bags for each pollution type and treatment combination were incubated in situ over a six day period and harvested at the beginning of the experiment (Day⁰), at 24 h (Day¹), 48 h (Day²), 72 h (Day³), 96 h (Day⁴), 120 h (Day⁵), and 144 h (Day⁶) totaling 168 individual samples. Each dialysis bag was transported to the University of South Florida (Tampa, FL) on ice in a sealed plastic bag containing 20 mL of ambient water to prevent desiccation (holding time < 60 min).

2.4. Fecal indicator enumeration by cultivation

Enterococci and E. coli were enumerated on mEI and modified mTEC agar as per standard methods (USEPA, 2002a, USEPA, 2002b). Somatic and F+ coliphage were enumerated by the double-agar overlay procedure according to standard methods (USEPA, 2001). Bacteroides fragilis GB-124 phage was also enumerated according to pre-existing protocols (ISO, 2001) with minor modifications as described (McMinn et al., 2014). Where necessary, ten-fold dilutions of water samples were made prior to analyses using sterile buffered water (0.0425 g/L KH₂PO₄ and 0.4055 g/L MgCl₂; pH 7.2) as the diluent. All data was expressed as either log₁₀ colony forming unit (CFU) or log₁₀ plaque forming unit (PFU) per 50 mL.

2.5. Preparation of reference DNA materials

Reference DNA sources consisted of plasmid constructs (Table S1; Integrated DNA Technologies, Coralville, IA) and salmon testes DNA (Sigma-Aldrich, St. Louis, MO). Please refer to Supplemental Information for detailed procedures on the preparation of plasmid constructs for calibration standards and internal amplification controls (IAC). All reference DNA material preparations were stored in GeneMate Slick low-adhesion microcentrifuge tubes (ISC BioExpress, Kaysville, UT) at −80 °C.

2.6. DNA purification

For each dialysis bag, 15–50 mL was filtered through a 0.45 μm polycarbonate filter (Fisher Scientific, Pittsburg, PA). Filters were placed in sterile 2 mL screw cap tubes containing silica bead mill matrix (GeneRite, North Brunswick, NJ) and stored at −80 °C until DNA purification (<2 months). Prior to DNA purification, 600 μL of 0.02 μg/mL salmon sperm DNA (Sigma-Aldrich, St. Louis, MO) was added to each bead mill tube. Bead beating was done with a MP FastPrep-24 (MP Biomedicals, LLC Solon, OH) at 4.0 m/s for 30 s. DNA purification was performed using DNA-EZ kit (GeneRite) according to manufacturer’s instructions. Extraction controls, with purified water substituted for ambient water, were performed with each sample batch (38 samples per batch). DNA was eluted with 100 μL elution buffer into low-retention microtubes (ISC BioExpress, Kaysville, UT) and a stored at 4 °C. All qPCR assays were performed within 24 h of DNA purification.

2.7. qPCR amplification

Six qPCR assays were used in the study including three general fecal indicator bacteria assays (Entero1a, GenBac3, and EC23S857), two human-associated methods (HF183/BacR287 and HumM2), as well as a sample processing control (SPC) assay (Sketa22) as previously reported (Bernhard and Field, 2000, Chern et al., 2011, Green et al., 2014, Haugland et al., 2010,Shanks et al., 2009, Shanks et al., 2014, Shanks et al., 2012) (Table S1) with the following modifications. Multiplex reaction mixtures contained 1X TaqMan^© Environmental Master Mix (version 2.0) except EC23S857 (Gene Expression MasterMix; Thermo Fisher Scientific, Grand Island, NY), 0.2 mg/mL bovine serum albumin (Sigma-Aldrich, St. Louis, MO), 1 μM each primer, 80 nM 6-carboxyfluorescein (FAM)-labeled probe, and 80 nM VIC-labeled probe. Multiplex reaction mixtures contained 10² copies (Entero1a, EC23S857, and HumM2), 5 × 10² copies (HF183/BacR287) or 10³ copies (GenBac3) of IAC template combined with either PCR grade water, 10 to 1 × 10⁵ target gene copies of respective calibration standard DNA, or 2 μL of DNA sample extract in a total reaction volume of 25 μL. All qPCR reactions were performed in triplicate using a 7900 HT Fast Real Time Sequence Detector (Thermo Fisher Scientific, Grand Island, NY). The thermal cycling profile for all assays was 2 min at 95 °C followed by 40 cycles of 5 s at 95 °C and 30 s at 60 °C. The threshold for each assay was manually set to 0.03 (GenBac3, Entero1a, HF183/BacR287, Sketa22 and EC23S857) or 0.08 (HumM2) and quantification cycle (C_q) values were exported to Microsoft Excel. To monitor for potential sources of extraneous DNA during qPCR amplification, six no-template amplifications with purified water substituted for template DNA were performed with each instrument run.

2.8. Amplification inhibition

Each test reaction was spiked with 10² to 10³ copies of an IAC plasmid construct to monitor for amplification inhibition. Amplification inhibition was identified in a two-step process starting with the determination of an instrument run-specific amplification interference threshold calculated from six NTC reactions [mean VIC NTC C_q + (3·standard deviation)]. Individual reactions from a test sample filter DNA extract can either “FAIL” (VIC C_q > interference threshold) or “PASS” (VIC C_q < interference threshold). If at least two of the three replicates “PASS”, then the filter DNA extract shows no evidence of amplification interference. However, if two or all three replicates “FAIL”, then data suggests the presence of amplification interference. Amplification interference can result from either inhibition (interference from substances that persist in the filter DNA extract after DNA purification) or competition between the native human-associated target sequence and IAC spike material.

To discriminate between inhibition and competition, an IAC range of quantification (ROQ) and competition threshold were determined for each instrument run. An instrument run-specific IAC ROQ was derived using VIC C_q data from the IAC spike associated with each standard concentration (10, 10², 10³, 10⁴, 10⁵ copies/reaction) in multiplex calibration curve reactions. The range of standard concentrations where at least two thirds or more of the replicates “PASS” (VIC C_q < interference threshold) for each standard dilution indicated the respective instrument run-specific IAC ROQ. Competition thresholds were defined as the calibration model FAM C_q that intersects the upper bound of a respective instrument run-specific IAC ROQ (Kelty et al., 2012). Any filter DNA extract exhibiting amplification interference (determined from IAC assay VIC C_q measurements), where the filter mean FAM C_q from the native sequence target assay (calculated from filter DNA extract triplicate FAM C_q measurements) was greater than the respective competition threshold indicated inhibition and was discarded from the study. Filter DNA extracts indicating evidence of amplification interference with filter mean FAM C_q values less than the respective competition threshold were influenced by competition between the IAC and the sample DNA target sequences rather than inhibition.

2.9. Sample processing controls

SPCs consisted of a fixed concentration spike of salmon testes DNA (0.02 μg/mL) followed by amplification of 2 μL DNA extract with the Sketa22 qPCR assay in test sample and extraction blank filters. For each DNA extraction batch preparation, a SPC acceptance threshold was calculated using C_q values from all three respective extraction blanks [Sketa22 extraction blank mean C_q + (3 ∙ standard deviation)]. Test sample mean C_q values below the respective SPC acceptance threshold indicated acceptable sample processing efficiency. Unacceptable values indicated that respective Entero1a, EC23S857, GenBac3, HF183/BacR287 and HumM2 C_q measurements were not suitable for data interpretation without accounting for sample matrix interference. For test samples that failed the SPC acceptance threshold with a sample mean Sketa22 C_q less than 37.7 C_q [Sketa22 LLOQ (data not shown)], each respective replicate Entero1a, EC23S857, GenBac3, HF183/BacR287 and HumM2 C_q measurement was adjusted as follows: Entero1a, EC23S857, GenBac3, HF183/BacR287 or HumM2 C_q - (test sample mean Sketa22 C_q – SPC acceptance threshold). Test samples that failed the SPC acceptance threshold and were ineligible for adjustment were discarded from study.

2.10. Calculations and statistical analysis

‘Mixed’ calibration models (generated from a master slope derived from six independent standard curves and instrument run-specific y-intercept control data), lower limit of quantification (LLOQ), and concentrations estimates of qPCR genetic markers were calculated using a Bayesian Markov Chain Monte Carlo approach on publicly available software WinBUGS, version 1.4.1 (Sivaganesan et al., 2008). LLOQ was defined as the upper bound of the 95% credible interval from repeated measurements of 10 copy per reaction calibration standard dilutions. Amplification efficiencies (E) were calculated as follows: E = 10^(−1/slope) −1 and qPCR target concentration estimates were reported as log₁₀ copy number per 50 mL. To evaluate trends in decomposition over the entire study period, concentration measurements for each pollution source, treatment and indicator combination were fitted by simple linear regression. A weighted two-way analysis of variance (ANOVA) comparing estimates of log₁₀ reduction at Day¹, Day², Day³, Day⁴ and Day⁵, was used to characterize treatment effects on a daily basis (weight was the reciprocal of the estimated variance of log₁₀ copies on a given Day¹⁻⁵). A Pearson correlation coefficient was used to characterize relationships between enterococci and E. coli cultivation and qPCR measurements. Calculations, unless otherwise noted, were performed using Statistical Analysis Software (Cary, North Carolina).

3. Results

3.1. Marine site conditions

The average water temperature and light intensity for the duration of the study were 24.9 ± 3.3 °C and 1299.1 lm/m² (Treatments A and B) and 24.1 ± 2.4 °C and 170.3 lm/m² (Treatments C and D). The average monthly solar insolation incident on a horizontal surface and monthly average daylight cloud coverage for the Fort De Soto Beach for the month of April was reported as 6.88 kWh/m²/day and 46.7%, respectively.

3.2. Trends in human fecal pollution sources over the study period

Distinct trends in fecal pollution type (sewage, septage, and fecal) decomposition were evident based on indicator concentration groupings (Supplemental Figs. S1 and S2). Simple linear regression analysis was used to determine whether concentrations of a particular indicator, pollution type, and treatment combination significantly changed over the course of the study period (Fig. 1). For indicator and treatment combinations with human fecal material as the pollution source, 96.4% (27/28) indicated a significant trend (p ≤ 0.05). Experiments with a sewage pollution source exhibited a similar trend to fecal material with 80% (32/40) of all indicator and treatment combinations yielding a significant (p ≤ 0.05) concentration change over the study period. However, unlike sewage and fecal material, septage experiments indicate that only 23.1% of indicator treatment combinations (6/26) significantly changed over the study period (p ≤ 0.05).

Figure 1.

Horizontal scatter plot showing the slope with 95% confidence interval (error bars) from fitted lines using log₁₀ concentration estimates per 50 mL from day⁰⁻⁵ samples for each indicator, treatment (A, B, C and D), and pollution type combination [Sewage (left), Fecal (middle), and Septage (right)]. The vertical solid line indicates a slope equal to zero. Slope ranges to the left of the zero line indicate a significant decay trend (p ≤ 0.05), while ranges that intersect the zero line indicate no significant change in concentration over the study period (p > 0.05).

3.3. Evaluation of treatment effects on sewage and fecal material

To assess indicator treatment effects on a daily basis, log₁₀ reduction values were calculated for each indicator, treatment (A, B, C and D), pollution type (sewage and fecal only), and sampling day (Day¹ to Day⁶) combination followed by a weighted two-way ANOVA. Due to the lack of change in indicator concentrations across the study period in almost 80% of experiments, septage data were excluded from these analyses. Findings are presented by indicator groupings consisting of cultivated bacteria (enterococci mEI and E. coli mTEC), cultivated phage (Somatic, F+, and GB-124), and qPCR (Entero1a, EC23S857, GenBac3, HF183/BacR287, and HumM2). For a complete list of all log₁₀ reduction values and weighted two-way ANOVA results, please refer to Supplemental Tables S2–S5.

3.3.1. Cultivated bacteria indicators

Two fecal indicator bacteria were measured by cultivation methods including E. coli mTEC and enteroccoci mEI (Fig. 2). In both pollution types, the largest log₁₀ reduction values were typically associated with treatment A (sun and microbiota). Log₁₀ reductions in treatment D (shaded, reduced microbiota) were minimal over time (<2 logs in both pollution types). Treatments were significantly different in 74.2% (89 of 120) of all pollution type, treatment, and day combinations (p < 0.046). Regardless of indicator or pollution type, there was a minimal significant treatment effect on day¹ (p > 0.05 in 80% of possible combinations). However, for days²⁻⁵, treatment D consistently yielded the smallest log₁₀ reduction values compared to other treatments (p < 0.026), except day² (E. coli mTEC; sewage treatment C and D; p > 0.05). Trends in treatments B and C differed between pollution types where they closely aligned with treatment A in sewage (Fig. 2). However, in fecal material experiments, treatments B and C resulted in intermediate log₁₀ reduction values.

Figure 2.

Scatter plot show ing log₁₀ reduction values for E. coli mTEC (Panels A and B) and enterococci mEI (Panels C and D) measurements in fecal (Panels A and C) and sewage (Panels B and D) pollution types from day⁰⁻⁵. Lines indicate average log₁₀ reduction values for each respective indicator, treatment and time point. Treatment A = Sunlight, Microbiota; Treatment B = Sunlight, Reduced Microbiota; Treatment C = Shaded, Microbiota; Treatment D = Shaded, Reduced Microbiota.

3.3.2. Phage indicators

Three phage indicators where measured including somatic and F+ coliphage, as well as GB-124 Bacteroides phage. For all feces material and septage experiments, phage indicator levels where too low to enumerate. For sewage experiments, log₁₀ reductions were greatest in treatments A and B compared to treatments C and D in all three phage indicator methods (Fig. 3). Overall, log₁₀ reductions in treatment A and B ranged from 0 to 2.5 while they were minimal in treatments C and D (< 0.25). For the somatic coliphage indicator on Day¹, there was no significant difference between all four treatments (p > 0.05). However, by Day² a significant difference (p < 0.001) between sunlight (A and B) and shaded (C and D) treatments was evident. No difference between shaded treatments (C and D) continued through Day⁵, while sun exposed treatments (A and B) decayed to extinction by Day⁴ (Fig. 3, Panel A). No significant difference between treatments for F+ and GB-124 indicators in sewage experiments was observed (p > 0.05), except for Day³ Treatment A with GB-124 (p = 0.041).

Figure 3.

Scatter plot reporting log₁₀ reduction values somatic coliphage (Panel A), GB-124 Bacteroides phage (Panel B), and F+ coliphage (Panel C) in sewage pollution type experiments from day⁰⁻⁵ samples. Lines indicate average log₁₀ reduction values for each respective indicator, treatment and time point. Treatment A = Sunlight, Microbiota; Treatment B = Sunlight, Reduced Microbiota; Treatment C = Shaded, Microbiota; Treatment D = Shaded, Reduced Microbiota.

3.3.3. qPCR indicators

Treatment effects on a daily basis were assessed with five qPCR indicators including EC23S857, Entero1a, GenBac3, HumM2, and HF183/BacR287. Decay of all qPCR assays in the fecal pollution type followed a biphasic decay pattern characterized by an initial phase with minimal log₁₀ reductions of approximately 1 log, followed by a phase with increased log₁₀ reductions ranging from 1 to 3 logs (Fig. 4). The length of each phase in days varied depending on pollution type with sewage exhibiting a much shorter initial phase compared to feces. The magnitude of decline in both phases was generally true for all qPCR indicators. Based on a two-way ANOVA comparison, only 29.3% (88 of 300) of pollution type, treatment and day combinations were significantly different (p < 0.046). Treatments were most influential on days²⁻⁵, accounting for 94.3% (83 of 88) of all significant treatment effects. Evaluation of log₁₀ reduction values across fecal and sewage pollution types indicated two distinct trends in treatment effects (Fig. 4). For feces, qPCR indicator measurements typically clustered closely together by day with significant differences observed in only 16.7% (25 of 150) of all qPCR indicator, treatment and day combinations (p < 0.05) (Fig. 4, Panels A–E). Sewage experiments showed a different trend (Fig. 4, Panels F–J), where treatments with reduced biota (B and D) and indigenous marine biota (A and C) begin to diverge on day² [6 of 30 combinations (30%); p < 0.039], peaking on day³ [18 of 20 combinations (90%); p < 0.016] regardless of qPCR indicator.

Figure 4.

Scatter plot showing log₁₀ reduction values for all qPCR indicators (GenBac3, HF183/BacR287, HumM2, Entero1a, and EC23S857) in fecal (Panels A–E) and sewage (Panels F–J) pollution sources from day⁰⁻⁵. Error bars indicate 95% Bayesian credible intervals. Lines indicate average log₁₀ reduction values for each respective indicator, treatment and time point. Treatment A = Sunlight, Microbiota; Treatment B = Sunlight, Reduced Microbiota; Treatment C = Shaded, Microbiota; Treatment D = Shaded, Reduced Microbiota.

3.4. Comparison of enterococci and E. coli culture-based and qPCR measurements

The strength of linear association between paired measurements of enterococci (mEI and Entero1a) and E. coli (mTEC and EC23S857) using cultivation and qPCR technologies is reported in Table 1. Pearson correlation coefficients for paired measurements were determined by human fecal pollution type (sewage, septage and fecal) and across all sources. The highest degree of correlation was for E. coli indicators across all pollution types (R² = 0.71) while the lowest was for enterococci in septage (R² = 0.03).

Table 1.

Correlations between cultivation and qPCR paired measurements of E.coli and enterococci.

E. coli			Enterococci
Fecal	Sewage	Septage	Fecal	Sewage	Septage
0.42(46)	0.49(43)	0.1(25)	0.38(46)	0.26(46)	0.03(37)
0.71(114)a			0.5(129)a

Parenthesis indicate number of measurements used to calculate R².

“a” denotes correlation with all samples combined.

3.5. Quality assurance and performance metrics

Calibration curve parameters and IAC thresholds for each qPCR assay are shown in Table 2. Calibration model R² values were all greater than 0.99, and E values ranged from 0.94 (GenBac3) to 1.03 (HumM2). Of the 168 total filters, nine filter DNA extracts (5.4%) were discarded from the study due to severe matrix interference based on SPC tests. SPC acceptance thresholds ranged from 30.9 C_q (Batches 3 and 8) to 33.8 C_q (Batch 5). Amplification inhibition was rarely identified in multiplex IAC experiments (0.08%). The majority of C_q measurements (91.2%) were above respective LLOQ thresholds regardless of qPCR assay with GenBac3 (3.0%) exhibiting the lowest and EC23S857 (18.5%) the highest percentages of failure. Results from 720 no-template and 360 extraction blank amplifications with purified water substituted for sample DNA indicated the absence of extraneous DNA molecules at concentrations above respective LLOQ levels in all experiments.

Table 2.

Summary of quality assurance and performance metrics for qPCR experiments.

Assay	Slope	Y-intercept		E	R2	LLOQ	IAC thresholds		Reference
		Estimate	Error				Interference	Competition
Entero1a	−3.38 ± 0.024	29.7 to 33.3	±0.10 to 0.36	0.98	0.997	31.9 to 33.8	29.7 to 35.1	21.5 to 22.9	(Haugland et al., 2010)
GenBac3	−3.47 ± 0.046	35.4 to 37.0	±0.16 to 0.42	0.94	0.991	32.3 to 33.8	26.0 to 27.3	21.5 to 23.1	(Haugland et al., 2010)
EC23S857	−3.33 ± 0.024	37.8 to 39.3	±0.13 to 0.21	1.00	0.997	34.7 to 36.3	33.5 to 35.1	27.8 to 29.3	(Chern et al., 2011)
HF183/BacR287	−3.40 ± 0.026	38.2 to 41.9	±0.14 to 0.63	0.97	0.997	35.1 to 39.0	31.1 to 32.8	28.0 to 31.7	(Bernhard and Field, 2000, Green et al., 2014)
HumM2	−3.25 ± 0.028	40.1 to 43.7	±0.19 to 0.61	1.03	0.996	37.3 to 39.9	32.9 to35.6	30.4 to 34.0	(Shanks et al., 2009)

E indicates amplification efficiency (E = 10^(−1/slope)−1) calculated from a master slope estimate.

LLOQ represents instrument run-specific lower limit of quantification values.

IAC threshold shows instrument run-specific internal amplification control interference and competition thresholds.

4. Discussion

4.1. Source of human fecal pollution matters

Decomposition trends of select cultivated and molecular indicators of fecal pollution originating from fresh human feces, septage, and primary effluent sewage was assessed over a six day period in a subtropical marine environment. Clear differences in decay over time were evident. The fresh fecal pollution type was most responsive to solar irradiation and indigenous microbiota experimental conditions (96.4% of all test conditions show significant change p < 0.05) closely followed by sewage (80%), while most indicators in septage experiments did not significantly change over the study period (23.1%). The observation that fresh feces was the most susceptible to environmental biotic and abiotic stressors is supported by the notion that this pollution type most closely represents the microbial community composition in the human gastrointestinal tract or primary habitat. In contrast, only an estimated 15% of untreated sewage microbial communities consist of human fecal derived bacteria suggesting that a majority of fecal bacteria die-off before sewage reaches a waste water treatment facility (Newton et al., 2015, Shanks et al., 2013). Our findings also suggest that the septage environment is subject to much harsher selective pressures compared to sewage resulting in a collection of well adapted indicator bacteria that are capable of persisting for more than 144 h in a marine environment. Extremely low correlation coefficients (≤ 0.1) between culture and qPCR paired measurements for E. coli and enterococci indicators further supports the idea that something distinctive occurs in the septage environment (Table 1). Septage is a complex mixture of household wastes such as feces, urine, food wastes, and cellulose combine with harsh chemicals and soil incubated in an enclosed, dark plastic or concrete tank for long periods of time. To date, little is known about the microbial communities in the septic system environment. Some have suggested that septage contains a higher abundance of general Bacteriodales per unit mass compared to sewage (Layton et al., 2013), as well as a greater abundance of E. coli uidA, Enterococcus and Bacteroides thetaiotaomicron 16S rRNA gene copies compared to sewage (Srinivasan et al., 2011). However, the decomposition of fecal microorganisms in septage remains poorly understood. Previous studies have shown that environmental stressors can be a dominant selective force during the transition from primary to secondary habitats (Anderson et al., 2005, Gordon et al., 2002, Whittam, 1989) in E. coli and enterococci populations, including septic systems (Gordon et al., 2002). Our experiments suggest that this phenomenon may not be restricted to E. coli and enterococci, but likely occurs in Bacteroides spp. populations based on qPCR measurements with GenBac3, HF183/BacR287, and HumM2 methods. Future research on Bacteroides spp. communities in the septage environment is needed to characterize the origin and diversity of these persistent fecal indicator subpopulations.

4.2. The role of solar irradiation and indigenous microbiota in human fecal pollution decomposition

The detrimental effects of ambient sunlight on the decomposition of culturable and genetic fecal indicators in recreational waters is the subject of numerous studies (Bae and Wuertz, 2009, Dick et al., 2010, Green et al., 2011, Korajkic et al., 2013a, Korajkic et al., 2014, Korajkic et al., 2013b, Menon et al., 2003, Walters et al., 2009, Wanjugi and Harwood, 2013, Wanjugi and Harwood, 2014); however, only a small number of these research efforts attempt to isolate the role of both indigenous microbiota and ambient sunlight in-situ (Korajkic et al., 2013b, Korajkic et al., 2014). Here we measure the influence of both solar irradiation and indigenous microbiota on a wide range of fecal indicators under ambient marine water conditions. Findings suggest that these stressors can alter fecal indicator decomposition, but to varying degrees depending on human fecal pollution type and incubation time. Similar to other studies (Green et al., 2011, Korajkic et al., 2014), both treatment conditions had a more profound impact on culturable indicators compared to genetic indicators and these effects mostly occurred on days²⁻⁵ (Green et al., 2011, Korajkic et al., 2014). Cultivated E. coli and enterococci exhibited different decomposition patterns in sewage compared to fresh feces (Fig. 2). In both pollution types, Treatment A (sunlight and microbiota) generally resulted in the greatest attenuation of indicator levels suggesting that the combination of solar irradiation and indigenous microbiota exerts the strongest influence on decomposition. However, Treatments B (sunlight and no microbiota) and C (shaded with microbiota) elicited intermediate log₁₀ reduction levels in fresh fecal experiments, while these same treatments in sewage were more closely aligned with Treatment A. Similar log₁₀ reduction trends between treatments in fresh feces material experiments implies that both stressors continue to influence decomposition evenly throughout the study period. In contrast, either solar irradiation or indigenous microbiota is sufficient to reduce cultivated bacterial indicators to Treatment A levels in sewage.

Trends in qPCR indicator data parallel other studies reporting that sunlight has a greater influence on genetic markers in the first 24–48 h in sewage seeded experiments (Korajkic et al., 2014, Sassoubre et al., 2015), but that indigenous microbiota play a key role in persistence after 48 h (Korajkic et al., 2014). These findings support previous in situ studies suggesting that predatory behaviors of indigenous protozoa may significantly contribute to sewage decomposition (Korajkic et al., 2013a, Korajkic et al., 2014). However, this delayed microbiota effect is absent in fresh fecal experiments exposed to the identical experimental conditions including the same marine indigenous microbial community. The exact mechanisms responsible for these differences remains unclear. A possible explanation may include predator and prey microorganism density- and trait-mediated effects (Banerji et al., 2015, Pernthaler, 2005, Surbeck et al., 2010, Wanjugi and Harwood, 2014) inherent to different pollution sources. For example, a previous freshwater decomposition study reports an abundance of Oligohymenophora ciliates in sewage seeded experiments (Korajkic et al., 2015). It is possible that interactions between these abundant sewage-derived protozoa and indigenous microbiota trigger increased predatory grazing of bacterial indicators. Additional research is warranted to characterize prokaryotic and eukaryotic communities in sewage and fresh fecal decomposition experiments, as well as any predator-prey modes of interaction.

4.3. Fecal indicator selection matters

There are numerous indicators available to characterize fecal pollution in ambient waters relying on cultivation of bacteria and viruses, as well as molecular methods targeting general and host-associated microbial genetic sequences. In this study, we measured 10 different fecal indicators to investigate decay trends of human fecal pollution originating from three fecal pollution types. Several trends emerged that could influence indicator selection in future water quality studies. In general, there was less than one log₁₀ reduction in indicator levels during the first 24 h of decomposition for all indicators except for phage methods where concentrations were too low to detect in most instances. The poor performance of phage indicators except somatic coliphage and Bacteroides fragilis phage (GB-124) in sewage (Fig. 1, Fig. 3) is likely due to the small volume that can be analyzed via double-agar overlay method (1–10 mL) indicating that larger volumes combined with concentration procedures may be necessary to ensure consistent detection of phage indicators in impaired recreational waters. Similar decay trends among all detectable indicators suggest that most indicators are suitable for recent human fecal pollution events (< 24 h) across different pollution types. However, after 24 h, different indicator groups (cultivated bacteria, cultivated phage, and molecular methods) exhibited distinct decomposition patterns in response to incubation time, pollution type, and environmental stressors. This suggests that one indicator may be more suitable than another, depending on the local climate conditions and intended application. For example, some applications employ more than one indicator utilizing a ratio-based approach to estimate the contribution of different animal groups to the total fecal load (Balleste et al., 2010, Reischer et al., 2011, Wang et al., 2010, Wang et al., 2013). Our findings imply that combining cultivated and molecular indicator measurements may confound validity of ratio-based interpretations, especially when more than one human fecal pollution type may be present and contaminants are more than 24 h old. Data also demonstrates that cultivated phage and bacterial fecal indicators respond to indigenous microbiota in markedly different ways suggesting that additional research is needed to explore the ramifications of differential decomposition mechanisms for water quality applications seeking to estimate health risk due to exposure to waterborne viral pathogens.

4.4. Water quality management implications

Current fecal indicators including most human-associated fecal source identification methods do not discriminate between fresh fecal, primary effluent sewage, and septage pollution types. We demonstrate that different fecal indicators do not always share similar fates in an ambient marine environment. Our results provide valuable information for water quality managers on future applications of fecal indicators tested in this study. Because there can be large discrepancies in the persistence of fecal indicators originating from different human wastes, we recommend that water quality managers document all possible human sources including the number of bathers at recreational sites especially young children including those in diapers and collect human waste reference samples in each area of interest. This information can help managers and researchers select the most suitable indicators based on anticipated human pollution types present and create an improved framework to interpret results. In addition to modifying field information collection procedures, our study provides useful information that could facilitate better implementation of fecal source identification methods for source allocation where the potential for differential decay is evident across human fecal pollution types and fecal indicator technologies. Researchers have suggested that allocating fecal sources with ratios of cultivated general indicators (E. coli or enterococci) and qPCR host-associated genetic indicators is likely not possible due to differential decay of these two different indicator groups, however it may be possible to combine qPCR measurements of general and host-associated indictors (Wang et al., 2013). While decomposition trends between qPCR measurements of general and host-associated indictors exhibited similar patterns in this study, results show clear differences between fecal pollution types further constraining the application of ratio approaches in recreational water settings due to this additional source of variation, especially when more than one human pollution type is present. Water quality managers attempting to establish links between cultivated measurements of waterborne viral pathogens and molecular measurements of general and host-associated indicators will likely face the same obstacles. For example, the lack of any measurable decomposition in septage experiments suggests correlations between estimated public health risks may also differ based on the pollution type. Future work must investigate the relative decomposition patterns of general, host-associated, and viral pathogens in a variety of ambient water environments and document levels of variability introduced into estimates of public health risk based on decay of fecal indicators across different human pollution types.

5. Conclusions

We describe the decomposition of 10 fecal indicators in three common human fecal pollution types in a subtropical marine environment. Efforts were made to mimic the release of each pollution type into surface waters by using an in situ submersible aquatic mesocosm combined with permeable dialysis tubing. For each indicator and human fecal pollution type combination, the influence of solar irradiation and ambient water indigenous microbiota was measured.

Highlights.

• Common human waste pollution types exhibit markedly different decomposition trends.

• Most indicator concentrations did not change over time in septage experiments (p > 0.05).

• Clear patterns are evident between viral, bacterial, cultivation and genetic indicator types.

• Indigenous microbiota can strongly influence indicator decay, especially after 48 h.

• Variation in decay trends can have serious implications for water quality applications.

Key findings include:

• Fresh fecal, septage, and primary effluent sewage human waste pollution types exhibit markedly different decomposition trends in an ambient marine environment.

• Cultivated bacteria, cultivated phage, and qPCR genetic indictors exhibit different decay trends regardless of human fecal pollution type.

• The concentration of most indicators in septage experiments did not significantly change (p > 0.05) over the study period.

• Exposure to ambient sunlight and indigenous microbiota have a cumulative influence on the decomposition of cultivated bacterial indicators in fecal pollution.

• Decay of sewage-borne somatic coliphage and Bacteroides fragilis phage were heavily influenced by the presence of ambient sunlight, but not indigenous microbiota.

• The influence of indigenous microbiota on qPCR indicator decay is strikingly different between fresh feces material and primary effluent sewage, especially between days²⁻⁵ where log₁₀ reduction values diverge in sewage experiments.

• The large range of decomposition trends observed between different fecal indicators and human fecal pollution types has serious implications for a number of water quality management and public health risk assessment applications.